Piezo-dependent surveillance of matrix stiffness generates transient cells that repair the basement membrane

Summary

Basement membranes are extracellular matrix sheets separating tissue layers and providing mechanical support, and Collagen IV (Col4) is their most abundant structural protein. Although basement membranes are repaired after damage, little is known about repair, including whether and how damage is detected, what cells repair the damage, and how repair is controlled to avoid fibrosis. Using the intestinal basement membrane of adult Drosophila as a model, we show that after basement membrane damage, there is a sharp increase in enteroblasts transiently expressing Col4, termed “matrix mender” cells. Enteroblast-derived Col4 is specifically required for matrix repair. The increase in matrix mender cells requires the mechanosensitive ion channel Piezo, expressed in intestinal stem cells. Matrix menders are induced by the loss of matrix stiffness, as specifically inhibiting Col4 crosslinking is sufficient for Piezo-dependent induction of matrix mender cells. Our data suggest that epithelial stem cells control basement membrane integrity by monitoring stiffness.

Graphical Abstract

eTOC Blurb:

Basement membranes are a conserved extracellular matrix that provide support to tissues, yet little is known about their repair. Stricker et al find that in the Drosophila midgut, basement membrane damage is detected by Piezo as a loss of stiffness, prompting local epithelial cells to express collagen IV for repair.

Introduction

Extracellular matrix (ECM) is an essential conserved component of all animal tissues. In addition to providing structural support, ECM influences cell adhesion, shape, signaling, and behaviors. When tissues are damaged, ECM needs to be repaired, and pathological ECM damage is a feature of many human diseases^1,2. Despite the critical importance of ECM repair, little is known about the cellular processes and molecular signals that orchestrate repair. At the most fundamental level, it is unknown whether and how cells detect ECM damage. Moreover, ECM proteins, including collagens, are among the most stable proteins in the body, with the longest-lived exhibiting half-lives on the order of an animal’s lifetime^3–6. ECM longevity presents a challenge, as repairing with excess ECM causes fibrosis. In this study, we identify a mechanism that cells use to sense damaged matrix, and they respond by transiently producing matrix until repair is completed.

ECM can be subdivided into two types, stromal matrix and basement membrane, the focus of this study. Basement membranes are thin sheets that underlie epithelia and wrap around muscles and organs. Their most abundant structural protein is Collagen IV (Col4)^7,8, a heterotrimer that self-assembles in vitro into networks and provides structural support⁹. Self-assembly raises questions about the nature of basement membrane repair. Is there a continuing self-assembly process for maintenance that also functions passively to repair damage? Or is repair an active process, with cells sensing damage and directing repair?

Basement membranes are ancient, arising along with multicellularity itself¹⁰. Given the evolutionary conservation, Drosophila offers an excellent model to address questions of basement membrane homeostasis and repair. Flies have the same basement membrane structure and composition as mammals, with all the main components but fewer family members. There is only one type IV collagen (Col4) heterotrimer composed of the two collagen subunits Col4a1 and Col4a2 (aka Viking, Vkg); there are two laminin heterotrimers, one nidogen gene, and one perlecan gene¹¹. Functional GFP-tagged alleles of Vkg offer tremendously powerful reagents for imaging and genetics¹². We use the Drosophila posterior midgut as a model to analyze basement membrane repair. The fly gut is a monolayer epithelium derived from stem cells that generate two lineages, enterocytes and enteroendocrine cells, with intermediates in between^13–15. Between the epithelium and the surrounding peristalsis muscles there is basement membrane, and it is continuous with the basement membrane enveloping the muscles (Fig. 1A). We damage the basement membrane by feeding the flies dextran sodium sulfate (DSS), which gets visibly lodged in the gut basement membrane, fractures its structure, and reduces its stiffness¹⁶. After stopping DSS feeding, basement membrane repair occurs within two days, providing a temporal window for analyzing repair¹⁶.

Figure 1. “Matrix mender” gut cells upregulate Col4 after basement membrane damage. (A) Left: Drosophila midgut.

(A) Left: Drosophila midgut. The gut tube is covered with peristalsis muscles (magenta) wrapped in basement membrane (green). Underneath the muscle layer, the gut monolayer epithelium (blue nuclei/DAPI) comprises absorptive enterocytes (blue), stem cells (brown), enteroblasts (gray), and enteroendocrine cells (red), all sitting on basement membrane continuous with that of the muscles (green). Scale bars, 10 μm.

Right: DSS damage. DSS (orange dots) is ingested, transported through the enterocytes, and deposited in the underlying basement membrane where it fractures the basement membrane making it less stiff, resulting in visibly dysmorphic muscles (described in Howard et al¹⁶). Basement membranes and muscle shapes are fully repaired 2 days after DSS is removed. Both repair and maintenance of the basement membrane require basement membrane components and regulators (see Fig. S1 and Howard et al¹⁶ for data). Scale bars, 10 μm.

(B-G) Collagen IV (Vkg) in the gut basement membrane comes from the fat body in undamaged conditions. B depicts the experiment knocking down GFP of Vkg-GFP specifically in the fat body with c564-Gal4, data in C (en face view). After DSS damage, some gut Vkg-GFP came from another source (D,E). Each dot in E represents a gut. Mean +/− SEM, significance by t-test. Cross-sections (F,G) show the new Vkg source deposits Vkg-GFP between the epithelium and the muscles. Full data set for fat-body knockdown and the specificity of the c564-Gal4 driver shown in Fig. S2. Scale bars, 10 μm.

(H-K) Col4 reporter Cg>GFP identified unknown cells that respond to DSS in the gut. Anterior (left), posterior (right). Posterior R4 midgut is the focus of this study. Scale bar, 100 μm in H,J, scale bar, 20 μm in I,K.

(L-M) Collagen-expressing cells (“matrix menders”) increased in number during basement membrane damage, then decreased during repair (L). Shaded regions represent SEM. 5–21 guts analyzed per timepoint and treatment, see Fig. S3A for dataset. Significance by t-test. In addition to number, expression level increased after DSS (M), see arrowheads in I,K. Each dot represents a gut. Mean +/− SEM, significance by t-test.

(N-Q) Feeding bleomycin did not damage the basement membrane, as indicated by muscle shape (N-P), even though it did cause DNA damage and stem cell proliferation (see Fig. S3B,C). Bleomycin did not induce Cg>GFP+ matrix menders (Q). Mean +/− SEM, significance by t-test. Scale bar, 10 μm.

(R-U) Feeding paraquat damaged the basement membrane, indicated by changes in muscle shape aspect ratio (R-T). Paraquat induced Cg>GFP+ matrix menders (U). Mean +/− SEM, significance by t-test. Scale bar, 10 μm.

In this study, we find that genes encoding all major basement membrane components are required for both repair and maintenance. However, after damage, the source of Col4 for the midgut changes: without damage, Col4 originates from the fat body, but after damage, Col4 is transiently expressed locally by gut epithelial cells we call “matrix menders,” which are specialized short-lived enteroblasts. Knocking down Col4 in enteroblasts specifically inhibits gut basement membrane repair. Damage induces matrix menders via the mechanosensitive channel Piezo, expressed in neighboring intestinal stem cells (ISCs). Indeed, specifically reducing basement membrane stiffness by reducing Col4 crosslinking is sufficient to induce matrix-mender cells in a Piezo-dependent manner. Our results suggest that epithelial stem cells surveil basement membrane integrity by monitoring its stiffness, responding to defects by generating short-lived repair cells.

Results

Basement membrane repair and maintenance require many of the same components.

DSS directly damages the posterior midgut basement membrane of Drosophila¹⁶. After ingestion, DSS is transported through enterocytes to accumulate in the adjacent basement membrane (Fig. 1A). This mechanism is different from mice, where DSS is excluded by the intact epithelium^16,17. In flies, DSS damages basement membranes, making them visibly fractured by TEM, less dense as measured with fluorescence and super-resolution microscopy, and less stiff as measured in a tensile strain assay¹⁶. The basement membrane mechanical damage changes the shape of the gut peristalsis muscles, which are outlined by Col4 (Fig. 1A, right side). A similar muscle shape is observed in fly guts lacking the Col4 crosslinking enzyme Peroxidasin (Pxn)^16,18,19, and these basement membranes also have reduced tensile stiffness^16,19,20. In both instances, gut muscles become dysmorphic when enveloped by weakened basement membrane, and thus we can infer basement membrane stiffness from gut muscle shape, measured as aspect ratio. Importantly, two days after removing DSS and feeding flies regular food, there is full restoration of basement membrane structure and the muscle shape¹⁶.

Previously, we determined that repair after DSS damage requires synthesis of new basement membrane proteins, including vkg, LamininB1, and Pxn¹⁶. We asked whether other conserved basement membrane components were required for repair, targeting Perlecan (trol), Nidogen (Ndg), and the Col4 extracellular chaperone, SPARC. We used two different RNAi lines to knock them down everywhere with TubP-Gal4, initiating knockdown in 6-day old adult females. After two days, flies were fed DSS for two days, then repair was allowed for two days (6 days total knockdown). By analyzing muscle shape, we found trol, Ndg, and SPARC were all required for basement membrane repair (Fig. S1A–F). The requirement for new basement membrane proteins indicates that damaged basement membranes are replaced with new matrix. These same proteins are required for basement membrane maintenance without externally applied damage, as knocking down trol, Ndg, or SPARC for 10 days in adults caused dysmorphic gut muscles (Fig. S1K–P), whereas knocking down non-matrix extracellular proteins Gbp1 or Gbp2²¹ had no effect (Fig. S1G–J, Q–T). From this, we conclude that although basement membrane is replaced during normal homeostatic maintenance, it is replaced faster during repair.

Collagen IV is upregulated in matrix mender cells following damage.

Although basement membranes are maintained throughout the body, we expect damage-triggered repair to be a localized process. The Col4 in many Drosophila tissues originates from a distant organ, the fat body^22,23, and we found that was true for the undamaged gut by expressing tissue-specific dsRNA against GFP in vkg-GFP/vkg+ heterozygotes. vkg-GFP/vkg+ gut basement membranes fluoresce green, but when dsRNA against GFP was expressed ubiquitously with TubP-Gal4, there was virtually no gut fluorescence, demonstrating the effectiveness of GFP knockdown (Fig S2 A–D). Importantly, the GFP-knockdown animal is functionally equivalent to a vkg heterozygote, viable and healthy, so we do not expect to trigger any compensatory mechanisms²³. When GFP was knocked down specifically in the fat body with c564-Gal4, gut GFP fluorescence was also extremely low, similar to ubiquitous GFP knockdown (Fig. 1B,C and Fig. S2A–D; specificity of c564-Gal4 shown in Fig. S2I,J), indicating most of the midgut Col4 originates from the fat body, a distant source. However, when these animals were fed DSS to damage the gut basement membrane, the gut GFP fluorescence increased significantly, indicating that during repair Col4 was coming from another source besides the fat body (Fig. 1C–G; see also Fig. S2E–H). In cross-section, the DSS-induced Vkg-GFP was deposited between the gut epithelium and peristalsis muscles, suggesting a repair-specific local source of Col4 in the gut epithelium (Fig. 1F,G).

To identify a local source of Col4, we used Cg-Gal4, a reporter of Collagen IV gene expression in which Gal4 is under control of a cloned collagen enhancer whose genomic location is between the head-to-head promoters of the two collagen genes Col4a1 and vkg²⁴. In control guts, Cg>GFP was very faintly expressed in only a few cells, but after 2 d of DSS damage the number and intensity of Cg>GFP cells in the gut increased substantially, from 1.8% to 5.7% of epithelial cells in the posterior midgut (R4) (Fig. 1H–M, Fig. S3A). Analyzing their numbers over time, we found that Cg>GFP cells increased after only one day of DSS feeding, peaked after the second day of DSS, and then subsided after DSS is removed. Interestingly, Cg>GFP expression was no longer observed 2 days after DSS removal, when the gut basement membrane was repaired (Fig. 1L). We dubbed these Cg>GFP cells “matrix menders” because they express Col4 strongly in response to basement membrane damage but not after repair is complete. Analyzing other genes required for basement membrane repair, we found that reporters for Pxn and Laminin expression (Pxn>GFP and LanB1>GFP) were active in the gut without basement membrane damage; the number of Pxn-expressing cells doubled after damage, but the number of LanB1-expressing cells did not increase (Fig. S4).

Matrix menders arise in response to basement membrane damage rather than epithelial damage.

In addition to damaging basement membranes, DSS also damages cells²⁵. To assess the damage profile that induces matrix menders, we tested two additional drugs known to damage the fly gut. Bleomycin is a DNA-damaging agent that targets enterocytes, causing tissue disorganization, cell loss, and a resulting increase in stem cell divisions²⁵. After feeding bleomycin for 2 days, DNA damage was evident by anti-H2AvD staining, and increased stem cell divisions were evident by anti-pH3 staining (Fig. S3B,C), yet the basement membranes remained undamaged as determined by muscle shape, and matrix mender cells were not evident (Fig. 1N–Q). Next we tested paraquat, which damages the fly gut by inducing reactive oxygen species and increasing stem cell divisions^26,27. Feeding flies paraquat for 16 h, reported to maximize damage without causing lethality²⁶, caused unexpected basement membrane damage, evident by altered muscle shape, and it significantly induced matrix menders (Fig. 1R–U), similar to the levels seen in response to DSS. These results indicate that matrix mender cells arise specifically to basement membrane damage, rather than epithelial damage.

Matrix menders are a subset of enteroblasts.

The cell types of the fly posterior midgut are well known (Fig. 2A) and can be identified with established cell markers. In DSS-fed guts, the Cg>GFP matrix menders did not express the intestinal stem cell marker Dl, but interestingly, about 75% of matrix menders were adjacent to Dl+ stem cells (174/236), suggesting that matrix menders are either enteroblasts or pre-enteroendocrine cells. Nearly all Cg>GFP matrix menders were positive for the enteroblast marker Su(H)GBE-lacZ, indicating that matrix menders are enteroblasts (Fig. 2B,C). Not all enteroblasts are matrix menders, however, as only 43% of the Su(H)GBE-positive enteroblasts were also positive for Cg>GFP (Fig. 2C). In guts that were not damaged by DSS, Cg-expressing cells were still enteroblasts (81%) but fewer enteroblasts expressed Cg (21%) (Fig. 2C). Thus, some but not all enteroblasts are matrix menders. The number of enteroblasts is known to increase in response to DSS²⁵, and we found that after 2 d of DSS, the number of enteroblasts increased from an average of ~7/100 cells to ~11/100 cells (Fig. 2D and Fig. S6A). Nearly all this increase in enteroblasts can be accounted for by the increase in collagen-expressing matrix mender cells (Fig. 1L).

Figure 2. Matrix menders are a subset of enteroblasts that provide Col4 necessary for repair and are short-lived.

(A)Lineage of gut epithelial cell types.

(B)Matrix menders (Cg>GFP, green) expressed an enteroblast marker (Su(H)GBE-lacZ, magenta). Scale bar, 10 μm.

(C)In both damaged and undamaged conditions, nearly all matrix menders were enteroblasts expressing Su(H)GBE-lacZ, but not all enteroblasts were matrix menders expressing Cg>GFP. After damage, more enteroblasts expressed Cg>GFP, evaluated by Fisher’s exact test. Number of cells scored is indicated.

(D)Total enteroblasts increased in response to DSS, SEM indicated. Each dot represents 5–19 guts. Complete dataset in Fig. S6A.

(E)Repair required Col4 expressed from enteroblasts. Col4a1 was knocked down in different tissues, and adult gut muscle shape was compared in damaged-and-repaired guts to unchallenged guts. When Col4a1 was fully expressed (first two columns), nearly all guts had undamaged healthy basement membranes, both unchallenged and after repair. Col4 from the fat body was required for development and/or homeostasis, as guts appeared damaged even without challenge. Enteroblast Col4 was required specifically for repair. Col4 was not required from enterocytes. Numbers of guts evaluated is shown at column base. Guts were scored blinded to sample identity. Knock-down in other gut cell types and knockdown of vkg confirmed that Col4 is specifically required from enteroblasts for repair, see Fig. S5A–C. Significance evaluated by Fisher’s exact test.

(F-H) Dynamics of matrix mender cells during and after collagen expression determined by G-TRACE. Overview (F) adapted from Evans et al.²⁹ DSS increased the number of real-time Cg-expressing cells (RFP+, magenta); after repair, Cg-expressing RFP+ cells declined quickly, but most did not transition into lineage traced GFP+ cells (green). Cells expressing both RFP+ and GFP+ were counted only as RFP+ (magenta) to count each cell only once. Complete dataset in Fig. S6C,D.

I,J) Cg>GTRACE real-time expressing cells (RFP+, magenta) underwent cell death, expressing Cleaved Caspase 3. Scale bar, 20 μm. Cell death of Cg-expressing cells was also observed without DSS, see Fig. S6G. Shading represents SEM in G,H,J.

Collagen IV from enteroblasts is required for repair.

We wanted to determine if the Col4 from matrix-mender enteroblasts was important for repair, but unfortunately knocking down Col4 with Cg-Gal4 would target collagen in other tissues, such as the fat body. Instead, we knocked down Col4a1 in specific cell types with different Gal4 drivers, including the enteroblast driver Su(H)GBE-Gal4 (Fig. 2E). With collagen removed from each potential source, we compared gut morphology in undamaged guts to those that were DSS-damaged and allowed to repair, scoring samples blinded to identity or treatment. When Col4a1 was knocked down in the fat body with c564-Gal4, many guts had dysmorphic muscles even without damage, and morphology was similar when treated with DSS and allowed to repair; these results indicate that Col4 from the fat body is important for gut basement membrane development or homeostasis as expected, but it is not specific for repair. We next knocked down Col4a1 in enteroblasts with Su(H)GBE-Gal4, which includes matrix menders; although Su(H)GBE is expressed in some larval tissues including wing and tracheae²⁸, it is not expressed in known collagen source tissues (adult fat body, Malpighian tubules, or ovary, Fig. S5D–G). Gut muscle morphology was normal without DSS treatment, but repair was severely compromised, with 70% remaining damaged two days after removing DSS. To confirm the specificity of RNAi-mediated knockdown, we knocked down vkg (Col4a2) with Su(H)GBE-Gal4, and this also interfered specifically with repair and not development (Fig. S5A,B). These results indicate that Col4 from the matrix mender cells is required specifically for midgut basement membrane repair. When Col4a1 was knocked down with Myo1a-Gal4 in enterocytes, the terminally differentiated cell type arising from enteroblasts, guts appeared like no-knockdown controls, normal both without damage and after repair; these results indicate that after differentiation, enterocytes do not supply Col4 for development, homeostasis, or repair. We also knocked down Col4a1 in the intestinal stem cells and enteroendocrine cells (Fig. S5C), and morphology appeared similar without damage and after repair. Only the collagen from enteroblasts is specific for gut basement membrane repair.

Most matrix menders are removed from the gut epithelium after repair.

We were interested in what happened to the matrix mender cells after repair, when they were no longer detectable. Did they turn off collagen expression, did they differentiate, or did they die? To address this question, we performed lineage tracing of the Cg-Gal4 expressing matrix menders using the G-TRACE system²⁹, in which cells actively expressing Col4 were labeled with nlsRFP and cells that previously expressed Col4 were permanently labeled with nlsGFP (Fig. 2F). We analyzed the number and fate of collagen-expressing cells in the gut, and as expected, the RFP+ cells behaved like the Cg>GFP matrix mender cells, except that their numbers peaked slightly higher and two days later (compare Fig. 2H to Fig. 1L). Although the difference is explained by nuclear RFP persisting longer than cytoplasmic GFP, it also indicates that matrix mender cells persist after they turn off collagen expression. After the peak, the number of RFP+ cells decreased rapidly. We expected that the RFP+ cells would transition into GFP+ lineage cells, because enteroblasts can remain dormant for weeks before differentiating into enterocytes in a healthy gut epithelium³⁰. Contrary to our expectations, however, as RFP+ cells disappeared, the corresponding pulse of GFP+ cells was much smaller (Fig. 2H). These results indicate that most matrix menders are removed from the gut epithelium after repair is complete. Matrix menders are capable of differentiating into enterocytes, however, as the enterocyte-specific antibody Pdm1 stained about 15% (50/340) of GFP+ lineage-expressing cells after 2 days recovery but did not stain RFP+ collagen-expressing cells (Fig. S6F,G). Matrix menders never became enteroendocrine cells, as no G-TRACE labeled cells were labeled with anti-Prospero (Fig. S6E). To determine if some matrix menders were dying, we stained Cg>G-TRACE flies with anti-Cleaved Caspase 3 (CC3). In vehicle-treated guts, we found ~9%of RFP+ matrix mender cells were positive for the cell death marker (Fig. S6H). In guts recovering from DSS damage, the percentage of dying matrix menders increased over time: at 0 d post DSS, 5% were labeled as dying, but by 4 d post DSS feeding (2 d recovery) 15% RFP+ cells were labeled as dying (Fig. 2I,J and Fig S6H). Although kinetic analysis is limited in fixed samples, it appears that most matrix menders do not differentiate into enterocytes and are short-lived, a property of enteroblasts generally after DSS treatment (Fig. S6B). Taken together, these results indicate that as repair is completed, matrix menders first turn off collagen, then a few differentiate while most die.

Piezo is required for matrix menders to arise in response to damage.

How does the gut sense damage to turn on Col4? We previously demonstrated that DSS damage makes basement membranes less stiff, with a reduced elastic modulus measured in a quantitative tensile strain assay¹⁶, and we envision that most types of matrix damage would result in reduced stiffness (increased flexibility). Thus, we reasoned that basement membrane damage might be detected by a mechanosensitive mechanism such as the ion channel Piezo, known to be active in the fly gut^13,31,32. To test Piezo’s role in basement membrane repair, we analyzed Piezo^KO null flies, which are homozygous viable. After damaging with DSS, Piezo^KO flies were unable to repair basement membranes (Fig. 3A–C), but unlike other matrix repair genes (Fig. S1), Piezo was not required for basement membrane maintenance (Fig. 3D–F). Analysis of a second independent mutation, Piezo^MI04294, confirmed its unusual repair-specific function (Fig. S7A–F). To determine if Piezo is required for detection of basement membrane damage, we quantified Cg>GFP matrix mender cells in Piezo^KO versus control midguts after DSS damage. Piezo was required for matrix mender induction: in controls, 5.8% of epithelial cells were matrix menders, but in Piezo^KO only 1.9% were and their Cg expression level was reduced (Fig. 3G–J). Thus, Piezo is required for matrix menders to arise in response to damage.

Figure 3. Piezo is required for basement membrane repair and for matrix menders to arise and is expressed in the intestinal stem cells.

(A-F) Piezo was required for basement membrane repair but not maintenance in adult guts. See Fig. S7A–F for confirmation with another allele. Mean +/−SEM. Significance by t-test. Scale bar in B for A-E, 10 μm.

(G-J) Piezo was required for Cg>GFP matrix mender cells to arise in response to DSS damage. Both the number and expression level of Cg>GFP cells were significantly reduced in Piezo^KO flies. Each dot represents a gut (I,J). Mean +/−SEM. Significance by ANOVA (I) and t-test (J). Scale bar, 20 μm.

(K-O) Piezo-transcriptional reporter (Piezo^KI-Gal4, green) was expressed in Dl+ intestinal stem cells (magenta), flies fed normal food. Almost all Piezo+ cells were Delta+ intestinal stem cells, yet only 36% of Dl+ intestinal stem cells express Piezo. For Venn diagram, the circle size and amount of overlap are representative. Scale bars, 10 μm in K, 5 μm in L.

(P-Q) Cg>GFP matrix mender cells (green) were adjacent to intestinal stem cells (anti-Delta, magenta). Scale bar, 10 μm. The number of cells analyzed is reported at column base for M,N,Q.

Piezo is expressed in a subset of intestinal stem cells.

For Piezo to monitor basement membrane damage, it must be present prior to damage. To localize Piezo, first we tried a functional genomic GFP-Piezo fusion protein reported to be extremely dim in neurons³³, but we were unable to detect GFP in the gut. Next, we turned to a Gal4-based Piezo>GFP transcriptional reporter, expressed outside the gut epithelium in tracheal cells that penetrate the gut muscles (Fig. S7G,H) and in small cells within the gut epithelium. On regular food, 93% of Piezo>GFP cells were positive for the intestinal stem cell marker Dl but only 36% of Dl+ stem cells expressed Piezo (Fig. 3K–O). Our results are similar to a previous report that Piezo is expressed in 40% of intestinal stem cells¹³. We found that stem cells are often adjacent to matrix menders (Fig. 3P–Q), suggesting that stem cell divisions give rise to matrix menders in a Piezo-dependent manner.

Piezo detects decreased basement membrane stiffness to activate matrix menders.

What aspect of basement membrane damage does Piezo detect? Piezo is a transmembrane ion channel activated by mechanical stretch to the plasma membrane³⁴, and the plasma membranes of intestinal stem cells are in direct contact with the underlying basement membrane³⁵. A damage-induced loss of tensile stiffness in the basement membrane is expected to transfer more tension to neighboring plasma membranes, resulting in membrane stretching and Piezo activation. Thus, we hypothesized that Piezo detects basement membrane damage via its loss of stiffness. As a test, we specifically reduced basement membrane stiffness in otherwise healthy and undamaged animals, by inhibiting the Col4 crosslinking enzyme Peroxidasin (Fig. 4A). Peroxidasin mouse mutant tissue has reduced basement membrane stiffness (elastic modulus), measured as increased deformation (stretch) in response to tension²⁰. Using the same quantitative tensile strain assay, we recently reported that inhibiting Peroxidasin in wild-type adult Drosophila by feeding 100 μM of the suicide substrate phloroglucinol (PHG) for 5 days decreases basement membrane stiffness by about half¹⁹. The PHG-induced decrease in basement membrane stiffness was evident in the muscle aspect ratio, both at 100 μM and even more strongly at 5 mM, indicating a dose response (Fig. 4B–D). Excitingly, 100 μM PHG alone without other damage induced matrix menders (3.5% of gut epithelial cells), and 5 mM PHG induced even more matrix menders (5.3%), at a level similar to DSS induction (Fig. 4E,F,H, compare to Fig. 1L and S3A). Thus, the loss of stiffness specifically activated matrix mender cells. Further, the mechanical induction of matrix menders required Piezo, as PHG-induced matrix menders were completely suppressed in the Piezo^KO background (Fig. 4G,H). These results indicate that Piezo responds to the reduction in basement membrane stiffness to give rise to collagen-expressing matrix mender cells, positioned adjacent to the basement membrane (Fig. 4I,J) where they can effect repair.

Figure 4. Piezo detects decreases in basement membrane stiffness to activate matrix menders.

(A) PHG inhibits collagen IV crosslinking enzyme Pxn.

(B-D) Feeding PHG reduced basement membrane stiffness in a dose-dependent manner, evidenced by muscle aspect ratio. Quantitative tensile stiffness measurements were reported previously by Peebles et al.¹⁹ Significance evaluated by ANOVA. Mean +/−SEM. Scale bar, 10 μm.

(E-F) Reducing collagen crosslinking induced matrix mender cells. Scale bar, 20 μm.

(G-H) Piezo was required for matrix mender cells to arise in response to reduced collagen crosslinking. Each data point represents an individual gut. Mean +/−SEM. Significance by ANOVA.

(I-J) Matrix mender cells have extensive contact with basement membrane, both with and without DSS (no recovery), positioning them well to repair it. Collagen is labeled with the triple-helix binding protein, CNA35 (green). Scale bar, 10 μm.

Discussion

Using the basement membrane around the Drosophila posterior midgut as an experimental model, we found that basement membrane damage gives rise to and is repaired by a distinct set of local epithelial cells, matrix mender cells, defined by their expression of the Collagen IV reporter Cg-Gal4. Matrix mender cells are a subset of enteroblasts that supply Col4 specifically required for basement membrane repair. After repair is completed, matrix menders turn off collagen expression, and most of them die. In contrast, Col4 is a long-lived molecule, and too much leads to fibrosis. Perhaps the fly gut avoids fibrosis by making a long-lived molecule from a short-lived cell.

In addition to the data shown here, supporting data for enteroblasts expressing Col4 come from mining published RNA sequencing data from single gut nuclei: transcripts for both Col4 subunits, Cg25C and Vkg, were identified almost exclusively in cells called progenitors of enterocytes (proECs), defined as a subset of enteroblasts; further, after DSS treatment, the number of proECs expressing the collagen transcripts doubled, results similar to our findings but using very different methods to identify collagen expression and cell identity^36,37.

We expect that even in undamaged control guts there may be intrinsic low levels of basement membrane damage, given the proximity of the gut to an unpredictable external environment and the mechanical wear and tear of peristalsis, and this low level of damage would induce a weak matrix-mender response. Indeed, in undamaged guts, about 20% of enteroblasts or 1.8% of epithelial cells express Cg>GFP, at generally low levels; our various control data show this to be a somewhat variable number, consistent with stochastic low levels of matrix damage. After experimentally induced basement membrane damage, the numbers of matrix mender cells increase dramatically: intestinal stem cell divisions generate more total enteroblasts, and a greater percentage of them express collagen (43%), with the result that about 6% of total epithelial cells express collagen, generally at high levels.

How is matrix damage detected to give rise to cells that repair it? Matrix menders arise in response to three independent challenges that damage the basement membrane – DSS, paraquat, and PHG – but not in response to the DNA damaging agent bleomycin. Although DSS and paraquat may cause non-specific damage, PHG is highly specific, reducing Col4 covalent crosslinking^18,38 and decreasing basement membrane stiffness in response to tensile strain, meaning that the basement membrane will stretch more in response to a given pulling force or tension^16,19. The gut is a mechanically active environment, and increased basement membrane stretching would stretch neighboring plasma membranes in the gut epithelium, activating Piezo, a stretch-activated transmembrane ion channel required specifically for matrix menders to arise and for basement membrane repair. Unlike other genes we identified that are required for repair – Nidogen, Perlecan, Laminin, Collagen IV, Peroxidasin, and SPARC – Piezo is not required for basement membrane maintenance but repair only. Our data indicate that Piezo detects reductions in basement membrane stiffness and signals for the generation of matrix menders.

Simple and general model of basement membrane repair

Our data are consistent with a simple stem-cell based model of basement membrane repair: intestinal stem cells directly monitor tissue stiffness, with matrix damage allowing greater cell stretching, activating Piezo. As a cation channel, Piezo activation increases intracellular calcium levels, and sustained cytoplasmic calcium is sufficient to induce intestinal stem cell proliferation^39,40. Thus, we propose that mechanically activated stem cells give rise to matrix menders that express and secrete Col4 locally, restoring the structure of the basement membrane. Their job complete, these cells turn off collagen expression and most of them die. Very recently it was shown that stiffness regulates mouse intestinal stem cells via Piezo⁴¹, and this may represent part of the mechanism we show here, as Piezo and basement membrane components are all highly conserved. In addition to being generalizable to other animals, our basement membrane repair model may apply to other epithelial tissues, as most epithelial stem cells contact basement membranes.

Limitations of the Study

It remains unclear where Piezo detects basement membrane stiffness. Piezo is expressed in Dl+ intestinal stem cells, but its function there has not been demonstrated as we have not had success with Piezo conditional knockdown (i.e. ubiquitous knockdown does not reproduce the mutant phenotype). Further, although it is likely that Piezo+ stem cells divide to give rise to daughter cells that are matrix menders expressing Col4, this has not yet been demonstrated and other mechanisms of activating Col4 expression have not been ruled out. Nevertheless, our data demonstrate a two-part repair detection and repair mechanism in the Drosophila gut: (1) mechanical surveillance of basement membrane stiffness by Piezo, which activates (2) Col4-based repair mediated by local epithelial cells.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Andrea Page-McCaw (andrea.page-mccaw@vanderbilt.edu)

Materials availability

Any requests for resources and reagents should be directed to the lead contact.

Data and code availability

• Microscopy data reported in this paper will be shared by the lead contact upon request.

• All original code has been archived at Zenodo (https://doi.org/10.5281/zenodo.14659622) and is publicly available as of the date of publication.

• Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

Star Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Fly husbandry

Flies were maintained on cornmeal-molasses food at 25 °C unless otherwise specified. Adult-onset RNAi-based gene knockdown was done under Gal80^ts control, with flies maintained at 18 °C until they eclosed. Female flies were then aged at 18 °C until they were 6-day adults (mated) before moving them to 29 °C to activate Gal4. Each experiment contains data from at least two independent crosses. For the list of fly stocks, please refer to Key Resource Table.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-β gal	Developmental Studies Hybridoma Bank	Cat# 40–1a; RRID:AB_528100
Rabbit anti-Cleaved Caspase 3	Cell Signaling Technology	Cat# 9661; RRID:AB_2341188
Mouse anti-Delta	Developmental Studies Hybridoma Bank	Cat# SA4; RRID:AB_2056641
Rabbit anti-H2AvD	Rockland	Cat# 600-401-914; RRID:AB_828383
Rabbit anti-pH3	Millipore	Cat# 06-570; RRID:AB_310177
Mouse anti-Prospero	Developmental Studies Hybridoma Bank	Cat# MR1A; RRID:AB_528440
Rabbit anti-Pdm1	Yang Xiaohang	n/a
Cy3 goat anti-mouse IgG1	Jackson ImmunoResearch Labs	Cat# 115-165-205; RRID:AB_2338694
Cy3 donkey anti-rabbit IgG	Jackson ImmunoResearch Labs	Cat# 711-165-152; RRID:AB_2307443
Cy5 goat anti-mouse IgG	Jackson ImmunoResearch Labs	Cat# 115-175-205; RRID:AB_2338715
Cy5 donkey anti-rabbit	Jackson ImmunoResearch Labs	Cat# 711-175-152; RRID:AB_2340607
Experimental models: Organisms/strains
D. melanogaster: VkgGFP791: y w; vkgP{GFP791, w+}	Carnegie Protein trap library via Sally Horne-Badovinac	CC00791
UAS-GFPRNAi: w[1118]; P{w[+mC]=UAS-GFP.RNAi.R}142	Bloomington Drosophila Stock Center	9330
VkgGFP, UAS-GFP RNAi /CyO, mCh	this study	n/a
Cg-Gal4: w[1118]; P{w[+mC]=Cg-GAL4.A}2	Bloomington Drosophila Stock Center	7011
UAS-GFP: w[*]; P{w[+mC]=UAS-GFP.S65T}eg[T10]	Bloomington Drosophila Stock Center	1522
Su(H)GBE-lacZ: w[*]; l(2)*[*]/CyO, P{ry[+t7.2]=en1}wg[en11]; P{ry[+t7.2]=Ddc.E(spl)m8-HLH-lacZ.Gbe}3	Bloomington Drosophila Stock Center	83352
Su(H)GBE-Gal4, UAS-mCD8GFP: w[*]; P{w[+mC]=Su(H)GBE-GAL4}2, P{w[+mC]=UAS-mCD8::GFP.L}LL5, P{UAS-mCD8::GFP.L}2/CyO	Bloomington Drosophila Stock Center	83377
UAS-Cg25cRNAi: w1118; P{GD12784}v28369	Vienna Drosophila Resource Center	28369
VkgRNAi (RNAi #1): P{KK111668}VIE-260B	Vienna Drosophila Resource Center	106812
VkgRNAi (RNAi #2): y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.HMC02400}attP2	Bloomington Drosophila Stock Center	50895
c564-Gal4: w[1118]; P{w[+mW.hs]=GawB}c564	Bloomington Drosophila Stock Center	6982
Myo1a-Gal4: w*; P{w+mW.hs=GawB}NP0001 / CyO, P{w-=UAS-lacZ.UW14}UW14	Drosophila Genomics Resource Center	112001
UAS-GTRACE: w[*]; P{w[+mC]=UAS-RedStinger}6, P{w[+mC]=UAS-FLP.Exel}3, P{w[+mC]=Ubi-p63E(FRT.STOP)Stinger}15F2	Bloomington Drosophila Stock Center	28281
PiezoKO: w[*]; PBac{w[+mC]=RB5.WH5}Piezo[KO]	Bloomington Drosophila Stock Center	58770
Piezo KO , VkgGFP 791	this study	n/a
Piezo KO , Cg-Gal4/CyO	this study	n/a
Piezo KO ; UAS-GFP/TM3	this study	n/a
PiezoKI-Gal4: w[*]; TI{GAL4}Piezo[KI]	Bloomington Drosophila Stock Center	78335
UAS-myr-RFP: w[*]; P{w[+mC]=Cg-GAL4.A}2, P{w+mC]=UAS-myr-mRFP}1/CyO; P{w[+mC]=UAS-GFP.RNAi.R}142	Bloomington Drosophila Stock Center	63147
VkgGFP 791 ; tubGal4, tubGal80 ts /S-T	this study	n/a
UAS-PerlecanRNAi (RNAi #1): P{KK110494}VIE-260B	Vienna Drosophila Resource Center	108157
UAS-PerlecanRNAi (RNAi #2): y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.GL01153}attP2	Bloomington Drosophila Stock Center	42783
UAS-NidogenRNAi (RNAi #1): y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.HMJ24142}attP40/CyO	Bloomington Drosophila Stock Center	62902
UAS-NidogenRNAi (RNAi #2): P{KK109625}VIE-260B	Vienna Drosophila Resource Center	102310
UAS-SPARCRNAi (RNAi #1): P{KK100566}VIE-260B	Vienna Drosophila Resource Center	103707
UAS-SPARCRNAi (RNAi #2): y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.HMS02133}attP40	Bloomington Drosophila Stock Center	40885
UAS-Gbp1RNAi: P{KK110262}VIE-260B	Vienna Drosophila Resource Center	108755
UAS-Gbp2RNAi: w[1118]; P{GD4931}v16696	Vienna Drosophila Resource Center	16696
Pxn-Gal4/TM3: y[1] w[67c23]; Mi{GFP[E.3xP3]=ET1}Pxn[MB00459]	Bloomington Drosophila Stock Center	22805
LanB1-Gal4/CyO: w[1118]; PBac{w[+mC]=IT. GAL4}LanB1[0987-G4]/CyO	Bloomington Drosophila Stock Center	63941
Tubulin-Gal4: y[1] w[*]; P{w[+mC]=tubP-GAL4}LL7/TM3, Sb[1] Ser[1]	Bloomington Drosophila Stock Center	5138
Prospero-Gal4: w[*]; P{w[+mC]=GAL4-pros.MG}3	Bloomington Drosophila Stock Center	80572
Btl-Gal4	Greg Beitel (Northwestern University)	FBti0148919
TubGal80ts/CyO; DlGal4/TM6B	Bruce Edgar (University of Utah)	FBal0256627
PiezoMI0429: y[1] w[*]; Mi{y[+mDint2]=MIC}Piezo[MI04294]	Bloomington Drosophila Stock Center	37433
Oligonucleotides
PCR PiezoKO F: GAGTGAGCTGATACCGCTCGC	this study	n/a
PCR PiezoKO R: CAGCGACGGATTCGCGC	this study	n/a
Software and algorithms
ImageJ 1.53v	National Institutes of Health	https://imagej.net/sof tware/fiji/
Graphpad Prism 9.5		https://www.graphpad.com/
Seedwater Segmenter	Mashburn et al. 201242	n/a
Mathematica	Wolfram Inc.	https://www.wolfra m.com/mathematic aL
Original code for muscle cell shape analysis	This study	https://doi.org/10.5 281/zenodo.l46596 22

METHOD DETAILS

Confirming Piezo^KO using PCR

Because Piezo^KO flies have no obvious phenotype, we verified the presence of the deletion by PCR in the starting stock and in recombinants. Primers were designed and optimized against the Piggybac insertion PBac{WH}CG8486-f02291 in Piezo^KO animals to confirm the loss of Piezo using the following primers – F: GAGTGAGCTGATACCGCTCGC, R: CAGCGACGGATTCGCGC. OneTaq2x Master Mix with standard buffer was used. The following PCR conditions were used – Initial Denaturation: 94 °C, 30 sec; denaturation: 94 °C, 30 sec; annealing: 65 °C, 60 sec; extension: 68 °C, 45 sec (30x cycles); final extension: 68 °C, 5 minutes.

DSS feeding

All DSS feeding experiments used mated female flies aged 3–5 d after eclosion before feeding DSS. Flies utilizing the Gal80^ts system were moved to 29 °C for 2 days to ensure the Gal4 was active prior to starting DSS feeding treatments. Flies not utilizing the Gal80^ts system were aged 3–5 days at 25 °C prior to starting DSS feeding treatments. DSS was administered as described in Amcheslavsky et al.²⁵ and Howard et al.¹⁶: a 2.5 cm × 3 cm piece of chromatography paper (Whatman 3030–861, Grade 3 MM CHR) was placed in an empty vial. 500 μl of a 5% sucrose solution with or without 3% 36–50 kDa DSS (dextran sulfate sodium salt colitis grade, MP Biomedicals, CAS number 9011-18-1, 36,000 – 50,000 MW) was added to the paper in the vial. Female flies were anesthetized and carefully placed in the vial. Flies were added to new vials with fresh media each day for 2 days. For recovery experiments, flies were transferred to a vial containing standard cornmeal-molasses food for 2 days. For imaging in 3 dimensions (Z-stacks), flies on standard cornmeal-molasses food were fed 5% sucrose for 4 hours before dissection to reduce gut autofluorescence from the food.

Phloroglucinol feeding

PHG was fed to flies by dissolving it in food prepared without any cornmeal, freshly made with or without 100 μM or 5 mM Phloroglucinol (PHG) (Sigma-Aldrich, CAS number 108-73-6) and solidified in a vial overnight. 20 aged female flies and 5 male flies were added to the vials with or without PHG. Flies were kept at 25 °C for 5 days. Prior to dissection, flies were fed 5% sucrose for 4 hours before dissection to reduce gut autofluorescence from the food.

Bleomycin feeding

Bleomycin (Sigma, 9041-93-4) was dissolved in water at a final concentration of 25 μg/mL and 5% sucrose, or 5% sucrose as a control. 500 μl of bleomycin or control solution were added to a 2.5 cm × 3 cm piece of chromatography paper at the bottom of an empty food vial. Flies were fed bleomycin for 2 days, changing the filter paper and vial daily.

Paraquat feeding

A 10 mM paraquat (Sigma, 75365-73-0) solution was made in water with 5% sucrose, or 5% sucrose alone as the control. Flies were starved in an empty vial for 4 hours before transferring them to a vial with a 2.5 cm × 3 cm piece of chromatography paper soaked with either 500 μl of paraquat or control solution. Flies were fed for 16 hours prior to dissection.

Gut dissections and preparations

Adult females were placed in cold Grace’s media and pinched between the abdomen and thorax using sharp #5 dissecting forceps (Dumont). The abdomen was separated from the thorax, and the integument surrounding the abdomen was peeled away, exposing the gut and Malpighian tubules, being careful not to pinch or stretch the gut. The Grace’s media was removed using a Pasteur pipette, followed by one quick wash with 1x PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄). The guts were fixed with 8% paraformaldehyde (Ted Pella, Paraformaldehyde 16%, Product #18505) in PBS for 60 minutes at room temperature, followed by 3 quick washes in PBS, and 4 × 15 min washes with PBS or PBT.

QUANTIFICATIONS AND STATISTICAL ANALYSIS

Fluorescence intensity measurements

For immunofluorescence experiments, blocking buffer was added to fixed guts for 2 hours at room temperature or overnight at 4 °C then incubated in primary antibody and 0.2 μM SiRActin (Spirochrome Cat# SC001) in PBS or PBT overnight at 4 °C, followed by 4 × 30 min washes with 1x PBS. Secondary antibodies were added for 2 hours at room temperature, followed by 4 × 30 min washes. Guts were then mounted in DAPI-containing mounting media (Vectashield, Vector Laboratories, H1200). Antibodies used were mouse anti-β gal (1:50), rabbit anti-Cleaved Caspase 3 (1:200), mouse anti-Delta (1:50), rabbit anti-H2AvD (1:1000), rabbit anti-pH3 (1:1000), mouse anti-Prospero (1:50), and rabbit anti-Pdm1 (1:1000). The following secondary antibodies were used at 1:200 Cy3 goat anti-mouse IgG1, Cy3 donkey anti-rabbit IgG, Cy5 goat anti-mouse IgG, and Cy5 donkey anti-rabbit.

CNA35 staining

Immediately after gut dissections, CNA35 recombinant monomer protein labeled with Alexa488 (1 mg/ml in PBS) was diluted to 1:100 in PBS and added to guts for 2 hours, followed by 1 quick wash in PBS and fixed for 1 hour with 8% PFA.

Light Microscopy

Single optical sections of the posterior midgut (R4) were imaged using a Zeiss Apotome mounted to an Axio Image M2 with a 63x/1.4 oil Plan-Apochromat objective. Z-stacks were acquired using a 40x/1.3 oil Plan-Apochromat objective at a 0.75 μm step size taken through the top half of the gut tube (lumen through the top of the muscle layer). Images were taken using an AxioCam MRm camera (Zeiss), X-Cite 120Q light source (Excelitas Technologies), and AxioVision 4.8 software. FIJI ImageJ (version 1.53v, National Institutes of Health) 16-bit, ZVI files were used for analysis and figures for images.

For Fig 1C,D and Fig S2A–H, the fluorescence intensity of the basement membrane was determined by taking single optical slices en face through the middle of the circumferential muscles to ensure the same plane of the basement membrane was imaged, and the exposures were kept the same. The fluorescence intensity of the basement membrane was measured in ImageJ using the “box” tool to measure the average intensity of 3 representative regions of the gut basement membrane. Similarly, the background intensity was determined by subtracting the average of 3 boxes outside of the gut region. An AVOVA test was performed (GraphPad Prism 9.5).

Fluorescence intensity measurements of gut cells to be compared were acquired at the same exposure from at least 3 experiments done on different days. Maximum intensity projections were created using FIJI ImageJ and the line tool was used to draw through the center of the 3 brightest nuclei in the GFP channel and the fluorescence intensity was measured and averaged. Background fluorescence was subtracted by taking the average intensity of 3 lines within the gut region that did not overlap with GFP nuclei. An unpaired t-test was performed.

Measurement of muscle shape

Using single-slice cross section images of the gut basement membrane, individual muscle cells were identified and segmented using the software Seedwater Segmenter⁴². Custom code in Mathematica (Wolfram Inc., Champaign, IL) was then used to analyze the segmented images and determine the aspect ratio of each muscle cell (a GitHub archive of this original code has been deposited at Zenodo and is publicly available as of the date of publication at https://doi.org/10.5281/zenodo.14659622). Specifically, to quantify the aspect ratio of gut muscle cells, we first fit a 4th-order polynomial to all the segmented cells to estimate a tangent line along the gut’s long axis. We then calculated the two-dimensional moment-of-inertia tensor, J, for each cell relative to the local tangent line, and calculated each cell’s aspect ratio as Sqrt[J11/J22]. This aspect ratio is thus the cell’s height normal to the gut’s long axis divided by the cell’s length along the gut’s long axis.

Quantifying number of cells

GFP+ cells were manually quantified by creating a maximum intensity projection of the Z-stack images of the top half of the gut and confirmed by the presence of a nucleus. The total number of nuclei in the posterior midgut was quantified by creating a maximum intensity projection and setting a threshold to select the DAPI-stained nuclei. Images were “despeckled,” and a watershed was applied to the image. Using the ImageJ “analyze particle” feature, the number of particles larger than 5 μm² was quantified. The areas where nuclei were fused together or unable to be accurately quantified by ImageJ were manually quantified. The total number of GFP+ cells captured in each image was divided by the total number of nuclei in the image and multiplied by 100 to report the #GFP cells/100.

Statistical reporting

All statistical analysis was performed using GraphPad Prism 9.5. ns indicates P > 0.05, * indicates P ≤ 0.05, ** indicates P ≤ 0.01, *** indicates P ≤ 0.001, **** indicates P ≤ 0.0001.

Highlights:

• In the Drosophila midgut, basement membrane is made of collagen IV from the fat body.

• After damage, “matrix mender” enteroblasts express collagen IV necessary for repair.

• Matrix menders are induced by Piezo, expressed in intestinal stem cells.

• Piezo detects loss of basement membrane stiffness to initiate its repair.