Matrix Metalloproteinases (Mmps) degrade glycoproteins and proteoglycans of the extracellular matrix (ECM) or cell surface and are crucial for morphogenesis. Mmps and their inhibitors are expressed during early stages of cardiac development in vertebrates and expression is altered in multiple congenital cardiomyopathies such as cardia bifida. Drosophila genome encodes two copies of Mmps, Mmp1 and Mmp2 whereas in humans up to 25 Mmps have been identified with overlapping functions. We investigated the role of Mmps during embryonic heart development in Drosophila, a process which is morphogenetically similar to early heart tube formation in vertebrates. We demonstrate that the two Mmps in Drosophila have distinct and overlapping roles in cell motility, cell adhesion and cardiac lumenogenesis. We determined that Mmp1 and Mmp2 promote Leading Edge membrane dynamics of cardioblasts during collective migration. Mmp2 is essential for cardiac lumen formation, and mutants generate a cardia bifida phenotype. Mmp1 is required for luminal expansion. Mmp1 and Mmp2 both localise to the basal domains of cardiac cells, however, occupy non-overlapping domains apically. Mmp1 and Mmp2 regulate the proteoglycan composition and size of the apical and basal ECM, yet only Mmp2 is required to restrict ECM assembly to the lumen. Mmp1 negatively regulates the size of the adhesive Cadherin cell surface domain, whereas in a complementary fashion, Mmp2 negatively regulates the size of the Integrin-ECM domain and thereby prescribes the domain to establish and restrict Slit morphogen signalling. Inhibition of Mmp activity through ectopic expression of Tissue Inhibitor of Metalloproteinase in the ectoderm blocks lumen formation. Therefore, Mmp expression and function identifies ECM differentiation and remodelling as a key element for cell polarisation and organogenesis.
Citation: Raza QS, Vanderploeg JL, Jacobs JR (2017) Matrix Metalloproteinases are required for membrane motility and lumenogenesis during Drosophila heart development. PLoS ONE12(2): e0171905. https://doi.org/10.1371/journal.pone.0171905
Editor: Juan-Pablo Labrador, University of Dublin Trinity College, IRELAND
Received: December 6, 2016; Accepted: January 28, 2017; Published: February 13, 2017
Copyright: © 2017 Raza et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by Natural Sciences and Engineering Research Council Discovery 46651 - JRJ. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Matrix Metalloproteinases (Mmps) are zinc dependent proteases which digest components of the extracellular matrix (ECM) and embedded signalling molecules. Mmps enable morphogenesis by modifying cell migration, cell polarization, ECM remodelling and lumenogenesis . Mmps are regulated by signalling pathways such as Wnt/β-catenin and can modulate signalling of growth factors such as Vascular Endothelial Growth Factors (VEGF) [2–4]. Due to genetic redundancy in the mammalian genome, it is challenging to examine Mmp function when genetic compensation can affect mutant analysis [1, 5–7]. In contrast, the Drosophila genome encodes two Mmps, a secreted protease Mmp1 and a glycophosphatidylinositol (GPI) anchored Mmp2 [8, 9]. Drosophila has a single Tissue Inhibitor of Metalloproteinase (Timp), which has been shown experimentally to be a potent inhibitor of vertebrate Mmps, Drosophila Mmp1 and Mmp2, and other extracellular proteases .
An in vitro model of human vasculogenesis reveals that endothelial cells require a membrane associated Mmp (MT1-Mmp) for luminal expansion and formation of vascular guidance tunnels . Branching morphogenesis in organs such as lungs, mammary and submandibular glands requires activity of Mmps for cell motility and lumen formation, mediated by ECM degradation . During tumour invasion, MT1-Mmp, targeted to the invadopodia, promotes metastasis by degrading ECM barriers [13, 14]. Although vertebrate Mmps have received considerable attention, their contributions to embryonic morphogenesis are less characterised because of the barriers to genetic approaches [5, 7, 15, 16]. Using Drosophila embryogenesis as a genetic model, we can study the effect of complete elimination or inhibition of Mmp activity in vivo. Conservation of activity in Drosophila homologues of Mmps has been well established [8, 9]. Intriguingly however, Mmp activity is not required for embryonic viability in Drosophila since single or double mmp mutant embryos hatch and survive until mid or late larval stages . Nevertheless, developmental processes such as motor axon fasciculation during embryogenesis require Mmp activity [18, 19]. Expression data demonstrated that Mmps are upregulated during late stages of embryogenesis in multiple tissues . Vertebrate Mmps and Timps are expressed in the cardiomyocytes during early heart tube assembly [20, 21] and modulate cardiac morphogenetic events such as heart tube formation, directional looping  and differentiation of ostial cells .
In this study, we tested the genetic requirement of Mmps during heart development in Drosophila, a process with striking similarities with early stages of vertebrate cardiogenesis [24, 25]. To form the embryonic heart, cardioblasts (CBs) in the lateral mesoderm align into bilateral rows and collectively migrate towards the dorsal midline. There, the CBs form specific medial adhesions with their contralateral partners and form an apical lumen, which expands as the embryo transitions through larval stages . We demonstrate that Mmp1 and Mmp2 are expressed in the CBs and both influence Collective Cell Migration (CCM) of CBs by regulating lateral adhesions between CBs and the activity of the migratory Leading Edge (LE). After migration, Mmp1 and Mmp2 play complementary roles during ECM remodelling, cell polarization and lumen formation. Mmp2 is essential for forming adhesions with contralateral CBs. Mmp1 is required to limit the domain of E-Cadherin based contralateral adhesions and its activity contributes to luminal expansion. We show that Mmp2 regulates apical localization of guidance signalling molecules Slit and Robo, and ECM receptors Integrin and Dystroglycan (Dg). Composition of the CB apical luminal ECM is modulated by both Mmps, whereas positioning is only dependent on Mmp2. Overall, we demonstrate that Mmps play essential roles promoting CCM, ECM remodelling, cell polarization and lumen formation during Drosophila cardiogenesis.
mmp1 and mmp2 mutants have distinct heart phenotypes
In Drosophila, polarised positioning of molecularly distinct ECMs to the basal and luminal domains is essential for the formation of the heart [27–29]. Since Mmps contribute to ECM degradation and remodeling in other systems, we reasoned that Mmps may contribute to Drosophila heart formation. To address this hypothesis, we examined the phenotypes of loss of function single and double mmp1 and mmp2 mutant hearts with luminal, junctional and nuclear markers and assessed embryonic heart structure (Fig 1). The cardioblasts of stage 16 wildtype embryos were aligned in bilateral rows and migrated to the midline collectively (Fig 1A). At stage 17, contralateral CBs contact at the midline and then reshape to enclose a medial lumen (Fig 1A’). Dystroglycan (Dg), an ECM receptor, localises to the luminal domain, whereas Discs-large (Dlg), an apical polarity scaffold protein, labels the junctions at the apical attachment sites (Fig 1E and 1E’). Although loss of either or both mmp1 and mmp2 resulted in disorganised CB arrangement along the bilateral rows, CBs eventually reached the midline (Fig 1B, 1B’,1C, 1C’, 1D and 1D’). Following migration, mmp1 and mmp2 mutant phenotypes diverge. In mmp1 mutants, a Dg rich lumen forms, but it is reduced in size and is enclosed by extended Dlg marked cell junctions (Fig 1F and 1F’). In mmp2 and mmp1,mmp2 double mutants, a lumen fails to form (Fig 1G, 1G’, 1H and 1H’). Prior to lumen closure, wildtype CBs curve to form a medial pocket. CB somas in mmp2 mutants are spherical and Dg localisation is circumferentially distributed, whereas Dlg marked junctions are absent (Fig 1G, 1G’, 1H and 1H’). We scored 20–25 βPS Integrin labelled posterior heart cross-sections and determined that 75% of the embryos mutant for either mmp2 or both mmp1,mmp2 were rounded and lacked contralateral contacts, whereas the most prevalent phenotype in mmp1 mutants was the presence of a small apical lumen (Table 1). These results suggest that Mmp2 is required to develop a heart tube, whereas Mmp1 is required for luminal expansion.
(A) In stage 16 wildtype embryos labelled with MEF-2 antibody, CBs form bilateral rows and migrate collectively towards the dorsal midline (arrow). (A’) At stage 17, contralateral CBs align at the midline and form a lumen (arrow). (B-B’) In mmp1Q112/ mmp1Q112 null mutants (from here on referred to as mmp1), organization of CB rows is uneven and delayed migration of CBs is observed (asterisk), however they reach the midline and form a reduced size lumen (arrow). (C-D’) In mmp2w307/ mmp2w307 and mmp1q112, mmp2w307/ mmp1q112, mmp2w307 null mutants (from here on referred to as mmp2 and mmp1,mmp2, respectively), organization of CBs along the bilateral rows is disrupted and CB migration is delayed (asterisk). The CBs in the heart region do not reach the midline by stage 17 (arrow). (E-H) Cross sectional views of stage 17 hearts labelled with α-Mef-2, α-Dg and α-Dlg are visualised (E-H). (E) In wildtype, Dg localises to the CB luminal domain (arrowhead) and pericardial cells (asterisk) while Dlg localises to the junctional domains (arrow). (F) In mmp1 mutants, a reduced Dg labeled lumen forms (arrowhead) and Dlg labels the enlarged junctional domain (arrow). (G) In mmp2 and mmp1,mmp2 mutants, CBs fail to adhere to contralateral partners (arrowhead). CBs exhibit rounded morphology and Dg localises along an extended apical domain (arrowhead). (E’-H’) Representation of CBs at stage 17 in wildtype and mmp1, mmp2 and mmp1,mmp2 mutants summarising distribution of Dg (green), Dlg (red) and Mef2 (blue). Posterior is to the right in dorsal view images (A-D). Scales—25 μm in A-D’ and 10 μm in E-H’.
Remodelling of apical and basal ECM requires Mmp1 and Mmp2 activity in the CBs
In Drosophila, Mmp2 is required to cleave components of the basement membrane and regulate its polarised deposition [19, 30–33]. To explore Collagen-IV deposition in the developing heart, we analysed time-lapse movies of embryos expressing a CollagenIV-GFP gene trap construct (VkgGFP). In wildtype embryos, Vkg is present at the basal and medially oriented apical surface of the CBs (Fig 2A, S1 Movie). At the apical side of CBs, Vkg is absent from the apical extensions and is restricted to the pre-luminal domain during CCM, and to the lumen thereafter (Fig 2A”). In mmp1 mutants, VkgGFP localises normally and is absent from the adhesive domains (Fig 2B–2B”, S2 Movie). In mmp2 and mmp1,mmp2 double mutants, VkgGFP localisation extended to the entire apical region of the CBs (Fig 2C–2C” and 2D–2D”, S3 and S4 Movies). The junctional (the dorsal-apical domain of CBs which adheres to contralateral CBs), luminal (the apical domain encompassing the lumen), basal and lateral (domain connecting ipsilateral CBs) domains in mmp2 and mmp1,mmp2 mutants significantly accumulated GFP signal relative to wildtype (Fig 2C’,2D’ and 2G). We reasoned that if spatial restriction of ECM around the CBs requires Mmp2, then Mmp2 would not be active in the pre-luminal domain. By employing a Mmp2-GFP construct  we determined that Mmp2 was enriched in the apical outgrowths and basal domains and was absent from the pre-luminal domain (Fig 2E and 2E’). After heart assembly, enriched localisation of Mmp2 was not detected at any particular CB cell domain (Fig 2F and 2F’). Mmp1 was detected in the pre-luminal and luminal domain of CBs during migratory and lumen formation stages, respectively, supporting a model in which Mmp1 excludes Cadherin from this domain (S1 Fig).
(A-D) Dorsal, (A’-D’) enlarged and (A”-D”) cross sectional views of live vkg-GFP embryos at stage 17 are shown. (A-A”) In wildtype embryos, Vkg (Collagen IV) localises to the luminal (arrow) and basal domains (A’ arrowhead). (B-B”) In mmp1 mutants, Vkg localises to the reduced luminal (arrow) and basal domains (arrowhead). (C”,D”) In mmp2 and mmp1,mmp2 mutants, Vkg localises to the entire apical domain (arrow) and basal domains (arrowhead). Ectopic lumens between lateral domains of CBs are observed in mmp2 and mmp1,mmp2 mutants (B’-D’ red arrows). (E,F) Dorsal and (E’,F’) cross sectional views of the heart in wildtype. (E-E’) Mmp2-GFP localises to the leading edge (arrows) and basal domain (arrowheads) of CBs during migratory stages. (F-F’) Mmp2 does not localise to the βPS integrin labeled luminal, basal domains or junctional domains of the CBs at stage 17 (arrows). (G) Mean Vkg-GFP fluorescence intensity in wildtype and mmp mutants are shown for junctional, luminal and basal domains. Asterisks indicate significant differences in fluorescence intensity between wildtype and mutants for each domain, respectively, as determined by one-tailed T test. *—p<0.05, **—p<0.01, ***—p<0.001. (A-D, F’) Embryonic hemocytes are marked with asterisks. Scales—25 μm in A-D, 10 μm in A’-D” and E-F’.
If Mmp2 limits the extent of the luminal ECM, does it also play a role in defining ECM identity? To address this question we examined the distribution of Pericardin (Prc), an ECM component expressed and deposited by the pericardial cells exclusively at the basal side of CBs . In contrast to wildtype, where Prc is excluded from the lumen (Fig 3E and 3E’), we found Prc localisation at the apical ECM as well as the basal ECM in mmp1, mmp2, and mmp1,mmp2 mutants (Fig 3F, 3F’, 3G, 3G’, 3H and 3H’). These results suggest that Mmp1 and Mmp2 collectively regulate the identity of the apical ECM, whereas Mmp2 is required to limit ECM to the pre-luminal and luminal domain.
Wildtype and mmp mutant embryos labelled with α-βPS(A-D’, red), α-Prc (E-H’,green) and α-Mef2 (A-H’, blue) are shown from both (A-H) dorsal and (A’-H’) cross sectional views of the heart. (A-A’) In wildtype, βPS localises to the luminal domain (arrow). (B-B’) In mmp1 mutant embryos, βPS is localised to the reduced lumen (arrow). (C-D’) In mmp2 and mmp1,mmp2 mutant embryos, βPS covers the entire medial apical surface of the CBs (arrow). (E-E’) In wildtype, Pericardin localises exclusively to the basal domain (arrowhead). In mmp1 (F-F’), mmp2 (G-G”) and mmp1,mmp2 (H-H’) mutants, Prc localises to the entire medial surface (arrow) as well as the basal side (arrowhead) of the CBs. Scales—10 μm.
CB polarization requires Mmp2 activity
Apicalisation of Integrin is an early event in CB polarization . Localised Integrin enables CBs to stabilise polarising morphogens such as Slit and its receptor, Robo, which are required for lumen formation [34–37]. However, it remains unclear what signals coordinate early Integrin apicalisation. Since Vkg accumulates ectopically in mmp2 mutants (Fig 2C–2C”, Fig 3G and 3G’), we sought to determine whether Integrin localisation could be informed by Mmp activity. In wildtype hearts, βPS Integrin localises to the luminal domain (Fig 3A and 3A’). In mmp1 mutants, βPS is restricted to the reduced lumen (Fig 3B and 3B’), whereas in mmp2, and mmp1,mmp2 mutants, βPS accumulates at the apical and basal sides of the CBs and some lateral locations (Fig 3C–3D’). These results suggest Mmp2 may restrict Integrin ligand binding to the luminal domain by negative regulation of ECM assembly elsewhere.
Signalling by the Slit morphogen is required for heart lumen formation, perhaps by destabilising Cadherin based CB adhesion [35–37]. If ECM assembly is altered in Mmp mutants, then the resulting expansion of Integrin distribution would extend the domain of Slit and Robo stabilisation. In wildtype stage 17 embryos, Slit and Robo accumulate along the apical domain and the lumen develops as a single, continuous midline tube (Fig 4A, 4E, 4I and 4I’). In mmp1 mutants, Slit and Robo localise predominantly to the luminal domain, however lateral accumulation is occasionally observed (Fig 4B, 4F, 4J and 4J’). More dramatically, in mmp2 and mmp1,mmp2 mutants, Slit and Robo fail to apicalise, but accumulate at the lateral domains (Fig 4C, 4D, 4G, 4H and 4K–4L’). These lateral domains resemble ectopic pockets with luminal composition. Integrin and Vkg are present in the ectopic lumens in mmp2 mutants, confirming the luminal characteristics of these lateral structures (data not shown). Even though Integrin localisation includes the apical domain in mmp mutants, the ability to direct Slit and Robo to the apical domain is lost in mmp2 and mmp1,mmp2 mutants and retained in mmp1 mutants. Collectively these results suggest that targeting of guidance molecules Slit/Robo to the luminal domain, downstream of Integrin function, requires Mmp2 activity.
(A-H) Wildtype and mmp mutant embryos labelled with α -Slit and α -Robo. (I-L) Cross-sectional view of wildtype and mmp mutant embryos labelled with α-Dg and α-Robo. (A) In wildtype, Slit localises to the luminal domain of the stage 17 CBs (arrow). (B) In mmp1 mutants, Slit localises to the luminal domains (arrow), however lateral mislocalisation is also observed (arrowhead). (C,D) In mmp2 and mmp1,mmp2 mutants, Slit is mislocalised, accumulating ectopically at the lateral domains of the CBs (arrowheads). (E, I-I’) In wildtype, Robo localises to the luminal domain (arrows). (F,J-J’) In mmp1 mutants, apical (arrow) and lateral (arrowhead) distribution of Robo is observed. In mmp2 (G,K-K’) and mmp1,mmp2 (H,L-L’) mutants, Robo localises to the lateral domains of CBs (arrowhead), whereas Dg is distributed throughout the plasma membrane (arrow). Scale—10 μm.
In the Drosophila nervous system, Mmp2 dependent cleavage of Faulty Attraction, an ECM protein containing multiple epidermal growth factor (EGF) domains, is required for motor axon targeting . Similarly, Slit, an EGF domain containing protein, can be cleaved in to a C-terminus fragment of unknown function and an active N-terminus fragment [38, 39]. Since Slit normally localises to the luminal domain we sought to establish if Mmp2 processed Slit outside the lumen. However, subsequent to ubiquitous transgene expression in embryos, no differences in full-Slit or C-Slit levels in Mmp2 overexpressing embryos were seen relative to control (data not shown). If Slit processing by Mmp2 is unlikely, other mechanisms account for Slit mislocalisation in mmp2 mutants. One possibility is that lateral accumulation of Slit and Robo in mmp2 and mmp1,mmp2 mutants is due to perturbed Integrin dependent targeting and ectopic stabilisation.
Mmp1 and Mmp2 are required for the motility of CB Leading Edge during collective cell migration
CBs collectively migrate towards their final destination—the dorsal midline. Since Mmps regulate cell migration in multiple vertebrate and invertebrate models [40–42], we assessed whether either Mmp contributes to CCM of CBs. To observe CCM, we analysed time-lapse movies of heart development in embryos expressing a nuclear marker, tail-up-GFP (tupGFP), and a fluorescent actin binding protein, Moesin-mCherry (MoeRFP) (Fig 5). This method allowed us to quantify aspects of CCM such as migration velocity, filopodial and lamellopodial activity . We noticed a significant reduction in migration velocity in both mmp2 and mmp1,mmp2 mutant CBs, and a non-significant reduction was observed in mmp1 mutants (Fig 6A, S2 Table). These results strongly suggest that ECM modification is required for the normal migration of CBs. To better understand why migration speed declined, we quantified the number of filopodia and lamellopodia at the Leading Edge (LE). As CBs approach the midline, the number of Moesin-labelled filopodia and lamellopodia increases at the LE (Fig 5A and 5A’, S2A Fig, S5 Movie). These protrusions mediate the formation of apical adhesions with the contralateral CBs upon medial contact (Fig 5A”) . In mmp1, mmp2, and mmp1,mmp2 mutants, filopodial and lamellopodial activities were significantly reduced (Fig 5B, 5B’,5C, 5C’, 5D and 5D’, Figs 6B and 5C, S6–S8 Movies, S2 Table). Although both Mmp1 and Mmp2 activity were required to promote LE activity and outgrowth formation during CCM stages of heart development, only Mmp2 was required for timely migration of the CBs. Additionally, the negative correlation between distance to the midline and number of filopodial extensions was reduced in mmp1, mmp2 and mmp1,mmp2 mutants compared to wildtype (S2B–S2D Fig, S1 Table). During early stages of migration, CBs maintain direct contact with Amnioserosa cells and follow behind the ectoderm LE as it migrates towards the dorsal midline . However, following ectoderm dorsal closure, CBs initiate autonomous movement to complete their migration to the midline. We used complimentary approaches to test whether Mmp2 is autonomously required in the CBs for CCM and lumen formation. First, we attempted to rescue mmp2 mutants by expressing a UAS-Mmp2 transgene using a mef2-GAL4 construct that drives the expression of GAL4 in heart and body wall muscle cells, but not in the Amnioserosa or the Ectoderm which are immediately ventral and dorsal to the heart tissue. Migration velocity, filopodial, lamellopodial activity of CBs and lumen formation (data not shown) were rescued (Fig 5E–5E”, Fig 6A–6C). The negative correlation between distance to the midline and number of filopodia per segment however was not restored (S2F Fig, S1 Table). Second, we reduced Mmp2 levels in the CBs by driving a UAS-mmp2-RNAi construct under the control of mef2-GAL4 driver. Recapitulating the phenotype of mmp2 mutants, CCM of CBs was delayed and frequent gaps (data not shown) and clumps of CBs were also observed in MMP2 reduced embryos (S3B Fig, S9 Movie). Reduction of filopodial and lamellopodial activities were also apparent and migration velocity was reduced (Fig 6A–6C). Formation of intermittent lumens was observed in Mmp2 reduced embryos (S3F Fig). Altogether these results suggest that CBs autonomously regulate CCM and lumenogenesis primarily via Mmp2 activity.
Live embryos expressing tup-GFP (to visualise CB nuclei) and UAS-Moesin-mCherry under the control of mef2-GAL4 driver are visualised. Thin cytoplasmic protrusions were identified as filopodia, whereas sheet-like extensions were categorised as lamellopodia. (A-A”) In wildtype embryos, CBs extend finger-like filopodia (arrow) and sheet-like lamellopodia (arrowhead) towards their contralateral partner cells. CBs make contact at the dorsal midline (arrowhead). (B-B”) In mmp1 mutants, filopodial and lamellopodial activity at the leading edge is reduced (arrow) and CBs form short outgrowths towards their contralateral partners (arrowhead). (C-D”) In mmp2 and mmp1,mmp2 mutants delayed migration is observed (asterisk). CBs extend reduced numbers of filopodia and lamellopodia (C’,D’ arrow). CBs appear rounded without apical extensions towards the midline (C”,D” arrowheads). (E-E”) Expression of a mmp2 transgene in the CBs of mmp2 mutants restores formation of filopodia, lamellopodia (arrows) and apical outgrowths (arrowhead). Scale—25 μm in A-E and 10 μm in A’-E”.
Analysis of time-lapse movies of live embryos expressing cardiac markers tup-GFP (nuclear) and mef2>moesin-mCherry. (A) Migration velocity is significantly reduced in mmp2, mmp1,mmp2 and mmp2-RNAi knockdown embryos. Knockdown and control embryos were raised at 29°C to maximise the effect of the GAL4/UAS system. Bars represent mean±s.e.m. Filopodial (B) and lamellopodial (C) activities are significantly reduced in mmp1, mmp2, mmp1,mmp2 mutants and mmp2-RNAi knockdown embryos. Migration velocity (A), filopodial (B) and lamellopodial (C) activities are restored in embryos where MMP2 rescue was performed. For Box plots, center lines show the medians and box limits indicate the 25th and 75th percentiles as determined by R software . Whiskers extend to the 5th and 95th percentile. Width of the boxes is proportional to the square root of the sample size. Notches represent 95% confidence intervals. One-tailed T-test was performed to determine significance of difference. *—P<0.05, **—P<0.01, ***—P<0.001.
Overexpression of Mmp2 in the CBs disrupts ECM, CCM and lumen formation
The observation that migratory and lumen formation phenotypes in mmp1 mutant were relatively less pronounced compared to mmp2 mutants suggests that Mmp2 plays the predominant role in regulating cardiac development. Since removing Mmp2 activity in the CBs results in disruption of CCM, lumen formation and accumulation of ECM around the CBs, we hypothesised that elevating Mmp2 levels in the CBs would increase ECM degradation and decrease the size of the CB lumen. We employed a UAS-Mmp2 construct under the control of heart specific mef2-GAL4. During CCM, the regular CB alignment was disrupted, but the disjointed CB rows were still able to extend filopodia and lamellopodia (Fig 7A and 7A’, S10 Movie). Quantification further demonstrated that migration velocity and lamellopodial activity in the CBs were not affected, however filopodial activity was reduced (Fig 6A–6C). The negative correlation between distance to the midline and number of filopodia per segment was also reduced (S2E Fig, S1 Table). Once migration was complete, the dramatic misalignment of the CB rows persisted in to late embryonic stages (Fig 7B and 7B’). We further assessed whether Integrin localisation was affected in the CBs. In control and Mmp2 overexpressing embryos, Integrin localised to the pre-luminal and luminal domains of the CBs (Fig 7E–7E” and 7F–7F”). At stage 17, in Mmp2 overexpressing embryos however, Integrin was observed at multiple small lateral and midline luminal pockets (Fig 7B and 7B’). Consistent with a role for Mmps in ECM degradation, overexpression of Mmp2 resulted in severe reduction of VkgGFP levels surrounding the CBs relative to control embryos (Fig 7C–7C” 7D–7D”, 7G). This suggests that presence of Vkg in the CB pre-luminal ECM is not essential for Integrin targeting but might be required for expansion of luminal ECM. However, disruption of the ECM during lumen formation fragments the continuous Integrin-ECM complex and generates ectopic luminal pockets.
UAS-mmp2 transgene was expressed in the CBs under the control of mef2-GAL4. (A-A’) Filopodial, lamellopodial activities and outgrowth formation are not affected (arrows). However, alignment of the bilateral row of nuclei is disrupted (arrowheads). (B-B’) Multiple ectopic lumens form at the heart proper region when MMP2 is overexpressed (arrowheads). (C-D”) Embryos expressing vkg-GFP and mef2>moesin-mCherry demonstrate the reduction of Vkg-GFP levels at the CB ECM in mef2>mmp2 embryos relative to control (arrows). In control embryos, ECM localises at the pre-luminal (C,C’ arrowhead), luminal (C”, arrowhead) and basal domains (C-C”, diamonds). After Mmp2 overexpression, ECM is reduced at the pre-luminal (D,D’ arrowhead), luminal (D”, arrowhead) and basal domains (D-D”, diamonds) (C-D’, arrow indicates the dorsal midline). (E-F’) Embryos labelled with α-βPS and α-Mef2 reveal the disorganization of CBs at the heart proper in Mmp2 overexpressing hearts relative to control (arrows). In control embryos, Integrin localises to the pre-luminal (E’ arrowheads) and luminal domains (E,E”, arrowheads). In mef2>mmp2 embryos, Integrin localises to the pre-luminal domain (F’, arrowheads) and ectopic lumens (F,F”, arrowhead). (G) Mean Vkg-GFP fluorescence intensity in control and Mmp2 overexpressing embryos for junctional, luminal and basal domains. Junctional, luminal and basal fluorescence intensity, respectively, differ significantly in Mmp2 overexpressing embryo compared to both controls as determined by one-tailed T test. ***—p<0.001. Scale– 10 μm.
Ectopic expression of TIMP in the ectoderm inhibits lumen formation in CBs
Mmp activity is regulated by Timp which inhibits both Mmp1 and Mmp2 in Drosophila . Timps are expressed in both mammalian and Drosophila hearts during cardiac development [45, 46], suggesting that CBs tightly regulate Mmp activity to control ECM assembly. Surprisingly, overexpressing Timp in the CBs using a mef2-GAL4 driver did not affect CCM or lumen formation (data not shown). Therefore, we expressed Timp ectopically in the tissue closest to the apical extensions of the CBs where Mmp2 accumulates. We used the paired-GAL4 driver to express Timp to higher levels in the ectodermal stripes immediately dorsal to the heart (Fig 8A). We reasoned that since Mmp2 is detected at the apical domain of CBs, expression of TIMP in the dorsally located ectodermal stripes would be sufficient to inhibit local Mmp2 activity. All CBs, located ventral to the paired expressing and the more distant non-paired expressing ectodermal stripes, failed to extend apical protrusions and form a lumen when Timp was ectopically expressed (Fig 8B, 8B’, 8C and 8C’). Next, we assessed whether CBs retained their ability to localise ECM receptor Dg to the appropriate CB domain. In the control embryo, Dg localised to the luminal and basal domains of the CBs (Fig 8D and 8D’). Subsequent to ectopic Timp expression, CBs failed to form a lumen and Dg label was dispersed around the CBs (Fig 8E and 8E’). In addition, misaligned and rounded CBs were also observed, phenocopying mmp2 and mmp1,mmp2 mutants. This phenotype, coupled with the in situ hybridisation data , suggests that Timp modulates Mmp function during heart development, and that expression by the CBs is spatially regulated.
(A) prd-GAL4 expression is reported by UAS-live-Actin-GFP (green) and CBs are expressing Moesin-mCherry under the control of hand promoter. Arrow indicates the CB midline. (B-C’) Cross sectional images were taken through the non-ostial cells under the non-paired expressing ectoderm. (B-B’) In stage 16 control embryos, CBs extend protrusions (arrow) towards the contralateral partner cells and by stage 17 form an enclosed lumen (arrowhead). (C-C’) When TIMP is expressed in paired expressing ectodermal stripes, apical protrusions (arrow) do not form and lumen formation does not occur (arrowhead). (D-D’) In stage 17 control embryos, Dg localises to the apical luminal domain of CBs (arrow). (E-E’) In embryos expressing TIMP ectopically, accumulation of Dg is observed around the CBs and a medial lumen fails to form (arrow). Gaps between contralateral CBs are also observed (asterisk). Scales—10 μm.
Multiple Mmps and Timps are expressed in embryonic vertebrate hearts and are required for key events such as tube formation, cardiac looping and heart septation [20, 21]. Although vertebrate studies highlight the importance of Mmps and Timps, the large number of Mmp and Timp genes complicate analysis of their function. Employing Drosophila as a genetic model, we demonstrate that Mmps are required for CCM, ECM degradation, cell polarization and lumen formation in Drosophila hearts.
Loss of Mmp2 activity results in severe embryonic heart development phenotypes that persist to larval stages. This activity is autonomously required in the CBs. Surprisingly, despite a malformed heart, mmp2 mutant larvae appear healthy and develop through to pupation, whereas mmp1 and mmp1,mmp2 mutant larvae die during late larval stages . This suggests that an effective heart is not required for embryonic and larval development. Peristaltic movements within the larva may be sufficient to circulate the hemolymph around the larval body in the absence of a functional heart.
Mmp1 and Mmp2 have distinct roles during embryonic heart development
Previous studies suggest that Mmp1 and Mmp2 play distinct roles in Drosophila. During fat body disintegration in pupae, Mmp2 is required to cleave components of the basement membrane and thereby disable cell-ECM interaction while Mmp1 disrupts cell-cell adhesions by cleaving cell adhesion molecule, E-Cadherin . Consistent with this, our data indicates that Mmp1 and Mmp2 perform divergent functions during heart development, specifically during CCM and lumen formation. Mmp1 is required to expand lumen size, which is consistent with E-cadherin as a key substrate . Since Mmp1 localises at the pre-luminal and luminal domains, its activity may restrict E-cadherin to less exposed dorsal and ventral attachment sites and remove E-cadherin from the more exposed medial luminal domain. In a complementary manner, Mmp2 localises to the apical outgrowths in CBs and assists in clearing the Collagen-IV containing apical ECM from the LE. The significant reduction of filopodia and lamellopodia at the LE of CBs and accumulation of Vkg around the CBs in mmp2 mutants suggests that Mmp2 activity promotes the formation of invasive protrusions by degrading the ECM barrier at the dorsal apical domain. Enlargement of the Cadherin domain in mmp1 mutants may reduce the concentration of LE motility factors below threshold for filopoda formation.
The contrasting roles for Mmp1 and Mmp2 reflect different substrate specificities and localization during heart development. Mmp1 can degrade Collagen-IV and Fibronectin in vitro, however cannot degrade laminins, Fibrinogen or type I and type II fibrillar collagens . Mmp2 on the other hand is proposed to degrade Collagen-IV as well as laminins . Laminins are required for proper localisation of ECM components around the CBs and are required for heart development [27, 28, 47, 48]. Stability of Laminin is key to lumen formation. In both lanb1 and mmp2 mutants, abnormal accumulation of Collagen-IV is observed around the CBs .
In mammals, GPI anchored MT1-Mmps accumulate at podosomes of normal cells and invadopodia of metastasising cancer cells and are responsible for removing the basement membrane barrier [13, 14, 49]. CBs extend apical outgrowths, morphologically similar to mammalian podosomes, which localise Mmp2. Therefore, Mmp2 may weaken or remove the apical ECM barrier to create the necessary environment for the formation of actin-based membrane extensions (Fig 9). Degradation of the ECM might further release embedded signalling molecules such as Slit and Netrin which are required for formation of filopodia and lamellopodia in the CBs .
(A) In wildtype stage 16 embryos visualised in cross-section, Mmp2 is targeted to the LE where it cleaves components of the ECM, allowing for stabilisation of a Cadherin domain, and releasing embedded morphogens, possibly Slit, which trigger filopodia and lamellopodia formation. Mmp1 is targeted to the preluminal domain, where it excludes assembly of Cadherin-based adherens junctions and partakes in homeostatic ECM turnover. Pericardin localization is limited to the basal side of the CBs. (B) In mmp1 loss-of-function embryos, Pericardin is present at a reduced luminal ECM, suggesting Mmp1 negatively regulates the size of the Cadherin domain and Pericardin deposition. Mmp2 is likely targeted to the LE where it degrades ECM components and releases embedded morphogens. (C) In mmp2 loss of function embryos, the apical identity of both junctional and preluminal domain is disrupted suggesting Mmp2 is essential for Integrin and Slit mediated CB polarization.
How might Mmps promote cell polarization?
An early feature of CCM of CBs is the targeting of Integrin and ECM components to the apical pre-luminal domain . Subsequently, Integrin at the pre-luminal domain stabilises lumen size determinants, Slit and Robo. In wildtype heart tubes, the apical domain is comprised of the luminal sub-domain bracketed by dorsal and ventral junctional domains that seal the heart tube medially. In mmp1 mutants this fundamental polarity is maintained, but the luminal sub-domain is reduced and the junctional domains are extended. In contrast, in mmp2 mutants, apical junctional domains are not established and a modified ECM domain extends over the entire CB apical surface. The mmp2 mutant apical ECM contains ectopic Pericardin, and Slit and Robo do not accumulate; neither is characteristic of a pre-luminal domain. Mmp2 is not required to target Integrin apically, since βPS1 accumulates apically in mutants. Rather, Mmp2 is required to limit the extent of the ECM within this apical domain. Mmp2 is also not required for proper targeting of Integrin in the Drosophila pupal fat body . Both Mmp function and Integrin accumulation at the incipient CB apical surface are required to stabilise (and possibly target) the luminal determinants Robo and Slit . In both the fat body and heart, Integrin localisation is not affected when Mmp2 is overexpressed suggesting that Mmp2 activity regulates Slit/Robo apicalisation downstream of Integrin activity and that Integrin is not a target substrate for Mmp2 in Drosophila. The spatial restriction of Fibroblast Growth Factor signalling activity by Mmps described during branching morphogenesis in Drosophila  may analogously apply to heart lumen formation signals Netrin and Slit.
Mmp2 sculpts the location and molecular composition of the apical CB lumen. In mmp2 mutant embryos, ectopic lumens (described by [51, 52]) form between lateral CBs which localise luminal markers, Integrin, Dg, Slit, Robo and Vkg. This suggests that the complex of luminal determinants can assemble in the absence of Mmp2, and thereby displace Cadherin from lateral pockets. In converse, increased Mmp2 expression results in degradation of Collagen-IV on all cell surface domains. Collagen-IV is not required for formation of luminal space between CBs since lateral luminal pockets containing Integrin and not Collagen-IV are also observed between Mmp2 overexpressing CBs.
Ectopic lumens, containing the canonical luminal determinants Slit and Robo, form when Mmp2 is absent or if Mmp2 is overexpressed. Both perturbations alter the molecular identity of the apical surface, but this does not reveal how pre-luminal Integrin leads to Slit and Robo accumulation. Two possible mechanisms emerge. The Syndecan family of proteoglycans work cooperatively with Integrins in ECM adhesion . Syndecans are substrates of MT1-Mmps [41, 54] and stabilise Slit and Robo in the heart lumen . Therefore, it is possible that Drosophila Sdc is an Mmp2 substrate, and its cleavage would prevent Slit accumulation. Alternatively, Timp may be localised by a component of the pre-luminal ECM and acts locally to inhibit Mmp2 activity.
Heart morphogenesis in vertebrates and Drosophila begins with polarised cell migration, medial cardiac cell fusion and lumen formation. Mmp action plays a key role in all three activities. Mmps regulate the sub-cellular pattern of ECM stabilisation and the local molecular composition of the ECM. This is required to localise membrane responses to guidance signals, including those for LE motility and for lumen specification.
Methods and materials
Drosophila melanogaster strains
mmp12, mmp1Q112, mmp2DfUba1, mmp2w307, UAS-TIMP, UAS-mmp2 and UAS-mmp2-RNAi (NIG-1794-1R-1) lines were obtained from A. Page-McCaw [17–19]. Tail-up-F4-GFP was provided by R. Schulz . UAS-moesin-mCherry was provided by T. Millard . paired-GAL4 was provided by B. Reed . Remaining stocks were obtained through the Bloomington stock center (NIH P40OD018537).
Embryo and larval fixation and staining were adapted from standard protocols . The primary antibodies used were rabbit α-Mef2 (1:5,000 ), rabbit α-Dg (1:150,), chicken α-GFP (1:1,000, (Novus Biologics)), α-Mmp1 (1:10, 1:1 cocktail of 3A6B4 and 3B8D12, Drosophila Studies Hybridoma Bank (DSHB)), α-Prc (EC11),α-Robo (13C9), α-Dlg (4F3), α-Slit (C555.6D) and α-βPS (CF.6G11) (all 1:30, DSHB). Alexa 488, 546, 594, and 647 secondary antibodies were used at a 1:150 dilution (Molecular Probes). Images were acquired using a Leica SP5 confocal microscope. Transverse images are single optical sections, while all other frontal images are projections of four to six sections. All images were processed using ImageJ and assembled with Adobe Photoshop.
Time-lapse live imaging, quantification and statistics
Embryos were prepared according to the hanging drop protocol . Mutants were identified by the absence of GFP expressing balancer. Live imaging was performed using a Leica SP5 confocal microscope.
Quantification of Viking levels was performed by measuring the GFP intensity of a 2 μm2 area at the junctional, luminal or basal domains of CBs by employing ImageJ. All values were normalised to wildtype embryos (vkgGFP/+) (n = 19 embryos).
Quantification of filopodial and lamellipodia activity is as previously described by Raza and Jacobs . One-tailed T-tests were performed to compare migration velocity, filopodial and lamellopodial activity between a specific pair of genotypes, with a p threshold of 0.05. Single cross-sectional images at the heart proper region per embryos were scored blindly and the averaged values are shown in Table 1.
S1 Fig. Mmp1 localization at the dorsal midline during embryogenesis.
Dorsal (A-A’,C-C’) and cross sectional (B-B’,D-D’) view of embryos immunolabelled with α-Mmp1 and α-Dg antibodies are shown. Mmp1 localizes to the pre-luminal (A-A’,B-B’, arrow) and basal (A-A’,B-B’, arrowhead) domain during migratory stages. At stage 17, Mmp1 localizes to the luminal domain of CBs (C-C’,D-D’,arrows). Scale– 10 μm
S2 Fig. Scatter plot of ‘distance of the cardiac segment to the midline’ for each segment and the ‘number of filopodia extended in cardiac segments’.
In wildtype embryos (A), an inverse correlation is observed between distan. The inverse correlation between ‘distance to the midline’ and the ‘number of filopodia of the segment’ is lost in mmp1 (B), mmp2 (C), mmp1,mmp2 (D) and mef2>mmp2 (E) embryos. Expression of Mmp2 transgenes under mef2 control in mmp2 mutants (F) partially restores the inverse correlation between ‘distance to the midline’ and ‘number of filopodia per segment’ relative to respective mutants.
S3 Fig. MMP2 RNA depletion in the CBs disrupts filopodial and lamellopodial activities and lumen formation.
UAS-mmp2-RNAi was expressed under the control of mef2-GAL4 driver. LE of CBs extends multiple filopodia and lamellopodia (A arrow). In MMP2 depleted embryos, filopodial and lamellopodial activity of the CBs is reduced (B arrow). (C-D) CBs do not extend protrusions towards contralateral partners when MMP2 is downregulated (D arrow) compared to control (C arrow). (E-F) Lumen formation does not occur in embryos where MMP2 is downregulated (F arrowhead) compared to control (E arrowhead). Accumulation of actin at the junctional domain is reduced in embryos where MMP2 is downregulated (E,F arrow). Scale– 10 μm.
S1 Movie. Cardiac ECM in wildtype embryo.
Time-lapse of embryos expressing Vkg-GFP construct. Vkg localises to the pre-luminal domain at the apical, and basal side of the CBs. Embryonic hemocytes migrate underneath the heart. All time-lapse movies were filmed over 30 minutes. A z-stack of 20–30 μm with 1 μm intervals was acquired every minute and Z projected. Movies are 3 fps or a time compression of 180 fold. Posterior of the heart is to the right in this and subsequent movies. Scale—25 μm for all movies.
S2 Movie. Cardiac ECM in mmp1 mutant embryo.
Vkg-GFP is present at the apical and basal side of the CBs, however migration appears disrupted and delayed.
S3 Movie. Cardiac ECM in mmp2 mutant embryo.
Apical accumulation of Vkg-GFP is observed. The polarised positioning of Vkg-GFP around the CBs is affected.
S4 Movie. Cardiac ECM in mmp1,mmp2 mutant embryo.
The ECM around the CBs is severely affected. Migration of the CBs is delayed and disrupted.
S5 Movie. CCM of CBs in wildtype embryo.
Time-lapse of embryos expressing a nuclear marker, tup-GFP and an actin binding fluorescent marker, moesin-mCherry. CBs extend multiple filopodial and lamellopodial extensions at the LE and migrate towards the dorsal midline while maintaining adhesion with ipsilateral cells.
S6 Movie. CCM of CBs in mmp1 mutant embryo.
Filopodial and lamellopodial activity of the CB LE is significantly reduced, however, some filopodial and lamellopodial processes are observed. A gap spanning 8 cell diameter is observed in the bottom LE.
S7 Movie. CCM of CBs in mmp2 mutant embryo.
Filopodial and lamellopodial activity of the CB LE is significantly reduced, however, some filopodial and lamellopodial processes are observed. Migration of the CBs is delayed.
S8 Movie. CCM of CBs in mmp1,mmp2 mutant embryo.
Filopodial and lamellopodial activity of the CB LE is significantly reduced, however, some filopodial and lamellopodial processes are observed. Migration of the CBs is delayed. The bilateral row structure of the heart is disrupted and clumping of CBs is noted.
S9 Movie. CCM of CBs in embryos expressing UAS-MMP2-RNAi under the control of mef2-GAL4.
Filopodial and lamellopodial activity of the CB LE is significantly reduced, however, some filopodial and lamellopodial processes are observed. Migration of the CBs is delayed. The bilateral row structure of the heart is disrupted and clumping of CBs is noted.
S10 Movie. CCM of CBs in embryos expressing UAS-MMP2 under the control of mef2-GAL4.
Filopodial and lamellopodial protrusions are observed at the CB LE. The bilateral row structure of the heart is disrupted and clumping of CBs is noted.
S1 Table. Comparison of correlation coefficients of ‘distance of CBs to the midline’ and ‘number of filopodia per heart segment’ in wildtype, mutant and rescued embryos.
Z values were obtained through Fischer r to z transformation. Z values were used to determine the significance of difference (p-value). ‘N’ represents the number of heart segments scored.
We thank A. Page-McCaw, T. Millard, R Schulz, B. Reed, Bloomington Stock Centre and the Vienna Drosophila RNAi Centre for fly stocks. Resources of the Bloomington Drosophila Stock Center and Flybase contributed to this work. We thank C. Hughes for his help with genetics.
- Conceptualization: QR RJ.
- Data curation: QR.
- Formal analysis: QR.
- Funding acquisition: RJ.
- Investigation: QR JV.
- Methodology: QR JV RJ.
- Project administration: QR.
- Resources: QR JV RJ.
- Supervision: RJ.
- Validation: QR JV RJ.
- Visualization: QR.
- Writing – original draft: QR.
- Writing – review & editing: QR JV RJ.
- 1. Sternlicht MD, Werb Z. How matrix metalloproteinases regulate cell behavior. Annual review of cell and developmental biology. 2001;17:463–516. PubMed Central PMCID: PMC2792593. pmid:11687497
- 2. Mill C, George SJ. Wnt signalling in smooth muscle cells and its role in cardiovascular disorders. Cardiovascular research. 2012;95(2):233–40. pmid:22492675
- 3. Lemaitre V, D'Armiento J. Matrix metalloproteinases in development and disease. Birth defects research Part C, Embryo today: reviews. 2006;78(1):1–10.
- 4. Lee S, Jilani SM, Nikolova GV, Carpizo D, Iruela-Arispe ML. Processing of VEGF-A by matrix metalloproteinases regulates bioavailability and vascular patterning in tumors. The Journal of cell biology. 2005;169(4):681–91. PubMed Central PMCID: PMC2171712. pmid:15911882
- 5. Page-McCaw A, Ewald AJ, Werb Z. Matrix metalloproteinases and the regulation of tissue remodelling. Nature reviews Molecular cell biology. 2007;8(3):221–33. PubMed Central PMCID: PMC2760082. pmid:17318226
- 6. Nagase H, Visse R, Murphy G. Structure and function of matrix metalloproteinases and TIMPs. Cardiovascular research. 2006;69(3):562–73. pmid:16405877
- 7. Egeblad M, Werb Z. New functions for the matrix metalloproteinases in cancer progression. Nature reviews Cancer. 2002;2(3):161–74. pmid:11990853
- 8. Llano E, Pendas AM, Aza-Blanc P, Kornberg TB, Lopez-Otin C. Dm1-MMP, a matrix metalloproteinase from Drosophila with a potential role in extracellular matrix remodeling during neural development. The Journal of biological chemistry. 2000;275(46):35978–85. pmid:10964925
- 9. Llano E, Adam G, Pendas AM, Quesada V, Sanchez LM, Santamaria I, et al. Structural and enzymatic characterization of Drosophila Dm2-MMP, a membrane-bound matrix metalloproteinase with tissue-specific expression. The Journal of biological chemistry. 2002;277(26):23321–9. pmid:11967260
- 10. Wei S, Xie Z, Filenova E, Brew K. Drosophila TIMP is a potent inhibitor of MMPs and TACE: similarities in structure and function to TIMP-3. Biochemistry. 2003;42(42):12200–7. pmid:14567681
- 11. Davis GE, Stratman AN, Sacharidou A, Koh W. Molecular basis for endothelial lumen formation and tubulogenesis during vasculogenesis and angiogenic sprouting. International review of cell and molecular biology. 2011;288:101–65. PubMed Central PMCID: PMC3891664. pmid:21482411
- 12. Bonnans C, Chou J, Werb Z. Remodelling the extracellular matrix in development and disease. Nature reviews Molecular cell biology. 2014;15(12):786–801. PubMed Central PMCID: PMC4316204. pmid:25415508
- 13. Jacob A, Prekeris R. The regulation of MMP targeting to invadopodia during cancer metastasis. Frontiers in cell and developmental biology. 2015;3:4. PubMed Central PMCID: PMC4313772. pmid:25699257
- 14. Poincloux R, Lizarraga F, Chavrier P. Matrix invasion by tumour cells: a focus on MT1-MMP trafficking to invadopodia. Journal of cell science. 2009;122(Pt 17):3015–24. pmid:19692588
- 15. Khokha R, Murthy A, Weiss A. Metalloproteinases and their natural inhibitors in inflammation and immunity. Nature reviews Immunology. 2013;13(9):649–65. pmid:23969736
- 16. Overall CM, Lopez-Otin C. Strategies for MMP inhibition in cancer: innovations for the post-trial era. Nature reviews Cancer. 2002;2(9):657–72. pmid:12209155
- 17. Page-McCaw A, Serano J, Sante JM, Rubin GM. Drosophila matrix metalloproteinases are required for tissue remodeling, but not embryonic development. Developmental cell. 2003;4(1):95–106. pmid:12530966
- 18. Miller CM, Page-McCaw A, Broihier HT. Matrix metalloproteinases promote motor axon fasciculation in the Drosophila embryo. Development. 2008;135(1):95–109. pmid:18045838
- 19. Miller CM, Liu N, Page-McCaw A, Broihier HT. Drosophila MMP2 regulates the matrix molecule faulty attraction (Frac) to promote motor axon targeting in Drosophila. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2011;31(14):5335–47. PubMed Central PMCID: PMC3486643.
- 20. Cai DH, Vollberg TM Sr., Hahn-Dantona E, Quigley JP, Brauer PR. MMP-2 expression during early avian cardiac and neural crest morphogenesis. The Anatomical record. 2000;259(2):168–79. pmid:10820319
- 21. Cantemir V, Cai DH, Reedy MV, Brauer PR. Tissue inhibitor of metalloproteinase-2 (TIMP-2) expression during cardiac neural crest cell migration and its role in proMMP-2 activation. Developmental dynamics: an official publication of the American Association of Anatomists. 2004;231(4):709–19.
- 22. Linask KK, Han M, Cai DH, Brauer PR, Maisastry SM. Cardiac morphogenesis: matrix metalloproteinase coordination of cellular mechanisms underlying heart tube formation and directionality of looping. Developmental dynamics: an official publication of the American Association of Anatomists. 2005;233(3):739–53.
- 23. Alexander SM, Jackson KJ, Bushnell KM, McGuire PG. Spatial and temporal expression of the 72-kDa type IV collagenase (MMP-2) correlates with development and differentiation of valves in the embryonic avian heart. Developmental dynamics: an official publication of the American Association of Anatomists. 1997;209(3):261–8.
- 24. Bodmer R, Venkatesh TV. Heart development in Drosophila and vertebrates: conservation of molecular mechanisms. Developmental genetics. 1998;22(3):181–6. pmid:9621426
- 25. Bier E, Bodmer R. Drosophila, an emerging model for cardiac disease. Gene. 2004;342(1):1–11. pmid:15527959
- 26. Tao Y, Schulz RA. Heart development in Drosophila. Seminars in cell & developmental biology. 2007;18(1):3–15.
- 27. Yarnitzky T, Volk T. Laminin is required for heart, somatic muscles, and gut development in the Drosophila embryo. Developmental biology. 1995;169(2):609–18. pmid:7781902
- 28. Haag TA, Haag NP, Lekven AC, Hartenstein V. The role of cell adhesion molecules in Drosophila heart morphogenesis: faint sausage, shotgun/DE-cadherin, and laminin A are required for discrete stages in heart development. Developmental biology. 1999;208(1):56–69. pmid:10075841
- 29. Chartier A, Zaffran S, Astier M, Semeriva M, Gratecos D. Pericardin, a Drosophila type IV collagen-like protein is involved in the morphogenesis and maintenance of the heart epithelium during dorsal ectoderm closure. Development. 2002;129(13):3241–53. pmid:12070098
- 30. Jia Q, Liu Y, Liu H, Li S. Mmp1 and Mmp2 cooperatively induce Drosophila fat body cell dissociation with distinct roles. Scientific reports. 2014;4:7535. PubMed Central PMCID: PMC4269897. pmid:25520167
- 31. Deady LD, Shen W, Mosure SA, Spradling AC, Sun J. Matrix metalloproteinase 2 is required for ovulation and corpus luteum formation in Drosophila. PLoS genetics. 2015;11(2):e1004989. PubMed Central PMCID: PMC4335033. pmid:25695427
- 32. Stevens LJ, Page-McCaw A. A secreted MMP is required for reepithelialization during wound healing. Mol Biol Cell. 2012;23(6):1068–79. PubMed Central PMCID: PMC3302734. pmid:22262460
- 33. Glasheen BM, Robbins RM, Piette C, Beitel GJ, Page-McCaw A. A matrix metalloproteinase mediates airway remodeling in Drosophila. Developmental biology. 2010;344(2):772–83. PubMed Central PMCID: PMC2914218. pmid:20513443
- 34. Vanderploeg J, Vazquez Paz LL, MacMullin A, Jacobs JR. Integrins are required for cardioblast polarisation in Drosophila. BMC developmental biology. 2012;12:8. PubMed Central PMCID: PMC3305622. pmid:22353787
- 35. Medioni C, Astier M, Zmojdzian M, Jagla K, Semeriva M. Genetic control of cell morphogenesis during Drosophila melanogaster cardiac tube formation. The Journal of cell biology. 2008;182(2):249–61. PubMed Central PMCID: PMC2483531. pmid:18663140
- 36. Santiago-Martinez E, Soplop NH, Patel R, Kramer SG. Repulsion by Slit and Roundabout prevents Shotgun/E-cadherin-mediated cell adhesion during Drosophila heart tube lumen formation. The Journal of cell biology. 2008;182(2):241–8. PubMed Central PMCID: PMC2483515. pmid:18663139
- 37. MacMullin A, Jacobs JR. Slit coordinates cardiac morphogenesis in Drosophila. Developmental biology. 2006;293(1):154–64. pmid:16516189
- 38. Nguyen Ba-Charvet KT, Brose K, Ma L, Wang KH, Marillat V, Sotelo C, et al. Diversity and specificity of actions of Slit2 proteolytic fragments in axon guidance. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2001;21(12):4281–9.
- 39. Wang KH, Brose K, Arnott D, Kidd T, Goodman CS, Henzel W, et al. Biochemical purification of a mammalian slit protein as a positive regulator of sensory axon elongation and branching. Cell. 1999;96(6):771–84. pmid:10102266
- 40. Itoh Y. MT1-MMP: a key regulator of cell migration in tissue. IUBMB life. 2006;58(10):589–96. pmid:17050376
- 41. Endo K, Takino T, Miyamori H, Kinsen H, Yoshizaki T, Furukawa M, et al. Cleavage of syndecan-1 by membrane type matrix metalloproteinase-1 stimulates cell migration. The Journal of biological chemistry. 2003;278(42):40764–70. pmid:12904296
- 42. Kajita M, Itoh Y, Chiba T, Mori H, Okada A, Kinoh H, et al. Membrane-type 1 matrix metalloproteinase cleaves CD44 and promotes cell migration. The Journal of cell biology. 2001;153(5):893–904. PubMed Central PMCID: PMC2174329. pmid:11381077
- 43. Raza Q, Jacobs JR. Guidance signalling regulates Leading Edge behaviour during collective cell migration of cardiac cells in Drosophila. Developmental biology. 2016.
- 44. Haack T, Schneider M, Schwendele B, Renault AD. Drosophila heart cell movement to the midline occurs through both cell autonomous migration and dorsal closure. Developmental biology. 2014;396(2):169–82. pmid:25224224
- 45. Godenschwege TA, Pohar N, Buchner S, Buchner E. Inflated wings, tissue autolysis and early death in tissue inhibitor of metalloproteinases mutants of Drosophila. European journal of cell biology. 2000;79(7):495–501. pmid:10961449
- 46. Brauer PR, Cai DH. Expression of tissue inhibitor of metalloproteinases (TIMPs) during early cardiac development. Mechanisms of development. 2002;113(2):175–9. pmid:11960708
- 47. Urbano JM, Torgler CN, Molnar C, Tepass U, Lopez-Varea A, Brown NH, et al. Drosophila laminins act as key regulators of basement membrane assembly and morphogenesis. Development. 2009;136(24):4165–76. PubMed Central PMCID: PMC2781052. pmid:19906841
- 48. Hollfelder D, Frasch M, Reim I. Distinct functions of the laminin beta LN domain and collagen IV during cardiac extracellular matrix formation and stabilization of alary muscle attachments revealed by EMS mutagenesis in Drosophila. BMC developmental biology. 2014;14:26. PubMed Central PMCID: PMC4068974. pmid:24935095
- 49. Watanabe A, Hoshino D, Koshikawa N, Seiki M, Suzuki T, Ichikawa K. Critical role of transient activity of MT1-MMP for ECM degradation in invadopodia. PLoS computational biology. 2013;9(5):e1003086. PubMed Central PMCID: PMC3667784. pmid:23737743
- 50. Wang Q, Uhlirova M, Bohmann D. Spatial restriction of FGF signaling by a matrix metalloprotease controls branching morphogenesis. Developmental cell. 2010;18(1):157–64. pmid:20152186
- 51. Vanderploeg J, Jacobs R. Talin is required to position and expand the luminal domain of the Drosophila heart tube. Developmental biology. 2015.
- 52. Vogler G, Liu J, Iafe TW, Migh E, Mihaly J, Bodmer R. Cdc42 and formin activity control non-muscle myosin dynamics during Drosophila heart morphogenesis. The Journal of cell biology. 2014;206(7):909–22. PubMed Central PMCID: PMC4178965. pmid:25267295
- 53. Roper JA, Williamson RC, Bass MD. Syndecan and integrin interactomes: large complexes in small spaces. Current opinion in structural biology. 2012;22(5):583–90. PubMed Central PMCID: PMC3712168. pmid:22841476
- 54. Manon-Jensen T, Multhaupt HA, Couchman JR. Mapping of matrix metalloproteinase cleavage sites on syndecan-1 and syndecan-4 ectodomains. The FEBS journal. 2013;280(10):2320–31. pmid:23384311
- 55. Knox J, Moyer K, Yacoub N, Soldaat C, Komosa M, Vassilieva K, et al. Syndecan contributes to heart cell specification and lumen formation during Drosophila cardiogenesis. Developmental biology. 2011;356(2):279–90. pmid:21565181
- 56. Millard TH, Martin P. Dynamic analysis of filopodial interactions during the zippering phase of Drosophila dorsal closure. Development. 2008;135(4):621–6. PubMed Central PMCID: PMC2440488. pmid:18184725
- 57. Yoffe KB, Manoukian AS, Wilder EL, Brand AH, Perrimon N. Evidence for engrailed-independent wingless autoregulation in Drosophila. Developmental biology. 1995;170(2):636–50. pmid:7649390
- 58. Patel NH. Imaging neuronal subsets and other cell types in whole-mount Drosophila embryos and larvae using antibody probes. Methods in cell biology. 1994;44:445–87. pmid:7707967
- 59. Reed BH, McMillan SC, Chaudhary R. The preparation of Drosophila embryos for live-imaging using the hanging drop protocol. Journal of visualized experiments: JoVE. 2009;(25). PubMed Central PMCID: PMC2789762.
- 60. Spitzer M, Wildenhain J, Rappsilber J, Tyers M. BoxPlotR: a web tool for generation of box plots. Nature methods. 2014;11(2):121–2. PubMed Central PMCID: PMC3930876. pmid:24481215