Laminin and Matrix metalloproteinase 11 regulate Fibronectin levels in the zebrafish myotendinous junction

Remodeling of the extracellular matrix (ECM) regulates cell adhesion as well as signaling between cells and their microenvironment. Despite the importance of tightly regulated ECM remodeling for normal muscle development and function, mechanisms underlying ECM remodeling in vivo remain elusive. One excellent paradigm in which to study ECM remodeling in vivo is morphogenesis of the myotendinous junction (MTJ) during zebrafish skeletal muscle development. During MTJ development, there are dramatic shifts in the primary components comprising the MTJ matrix. One such shift involves the replacement of Fibronectin (Fn)-rich matrix, which is essential for both somite and early muscle development, with laminin-rich matrix essential for normal function of the myotome. Here, we investigate the mechanism underlying this transition. We show that laminin polymerization indirectly promotes Fn downregulation at the MTJ, via a matrix metalloproteinase 11 (Mmp11)-dependent mechanism. Laminin deposition and organization is required for localization of Mmp11 to the MTJ, where Mmp11 is both necessary and sufficient for Fn downregulation in vivo. Furthermore, reduction of residual Mmp11 in laminin mutants promotes a Fn-rich MTJ that partially rescues skeletal muscle architecture. These results identify a mechanism for Fn downregulation at the MTJ, highlight crosstalk between laminin and Fn, and identify a new in vivo function for Mmp11. Taken together, our data demonstrate a novel signaling pathway mediating Fn downregulation. Our data revealing new regulatory mechanisms that guide ECM remodeling during morphogenesis in vivo may inform pathological conditions in which Fn is dysregulated.


Background
The muscle extracellular matrix (ECM) (called the myomatrix) plays a critical role in muscle development, physiology, homeostasis, and disease. Adhesion of muscle fibers to their surrounding myomatrix is required for muscle development and function, as evidenced by congenital muscle diseases -including Duchenne, Becker, and Merosin-deficient muscular dystrophies-which result from mutations that disrupt adhesion of muscle fibers to the myomatrix. The myomatrix is also critical for muscle homeostasis because it bears much of the passive load [1,2], and many clinical symptoms involving range of motion and stiffness are likely derived from changes to the myomatrix [3]. The macromolecular composition of the myomatrix changes during muscle development and regeneration [4][5][6][7][8][9]. These changes in matrix composition define the biochemical and biophysical properties of the myomatrix; yet, the mechanisms that mediate dynamic changes in myomatrix composition during development and regeneration are unknown. In this study, we focused on elucidating mechanisms that underlie regulation of the myomatrix during development because this knowledge could inform efforts to maintain muscle health and prevent muscle diseases.
Remodeling of the ECM at the nascent myotendinous junction (MTJ) during zebrafish skeletal muscle development provides an ideal paradigm with which to investigate dynamic changes in the extracellular milieu. In teleost fishes, the ECM-rich somite boundaries give rise to MTJs that separate segmentally reiterated myotomes [10]. Elegant experiments have recently provided insight into how the ECM protein Fibronectin (Fn) is first polymerized at somite boundaries. During somite formation, integrin alpha5 activation, clustering, and signaling through Rap1b promote Fn polymerization [11]. Fn polymerization is limited to nascent somite boundaries by Eph/Ephrin signaling [12], reviewed in [13]. Subsequent to somitogenesis, a laminin-containing basement membrane is enriched at MTJs and Fn is downregulated [4,9]. The mechanism underlying this downregulation of Fn at MTJs is unknown.
Understanding the regulation of Fn levels in the myomatrix is particularly important because Fn can either positively or negatively affect skeletal muscle depending on the context. Fn is an interstitial matrix protein best known for its functions during branching morphogenesis and cell migration [14][15][16][17][18][19][20][21]. Fn is necessary for muscle development: muscle is disorganized in zebrafish with reduced levels of Fn [22]. Although disrupted mesoderm development in Fn-null mice precludes analysis of how Fn contributes to muscle organization [23], elongating myocytes in wild-type mice express Fn receptors and appear to attach to Fn in the intersegmental ECM [4]. This result suggests that Fn is important for mammalian muscle development as well. Fn is also critical for muscle regeneration: Fn in the stem cell niche is both necessary and sufficient for satellite cell expansion [5]. These data clearly indicate that Fn is necessary for muscle development and regeneration, perhaps by generating a supportive ECM microenvironment. However, Fn expression in these contexts is transient. Sustained Fn expression may contribute to pathology in aging and diseased muscle. Abnormal Fn deposition is a hallmark of fibrosis in cardiac and skeletal muscle [24][25][26]. The replacement of contractile muscle tissue by Fn-rich fibrotic material leads to diminished organ function. Therefore, it is critical to understand the mechanisms underlying both Fn assembly and downregulation during muscle development and regeneration.
We utilized the zebrafish MTJ, where the characteristic downregulation of Fn is documented but remains mechanistically unknown, to identify regulators of Fn levels in vivo. We found that Fn levels are elevated in laminin mutant embryos. This result suggests that laminin deposition is a permissive cue for Fn downregulation. As laminin is not known to be a protease, we hypothesized that laminin regulates a member of the matrix metalloproteinase (MMP) family. Collectively, MMPs are able to cleave all known ECM proteins. MMPs are tightly regulated during development and homeostasis, and changes in MMP expression occur in many diseases [27,28]. We identified Mmp11 (also known as stromelysin-3) as both necessary and sufficient for Fn downregulation in vivo. Significantly, Mmp11 localization to the MTJ is regulated by laminin. Inhibition of the residual Mmp11 activity in laminin mutants increases Fn levels and improves, but does not completely rescue, muscle structure. These results suggest that although decreased downregulation of Fn can improve muscle structure, the specific biochemical and biophysical properties of laminin are critical for normal muscle development. Taken together, our results highlight a novel crosstalk between myomatrix proteins and identify a MMP that regulates Fn levels in vivo.

Zebrafish husbandry, mutant, and transgenic lines
Zebrafish embryos were collected from natural spawnings of adult fish kept at 28.5°C on a 14-h light/10-h dark cycle and staged according to [29]. The following strains were used: AB, Gup/laminin beta1 tg210 , Sly/laminin gamma1 ti263A (a generous gift from the Tuebingen stock center), laminin gamma1 wi390Tg , and Tg(HSP:mmp11-EGFP). All embryos were grown in embryo rearing media (1X ERM), which contained methylene blue to prevent microbial growth. All protocols were conducted in accordance with the University of Maine Institutional Animal Care and Usage Committee's guidelines, which do not require approval for the use of zebrafish 3 days old and younger.

Morpholino and transgenic construct injections
Morpholinos (MOs) were obtained from Gene Tools, LLC (Philomath, OR, USA). All MOs were diluted in sterile water and injected with a MPPI-2 Pressure Injector from ASI. Embryos were injected at the one-cell stage in the yolk with approximately 3 nL of MO. The nucleotide sequences of the MOs are as follows: mmp11a MO is 5′-GAGACCCGCATCCTCCTCCTG TATC-3′, mmp11b MO is 5′-CATGATGACGGA GAGCCGCACGCAG-3' , and standard control MO is 5′-CCTCTTACCTCAGTTACAATTTATA-3′. Mmp11a and mmp11b MOs were mixed, and embryos were injected such that they received a final concentration of approximately 2.5 ng of each MO. For controls, 12 ng of the standard control MO was injected and did not elicit a phenotype (data not shown). The laminin gamma1 (lamc1) MOs used have been previously described [30] and recapitulate the mutant phenotype. The nrk2b MOs were injected as previously described [31]. The Tol2Kit gateway system [32] was used to create the Tg(HSP:mmp11-EGFP) construct. The expected product of this construct was a fusion protein of Mmp11 and EGFP, driven by a heat shock promoter (HSP). Primers used to create the mmp11a middle entry vector were mmp11a F 5′-GGGGACAAGTTTGTA CAAAAAAGCAGGCTAGCACTTCCCACTTCACTTTC AGC-3′ and mmp11a R 5′-TCTTCGGATGTAACATGC CAcACCCAGCTTTCTTGTACAAAGTGGTCCCC-3′. The PCR product was recombined with the donor vector pDONR221 from the Tol2Kit via BP clonase (Invitrogen). This plasmid was then recombined with p5E-hsp70l, p3E-EGFPpA, and pDestTol2pA2 from the Tol2Kit, via LR clonase (Invitrogen) to create the final Tg(HSP:mmp11-EGFP) construct. Linearized pCS2-TP plasmid (kindly provided by Koichi Kawakami) was used as a template to generate capped mRNA encoding Tol2 transposase by in vitro transcription using SP6 polymerase (Message Machine, Ambion). The Tg(HSP:mmp11-EGFP) construct was co-injected with mRNA encoding Tol2 transposase to generate transient transgenic embryos (tTg). tTg embryos were injected at the one-cell stage into the cell and, at the designated developmental time point, heat-shocked at 40°C for 1 h. A stable Tg(HSP:mmp11-EGFP) line was also generated, and embryos collected from natural spawnings of this line were heat-shocked as above and used for experiments.

Fixation/phalloidin staining/immunocytochemistry
Dechorionated embryos were fixed in 4 % paraformaldehyde (PFA) for 4 h at room temperature (RT) or overnight at 4°C, and fix was removed by washing five times for 5 min each in 0.1 % PBS-Tween20.
Alexa Fluor 546 phalloidin (Molecular Probes) staining required permeabilizing fixed embryos for 1.5 h in 2 % PBS-Triton and then incubating in phalloidin (1:20 in 2 % PBS-Triton) for 1-4 h at RT or overnight at 4°C. Embryos were then deyolked and imaged, or subsequent antibody staining was performed.
Anti-Mmp11 antibody staining involved fixing embryos overnight at 4°C in Dent's fixative (80 % MeOH, 20 % DMSO). Embryos were then washed three times for 15 min each in PBS with 0.1 % Tween20, blocked for 4 h at RT, and incubated in an antibody raised against zebrafish Mmp11 antigen (Anaspec, 1:50) overnight at 4°C. Embryos were washed three times for 15 min each and then incubated in goat anti-rabbit secondary (Invitrogen, 1:200) overnight at 4°C. Embryos were washed, deyolked, mounted, and imaged.

Imaging
Brightfield images of live embryos were taken on the Zeiss SteREO Discovery V12 microscope at ×25 or the Zeiss Axio Imager Z1 microscope with the ×5 objective. Live embryos were anesthetized using a non-lethal dose of MS-222. Fixed embryos were deyolked in PBS, sidemounted in 80 % glycerol/20 % PBS, and imaged with the ×20 objective of a Zeiss Axio Imager Z1 microscope with a Zeiss ApoTome attachment and optimized by averaging five frames. For images where fluorescence levels were to be compared, exposure times were kept constant throughout the imaging of that experiment. All image modifications were performed in Adobe Photoshop with Gaussian filtering (0.3 pixel) and unsharp mask (50 %, 1 pixel) prior to being collated in Adobe Illustrator.

Histograms
Images were imported into ImageJ software, and the intensity profiles were plotted. The resulting data were analyzed statistically using Microsoft Office Excel software. We calculated the maximum intensity, and all values were normalized to it. The percentage of maximum intensity was plotted.

Results
Reciprocal expression of laminin and Fn at the fast-twitch muscle MTJ during muscle development During axial skeletal muscle development, somites (blocks of mesodermal cells) transform into myotomes (groups of muscle fibers). Initially short myoblasts ( Fig. 1 A1 inset, arrow) undergo elongation ( Fig. 1 A2-A3 insets, arrows) and attach to MTJs ( Fig. 1 A2-A3), arrowheads) that are derived from initial somite boundaries ( Fig. 1 A1 , arrowheads). Slow-and fast-twitch muscle fibers are spatially segregated during zebrafish development [34]. Slow-twitch muscle fibers are specified medially and then migrate laterally. Slow-twitch muscle fiber migration triggers the elongation of fast-twitch muscle fibers [35]. These morphological changes are accompanied by molecular changes in the ECM: laminin increases and Fn is downregulated [9]. At 18 hpf, in segments where slow-twitch muscle migration has not yet occurred, Fn is present throughout the mediallateral extent of the MTJ (Fig. 1B1, see 3D projection and transverse view (T)). Fn is then downregulated medial to slow-twitch muscle fibers as they migrate laterally. In other words, Fn is downregulated at the MTJ adjacent to fasttwitch fibers after slow-twitch fibers migrate through the fast-twitch domain. This downregulation results in Fn concentrating adjacent to slow-twitch but not fast-twitch muscle fibers at 24 hpf (Fig. 1B2, arrow). Thus, migrating slow-twitch muscle fibers provide a spatial and temporal marker for normal Fn downregulation. By 48 hpf, Fn is detected at relatively low levels throughout the MTJ. Fn concentrates at a subset of slow-twitch muscle fibers called muscle pioneers (MPs) (Fig. 1B3, asterisk in T). The pattern of laminin-111 accumulation at the MTJ adjacent to fasttwitch fibers is reciprocal to that of Fn. Laminin-111 deposition at the MTJ increases through 48 hpf (

Laminin organization is required for Fn downregulation at the MTJ
The reciprocal regulation of Fn and laminin in the fasttwitch muscle domain suggests the hypothesis that laminin organization/deposition/signaling acts as a cue/ checkpoint that is permissive for Fn downregulation (Fig. 2a). If this is the case, absence of laminin at the MTJ would result in increased levels of Fn. Laminins are cross-shaped heterotrimeric proteins comprised of an alpha, beta, and gamma chain. Translated laminin gene products (one alpha, one beta, and one gamma) selfassemble into cross-shaped laminin proteins with the alpha subunit constituting the long arm of the cross and the beta and gamma subunits forming the short arms of the cross [36,37]. Assembled laminin proteins then polymerize with other laminin proteins and other ECM molecules to form specialized basement membranes. Laminin-111 (assembled from laminin alpha1, beta1, and gamma1 gene products) is the developmental isoform of laminin relevant for early muscle development [30,38,39]. We tested the hypothesis that laminin organization/deposition/signaling acts as a cue/checkpoint that is permissive for Fn downregulation by assessing Fn protein levels in laminin beta1 or laminin gamma1 mutant zebrafish (Fig. 2a). As laminin beta1 and laminin gamma1 gene products are both part of the assembled laminin-111 protein, neither of these mutant zebrafish lines will have the laminin-111 isoform or any other isoform of laminin that utilizes laminin beta1 or laminin gamma1 gene products, respectively. Protein immunoblots of whole embryo extracts at 48 hpf showed increased Fn levels in laminin gamma1 mutants compared to siblings (Fig. 2b). Slow-twitch fiber migration is slightly delayed in laminin beta1 and gamma1 mutant embryos but recovers by 26 hpf [40]. Thus, at 32 hpf, slow-twitch fibers can be used as a marker for Fn: Fn  E2 After slow-twitch fiber migration is complete, the MTJ is primarily laminin-111 adjacent to fast-twitch fibers while Fn remains adjacent to slow-twitch fibers. Scale bars are 50 μm should be downregulated medial to slow-twitch fibers. Fn was abnormally abundant throughout the mediallateral extent of the MTJ in laminin beta1 or laminin gamma1 mutants compared to siblings at 32 hpf (compare Fig. 2D to C and Fig. 2F to E). To test whether laminin regulates Fn at the transcriptional level, we assayed fn1a and fn1b mRNA abundance in laminin gamma1 mutants and their siblings at 26 hpf. Fn1a transcript abundance was not changed in laminin gamma1 mutants compared to sibling controls; however, fn1b transcript abundance was approximately fourfold higher in laminin gamma1 mutants compared to controls (Fig. 2g, fn1a: 0.9 +/− 0.2 or 0.9 +/− 0.1 in laminin gamma1 mutants relative to controls normalized to beta-actin or gapdh, respectively; fn1b: 3.6 +/− 0.5 or 4.1 +/− 0.5 in laminin gamma1 mutants relative to controls normalized to beta-actin or gapdh, respectively). The fact that multiple assays-qPCR, Western blotting, and immunocytochemistry-show increased Fn in laminin mutant embryos supports the hypothesis that laminin is required for Fn downregulation. The phenotype of laminin beta1 or laminin gamma1 mutants is fairly severe: embryos are truncated, somites are much narrower in the anteriorposterior axis, and muscle fiber elongation is delayed [41]. We previously showed that Nicotinamide Riboside Kinase 2b (Nrk2b) is required for organized laminin at MTJs during muscle development [31]. The phenotype of nrk2b morphant embryos is milder than that of laminin beta1 or laminin gamma1 mutants: embryos are longer, somites are wider, and neither development nor muscle fiber elongation is delayed. We used nrk2b morphants to ask whether Fn remains at MTJs in a situation where laminin protein is  present, but the organization of laminin-111 at MTJs is disrupted compared to controls (Fig. 2h, i). In medial focal planes of control embryos at 32 hpf, Fn was concentrated at the MTJ adjacent to MPs and the horizontal myoseptum and was not visible at the MTJ adjacent to fast-twitch muscle fibers (Fig. 2j). Fn was also concentrated at the MTJ adjacent to medial MPs in nrk2b morphant embryos at 32 hpf; however, Fn concentration at the MTJ adjacent to fast-twitch muscle fibers was much higher in nrk2b morphant embryos compared to controls at 32 hpf (Fig. 2k). As stated above, the relative abundance of fn1b transcripts was upregulated in in laminin gamma1 mutants. The relative abundance of fn1b is similarly upregulated in nrk2b morphants compared to controls (fn1b: 2.9 +/− 0.6 or 3.3 +/− 1.2 in nrk2b morphants relative to controls normalized to beta-actin or gapdh, respectively). Taken together, the above data indicate that organized laminin, and not just the presence of laminin protein, is necessary for Fn downregulation during MTJ development.
Laminin is necessary to recruit/retain Mmp11 at the MTJ Our data show that increased transcription of fn1b (Fig. 2g) could explain the persistence of Fn protein at MTJs in laminin mutants; however, the low level of expression of fn1a and fn1b mRNA after segmentation is complete [21,42] also suggests the possibility that changes in Fn protein levels at the MTJ may be mediated posttranslationally. Although laminin is necessary for normal Fn downregulation, laminin itself is not a candidate for degrading Fn because laminin has not been shown to have protease activity. Collectively, the members of the MMP family of proteases are able to degrade all proteins found in the ECM [43]. We therefore looked for MMPs that fit two criteria: (1) MMP "X" would need to be expressed at the right time and place to play a role in Fn degradation at the MTJ, and (2)  laminin would need to regulate MMP "X." The rationale for the second criteria is that if laminin is required for Fn downregulation, one potential mechanism would be that laminin regulates MMP "X" expression, localization, or activity. Mmp11 (stromelysin-3) is strongly expressed at the MTJ in zebrafish at 2 days post fertilization [44]. We analyzed Mmp11 expression earlier in muscle development and found that Mmp11 was initially distributed throughout the somitic mesoderm (Fig. 3A1). Mmp11 was more concentrated at the MTJ at 26 hpf (Fig. 3A2) and was robustly localized to the MTJ at 48 hpf (Fig. 3A3). Thus, Mmp11 is expressed in the right time and place to play a role in Fn downregulation. We next asked if laminin deposition/organization is necessary for Mmp11 expression by analyzing Mmp11 protein levels in laminin-deficient embryos. Mmp11 protein levels were decreased in whole embryo homogenates of laminin gamma1 morphant embryos compared to controls at 26 hpf (Fig. 3B1); however, overall Mmp11 protein levels were unchanged in whole embryo homogenates of 26 hpf laminin gamma1 mutants compared to siblings (Fig. 3B2). This discrepancy could be due to the function of maternally provided laminin gamma1 in the mutants. Laminin gamma1 is maternally expressed [30]. The phenotype of morphants is sometimes more severe than mutants because morpholinos can block translation of both maternal and zygotic transcripts [45]. Therefore, laminin gamma1 morphants and mutants would both lack zygotic laminin gamma1 protein. In contrast, mutants would express maternal laminin gamma1 protein but morphants would not. The reduction in Mmp11 levels in laminin gamma1 morphants could be a nonspecific artifact of morpholino knockdown technology or the maternally supplied laminin gamma1 in laminin gamma1 mutants could be sufficient for normal Mmp11 expression through primary muscle development.
As the effect of laminin on Mmp11 was not clear from Western analysis, we performed immunocytochemistry to visualize Mmp11 localization in embryos with disrupted laminin-111 at MTJs. Control, mutant, and morphant images were imaged with the same settings and postprocessed identically with Photoshop. In control embryos at 27 hpf, Mmp11 was strongly concentrated at MTJs (Fig. 3 C1, C2). A histogram of relative pixel intensities across the white rectangle in Fig. 3C2 corroborates that Mmp11 localized to MTJs: tall, tight peaks of bright pixel intensities occurred at MTJs and low valleys between peaks represent dim pixel intensities throughout myotomes. We used nrk2b morphants, where laminin at the MTJ is significantly less organized, to ask whether laminin organization regulates Mmp11 localization. Mmp11 concentrated at the MTJ in nrk2b morphants (Fig. 3 D1, D2, white arrow); however, Mmp11 was also observed throughout the myotome (Fig. 3D2). In the histogram of relative pixel intensities, tall peaks of bright pixel intensities occurred at MTJs (Fig. 3D3), supporting the observation that Mmp11 localized to MTJs in nrk2b morphants. However, pixel intensities between peaks were higher in nrk2b morphants compared to controls (Fig. 3  C3, D3). This reflects the observation that more Mmp11 localized within the myotome in nrk2b morphants than controls. Mmp11 concentration at the MTJ was even more disrupted in laminin gamma1 mutant embryos. Mmp11 only concentrated at MTJs in small patches (Fig. 3E2, arrowhead), and the histogram of relative pixel intensities showed wide, irregular peaks, and high pixel intensities within myotomes (Fig. 3(E3)). These data suggest that laminin adhesion and/or signaling modulate Mmp11 localization.

Mmp11 is necessary and sufficient for Fn downregulation
The above data indicate that laminin is necessary for both Mmp11 localization to the MTJ and Fn downregulation at the MTJ. Therefore, Mmp11 cleavage of Fn could be a mechanism through which laminin promotes downregulation of Fn at the MTJ. However, MMP11 shows only weak protease activity against Fn in vitro [46]. As in vivo functions for Mmp11 are not fully elucidated, Mmp11 could directly or indirectly modulate Fn levels at the MTJ. We used a reverse genetics approach to knockdown Mmp11 protein levels and assay the effect on Fn. The human MMP11 (ST-3) gene has two zebrafish paralogs, mmp11a and mmp11b [44], so we designed translation-blocking morpholinos (MOs) against both. Injection of either MO alone did not appreciably reduce Mmp11 expression on immunoblots of whole embryo homogenates; however, co-injection of both MOs dramatically decreased Mmp11 levels (Fig. 4A). We therefore co-injected mmp11a and mmp11b MOs in all subsequent experiments.
Injection of mmp11 MOs did not overtly affect body, slow-twitch muscle, or fast-twitch muscle morphology compared to control embryos (Fig. 4B, C and data not shown). We used mmp11 morphants to ask whether Mmp11 is required for Fn downregulation. We found Fn protein to be increased in whole embryo homogenates of mmp11 morphants compared to controls (Fig. 4D). The increase in Fn was also apparent with immunostaining. In control embryos, Fn was downregulated medial to migrating slow-twitch muscle fibers, resulting in very little Fn at the MTJ adjacent to fast-twitch muscle fibers (Fig. 4E1,  E1'). In contrast, Fn remained at the MTJ adjacent to slowtwitch and fast-twitch fibers in mmp11 morphants (Fig. 4E2, E2' , arrowhead and arrow, respectively). Thus, multiple types of analysis-western blotting and whole mount immunocytochemistry-show increased Fn levels in mmp11 morphants. The fact that embryonic development is grossly normal in mmp11 morphant embryos suggests that the requirement of Mmp11 for Fn downregulation is specific and not due to confounding requirements for Mmp11 during early development. Taken together, these data suggest that Mmp11 is necessary for normal downregulation of Fn during muscle development.
We next asked whether Mmp11 is sufficient to downregulate Fn. We found that ubiquitous overexpression of Mmp11 during early development disrupted morphogenesis (data not shown). We therefore expressed Mmp11 under control of the heat shock promoter to allow for control over the timing of Mmp11 overexpression. Embryos were injected with the Tg(HSP:mmp11-EGFP) construct and tol2 transposase mRNA at the one-cell stage, grown at normal temperature, and then heat-shocked at 18 hpf. At 5 hours post-heat shock, transient overexpression of Mmp11 resulted in decreased levels of Fn compared to controls on immunoblots from whole embryo lysates (Fig. 4D). In embryos derived from matings of the stable Tg(HSP:mmp11-EGFP) line, Mmp11-EGFP expression was observed throughout the myotome and concentrated at MTJs (Fig. 4E3). Fn immunostaining revealed reduced Fn throughout the MTJ, even in the muscle pioneer domain, compared to controls (Fig. 4E3'). The effects of ectopic Mmp11 on Fn levels were even more pronounced in mmp11 morphants. In mmp11 morphants that express ectopic Mmp11-EGFP, Fn was drastically reduced at the MTJ compared to mmp11 morphant controls (Fig. 4E4,  E4'). These data indicate that Mmp11 overexpression is sufficient to downregulate Fn at the MTJ. Furthermore, as Mmp11 expression rescues the mmp11 morphant phenotype, this experiment provides evidence that mmp11 MOs are specific and suggests that the aberrant upregulation of Fn seen in mmp11 morphants is not an off-target effect of mmp11 MOs. . Note that Fn concentrates adjacent to MPs in both controls and mmp11 morphants (arrowheads in E1' and E2') but only concentrates adjacent to fast-twitch muscle fibers in mmp11 morphants (arrow E2'). E3-E3' mmp11-EGFP transgenic. Note that Fn is downregulated adjacent to MPs in transgenics compared to controls. E4-E4' mmp11 morphant overexpressing mmp11-EGFP (n > 50 embryos). Mmp11-EGFP overexpression reduced Fn levels throughout the MTJ in mmp11 morphants. Scale bars are 50 μm

Laminin functions upstream of Mmp11 in the regulation of Fn at MTJs
The data presented thus far fit a unidirectional model where laminin is necessary for localization of Mmp11 to MTJs, and Mmp11 is then necessary and sufficient for the regulation of Fn. As many regulatory networks involve bidirectional signaling interactions, it is also possible that Mmp11 regulates laminin and laminin, in turn, affects Fn levels at MTJs. To investigate this possibility, we analyzed laminin mRNA and protein localization in  K1). Note that the timing of Mmp11-GFP MTJ localization recapitulates that of native Mmp11 protein, Mmp11-GFP and laminin-111 co-localize at MTJs, and laminin-111 appears normal in Tg(HSP:mmp11-GFP) embryos (n = 5 embryos). Scale bars are 50 μm mmp11 morphants and/or transgenics that overexpress Mmp11. Deposition of laminin-111 protein at MTJs in mmp11 morphants was similar to controls at both 24 and 48 hpf (Fig. 5A-D). Furthermore, antibody staining for beta-dystroglycan (part of the dystrophinglycoprotein complex receptor for laminin-111) in 27 hpf mmp11 morphants compared to controls was also similar (Fig. 5E-F1). This result supports the observation that laminin-111 deposition is unaffected in mmp11 morphants. Next, we analyzed mRNA expression of laminin genes and laminin-111 protein localization in zebrafish stably overexpressing Mmp11-GFP. Expression of relevant laminin gene products for the isoforms of laminin expressed in skeletal muscle development and homeostasis (i.e., laminin alpha1, alpha2, beta1, and gamma1) was not changed in 26 hpf Tg(HSP:mmp11-GFP) embryos compared to heat-shocked AB controls (Fig. 5G). Additionally, laminin-111 deposition and localization at MTJs appeared normal at 26 and 48 hpf in embryos overexpressing Mmp11-GFP (Fig. 5H, I and Fig. 5J, K,  respectively). The expression and localization of Mmp11-GFP recapitulated that of the endogenous protein: Mmp11-GFP began to localize to MTJs at 26 hpf and was robustly localized to MTJs by 48 hpf (Fig. 5 I1, J1, compare to Fig. 3A2, A3). Mmp11-GFP and laminin-111 were found to co-localize at MTJs in Tg(HSP:mmp11-GFP) embryos at 26 and 48 hpf (Fig. 5 I1, J1 insets). This result suggests that laminin-111 deposition was unaffected by Mmp11 overexpression. Therefore, laminin-111 regulates Mmp11 localization to MTJs but not vice versa. This suggests that laminin is upstream of Mmp11 in the regulation of Fn. Although our data do not rule out the possibilities of bidirectional or non-linear signaling, they do suggest that laminin-111 functions upstream of Mmp11 in this context.

Fn levels correlate with phenotypic severity of laminin mutants
We showed that laminin is required for normal Mmp11 expression and Fn levels at the MTJ: in the absence of laminin, Fn levels are increased and Mmp11 levels are decreased. Interestingly, the perdurance of Fn at the MTJ laminin gamma1 mutants is transient. Increased Fn was clear at 32 hpf ( Fig. 6A) but Fn was not observed in mutant MTJs at 48 hpf (Fig. 6C). Injection of mmp11 MOs into laminin gamma1 mutants resulted in the expected persistence of Fn at the MTJ at 32 hpf ( Fig. 6B) but strikingly, Fn persisted in morphant/mutants even at 48 hpf (Fig. 6D). We next tested whether prolonging the transient increase in Fn at MTJs in laminin mutants had any effect on the laminin mutant phenotype by injecting mmp11 MOs into laminin gamma1 mutant zebrafish. Neither the whole embryo morphology nor the slowtwitch muscle of laminin mutants appeared different from mutant morphants at 24 or 32 hpf (Fig. 6E-F). By 72 hpf, slow-twitch muscle fibers had degenerated and could be visualized as spheres instead of cylindrical fibers in laminin mutants (Fig. 6G). Slow-twitch muscle fiber degeneration was drastically reduced in laminin gamma1 mutants injected with mmp11 MOs compared to mutant controls (Fig. 6H). Thus, Mmp11 knockdown can partially compensate for the muscle defects seen in laminin gamma1 mutants, likely via prolonging the presence of Fn at the MTJ. These data suggest that, in the absence of laminin, the amount of Fn has a significant effect on the rate and amount of muscle degeneration. All together, we find that myomatrix crosstalk between laminin and Fn regulates Mmp11-dependent remodeling of zebrafish MTJs during development and that this remodeling can be manipulated to improve muscle tissue structure.

Discussion
The ECM is an essential component of metazoan tissues and exhibits extraordinary complexity and dynamics at a molecular level. The delicate balance between the degradation and deposition of ECM constituents is constantly changing, and its regulation is central to normal development and tissue homeostasis. Thus, it is not surprising that misregulation of ECM remodeling plays a role in aging and disease. ECM components are regulated on many different levels: regulation of transcription (including alternate splicing and miRNA-mediated silencing), translation, protein trafficking, posttranslational modifications, and formation of multiprotein complexes. Most of what is known about ECM remodeling is based on in vitro studies, and a current challenge is to relate these insights to in vivo processes. Towards this end, we investigated mechanisms of Fn downregulation during MTJ development. We show that one initial cue for Fn downregulation is organized deposition of a different ECM protein, laminin. Laminin downregulates fn1b transcription and potentiates Mmp11 localization to the MTJ. Mmp11 is then both necessary and sufficient for Fn downregulation (Fig. 7). Therefore, we have elucidated a novel regulatory pathway that mediates Fn downregulation in vivo.

Mmp11: a critical regulator of Fn levels in vivo
The matrix metalloproteinases (MMPs) cleave all known ECM proteins. MMPs are tightly regulated during development and homeostasis, and changes in MMP expression occur in many diseases [27,28]. For example, several MMPs were originally identified because of their upregulation in cancer cell lines. Stromelysin-3 (MMP11 in mammals or Mmp11 in zebrafish) is over expressed in several cancers, where its abundance correlates with tumor aggressiveness and concomitant mortality rates [47][48][49]. Specifically, increased levels of MMP11 in human and mouse breast cancers are associated with increased metastasis and poor patient prognosis [49][50][51][52][53].
Multiple lines of evidence suggest that MMP11 is a protein with important functions in normal development as well as disease. MMP11 expression is found in remodeling ECM and is required for cell migration during normal morphogenesis in Xenopus [54]. Here, we show that (1) Mmp11 protein localizes to the MTJ during myogenesis in a laminin-dependent manner, (2) Mmp11 is necessary and sufficient for Fn downregulation at the MTJ, and (3) inhibiting Mmp11 activity in laminin mutants increases Fn levels and improves muscle structure. Thus, our data extend and support previous experiments indicating that Mmp11 plays crucial roles in morphogenesis. The data shown here do not resolve whether Mmp11 degrades Fn in vivo directly or indirectly due to technical limitations. MMP11 shows weak protease activity against Fn in vitro [46]. However, it remains largely undetermined whether the ability or inability of MMPs to cleave substrates in vitro corresponds with their in vivo activities. It is possible that MMP11 cleaves Fn in vivo more efficiently, perhaps due to mechanical load on the Fn fibrils, contractile forces, or other interactions that position scissile bonds such that they become accessible to the protease. It is also possible that the effects of MMP11 on Fn are indirect. MMP11 is activated intracellularly by Furin [55] and is capable of cleaving and activating both extracellular and intracellular proteins, such as proMMPs [56]. Therefore, MMP11 may have indirect effects on Fn downregulation by activating other proteases. Recently, it has been shown that Mmp14a is expressed at the zebrafish MTJ at 24 and 48 hpf [57], which makes it a good candidate to be involved in the regulation of Fn. MMP14 is known to degrade Fn [58] and is critical for myoblast differentiation in vitro [59]. In addition, MMP11 is a MMP14 substrate and MMP14 thereby negatively regulates MMP11 [60]. Interactions between MMP11, MMP14, laminin, and Fn will be interesting to identify in future studies. Alternatively, MMP11 could affect Fn levels indirectly via MMP11-dependent regulation of receptors for laminin. MMP11 cleaves the 67kD non-integrin laminin receptor (LR) in vivo [61,62]. It is possible that MMP11 does not directly degrade Fn but rather releases LR-mediated inhibition of some other protease. Regardless, our data clearly show that Mmp11 is necessary and sufficient for Fn downregulation in vivo. It will be interesting in the future to identify molecular mechanisms underlying this ECM remodeling, such as the mechanisms of Mmp11 action and the effect of ectopic Mmp11 in disease states associated with excess Fn (i.e., fibrosis).

Fn levels are dynamically regulated in development and disease
Fn plays important roles in development, repair/regeneration, and the transition to disease states in many organs. Fn is required for mesoderm specification, left-right patterning, trunk elongation, angiogenesis, heart morphogenesis, and migration of myocardial cells [21,[63][64][65][66][67][68][69]. In skeletal muscle, Fn is required for muscle development [22,42,70] and is both necessary and sufficient in the stem cell niche for muscle regeneration [5]. However, upregulation of Fn, while critical for certain developmental processes, is also linked to tumor metastasis and fibrosis. Fn is upregulated in many cancers and promotes formation of the premetastatic niche, which generates a microenvironment more hospitable for tumor cell adhesion and proliferation [71][72][73][74]. Fibrosis, the aberrant deposition of Fn and various collagens, diminishes organ function. Inhibiting Fn deposition not only prevents but can reverse liver fibrosis, possibly via both reducing cell infiltration and not providing a scaffold for collagen deposition [75]. Fibrosis occurs in skeletal muscle during aging, atrophic conditions, and disease. Fn is a serum biomarker for Duchenne muscular dystrophy, suggesting that excess Fn contributes to muscle pathology [76]. Thus, the importance of precise Fn regulation for skeletal muscle development, regeneration, and function is mirrored in other organ systems. While these multifactorial roles for Fn demonstrate the importance of dynamically regulating Fn levels, the molecular mechanisms regulating Fn expression and polymerization are just beginning to be elucidated and exceedingly little is known about how Fn is degraded in vivo. Fn downregulation during development is a normal part of the transition to a laminin-rich basement membrane at the myotendinous junction (MTJ). Our data show that laminin, via subcellular localization of Mmp11 and inhibition of fn1b transcription, is necessary for Fn downregulation. While likely an indirect effect of Mmp11, our data implicate Mmp11 and laminin in a novel network that regulates Fn downregulation in vivo. As Fn is dynamically regulated in other systems, such as its degradation behind migrating myocardial cells [21], it will be interesting to determine if there is a similar regulation of Fn by laminin in other developmental, cell migratory, and/or disease contexts.

Crosstalk between myomatrix proteins and cellular adaptation in development and disease
Our data provide compelling evidence that there is crosstalk between the ECM proteins Fn and laminin during muscle development and that this crosstalk results in cellular adaptations that protect muscle structure when development is disrupted. We previously showed an inverse expression pattern of Fn and laminin during MTJ development: Fn is downregulated and laminin is polymerized adjacent to fast-twitch muscle fibers as they elongate and attach to the MTJ [9]. Here, we show that laminin deposition/organization acts as a "checkpoint" for Fn downregulation. To our knowledge, this is an unrecognized interaction between Fn and laminin and represents a novel paradigm in which laminin is permissive for Fn downregulation. In a developmental context, such a mechanism ensures that an established matrix is continuously present, while simultaneously permitting biochemical and structural remodeling. Crosstalk between ECM molecules Fn and Collagen1 has been shown to regulate MMP expression in vitro. Exogenous Fn caused human fibroblasts to upregulate MMP15 and MMP9, and subsequent addition of Colla-gen1 to the Fn-rich matrix attenuated this response [77].
Thus, there appears to be crosstalk between these ECM molecules as well as between these ECM molecules and enzymes that remodel ECM. It will be an interesting line of future investigation to look at such interactions in vivo as well as to elucidate roles for collagen molecules in normal and pathological remodeling of the myomatrix.
The crosstalk between laminin and Fn presented here is reminiscent of the crosstalk between receptors for ECM proteins that occurs and can be beneficial in muscle diseases. Adhesion of muscle fibers to the laminin-rich basement membrane (BM) is critical for muscle development and homeostasis. Many muscular dystrophies, including Duchenne, Becker, and Congenital muscular dystrophy with integrin alpha7 deficiency, result from mutations that disrupt adhesion of muscle fibers to their BM. There are two main receptor complexes that anchor muscle cells to the BM: the dystrophin-glycoprotein complex (DGC) and integrinbased adhesions. Although both of these receptors have unique functions [78], they can partially compensate for each other in adhesion to laminin. The fact that patients with Duchenne muscular dystrophy (DMD), which affects a member of the DGC, show increased expression of integrin alpha7 [79], highlights the intrinsic crosstalk between cell-ECM adhesion complexes that occurs in diseased muscle.
Cell-ECM adhesion complexes sense multiple types of physiological changes and interface with every major signaling pathway. Thus, cell-ECM adhesion complexes are ideally situated to facilitate appropriate responses to physiological change. One approach towards elucidating the roles of cell-ECM crosstalk in cellular adaptation is to investigate the dynamic regulation of cell-ECM adhesion during embryonic development. We focus on zebrafish MTJ development as a model of in vivo cell-ECM adhesion and cell adaptation because muscle tissue is strikingly adaptable. While the increased Fn at the MTJ of laminin gamma1 mutants did not completely rescue muscle morphogenesis in these mutants, just as the increased integrin alpha7 expression observed in humans with DMD does not prevent muscle atrophy, one key experiment shows that reducing Fn downregulation in laminin mutants did have a positive effect on slowtwitch muscle fibers. Injection of mmp11 MOs into laminin gamma1 mutants increased Fn abundance and decreased slow-twitch muscle degeneration. These data support the hypothesis that reciprocal regulation of ECM proteins plays a role in cell adaptation in adverse conditions (such as the absence of laminin-111 in laminin gamma1 mutants) and suggests that this line of investigation (elucidating mechanisms of cell-ECM adhesion that regulate cell adaptation during development) can potentially be translated into meaningful therapies for certain diseases.

Conclusions
We have addressed the regulatory mechanism of Fibronectin downregulation during MTJ development in vivo. Our data show that normal laminin organization acts as a "checkpoint" for Fn downregulation. Furthermore, laminin signaling modulates both the expression of fn1b and localization of Mmp11 to MTJs. The augmentation of Fn at MTJs in mmp11 morphants was reversed by driving overexpression of Mmp11-GFP, demonstrating that Mmp11 is both necessary and sufficient for Fn downregulation (and the specificity of our morpholinos). Overall, we identify a new mechanism in the myomatrix that regulates the remodeling process during morphogenesis in vivo, and that may be profitably targeted in the many pathological conditions in which Fn is dysregulated, to improve muscle tissue structure. equipment. Additional thanks go to Mark Nilan for the fish care. This work was