C 2014 Wiley Periodicals, Inc. V

genesis 52:833–848 (2014)

RESEARCH ARTICLE

Mmp25b Facilitates Elongation of Sensory Neurons During Zebrafish Development Bryan D. Crawford,1,2,3* Michelle D. Po,2 Pillai V. Saranyan,1 Daniel Forsberg,2 Richard Schulz,3 and Dave B. Pilgrim2 1

Department of Biology, University of New Brunswick, New Brunswick, Canada

2

Department of Biological Sciences, University of Alberta, Alberta, Canada

3

Department of Pharmacology, University of Alberta, Alberta, Canada

Received 17 May 2014; Revised 23 July 2014; Accepted 25 July 2014

Summary: Matrix metalloproteinases (MMPs) are a large and complex family of zinc-dependent endoproteinases widely recognized for their roles in remodeling the extracellular matrix (ECM) during embryonic development, wound healing, and tissue homeostasis. Their misregulation is central to many pathologies, and they have therefore been the focus of biomedical research for decades. These proteases have also recently emerged as mediators of neural development and synaptic plasticity in vertebrates, however, understanding of the mechanistic basis of these roles and the molecular identities of the MMPs involved remains far from complete. We have identified a zebrafish orthologue of mmp25 (a.k.a. leukolysin; MT6-MMP), a membrane-type, furin-activated MMP associated with leukocytes and invasive carcinomas, but which we find is expressed by a subset of the sensory neurons during normal embryonic development. We detect high levels of Mmp25b expression in the trigeminal, craniofacial, and posterior lateral line ganglia in the hindbrain, and in Rohon-Beard cells in the dorsal neural tube during the first 48 h of embryonic development. Knockdown of Mmp25b expression with morpholino oligonucleotides results in larvae that are uncoordinated and insensitive to touch, and which exhibit defects in the development of sensory neural structures. Using in vivo zymography, we observe that Mmp25b morphant embryos show reduced Type IV collagen degradation in regions of the head traversed by elongating axons emanating from the trigeminal ganglion, suggesting that Mmp25b may play a pivotal role in mediating ECM remodeling in the vicinity of these elongating axons. genesis C 2014 Wiley Periodicals, Inc. 52:833–848, 2014. V Key words: Mmp25; Leukolysin; MT6-MMP; axon elongation; axon pathfinding; ECM remodeling; in vivo zymography

INTRODUCTION Development of the nervous system requires neurons to extend lengthy processes away from their cell bodies to targets situated throughout the embryonic body. The guidance of these cell processes has been the subject of extensive research and a sophisticated understanding of the cell biology of neuronal growth cone guidance has, and is continuing to emerge (reviewed in Charron and Tessier-Lavigne, 2005; Farrar and Gaynor, 2008; Heckman and Plummer, 2013; Kolodkin and Tessier-Lavigne, 2011). However, the mechanisms by which these elongating cell processes traverse the extracellular matrix (ECM)-rich interstitial spaces, modify extracellular macromolecules, remove inhibitory molecules, and penetrate the basement membranes of epithelia they innervate remains a topic of active investigation. The matrix metalloproteinases (MMPs), a family of secreted and membrane-bound zinc-dependent endopeptidases capable of degrading ECM components (reviewed in Murphy and Nagase, 2008; Nagase et al., 2006; PageMcCaw et al., 2007; Sbardella et al., 2012), seem Additional Supporting Information may be found in the online version of this article. * Correspondence to: Bryan D. Crawford, Department of Biology, University of New Brunswick, New Brunswick, Canada, E-mail: [email protected] Contract grant sponsor: Discovery Grant (D.B.P.); Contract grant sponsor: Discovery Grant to (B.D.C.), Contract grant number: BB/J012866/1; Contract grant sponsor: National Sciences and Engineering Research Council (NSERC) (D.B.P. and B.D.C.) Published online 30 July 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/dvg.22803

834

CRAWFORD ET AL.

obvious candidates for components of such mechanisms. Several roles for MMPs in the development and maintenance of the nervous system have emerged (reviewed in Agrawal et al., 2008; Huntley, 2012; McFarlane, 2003). There are two MMPs in the Drosophila genome, one of which appears to encode an alternatively spliced secreted MMP expressed by the developing glial cells of the CNS (Llano et al., 2000) and the second is a membrane-type MMP expressed in the differentiating ommatidia and brain (Llano et al., 2002; Page-McCaw et al., 2003). Misexpression of either of these MMPs results in axonal defasciculation and motor neuron pathfinding defects during embryonic development (Miller et al., 2008). Migration and axon-outgrowth of granular cells during the postnatal maturation of the rat cerebral cortex correlates with MMP-9 expression and is inhibited by blockade of MMP-9 activity using antibodies targeting the active site of the enzyme (Vaillant et al., 2003). Significant changes in development of purkinje cells are also observed in MMP-2 null mice (Verslegers et al., in press). MMP inhibitors block deafferentation-induced neural sprouting after injury in the rat brain (Reeves et al., 2003), and cause axonpathfinding defects in Xenopus (Hehr et al., 2005; Webber et al., 2002), suggesting that MMP activity is required for the axonal outgrowth and ECM-remodeling involved in neural healing and development. The functional roles of MMPs in neural development remain obscure at the molecular level, but some plausible activities include mechanically opening a path for axon elongation, exposing, and/or releasing cryptic guidance cues present in the ECM, processing growth factors or other signaling molecules and/or their receptors, and removing inhibitory molecules such as chondroitin sulfate proteoglycans (CSPGs). MMP-3 and to a lesser extent MMP-2 digest proteoglycans that are inhibitory to axon outgrowth and these MMPs and their inhibitors are dynamically expressed during regeneration of injured brain tissue in rats (Muir et al., 2002). Consistent with this, the neurite promoting potential of basement membranes synthesized by cultured Schwann cells is increased by the activities of MMP-2 and 9 (Ferguson and Muir, 2000), and MMPs 1, 2, and 9 are upregulated (and their endogenous inhibitors down-regulated) in regenerating optic nerve with respect to levels in scar-forming nerves (Ahmed et al., 2005). In situ zymography using fluorescently quenched gelatin reveals gelatinolytic activity at sites of neurite regrowth during optic nerve regeneration that correlates with MMP-2 and MMP-9 expression (Duchossoy et al., 2001). MMP-24 (MT5-MMP) is a membrane type MMP expressed specifically in the brain and testis (Pei, 1999) and is hypothesized to be required for neuronal plasticity (Sekine-Aizawa et al., 2001). It is expressed dynamically in the developing CNS of the mouse, and the

protein is observed to concentrate in growth cones of elongating neurons (Hayashita-Kinoh et al., 2001). Surprisingly, mice deficient for MMP-24 are born without obvious phenotypes, although subtle behavioural differences and defects in neural remodeling after sciatic injury are observed (Komori et al., 2004). This suggests that multiple MMPs may be functioning redundantly during the neural development of the mouse, and/or that the prodigious capacity for self-modification characteristic of the CNS provides a mechanism to compensate for this loss of function. It is increasingly suspected that properly regulated MMP activity in the brain is required for the synaptic remodeling responsible for learning and memory (Brown et al., 2008; Chaillan et al., 2006; Kim et al., 2008; Mizoguchi et al., 2008; Nagy et al., 2006; Paggi et al., 2006; Takacs et al., 2010; Wlodarczyk et al., 2011; reviewed in Huntley, 2012), suggesting that ECM dynamics are of fundamental importance in higher cognitive functions. This presents a particularly exciting new avenue for matrix biology research, but it will first require the identification and functional analysis of MMPs expressed in the nervous system. MMP-25 (MT6-MMP, a.k.a. leukolysin) is the most recently described membrane-type MMP (Pei, 1999) and its functional roles and biochemical properties are only beginning to be established (Radichev et al., 2010; Starr et al., 2012). In adult humans, MMP-25 is expressed in leukocytes and appears to function at three levels during extravasation. It acts directly on basement membrane (Type IV) collagen, as well as indirectly by activating MMP-2 (Nie and Pei, 2003) and by inactivating alpha-1-protease inhibitor (Nie and Pei, 2004). In addition to Type IV collagen, MMP-25 is known to degrade fibronectin, chondroitin- and dermatan-sulfate proteoglycans (Kang et al., 2001). Using a combination of targeted and unbiased proteomic analysis of substrates, Starr et al. (2012) demonstrated that MMP-25 also cleaves several cytokines, vimentin, galectin-1, and a variety of other molecules that may be important in axonal guidance in addition to the inflammatory response, making it a good candidate for the redundancy of MMP activity during neural development discussed above. Here, we report the identification and characterization of a novel zebrafish mmp25 orthologue that is expressed in the developing CNS, particularly in sensory ganglia and primary sensory cells of the spinal cord. Morpholino-mediated inhibition of Mmp25b expression results in larvae that are insensitive to touch and uncoordinated, often with mispatterned components of their sensory nervous system, including reduced or absent lateral line axons and disrupted trigeminal and craniofacial ganglia. Using in vivo zymography, we show that Type IV collagen is broken down along the paths followed by the processes extending from Mmp25b-expressing sensory ganglia, and that this remodeling is reduced in morpholino knock-downs of

MMP25b FACILITATES ELONGATION OF SENSORY NEURONS

Mmp25b. Thus, we conclude that Mmp25b plays a role in facilitating the elongation of neuronal processes from sensory ganglia in the zebrafish. This poorly characterized MT-MMP may also play similar roles, perhaps in partial redundancy with MMP-24, in the CNS of other vertebrates. MATERIALS AND METHODS Fish Husbandry Zebrafish (Danio rerio) embryos were obtained from AB (wild-type) fish (originally obtained from the Zebrafish International Resource Center at the University of Oregon) maintained on a 14 h light, 10 h dark cycle (as described by Westerfield, 1995) in either the University of Alberta Zebrafish Facility or the University of New Brunswick Zebrafish Facility. Embryos were reared at 28.5 C and staged according to Kimmel et al. (1995). Sequence Analysis The fj85c03.y1 cDNA (accession number AW419920) in vector pBK-CMV was obtained from RZPD (Berlin, Germany), on the basis of its partial sequence in dbEST having apparent similarity to known MMPs. Complete sequence for the insert was determined using the BigDye Terminator v3.1 Cycle Sequencing Kit (ABI) with standard T7 and T3 sequencing primers. The insert sequence was 97% identical to the 50 end of the predicted transcript from locus ENSDARG00000010556, which was initially annotated as MMP25-like, and is now annotated as mmp25b (ZDB-GENE-100308-4). The identification as an mmp25 orthologue was verified by phylogenetic reconstruction using the Unweighted Pair Group Method with Arithmetic Mean algorithm (Nie and Kumar, 2000), using Mmp9 as an out-group. The zebrafish genome also contains another mmp25 paralogue, ENSDARG00000077290, annotated as mmp25a (ZDB-GENE-070820-22). We were unable to detect the expression of mmp25a by RT-PCR until after 72 hpf, and do not consider it further in this study. The genomic sequence of mmp25b (ZDB-GENE100308-4) was used in all subsequent analysis (see Supporting Information Fig. 1 for a schematic illustration of the relationship of all primers, morpholinos and in situ probes with respect to the structure of this gene). RT-PCR For expression ontogeny, RNA was collected from 60 embryos of each stage by TRIzol extraction and purification using RiboPure filters (Ambion). cDNA was synthesized from 5 mg total RNA using anchored oligodT primers and enhanced Avian Reverse Transcriptase (Sigma). Mmp25b and Gapdh transcripts were amplified from 1 mg of cDNA template from each stage using exon-junction-spanning primers designed using

835

PerlPrimer (Marshall, 2004) (50 -TCATCTTCTTCATAGG CTCACAG-30 and 50 -ATCCACTCTCATCACCTTCC-30 , amplifying a 125 bp product from the Mmp25b transcript; 50 -AATGTCTCTGTTGTGGATCTG-30 and 50 -TCA TTGTCATACCATGTGACC-30 , amplifying a 235 bp product from the Gapdh transcript). To verify the efficacy of the splice blocking morpholino, primers were designed that will hybridize to sequences in the first and fourth exon of the mmp25b gene (50 -GGCTCCGGTAAAGG ATCAGT-30 and 50 -TCGACTTGAGTGGGTGTCAG-30 , respectively), and thus spanning the second exon that is expected to be deleted. It was expected that these primers would amplify a 360 bp product from the normal transcript and a 227 bp product from the spliceblocked transcript. However, these primers consistently failed to amplify any product from cDNA made from embryos injected with the splice-blocking morpholino [they did, however, generate the predicted 360 bp product from control cDNA (data not shown)]. Nonsense-mediated decay is frequently observed as a result of incorrect splicing (Rodriguez-Pascau et al., 2009) and this is a common occurrence when using splice-blocking morpholinos in zebrafish (Morcos, Gene-Tools, personal communication). To verify this effect we used primers that amplify a 125 bp product from the 30 end of the mmp25b transcript (50 TCATCTTCTTCATAGGCTCACAG 30 and 50 ATCCACTC TCATCACCTTCC 30 ). In Situ Hybridization In situ hybridization was performed on zebrafish embryos essentially as previously described (Jowett, 1999). Briefly, sense and antisense digoxigenin (DIG) labeled riboprobes were produced by in vitro transcription (Roche). Paraformaldehyde fixed embryos were manually dechorionated, treated briefly with Proteinase K, postfixed for 20 min in 4% paraformaldehyde, hybridized overnight at 70 C, washed in 0.23 salt/sodium citrate for 1 h at 70 C and DIG detected using alkaline phosphatase conjugated anti-DIG F’Ab (Roche) followed by 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium chloride staining. Staining was stopped using phosphate buffered saline (PBS) pH 5.5 with 1 mM ethylenediaminetetraacetic acid (EDTA). Embryos were then photographed using a Zeiss Axiophot fitted with a Retiga Qxi CCD camera and Empix Northern Eclipse image capture and analysis software (Empix Imaging). Extended depth-of-focus projections comprising multiple focal planes were produced using ImageJ (Rasband, 2008). Composite images and figures were assembled using Photoshop (Adobe). Immunohistochemistry Embryos were fixed either overnight at 4 C or for 2 h at room temperature in 4% paraformaldehyde in PBS with 2% sucrose. The embryos were blocked for 1 h at

836

CRAWFORD ET AL.

room temperature with PBSDTx (PBS with 0.1% dimethyl sulfoxide (DMSO) and 0.1% Triton X-100, pH 7.4) with 5% bovine serum albumin (BSA). The monoclonal antibody Zn-12, which recognizes an L2/HNK-1 carbohydrate epitope present on the primary sensory neurons of zebrafish (Metcalfe et al., 1990), was used at a concentration of 1:50 in blocking buffer and incubated at 4 C overnight. Embryos were washed 3 3 15 min in PBSDTx at room temperature then incubated in Alexa 488 anti-mouse secondary (Invitrogen) antibody diluted 1:500 in PBSDTx with 5% BSA. After washing 3 3 15 min in PBSDTx at room temperature, the embryos were mounted in 3% methylcellulose on a depression slide and viewed using a Leica SP-2 confocal microscope fitted with a 20 3 0.7 NA water immersion lens and/or a 63 3 1.4 NA oil immersion lens. Negative controls (incubated in blocking buffer without primary antibody, but otherwise processed as above) showed negligible background staining (not shown). Double Labeling: In Situ Hybridization and Immunostaining In situ hybridization was performed as described above. Subsequently, embryos were washed with PBS and sterile water, and permeabilized with acetone at 220 C overnight. Embryos were rehydrated in PBS, blocked with 2% goat serum, and 1% BSA in PBSDTx, then probed with Zn-12 diluted 1:50 in this same solution. Anti-mouse horse radish peroxidase (HRP) secondary (Jackson Immunolabs) was used at 1:5000, followed by HRP detection using 3,3’-diaminobenzidine (DAB). Morpholino Injection Two nonoverlapping translation-blocking morpholinos targeting the 50 UTR of Mmp25b (Mmp25bMO1 50 and Mmp AAACTCATGTGTGTGCTGGTGGAAA-30 25bMO2 50 -ACCTTTATTCACAGATGAGTGTTGC-30 ) as well as a morpholino targeting the splice acceptor site between the first intron and the second exon (SBMO 50 (GeneTools, CAGTCCTACAGGTCAGAGAAATACC-30 ) LLC) were diluted at 500 mg/ml in Danieau buffer [58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO4, 0.6 mM Ca(NO3)2, 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.6] and approximately 3–5 nl was injected into freshly fertilized embryos before the end of the second cell division. When embryos were injected with a mixture of two morpholinos, the total concentration of morpholino was kept at 500 mg/ ml. Control embryos were injected with either a morpholino with no complementarity to any known zebrafish sequences (50 -GACGTTGTCATTTATTTGATTTTCG30 ), and which has no detectable effects on development, or, in the case of the time-lapse and confocal imaging of Islet-1:GFP embryos, with a morpholino targeting the translational start site of Unc45b, (50 -ATCTC-

CAATTCTCCCATCGTCATT-30 ) a gene product necessary for contractility of striated and cardiac muscle (Wohlgemuth et al., 2006), but which was not expected to alter development of Mmp25b-expressing neurons. This latter control was used to ensure that effective dosages of translation blocking morpholinos were being used (100% of Unc45 morphant embryos were completely paralyzed). Efficacy of the splice blocking morpholino was verified by RT-PCR (Supporting Information Fig. 2). Quantitative Analysis of Touch Sensitivity by Video Microscopy Control and Mmp25b morphant embryos were allowed to develop until 30-h postfertilization (hpf), manually dechorionated, and then assayed for sensitivity to touch while being video recorded at 24 framesper-second using a Nikon D90 camera mounted on a Leica MZ6 stereo microscope. Touch sensitivity assays consisted of brushing embryos with an eyelash mounted on the end of a Pasteur pipette. Subsequently, video records were analyzed, frame-by-frame, to determine the speed of the probe and whether the animal responded to the stimulus or not (Low et al., 2010). Finally, these data were analyzed using a binomial general linear model in R (R Development Core Team, 2007; vsn 11.1) to determine whether there were statistically significant differences in the probability that individuals of different treatment groups would respond to a stimulus of a given speed (Fig. 5). Quantitative Analysis of Neuromasts Numbers Embryos were injected with control or Mmp25b morpholinos as described above, and allowed to develop to 24 hpf. Embryos were then manually dechorionated and transferred to fresh embryo medium with 0.1 mM DASPEI [2-(4-dimethyl-aminostyryl)-N-ethyl pyridinium iodide]. At 48 hpf, embryos were transferred to fresh medium, and cooled to 4 C to reduce activity while scoring. Embryos were examined under epifluorescent illumination using a Leica M205FA stereomicroscope, and clusters of two or more DASPEI-stained cells along the lateral line were counted as neuromasts. Images were captured using a Leica DFC camera, and assembled into composites using Adobe Photoshop. Counts of lateral line neuromasts from each treatment group were subjected to two-tailed t-tests and graphed using GraphPad Prism (version 5.0d for Macintosh). Timelapse Microscopy Islet1:GFP transgenic embryos were injected with Mmp25b or control morpholinos as described above, then manually dechorionated at 24 hpf, mounted in methylcellulose, and GFP fluorescence was imaged every 2 min for 8 h at RT using a 20x 0.3 NA lens. These

MMP25b FACILITATES ELONGATION OF SENSORY NEURONS

837

FIG. 1. Alignment of the human MMP-25 amino acid sequence and the putative zebrafish orthologue Mmp25b. Bold face indicates sequence similarity, and underlining indicates identity. Functional regions (from N-to-C) are as follows: signal peptide (purple, with cleavage site indicated by a vertical line), propeptide (gray) with “cysteine-switch” (arrowhead), furin cleavage site red (cleavage site again indicated with a vertical line), catalytic domain (yellow) with zinc-binding motif (turquoise) and “M-turn” (orange), hinge (blue), hemopexin domain (green), and hydrophobic carboxyl-terminal GPI-anchoring sequence (fuchsia) (Based on Pei et al., 1999; Sohail et al., 2008).

images were assembled into time-lapse movies using Northern Eclipse software (Empix Imaging). Individual morphant and control embryos were fixed in 4% paraformaldehyde in PBS at 24 hpf, and high resolution confocal stacks were obtained using a 63x 1.4 NA lens. In Vivo Zymography Embryos developing from Mmp25b or control morpholino injected zygotes were manually dechorionated at 28 hpf, mounted in 3% methyl cellulose, and injected in the trunk with Type IV DQ collagen (Invitrogen) (1 mg/ml in Danieau buffer). Embryos were transferred to fresh embryo medium and allowed to recover for 1 h before being fixed in 4% paraformaldehyde in PBS overnight at 4 C. Fixed embryos were washed in PBS, mounted and imaged by confocal microscopy.

sequence alignment of the human MMP-25 with that of this zebrafish homologue shows strong conservation in regions of known functional importance, including the propeptide with its “cysteine-switch” and furin cleavage site, the catalytic domain with its zinc-binding motif and “M-turn,” the hemopexin domain, and the hydrophobic glycophosphatidylinositol (GPI)-anchoring sequence at the carboxyl terminus (Fig. 1). Given that the duplicated zebrafish furin orthologues are both expressed ubiquitously at the developmental stages examined here (Walker et al., 2006), it is reasonable to predict that this MMP-25 orthologue will likely be activated cell autonomously, presented as a membranetethered protein in lipid rafts on the cell surface, and will likely participate in the dimerization and other protein–protein interactions that depend on its hemopexin domain (Sohail et al., 2008; Sun et al., 2007).

RESULTS We Have Identified a Homologue of mmp25 in the Zebrafish Searches of expressed sequence tag (EST) databases revealed the existence of several uncharacterized zebrafish MMP homologues. One of these, fj85c03.y1, was chosen for further study on the basis of its ambiguous sequence similarity. This cDNA was sequenced and was found to encode a single open reading frame (ORF), encoded by Ensembl Locus ENSDARG00000010556 on linkage group 12. We verified the annotation of this as an mmp25 orthologue by phylogenetic reconstruction of the translated ORF from this genomic sequence with other known MT-MMPs and found 100% bootstrap support for this identification (data not shown). A protein

The Zebrafish mmp25b Homologue is Expressed in the Developing Sensory Nervous System We detect expression of Mmp25b mRNA beginning after early somitogenesis (11 hpf) and persisting until at least 72 h of development by RT-PCR (Fig. 2). No evidence of maternal Mmp25b transcripts is detected in eggs or early embryos [also, no evidence of Mmp25a expression is detected until 72hpf (data not shown)]. Embryos fixed at various stages and processed for in situ hybridization with antisense riboprobes complementary to zebrafish Mmp25b reveals expression concentrated in paired foci lateral to the hindbrain at 14 hpf (Fig. 3b). By late somitogenesis (18 hpf), three paired foci of Mmp25b expression are apparent lateral to the hindbrain (Fig. 3c, arrows), as well as expression in a population of cells in

838

CRAWFORD ET AL.

FIG. 2. RT-PCR detects Mmp25b transcripts accumulating during late somitogenesis and persisting through at least the first three days of development. cDNA synthesized from RNA purified from cleavage (