Molecular and Cellular Neuroscience 69 (2015) 30–40

Contents lists available at ScienceDirect

Molecular and Cellular Neuroscience journal homepage: www.elsevier.com/locate/ymcne

Rho kinase is required to prevent retinal axons from entering the contralateral optic nerve Paula B. Cechmanek, Carrie L. Hehr, Sarah McFarlane ⁎ Department of Cell Biology and Anatomy, Hotchkiss Brain Institute, University of Calgary, 3330 Hospital Dr., NW, Calgary, AB, Canada

a r t i c l e

i n f o

Article history: Received 12 January 2015 Revised 16 September 2015 Accepted 8 October 2015 Available online 9 October 2015 Keywords: Rock Optic nerve Visual system Growth cone Axon guidance Xenopus

a b s t r a c t To grow out to contact target neurons an axon uses its distal tip, the growth cone, as a sensor of molecular cues that help the axon make appropriate guidance decisions at a series of choice points along the journey. In the developing visual system, the axons of the output cells of the retina, the retinal ganglion cells (RGCs), cross the brain midline at the optic chiasm. Shortly after, they grow past the brain entry point of the optic nerve arising from the contralateral eye, and extend dorso-caudally through the diencephalon towards their optic tectum target. Using the developing visual system of the experimentally amenable model Xenopus laevis, we find that RGC axons are normally prevented from entering the contralateral optic nerve. This mechanism requires the activity of a Rhoassociated kinase, Rock, known to function downstream of a number of receptors that recognize cues that guide axons. Pharmacological inhibition of Rock in an in vivo brain preparation causes mis-entry of many RGC axons into the contralateral optic nerve, and this defect is partially phenocopied by selective disruption of Rock signaling in RGC axons. These data implicate Rock downstream of a molecular mechanism that is critical for RGC axons to be able to ignore a domain, the optic nerve, which they previously found attractive. © 2015 Elsevier Inc. All rights reserved.

1. Introduction A growing axon's distal tip, the growth cone, is directed through a series of guidance choice points to a final target by repellent and attractive guidance cues. The response of a growth cone to any given guidance cue is context dependent, such that a single cue can be attractive to axons in one circumstance and repulsive in another. This is true at the floorplate of the spinal cord, where the axons of dorsal commissural neurons are attracted to the ventral floor plate by the attractive cues Netrin (Ntn) and Sonic Hedgehog (Shh), but are repelled by Shh upon crossing the floor plate (Chedotal, 2011). Similarly, the presence or absence of laminin alters the response of the retinal ganglion cell (RGC) axons to Ntn expressed within the optic nerve head (ONH) (Hopker et al., 1999). RGC axons make a sharp turn into the ONH, repelled by Ntn co-presented with laminin on the vitreal surface of the retina. These axons are subsequently attracted towards the laminin poor, Ntn-rich ONH interior. RGC axons leave the eye, extend through the optic nerve, enter the ventral brain and then cross to the contralateral side of the brain at a ventral midline structure called the optic chiasm. After crossing the ventral midline, RGC axons encounter the environment of the optic ⁎ Corresponding author at: 3330 Hospital Dr., NW, University of Calgary, T2N 4N1 Calgary, AB, Canada. E-mail address: [email protected] (S. McFarlane).

http://dx.doi.org/10.1016/j.mcn.2015.10.001 1044-7431/© 2015 Elsevier Inc. All rights reserved.

nerve entering the brain from the other (contralateral) eye, yet bypass the nerve and continue dorsally through the diencephalon enroute to their major midbrain target, the optic tectum. The molecular mechanisms that ensure that RGC axons from one eye do not grow into the optic nerve of the other eye, which is permissive to RGC axon growth, are not well understood. Rho-associated kinase (Rock) is a downstream effector of several axon guidance cues. Rock is activated downstream of the guidance receptors Plxna1 and Epha4, and downregulated downstream of the Ntn receptor, Deleted in Colorectal Cancer (Gallo, 2006; Hall and Lalli, 2010). Rock signaling is controlled by Rho GTPases, and feeds into effector proteins that include Myosin light chain, Lim kinase-1 and Collapsin response mediator protein-2, which act on the cytoskeleton (Gallo, 2006; Loudon et al., 2006; Schofield and Bernard, 2013). We show that Xenopus laevis RGCs and their growth cones express Rock protein during development of the optic projection. Pharmacological inhibition of Rock activity reveals that Rock is required for RGC axons of X. laevis to extend past the brain entry point (BEP) of the contralateral optic nerve. With Rock inhibition a subset of RGC axons grow aberrantly into the optic nerve and eye on the contralateral side of the embryo. These data are partially phenocopied by overexpression of wildtype (wt)-Rock2, indicating that tightly controlled Rock activity in growth cones functions to prevent a population of RGC axons from exiting the brain and erroneously extending into an area, the optic nerve, which is seen earlier in their trajectory as attractive.

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

2. Materials and methods 2.1. Animals Adult female X. laevis were injected with Human Chorionic Gonadotropin (Intervet) to stimulate egg production for in vitro fertilization. Embryos were reared in 0.1× Marc's Modified Ringers solution (MMR; 0.1 M NaCl, 2 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 5 mM HEPES, pH 7.5) and the temperature was varied between 16 and 23 °C to control the rate of development. Stages were assessed according to guidelines described by Nieuwkoop and Faber (Nieuwkoop and Faber, 1994), and optic tract development was staged according to Chien and colleagues (Chien et al., 1993). Frogs and embryo procedures were approved by the Animal Care and Use Committee, University of Calgary. 2.2. Identification of rock1 The human ROCK1 and Xenopus tropicalis rock1 nucleotide sequences were used to identify expressed sequence tags (ESTs) from the X. laevis genome (Gurdon Institute's Sequence Database, Cambridge). One EST (Clone IMAGE, 6641103) mapped to the 3′ end of rock1. A reverse primer AGCTGGCTTTTCCAGATGAAGTTTTTGC was designed based on this EST, and 5′ rapid amplification of cDNA ends polymerase chain reaction (RACE-PCR, Clonetech) was used to amplify the rock1 coding sequence from cDNA (whole embryos Stages 32 to 40). Gene walking in the 3′ to 5′ direction was used to identify a 4083 base pair (bp) sequence that encodes a full-length Rock1 protein. ClustalW (Goujon et al., 2010; Larkin et al., 2007) and Jalview 2 (Waterhouse et al., 2009) web applications were used to align and annotate rock sequences from multiple species. 2.3. In situ hybridization Probe synthesis and in situ hybridizations were performed as outlined by Sive et al., 2000. The following plasmids were linearized and used as templates for the synthesis of Digoxygenin-labeled riboprobes: pBSK-xfgfr1 and pCS2-xslit2 as per Atkinson-Leadbeater et al., 2010, pGEMT-xlhx9 from S. Retaux (Institut Alfred Fessard, Gif sur Yvette, France) (Bachy et al., 2001), and pBSK-ntn1 from C. Holt (Univ. Cambridge, England). For xrock1, the plasmid Clone IMAGE, 6641103 was linearized with Sma1 enzyme and transcribed with T7 RNA polymerase. The antisense xrock2 riboprobe was synthesized from a plasmid (J. Han, Pohang Univ., Korea) encoding full-length Xenopus rock2 (NCBI Accession NM_001087476.1) by linearizing with Ava1 enzyme and transcribing with T7 RNA polymerase. Wholemount in situ hybridization was performed as described previously (Sive et al., 2000). Briefly, embryos were fixed in MEMFA (0.1 M MOPS, 2 mM EGTA, 1 mM MgSO4, 4% formaldehyde) overnight at 4 °C and stored in 100% ethanol at −20 °C. Embryos were rehydrated and permeabilized with Proteinase-K (Sigma, 10 μg/mL), and hybridized at 60 °C to 65 °C with riboprobe in hybridization buffer (10% MEMFA, 50% formamide, 0.1 g/mL dextran sulfate, 1 mg/mL torula RNA (Sigma), 1 × Denhardt's solution). Embryos were blocked in 2% Boehringer Mannheim Blocking (BMB) reagent (Roche), and incubated overnight at 4 °C with alkaline phosphatase-coupled anti-DIG antibody (Roche, 1:2000). In situ signal was detected with a colorimetric reaction using BM purple (Roche) or NBT/BCIP (Roche). Staining was stopped by fixation in modified Bouin's fix (10% formaldehyde, 5% glacial acetic acid). For in situ hybridization on sections, Stage 33/34 and 35/36 embryos were fixed in MEMFA overnight at 4 °C, then sunk in 30% sucrose and embedded in Optimal Cutting Temperature compound (OCT, TissueTek). Twelve micrometer transverse sections through the brain and eyes were cut with a cryostat (Microm) and placed on glass slides. Slides were incubated at 60 °C to 65 °C in hybridization buffer containing DIG-labeled sense or antisense riboprobes. The slides were washed

31

and anti-DIG antibody (Roche, 1:2500) conjugated to alkaline phosphatase was used to detect the riboprobe with NBT/BCIP staining. Slides were washed and mounted with Aquapolymount (Polysciences Inc.). Here, and elsewhere in the paper, photographic images were taken with a Zeiss Stemi SV 11 microscope with AxioCam HRc camera or Zeiss Axioplan2 microscope with MRc camera and Axiovision software. Images were compiled and edited with Adobe Photoshop. 2.4. Eye explant cultures Stage 24 embryos were collected and anesthetized in Modified Barth's Saline (MBS, 88 mM NaCl, 1 mM KCl, 0.82 mM MgSO4, 0.33 mM Ca(NO3)2, 0.41 mM CaCl2, 10 mM HEPES, pH 7.5) containing 0.4 mg/mL tricaine (ethyl 3-aminobenzoate methanesulfonic acid, Sigma). Intact eye buds were placed on coverslips coated with poly-L-ornithine and 10 μg/mL laminin (Sigma), grown for 24 h at 22 °C in culture media (60% L15 (Invitrogen), 0.01% (w/v) bovine serum albumin (BSA, Sigma), 1% antibiotic/antimycotic (Invitrogen)). Of note, sister embryos had reached Stage 35/36, with axons in the middiencephalon over the same period. Cultures were then fixed in 2% paraformaldehyde (PFA). 2.5. Immunostaining Immunostaining was performed as described previously (McFarlane et al., 1995). Briefly, for eye bud cultures, fixed explants were washed with phosphate buffered saline (PBS) containing 0.25% Triton-X100 (Sigma) and 0.2% BSA, blocked with 5% goat serum, and incubated with primary antibody. For immunostaining on slides, whole embryos were fixed in 4% PFA at 4 °C, sunk in 30% sucrose and embedded in OCT reagent. Twelve μm thick tissue sections were cut transversely through the brain and eyes using a cryostat (Microm), then washed with PBS containing 0.5% Triton-X100 and 0.2% BSA, blocked with 5% goat serum, and incubated with primary antibody. Samples were mounted with Aquapolymount and imaged. Primary antibodies included rabbit polyclonal anti-human ROCK1/2 (Millipore 07–1458, 1:1000), rabbit polyclonal anti-human Rock1 (Abcam ab97592, 1:200), rabbit polyclonal anti-Xenopus Rock2 (1:2000) (a gift from J. Liu (Univ. of Ottawa) (Farah et al., 1998)) (see Supp. Fig. 1 for epitope/immunogen sequences), mouse monoclonal Islet1/2 (Developmental Studies Hybridoma Bank, 39.4D5), and mouse monoclonal anti-myc (9E10, 1:1000). Secondary antibody (Alexa 488 or Alexa 546, Invitrogen) was used at 1:500 for explants and 1:1000 for tissue sections. 2.6. Western blot Proteins extracted from Xenopus embryos Stages 28–40 were separated on a 6% polyacrylamide gel, transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad, Hercules, CA) and immunoblotted with rabbit polyclonal anti-human ROCK1/2 antibody (Millipore, 1:1000), anti-human Rock1 antibody (Abcam, 1:1000), and rabbit polyclonal anti-Xenopus Rock2 antibody (Farah et al., 1998, 1:1000). Antibodies were pre-incubated for 30 min at room temperature with homogenized Stages 16–24 embryos. Goat anti-rabbit peroxidase-conjugated secondary antibody was used to detect protein expression by enhanced chemiluminescense (Perkin Elmer Corp, UK). 2.7. Exposed brain preparation The exposed brain preparation was performed as described previously (Chien et al., 1993). Stages 33/34 and 35/36 embryos were anesthetized in 1X MBS containing 0.4 mg/mL tricaine (pH 7.4). The outer skin and dura lining the brain were peeled back on the left side of the embryo, exposing the neuroepithelium to MBS with or without the water-soluble Rock inhibitors Y-27632 (Sigma) and HA-1077 (Sigma), or forskolin (dimethyl sulfoxide) (gift from the Giembycz lab,

32

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

Univ. Calgary). At Stage 40, approximately 20–22 h later, the optic tract was anterogradely labeled with horseradish peroxidase (HRP) (Atkinson-Leadbeater et al., 2010), and then fixed brains reacted with 0.7 mg/mL diaminobenzidine (DAB, Sigma) and Urea (Sigma) dissolved in 0.5% Triton-X100. For sectioning, embryos were embedded in gelatin/ albumin set with 25% glutaraldehyde, and 50 μm transverse sections were cut with a vibratome (Leica VT1000S) and collected on glass slides. Dehydrated, cleared sections and whole brains were mounted with Permount (Fisher). Brains with abnormal morphologies or where the optic tract was poorly labeled were discarded. Embryos were scored in wholemount as having a contralateral optic nerve defect if N10 axons were present. Data from multiple independent experiments were pooled.

the EST Clone IMAGE 6641103, which contained a portion of the 3′ end coding sequence and 3′ untranslated region of rock1 in X. laevis. Gene walking in the 5′ direction was used to identify a 4083 base pair sequence that encodes a full-length Rock1 protein. ClustalW (Goujon et al., 2010; Larkin et al., 2007) was used to align Rock1 protein sequences from various species and a phylogeny tree was generated (Fig. 1A). X. laevis Rock1 and Rock2 share 91% amino acid identity between their kinase domains, 48% identity between their Rho binding domains (RBDs), and 77% between their pleckstrin homology domains (PHDs). Overall, the two proteins share 63% amino acid identity and 79% conservation of functional residues (Fig. 1B).

2.8. Wildtype and dominant negative Rock constructs

Next we characterized the expression of rock genes in the developing visual system of X. laevis. The first RGCs are born at Stage 24 (one day post-fertilization), and several hours later, at Stage 28, these RGCs extend axons that begin their journey towards their midbrain target, the optic tectum. The first axons reach the ipsilateral ventral diencephalon at Stage 30, cross at the optic chiasm to the contralateral diencephalon by Stage 32, extend dorsally in the diencephalon, and enter the tectum at Stage 37/38 (Chien et al., 1993). Of note, most axons present at Stage 40 have arrived at the tectum. We first performed RNA in situ hybridization with antisense riboprobes against Xenopus rock1 and rock2. Sense probes gave no signal (Supp. Fig. 2A,B) and the rock2 riboprobe, but not the rock1 riboprobe, was able to identify wtRock2-overexpressing cells in the eye (Supp. Fig. 2C,D). In wholemount embryos, the two antisense riboprobes showed distinct in situ hybridization label in the brain and trunk (Fig. 2A,C), though both mRNAs were expressed in the developing eye (Fig. 2B,D). In transverse sections of the Stage 35/36 eye in situ label for rock1 (Fig. 2B) and rock2 (Fig. 2D) was observed in a salt and pepper fashion, with label present in some but not all cells in the newly formed RGC layer, and of the outer and inner nuclear layers. In situ signal was also observed in the lens and the proliferative ciliary marginal zone. We next determined whether Rock protein was expressed by RGCs over the period of growth of their axons to the optic tectum. We used a rabbit polyclonal antibody (Millipore) against human ROCK that was raised against an epitope in the N-terminal domain of the protein (Supp. Fig. 1) that is 90% identical at the amino acid level to X. laevis Rock1 and Rock2. The ROCK1/2 antibody recognized a protein band around 170 kDa, which is close to the predicted molecular weight of X. laevis Rock1 (159 kDa), and Rock2 (164 kDa, Amano et al., 1996) (Fig. 2I). Of note, the ROCK1/2 antibody recognized brain cells that overexpress Xenopus Rock2 in transverse cryostat sections (Supp. Fig. 3A,A’), as well as those overexpressing human ROCK1 (data not shown). We used the ROCK1/2 antibody to investigate the expression of Rock proteins in developing Xenopus RGCs. At Stage 33/34, when the first RGC axons reach the contralateral diencephalon, the ROCK1/2 antibody labeled the lens and cells throughout the retina, including somata in the RGC layer adjacent to the lens, and RGC axons within the ONH, optic nerve and ventral diencephalon (Fig. 2E,F, Supp. Fig. 3C). Similar labeling was observed with a Rock2 antibody against X. laevis Rock2 (Farah et al., 1998) (Fig. 2G), which we found recognized a band of approximately 170 kDa on a Western of Xenopus embryos (Fig. 2I), and with a ROCK1 antibody (Abcam) that was raised against a sequence in the N-terminal domain that is 97% identical at the amino acid level to X. laevis Rock1 and 88% to Rock2 (Supp. Figs. 1, 3C). To determine if Rock is present in the growth cones of extending RGC axons, we performed Rock immunolabeling on Stage 24 eye buds cultured as explants for 24 h (only RGC axons extend any distance from the explant). Of note, sister embryos had axons with growth cones at the optic chiasm and contralateral, ventral optic tract. The ROCK1/2 antibody immunolabeled the filopodia and central region of the growth cone, as well as the length of the axon shaft (Fig. 2H,H’). In summary, the protein expression and in situ hybridization data argue

To engineer a wtrock2 expression construct, full-length Xenopus rock2 sequence (plasmid from J. Liu, Univ. Ottawa) (Farah et al., 1998) was PCR amplified using the forward primer CCAGAGCAGAAATGTCTC and reverse primer CTGGGTTTATTTGGAGGAA, and sub-cloned into a pCR2 expression vector (Invitrogen), containing six in-frame 3′ myc tags, with EcoRI and Xba1 restriction enzymes (New England Biolabs). A dominant negative (dn) Xenopus Rock2 (dnrock2-myc) was designed by removing the kinase domain of the wtrock2-myc construct by digestion with Ava1 and EcoRV restriction enzymes (New England Biolabs) (Amano et al., 1998; Kim and Han, 2005; Zhang et al., 2009). The vector was blunted with Mung Bean Nuclease (New England Biolabs) and re-ligated to produce an in-frame deletion of the kinase domain. A human DNROCK1 construct with a 5′myc tag was provided as a gift from B. Winning (Eastern Michigan University), which contained full-length human ROCK1 Y39H K105A (Winning et al., 2002). This was reverse-mutated to create the WTROCK1-myc expression construct. 2.9. Gene transfer via electroporation dnrock2-myc, wtrock2-myc, WTROCK1-myc, or DNROCK1-myc were co-electroporated along with pCS2gfp into the eye of Stage 28 embryos as described previously (Chen et al., 2007; Haas et al., 2002). A fine tipped needle was made from a borosilicate glass capillary tube (World Precision Instruments) pulled on an electrode puller, and filled with 10% fast green and DNA solution. A Picospritzer (General Valve) was used to pressure eject solution medial to the right eye and then a brief current was passed across the head of the embryo using two platinum electrodes attached to a stimulator (Grass Technologies). Electroporated embryos were left in 0.1 × MMR and then fixed at Stage 39 in 4% PFA at 4 °C. Analysis was done on transverse, retinal cryostat sections processed for anti-myc immunohistochemistry. The numbers of GFP and/or myc positive RGCs with axons were quantified in central retinal sections that had a distinct RGC layer. The location of axon termination within the optic pathway was scored for those axons that could be followed over multiple sections and which terminated with a growth cone. Data from 3 to 5 independent experiments were pooled. 3. Results 3.1. Identification of X. laevis rock1 We investigated whether Rock functions as a downstream effector of axon guidance signaling in the developing visual system of X. laevis. In other species, rock1 (or rokβ) and rock2 (or rokα) have been identified (reviewed in Riento and Ridley, 2003). To date only rock2 is described in X. laevis (Farah et al., 1998). Thus, we first identified the Xenopus rock1 ortholog. 5′ RACE-PCR was used to amplify the rock1 coding sequence from X. laevis cDNA using a reverse primer based on

3.2. rock1 and rock2 are expressed in the developing visual system

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

33

Fig. 1. Gene structure and conservation of X. laevis and X. tropicalis rock1 and rock2. A: ClustalW was used to align the rock1 nucleotide sequences of X. laevis and X. tropicalis, zebrafish, chicken, mouse, and human, and a phylogeny tree was generated. The phylogram depicts the inferred evolutionary relationships between organisms based on rock1 divergence. B: ClustalW alignment of the predicted protein sequences for X. laevis Rock1 and Rock2 indicates that the two proteins share 63% amino acid identity, and when the functional redundancy of certain amino acids is considered, their homology match is increased to 79%. Rock1 residues (light gray) flanking the kinase, rho-binding (RBD) and pleckstrin homology (PHD) domains are predicted based solely on the alignment with Rock2.

that RGCs and their axons express both Rock1 and Rock2 at the time the axons are extending through the ventral diencephalon. 3.3. A subset of RGC axons invade the contralateral optic nerve with Rock inhibition The expression data indicate that Rock protein is expressed in many developing Xenopus RGCs and their axons. Given the importance of Rock in signaling downstream of axon guidance cues in vitro (Amano et al., 2010; Schofield and Bernard, 2013), we investigated the in vivo role of Rock in RGC axon guidance. We used an in vivo exposed brain preparation (Chien et al., 1993) to pharmacologically inhibit Rock function during the time of RGC axon growth to the optic tectum. The outer skin and dura were peeled back on one side of the brain at Stage 33/34, exposing the neuroepithelium to the bathing solution. In this way, the skin restricts the access of drugs, and only the anterior brain on one side of the embryo is exposed to the inhibitors. Embryos were left to develop in control solution or the selective Rock inhibitor Y-27632 until Stage 40. Y-27632 competes with endogenous adenosine triphosphate for

the binding of conserved amino acid residues in the kinase domain of both Rock1 and Rock2 (Breitenlechner et al., 2003; Narumiya et al., 2000). Y-27632 was shown previously to inhibit 50% of Xenopus Rock2 kinase activity at 10 μM (Ohan et al., 1999), and has been used in Xenopus in situ preparations to block Rock signaling (Broders-Bondon et al., 2007; Harata et al., 2013). At Stage 40, RGC axons were anterogradely labeled with HRP and then assayed for guidance defects in a blinded fashion. In controls, RGC axons left the eye and extended through the diencephalon to innervate the optic tectum (Fig. 3A). When assessed in a wholemount preparation, occasionally (8/51 embryos), one or two axons were seen to enter the contralateral optic nerve (Fig. 3D), and in the vast majority of embryos (38/51) no axons were seen to enter. In 25 μM Y-27632-treated brains, significant numbers of HRP-labeled axons continued along the correct path and extended dorsally in the diencephalon. HRP-labeled axons (N10 axons), however, were frequently observed (81.8%, 18/22 embryos, Fig. 3D) in the contralateral optic nerve (black arrowheads in Fig. 3B,B′,C,C′), a behavior that was only observed in a small number of control embryos (9.8%, 5/51 embryos). This behavior was particularly evident in

Fig. 2. Rock is expressed by RGCs and their axons. AD: in situ hybridization with antisense riboprobes for rock1 (A,B) and rock2 (C,D) of wholemount Stage 32 embryos (A,C) and Stage 35/ 36 transverse cryostat retinal sections (B,D). In A and C, arrows point to the eye, white arrowheads to rock1 in situ label in the trunk and black arrowheads to rock2 in situ label in the brain. E–F: Transverse cryostat sections at Stage 35/36 immunostained with an antibody against human ROCK1/2 (Millipore). Immunoreactivity is observed within the optic nerve head, optic nerve (arrow in E), axons at the optic chiasm and within the brain neuropil, and within somata in the RGC layer adjacent to the lens. Orientation bar in B applies to panels A–G. G: Transverse cryostat section through the eye at Stage 35/36 immunostained with an antibody against Xenopus Rock2 (Farah et al., 1998). H: Immunocytochemistry with the anti-human ROCK1/2 antibody of RGC growth cones in culture. I: Western blot analysis of protein isolated from Stages 32–33/34 Xenopus embryos for the hROCK1/2 antibody (Millipore) and for Xenopus Rock2 antibody. Scale bar in F is 300 μm for A and C, 50 μm for B, D, E, and F, 30 μm for G, and 5 μm for H. br, brain; ba, branchial arches; cmz, ciliary marginal zone; D, dorsal; l, lens; oc, optic chiasm; onh, on, optic nerve, optic nerve head; onl, outer nuclear layer; rgc, retinal ganglion cell layer; rpe, retinal pigment epithelium; and V, ventral.

34

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

Fig. 3. RGC axons invade the contralateral optic nerve with pharmacological inhibition of Rock. Rock signaling was inhibited pharmacologically in the exposed brain from Stages 33/34 through 40. RGC axons were visualized in wholemount brains by HRP anterograde labeling. Lateral (A–C) and ventro-lateral (A′–C′) views of HRP-labeled optic projections in wholemount brains treated with (B,C) and without (A) Y-27632. Dotted white line marks the contralateral optic nerve. A small projection to the basal optic nucleus is present in control and treated brains. In some Y-27632-treated brains, small numbers of axons are seen to stray into the telencephalon and/or dorsal diencephalon (white arrowhead). D: percent of embryos with a contralateral optic nerve (CON) phenotype. Data pooled from five independent experiments; ***, p b 0.001 according to Fisher's exact test. E–G: 50 μm vibratome cross-sections through the eyes and diencephalon of control (E) and 25 μM Y-27632 (F,G) treated embryos. Note the extension of axons (black arrowheads) in the Y-27632 treated embryos into the contralateral optic nerve and eye. Orientation bar in A applies to panels A–C and E–I. Scale bar in A is 100 μm for A–C and E–F, and 50 μm for A′–C′,G. H–I: Nomarski image of the RGC optic projection at the contralateral optic nerve (H), and a schematic showing the behavior of a subset of the HRP-positive axons over multiple focal planes (I). bon, basal optic nucleus; di, diencephalon; cbep, contralateral brain entry point; D, dorsal; e, eye; ibep, ipsilateral brain entry point; oc, optic chiasm; ot, optic tract; pi, pineal gland; tec, tectum; V, ventral.

transverse vibratome sections, where some RGC axons misrouted into the contralateral optic nerve and invaded the contralateral eye (compare Fig. 3E with F,G). With control, we found that most of the sectioned embryos had 0–2 axons in the contralateral optic nerve (13/17), with only one embryo showing a significant contralateral optic nerve projection. In contrast, all 25 μM Y-27632-treated embryos (n = 10 sectioned) had greater than 10 axons in the contralateral optic nerve. Of note, no aberrant branching of RGC axons was observed in the vicinity of the contralateral brain entry point with Y-27632 treatment (Fig. 3H,I). In a dose response assay, the presence of a bundle of axons in the contralateral optic nerve was seen occasionally at 5 μM (22%, n = 9) (Fig. 3B), and in the vast majority of embryos treated with 25 μM Y-27632 (81.8%, n = 22, Fig. 3C) and 100 μM Y-27632 (90.9%, n = 11), as compared to only 9.8% in controls (n = 51) (Fig. 3D). 25 μM Y-27632 was used for further experimentation. To verify the specificity of the guidance defect for Rock inhibition we asked whether a second Rock inhibitor generated a similar RGC axon defect. We exposed the developing brains of Stage 33/34 embryos to HA-1077 (Fausadil) (Lingor et al., 2007), and assayed in a blinded fashion for defects in the behavior of HRP-labeled RGC axons (Fig. 4A-C). In control (20/21), and with 5 μM and 25 μM HA-1077 (12/12 and 4/4, respectively), most optic projections had no or 1–3 axons that invaded the contralateral optic nerve. At 75 μM HA-1077, most embryos had a subpopulation of RGC axons that invaded the

contralateral optic nerve and eye (arrowheads in Fig. 4B, 8/13). At 90–100 μM, the vast majority of embryos (15/17) exhibited the contralateral optic nerve phenotype, as confirmed in transverse vibratome sections (10/10 embryos with N10 axons in the contralateral optic nerve). Of note, the use of pharmacological inhibitors of other intracellular kinases in the exposed brain preparation does not produce a similar defect (Webber et al., 2005). As a final pharmacological means of manipulating Rock activity to verify the specificity of the effects we observed with the Rock inhibitors, we took advantage of the known negative regulation of Rock by the Protein kinase A (PKA) pathway (Brown et al., 2012; Park et al., 2006; Tkachenko et al., 2011). Brains were exposed to forskolin, an activator of adenylate cyclase that should increase cAMP and inhibit Rock. Forskolin resulted in a significant number of optic projections (17/19) with axons that entered the contralateral optic nerve, phenocopying the data obtained with Rock inhibition (Fig. 4D–F). The forskolin data further support the idea that RGC axons are prevented normally from entering the contralateral optic nerve, and that a cAMP-dependent mechanism, possibly involving the regulation of Rock activity, is involved. Interestingly, many axons that extended correctly past the contralateral optic nerve with Y-27632 treatment from Stage 32–33/34 onwards continued through the diencephalon to reach the optic tectum, with no axon extension or guidance defects (Fig. 3B,C). In control,

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

35

Fig. 4. HA-1077 and forskolin phenocopy the optic nerve defect seen with Y-27632. HRP-labeled optic projections viewed in 50 μm transverse vibratome sections (A,B,D,E). In embryos treated with 75 μM HA-1077 (B) or 5 μM forskolin (E) many axons inappropriately invade the contralateral optic nerve and eye (arrowheads). In control, few if any axons enter the contralateral optic nerve (A,D). Dotted white lines mark the optic nerves. C,F: percent of embryos showing the contralateral optic nerve defect with 75 μM HA-1077 (C) and 5 μM forskolin (F). Data pooled from four independent experiments; ***, p b 0.001 according to Fisher's exact test. D, dorsal; e, eye; oc, optic chiasm; tec, tectum; V, ventral.

96.1% (n = 51) of optic tracts guided properly, and the tectum of 98% (n = 51) of brains was well innervated. At a dose of 25 μM Y-27632, where 81.8% (n = 22) of embryos exhibited the contralateral optic nerve phenotype, the majority of optic tracts (81.8%, n = 22, p N 0.05 Fisher Exact test) showed no other defects in axon guidance. In some embryos (9.1%, n = 22), the occasional RGC axon (b 10 axons) strayed into the telencephalon and/or dorsal diencephalon (Fig. 3B). For the vast majority of the embryos (90.9%, n = 22), however, the bulk of the axons that reached the contralateral diencephalon went on to innervate the optic tectum (Fig. 3B). Even, in the two embryos (9.1%, n = 22) when more axons (≫10 axons) strayed from the optic tract, large numbers of axons still innervated the optic tectum. These data argue that, unlike what happens with RGC axon behavior at the contralateral optic nerve, there is no strong requirement for Rock for most RGC axons to make the appropriate guidance decisions within the contralateral brain and to reach the optic tectum. The fact that a higher dose of Y-27632 (75 μM) only marginally increased the numbers of optic tracts with large numbers (N20 axons)

of misguided axons (23.8%, n = 21), suggests that the minor effect on guidance within the diencephalon was not explained by sub-maximal inhibition of Rock. Indeed, the kinase activity of Xenopus Rock2 is 50% inhibited by 10 μM Y-27632 (Ohan et al., 1999). The minimal effect of Y-27632 on RGC axon guidance within the diencephalon and midbrain could arise if a sub-population of RGC axons that expressed Rock and were sensitive to Y-27632 were already lost into the contralateral optic nerve with Y-27632 treatment starting at Stages 32–33/34. To address this issue, we delayed the brain exposure assay and drug application by five hours to Stage 35/36, when pioneering axons normally reach the mid-diencephalon. By this means we would allow many more of a potential subset of Rock-expressing axons to cross at the optic chiasm and reach the contralateral brain. Allowing more of the RGC axons to reach the diencephalon before exposure to Y-27632 did decrease the numbers of optic tracts without guidance errors (33.3%, n = 15 25 μM Y-27632 vs 91.7%, n = 24, control) as compared to Y27632 treatment starting at Stage 33/34 (81.8%, n = 22), though the majority of optic tracts (66.7%, n = 15) had only a handful (b10

Fig. 5. Delayed application of the Rock inhibitor produces no additional guidance defects. A–D: Lateral (A,B) or ventro-lateral (A′,B′) wholemount views at Stage 40 of HRP-labeled optic projections exposed at Stage 35/36 to either control solution (A,A′) or 25 μM Y-27632 (B,B′). In embryos treated with Rock inhibitor some RGC axons inappropriately invade the contralateral optic nerve (CON; arrowhead in B′), and in some embryos a handful of axons stray from the optic tract (arrowheads in B). Dotted white lines mark the optic nerves. C: percent of control and 25 μM Y-27632 treated embryos that show few or no defects in axon guidance in the diencephalon (guidance) and with most axons properly innervating the optic tectum (target innervation). Data pooled from four independent experiments; **, p b 0.01 according to Fisher's exact test. Scale bar in A is 100 μm for A,B and 50 μm for A′,B′. di, diencephalon; D, dorsal; oc, optic chiasm; ot, optic tract; tec, tectum; V, ventral.

36

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

axons) that misrouted (Fig. 5C). Indeed, the vast majority of Y-27632treated embryos had optic tecta that were well innervated (93.3%, n = 15) (Fig. 5B). These data argue against a major role for Rock in

axon extension and guidance in the optic tract, though some axons appear to rely on the activity of the Rock signaling pathway to avoid straying into the telencephalon and/or dorsal diencephalon. 3.4. Gross patterning of the diencephalon is normal in the presence of Rock inhibitor With the exposed brain preparation, the Rock inhibitor has access to both the RGC axons and the brain neuroepithelium. Thus, it is possible that Rock works cell non-autonomously to guide RGC axons, by controlling the patterning of the diencephalic neuroepithelium. While we could not assess all means by which Rock inhibition could affect the neuroepithelium, we could determine whether at a gross level the patterning of the diencephalon, with specific emphasis on the ventral region, was affected. Thus, we assessed the expression of dorsally and ventrally expressed genes. mRNA for lhx9, a LIM homeodomain transcription factor that regulates neuronal differentiation (Molle et al., 2004; Moreno et al., 2005), was expressed in the dorsal brain of both control embryos and those treated with 25 μM Y-27632 (n = 11 control, n = 11 Y27632, Fig. 6C,D). Similarly, several ventrally expressed genes were expressed normally in inhibitor treated brains. These included Islet1/2 (n = 6 control, n = 6 Y-26732, Fig. 6A,B), fibroblast growth factor receptor-1 (fgfr1) (n = 8 control, n = 6 Y-27632, Fig. 6E,F), and the guidance molecules ntn1 (n = 6 control, n = 7 Y-27632, Fig. 6G,H) and slit2 (n = 4 control, n = 4 Y-27632, Fig. 6I,J) (Atkinson-Leadbeater et al., 2010; Elshatory et al., 2007; Hocking et al., 2010; de la Torre et al., 1997). Together with the observation that numerous RGC axons still navigate the ventral diencephalon and reach the optic tectum, these data argue that ventral patterning, at least at a gross level, occurs normally when Rock function is inhibited. 3.5. Disruption of Rock results in RGC axon initiation and guidance defects While the marker analysis showing no dramatic change in the ventral diencephalon with Rock inhibition were supportive of a role for Rock in the RGC axons themselves, to test directly for such a cellautonomous role we used a molecular approach to manipulate Rock function selectively in RGC axons. We used a myc-tagged human DNROCK1 (Y39H and K105A) plasmid in an eye electroporation assay that was used previously in Xenopus (Winning et al., 2002), as well as a myc-tagged human WTROCK1 plasmid we engineered. A Xenopus dnrock2 construct was created based on published reports (Amano et al., 1998; Farah et al., 1998; Kim and Han, 2005; Zhang et al., 2009) by deleting the kinase domain (amino acids 1 to 447) and adding myc tags for visualization purposes. An identical construct causes severe convergent-extension defects in both zebrafish and Xenopus, which are phenocopied by inhibition of the upstream pathway component RhoA, and rescued in zebrafish by co-injection of wildtype Rock2 mRNA (Kim and Han, 2005; Marlow et al., 2002). Stage 28 eye electroporations were used to introduce the constructs around the peak of RGC genesis (Chen et al., 2007; Haas et al., 2002). Importantly, because only a small subset of retinal cells is electroporated, RGCs that express the rock transgenes develop in an otherwise wildtype background (McFarlane and Lom, 2011). CS2gfp alone was used as a control, Fig. 6. Patterning of the forebrain not grossly affected by Rock inhibition. The brains of Stage 33/34 embryos were exposed to control solution (A,C,E,G) or 25 μM Y-27632 (B,D,F,H) and fixed at Stage 40. A–B: immunohistochemistry with an antibody against Islet-1/2 performed on 12 μm transverse sections. Brain and one eye outlined with dots. C–H: lateral (C–F), ventro-lateral (E′,F′,G,H), and ventral (I,J) views of wholemount brains processed for in situ hybridization with antisense riboprobes against the transcription factor lhx-9 (C,D), fgfr1 (E–F), ntn1 (G–H), and slit2 (I,J). Dotted white lines mark the optic nerves. Contralateral optic nerve guidance defect (arrows) evident by visualization of HRP-labeled RGC axons (brown). fgfr1 (E′,F′), ntn1 (G,H) and slit2 (I,J) mRNAs are present just anterior to the optic chiasm (arrowheads) in both control and Y-27632-treated brains. Scale bar in A is 100 μm. di, diencephalon; D, dorsal; e, eye; hy, hypothalamus; oc, optic chiasm; ot, optic tract; tec, tectum; V, ventral.

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

and was co-electroporated along with the rock constructs to help visualize the entire morphology of the electroporated cells. After electroporation, embryos were left to develop until Stage 39 and myc-immunolabeled RGC axons were followed in serial transverse cryostat sections through the forebrain and midbrain. To determine the percent of RGCs with an axon, we looked for myc + cells in the central retina, where we were confidently able to identify the RGC layer, and asked if they had initiated an axon from their basal surface. The vast majority of DNROCK1 (93.8%, n = 80, Fig. 7D) and dnRock2 (76.9%, n = 57, Fig. 7E) expressing RGCs initiated an axon. In contrast, excess Rock activity appeared to inhibit the ability of RGCs to initiate an axon. Less than half of wtRock2-expressing RGCs (39.8%, n = 221, Fig. 7E), and almost no WTROCK1 expressing RGCs (11.1%, n = 99, Fig. 7D), exhibited an axon. Many of the wtRock-expressing cells had dendritic arbors (Fig. 7C), arguing against the possibility that failure to initiate axons reflected sick RGCs. Rather, the data suggest that Rock activity must be kept low for RGCs to initiate axons. For each construct we assessed the final location of the transgeneexpressing axons in the optic pathway. Many control GFP-expressing axons had reached the contralateral optic tectum by Stage 39 (Fig. 7A,D). Normal Rock1 activity, however, appeared critical for axon extension in the early part of the optic pathway. Disruption of rock1 function, either by loss-of-function or overexpression, resulted in many axons terminating in the (ipsilateral) eye (WTROCK1, 45.5% n = 11, DNROCK1, 16% n = 75) (Fig. 7F). Indeed, few axons made it to the optic tectum (WTROCK1, 0% n = 11, DNROCK1, 16% n = 75) (Fig. 7D). Rock2 activity seemed less critical for RGC axon extension. Most dnRock2-expressing axons exited the eye (Fig. 7G), and a significant number reached the optic tectum (37.5%, n = 40, Fig. 7E), reflecting possibly low levels of endogenous Rock2 activity in RGCs. Overexpression of wtRock2 resulted in a small number of axons reaching the tectal target (22.7%, n = 88, Fig. 7B,E). The dramatic axon initiation and extension defects observed with the Rock1 constructs precluded any assessment of whether transgeneexpressing RGC axons made guidance errors at the BEP of the contralateral optic nerve. Instead, we focused on the results with Rock2 overexpression. Importantly, a significant number of wtRock2-expressing RGC axons (15.9%, n = 88) misrouted into or stalled at the BEP of the contralateral optic nerve (Fig. 7B’,H), a behavior not observed for any control GFP positive axon (Fig. 7A,H). A small percent of dnRock2-expressing axons (5.0%, n = 40) also terminated at the contralateral BEP (Fig. 7H). These data argue that balanced Rock activity within the RGC axons themselves is required for the axons to properly navigate into and away from the optic nerve. 4. Discussion In this paper we explore a specific role for Rock in guidance decisions made by RGC axons. First, we identify the X. laevis ortholog of rock1 and show that both rock1 and rock2 mRNAs are expressed by RGCs at the time their axons extend through the ventral diencephalon. Further, ROCK1/2 antibody immunostaining confirms the presence of protein in RGC axons and their growth cones. Pharmacological inhibition of Rock function in vivo results in RGC axons aberrantly entering the optic nerve on the contralateral side of the embryo, while axon extension appears relatively unaffected. Interestingly, despite Rock inhibition, most RGC axons make appropriate guidance decisions within the contralateral optic tract. Molecular manipulation of Rock function within RGC axons, but not their neuroepithelial environment, provides additional evidence to argue for a cell-autonomous guidance role for Rock within RGC axons at the BEP of the contralateral optic nerve. Rock is regarded as a ubiquitously expressed protein that lies downstream of many signaling cascades (Schofield and Bernard, 2013; Thumkeo et al., 2013). In the developing X. laevis visual system, however, the salt and pepper expression of rock1 and rock2 mRNA suggests that not all retinal cells, including not all RGCs, express the kinase.

37

rock expression in developing zebrafish is also not ubiquitous. rock1 mRNA is present in anterior brain and head by 36 h post fertilization (hpf) (Chu et al., 2012), while rock2a (Marlow et al., 2002) and rock2b (Thisse et al., 2001) mRNAs are localized discretely to the developing eye and at the midbrain/hindbrain boundary, and a small subset of hindbrain cells, respectively. Thus, Rock likely has specific roles in the developing nervous system. In agreement, we found preliminarily that pharmacological inhibition of Rock produced no global defects in brain neuroepithelial patterning, and the majority of RGC axons in the distal portion of the optic projection extended appropriately. Instead, we observed defects in the guidance of a subset of RGC axons at the BEP of the contralateral optic nerve, and only small numbers of axons within the optic tract. The molecular functional data suggest that careful regulation of Rock activity is required for RGC axon initiation and extension within the eye. Of note, the pharmacological inhibitors were applied at a later stage and thus did not address a role for Rock in axon initiation. It appears that low levels of Rock activity are required for axon initiation, as increased Rock expression significantly reduced the number of RGCs that initiated axons, whereas DN inhibition increased axon initiation with respect to controls. Previous work in vitro on cerebellar granule neurons supports the idea of low Rock activity being a general requirement for neurons to initiate axons (Bito et al., 2000). Once initiated, however, we find that normal levels of Rock activity are required in the axons for them to extend out of the eye vesicle. Too much Rock activity (seen with wtRock1 overexpression), or too little (as produced with dnRock1), increased dramatically the percent of Rock-expressing RGC growth cones within the eye, at a time when their control counterparts were starting to reach the optic tectum. Interestingly, axon extension in the distal, contralateral portion of the optic project was unaffected by pharmacological inhibition of Rock. Together these data raise the possibility that normal Rock activity is required for RGC axons to extend, but only in the early stages of their journey to their tectal target. Of note, pharmacological and molecular disruption of Rock signaling in cerebellar granule cells in vitro produced only mild if any disruption in axon extension (Bito et al., 2000), indicating that Rock control of axon extension is dependent on signaling via cues present in the in vivo environment. The exposed brain experiments argue that there is an active mechanism to keep RGC axons from entering the contralateral optic nerve. This mechanism requires Rock activity and is disrupted by high cAMP levels, possibly because of a known role for cAMP in inhibiting Rock function (Brown et al., 2012; Park et al., 2006; Tkachenko et al., 2011). Our data argue that this Rock-dependent mechanism functions cell autonomously in axons. First, RGC growth cones express Rock protein. Second, with the Rock inhibitor no gross defect in the expression of several molecular markers, including the axon guidance molecules ntn1 and slit2, were observed in the ventral diencephalon. More significant, however, was the observation that disruption of Rock activity with wtRock2 results in approximately 15% of the transgene-expressing axons entering the contralateral optic nerve or collecting at its BEP. Interestingly, at concentrations of Y-27632 where most embryos exhibited a contralateral optic nerve phenotype, most axons still appropriately targeted the optic tectum. These data suggest that not all RGC axons are sensitive to alterations in Rock activity. In support, rock1 and rock2 mRNAs are not expressed by all RGCs, a higher dose of Y-27632 did not dramatically worsen axon pathfinding by RGC axons, and only a subset of wtRock2-expressing axons enter the contralateral optic nerve. In the future, it will be important to determine the RGC identity of this population of axons. Of note, the axon initiation and extension defects observed with Rock1 manipulation prevented us from properly assessing Rock1 in guidance at the BEP of the contralateral optic nerve. Nevertheless, it is significant that, with the exception of the wtRock2-expressing RGC axons at the contralateral optic nerve BEP, none of the Rock transgeneexpressing axons deviated significantly away from the optic projection.

38

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

Similarly, while pharmacological inhibition of Rock did result in straying of some RGC axons into the telencephalon and dorsal diencephalon, even at high doses the majority of RGC axons reached the optic tectum. These

data argue that there is not a strong dependence on Rock activity for guidance of most RGC axons through the optic tract. Given the demonstrated in vitro importance of Rock in signaling downstream of multiple

Fig. 7. Some wtRock2-expressing RGC axons invade the contralateral optic nerve. Single eyes of Stage 28 embryos were electroporated with a wtrock2, dnrock2, WTROCK1 or DNROCK1 construct and pCS2gfp, or pCS2gfp alone for control, and left to develop to Stage 39. Embryos were fixed and anti-myc immunohistochemistry was performed to visualize transgene expressing cells in transverse cryostat sections. A: GFP-positive RGC axons from the right eye target the contralateral optic tectum (tec) and avoid the contralateral optic nerve. B: some wtRock2-myc expressing RGC axons extend towards the optic tectum (arrow) while others enter the contralateral optic nerve (arrowhead), as seen at higher magnification in B′. Boxed regions in A and B highlight the brain entry point of the contralateral optic nerve. C: wtRock2-myc expressing RGC with dendrites (arrowhead). D–E: graphs for ROCK1 (D) and Rock2 (E) transgenes showing the percent of RGCs with an identifiable axon, and the percent of axons that reached their target, the optic tectum, expressing either GFP as control, or one of the Rock constructs. F–G: graphs for ROCK1 (F) and Rock2 (G) showing the percent of RGCs with an axon that terminated within the eye. H: graph for Rock2 showing the percent of RGCs with an axon that terminated at the BEP of the contralateral optic nerve or within the contralateral optic nerve. Numbers above bars are the numbers of RGCs for the black bars in D and E, axon numbers for the white bars in D and E, and axons numbers in F–H. Chi-square values (F) from Chi-square statistical analysis of the distributions are shown, showing that for each data set represented there are significant statistical differences between the distributions for the different experimental groups. di, diencephalon; D, dorsal; e, eye; inl, inner nuclear layer; l, lens; oc, optic chiasm; tec, tectum; V, ventral.

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

axon guidance cues, including Semas and Ephrins, these data are surprising (Govek et al., 2005; Hall and Lalli, 2010). Potentially they reflect the likelihood that not all RGCs express Rock, and/or that redundant signaling pathways ensure the making of proper guidance decisions. We know for example that while Slit1 and Sema3a co-regulate the behavior of RGC axons at a caudally-directed turn in the mid-diencephalon, the absence of one or the other protein has little effect on RGC axon behavior (Atkinson-Leadbeater et al., 2010). If at a given choice point Rock functions downstream of only one cue, the signaling induced by the redundant cues may be sufficient to ensure that proper guidance decisions are made. Whether Rock activity is required in RGC axons to move into or out of the ipsilateral optic nerve is unclear. The surgery for the in vivo pharmacological manipulations would not have exposed the ipsilateral optic nerve to the drugs. Our molecular data hint at a role for Rock at the ipsilateral optic nerve, in that a significant number of DNROCK1(27%, n = 11) and dnRock2- (10.7%, n = 40) expressing axons were found at the entrance or exit of the ipsilateral nerve, which was not true of control (0.8%, n = 133). Future experiments will need to address whether Rock controls entry into and exit from both the ipsilateral and contralateral optic nerve. The molecular manipulation experiments lend preliminary support for distinct roles for Rock1 and Rock2 in RGC axons. While our antibody and pharmacological experiments could not differentiate between Rock1 and Rock2, the differences observed with the DNROCK1 and dnRock2 constructs provide some support for a greater role for Rock1 in RGC axon development. DNROCK1 dramatically altered RGC axon initiation and extension in the early optic projection, whereas the dnRock2 had a less prominent effect on the ability of RGC axons to reach the optic tectum. The fact that wtRock2 overexpression affected RGC axon guidance at the contralateral BEP could reflect a specific function of Rock2 in the guidance of some RGC axons at this decision point, or a disruption of Rock1 activity by excess levels of the highly related Rock2. Thus, the distinct roles of Rock1 and Rock2 in axon initiation and extension in the eye, and at regulating axon behavior at the contralateral optic nerve, remain to be determined. Several models could explain the requirement for Rock in preventing RGC axons to enter the contralateral optic nerve. First, given the demonstrated in vitro importance of Rock in axon guidance signaling (Govek et al., 2005; Hall and Lalli, 2010), is that Rock functions in RGC axons as a downstream effector of the actions of a guidance cue. This cue could be a repulsive cue in the optic nerve, or an attractive cue that pulls axons on past the entry point of the contralateral optic nerve. Alternatively, Rock may prevent RGC axons from sensing an attractant expressed by the optic nerve. An excellent candidate in this last scenario is the secreted chemoattractant, Ntn1, which is expressed in the optic nerve (de la Torre et al., 1997). The receptor for Ntn1, deleted in colorectal cancer (DCC), is expressed by Xenopus RGCs, and Rock signaling downregulates DCC at the plasma membrane (Moore et al., 2008). Alternatively, Rock may participate in an error-correction mechanism, which would normally redirect axons that accidentally stray into the contralateral optic nerve. Such a mechanism functions at the optic chiasm in zebrafish, where guidance signaling involving the Slit ligand and Robo receptor redirects the occasional straying RGC axon back to the optic tract (Hutson and Chien, 2002). Thus, Rock could act downstream of a guidance molecule that normally corrects the occasional misrouting of axons at the contralateral BEP. An additional possibility is that ipsilateral RGC axons are never in proximity to the contralateral optic nerve, as a consequence of a mechanism that normally segregates and organizes eye specific optic projections through the optic chiasm. This mechanism could require Rock activity. In support, previous reports in zebrafish and mouse models indicate that disruption of RGC axon guidance and/or organization at the optic chiasm results in aberrant entry of RGC axons into the contralateral optic nerve (Macdonald et al., 1997; Plump et al., 2002). Yet, in Xenopus at least, RGC axons appear to grow close to the BEP of the contralateral optic nerve (see Fig. 3H). Further, there were no obvious gross defects in

39

the guided growth of axons at the optic chiasm with our pharmacological inhibitors (see Fig. 3F,4B). Finally, Rock signaling may normally promote selective fasciculation and segregation of axons from the two eyes, or prevent axon branching. If true, one might have expected defasciculated optic tracts with Rock inhibition in the former scenario, and enhanced axon branching in the latter, but at first glance this does not appear to be the case. In the future, in vivo time lapse microscopy of RGC axons in the presence and absence of Rock inhibition could start to distinguish between these possibilities. An interesting aspect of the axon behavior revealed by Rock inhibition is the presumed similarity of the two optic nerves. As discussed above, the fact that RGC axons normally grow in the ipsilateral optic nerve, but avoid the contralateral optic nerve, may simply reflect a Rock-dependent sorting of RGC axons at the optic chiasm that ensure the axons never sample the environment of the contralateral optic nerve. Alternatively, axons may switch their responsiveness to cues provided by the optic nerve, whether these control branching, error correction or guidance. Switches in the behavior of axons to presented cues are key for the crossing of the midline by the axons of dorsal commissural neurons in the spinal cord, and can be time or local environment dependent (Chedotal, 2011; Yam et al., 2012). Such a switch could also be key to the behavior of RGC axons with regard to the contralateral optic nerve. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.mcn.2015.10.001. Acknowledgments The authors would like to thank J. Johnston for excellent technical assistance, B. Winning for the DNROCK1 construct and J. Han for the wtrock2 sequence. Thank you to Dr. Giembycz for providing the forskolin. PC was supported by a studentship from Natural Science and Engineering Research Council (NSERC) of Canada, and SM was a Tier II Canada Research Chair in Developmental Neurobiology and is an Alberta Innovates-Health Solutions (AI-HS) scientist. Funds were provided by an operating grant from Canadian Institutes for Health Research (CIHR #14138) to SM. References Amano, M., Chihara, K., Nakamura, N., Fukata, Y., Yano, T., Shibata, M., Ikebe, M., Kaibuchi, K., 1998. Myosin-II activation promotes neurite retraction during the action of Rho and Rho-kinase. Genes Cells 3, 177–188. Amano, M., Mukai, H., Ono, Y., Chihara, K., Matsui, T., Mamajima, Y., Okawa, K., Iwamatsu, A., Kaibuchi, K., 1996. Identification of a putative target for Rho as the serine–threonine kinase protein kinase N. Science 271, 648–650. Amano, M., Nakayama, M., Kaibuchi, K., 2010. Rho-kinase/ROCK: a key regulator of the cytoskeleton and cell polarity. Cytoskeleton 67, 545–554. Atkinson-Leadbeater, K., Bertolesi, G., Hehr, C., Webber, C., Cechmanek, P., McFarlane, S., 2010. Dynamic expression of axon guidance cues required for optic tract development is controlled by fibroblast growth factor signaling. J. Neurosci. 30, 685–693. Bachy, I., Vernier, P., Retaux, S., 2001. The LIM-homeodomain gene family in the developing Xenopus brain: Conservation and divergences with the mouse related to the evolution of the forebrain. J. Neurosci. 21 7620–2629. Bito, H., Furuyashiki, T., Ishihara, H., Shibasaki, Y., Ohashi, K., Mizuno, K., Maekawa, M., Ishizaki, T., Narumiya, S., 2000. A critical role for a Rho-associated kinase, p160ROCK, in determining axon outgrowth in mammalian CNS neurons. Neuron 26, 431–441. Breitenlechner, C., GaBel, M., Hidaka, H., Kinzel, V., Huber, R., Engh, R., Bossemeyer, D., 2003. Protein kinase a in complex with Rho-kinase inhibitors Y-27632, fasudil and H-1152P: structural basis of selectivity. Structures 11, 1595–1607. Broders-Bondon, F., Chesneau, A., Romer-Oliva, F., Mazabraud, A., Mayor, R., Thiery, J., 2007. Regulation of XSnail2 expression by Rho GTPases. Dev. Dyn. 236, 2555–2566. Brown, J., Diggs-Andrews, K., Gianino, S., Gutmann, D., 2012. Neurofibromatosis-1 heterozygosity impairs CNS neuronal morphology in a cAMP/PKA/ROCK-dependent manner. Mol. Cell. Neurosci. 49, 13–22. Chedotal, A., 2011. Further tales of the midline. Curr. Opin. Neurobiol. 68-75. Chen, Y., Hehr, C., Atkinson-Leadbeater, K., Hocking, J., McFarlane, S., 2007. Targeting of retinal axons requires the metalloproteinase ADAM10. J. Neurosci. 27, 8448–8458. Chien, C.-B., Rosenthal, D., Harris, W., Holt, C., 1993. Navigational errors made by growth cones without filopodia in the embryonic Xenopus brain. Neuron 11, 237–251. Chu, L.-T., Hong, S., Kondrychyn, I., Loh, S.L., Ye, Z., Korzh, V., 2012. Yolk syncytial layer formation is a failure of cytokinesis mediated by Rock1 function in the early zebrafish embryo. Biol. Open 1, 747–753.

40

P.B. Cechmanek et al. / Molecular and Cellular Neuroscience 69 (2015) 30–40

de la Torre, J., Hopker, V.H., Ming, G.-l., Poo, M.-m., Tessier-Lavigne, M., HemmatiBrivanlou, A., Holt, C.E., 1997. Turning of retinal growth cones in a netrin-1 gradient mediated by the netrin receptor DCC. Neuron 19, 1211–1224. Elshatory, Y., Everhart, D., Deng, M., Xie, X., Barlow, R., Gan, L., 2007. Islet-1 controls the differentiation of retinal bipolar and cholinergic amacrine cells. J. Neurosci. 14, 12707–12720. Farah, S., Agazie, Y., Ohan, N., Ngsee, J., Liu, J., 1998. A Rho-associated protein kinase, roka, binds insulin receptor substrate-1 and modulates insulin signaling. J. Biol. Chem. 273, 4740–4746. Gallo, G., 2006. RhoA-kinase coordinates F-actin organization and myosin II activity during semaphorin-3A-induced axon retraction. J. Cell Sci. 119, 3413–3423. Goujon, M., McWilliam, H., Li, W., Valentin, F., Squizzato, S., Paem, J., Lopez, R., 2010. A new bioinformatics analysis tools framework at EMBL-EBI. Nucleic Acids Res. 38, W695–W699. Govek, E.-E., Newey, S., VanAelst, L., 2005. The role of the Rho GTPases in neuronal development. Genes Dev. 19, 1–49. Haas, K., Jensen, K., Sin, W.C., Foa, L., Cline, H., 2002. Targeted electroporation in Xenopus tadpoles in vivo — from single cells to the entire brain. Differentiation 70, 148–154. Hall, A., Lalli, G., 2010. Rho and Ras GTPases in axon growth, guidance, and branching. Cold Spring Harbor Perspectives in Biology, pp. 1–18. Harata, A., Matsuzaki, T., Nishikawa, A., Ihara, I., 2013. The cell sorting process of Xenopus gastrul cells involves the acto-myosin system and TGF-B signaling. In Vitro Cell. Dev. Biol. 49, 220–229. Hocking, J., Hehr, C., Bertolesi, G., Wu, J., McFarlane, S., 2010. Distinct roles for Robo2 in the regulation of axon and dendrite growth by retinal ganglion cell. Mech. Dev. 127, 36–48. Hopker, V., Shewan, D., Tessier-Lavigne, M., Poo, M.-m., Holt, C., 1999. Growth-cone attraction to netrin-1 is converted to repulsion by laminin-1. Nature 401, 69–73. Hutson, L., Chien, C.-B., 2002. Pathfinding and error correction by retinal axons: the role of astray/robo2. 33, 205–217. Kim, G.-H., Han, J.-K., 2005. JNK and ROKa function in the noncanonical Wnt/RhoA signaling pathway to regulate Xenopus convergent extension movements. Dev. Dyn. 232, 958–968. Larkin, M., Blackshields, G., Brown, N., Chenna, R., McGettigan, P., McWilliam, H., Valentin, F., Wallace, I., Wilm, A., Lopez, R., Thompson, J., Gibson, T., Higgins, D., 2007. ClustalW and ClustalX version 2. Bioinformatics 23, 2947–2948. Lingor, P., Teusch, N., Schwarz, K., Mueller, R., Mack, H., Bahr, M., Mueller, B., 2007. Inhibition of Rho kinase (ROCK) increases neurite outgrowth on chondroitin sulphate proteoglycan in vitro and axonal regeneration in the adult optic nerve in vivo. J. Neurochem. 103, 181–189. Loudon, R., Silver, L., Yee, H., Gallo, G., 2006. RhoA-kinase and myosin-II are required for the maintenance of growth cone polarity and guidance by nerve growth factor. J. Neurobiol. 66, 847–867. Macdonald, R., Scholes, J., Strahle, U., Brennan, C., Holder, N., Brand, M., Wilson, S., 1997. The pax protein Noi is required for commissural axon pathway formation in the rostral forebrain. Development 124, 2397–2408. Marlow, F., Topczewski, J., Sepich, D., Solnica-Krezel, L., 2002. Zebrafish Rho kinase 2 acts downstream of Wnt11 to mediate cell polarity and effective convergence and extension movements. Curr. Biol. 12, 876–884. McFarlane, S., McNeill, L., Holt, C., 1995. FGF signaling and target recognition in the developing Xenopus visual system. Neuron 15, 1017–1028. McFarlane, S., Lom, B., 2011. The Xenopus retinal ganglion cell as a model neuron to study the establishment of neuronal connectivity. Dev. Neurobiol.

Molle, B., Pere, S., Failli, V., Bach, I., Retaux, S., 2004. Lhx9 and lhx9alpha: differential biochemical properties and effects on neuronal differentiation. DNA Cell Biol. 23, 761–768. Moore, S., Correia, J., Sun, K., Pool, M., Fournier, A., Kennedy, T., 2008. Rho inhibition recruits DCC to the neuronal plasma membrane and enhances axon chemoattraction to netrin-1. Development 135, 2855–2864. Moreno, N., Bachy, I., Retaux, S., Gonzalez, A., 2005. LIM-homeodomain genes as territory markers in the brainstream of adult and developing Xenopus laevis. J. Comp. Neurobiol. 9, 240–254. Narumiya, S., Ishizaki, T., Uehata, M., 2000. Use and properties of ROCK-specific inhibitor Y-27632. Methods Enzymol. 325, 273–284. Nieuwkoop, P., Faber, J., 1994. Normal table of Xenopus laevis. Garland, (Daudin) New York. Ohan, N., Agazie, Y., Cummings, C., Booth, R., Bayaa, M., Liu, J., 1999. RHO-associated protein kinase alpha potentiates insulin-induced MAP kinase activation in Xenopus oocytes. J. Cell Sci. 112, 2177–2184. Park, E., Kim, G.-H., Choi, S.-C., Han, J.-K., 2006. Role of PKA as a negative regulator of PCP signaling pathway during Xenopus gastrulation movements. Dev. Biol. 292, 344–357. Plump, A., Erskine, L., Sabatier, C., Brose, K., Epstein, C., Goodman, C., Mason, C., TessierLavigne, M., 2002. Slit1 and Slit2 cooperate to prevent premature midline crossing of retinal axons in the mouse visual system. Neuron 33, 219–232. Riento, K., Ridley, A., 2003. ROCKs: multifunctional kinases in cell behaviour. Nat. Rev. 4, 446–456. Schofield, A., Bernard, O., 2013. Rho-associated coiled-coil kinase (ROCK) signaling and disease. Crit. Rev. Biochem. Mol. Biol. 48, 301–316. Sive, H., Grainger, R., Harland, R., 2000. Early Development of Xenopus laevis. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Thisse, B., Obrecht-Pflumio, S., Furthauer, M., Loppin, B., Hayer, V., Degrave, A., Woehl, R., Steffan, T., Thisse, C., 2001. Expression of the Zebrafish Genome During Embryogenesis, ZFIN Direct Data Submission, pp. ZFIN ID: ZDB-PUB-010810-010811. Thumkeo, D., Watanabe, S., Narumiya, S., 2013. Physiological roles of Rho and Rho effectors in mammals. Eur. J. Cell Biol. 92, 303–315. Tkachenko, E., Sabouri-Ghomi, M., Pertz, O., Kim, C., Gutierrez, E., Machacek, M., Groisman, A., Danuser, G., Ginsberg, M., 2011. Protein kinase A governs a RhoARhoGDI protrusion-retraction pacemaker in migrating cells. Nat. Cell Biol. 13, 660–667. Waterhouse, A., Procter, J., Martin, D., Clamp, M., Barton, G., 2009. Jalview version 2 — a multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191. Webber, C., Chen, Y.-Y., Hehr, C., Jonhnston, J., McFarlane, S., 2005. Multiple signaling pathways regulate FGF-2-induced retinal ganglion cell neurite extension and growth cone guidance. Mol. Cell. Neurosci. 30, 37–47. Winning, R., Ward, E., Scales, J., 2002. EphA4 catalytic activity causes inhibition of RhoA GTPase in Xenopus laevis embryos. Differentiation 46–55. Yam, P., Kent, C., Morin, S., Farmer, T., Alchini, R., Lepelletier, L., Coleman, D., TessierLavigne, M., Fournier, A., Charron, F., 2012. 14-3-3 proteins regulate a cell-intrinsic switch from sonic hedgehog-mediated commissural axon attraction to repulsion after midline crossing. Neuron 76, 735–749. Zhang, Y., Li, X., Qi, J., Wang, J., Liu, X., Zhang, H., Lin, S.-C., Meng, A., 2009. Rock-2 controls TGFb signaling and inhibits mesoderm induction in zebrafish embryos. J. Cell Sci. 122, 2197–2207.