Dynein/dynactin is necessary for anterograde transport of Mbp mRNA in oligodendrocytes and for myelination in vivo

Significance

Oligodendrocytes in the brain insulate neuronal axons in layers of fatty myelin to facilitate fast electrical signaling. Myelin basic protein (MBP), an important myelin component, is transported as mRNA away from the cell body before being translated into protein. In zebrafish, the anterograde motor kinesin transports mbp mRNA away from the cell body. We now identify myelination defects in zebrafish caused by a mutation in the retrograde motor complex dynein/dynactin, which normally transports cargos back toward the cell body. However, this mutant displays defects in anterograde mbp mRNA transport. We confirm in mammalian oligodendrocyte cultures that drug inhibition of dynein arrests transport in both directions and decreases MBP protein levels. Thus, dynein/dynactin is paradoxically required for anterograde mbp mRNA transport.

Abstract

Oligodendrocytes in the central nervous system produce myelin, a lipid-rich, multilamellar sheath that surrounds axons and promotes the rapid propagation of action potentials. A critical component of myelin is myelin basic protein (MBP), expression of which requires anterograde mRNA transport followed by local translation at the developing myelin sheath. Although the anterograde motor kinesin KIF1B is involved in mbp mRNA transport in zebrafish, it is not entirely clear how mbp transport is regulated. From a forward genetic screen for myelination defects in zebrafish, we identified a mutation in actr10, which encodes the Arp11 subunit of dynactin, a critical activator of the retrograde motor dynein. Both the actr10 mutation and pharmacological dynein inhibition in zebrafish result in failure to properly distribute mbp mRNA in oligodendrocytes, indicating a paradoxical role for the retrograde dynein/dynactin complex in anterograde mbp mRNA transport. To address the molecular mechanism underlying this observation, we biochemically isolated reporter-tagged Mbp mRNA granules from primary cultured mammalian oligodendrocytes to show that they indeed associate with the retrograde motor complex. Next, we used live-cell imaging to show that acute pharmacological dynein inhibition quickly arrests Mbp mRNA transport in both directions. Chronic pharmacological dynein inhibition also abrogates Mbp mRNA distribution and dramatically decreases MBP protein levels. Thus, these cell culture and whole animal studies demonstrate a role for the retrograde dynein/dynactin motor complex in anterograde mbp mRNA transport and myelination in vivo.

In the central nervous system (CNS), specialized glial cells called oligodendrocytes wrap axons in many layers of plasma membrane to form the myelin sheath. Oligodendrocytes originate from neuroepithelial precursors that develop into oligodendrocyte precursor cells (OPCs), which are migratory and proliferative, extending numerous processes to sample the environment (1). OPCs differentiate into postmitotic oligodendrocytes, which activate expression of mature myelin proteins and ensheathe multiple axon segments with loose membrane spirals that are eventually compacted to form a functional myelin sheath (2). Disruption of the myelin membrane can cause debilitating human conditions, including multiple sclerosis. However, although the clinical applications of myelin research are clear, molecular mechanisms regulating basic oligodendrocyte development are not well understood.

A critical protein generated during oligodendrocyte differentiation is myelin basic protein (MBP), which is essential for proper compaction of the myelin sheath. Due to its highly basic charge and propensity to promote membrane adherence, Mbp translation is tightly regulated during oligodendrocyte development (3). Mbp mRNA is trafficked to the developing sheath and translated locally (4, 5). Translation at the membrane and formation of myelin sheaths is stimulated by Fyn kinase, which is phosphorylated in response to axonal electrical activity (6–8). In addition, MBP acts as an important spatial and temporal regulator of myelination, by triggering disassembly of the actin cytoskeleton to promote initiation of myelin membrane wrapping (9, 10).

Classic experiments in cultured oligodendrocytes demonstrated that Mbp mRNA trafficking in the anterograde direction (away from the cell body) relies on microtubules (11). By electron microscopy, these microtubules are uniformly oriented with polymerizing plus ends directed away from the cell body (12). Two types of motors move along microtubules: a vast family of kinesin motors, the majority of which move toward the plus end, and a single cytoplasmic dynein motor that moves toward the minus end when bound to the dynactin activator complex (13). Previously, a genetic screen in zebrafish identified a kinesin kif1b mutant in which anterograde mbp mRNA transport is disrupted, resulting in mislocalization of mbp mRNA in oligodendrocyte cell bodies (14). Importantly, another zebrafish genetic screen for myelination defects uncovered a dynein mutation that results in decreased mbp mRNA levels in both the peripheral nervous system (PNS) and CNS, highlighting the necessity of molecular motors in myelination (15, 16).

Classic transport studies have demonstrated the interdependence of oppositely directed motors. Many vesicular cargos are simultaneously associated with both dynein and kinesin motors, and inhibition or loss of either motor results in transport arrest in both anterograde and retrograde directions (17). This has been observed for organelles in squid axoplasm treated with anti-dynactin antibodies (18); for synaptic vesicles in axons of Drosophila with mutations in kinesin, dynein, or dynactin (19); and for vesicles and lysosomes in mammalian axons in which subunits of kinesin and dynactin have been knocked down (20, 21). Though less is known about how opposing motors regulate mRNA transport, one early study in Drosophila S2 macrophage-like cells showed that dynein knockdown leads to arrested transport in both anterograde and retrograde directions of fluorescently tagged fragile X mental retardation protein (FMRP), which is an RNA-binding protein (22). However, a role for the interdependence of kinesin and dynein motors in mRNA transport has yet to be shown in glial cells.

Here, we demonstrate in vivo in zebrafish and in mammalian oligodendrocyte cultures that dynein/dynactin is required for anterograde mbp mRNA transport. In a myelination screen in zebrafish, we discovered a mutation in a subunit of the dynein activator dynactin. This mutation in actr10, which encodes the Arp11 protein, results in dynein/dynactin loss-of-function phenotypes, such as photoreceptor loss and aberrant melanosome distribution in pigment cells. Mutants have OPC proliferation defects, fewer myelinated axons by transmission electron microscopy (TEM) and reduced levels of mbp mRNA, all of which phenocopy a previously published zebrafish dynein mutant (16). Interestingly, of the oligodendrocytes that are present in the mutants, we noticed a reduction in mbp mRNA localized to processes, which is similar to the previously published kinesin kif1b zebrafish mutant and therefore may be caused by mRNA transport defects (14). To test whether dynein/dynactin directly plays a role in mbp mRNA transport, we turned to primary rodent oligodendrocyte cultures. Coimmunoprecipitations demonstrated that Mbp mRNA granules indeed associate with dynactin and dynein. Furthermore, imaging of Mbp mRNA granules in primary oligodendrocytes revealed movement in the anterograde direction, as expected; unexpectedly, the granules also move transiently backward in the retrograde direction. To test how dynein activity affects Mbp mRNA transport, we pharmacologically inhibited dynein using ciliobrevin in both zebrafish larvae and cell culture. In zebrafish, ciliobrevin treatment reduced mbp mRNA localization in oligodendrocyte processes; in primary oligodendrocytes, this resulted in arrested transport in both anterograde and retrograde directions and also dramatically reduced MBP protein levels. Taken together, our cellular and in vivo data demonstrate a previously unappreciated role for dynein/dynactin in anterograde Mbp mRNA transport in developing oligodendrocytes.

Results

actr10 Mutations in Zebrafish Cause Reduced mbp mRNA Levels.

To investigate the development of myelinated axons, we performed a large-scale N-ethyl-N-nitrosourea (ENU)-based forward genetic screen in which we visualized axons using the transgenic reporter tg(lhx1a:GFP) (23), then imaged myelin using mbp in situ hybridization (ISH). We uncovered a mutant, designated stl83, with both GFP-positive axonal swellings (Fig. 1 A and B) and reduced levels of mbp by ISH (Fig. 1 C and D). Whole genome sequencing of equimolar pools of mutants and siblings and subsequent analysis using an in-house pipeline (24) were used to determine that the gene of interest likely resided on chromosome 20 (Fig. 1E). Further single-nucleotide polymorphism (SNP) subtraction analysis to compare these data to other whole genome sequencing datasets reduced the number of candidate genes within our mapped locus to five, including one in actr10 (Fig. 1F). To test if the stl83 mutation was indeed in actr10, we performed a complementation test using a second actr10 mutant allele (actr10^nl15) (25). While the putative actr10^stl83 allele results from a G-to-T transversion in exon 12, resulting in a glycine to tryptophan amino acid change, the actr10^nl15 allele is a putative null mutation in the start site of actr10 (Fig. 1G and Fig. S1A). Complementation analysis confirmed that the stl83 phenotype results from a mutation in actr10, as transheterozygous animals have axonal swellings, which are never seen in either heterozygous or wild-type (WT) animals from either genotype (Fig. 1 H and I). Although the actr10^stl83 mutation occurs late in the genomic sequence, both mutants are lethal at the larval stage. The actr10^stl83 mutants are grossly healthier than the actr10^nl15 mutants (Fig. S1D), indicating that actr10^stl83 may be a hypomorphic allele.

Fig. 1.

A forward genetic screen uncovers actr10 mutants. (A) tg(lhx1a:gfp) marks axons in the tail of a control zebrafish larva at 4 dpf. (B) stl83 mutants exhibit axonal swellings in the CNS (arrows). (C) mbp ISH in a control animal at 5 dpf shows robust mbp mRNA levels. (D) An stl83 mutant animal exhibits reduced mbp mRNA in the hindbrain (arrow) and spinal cord (arrowheads). (E) Analysis of whole genome sequencing data revealed a higher mutant-to-WT allele ratio on chromosome 20, and SNP subtraction analysis (F) narrowed the number of candidate genes to five. (G) Diagram of actr10 genomic structure shows the location of the nl15 mutation in exon 1 and the stl83 mutation in exon 12. (H) In a tg(neurod:egfp) background, the lateral line is normal in the tail of control animals at 2 dpf. (I) A complementation cross demonstrated that transheterozygous stl83/nl15 animals have axonal swellings in the PNS (actr10^stl83/nl15: n = 22) (arrows), which are never seen in either heterozygous or WT animals from either genotype (actr10^+/+: n = 47; actr10^stl83/+: n = 13; actr10^nl15/+: n = 9). A, B, H, and I, lateral views, dorsal up; C and D, dorsal views, anterior up. (Magnification: A and B, 80×; C and D, 50×; H and I, 160×.)

Fig. S1.

nl15 and stl83 are alleles of actr10 (related to Figs. 1 and 2). (A) The nl15 allele is the result of a mutation in the start site (T to G), while the stl83 allele results from a G-to-T transition. (B) The Actr10/Arp11 amino acid sequence is highly conserved across species, and the glycine altered in the stl83 allele is completely conserved from zebrafish to human (black rectangle). (C) A dCAPS genotyping assay for stl83 inserts an AvaI restriction site in WT, leading to a 224-bp major band in WT used for genotyping (arrowhead) and a minor fragment of 22 bp. Mutants have a 244-bp fragment (asterisk). The nl15 mutation introduces an HaeIII site in mutants resulting in a major band at 112 bp (asterisk) and shorter bands at 50 bp and 14 bp, while WT has a major band at 126 bp (arrowhead) and a minor band of 50 bp. (D) Neither stl83 nor nl15 mutants develop a swim bladder. However, stl83 mutants appear generally healthier than nl15 mutants at 6 dpf. (Magnification: 20×.) (E) There is no significant difference in number of myelinated axons in the ventral spinal cord between sibling actr10^+/+ (n = 3) and actr10^stl83/+ (n = 4) animals (P < 0.61). (F) Similarly, there is no difference in number of myelinated axons between sibling actr10^+/+ (n = 3) and actr10^nl15/+ (n = 3) animals. Unpaired t test with Welch’s correction used for statistical analyses. N.S., nonsignificant.

actr10 Mutants Have Fewer Myelinated Axons.

To directly observe the ultrastructure of the myelin sheath, we performed TEM of both actr10 mutant alleles and counted the number of myelinated axons in a hemisegment of the ventral spinal cord. Of note, heterozygous actr10^stl83/+ and actr10^nl15/+ animals are viable and have no discernable phenotype compared with WT siblings as assessed by tg(neurod:egfp) from the complementation cross and number of myelinated axons by TEM (Fig. S1 E and F) and are therefore combined with WT as “controls.” Consistent with our mbp ISH analysis, both actr10 mutant alleles have significantly fewer myelinated axons compared with control animals by TEM at 5 d postfertilization (dpf) (Fig. 2), demonstrating that dynein/dynactin dysfunction has functional consequences for myelin in the CNS.

Fig. 2.

actr10 mutants have fewer myelinated axons in the ventral spinal cord. (A) Myelinated axons in a control larva in a hemisegment of the ventral spinal cord at 5 dpf. (B) Pseudocoloring and higher magnification enhance visualization of control myelinated axons. (C and D) Myelinated axons are similarly shown in a quadrant of the ventral spinal cord in an actr10^nl15/nl15 mutant animal. (E and F) Myelinated axons in a control larva in a quadrant of the ventral spinal cord at 5 dpf. (G and H) Myelinated axons are similarly shown in a quadrant of the ventral spinal cord in an actr10^stl83/stl83 mutant animal at 5 dpf. (I and J) Quantification revealed a significant difference in the number of myelinated axons in a ventral quadrant between control animals (n = 6) and actr10^nl15/nl15 mutants (n = 4; ***P < 0.0006) (I) and between control animals (n = 8) and actr10^stl83/stl83 mutants (n = 5; ****P < 0.0001) (J) at 5 dpf. Unpaired t tests with Welch’s correction were used for statistical analysis.

actr10^nl15/nl15 Mutants Have Fewer OPCs.

One explanation for reduced myelinated axon numbers in actr10 mutants could be fewer OPCs. To test this, we examined spinal cord cross-sections from double transgenic tg(olig2:dsred);tg(sox10:mgfp) (26, 27) zebrafish to quantify OPC number in putative null actr10^nl15/nl15 mutants. olig2 is expressed in OPCs and motor neurons at 3 dpf, sox10 is expressed in OPCs, and the tg(olig2:dsred);tg(sox10:mgfp) transgenes combined allow for visualization of OPCs with red cytoplasm and green membrane. At 3 dpf, actr10^nl15/nl15 mutants had fewer OPCs compared with controls (Fig. 3 A–C), indicating a role for actr10 in OPC development.

Fig. 3.

actr10^nl15/nl15 mutants have fewer OPCs. (A and B) Double transgenic tg(sox10:mgfp);tg(olig2:dsred) larvae were used to identify OPCs in spinal cord cross-sections at 3 dpf. Arrowheads mark OPCs, which are labeled by cytoplasmic DsRed and membrane-tagged GFP. (C) Mutants (n = 7) have significantly fewer OPCs in the spinal cord compared with sibling controls (n = 16; ***P < 0.0002). (D–G) The tg(olig2:dsred) line was used to count the number of olig2-labeled elongated cells in control animals (D and D′, arrows) compared with actr10^nl15/nl15 mutants (E and E′, arrows) at 3 dpf and at 4 dpf (pictured). (F) Quantification revealed a significant difference between the number of olig2-labeled dorsal cells in controls (n = 48) compared with actr10^nl15/nl15 mutants (n = 16; **P < 0.007) at 3 dpf. (G) Similarly, there was also a significant difference between controls (n = 28) and actr10^nl15/nl15 mutants (n = 16; ****P < 0.0001) at 4 dpf. Unpaired t tests with Welch’s correction were used for statistical analysis.

In the zebrafish spinal cord, OPCs originate ventrally and then migrate dorsally on either side of the neuronal cell bodies located in the center of the spinal cord (28). A population of OPCs remains in the ventral cord to differentiate and myelinate ventral axons, while the dorsally migrated OPCs differentiate and myelinate axons in the dorsal spinal cord. Dorsal OPCs can be visualized using the tg(olig2:dsred) transgene and are easily distinguished by their elongated cell bodies (Fig. 3 D–E') (28). Using this elongated cellular morphology as a metric, we found that the number of DsRed-positive elongated cells in the dorsal spinal cord was reduced at 3 dpf in actr10^nl15/nl15 mutants compared with controls (Fig. 3F). To determine whether this resulted from a delay in migration and maturation, we similarly assayed 4 dpf animals and also found a significant reduction at this later time point (Fig. 3G). Together, these data demonstrate that actr10^nl15/nl15 mutants have reduced numbers of OPCs and fewer dorsally migrated OPCs, recapitulating early defects described in zebrafish dynein mutants (16). Importantly, however, some maturing dorsal OPCs are present in actr10^nl15/nl15 mutants, indicating that migration defects alone are not the sole cause of the myelination phenotypes observed in mutants.

OPC Proliferation Is Slower in actr10^nl15/nl15 Mutants.

Interestingly, depletion of Arp11 in mammalian COS7 resulted in the striking formation of multiple spindles during mitosis (29). To better understand the mechanistic role of Arp11 in OPC proliferation and cell division, we performed live imaging using the transgenic line tg(nkx2.2a:megfp), which also marks OPCs (1). Transgenic zebrafish were imaged for up to 12 h starting at ∼57 h postfertilization (hpf). During this time period in oligodendrocyte development, OPCs have been specified and continue to divide and migrate from the ventral to the dorsal spinal cord. They continuously extend and retract exploratory processes to sample the environment before becoming postmitotic and beginning to myelinate (1). Live imaging revealed that mutant nkx2.2a-labeled cells took longer to divide compared with cells in control sibling actr10^nl15/+ animals (Fig. 4 and Movies S1–S4). During division, the nkx2.2a-labeled cells retract processes and become distinctly round before dividing. In the control animals, the time from beginning of rounding until division or the end of live imaging takes an average of 46 min, compared with 159 min in mutants, suggesting that reduced OPC numbers are in part due to slower OPC proliferation.

Fig. 4.

Cell division is delayed in actr10^nl15/nl15 mutants. (A–E) Time-lapse imaging from ∼57 hpf using the tg(nkx2.2a:megfp) line to mark OPCs allows for visualization of individual nkx2.2a-labeled cells that retract processes and become round in preparation for division (arrowhead, A). In actr10^nl15/+ heterozygous control animals, the cell remains in this state for only two panels of 10 min (B and C) before becoming two cells (arrowheads, D) that divide rapidly (E). (F–R) In contrast, a cell from an actr10^nl15/nl15 mutant animal that has already retracted its processes (arrowhead, F) takes over 3 h to divide (G–R). (S) Quantification revealed a significant difference in either time to cell division or the end of imaging between heterozygous control animals (n = 11 cells from five animals) and mutants (n = 7 cells from six animals; *P < 0.02, unpaired t test with Welch’s correction). See also Movies S1–S4.

Dynein Activity Is Impaired in Multiple Tissues in actr10^nl15/nl15 Mutants.

Previous studies have found that loss of Arp11 disrupts the assembly of the dynactin complex in mammalian cell culture and in the filamentous fungi Aspergillus nidulans (29, 30). Actr10 has recently been shown to have a specific role in retrograde mitochondrial transport in peripheral axons without disrupting dynein/dynactin complex integrity (25). We hypothesized, however, that oligodendrocyte cell lineage and myelin phenotypes in actr10^nl15/nl15 mutants result from general disruption to dynein/dynactin function for several reasons. Firstly, mutants have defects in OPC development and myelination that are similar to those previously defined in zebrafish dynein mutants (16). Secondly, mutants phenocopy the photoreceptor loss observed in zebrafish with mutations in other subunits of dynactin, including p150^Glued and p50 (31–33) (Fig. S2 A and B). The photoreceptor defects observed indicated that the fish might be blind. To this end, we used the Noldus behavior unit, DanioVision, to assay the response of WT, actr10^nl15/+, and actr10^nl15/nl15 animals during light/dark cycles. Larval zebrafish, unlike their adult counterparts, avoid the dark and respond with increased movement (34, 35). The movement of larvae in a 96-well plate was tracked during an alternating light/dark cycle. Both WT (Fig. S2G) and actr10^nl15/+ animals (Het) (Fig. S2H) showed a significant increase in movement in the dark compared with the light. However, actr10^nl15/nl15 animals (Mut) showed no significant difference in movement between the light baseline and dark response (Fig. S2I), indicating that the observed photoreceptor death likely leads to blindness.

Fig. S2.

actr10^nl15/nl15 mutants have similar phenotypes to dynactin and dynein mutants (related to Fig. 6). (A and B) Photoreceptor layer analysis by DAPI staining at 3 dpf. Controls (n = 4) have a normal photoreceptor layer (A, arrowheads), while actr10^nl15/nl15 mutants (n = 3) have a disrupted photoreceptor layer (B, arrowheads). (C and D) Head pigment analysis at 5 dpf. WT controls have normal pigmentation (C), while actr10^nl15/nl15 mutants have expanded pigment (D, arrows). (E and F) Lateral line pigment analysis at 5 dpf. WT control animals have intermittent pigment along the lateral line (E), while actr10^nl15/nl15 mutants have increased pigment (F, arrows). (G–I) The Noldus DanioVision behavioral system was used to track larval movement at 6 dpf. WT (G, n = 23, ***P < 0.0002) and heterozygous (H, n = 36, ****P < 0.0001) controls showed a significant increase in movement in response to dark, while actr10^nl15/nl15 mutants did not (I, n = 26, P < 0.98). An unpaired t test with Welch’s correction was used for statistical analysis. (J–S) Epinephrine treatment to assay retrograde melanosome transport. Before epinephrine treatment at 3 dpf, control animals (n = 19) exhibit similar pigment patterns. After 5 min of epinephrine exposure (K), melanosomes have begun to move back toward the melanophore cell body via retrograde transport. After 10 (L), 15 (M), and 30 min (N), all control animals have small pigment patches representative of aggregated melanosomes. Before epinephrine treatment, actr10^nl15/nl15 mutants (n = 11) can be distinguished phenotypically from control animals by excessive pigment (O). After 5 (P), 10 (Q), 15 (R), and 30 (S) min of epinephrine treatment, melanosomes in mutants are still more dispersed relative to those in control animals. Note that representative animals were selected at random, and images may be different animals. (Magnification: C and D, 80×; E and F, 20×; J–S, 80×.)

Thirdly, actr10^nl15/nl15 animals phenocopy the excessive pigment observed in the dorsal head region of zebrafish dynein mutants (16) (Fig. S2 C and D) and also have increased pigment along the lateral line (Fig. S2 E and F). Interestingly, blindness has been shown to cause background adaptation in which zebrafish up-regulate pigment production in response to constant darkness (36). Thus, the increase in basal levels of pigmentation could be connected to photoreceptor loss. In addition, pigment cells, called melanophores, can change skin color in response to environmental stimuli. This is mediated by release of the hormone epinephrine, which triggers intracellular changes in protein kinase A activity that result in increased dynein-mediated retrograde run lengths, leading to melanosome aggregation toward the melanophore cell body (16, 37, 38). Thus, to further test whether dynein activation is perturbed in actr10^nl15/nl15 mutant melanophores, we performed a classic melanosome aggregation assay (Fig. S2 J–S). In control animals, 5 min of exposure to epinephrine induced noticeable melanosome aggregation toward the center of the cell (Fig. S2K), whereas actr10^nl15/nl15 mutants maintained relatively dispersed melanosomes (Fig. S2P). After 30 min, whereas melanosomes in all control animals were aggregated (Fig. S2N), mutants continued to exhibit widespread pigment distribution (Fig. S2S), indicating that activated retrograde melanosome transport is compromised in mutant melanophores.

Actr10 Function in Oligodendrocytes Can Partially Suppress Myelin Defects.

To understand the function of actr10 in oligodendrocytes, a construct in which the glial sox10 promoter drives expression of actr10 tagged with monomeric RFP (sox10:mRFP-actr10) was transiently expressed in larvae from a cross between tg(mbp:gfp-caax);actr10^nl15/+ and actr10^nl15/+ fish. sox10:mRFP-actr10–injected larvae (sox10-injected) exhibit RFP fluorescence, which is not seen in uninjected mutants (Fig. 5 A–B′′). TEM analysis and quantification at 5 dpf showed partial suppression of the mutant phenotype in the dorsal spinal cord (Fig. 5 C–E). Although sox10:mRFP-actr10–injected mutants were not rescued to control levels, sox10:mRFP-actr10–injected mutants possessed significantly more myelinated axons compared with uninjected mutant siblings, demonstrating that glial actr10 rescue can promote proper myelination in vivo (Fig. 5F).

Fig. 5.

Transient expression of actr10 in oligodendrocytes partially suppresses myelination defects in actr10^nl15/nl15 mutants. (A–B′′) Confocal images of mbp-labeled oligodendrocytes in uninjected (A) and sox10:mRFP-actr10 (“sox10 injected”) actr10^nl15/nl15 mutants (B). While uninjected actr10^nl15/nl15 mutants do not exhibit RFP fluorescence (A′ and A′′), injection of sox10:mRFP-actr10 results in monomeric RFP fluorescence in actr10^nl15/nl15 mutants (B′ and B′′). (C–E) TEM images show dorsal spinal cords of uninjected WT and actr10^nl15/+ controls (C), uninjected actr10^nl15/nl15 mutants (D), and sox10:mRFP-actr10–injected actr10^nl15/nl15 mutants (E). Myelinated axons are pseudocolored in purple. (F) Quantification shows that sox10-injected actr10^nl15/nl15 mutants (n = 6) have significantly greater numbers of myelinated axons in the dorsal spinal cord compared with uninjected actr10^nl15/nl15 mutant siblings (n = 5, P < 0.0377), although sox10:mRFP-actr10 injection does not restore myelination to WT control levels (n = 5, P < 0.0001), indicative of partial rescue. One-way ANOVA with Tukey’s multiple comparisons test was used for statistical analyses.

Analysis of a stable transgenic neuronal rescue line in which mRFP-actr10 is driven by the neuronal-specific neurod promoter tg(neurod:mRFP-actr10) demonstrates that Actr10 also functions in neurons to promote oligodendrocyte myelination (Fig. S3). Given the critical functions of dynein/dynactin in all cells, it is not surprising that Actr10 functions in both neurons and oligodendrocytes to promote myelination. Of note, injection of sox10:mRFP-actr10 resulted in lower levels of actr10 as assayed by RFP fluorescence than were seen in the tg(neurod:mRFP-actr10) stable line. Thus, one reason a neuronal rescue was more effective could be due to the differences between transient injection with weak oligodendrocyte actr10 expression and a stable transgenic line with strong actr10 expression. The function of Actr10 in neurons to promote myelination forms the basis of future work; here, we focus on the cell-autonomous function of Actr10 in oligodendrocytes.

Fig. S3.

Stable expression of actr10 in neurons suppresses myelination defects in actr10^nl15/nl15 mutants (related to Fig. 5). (A–D) Fluorescent micrographs showing tg(mbp:gfp-caax) and tg(neurod:mRFP-actr10) expression in zebrafish of the indicated genotypes harboring the indicated transgenes at 5 dpf. (A) The ventral and dorsal spinal cords are normal in a WT control animal lacking the neurod:mRFP-actr10 transgene. (B) Large caliber swellings (arrows) are present in an actr10^nl15/nl15 mutant lacking the neurod:mRFP-actr10 transgene. (C and D) Swellings are not observed in a WT control (C) and are ameliorated in actr10^nl15/nl15 mutants harboring the neurod:mRFP-actr10 transgene (n = 15/17). (E–G) TEM images of the dorsal spinal cord show myelinated axons (pseudocolored purple) in WT control animals (E) (n = 4), actr10^nl15/nl15 mutants lacking the stable neurod:mRFP-actr10 transgene (F) (n = 3), and actr10^nl15/nl15 mutants harboring the stable neurod:mRFP-actr10 transgene (G) (n = 4). (H) Quantification shows that stable neuronal expression of actr10 suppresses actr10^nl15/nl15 mutant myelination defects to WT levels (P < 0.99, N.S.), while there is a significant difference between control and actr10^nl15/nl15 mutants (*P < 0.018) and actr10^nl15/nl15 mutants and actr10^nl15/nl15 mutants with the neurod:mRFP-actr10 transgene (*P < 0.022).

actr10^nl15/nl15 Mutants Have Reduced mbp mRNA Localization in Oligodendrocyte Processes.

All actr10^nl15/nl15 mutants observed have reduced mbp mRNA localization to processes at 4 dpf (Fig. 6 A–F). This phenotype cannot be explained by OPC proliferation defects alone, as the observation of dorsal elongating cells in mutants (Fig. 3) indicates that some oligodendrocytes are indeed migrating and maturing. Moreover, this phenotype is reminiscent of the kinesin kif1b zebrafish mutant in which disruption of anterograde transport results in accumulation of mbp mRNA in oligodendrocyte cell bodies (14). This led us to hypothesize that dynein/dynactin may play a role in efficient transport of mbp mRNA to distal oligodendrocyte processes.

Fig. 6.

Dynein and dynactin are associated with Mbp mRNA granules. (A) ISH shows robust levels of mbp mRNA in a representative control larva at 4 dpf, and higher magnification (B) shows mbp processes in the control (arrowheads) (n = 47/47). In contrast, an actr10^nl15/nl15 mutant (C) has a punctate mbp phenotype, and higher magnification of the same image (D) shows mbp-positive cell bodies (arrows) but no processes (n = 18/18). (E) A lateral view of the brain and spinal cord of a WT animal shows mbp mRNA in processes (arrowheads), while actr10^nl15/nl15 mutant animals (F) have cell bodies (arrows) but reduced mbp mRNA-bearing processes. (G) A bidirectional construct expressing MS2-GFP, a RNA-binding reporter, as well as MBP 5′UTR, CDS, and 3′UTR tagged with MS2 stem loops was electroporated into purified rat oligodendrocytes. MS2-GFP binds to stem loops to allow visualization of Mbp mRNA motility. (H) A primary rat oligodendrocyte expressing MS2-GFP–labeled Mbp mRNA imaged using spinning-disk confocal microscopy shows distribution of Mbp mRNA throughout the oligodendrocyte processes. (I) Lysates from primary rat oligodendrocytes expressing the MS2-GFP–labeled Mbp mRNA construct were immunoprecipitated using an anti-GFP antibody and probed with p150^Glued, DIC (dynein intermediate chain), and Actr10/Arp11 antibodies (n = 4 independent experiments).

Dynein and Dynactin Associate with Mbp mRNA Granules.

To visualize Mbp mRNA transport, we expressed the MS2 reporter in primary rat oligodendrocyte cultures that were purified using the immunopanning technique and are free of neuronal contamination (39). MS2 is an RNA-binding protein derived from bacteriophage that binds with high affinity to specific RNA sequences that form stem loop structures (40). Using a bidirectional construct, we coexpressed two transcripts: (i) Mbp containing 5′UTR, CDS, and 3′UTR followed by MS2-binding stem loops and (ii) a GFP-tagged MS2 reporter (Fig. 6G). In oligodendrocytes differentiated in culture for 4 d, MS2-GFP–positive puncta associated with Mbp mRNA are distributed throughout the many processes that emanate outward from the cell body (Fig. 6H), while in control cells not expressing stem loops, MS2-GFP is retained in the cell body (Fig. S4A).

Fig. S4.

Coimmunoprecipitation pulls out p135 in addition to p150^Glued (related to Figs. 6 and 7). (A) Image of a cell expressing a control construct for MS2-GFP, which contains a nuclear localization sequence (NLS). In the absence of any stem loop containing constructs, such as MBP, MS2-GFP localizes to the nucleus. (B) Identity of the p135 band from an anti-GFP immunoprecipitation against MS2-GFP–associated MBP mRNA granules was confirmed by comparing it to the eluted band from an immunoprecipitation using an antibody targeted against the N terminus of p150^Glued, which contains its microtubule-binding domain. (C) Ribbon diagram of DCTN1 isoform 1 (p150^Glued) shows a CAP-Gly microtubule-binding domain, which is lacking in DCTN1 isoform 2 (p135). (D) Kymograph analysis shows rapid back-and-forth movement of Mbp mRNA granules.

We can use this system to biochemically validate the association of the dynein motor and its adaptor dynactin to Mbp mRNA granules. Indeed, immunoprecipitations using a GFP antibody to isolate MS2-GFP–tagged Mbp mRNA granules also pulls out the dynactin subunit p150^Glued, dynein intermediate chain (DIC), and Arp11 (Fig. 6I). Eluates more robustly coimmunoprecipitated p150^Glued than DIC. Interestingly, two bands are visible using the p150^Glued antibody, which indicates that the lower band is p135, a shorter splice isoform of p150^Glued that does not contain the N-terminal microtubule-binding domain (41). This was confirmed by comparing the p135 band in this anti-GFP immunoprecipitation to another immunoprecipitation using an antibody against the N terminus of p150^Glued that preferentially isolates full-length p150^Glued but not p135 (Fig. S4 B and C). Thus, we have now demonstrated that the p135 isoform is expressed in oligodendrocytes and can associate with mRNA granules.

Acute Dynein Inhibition Arrests Mbp mRNA Transport in both Anterograde and Retrograde Directions.

Live-cell imaging in this system using spinning-disk confocal microscopy reveals that processive Mbp mRNA transport occurs in both anterograde and retrograde directions with average speeds of ∼0.21 μm/s and 0.16 μm/s, respectively (Fig. 7 A and B and Movie S5), consistent with previously measured speeds for Mbp mRNA (4). To investigate the functional contribution of dynein in Mbp mRNA transport in oligodendrocytes, we pharmacologically inhibited dynein activity. The small-molecule cell-permeable dynein inhibitor ciliobrevin D works by disrupting the ATPase activity of the dynein motor and has a half-minimum inhibitory concentration (IC₅₀) ∼5–15 μM in the cilia (42). Cells acutely treated with 15 μM ciliobrevin were imaged for 60-s durations for up to an hour with no observed morphology changes or toxicity. As early as 4 min after ciliobrevin treatment, Mbp mRNA net speed decreases in both anterograde and retrograde directions (Fig. 7 A and B; n = 20 cells from two biological replicates). At around 10–15 min after ciliobrevin treatment, Mbp mRNA transport was arrested almost completely in both directions. When speeds are binned in 7-min increments, this decrease in speed is significantly different compared with untreated cells (Fig. 7 C and D).

Fig. 7.

Acute dynein inhibition arrests both anterograde and retrograde Mbp mRNA transport in cultured oligodendrocytes and perturbs mbp localization in zebrafish. (A) In kymographs representing 60 s of live-cell imaging, MS2-GFP–labeled Mbp mRNA can be seen moving in the retrograde direction in untreated cells (Top). Following acute ciliobrevin D (15 μM) addition, Mbp mRNA displays bidirectional motility characterized by frequent back-and-forth movement and many directional switches (Middle). Finally, around 15 min after ciliobrevin D (15 μM) addition, most cells display arrested motility (Bottom). (B) A scatter plot represents the average anterograde or retrograde net speeds for individual oligodendrocytes. (C and D) Anterograde or retrograde net speeds were averaged for each cell and binned across 27-min time periods following ciliobrevin treatment. At 8–15 min following ciliobrevin treatment, speeds significantly decreased compared with earlier time points. One-way ANOVAs with post hoc Tukey’s test were performed. Anterograde speeds: No CilioD vs. 1–7 min, P = 0.11; No CilioD vs. 8–15 min, **P = 0.0011; 1–7 min vs. 8–15 min, P = 0.077. Retrograde speeds: No CilioD vs. 1–7 min, P = 0.67; No CilioD vs. 8–15 min, P = 0.11, 1–7 min vs. 8–15 min, *P = 0.027. (E) A model shows that Mbp mRNA granules can move processively in the anterograde and retrograde directions as well as bidirectionally and that each Mbp mRNA granule can simultaneously bind to both kinesin and the dynein/dynactin complex. (F–I) mbp ISH of zebrafish larvae treated with ciliobrevin D for 21 h. Larvae were scored as having strong (F), normal (G), reduced (H), or absent (I) mbp localization in oligodendrocyte processes (arrowheads, processes; arrows, cell bodies). (J) Quantification of phenotypic distribution shows that there was a significant difference in scores for DMSO (n = 71) and ciliobrevin D-treated larvae (n = 68), with the latter exhibiting more larvae with reduced or absent mbp localization in oligodendrocyte processes (P < 4.5E-14, Fisher’s exact test). (Scale bar, 50 μM.)

Using faster frame rates to interrogate these possibilities, we saw as late as 11 min after ciliobrevin treatment Mbp mRNA that is engaged in bidirectional motility characterized by many frequent back-and-forth movements and directional switches (Fig. 7A, Middle). After ciliobrevin treatment, we also observe Mbp mRNA granules with zero net motility that are engaged in very rapid and frequent directional switches for durations of up to 1 min (Fig. S4D). This may represent a tug-of-war state during which anterograde and retrograde motor forces are roughly balanced and suggests that a single Mbp mRNA granule can associate simultaneously with both anterograde kinesin motors as well as retrograde dynein motors (Fig. 7E).

To confirm our findings in vivo, we treated WT zebrafish with ciliobrevin D for 21 h starting ∼3.5 dpf. Importantly, treating at this time point permits normal OPC development so that we can specifically examine the role of dynein in mbp mRNA transport. Although there was 15% mortality overall, importantly, the ciliobrevin-treated fish that were fixed at 4.5 dpf appeared grossly normal compared with vehicle-treated fish (Fig. S5 A and B). ISH for mbp mRNA revealed that ciliobrevin-treated fish had a significant reduction in mbp mRNA localized to oligodendrocyte processes compared with vehicle-treated controls (Fig. 7 F–J). Taken together, our in vitro and in vivo results show that dynein inhibition leads to reduced mbp mRNA distribution.

Fig. S5.

Effects of ciliobrevin D treatment on zebrafish and cultured oligodendrocytes (related to Fig. 7). (A and B) DMSO and ciliobrevin D-treated fish at 4.5 dpf. Compared with fish treated with DMSO for 21 h (A), ciliobrevin D-treated fish look relatively normal. (Magnification: A and B, 16×.) (C) Treatment of oligodendrocytes differentiated for 3 d in vitro with different concentrations of the dynein inhibitor ciliobrevin for 21 h. Cell death was assayed by morphology using DIC microscopy; living cells contained continuous, nonbulbous branches and adhered to coverslips. Concentrations of 1 μM and 5 μM did not significantly alter the percentage of living cells, while 10 μM and 15 μM significantly decreased the percentage of living cells (n = 4 coverslips per group; *P = 0.023, ***P = 0.0001, one-way ANOVA).

Chronic Dynein Inhibition Disrupts MBP Protein Translation in Cultured Oligodendrocytes and Disrupts Myelination in Zebrafish.

To understand how Mbp mRNA transport inhibition can lead to myelination defects, we asked whether Mbp mRNA transport is necessary for local MBP protein translation. We hypothesized that proper translocation of Mbp mRNA granules precedes local translation and recruitment of ribosomes (Fig. 8A). To test this, we treated cultured rat oligodendrocytes for 21 h with the dynein inhibitor ciliobrevin, then visualized Mbp mRNA localization using single molecule FISH (fluorescent ISH) and simultaneously immunostained for MBP protein. Previously, oligodendrocytes were treated acutely with 15 μM ciliobrevin for less than 1 h for live-cell imaging. However, chronic treatment for 21 h at this concentration resulted in some toxicity compared with control and vehicle-treated conditions (Fig. S5C). Thus, cells that had been differentiated for 3 d in vitro were treated for 21 h at the nontoxic concentration of 5 μM ciliobrevin (Fig. S5C), which had been previously used for overnight dynein inhibition in neurons (43).

Fig. 8.

Chronic dynein inhibition disrupts MBP protein translation in cultured oligodendrocytes and disrupts myelination in zebrafish. (A) A model shows that transport of Mbp mRNA granules by microtubule-based motors precedes local translation and recruitment of ribosomes. (B) Quantification of the percentage of DIV3 oligodendrocytes treated for 21 h with 5 μM ciliobrevin that express MBP protein demonstrate that a lower percentage of ciliobrevin-treated cells (42%) are positive for MBP protein compared with control cells (63%, P < 0.001) and DMSO-treated cells (62%, P < 0.001). Of these MBP-containing cells, only 8.5% of ciliobrevin-treated cells highly expressed MBP protein compared with 29% (P < 0.001) and 27% (P < 0.001) in control and DMSO-treated cells. (C) Cells treated with 5 μM ciliobrevin for 21 h were costained with single-molecule FISH probes against Mbp mRNA and with a monoclonal antibody against MBP protein. MBP protein images (Middle) show examples of high expressers (arrows) in control cells and low expressers (arrowheads) in ciliobrevin-treated cells (n = 2 biological replicates using primary cultures from two different animals, four coverslips, 23–32 fields of view per group).

Ciliobrevin-treated cells had remarkably decreased Mbp mRNA distribution and less MBP protein (Fig. 8 B and C). Whereas Mbp mRNA granules in control cells were dispersed throughout the oligodendrocyte processes as distally as the cell periphery, Mbp mRNA granules in ciliobrevin-treated cells were restricted to the cell body and to proximal processes. Consistent with live-cell imaging results from acute ciliobrevin treatment, this chronic 21-h treatment also demonstrates that dynein inhibition disrupts distal Mbp mRNA localization. Quantification of MBP protein staining revealed that only 42% of ciliobrevin-treated cells contained MBP protein, compared with 63% and 62% of control and vehicle-treated cells, respectively. These MBP-containing cells can be further binned as low expressers that have punctate MBP protein localization and high expressers that display pancake-like widespread MBP protein distribution. Binning demonstrates that while 29% and 27% of control and vehicle-treated cells are high expressers, only 8.5% of ciliobrevin-treated cells are high expressers. These data indicate that dynein inhibition leads to inefficient MBP translation.

Discussion

Myelination by oligodendrocytes is critical for proper function of the nervous system. In a forward genetic screen in zebrafish, we identified the dynactin subunit Actr10 as a regulator of myelination. actr10 mutants are lethal and display signs of impeded dynein loss of function in some tissues, including blindness due to photoreceptor loss and aberrant pigment cells with defects in melanosome motility. In the nervous system, mutants have fewer myelinated axons by TEM and fewer OPCs in part due to proliferation defects, phenocopying a previously described dynein zebrafish mutant (16). In addition, mutants exhibit less mbp mRNA in oligodendrocyte processes, similar to previous observations made in kinesin kif1b zebrafish mutants (14). Thus, we propose that proper function of the retrograde motor dynein and its adaptor dynactin is paradoxically required for anterograde transport of mbp mRNA. Indeed, we demonstrate in purified rodent oligodendrocyte cultures that acute pharmacological inhibition of dynein activity arrests Mbp mRNA transport in both anterograde and retrograde directions. Moreover, ciliobrevin treatment also decreases MBP protein levels, suggesting that mRNA transport is necessary for local translation of MBP. We confirmed these results in vivo in ciliobrevin-treated zebrafish, in which mbp mRNA distribution is perturbed in oligodendrocytes.

The combination of our in vivo and cellular data suggests that dynein/dynactin activity is necessary for anterograde Mbp mRNA transport. In zebrafish, both Actr10/Arp11 loss of function and chronic dynein inhibition prevent proper anterograde distribution of mbp mRNA in oligodendrocytes. In cultured oligodendrocytes, acute dynein inhibition disrupts Mbp mRNA transport in both anterograde and retrograde directions. Our transport results are consistent with ciliobrevin treatment of cultured dorsal root ganglion neurons, which display arrest of mitochondrial and lysosomal axonal transport in both anterograde and retrograde directions (43, 44). Furthermore, our observations are consistent with earlier experiments in Drosophila macrophage-like S2 cells that demonstrate dynein knockdown leads to arrested transport in both anterograde and retrograde directions of fluorescently tagged RNA-binding protein FMRP (22).

Our results also demonstrate that mRNA granules are highly heterogeneous, both in motility and in composition. We observe Mbp mRNA granules that move quickly with average net speeds ∼0.21 and 0.16 μm/s in anterograde and retrograde directions, respectively; we also observe Mbp mRNA granules that have very little net motility and move bidirectionally with frequent directional switches. Each bidirectional mRNA granule likely associates with multiple opposing motors that are engaged in a “tug of war” in which opposing motor forces are nearly balanced. Since one kinesin has 7 pN of unitary force while one dynein only has 1 pN of unitary force, this suggests that a bidirectional mRNA granule likely associates with at least 1 kinesin and 6–8 dyneins. Thus, our data support a model whereby each Mbp mRNA granule simultaneously associates with kinesin and dynein/dynactin and the full assembly of opposing motors may be required to sustain transport (Fig. 7E). Though it is unclear why both kinesins and dyneins are simultaneously associated with individual cargos, one possibility is that the ability to transiently step back and forth may allow more flexibility in circumnavigating roadblocks or switching microtubules (45). This may be particularly important in the crowded cytoskeletal environment of the oligodendrocyte, in which microtubules penetrate many concentric layers of the dense myelin sheath (46).

Moreover, association of different dynactin isoforms with mRNA granules may play a role in regulating Mbp mRNA transport as well. Coimmunoprecipitations revealed that Mbp mRNA granules associate with not only p150^Glued but also its shorter isoform p135, which is unable to bind to microtubules (41). Though a human neuronal cell line was used for the original characterization of p135, we now show that p135 is also present in primary oligodendrocytes. Though it is still unclear what functional role p135 may play in transport regulation, a classic in vivo labeling study in the optic nerve demonstrated that p150^Glued-associated cargos traveled much faster than p135-associated cargos, arriving past the optic chiasm at 2 d and 4 d after initial labeling, respectively (47).

These data highlight that there is much more to learn about the composition of vertebrate mRNA granules, including how many motors are associated and what adaptor proteins link these motors to mRNAs. In neuronal axons, the RNA-binding protein La is transported in the anterograde direction in its native form and in the retrograde direction in its sumoylated form; sumoylation is required for dynein association to La and may act as a mechanism to return mRNA-free La protein to the nucleus by decreasing its ability to bind to mRNA (48). Furthermore, transport of mRNA in neurons can be stimulated in response to neuronal activity (49, 50), and in oligodendrocytes, activation of Fyn kinase has been shown to trigger MBP translation (6–8). Further research on regulation at the adaptor level may yield clues on how environmental stimuli can trigger Mbp mRNA transport in oligodendrocytes.

Together, our results suggest that actr10 mutants display phenotypes similar to dynein/dynactin loss of function in photoreceptors, melanophores, and oligodendrocytes. Interestingly, recently identified human patients with mutations in dynein and dynactin primarily present with sensory and motor neuron symptoms and are diagnosed with Charcot–Marie–Tooth distal neuropathy or spinal muscular atrophy (51–60). Thus, in both humans and zebrafish, the nervous system is exquisitely sensitive to mutations in dynein and dynactin. Though a link between myelination and dynein/dynactin mutations has not been established in humans, our work in zebrafish and mammalian oligodendrocytes demonstrates that dynein/dynactin function in Mbp mRNA transport is critical for proper myelination in vivo.

Materials and Methods

Zebrafish Lines and Rearing Conditions.

All zebrafish work was approved by the Washington University IRB and was performed in compliance with Washington University institutional animal protocols (animal protocol no. 20160174; Institutional Biological & Chemical (IBC) Safety Committee protocol no. 10-643), and zebrafish were housed in the Washington University zebrafish facility (zebrafish.wustl.edu). Embryos were raised in incubators at 28.5 °C in egg water (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄). ENU mutagenesis was performed according to standard protocols (61), and larvae were screened at 5 dpf for myelin abnormalities by two transgenes, tg(lhx1a:gfp) and tg(mbp:mCherry), as well as by ISH for mbp mRNA levels. The following transgenic and mutant zebrafish lines were used: SAT, SJD (62), actr10^stl83, actr10^nl15 (25), tg(lhx1a:gfp) (23), tg(mbp:mCherry) (a kind gift from Dave Lyons, University of Edinburgh, Edinburgh), tg(neurod:egfp) (63), tg(olig2:dsred) (26), tg(sox10:mgfp) (27), tg(nkx2.2a:megpf) (1), and tg(mbp:gfp-caax) (64).

Whole Genome Sequencing and Mapping.

The F2 generation of stl83 heterozygotes was outcrossed to SJD and raised. These animals were intercrossed, and F4 progeny were screened for axonal swellings. DNA was extracted from phenotypically WT and mutant larvae at 5 dpf and sent to the Genome Technology Access Center (GTAC) at Washington University for whole genome sequencing. An in-house analysis pipeline (24) was used to determine that the mutant-to-WT allele ratio was highest on chromosome 20. A SNP subtraction analysis using other whole genome sequencing datasets narrowed the number of possible mutations down to five. The gene was confirmed by a complementation test using the actr10^nl15 allele (25).

Genotyping.

A derived cleaved amplified polymorphic sequences (dCAPS) assay was developed to genotype the actr10^stl83 allele (helix.wustl.edu/dcaps/dcaps.html). An A-to-C mismatch is introduced in the forward primer (forward primer: 5′-GTCAGAGAACTTGCTGATACTC-3′; reverse primer: 5′-AGCAGTTCAGGGCAGCTTTA-3′). This causes an AvaI restriction digest site in the WT haplotype, and digestion generates two visible WT products of 22 and 224 bp and a mutant product of 244 bp. To genotype the actr10^nl15 allele, PCR was used to amplify the first exon (forward primer: 5′-ACCCAGCCGTTCCTCTAATG-3′; reverse primer: 5′-CCGCTCCTAAATCAATCACC-3′). Mutagenesis introduced a T-to-G mutation in the start site. This inserts an HaeIII site, and digestion results in two WT fragments of 126 and 50 bp and three mutant fragments of 112, 50, and 14 bp.

ISH.

In brief, embryos were treated with 0.003% phenylthiourea (PTU) at 24 hpf to prevent pigmentation. Embryos were reared in egg water with PTU until the desired age and fixed in 4% paraformaldehyde overnight at 4 °C in 1.5 mL Eppendorf tubes with ∼20 embryos per tube. A standard ISH protocol was then used (65), and all phenotypes were scored blindly.

TEM.

TEM was done according to standard protocols (66). More detail can be found in SI Materials and Methods.

Marker Analysis Using Transgenes.

Two transgenes, tg(olig2:dsred) and tg(sox10:mgfp), were crossed into an actr10^nl15/+ background and grown to adulthood. Animals were intercrossed to produce double transgene labeling, screened for double fluorescence, and fixed at 3 dpf for OPC cross-section analysis. Single transgene tg(olig2:dsred) animals were used to count the number of dorsal, elongated olig2-labeled cells at 3 and 4 dpf. For details, see SI Materials and Methods.

Time-Lapse Imaging.

tg(nkx2.2a:megfp);actr10^nl15/+ and actr10^nl15/+ animals were crossed, and live imaging was performed for up to 12 h beginning at 57 hpf on a Zeiss LSM 880 confocal microscope. For details, see SI Materials and Methods.

Melanosome Aggregation.

Control and actr10^nl15/nl15 mutants were separated at 3 dpf by pigment phenotype. Larvae were treated with 0.5 mg/mL epinephrine (Sigma-Aldrich) dissolved in egg water. One representative animal from control and mutant groups was selected at random, treated with tricaine, mounted on an agarose mold, imaged, and returned to the epinephrine treatment. This was repeated approximately every 5 min, from 5 to 15 min, and again at 30 min. After 30 min, larvae were collected and genotyped.

Noldus DanioVision Assay.

The Noldus DanioVision system was used to determine whether mutants were blind. Larvae were placed individually in a 96-well plate, and the following protocol was used: light on (baseline) 0–4 min, light off (dark response) 4–8 min, light on (recovery) 8–12 min. The average distance moved from the center point during the baseline, and dark response was calculated and analyzed for statistical significance. Two technical replicates were performed.

Primary Rat Oligodendrocyte Culture and Electroporation.

All rodent work was approved by the Stanford University IRB and was performed in compliance with Stanford University institutional animal protocols (animal protocol no. 10726). OPCs were purified from enzymatically dissociated P6–P8 Sprague–Dawley (Charles River) rat brain cortices by immunopanning and differentiated in serum-free defined medium containing T3, as previously described (39). Cells were electroporated using Amaxa Nucleofector (Program O-17 for rat oligodendrocytes; Lonza) with a bicistronic construct (pBI; Clontech) simultaneously expressing MBP 5′UTR, CDS, 3′UTR, and MS2-binding stem loops (24x) as well as MS2-GFP (40, 67).

Live-Cell Imaging and Ciliobrevin D Treatment.

Live-cell imaging was performed at the Stanford Cell Sciences Imaging Facility on a Nikon spinning disk confocal (TiE inverted microscope body equipped with Perfect Focus mechanism, Yokogawa spinning disk) with an Andor EMCCD camera inside an environmental chamber (37 °C, CO₂). Cells were treated with 15 μM ciliobrevin D (Calbiochem), and images were acquired at 1–3 frames per second for 60 s for a different cell at each time point to minimize photobleaching. Kymographs were generated from these movies using the Multiple Kymographs plug-in (FIJI, NIH) with a line width of 3 pixels. Net speed was calculated from the net duration and net distance of each motile event and represented as the average net speed for each cell. n = 2 biological replicates (two OPC preps from two different animals), 10 imaging plates, 20 cells. WT zebrafish at 3.5 dpf were separated into six-well plates with 20 larvae per well. Larvae were treated with either 2.5 μM ciliobrevin D (Calbiochem) or DMSO vehicle in egg water for 21 h. Larvae were fixed in 4% paraformaldehyde for ISH after 21 h and scored blindly for localization of mbp mRNA in oligodendrocyte processes.

Coimmunoprecipitations.

We lysed 1–4 million primary oligodendrocytes per condition in HEM buffer (50 mM Hepes, 1 mM EDTA, 1 mM MgCl₂) containing 25 mM NaCl, 0.5% Triton X-100, and protease inhibitors (P8340; Sigma). Cell lysates were incubated with Protein-G Dynabeads (Invitrogen), and coimmunoprecipitations were performed following the manufacturer’s instructions using a monoclonal antibody against GFP (JL-8; Clontech). Western blots were probed with antibodies against DIC (Chemicon), p150^Glued (610474; BD Transduction), and Actr10 (rabbit polyclonal, 20101–1-AP; Proteintech).

Single-Molecule FISH.

Single-molecule FISH probes were designed against Mbp mRNA CDS (Stellaris; LGC Biosearch Technologies). Cells were stained according to the manufacturer’s protocol. Briefly, cells were fixed in 4% paraformaldehyde, permeabilized in 70% ethanol for 1 h at 4 °C, hybridized with single-molecule FISH probes and primary antibody against MBP (ab7349; Abcam) for 4–16 h at 37 °C, washed for 30 min at 37 °C, and then stained with secondary antibody.

Statistical Analysis.

GraphPad Prism 7 and RStudio were used for statistical analysis. Unpaired t tests with Welch’s correction were used for comparing all experiments with two variables. A one-way ANOVA was used to calculate the glial and neuronal TEM rescue experiments. Fisher’s exact test in R was used for calculating the ciliobrevin D experiments in zebrafish.

SI Materials and Methods

TEM.

Zebrafish larvae (5 dpf) were fixed in modified Karnovsky’s solution (2% glutaraldehye, 4% paraformaldehyde, and 0.1 M sodium cacodylate buffer, pH 7.4) using a Pelco BioWave Pro with Steady Temp Plus water recirculation system. To ensure accurate quantification, samples were cut between body segments 5–7. Preparation of tissue for analysis was performed as described (66). We cut ∼70-nm thin sections using an ultramicrotome and mounted them on copper mesh grids. Sections were stained using uranyl acetate and Sato’s lead stain. A Jeol JEM-1400 (Jeol USA) electron microscope was used to view samples, and images were obtained using an AMT V601 digital camera. Images were analyzed using FIJI software (NIH), and all quantification was performed blinded to genotype.

Spinal Cord OPC Quantifications.

tg(sox10:mgfp);actr10^nl15/+ animals were crossed to tg(olig2:dsred);actr10^nl15/+ animals. At 3 dpf, double transgenic larvae were identified on a Zeiss fluorescent stereoscope by screening for dual-color fluorescence. Larvae were fixed for 30 min at room temperature using 4% EM-grade paraformaldehyde diluted in 1× PBS. Larvae were then transferred to 30% sucrose for overnight immersion at 4 °C. The next day, larvae were transected between segments 5–7. Heads were saved for genotyping, and bodies were immersed and frozen in Tissue-Tek OCT compound using dry ice. Samples were sectioned at 10 µm on a cryostat and mounted on charged slides. Seven sections per animal were imaged at 20× on an inverted Zeiss LSM 880 confocal microscope, and OPC numbers were counted and averaged. All work was performed blinded to genotype.

Quantification of Dorsal Elongated OPCs.

tg(olig2:dsred);actr10^nl15/+ animals were crossed to actr10^nl15/+ animals. At 3 and 4 dpf, larvae were screened for DsRed on a Zeiss fluorescent stereoscope. Larvae were then immobilized in tricaine and mounted on an agarose mold. The numbers of dorsally migrated olig2-labeled cells on either side of the spinal cord along the entire length of the animal were counted using a Zeiss fluorescent stereoscope. Quantifications were notated, and animals were saved for DNA extraction.

Time-Lapse Imaging.

tg(nkx2.2a:megfp);actr10^nl15/+ animals were crossed to actr10^nl15/+ animals. Larvae were screened for fluorescence at 2 dpf. At ∼55 hpf, mEGFP-positive embryos were immersed in 0.8% low-melting point agarose and mounted on 22 × 22 mm² glass coverslips. Vacuum grease was applied to a glass slide, and egg water with 0.2% tricaine was used to fill the space. The coverslip containing mounted embryos was placed on top of the water to immerse embryos. At ∼57 hpf, larvae were imaged on an inverted Zeiss LSM 880 confocal microscope at 20× for up to 12 h, and Z-stacks were collected every 10 min. Blinded analyses were performed using FIJI software (NIH). Cell division times were calculated blinded to genotype from TIFF images of each time point.

Rescue Experiments.

The transgenic tg(neurod:mrfp-actr10) rescue line was generated by concurrent injection of 5kbneurod:mRFP-actr10 (12–25 pg) and transposase RNA (50–100 pg) into one-cell stage embryos from an actr10^nl15/+ intercross. Larvae with green hearts and mRFP fluorescence were grown to adulthood and genotyped for actr10^nl15/+. These animals were then crossed to actr10^nl15/+ animals, and progeny were scored for germ-line transmission of 5kbneurod:mRFP-actr10. Animals were again grown, genotyped for actr10^nl15/+, and crossed to actr10^nl15/+ for experiments. To generate the sox10:mRFP-actr10 construct, primers were designed to amplify the sox10 promoter (68) and to add Fse1 and Asc1 restriction enzyme sites to the 5′ and 3′ ends, respectively (forward primer: 5′-GACTACGGCCGGCCGGATCCCCTTATCAGAGTCAACATTCATGGATAATGAAAAAGGC-3′; reverse primer: 5′-TGGCAAGGCGCGCCACTAGTCGGTCCACTCGTTCTGCGGCCA-3′). The sox10 promoter fragment (insert) was amplified using a 2X Phusion Master Mix (Thermo Fisher Scientific) and PCR purified using the QIAquick PCR purification kit (Qiagen). The insert and 5kbneurod:mRFP-actr10 plasmid (vector) were digested with Fse1 and Asc1, and the vector was subsequently treated with Antarctic phosphatase. Both vector and insert were visualized using gel electrophoresis, and the correct bands were excised and gel purified. Vector and insert were ligated overnight using T4 ligase (New England Biolabs). The next day, ligated product was transformed and grown overnight on antibiotic plates. Subsequently, colonies were grown in liquid culture overnight, and mini preps were made the next day. We simultaneously injected 25 pg of mini prep and 100 pg of transposase RNA into 1-cell zebrafish embryos from tg(mbp:egfp-caax);actr10^nl15/+ fish crossed to actr10^nl15/+ fish. Injected larvae were screened for green hearts and mRFP fluorescence as early as 24 hpf. tg(mbp:egfp-caax) uninjected actr10^nl15/nl15 mutants and mRFP-positive–injected actr10^nl15/nl15 mutants were imaged on the two-photon LSM 880 at 4 dpf. In a separate experiment, tg(mbp:egfp-caax) uninjected actr10^nl15/nl15 mutants and mRFP-positive–injected actr10^nl15/nl15 mutants were saved for TEM analysis at 5 dpf.