Loss of mbl-1 alters synapse formation in the cholinergic motorneuron DA9
Located in the preanal ganglion, the cholinergic motorneuron DA9 is the most posterior of the DA-type motorneurons with a dendrite that extends anteriorly from the cell body and an axon that extends posteriorly, then crosses the midline of the animal via a commissure and extends anteriorly in the dorsal nerve cord (DNC). While in the DNC, DA9 forms approximately 25 en passant dyadic synapses onto body wall muscles and reciprocal inhibitory VD neurons [20].
To study synapse formation in DA9, we used a cell-specific promoter to express a green fluorescent protein (GFP)-tagged version of the synaptic vesicle associated protein RAB-3 (GFP::RAB-3) [21]. In wild-type animals, RAB-3 accumulates in discrete puncta along the axon of DA9 at stereotyped locations and co-localizes with presynaptic active zone proteins. These puncta correspond to the location of synapses in electron microscopy reconstructions of DA9 [19, 22] (Figure 1A).
While working with the strain RB771, we identified an unmapped locus that disrupts the normal pattern of synapses in DA9. Using SNP polymorphisms between N2 and Hawaiian strains [23], we mapped this mutation to the far right arm of the X chromosome. After whole genome sequencing we identified a 70-kb deletion that eliminates eight genes, one of which is mbl-1. Since the wy560 mutation completely eliminates the mbl-1 locus, it represents a null allele for this gene. A second allele, mbl-1(tm1563), is a 513-bp deletion that eliminates all of exon 3 of mbl-1, an exon shared among all mbl-1 isoforms. Exon 3 skipping is predicted to cause an in-frame deletion of some isoforms; however, for the three isoforms that utilize a start codon in exon 3, the next downstream start codon would result in a frame-shift. Further analysis revealed that trans-heterozygous mbl-1(tm1563)/mbl-1(wy560) animals have a synapse number phenotype identical to that of the homozygous single mutants (Figure 2) and because the two alleles of mbl-1 showed an indistinguishable defect for DA9 synapse number, the tm1563 allele is likely acting as a null with respect to the DA9 synapse number phenotype.
In both mbl-1(wy560) and mbl-1(tm1563) mutant animals, several aspects of DA9 presynaptic patterning are disrupted (Figure 1). The most penetrant phenotype is the loss of approximately ten synapses from the DA9 synaptic region: 24.4 ± 3.6 RAB-3 puncta in wild-type animals versus 14.3 ± 2.2 in mbl-1(wy560) mutants and 15.0 ± 2.4 in mbl-1(tm1563) mutants (standard deviation, P < 0.0001, t-test; Figure 1D). To quantitatively analyze this phenotype, we imaged the DA9 synaptic region in ten wild-type and mutant animals by confocal microscopy and constructed composite images of the presynaptic regions of all ten animals by aligning them along their anteroposterior axes (Figure 1E-G). We observe that the most distal GFP::RAB-3 puncta are absent in mbl-1(wy560) and mbl-1(tm1563). However, the spacing between the remaining synaptic puncta in both mbl-1(wy560) and mbl-1(tm1563) is similar to the spacing between GFP::RAB-3 puncta in wild-type animals. Examination of a transgenic line (wyEx1902) in which DA9 is labeled with a cytoplasmic fluorophore demonstrates that DA9 neuronal morphology is normal in mbl-1(tm1563) and loss of these distal synapses is not a result of axonal truncation (data not shown.)
Measurement of the total fluorescence intensity of the composite images along the x-axis demonstrates that the peak fluorescence is indeed more posterior in mbl-1(wy560) and mbl-1(tm1563) mutant animals compared to wild type (Figure 1H). Furthermore, it appears that the highest point of fluorescence intensity is lower in mbl-1(wy560) and mbl-1(tm1563). Thus, in mbl-1 mutants, fewer synapses form and the synapses that are present seem to contain less RAB-3 than their wild-type counterparts.
Imaging analysis of mbl-1(wy560) and mbl-1(tm1563) mutants also revealed the presence of ectopic RAB-3 in compartments of DA9 from which RAB-3 is normally excluded. In 59% of mbl-1(wy560) and 26% of mbl-1(tm1563) animals, we observe puncta in the DA9 dendrite, compared to 0% for wild type (Figure 1I). When visible, these ectopic dendritic puncta are significantly smaller than dorsal synaptic puncta and mutants typically have one to six puncta. We also observe ectopic puncta in other segments of DA9, including the posterior dorsal asynaptic region, commissure, and ventral axon.
Other pre-synaptic markers are affected in mbl-1, but post-synaptic markers are unchanged
The RAB-3 phenotype in mbl-1 mutants prompted us to ask whether this defect is a result of impaired synaptic vesicle transport or a failure of de novo synapse formation by analyzing the localization of other pre-synaptic proteins. The presynapse is a complex structure composed of many active zone proteins required to facilitate the exocytosis of synaptic vesicles in response to an action potential, as well as vesicle-associated proteins that allow active zone machinery to interact with synaptic vesicles [24]. We expressed two fluorescently tagged active zone proteins, UNC-10 (Rim) [25] and SYD-2 (Liprin-α) [26], in DA9 and examined their localization in the wild-type and mbl-1(tm1563) DA9 synaptic regions (Figure 3A). We find that UNC-10::GFP co-localizes with mCherry::RAB-3 at DA9 synapses in both wild-type and mbl-1(tm1563) animals, though there are fewer puncta overall in mbl-1(tm1563) mutant animals compared to wild-type animals (Figure 3B-G). Similarly, we find that GFP::SYD-2 and mCherry::RAB-3 co-localize, but there are again fewer total DA9 presynaptic puncta in mbl-1(tm1563) (Figure 3H-M). Taken together, these data suggest the defect in mbl-1(tm1563) represents the failure of synapse formation in the most distal part of the DA9 synaptic region.
To clarify whether the DA9 defect we observe is the result of improper synapse maintenance during development or an acute loss of synapses at the adult stage, we measured the length of the DA9 synaptic region in animals at the L2, L4, and adult stages (Figure 4). We measured the synaptic length rather than counting individual puncta because puncta are densely clustered in young animals, making it difficult to count individual punctum. We observe that there is a consistently shorter synaptic region in DA9 in mbl-1(tm1563) at all timepoints measured. This absolute difference translates to a 28% decrease in synapse length at the L2 stage, a 28% decrease at the L4 stage, and a 35% decrease in young adults. These results demonstrate that mbl-1 mutants animals consistently fail to accumulate the appropriate amount of synaptic material at the DA9 synaptic region throughout development.
We next asked whether the change in DA9 synaptic patterning is accompanied by changes in the postsynaptic muscle partner as loss of MBNL in human, mouse, and Drosophila all result in muscle changes. DA9 forms dyadic synapses onto body wall muscles and VD inhibitory motorneurons. Upon DA9 acetylcholine release, acetylocholine binds to postsynaptic acetylcholine receptors on body wall muscles and causes contraction of dorsal body wall muscles. The C. elegans acetylcholine receptor, ACR-16, is important for excitatory neuronal transmission at the NMJ [27] and localizes in a discrete line along the dorsal body wall muscles.
We examined ACR-16::GFP localization in wild-type and mbl-1(tm1563) mutants (Figure 5A-F). ACR-16:GFP localization was qualitatively similar between wild-type and mutant animals, with relatively continuous staining along the DNC. ACR-16 staining appears to cover more area compared with the RAB-3. This is to be expected because the ACR-16 staining represents postsynaptic specializations with both DA and DB classes of neurons, while the RAB-3 staining is only in the DA9 neuron. Qualitative observation of mCherry::RAB-3 puncta in both wild-type and mbl-1(tm1563) worms reveals that most RAB-3 co-localizes with ACR-16::GFP in the DNC (data not shown).
Previous reports have documented muscle defects in mbl-1 mutant animals [11]. We examined muscle morphology in two transgenic lines in which muscles are labeled with a cytoplasmic fluorophore, wyEx661 [20] and trIs30 [28] (Figure 5G-I). We find that both dorsal and lateral body wall muscle morphology is qualitatively normal in mbl-1 mutants. Muscles in both wild type and mutant are of similar size and shape.
In C. elegans, specialized extensions called muscle arms extend from the muscle to reach motor axons to form NMJs. Therefore, similar to dendritic spines, muscle arms represent postsynaptic components for NMJ formation in C. elegans. We found that muscle arms from both the dorsal and lateral muscles in mbl-1(tm1563) appear qualitatively normal in size and shape (Figure 5G-J). We further quantified the number of muscle arms from the lateral muscles and found that the number of muscle arms per muscle for wild-type and mbl-1(tm1563) animals are not significantly different from each other (3.7 ± 0.5 for wild type and 3.9 ± 0.7 for mbl-1(tm1563)) (Figure 5I, J). Taken together, these results demonstrate that the mbl-1 mutant phenotype does not affect muscle arm outgrowth or postsynaptic receptor localization, and we hypothesized that the phenotype reflects a defect in synapse formation on the presynaptic side.
mbl-1 functions cell autonomously in DA9
To confirm that mutations in mbl-1 cause the synaptic defects observed, we rescued the mbl-1 phenotype by injecting a fosmid (WRM0616bE04) that contained genomic DNA encompassing all of the mbl-1 exons and introns as well as upstream and downstream non-coding sequences. We found the presence of the genomic fragment completely restored the wild-type number of synapses in a mbl-1(tm1563) mutant background (25.2 ± 3.5 and 25.7 ± 3.6; Figure 6B). Rescue was determined by counting the number of visible DA9 synaptic puncta on an epifluorescence microscope.
To determine if the synaptic defects we observe in mbl-1 were due to changes in the presynaptic neuron DA9 or to changes in the postsynaptic body wall muscle, we performed cell autonomous rescue experiments (Figure 6). We expressed a genomic fragment including most of the mbl-1 ORF under the control of promoters that express in neurons including DA9 (Pmig-13 or Punc-4c) or body wall muscles (Phlh-1) and scored rescue of the DA9 synaptic phenotype in mbl-1(tm1563) mutant animals (Figure 6B). While expression of mbl-1 in muscle failed to rescue the DA9 synaptic defect (wyEx4028, 16.0 ± 3.3; wyEx4046, 15.3 ± 2.8), expression of mbl-1 in A-type neurons showed significant rescue of the DA9 synaptic defect. Punc-4c, a truncated version of the unc-4 promoter (unpublished reagent via MV and KS), drives expression in only the DA-type motorneurons. Expression of the mbl-1 genomic fragment under the control of the unc-4c promoter rescues the DA9 synaptic defect (wyEx4025, 23.4 ± 3.8; wyEx4026, 23.0 ± 3.7). The mig-13 promoter is expressed in DA9, VA12 and other anterior DA neurons and is expressed beginning at embryonic stages. When we used Pmig-13 to drive expression of the mbl-1 genomic fragment, we found that it also significantly restores the correct number of synapses to the dorsal axon in DA9 (wyEx4030, 22.4 ± 4.0; wyEx4234, 21.5 ± 3.5). Thus, we conclude that mbl-1 is required cell autonomously in the presynaptic neuron to regulate the number of synapses.
Expression of genomic mbl-1 in DA9 can also rescue the appearance of ectopic dendritic puncta in mbl-1(tm1563). Expression of the mbl-1 genomic fragment under the control of the Pmig-13 or Punc-4c promoters can rescue or partially rescue the appearance of ectopic dendritic puncta. In mbl-1(tm1563), 26% of animals have some dendritic puncta versus 0% and 13% with the Pmig-13 promoter, 4% and 10% with the Punc-4c promoter, versus 18% and 19% with the Phlh-1 promoter.
mbl-1 is expressed in the C. elegans nervous system
In order to examine the expression pattern of mbl-1, we inserted an SL2::mCherry sequence into the fosmid (WRM0616bE04) containing the mbl-1 ORF by homologous recombination. The SL2 sequence acts as a trans-splice acceptor [29] whereby the transcript is expressed under the control of the mbl-1 promoter and is later spliced into two separate mRNAs that are translated independently. The recombineered fosmid results in expression of endogenous, untagged MBL-1 and cytoplasmic mCherry, both under the control of the endogenous mbl-1 promoter. Examination of transgenic worms carrying the recombineered fosmid reveals that the mbl-1 promoter drives expression of mCherry in many cell bodies along the ventral nerve cord (Figure 6C, F). Thin neuronal processes emanate from these cell bodies and fasciculate in both the dorsal and ventral nerve cords.
To determine the neuronal subtypes that express MBL-1, we co-injected the recombineered fosmid with Punc-4::GFP, a construct that drives expression of GFP in the A-type (DA and VA) motorneurons [30]. By examining expression patterns of GFP and mCherry, we find that MBL-1 is expressed strongly in all of the A-type motorneurons, and also in other neurons in the ventral cord and several unidentified cells in the tail. mCherry expression was seen in embryonic and L1 animals, indicating that mbl-1 is expressed early in development. Our expression data are largely consistent with previously reports using a smaller fragment of the mbl-1 promoter [31].
To determine the subcellular localization of mbl-1, we created a translational fusion construct by adding a GFP to the carboxyl terminus of mbl-1. Expression of MBL-1::GFP was driven by the endogenous mbl-1 promoter and we observed that MBL-1::GFP is concentrated in the nucleus of ventral cord neurons, consistent with its putative function in pre-mRNA splicing, which occurs in the nucleus (Figure 7).
DA synapse density is altered in mbl-1
Because mbl-1 is expressed in all A-type motorneurons, we tested whether synapse formation is defective in other neurons of this class. Using a transgenic line that expresses GFP::RAB-3 in all DA neurons under the control of the unc-4 promoter, we found that DA synapse density is decreased from 43.2 synapses per 100 μm in wyIs85 to 37.3 synapses per 100 μm in mbl-1(tm1563) (P < 0.0001) (Figure 8). Individual DA synaptic regions cannot be discerned in this transgenic line, but changes in synapse density arise from significantly large gaps between puncta at regular intervals (Figure 8, arrowheads), rather than a change in the distance between all puncta, implying that all DA synaptic regions are shortened in a similar fashion.
Loss of mbl-1 affects backwards locomotion
To understand whether the synaptic phenotypes in the mbl-1 mutant affect behavior, we specifically examined the backward locomotion of mbl-1 mutants because the A-type motor neurons are involved in backward locomotion [32]. We observed that as mbl-1 animals move backwards, they do not back in a straight line as wild-type animals do. Instead, they move backwards more slowly and curve ventrally (Figure 9A-D).
To quantify the backwards locomotion defect, we tapped worms on the head with a platinum wire and recorded the angle of backwards movement where the axis of the preceding forward movement is defined as 0°. A ventral turn during backing is recorded as a positive angle (+) and a dorsal turn as negative (-). The average angle after moving backwards two body lengths for wild type was 0.8 ± 1.5° and 42.0 ± 7.2° for mbl-1(tm1563) (Figure 9E). Often the worms will curve 45° per body bend. In addition, mbl-1(tm1563) worms often back over many more body lengths than wild-type worms and rarely perform omega bends. This change in backwards locomotion could be explained by the hypocontraction of dorsal body wall muscles in response to a decrease in the density of DA9 motorneuron synapses.
When the same genomic fragment that rescues the DA9 synaptic defect (mbl-1 genomic exons 3 to 8) is expressed under the control of the neuronal-specific promoters Pmig-13 and Punc-4c, we find that the backing behavior is partially rescued. In contrast, when mbl-1 is expressed under the control of a muscle-specific promoter (Phlh-1), the backing behavior is not rescued. The angles observed after two lengths of backwards movement were 9.0 ± 2.4° for Pmig-13, 19.8 ± 5.0° for Punc-4c, and 41.3 ± 5.7° for Phlh-1 (Figure 9E). These results suggest that the loss of dorsal synapses in DA9 has a measurable affect on backwards locomotion.