1. Neuroscience

Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons

  1. Sahana Sitaraman
  2. Gnaneshwar Yadav
  3. Vandana Agarwal
  4. Shaista Jabeen
  5. Shivangi Verma
  6. Meha Jadhav
  7. Vatsala Thirumalai  Is a corresponding author
  1. National Centre for Biological Sciences, Tata Institute of Fundamental Research, India
https://doi.org/10.7554/eLife.68124
Share this article
Cite this article
  1. Sahana Sitaraman
  2. Gnaneshwar Yadav
  3. Vandana Agarwal
  4. Shaista Jabeen
  5. Shivangi Verma
  6. Meha Jadhav
  7. Vatsala Thirumalai
(2021)
Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons
eLife 10:e68124.
https://doi.org/10.7554/eLife.68124
  2. Download BibTeX
  3. Download .RIS

Abstract

Gap junctions between neurons serve as electrical synapses, in addition to conducting metabolites and signaling molecules. During development, early-appearing gap junctions are thought to prefigure chemical synapses, which appear much later. We present evidence for this idea at a central, glutamatergic synapse and provide some mechanistic insights. Loss or reduction in the levels of the gap junction protein Gjd2b decreased the frequency of glutamatergic miniature excitatory postsynaptic currents (mEPSCs) in cerebellar Purkinje neurons (PNs) in larval zebrafish. Ultrastructural analysis in the molecular layer showed decreased synapse density. Further, mEPSCs had faster kinetics and larger amplitudes in mutant PNs, consistent with their stunted dendritic arbors. Time-lapse microscopy in wild-type and mutant PNs reveals that Gjd2b puncta promote the elongation of branches and that CaMKII may be a critical mediator of this process. These results demonstrate that Gjd2b-mediated gap junctions regulate glutamatergic synapse formation and dendritic elaboration in PNs.

Introduction

The formation of synapses is an elaborate multi-step process involving several classes of signaling molecules and electrical activity (Waites et al., 2005; McAllister, 2007). Several studies support the view that connections that favor correlated activity between presynaptic and postsynaptic neurons are strengthened while those that are poorly correlated are weakened and eliminated (Katz and Shatz, 1996; Kirkby et al., 2013). Correlated activity promotes dendritic elaboration (Wong and Ghosh, 2002; Parrish et al., 2007) and the subsequent formation of synaptic sites on those newly formed arbors (Niell et al., 2004). However, we understand little about the molecular mechanisms linking correlated activity to dendritic arbor elaboration and synaptogenesis (Sin et al., 2002; Redmond and Ghosh, 2005; Schwartz et al., 2009; Chen et al., 2012).

Electrical synapses, formed via gap junctions between connected neuronal pairs, allow the passage of ions, metabolites, and second messengers, and are ideally suited for enhancing correlations in activity between connected neurons (Pereda, 2014; Connors, 2017). Indeed, several lines of evidence suggest that gap junctions play critical roles in circuit assembly. First, neurons show increased gap junctional connectivity early on in development at stages that precede chemical synapse formation (Montoro and Yuste, 2004; Marin-Burgin et al., 2008; Jabeen and Thirumalai, 2013). Second, knocking out or knocking down gap junction proteins at these stages results in decreased chemical synapse connectivity at later stages (Maher et al., 2009; Todd et al., 2010). Gap junctions could mediate chemical synaptogenesis by increasing correlations in activity, transmitting synaptogenic signaling molecules, or providing enhanced mechanical stability at junctional sites between connected pairs. It is not clear which of these functions of gap junctions are critical for synaptogenesis. In addition, while the role of excitatory chemical synapses in sculpting neuronal arbors has been investigated in several circuits (Inglis et al., 2002; Haas et al., 2006; Cline and Haas, 2008), little is known regarding such a role for electrical synapses. We set out to investigate whether gap junctions regulate structural and functional synaptic development of cerebellar Purkinje neurons (PNs), and if yes, what mechanisms may be involved using larval zebrafish as our model system.

The cerebellum is critical for maintaining balance and for coordination of movements. It is one of the most primitive organs of the vertebrate central nervous system, has a layered structure, and its circuitry is conserved from fish to mammals (Nieuwenhuys, 1967). PNs are principal output neurons of the cerebellar cortex and receive excitatory and inhibitory synaptic inputs on their elaborate dendritic arbors. PNs receive thousands of glutamatergic inputs from parallel fiber axons of granule cells and relatively fewer inhibitory inputs from molecular layer interneurons. They also receive strong excitatory inputs from inferior olivary climbing fiber (CFs) axons on their proximal dendrites. In zebrafish, PNs are specified by 2.5 days post fertilization (dpf), begin elaborating dendritic arbors soon after, and a distinct molecular layer consisting of PN dendritic arbors becomes visible by 5 dpf (Bae et al., 2009; Hamling et al., 2015). In addition, excitatory and inhibitory synaptic currents can be recorded in PNs of 4 dpf zebrafish larvae, evidencing a nascent functional circuit at this stage (Sengupta and Thirumalai, 2015).

The gap junction delta 2b protein (Gjd2b), also referred to as Connexin 35b or 35.1 (Cx35/Cx35.1), is the teleostean homolog of the mammalian Cx36, which is the predominant neural gap junction protein. PNs begin to express Gjd2b by about 4 dpf, and the level of expression increases steadily at least until 15 dpf (Jabeen and Thirumalai, 2013). We asked whether Gjd2b regulates structural and functional synaptic development of PNs. Using knockdown and knockout approaches, we show here that Gjd2b is indeed required for the formation of glutamatergic synapses and for normal dendritic arbor growth of PNs.

Results

Expression and knockdown of Gjd2b in PNs using morpholinos

Gjd2b is expressed at high levels in the cerebellum of larval zebrafish beginning at 4 dpf, a stage at which cerebellar neurons have been specified but chemical synaptic connections are still forming (Bae et al., 2009; Jabeen and Thirumalai, 2013; Figure 1—figure supplement 1A). Using an antibody that recognizes both Gjd2a and Gjd2b, we confirmed that Gjd2a/b puncta localized to PN cell membrane in their cell bodies and dendrites (Figure 1—figure supplement 1B). To test whether Gjd2b mediates chemical synaptogenesis in PNs, we knocked it down with a splice-blocking morpholino antisense oligonucleotide (referred to as Gjd2b-MO) targeted to the splice junction between exon 1 and intron 1 of gjd2b (Figure 1—figure supplement 2A, top). Injection of this splice blocking morpholino (Gjd2b-MO) into 1–4 cell stage embryos interferes with normal splicing of the gjd2b gene product, resulting in the inclusion of a 40 base intronic segment in the mature mRNA (Figure 1—figure supplement 2C, left). The mis-spliced gene product was detected in 2 dpf and 5 dpf larvae (Figure 1—figure supplement 2B) and the sequence reveals a premature stop codon resulting in a putative truncated protein of 24 amino acid residues in the N-terminus (Figure 1—figure supplement 2C, right). Gjd2a/b immunoreactivity in the PN and molecular layers of the cerebellum is reduced in Gjd2b-MO larvae compared to uninjected larvae (Figure 1—figure supplement 2D, Mann–Whitney test, p<0.001) confirming effective knockdown. Injection of a morpholino with a 5-base mismatch (CTRL) did not alter Gjd2a/b protein levels (Figure 1—figure supplement 2E, Mann–Whitney test, p=0.14).

Morphant PNs exhibit deficits in AMPAR-mediated synaptic transmission

We followed physiological changes induced by Gjd2b loss by recording AMPAR-mediated miniature excitatory postsynaptic currents (mEPSCs) in PNs of uninjected and Gjd2b-MO larvae using previously established methods (Sengupta and Thirumalai, 2015). mEPSCs in morphants occurred less frequently than in uninjected larvae (Figure 1A), reflected as increased inter-event intervals in the morphants (Figure 1C, Mann–Whitney test, U = 37,285, p<0.0001). mEPSCs in morphants also showed a small increase in the peak amplitude (Figure 1D, Mann–Whitney test, U = 25,537, p=0.002) and faster decay time constants (Figure 1B, F, Mann–Whitney test, U = 25,018, p=0.014). The increase in inter-event intervals of mEPSCs suggests that knocking down Gjd2b leads to a decrease in the density of synapses impinging on PNs.

Figure 1 with 2 supplements see all
Download asset Open asset
Knocking down Gjd2b reduces glutamatergic miniature excitatory postsynaptic current (mEPSC) frequency in Purkinje neurons (PNs) by potentially decreasing synaptic number.

(A) Raw traces of mEPSC recordings from PNs in control (black) and Gjd2b-MO-injected (red) larvae. (B) Average mEPSC waveforms from control and Gjd2b-MO-injected larvae. Inset: scaled mEPSC to show faster decay time of mEPSCs in morphants. Neurons were held at −65 mV. (C–F) Cumulative probability distributions and boxplots of mEPSC inter-event intervals (C), peak amplitudes (D), 10–90% rise times (E), and decay tau (F) in control (black) and Gjd2b-MO-injected (red) larvae. N = 7 cells from 7 larvae from 5 clutches for control and 12 cells from 12 larvae from 8 clutches for the Gjd2b-MO group. **p<0.01; ***p<0.001; ****p<0.0001; Mann–Whitney U test. See also Figure 1—figure supplements 1 and 2. Data used for quantitative analyses are available in Figure 1—source data 1.

Figure 1—source data 1

mEPSC data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig1-data1-v2.xlsx

Generation of gjd2b-/- zebrafish using TALENs

As an independent but stable approach (Kok et al., 2015; Stainier et al., 2017) to verify effects of a complete loss of Gjd2b on glutamatergic synapses, we generated gjd2b-/- fish using TALENs (Christian et al., 2010) targeted against exon 1 of gjd2b (Figure 2A). We isolated several alleles with indels in the target region and chose one allele, gjd2bncb215 (hereinafter referred to as gjd2b-/-), for further analysis. In this allele, insertion of a single G after the translation start site abolished an XhoI restriction enzyme recognition site within the target region (Figure 2B) and resulted in failure of XhoI restriction digestion in homozygotes and partial digestion in heterozygotes (Figure 2C). This single-nucleotide insertion caused a frame shift and insertion of a premature stop codon, with a predicted truncated protein of 55 amino acid residues (Figure 2D). Homozygotes with this mutation showed reduced Gjd2b-like immunoreactivity in their cerebellum (Figure 2E, F).

Figure 2
Download asset Open asset
Generation of gjd2b mutant zebrafish.

(A) Genomic region around the start codon (green box) of gjd2b gene was selected for TALEN design, where TALEN-1 recognition sequence (blue line) spans the 5′UTR, the start codon sequence, and a few nucleotides after the start codon. The TALEN-2 recognition sequence (purple line) begins after the 18 nucleotide spacer region. (B) Chromatograms of sequence read from wild-type (WT) and homozygous gjd2bncb215 fish generated in this study indicate an insertion of nucleotide G (red arrowhead) within the Xho1 restriction site (blue line). (C) Representative gel image after Xho1 restriction digestion analysis of an 800 bp amplicon (includes ~300 bp upstream and 500 bp downstream sequences from the point of insertion) shows undigested and partially digested bands in the homozygous and heterozygous mutants, respectively, compared to complete digestion in WT siblings. (D) Predicted amino acid sequences of various gjd2b mutant alleles generated in this study aligned with the WT Gjd2b sequence. gjd2bncb215 is predicted to code for the first six amino acid residues of Gjd2b followed by a nonsense sequence up to the 54th amino acid position. Presence of a premature stop codon terminates translation at this position. (E) Representative images of Gj2b-like immunoreactivity from WT and mutant fish. Staining was not completely abolished as the antibody also recognizes Gjd2a. (F) Gjd2b-like immunoreactivity is reduced in PNs of gjd2bncb215 compared to WT larvae (Mann–Whitney U test; ***p<0.001). Number of images analyzed is indicated in parentheses. Data used for quantitative analyses are available in Figure 2—source data 1.

Figure 2—source data 1

Fluorescence intensity data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig2-data1-v2.xlsx

Impaired AMPAR-mediated synaptic transmission in PNs of gjd2b-/- larvae

We recorded mEPSCs from PNs in gjd2b-/- larvae to determine if loss of Gjd2b leads to reduced glutamatergic synaptic contacts, as observed in the morphants (Figure 1). These recordings were performed with cesium gluconate internal solution, and under these conditions, the input resistance of mutant neurons was significantly higher compared to wild type (WT; Figure 3—figure supplement 1B). gjd2b-/- larvae showed a significant increase in mEPSC inter-event intervals (Figure 3A, C, Mann–Whitney test, U = 4906, p=0.02), recapitulating the results observed after knock down of Gjd2b with Gjd2b-MO (Figure 1C). Further, homozygous mutants also showed an increase in peak amplitudes (Figure 3A, B, D, Mann–Whitney test, U = 3385, p<0.0001), and faster kinetics as revealed by decreases in rise times (Figure 3B, E, Mann–Whitney test, U = 2322, p<0.0001) and decay time constants (Figure 3B, F, Mann–Whitney test, U = 2791, p<0.0001). These results are consistent with the results obtained after knocking down Gjd2b with Gjd2b-MO and point to a decrease in the number of glutamatergic synaptic contacts impinging on PNs.

Figure 3 with 1 supplement see all
Download asset Open asset
Knocking out Gjd2b results in decrease of glutamatergic synaptic number.

(A) Representative miniature excitatory postsynaptic current (mEPSC) recordings from Purkinje neurons (PNs) of wild-type (black trace) and gjd2b-/- (red) larvae. (B) Average mEPSC shown on expanded time base recorded from wild-type (black) and gjd2b-/- (red) larvae. Neurons were held at −65 mV. Inset: scaled mEPSC to show faster rise time and decay time of mEPSCs in mutants. (C–F) Cumulative probability histograms and boxplots reveal increased inter-event intervals (C), increased peak amplitudes (D), decreased 10–90% rise times (E), and decreased decay time constants (F) of mEPSCs in gjd2b-/- larvae (red lines) compared to wild type (black lines). N = 8 cells in wild type and 10 cells in gjd2b-/- larvae. *p<0.05; ****p<0.0001; Mann–Whitney U test. Data used for quantitative analyses are available in Figure 3—source data 1. See also Figure 3—figure supplement 1.

Figure 3—source data 1

Mutant mEPSC and Rin data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig3-data1-v2.xlsx

However, this change may also be due to a decrease in presynaptic probability of transmitter release or an increase in the number of N-methyl-D-aspartate receptor (NMDAR)-only ‘silent’ synapses (Liao et al., 1995; Wu et al., 1996). To test if these possibilities are likely, we recorded synaptic currents evoked in WT PNs after stimulation of CFs (Figure 4A). These recordings were performed with potassium gluconate internal solution, and under these conditions, the input resistance of mutant neurons was comparable to WT (Figure 3—figure supplement 1A). As is the case in developing mammalian PNs (Piochon et al., 2007), we recorded no NMDAR-mediated component of the evoked EPSC in PNs (Figure 4B), suggesting that the decrease in mEPSC frequency after Gjd2b knockdown and knockout is likely not due to silent synapses. To further strengthen our results, we performed voltage clamp recordings from WT and gjd2b-/- PNs after administering tetrodotoxin (TTX) in the bath and brief puffs of 10 mM NMDA close to the soma. The cells were held at a depolarized potential (−20 mV) to remove the Mg2+ block from NMDARs. No current response was observed after application of NMDA (Figure 4C), suggesting the absence of NMDARs on PNs in 7 dpf zebrafish larvae.

Figure 4
Download asset Open asset
Reduction in miniature excitatory postsynaptic current (mEPSC) frequency in Gjd2b-KD and KO animals is not due to silent synapses or change in probability of transmitter release.

(A) Schematic of experimental setup for stimulating climbing fibers (CFs) while recording EPSCs in Purkinje neurons (PNs; blue). (B) EPSCs recorded at a hyperpolarized holding potential (bottom row traces) and at a depolarized holding potential (top traces) in normal saline (left side traces) and in saline containing the AMPAR blocker CNQX (right-side traces). No EPSCs were detected in the presence of CNQX at −65 or +60 mV at 7 days post fertilization (dpf) (N = 4 cells) or at 19 dpf (N = 2 cells). Data from 19 dpf shown above. (C) Voltage clamp recordings from wild-type (WT) and mutant PNs at 7 dpf at a holding potential of −20 mV show no detectable response to brief pulses (blue bars) of N-methyl-D-aspartate (NMDA). Recordings were done in the presence of 1 µM tetrodotoxin (TTX). (D) Paired pulse depression of EPSCs in PNs of control (black) and Gjd2b-MO-injected (red) larvae. (E) Paired pulse ratios were not significantly different between control and Gjd2b-MO-injected larvae at any of the interstimulus intervals (ISIs) tested (mean ± SD; N = 5 cells from five larvae each in control and Gjd2b-MO groups; two-way repeated-measures ANOVA, p=0.081 for groups [control, Gjd2b-MO] and p<0.001 for ISIs). (F) Paired pulse depression of EPSCs in PNs of WT (black) and gjd2b-/- (red) larvae. (G) Paired pulse ratios were not significantly different between WT and mutant larvae. Mean ± SD; two-way repeated-measures ANOVA, p=0.17 for groups (control, Gjd2b-MO) and p<0.001 for ISI. Data used for quantitative analyses are available in Figure 4—source data 1.

Figure 4—source data 1

Morphant and mutant PPR data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig4-data1-v2.xlsx

Secondly, we measured paired pulse ratios (PPRs) as an indicator of changes in presynaptic vesicle release probability. The CF-PN synapse shows paired pulse depression when pulses are placed 35 ms apart (Figure 4D, F). We tested a range of inter-stimulus intervals (ISIs) from 30 ms till 550 ms (in morphants) and till 1000 ms (in mutants) and found that the PPR varied as a function of the ISI but did not vary significantly between the uninjected and Gjd2b-MO groups (Figure 4E, two-way repeated-measures ANOVA, F = 3.979, df = 1, p=0.081) or WT and gjd2b-/- groups (Figures 4G, two-way repeated-measures ANOVA, F = 2.039, df = 1, p=0.17).

Taken together, these results suggest that knocking down or knocking out Gjd2b leads to a decrease in the number of glutamatergic synapses impinging on PNs. In subsequent experiments, we investigate this phenomenon in greater detail using the gjd2b-/- mutants.

Loss of Gjd2b reduces synapse density in the molecular layer

To finally confirm if the loss of Gjd2b indeed results in a decrease in synapse density, we quantified synapse density at the ultrastructural level. PNs send elaborate dendritic arbors into the molecular layer of the cerebellum where they make excitatory synapses with parallel fibers and CFs and inhibitory synapses with axons of molecular layer interneurons. Transmission electron micrographs (TEMs) were obtained from the molecular layer of the corpus cerebelli (CCe) of 7 dpf WT and mutant larvae from 60-nm-thick sections of the brain. Sections were taken at an interval of 1.2 µm to avoid oversampling the same synapses. Synapses were counted as membrane appositions with presynaptic vesicles on one side and electron-dense postsynaptic density on the other (Figure 5A). We found that the density of synapses per cubic micrometer was significantly lower in gjd2b-/- larvae compared to WT larvae (Figure 5B, E, Mann–Whitney test, U = 86,268, p<0.0001). To understand if Gjd2b also regulates the maturation of synapses, we calculated the synapse maturation index (Blue and Parnavelas, 1983; Haas et al., 2006) for individual synaptic profiles, measured as the ratio of the area occupied by vesicles to the entire area of the presynaptic terminal (Figure 5C). The distribution of synapse maturation indices was not different between WT and mutant larvae (Figure 5D, F, Mann–Whitney test, U = 4922, p=0.12). In sum, these results indicate that Gjd2b is involved in the formation of synapses in the molecular layer of the cerebellum, to which PNs significantly contribute, but once formed, the maturation of synapses is independent of Gjd2b.

Figure 5
Download asset Open asset
Knocking out Gjd2b leads to reduction in synaptic density in the cerebellar molecular layer.

(A) Transmission electron micrograph illustrating synapses identified using clustered vesicles (pink areas) in presynaptic terminals (yellow areas) apposed to postsynaptic density in dendritic profiles (green areas). (B) Cumulative probability plot and (E) boxplot, showing distribution of synapse density per cubic micrometer in wild type (black) and gjd2b-/- larvae (red). 637 micrographs from three wild-type larvae and 550 micrographs from three mutant larvae were analyzed. ****p<0.0001; Mann–Whitney U test. (C) Transmission electron micrograph at high magnification used for quantification of synapse maturation index. The area occupied by clustered vesicles (pink, Ac) was divided by the total area of the presynaptic terminal (yellow, At) to obtain the maturation index. (D) Cumulative probability plot and (F) boxplot, showing the distribution of synapse maturation indices in wild type (black) and gjd2b-/- larvae. 106 micrographs from three larvae each in wild-type and mutant groups were analyzed. p=0.12, Mann–Whitney U test. See also Supplementary file 1. Data used for quantitative analyses are available in Figure 5—source data 1.

Figure 5—source data 1

Synapse density and maturity index data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig5-data1-v2.xlsx

Stunted dendritic arbors in gjd2b-/- PNs

An increase in peak amplitude combined with faster kinetics of mEPSCs in gjd2b-/- mutants (Figure 3) suggested that synapses are placed electrotonically closer to the soma in mutant PNs than in WT larvae. In addition, input resistance measured with potassium gluconate internal solution did not reveal any difference between WT and mutant PNs, while recordings with cesium gluconate showed that the input resistance was higher for mutant PNs (Figure 3—figure supplement 1), suggesting a dendritic leak conductance that was larger for WT neurons compared to mutant neurons. These data imply that loss of Gjd2b could result in stunted dendritic arbors and loss of distally located synapses. To understand if this is indeed the case, we labeled PNs of WT and mutant larvae in a mosaic fashion and imaged them daily from 5 dpf till 8 dpf (Figure 6A, B), a period when PNs in larval zebrafish are growing and making synaptic connections (Bae et al., 2009; Hamling et al., 2015).

Figure 6 with 1 supplement see all
Download asset Open asset
Dendritic arbor growth of gjd2b-/-Purkinje neurons (PNs) is impaired.

(A) Representative traces of a wild-type (WT) PN from 5 to 8 days post fertilization (dpf). (B) Representative traces of a gjd2b-/- mutant PN from 5 to 8 dpf; scale bar is 10 µm; circle represents the position of the soma and is not to scale. (C) Total dendritic branch length (TDBL) of WT (black) and gjd2b-/- (red) PNs from 5 to 8 dpf. WT and gjd2b-/- neurons show significant growth from 5 to 6 dpf (p=0.0021). WT neurons show significantly higher TDBL values at all days (p=0.048). Statistical comparison was done using a general linear model with an inverse Gaussian error distribution. Post-hoc comparisons were done with the emmeans package in R. (D) Average rate of normalized hourly net branch growth in WT and gjd2b-/- PNs at 6 dpf. Mutant neurons show a significantly reduced rate over 10 hr of observation (p=0.005). (E) Rate of branch elongation in WT and mutant PNs at 6 dpf. Mutant PNs show a significantly reduced rate of branch length elongations (p=0.022). (F) Rate of branch length retractions in WT and mutant PNs at 6 dpf. No significant difference is observed between the two groups (p=0.185). Statistical comparisons in (D–F) were done using general linear models with Gaussian error distribution. Ns (number of neurons sampled) are indicated in parentheses in (C–E). See also Figure 6—figure supplement 1. Data used for quantitative analyses are available in Figure 6—source data 1.

Figure 6—source data 1

Daily and hourly imaging analysis.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig6-data1-v2.xlsx

We measured the total dendritic branch length (TDBL) of PNs in WT and mutant larvae at 5, 6, 7, and 8 dpf. WT PNs showed a significant increase in their TDBL from 5 dpf to 6 dpf, but not at later stages (Figure 6C, generalized linear model [GLM], 5 dpf vs. 6 dpf difference estimate = –2.26e-05, 95% CI = −3.84e-05 to −6.68e-06, z ratio = –3.357, p=0.002; 6 dpf vs. 7 dpf difference estimate = –9.73e-06, 95% CI = –2.25e-05 to 3.00e-06, z ratio = –1.806, p=0.1821; 7 dpf vs. 8 dpf difference estimate = –1.48e-06, 95% CI = −1.35e-05 to 1.05e-05, z ratio = –0.291, p=0.98). gjd2b-/- PNs also showed a similar pattern of branch length growth, with an increase in TDBL from 5 to 6 dpf, followed by no significant increase until 8 dpf (Figure 6C). Post-hoc analysis revealed that there was a significant difference in TDBL between WT and gjd2b-/- on all 4 days, with the mutant group having consistently lower values (Figure 6A–C, GLM, WT vs. mutant difference estimate = 8.17e-06, z ratio = 1.974, p=0.048). PN somata in WT and gjd2b-/- larvae were comparable in diameter (Figure 6—figure supplement 1A, linear model, WT vs. mutant difference estimate = 0.3227, t = 1.935, p=0.055). These analyses reveal that loss of Gjd2b leads to a stunted dendritic arbor having fewer synapses. Since the arbor is stunted in mutants, the synapses that are present on the arbor are electrically closer to the soma compared to WT. This may in turn lead to the increased amplitudes and faster kinetics of mEPSCs recorded from the soma. These results are also consistent with the increased input resistance of mutant neurons compared to WT, seen only under conditions of better dendritic space clamp.

We also quantified the total dendritic branch numbers (TDBNs) of WT and mutant PNs at all 4 days. WT neurons showed a significant increase in branch numbers from 6 dpf to 7 dpf, but not at any other stages (Figure 6—figure supplement 1C, GLM with Poisson error distribution, 5 dpf vs. 6 dpf ratio = 1.132, z ratio = 2.175, p=0.0797; 6 dpf vs. 7 dpf ratio = 1.166, z ratio = 2.862, p=0.012; 7 dpf vs. 8 dpf ratio = 0.949, z ratio = –0.985, p=0.65). gjd2b-/- PNs also showed a similar pattern of branch number growth, with an increase in TDBN from 6 to 7 dpf, followed by no significant increase until 8 dpf (Figure 6—figure supplement 1C). Post-hoc analysis revealed that there was a significant difference in TDBN between WT and gjd2b-/- on all 4 days, with the mutant group having consistently higher values (GLM with Poisson error distribution, WT vs. mutant ratio = 1.16, z ratio = 3.708, p=0.0002). Despite these differences, WT and mutant neurons did not show any difference in the branching order at 7 dpf, as analyzed by Sholl analysis (Figure 6—figure supplement 1B, linear mixed model, WT vs. mutant difference estimate = –0.003208, z = –0.115, p=0.91).

To understand the dynamics that result in stunted and branched arbors, we imaged PNs every hour for 10 hr at 6 dpf. Mutant neurons had a significantly reduced growth rate compared to WT PNs (Figure 6D, linear model, WT vs. mutant difference estimate = 0.025462, t = 2.777, p=0.006). When dissected further into branch elongations and retractions, mutant neurons had a significantly lower rate of branch elongations than WT neurons (Figure 6E, linear model, WT vs. mutant difference estimate = 0.015030, t = 2.307, p=0.022), but their rates of branch length retractions were similar (Figure 6F, WT vs. mutant difference estimate = 0.009621, t = 1.33, p=0.185). This suggests that Gjd2b regulates dendritic growth by promoting branch elongations and is unlikely to be involved in regulating branch retractions.

Functional Gjd2b in PNs alone is sufficient to rescue dendritic growth deficits

We next wished to determine which neurons are electrically coupled to PNs. Electroporation of single PNs with a combination of a high molecular weight dye (tetramethyl rhodamine dextran) and a low molecular weight tracer (neurobiotin/serotonin) failed to reveal any dye-coupled cells. However, when non-PNs were electroporated with neurobiotin or serotonin, one or two PNs along with several non-PNs were detected (Figure 7—figure supplement 1, Supplementary file 1), indicating that PNs are likely to be coupled to other cerebellar cell types via rectifying junctions. To determine if the observed dendritic growth deficits in gjd2b-/- mutant PNs are due to lack of Gjd2b in PNs specifically, we introduced Gjd2b tagged to mCherry into single PNs in gjd2b-/- fish (Figure 7A). TDBL in mutant PNs expressing Gjd2b were significantly larger than mutant PNs and were rescued to WT levels (Figure 7B, ANOVA post-hoc Tukey HSD, difference estimate = –74.6, t-ratio = –5.336, p<0.0001). This suggests that the presence of Gjd2b in single PNs in otherwise Gjd2b null larvae is sufficient to guide dendritic arbor elongation. Heterotypic gap junctional channels formed by Gjd2b on PN membranes and a different connexin isoform in the coupled cell could mediate this process.

Figure 7 with 1 supplement see all
Download asset Open asset
Expressing Gjd2b in Purkinje neurons (PNs) alone is sufficient to rescue dendritic growth deficits.

(A) Representative image of a zebrafish PN expressing cytoplasmic GFP (green) and Gjd2b tagged with mCherry (magenta) at 7 days post fertilization (dpf); scale bar is 5 µm. (B) Total dendritic branch length (TDBL) of wild type (WT), gjd2b-/-, gjd2b-rescue, and gjd2bΔ5-21 rescue (pore dead variant) PNs at 7 dpf. TDBL of gjd2b-rescue neurons is significantly increased from that of mutant PNs. TDBL of gjd2bΔ5-21 rescue PNs is not significantly different from that of gjd2b-/- PNs. (C) Change in the lengths of WT PN dendritic branches with and without Gjd2b-mCherry expression. Elongation of branches with Gjd2b is significantly more than branches without (p<0.05). Retraction of branches is similar with and without Gjd2b-mCherry puncta (Mann–Whitney U test; N = 7 neurons). (D) Copy number of CaMKII mRNA in WT and gjd2b-/- larvae. The normalized copy number of CaMKII in the mutant group is significantly higher than the WT group (p<0.05). (E) TDBL of WT, untreated gjd2b-/-, 1 µM KN-93 treated, 1 µM KN-92 treated, and 1% DMSO-treated gjd2b-/- PNs at 7 dpf. TDBL of only KN-93-treated gjd2b-/- PNs are rescued to WT levels (WT vs. KN-93, p=0.23), whereas all other groups are significantly different and lower than WT (Kruskal–Wallis, post-hoc comparison with Mann–Whitney test). Ns (number of neurons sampled) are indicated in parentheses in (B) and (D). See also Figure 7—figure supplement 1 and Supplementary file 2. Data used for quantitative analyses are available in Figure 7—source data 1.

Figure 7—source data 1

Rescue and Branch dynamics data.

https://cdn.elifesciences.org/articles/68124/elife-68124-fig7-data1-v2.xlsx

To test if functional Gjd2b-mediated gap junctions are required for dendritic elaboration, we generated a construct coding for an N-terminal deleted version of Gjd2b (Gjd2bΔ5-21) as the N-terminus of connexins has been shown to be required for channel function but not for assembly into gap junctional plaques (Kyle et al., 2008). Expression of Gjd2bΔ5-21 in PNs of gjd2b-/- larvae resulted in dendritic arbors that remained stunted and not significantly larger than mutant PNs lacking Gjd2b (Figure 7B, ANOVA post-hoc Tukey HSD, difference estimate = –18.5, t-ratio = –1.506, p=0.4370). These results point to the need for functional gap junctional channels in regulating PN dendritic elaboration.

Presence of Gjd2b puncta on branches promotes their elongation

Next we wished to determine if the presence of Gjd2b puncta is sufficient to promote dendritic arbor growth at the level of single branches. We overexpressed full-length Gjd2b in PNs of WT larvae in a mosaic fashion. We imaged PNs at 5 dpf at 5 min intervals and observed the behavior of single dendritic branches. Dendritic branches having at least one Gjd2b punctum elongated more in length during the 5 min observation windows compared to branches that did not possess any puncta (Mann–Whitney test, U = 3779.5, p=0.022). Presence of Gjd2b puncta did not affect branch retraction lengths during the same window (Figure 7C, Mann–Whitney test, U = 4048, p=0.098). These results are consistent with the loss of Gjd2b affecting dendritic elongation but not retraction (Figure 6E, F). In sum, these results show that functional Gjd2b promotes the elongation of dendritic branches on which it is present.

CaMKII inhibition can also rescue dendritic arbor growth in gjd2b-/-larvae

To further understand how Gjd2b-mediated gap junctions regulate synaptogenesis and dendritic growth, we focused on cytoplasmic binding partners of Gjd2b that are also known to play a significant role in these processes. The calcium and calmodulin-dependent kinase II (CaMKII) has been shown to associate with Cx35/36-mediated gap junctions and modulate their function (Pereda et al., 1998; Alev et al., 2008; Flores et al., 2010; Tetenborg et al., 2017). CaMKII is localized in dendrites and in spines and is a critical regulator of dendritic development (Wu and Cline, 1998; Zou and Cline, 1999; Wayman et al., 2008). We first asked whether expression levels of the various isoforms of CaMKII are altered in gjd2b-/- larvae compared to WT. To our surprise, we observed increased copy numbers for all isoforms of camk2 we tested in gjd2b-/- compared to WT (Figure 7D). CaMKII has previously been shown to reduce dendritic dynamics and stabilize branches (Wu and Cline, 1998; Zou and Cline, 1999). To test if the stunted arbors observed in gjd2b-/- PNs were due to premature stabilization of dendritic arbors, mediated by increased CaMKII activity, we inhibited CaMKII in gjd2b-/- larvae using the drug KN-93 in the embryo medium. Incubation in KN-93 was able to rescue dendritic arbor lengths of PNs in gjd2b-/- larvae to WT levels (Figure 7E). No changes in dendritic arbor lengths were observed in larvae incubated in DMSO (vehicle control) or in KN-92, an inactive analog of KN-93 (Figure 7E). Taken together, these results show that conduction via Gjd2b-mediated electrical synapses is critical for proper glutamatergic synaptogenesis in PNs and that this process affects dendritic arbor growth.

Discussion

Gap junctions in PNs

Gap junction proteins are widely expressed in the cerebellum of developing and adult vertebrates. Cx36, the mammalian homolog of Gjd2b, has been shown to be expressed in the rodent cerebellar cortex in the molecular layer and granule cell layer (Belluardo et al., 2000; Degen et al., 2004; Nagy and Rash, 2017), and Cx36 puncta were observed localized to PNs (Alcami and Marty, 2013), although the mRNA was not found in PNs (Belluardo et al., 2000). In larval zebrafish, cell type-specific transcriptome analysis in larval zebrafish revealed gjd2b expression in PNs, granule cells, eurydendroid cells, and Bergmann glial cells (Takeuchi et al., 2017). We found Gjd2b puncta localized to PN cell membrane, but when dye was injected into PNs, we failed to observe any dye-coupled cells. This may be attributable to the low conductance of zebrafish Gjd2b channels, which have a unitary conductance of around 24 pS (Valiunas et al., 2004). This is also true of Cx36, which have very small unitary conductances of 10–15 pS (Srinivas et al., 1999). Cx36 has been shown to not support dye coupling (Teubner et al., 2000; Quesada et al., 2003). Further, when these same cells were made to overexpress Cx32, which has a larger conductance compared to Cx36, dye coupling could be observed (Quesada et al., 2003). Electrical coupling in the absence of dye coupling has been widely observed (Ransom and Kettenmann, 1990; Meda et al., 1991; Pérez-Armendariz et al., 1991; Moser, 1998; Teubner et al., 2000; Quesada et al., 2003). From these studies, it appears that PNs in fish and mammals are electrically coupled to other neurons, likely mediated by Cx36 in mammals and Cx35 in zebrafish. We suggest that these gap junctions on PNs could serve important developmental functions in all vertebrates.

Regulation of chemical synapse formation by electrical synapses

Our results indicate that Gjd2b-mediated functional electrical synapses are important regulators of glutamatergic synapse formation and dendritic elaboration. These results are in agreement with earlier studies on innexin-mediated gap junctions in invertebrates. Knockdown of the innexin inx1 in leech resulted in loss of electrical coupling between identified neurons at embryonic stages and decreased chemical synaptic strength between the same neurons at much later stages (Todd et al., 2010). In the neocortex of mice, sister neurons born from the same radial glia make transient electrical synapses, which are required for the formation of excitatory connections between them (Yu et al., 2009; Yu et al., 2012). Mice lacking Cx36 exhibit reduced synaptic connectivity between mitral cells in the olfactory bulb (Maher et al., 2009). Interestingly, though Cx36 is the predominant neural connexin, deficits in glutamatergic synapse formation have hitherto not been reported from other regions of the CNS in the Cx36-/- mouse, to the best of our knowledge. However, an increase in the number of inhibitory synapses was observed in Cx36-/- mice in thalamocortical relay neurons with a concomitant decrease in their dendritic complexity (Zolnik and Connors, 2016). Using morpholino-mediated knockdown and knockout approaches, we show that glutamatergic synapse number is decreased significantly when Gjd2b/Cx35b is perturbed. The decrease in synapse number was seen using both structural (transmission electron microscopy) and functional (electrophysiology) assays. In addition, there was a concomitant decrease in dendritic arbor size. The amplitude and kinetics of mEPSCs of mutant PNs followed a trend that was consistent with smaller dendritic arbors. However, alternate explanations such as changes in receptor numbers and subunit types are also possible. We present the results, interpretation, and caveats of all the experiments in this work in tabular form in Supplementary file 4. The collective evidence presented in this paper shows that Gjd2b-mediated functional gap junctions are important regulators of chemical synapse formation and dendritic elaboration in PNs.

Dendritic development of PNs

PNs exhibit one of the most elaborate and beautiful dendritic arbors known, and a number of studies have examined factors that determine arbor structure in PNs. In rodents, PN dendritic development occurs over a period of 1 month after birth and involves multiple steps. Rodent PNs undergo morphological changes soon after they reach the Purkinje cell layer. They retract their simple fusiform dendrites and then acquire a stellate morphology with multiple dendrites sprouting from their somata (Kapfhammer, 2004; Sotelo and Dusart, 2009; Tanaka, 2009). This has also been observed in zebrafish PNs (Tanabe et al., 2010 and Sitaraman and Thirumalai, unpublished observations). Later, one of these becomes the primary dendrite and the others are retracted. This process involves the localization of Golgi organelles at the base of the primary dendrite and is mediated by an atypical PKC (Tanabe et al., 2010). These early steps occur roughly before 4 dpf in zebrafish and before P10 in rodents. The early remodeling is mainly dependent on intrinsic factors, and the overall architecture of PN dendrites is maintained even in the absence of afferent input or activity. In the second postnatal week, the apical dendrite of rodent PNs is spiny and undergoes growth and branching to occupy the molecular layer in a planar manner. By P20 in mice and P30 in rats, PN dendrites have achieved their maximal length (Kapfhammer, 2004; Sotelo and Dusart, 2009). From our results, it appears that significant growth of PN dendritic arbors occurs in zebrafish between 5 and 8 dpf. At these stages, the neurons are spiny and receive parallel and CF inputs (Sengupta and Thirumalai, 2015). Both WT and mutant neurons show large variability in their TDBLs and numbers at all days of observation. Such variability could reflect distinct subtypes within the PN population or could be related to their time of birth. We also observed that at these stages the dendritic branches are dynamic and undergo elongations and retractions. In rodents, lack of afferent input during this phase results in abnormal orientation, reduced size, and lack of higher-order branches in PN dendritic arbors (Altman and Anderson, 1972; Rakic and Sidman, 1973). In both slice cultures and dissociated cell cultures, blockade of glutamatergic transmission reduces dendritic arbor size of PNs (Catania et al., 2001; Adcock et al., 2004). We observed that dendritic arbors of gjd2b-/- PNs were smaller compared to WT even at 5 dpf and stayed smaller at least until 8 dpf. During these stages, they also received fewer glutamatergic mEPSCs, suggesting fewer synaptic contacts. It is likely that the reduced glutamatergic synapses in gjd2b-/- PNs lead to a stunted dendritic arbor via a synaptotrophic mechanism (Haas et al., 2006; Cline and Haas, 2008). It is also likely that Gjd2b directly affects dendritic growth by promoting branch elongations. In gjd2b-/- PNs, lack of Gjd2b puncta results in shorter branch elongations as demonstrated in Figure 7C, and therefore a stunted arbor. A ping-pong mechanism, whereby the smaller dendritic arbor of gjd2b-/- PNs restricts the number of functional synapses that can be formed and the reduced afferent input in turn restricts further dendritic growth, could underlie this process.

We could rescue dendritic arbor growth deficits in Gjd2b mutant zebrafish by expressing full-length Gjd2b in single PNs. In addition, we found that expressing an N-terminal deleted, pore-dead version of Gjd2b could not rescue the dendritic growth deficit. These results suggest that conduction of signaling molecules through Gjd2b-mediated gap junctions regulates dendritic arborization. Further, in WT PNs, Gjd2b puncta facilitate dendritic branch elongation while not affecting branch retractions (Figure 7C). These data together lead us to a model whereby Gjd2b-containing gap junctions conduct signals that promote the elongation of dendritic branches and the formation of glutamatergic synapses locally.

Role of CaMKII in gap junction-mediated PN development

Signaling via Gjd2b-containing gap junctions could lead to long-lasting global changes such as the increased expression levels of α, β1, δ2, γ1, and γ2 isoforms of CaMKII that we observed. Further experiments are required to verify whether this increase in expression level of CaMKII isoforms translates to increased enzymatic activity. Nevertheless, when CaMKII levels and/or activity increase, dendrites are stabilized at their mature lengths. Lack of gap junctional signaling leads to premature stabilization of dendritic branches leading to stunted growth. In Xenopus tectum, immature neurons with simple arbors and low levels of CaMKII continue to grow while mature neurons have high levels of CaMKII and their dendritic structure is more or less stable. Expression of constitutively active CaMKII in tectal neurons causes them to grow slower and have relatively less dynamic arbors. In addition, inhibition of CaMKII in mature neurons causes them to grow at a higher rate (Wu and Cline, 1998). More recently, a human CAMK2A mutation, isolated from an ASD proband, was shown to cause increased dendrite arborization, when the mutant CAMK2A was introduced into cultured mouse hippocampal neurons (Stephenson et al., 2017). Our results are consistent with these earlier findings and suggest a stabilizing role for CaMKII in PN dendritic arbor elaboration. The mechanisms by which signaling via Gjd2b gap junctions regulate CaMKII levels will have to be investigated in future experiments.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Genetic reagent (Danio rerio)gjd2bncb215This paperRRID:ZDB-ALT-201215-7Gjd2b null zebrafish allele ncb215
Recombinant DNA reagentaldoca:gap43-VenusProf. Masahiko Hibi, Nagoya University, Japan
Tanabe et al., 2010		Microinjected in single-cell zebrafish embryos
Recombinant DNA reagentArch:TagRFP-T:PC:GCAMP5GDr. Hideaki Matsui, Niigata University, Japan
Matsui et al., 2014		Microinjected in single-cell zebrafish embryos
Recombinant DNA reagentCa8-cfos:GFPThis paperPN enhancer to drive GFP expressionMicroinjected in single-cell zebrafish embryos
Recombinant DNA reagentCa8-cfos:Gjd2b-mCherryThis paperPN enhancer to drive Gjd2b-mCherry expressionMicroinjected in single-cell zebrafish embryos
Recombinant DNA reagentCa8-cfos:Gjd2b⊗5-21-mCherryThis paperPN enhancer to drive Gjd2b deletion mutantMicroinjected in single-cell zebrafish embryos
Recombinant DNA reagentpTNTPromega Corp, Madison, WIVector for in vitro transcription
Transfected construct (Danio rerio)gjd2b-TALEN-1This paperTALEN constructGAACAGCCATGGGGGAATGGA
Transfected construct (Danio rerio)gjd2b-TALEN-2This paperTALEN constructGCTGTTGGACAGCCGCCTCCA
Commercial assay or kitT7-mMessage mMachineLife TechnologiesFor generating TALEN mRNAs
AntibodyAnti-Cx35/36 (Mouse)MilliporeCat# MAB30451:250
AntibodyAnti-Parvalbumin-7 (mouse)MilliporeMAB15721:1000
AntibodyDonkey anti-mouse Alexa Fluor 488InvitrogenA212021:500
AntibodyGoat anti-rabbit Alexa Fluor 488InvitrogenA210521:1000
OtherProlong gold antifade reagentMolecular ProbesCatalog #P10144Mounting reagent
Sequence-based reagentGjd2b-MOGene ToolsSplice block morpholino5′ACAACACTTTTTCCCCTCACCTCCC3′
Sequence-based reagentCTRLGene ToolsControl morpholino5′ACTAGACTTATTCCCGTGACCTCCC3′
Sequence-based reagentGjd2b forwardThis paperPCR primers5′GATCGGTACCTCCGAATGAACAGCCAT3′
Sequence-based reagentGjd2b reverseThis paperPCR primers5′TAGCGCTAGCAACGTAGGCAGAGTCACTGG3′
Sequence-based reagentGjd2b⊗5-21 forwardThis paperPCR primers5′ATTGCCATGGGGGAATGGATTGGGAGGATCCTGCTAAC3′
Sequence-based reagentGjd2b⊗5-21 reverseThis paperPCR primers5′TAGCGCTAGCAACGTAGGCAGAGTCACTGG3′
Chemical compound, drugMS-222 (Tricaine)Sigma-AldrichCAS #886862
Chemical compound, drugParaformaldehydeAlfa AesarCatalog #47392-9M
Chemical compound, drugTetrodotoxinHello BioCatalog #10691 µM
Chemical compound, drugNMDATocrisCatalog #011410 mM
SoftwareR statistical softwareR core team https://www.r-project.orgRRID:SCR_001905
SoftwareMATLABMathWorks https://www.mathworks.com/products/matlab.htmlRRID:SCR_001622
SoftwareClampfit 10.2Molecular DevicesRRID:SCR_011323
SoftwareFijiNIHhttp://fiji.sc RRID:SCR_002285
SoftwareQuantsoftBio-RadVersion 1.7.4
OtherQX200 AutoDG Droplet Digital PCR systemBio-Rad1864100Equipment
OtherLSM 780 confocal microscopeZeissRRID:SCR_020922Equipment
OtherSP5 point scanning confocal microscopeLeicaEquipment
OtherFV3000 confocal microscopeOlympusRRID:SCR_017015Equipment
OtherUltramicrotomePower Tome-PCEquipment
OtherDiamond knifeElectron Microscopy Sciences
OtherFormavar/Carbon 2 × 1 mm copper or nickel slot gridsElectron Microscopy Sciences
OtherTECNAI T12 G2Spirit BioTWIN transmission electron microscopeFEI CompanyEquipment
OtherBorosilicate glass capillariesWarner InstrumentsOD: 1.5 mm; ID: 0.86 mm
OtherFlaming-Brown P-97 pipette pullerSutter InstrumentsRRID:SCR_020540Equipment
OtherBipolar electrodeFHC, Bowdoin
OtherMulticlamp 700b amplifierMolecular DevicesRRID:SCR_018455Equipment
OtherDigidata 1440A digitizerMolecular DevicesEquipment

Zebrafish and animal husbandry

Request a detailed protocol

All experiments were performed using Indian WT zebrafish. Institutional Animal Ethics and Biosafety committee approvals were obtained for all procedures adopted in this study. Larvae and adults were reared using standard procedures (Westerfield, 2000).

Generation of transient transgenic larvae

Request a detailed protocol

To label single PNs for some of the experiments, single-celled embryos were microinjected with one of the following constructs along with Tol2 transposase mRNA (Urasaki et al., 2006): aldoca:gap43-Venus (Tanabe et al., 2010) (gift from Prof. Masahiko Hibi, Nagoya University, Japan); Ca8-cfos: RFP (Matsui et al., 2014) (gift from Dr. Hideaki Matsui, Niigata University, Japan); Ca8-cfos:GFP; Ca8-cfos:Gjd2b-mCherry; Ca8-cfos:Gjd2bΔ5-21-mCherry. Ca8-cfos: GFP was constructed by amplifying Ca8-cfos from the parent plasmid and ligating it with sequences coding for GFP. To construct the last two plasmids, full-length Gjd2b and Gjd2bΔ5-21 were first amplified from total cDNA using the following primers: Gjd2b forward: 5′GATCGGTACCTCCGAATGAACAGCCAT3′; Gjd2b reverse: 5′TAGCGCTAGCAACGTAGGCAGAGTCACTGG3′; Gjd2bΔ5-21 forward: 5′ATTGCCATGGGGGAATGGATTGGGAGGATCCTGCTAAC3′; Gjd2bΔ5-21 reverse: 5′TAGCGCTAGCAACGTAGGCAGAGTCACTGG3′. The amplified regions were digested using NcoI and NheI and ligated with Ca8-cfos on the 5′ end and mCherry at the 3′ end to generate the respective plasmids. Microinjected embryos were reared in embryo medium containing 0.003% of 1-phenyl-2-thiourea (PTU) for imaging experiments.

Morpholino antisense oligonucleotide-mediated knockdown of Cx35

Request a detailed protocol

Morpholino antisense oligonucleotides (Gene Tools LLC) were designed to bind at the junction between exon 1 and intron 1 of the gjd2b mRNA to block proper splicing of the mRNA (Gjd2b-MO; Figure 1—figure supplement 2A). Control morpholinos (CTRL) were designed to incorporate mismatches at five positions within the gjd2b recognition sequence. The morpholino sequences were:

  • Gjd2b-MO: 5′ ACAACACTTTTTCCCCTCACCTCCC 3′

  • CTRL: 5′ ACTAGACTTATTCCCGTGACCTCCC 3′

Either Gjd2b-MO or CTRL were injected into single-celled zebrafish embryos at 0.05 pmoles per embryo.

Generation of gjd2b-/- zebrafish

Request a detailed protocol

Transcription activator like effector nucleases (TALENs) recognizing nucleotide sequences near the start codon of gjd2b gene were used to generate the gjd2b mutant (gjd2b-/-) lines of zebrafish used in this study (Figure 2). A pair of TALEN vector constructs were designed and assembled to generate gjd2b-TALEN-1 and gjd2b-TALEN-2, which bind the plus and minus strands of gjd2b gene (Figure 2A), respectively, by following published protocols (Sanjana et al., 2012). These TALEN sequences were later moved to a pTNT (Promega Corp, Madison, WI) vector. TALEN mRNAs that encode gjd2b-TALEN-1 and gjd2b-TALEN-2 proteins were synthesized in vitro from the above vectors using the T7-mMessage mMachine kit (Life Technologies, Thermo Fisher Scientific, USA) and micro-injected into one-cell stage WT zebrafish embryos at a concentration of 50 ng/µl of each mRNA. TALENs were designed such that the spacer region incorporated an Xho1 restriction site enabling an easy screen for mutations in this locus. Up to 10 embryos were taken from every clutch of TALEN-injected embryos and screened for the presence of mutations using Xho1 restriction digestion. Clutches of embryos that showed a high percentage of mutants were then grown up into adults. The TALEN-injected founder generation (F0) was raised in the facility and the adult fish were out-crossed with WT fish to get heterozygous F1 progeny. These were screened for germline transmission of mutations in the gjd2b locus. Whole embryos or adult fish tail clips were screened using Xho1 restriction analysis of 800 bp DNA band around the TALEN-target site followed by sequencing of this region (Figure 2B, C). F1 heterozygous mutants (gjd2b+/-) were inbred to obtain F2 WT siblings, heterozygotes, and homozygotes (gjd2b-/-).

Single-cell electroporation

Request a detailed protocol

5 dpf zebrafish larvae were embedded in 1.5% low gelling agarose (Sigma) in a dorsoventral position. The Purkinje cell layer was observed in these larvae under the 63× water immersion objective of a Nikon compound microscope. Patch pipettes (OD: 1.5 mm, ID: 0.86 mm) were pulled using borosilicate glass capillaries and a P-97 pipette puller (Sutter Instruments). A single pipette was backfilled with a mixture of tetramethylrhodamine dextran (TMR-dextran) and serotonin/neurobiotin and inserted through the skin of the larva. The pipette tip was positioned near a cell body in the Purkinje cell layer and 3–5 electric pulses (30 V, 30 ms) were administered, until the neuron was completely filled with the dye. The larva was then released from the agarose and fixed in 4% paraformaldehyde after 30 min and processed for visualizing the injected serotonin or neurobiotin.

Whole-mount immunohistochemistry

Request a detailed protocol

Whole-mount immunofluorescence was performed as described in Jabeen and Thirumalai, 2013. Briefly, 5–7 dpf larvae were anesthetized in 0.01% chilled tricaine (Sigma-Aldrich) and then fixed overnight at 4°C in 4% paraformaldehyde (Alfa Aesar, Thermo Fisher Scientific, UK), followed by several washes in 0.1 M phosphate buffered saline (PBS) at room temperature. The eyes, jaws, and yolk sac were carefully dissected out and the skin covering the brain was peeled to expose the brain. Dissected larvae were kept overnight in 5% normal donkey serum and 0.5% Triton-X100 in 0.1M PBS (PBST) and at 4⁰ C. Next, they were incubated in a mouse anti-Cx35/36 antibody (MAB3045, EMD Millipore, Merck, USA) at a dilution of 1:250 for labeling Gjd2b puncta or rabbit anti-serotonin antibody (Sigma-Aldrich) at a dilution of 1:500 for labeling the serotonin electroporated cells. In larvae in which PNs were stochastically labeled with EGFP, we used the chicken anti-EGFP antibody (Ab13970, Abcam, UK). Primary antibody treatments were done in blocking solutions for 48 hr. After several washes in PBST, larvae were incubated overnight in donkey anti-mouse Alexa Fluor 488 antibody to label Gjd2b puncta or goat anti-mouse Alexa Fluor 633 to label PNs (A21202 or A11034 and A21052, Invitrogen, USA) at a dilution of 1:500 or 1:1000, respectively, in blocking solutions. Following this, the larvae were washed in 0.1 M PBS several times and then mounted between two coverslips using Prolong Gold antifade reagent (Molecular probes, Life Technologies, Thermo Fisher Scientific) and stored in the dark at 4⁰ C until imaging.

Confocal imaging and image analysis

Request a detailed protocol

Images were acquired on these confocal laser scanning microscopes: Zeiss LSM 780 using 63× oil immersion objective, Olympus FV1000 using a 60× oil immersion objective and Olympus FV3000 confocal laser scanning microscope with a 60× oil immersion objective. Imaging parameters and conditions were the same for all larvae imaged in their respective groups. Dorsal-most regions of the cerebellum were imaged. Two images were taken from each animal, one in each hemisphere. Image analysis was done using Fiji (https://fiji.sc/; Schindelin et al., 2012). For Gjd2b puncta analysis, a median filter was applied to remove salt and pepper noise from images after a background subtraction (rolling ball method). Average intensity of the z-projections of image stacks was then measured and normalized with respect to the intensity of uninjected (Figure 1—figure supplement 2) or WT animals (Figure 2). These were then plotted using R statistical software (https://www.r-project.org).

Transmission electron microscopy and image analysis

Request a detailed protocol

Zebrafish larvae at 7 dpf were fixed overnight in 4% paraformaldehyde and 2.5% glutaraldehyde prepared in EM buffer (70 mM sodium cacodylate and 1 mM CaCl2, pH 7.4). Secondary fixation was done with 1% OsO4 made in EM buffer for 90 min on ice. For additional contrast, samples were incubated in aqueous 2% uranyl acetate for 1 hr at room temperature. Samples were dehydrated serially in 30%, 50%, 70%, 90%, and finally in absolute ethanol for 15 min each. After two changes of ethanol (15′ each), samples were incubated in acetone with two changes for 10 min each. Samples were infiltrated with Epon812-Araldite resin mix and acetone in the ratio of 1:3, 1:1, 3:1 and in pure Epon812-Araldite mix for 2 hr, overnight, 2 hr and overnight, respectively. Samples were then embedded in the Epon812-Araldite resin mix for polymerization at 60°C for 48 hr. After polymerization, thick transverse sections were cut using a glass knife till the region of interest was obtained. Then, 60 nm ultrathin sections were cut from the anterior end of the cerebellum on an ultramicrotome (Power Tome-PC, RMC Boeckeler) by using 3.5 size Ultra 45° diamond knife (Electron Microscopy Sciences, USA). Sections were collected on Formavar/Carbon 2 × 1 mm copper or nickel slot grids (FCF 2010-Cu/Ni, Electron Microscopy Sciences) for imaging. Images were acquired on FEI TECNAI T12 G2 Spirit BioTWIN transmission electron microscope. For counting synapses and measuring synapse maturation indices, images were taken at 30K and 50K magnification, respectively. Total volume of a single imaged micrograph was 0.924 cubic micrometers. Images were analyzed using Fiji. Analysis was done blind to the genotype.

Electrophysiology

Request a detailed protocol

Whole-cell patch clamp recording from larval PNs was performed as described in Sengupta and Thirumalai, 2015. Briefly, 7 dpf larvae were anesthetized in 0.01% MS222 and then pinned onto a piece of Sylgard (Dow Corning) glued to a recording chamber. Then larvae were submerged in external saline (134 mM NaCl, 2.9 mM KCl, 1.2 mM MgCl2, 10 mM HEPES, 10 mM glucose, 0.01 D-tubocurarine, 2.1 mM CaCl2; pH 7.8; 290 mOsm). The cerebellum was exposed by peeling the skin from the top of the head. Cells were viewed using the 63× water immersion objective of a fixed stage compound microscope (Nikon Ni-E or Olympus BX61WI). Patch pipettes were pulled from borosilicate glass (OD: 1.5 mm; ID: 0.86 mm; Warner Instruments) on a Flaming-Brown P-97 pipette puller (Sutter Instruments), filled with pipette internal solution, and had a resistance of 10–12 MΩ. For mEPSC recordings, cesium gluconate pipette internal solution was used (115 mM CsOH, 115 mM gluconic acid, 15 mM CsCl, 2 mM NaCl, 10 mM HEPES, 10 mM EGTA, 4 mM Mg-ATP; pH 7.2; 290 mOsm). mEPSCs were recorded after bath application of TTX (1 µM), gabazine (10 µM), and APV (40 µM). Cells were held in voltage clamp at −65 mV for recording AMPAR-mediated currents. In a few cells, the holding potential was varied to measure reversal potential for the mEPSCs. Pipettes also contained sulforhodamine dye (Sigma), and only those cells filled with the dye at the end of the recording and showing Venus expression were considered for further analysis. In addition, cells whose series resistance varied by more than 20% during the recording session or those that had input resistances lower than 1 GΩ were excluded.

For evoked synaptic current recording, potassium gluconate-based internal solution was used (115 mM K gluconate, 15 mM KCl, 2 mM MgCl2, 10 mM HEPES, 10 mM EGTA, 4 mM Mg-ATP; pH: 7.2; 290 mOsm). Evoked synaptic currents were recorded by stimulating CFs with a bipolar electrode (FHC, Bowdoin, ME, USA). Stimulus strength was gradually increased until no failures occurred. PPR was measured by stimulating CFs at various ISIs ranging from 30 ms to 550 ms (in morphants) and 1000 ms (in mutants) and calculating the ratio of the peak amplitude of the second EPSC to the first.

For measuring NMDAR-mediated currents, all recordings were done with 1 µM TTX (Hello Bio catalog no. 1069) to block action potentials and with cesium gluconate internal in the recording pipette. Cells were held at a depolarized potential (−20 mV) to remove the Mg2+ block from the NMDAR. 10 mM NMDA (Tocris catalog no. 0114) was released close to the soma that was being recorded for 500 ms with a Picospritzer. Three trials per cell were recorded with each trial lasting 12 s.

Signals were acquired using Multiclamp 700b amplifier and digitized with Digidata 1440A digitizer (Molecular Devices). Data analysis was done offline in Clampfit 10.2 (Molecular Devices), and statistical analysis was performed using in-built functions in MATLAB (The MathWorks, Natick, MA). mEPSCs were analyzed blind to the experimental group.

In vivo time-lapse imaging and analysis

Request a detailed protocol

For daily imaging, 5 dpf larvae with sparse fluorescent protein labeling of PNs were anesthetized in 0.001% MS222 (Sigma) for 1 min. They were then mounted in 1.5% low gelling agarose (Sigma) on a custom-made confocal dish with their dorsal side towards the coverslip and covered completely with embryo medium. PNs in embedded larvae were imaged using a Leica SP5 point scanning microscope. Neurons were chosen based on their morphology, level of RFP/GFP expression and traceability. Imaging parameters were kept constant across samples. Post imaging, the larvae were released from agarose and allowed to recover in embryo medium for 24 hr. They were imaged again at 6, 7, and 8 dpf. For the hourly imaging, PNs in 6 dpf larvae were imaged at 1 hr intervals for 10 hr. The relationship between Gjd2b puncta and branch dynamics was investigated by imaging PNs in 5 dpf larvae at 5 min intervals for 30 min. For gjd2b and gjd2bΔ5-21 rescue experiments, PNs with both GFP and punctate mCherry in 7 dpf larvae were imaged using an Olympus FV3000 confocal microscope. Neurons were traced using the Simple Neurite tracer plugin in Fiji and their TDBL and TDBN were calculated.

For analyzing the daily imaging data, TDBL and TDBN of the two genotypes, across days, were statistically compared using GLM (inverse Gaussian and Poisson, respectively). For post-hoc analysis, the emmeans package in R was used to compare the TDBL and TDBN of consecutive days within each genotype and that of same days across the two genotypes. Hourly growth rates, elongation rates, and retraction rates of WT and mutant PNs were compared using GLM with Gaussian error distribution. For analyzing the effects of gjd2b and gjd2bΔ5-21 rescue, the TDBL and TDBN across groups were statistically compared using ANOVA and post-hoc Tukey HSD.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files.

References

  34. Book
    1. Nieuwenhuys R
    (1967) Comparative Anatomy of the Cerebellum
    In: Fox C. A, Snider R. S, editors. Progress in Brain Research the Cerebellum. Elsevier. pp. 1–93.
    https://doi.org/10.1016/S0079-6123(08)60962-0
  63. Book
    1. Westerfield M
    (2000)
    The Zebrafish Book: A Guide for the Laboratory Use of Zebrafish
    University of Oregon Press.

Decision letter

  1. Hollis T Cline
    Reviewing Editor; The Scripps Research Institute, United States
  2. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany
  3. Hollis T Cline
    Reviewer; The Scripps Research Institute, United States
  4. Alanna J Watt
    Reviewer; McGill University, Canada

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Acceptance summary:

Gap junctions form between neurons throughout the brain and can provide electrical coupling between neurons, thereby coordinating their activity. Coordinated activity in pre- and postsynaptic neurons is thought to enhance synapse formation. This study demonstrates a role for gap junctions in the development of chemical glutamatergic synapses and dendritic arbor development in Zebrafish cerebellar Purkinje neurons, providing mechanistic insight into early stages of synaptogenesis.

Decision letter after peer review:

Thank you for submitting your article "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Didier Stainier as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

Sitaraman et al., reveal in their manuscript that cerebellar Purkinje neurons (PN) contain gap junctions based on Gjd2b expression prior to and accompanying chemical synapse formation. The authors use morpholino (MO) knockdowns and gjd2b-mutant larval zebrafish to explore the contributions that Gjd2b (and therefore neural gap junctions) makes during dendrite elaboration and synapse formation in Purkinje neurons (PNs). As consequence of these loss of function approaches the number of gap junctions in the cerebellum is reduced which is claimed to lead to a decrease of glutamatergic synapses in PNs. This is approached by electrophysiological studies and further suggested by ultrastructural analysis. Furthermore, a reduced dendritic branch outgrowth is found in gjd2b-mutant PNs, suggesting that synapses are placed closer to the soma in these mutant PNs. Dendrite outgrowth defects can be rescued by functional but not by non-functional Gjd2b-containing gap junctions in PNs, supporting a role for Gjd2b in PN dendrite outgrowth regulation

Furthermore, a correlation between CaMKII signaling, dendritic expansion, and Gjd2b is reported, and pharmacological inhibition of CaMKII can restore the dendritic expansion of the PCs.

The three reviewers agree that the manuscript's concept is novel and essential for the understanding of the cerebellar networks and particular of Purkinje cell development.

For example, the generation of the knocked-down/-out zebrafish where the gap junctions are eliminated is an excellent tool for the cerebellar development and network organization's dissection. Yet, all three reviewers also expressed that the manuscript is too premature to warrant publishing in its current form and that substantial additional data are required.

Essential revisions:

Morphants: The MO experiments are carefully executed and properly controlled, with a clear confirmation of protein knockdown. But the subsequent electrophysiological analysis requires further characterization. For example, a conclusive negative result in the PPR experiment (Figure 2D) is not clearly seen. A consistent separation between the groups across three ISIs from 100-200ms is shown, with an adjusted p value of 0.08 with an n of only 5. This looks like a preliminarily positive result, not a conclusively negative one. The experimental n should be increased to better test the null hypothesis, and/or the alternative interpretation should be presented and discussed.

Also, the interpretation of the peak amplitude needs attention, as the cumulative distributions show different shapes. A greater proportion of control events reach 15pA, but a larger proportion of morphant events reach 40pA. Clear evidence for increased amplitude from these data cannot be deduced.

Given that data from mutant analysis is stronger, the authors could consider to move the morphant analysis to the supplementary part. If the morphant analysis remains in the main body of the manuscript, data in Figure 1 and Figure 3 should be presented with the same scale values to allow for a better comparison between wild type, knock-down and knock-out data.

TALEN mutants: Immunohistochemistry against Cx35/36 clearly shows expression in the membrane of PNs (Figure S1B). In homozygous gjd2b-mutants immunoreactivity against Cx35/36 is not absent but reduced in the cerebellum (Figure S3E, to show exemplary images in addition would be helpful). This is explained by the antibody recognizing both Gjd2a and Gjd2b. It remains unclear then, whether Gjd2b is expressed in PNs. Likewise, if Cx35/36 staining remains in homozygous gjd2b-mutants due to parallel Gjd2a expression, then PNs should still contain a number of gap junctions. This needs to be clarified (see also dye-coupling experiments).

A nice approach is the rescue of dendrite outgrowth by PN-specific overexpression of Gjdb-mCherry in gjdb mutants. This is the only cell-autonomous approach presented. Does this approach also rescue the number of glutamatergic synapses? How about a rescue of electrophysiological properties of PNs?

Finally, it would be interesting to know whether any gross functional consequences/behavioral abnormalities have been observed in homozygous gjdb-mutants.

Electrophysiological recordings: It remains unclear how differences in voltage clamp recordings can be assigned to a reduced number of glutamatergic synapses. Inhibitors should be used to support that the recording inputs are indeed glutamatergic.

An alternative explanation is that the number of synapses could remain the same, but the release is affected; thus, gap junctions can modulate the synaptic strength.

The reported absence of NMDA mediated responses should be supported by data and control experiments. Can the authors record from a PC and apply NMDA in TTX? If there is no increase in the activity, this will be a piece of strong evidence.

Regarding climbing fiber stimulation, the authors should provide further evidence for the specificity of their stimulation to exclude that parallel fibers are not activated as well.

EM analysis: This is a labor-intensive approach that was nicely performed. Yet, the ultrastructural analysis of synapse density and the synapse maturation index in the molecular layer of the cerebellum is not PN specific. A potential contribution of eurydendroid cells to the reduction of synapse number and mature synapses should be taken into account. Also, it remains unclear whether the reduction in synapses is specific to glutamatergic synapses. An immunohistochemical analysis of excitatory synapses e.g. by PSD95-staining in wildtype versus gjd2b-mutants could reveal both whether the number of excitatory synapses in PNs is reduced and whether remaining synapses are indeed placed closer to the soma. Both images for controls and mutants should be shown.

Understanding the labor intensity of EM, and the 1.2um spacing that avoids resampling the same synapses, these data cannot be analysed as an n of (for example) 637 slices. This is an n of 3 animals. Strictly speaking, this is the only analysis that should be done (average value for each animal as a data point). Minimally, these fish-level data should be shown (as a scatter plot) alongside the cumulative probabilities across the numerous slices, with accompanying statistics.

Dye-coupling experiments: Electroporation of rhodamine dextran and neurobiotin into PNs does not cross to other neurons, although electroporation into non PNs occasionally labels PNs. This is not an especially strong argument for functional gap junctions. In addition, dye coupling in knocked-down and knocked-out animals should be shown to verify that the animals are indeed missing gap junctions in Purkinje cells. In addition, it remains unclear whether gap junctions are unidirectional and which cell type is electrically coupled to PNs.

CaMKII-signaling: The expression analysis regarding the relationship between gap junctions and CaMKII signalling is intriguing as it suggests that loss of gap junctions increases CaMKII signalling, which is in turn responsible for stabilizing the dendrites in a smaller/stunted form. This analysis though was performed on total RNA extractions of the entire brain, which does not resolve expression in Purkinje neurons, in addition CaMKII activity could also be regulated at the translational and posttranslational level. While the successful KN-93 rescue places CaMKII signaling downstream of Gjd2b, it could occur in a cell-autonomous or non-cell-autonomous fashion. These investigations lack information about direct or indirect functions of CaMKII and should be considered to be left out.

Statistics: All statistical analysis values are missing, e.g., U value, F values and degrees of freedom, and the actual and adjusted P values. The asterisks are just representative of the actual values and cannot replace the actual statistics.

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the second round of review.]

Thank you for resubmitting your work entitled "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons" for consideration by eLife. Your revised article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

We are sorry to say that, after consultation with the reviewers, we have decided that your work will not be considered further for publication by eLife.

In order to provide orientation about the points of concern and strategies for adding experimental insight, I am listing below a collection of specific concerns raised by individual reviewers:

Immunohistochemistry and TEM data:

The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

The authors report that anti-PSD95 immunohistochemistry analysis results in a higher number of synapse-counts on PNs that is contradictory to their results from TEM analysis and explain this with a limit in resolution by light microscopy (line 231) that does not allow one to unequivocally assign synapses to PNs. But should this erroneously assigning of synapses to PNs be the same in wild type and mutants? Why should this effect occur more often in the gjd2b mutants? This argument remains unclear.

The new PSD95 results are concerning. They are among the strongest results in the paper, and while the authors give two caveats associated with interpreting these data, the overall picture is not, on the whole, supportive of their interpretations.

I do not see that fish-level data have been added to Figure 5, where box plots continue to show data from hundreds of micrographs drawn from a small number of fish.

Electrophysiology:

I remain unconvinced that a conclusive negative result is shown for PPR in Figure 4 (D-G). There is a nonsignificant drop in PPR in the morphants and a sometimes-significant rise in PPR at short ISIs in the mutants. I understand that the distributions overlap, and that this is why most results are not significant, but I do not believe that the results and the experimental n are sufficient to support the claim that this is a negative result. I do not believe that this argument has changed appreciably since the initial submission

The authors suggest that the increase in peak amplitude and faster kinetics of mEPSCs in gjd2b mutants results from excitatory synapses being placed closer to the PN soma, implying a stunted dendritic arbor and loss of distal synapses. This line of arguments is difficult for me to understand. Why should the loss of distal excitatory synapses lead to increases in peak amplitude?

Or do the authors suggest that the stunted dendrites represent "compressed" dendrites in which synapses are moved closer to the soma? Then synapse density close to the soma should be investigated rather than distal synapses.

Then the analysis reveals a reduced total dendritic branch length in gjd2b mutant PNs and confirms a stunted dendritic arbor from which the authors conclude that fewer synapses are placed proximally to the PN somata in gjd2b mutants (line 260). Is this not contradictory to the argument presented in line 240? For me this was confusing and I suggest to reword this paragraph. Did the immunohistochemistry analysis against PSD95 reveal an insight into the localization of excitatory synapses closer to PN somata as suggested in the previous review?

Dye-coupling experiments:

These seem inconclusive, and I am not convinced that the reviewer's concerns have been addressed.

Presentation of Data

The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

In the discussion (line 365-370) the authors emphasize their findings that glutamatergic synapses are decreased based on structural and physiological data but do not mention their contradictory finding with PSD95 immunohistochemistry. This should be discussed more cautiously.

"The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

The authors should try to stimulate the inferior olive to specifically address the input of climbing fibers.

The authors should discuss the discrepancy between their findings and the observation of gap junction coupling between cerebellar neurons recently found in adult zebrafish (Chang et al., 2020). That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

Summary:

Thanks a lot for having submitted a revised version of your manuscript entitled "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons". This revised version has now been seen by three reviewers and they share the opinion that the manuscript's concept is novel and essential for understanding cerebellar circuitry development. Also, all three reviewers were pleased by the new experimental data and changes to the manuscript that have been added by the authors. The manuscript has been improved addressing some of the concerns that have been raised during the initial review. Yet, the different reviews also coincide in the view that the presented data still fall short in being consistently convincing and that the presented study remains premature at this point and regretfully lacks sufficient support for publishing this experimental study.

Reviewer #1:

Reading the revised version of the manuscript "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons" by Sitaraman et al., I am pleased by the new experimental data and changes to the manuscript that have been added by the authors. The data is in most parts presented clearly, but a few points should be addressed to make some of their arguments more clear or to avoid misunderstandings of some of their claims.

1) According to the guidelines of eLife the title of the manuscript should mention the use of zebrafish as model system e. g.: "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in zebrafish Purkinje neurons"

2) The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

3) Line 221: the reduced number of synapses observed in TEM recordings could suggest but does not indicate a reduced number of formed synapses on PNs as this analysis is not able to distinguish between PNs and e.g. eurydendroid cells. I suggest to reword this sentence into "..these results indicate that Gjd2b is involved in the formation of synapses in the molecular layer of the cerebellum to which PN synapses significantly contribute.."

4) The authors report that anti-PSD95 immunohistochemistry analysis results in a higher number of synapse-counts on PNs that is contradictory to their results from TEM analysis and explain this with a limit in resolution by light microscopy (line 231) that does not allow one to unequivocally assign synapses to PNs. But should this erroneously assigning of synapses to PNs be the same in wild type and mutants? Why should this effect occur more often in the gjd2b mutants? This argument remains unclear to me.

5) Line 240: the authors suggest that the increase in peak amplitude and faster kinetics of mEPSCs in gjd2b mutants results from excitatory synapses being placed closer to the PN soma, implying a stunted dendritic arbor and loss of distal synapses. This line of arguments is difficult for me to understand. Why should the loss of distal excitatory synapses lead to increases in peak amplitude?

Or do the authors suggest that the stunted dendrites represent "compressed" dendrites in which synapses are moved closer to the soma? Then synapse density close to the soma should be investigated rather than distal synapses.

Then the analysis reveals a reduced total dendritic branch length in gjd2b mutant PNs and confirms a stunted dendritic arbor from which the authors conclude that fewer synapses are placed proximally to the PN somata in gjd2b mutants (line 260). Is this not contradictory to the argument presented in line 240? For me this was confusing and I suggest to reword this paragraph. Did the immunohistochemistry analysis against PSD95 reveal an insight into the localization of excitatory synapses closer to PN somata as suggested in the previous review?

6) In the discussion (line 365-370) the authors emphasize their findings that glutamatergic synapses are decreased based on structural and physiological data but do not mention their contradictory finding with PSD95 immunohistochemistry. This should be discussed more cautiously.

7) Also, the authors should point out in the discussion that currently besides the cell type specific rescue of PN dendrite outgrowth in gjd2b mutants, they can currently not distinguish between cell-autonomous and non-cell autonomous effects of gjd2b loss on PNs. This should be made clear to the reader.

Reviewer #2:

The manuscript is now improved in several aspects. The authors performed additional experiments to verify their observations. They followed most of the reviewers' recommendations, yet the overall impression is that the Authors did not consider some critical comments. Specifically, the ones that aimed to improve the data's presentation.

Specifically:

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

Regarding the question of the specificity of the CF stimulation, the answer is not convincing. The fact that the authors used this approach in the past is not enough to claim specific. Why did the authors not try to stimulate the inferior olive?

Regarding the Dye-coupling experiments where the authors state that they do not observe any dye coupling between PCs. In light of the new paper published in PNAS (Chang et al., 2020), the authors should discuss this discrepancy between their findings and what is observed in adult zebrafish. That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

Regarding our previous comment: "The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

Recommendations for the authors:

The manuscript is now improved in some aspects; however, the overall impression that I have is that the Authors did not take into consideration all the comments, and I found somehow quite lousy the revision of the previous manuscript. Specifically:

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

Regarding the question of the specificity of the CF stimulation, the answer is not convincing. The fact that the authors used this approach in the past is not enough to claim specific. Why did the authors not try to stimulate the inferior olive?

Regarding the Dye-coupling experiments where the authors state that they do not observe any dye coupling between PCs. In light of the new paper published in PNAS (Chang et al., 2020), the authors should discuss this discrepancy between their findings and what is observed in adult zebrafish. That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

Regarding our previous comment: "The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

Reviewer #3:

As outlined in my original review of this manuscript, I view it as interesting and potentially impactful, but preliminary in its interpretations and not consistently convincing. I do not view this has having changed in a meaningful way with these revisions. The new data provided do not particularly strengthen conclusions, and in one case conflict with the manuscript's narrative. Most reviewers' comments have been addressed without further experiments, but not in a way that is consistently satisfying. I address these revisions below, broken down by sections of the rebuttal letter.

Morphants:

I remain unconvinced that a conclusive negative result is shown for PPR in Figure 4. There is a nonsignificant drop in PPR in the morphants and a sometimes-significant rise in PPR at short ISIs in the mutants. I understand that the distributions overlap, and that this is why most results are not significant, but I do not believe that the results and the experimental n are sufficient to support the claim that this is a negative result. I do not believe that this argument has changed appreciably since the initial submission.

I accept the authors' assertion that increased peak amplitude for the mutant is convincingly demonstrated in Figure 3D.

I continue to think that the mutant provides stronger support for the authors' claims than the MOs do.

TALEN Mutants:

I am generally convinced by the authors' responses to these questions and with the added data, although I defer to the reviewer who originally raised the issue of the two orthologs with regard to whether this has been adequately addressed.

Electrophysiological recordings:

As described above, I continue not to be convinced of a negative result for the PPD data shown in Figure 4D-G.

The arguments presented about blockers and the new NMDA pulse data appear to be valid to me, but I defer to the relevant reviewer.

EM Analysis:

I understand the converging lines of evidence that the authors refer to (ephys and EM), and understand that these could be viewed as complementary, given each line's strengths and caveats. This does not really address the reviewer's concern, although I defer to him/her with regard to whether they are convinced.

The new PSD95 results are concerning. They are among the strongest results in the paper, and while the authors give two caveats associated with interpreting these data, the overall picture is not, on the whole, supportive of their interpretations.

I do not see that fish-level data have been added to Figure 5, where box plots continue to show data from hundreds of micrographs drawn from a small number of fish.

Dye-coupling experiments:

These seem inconclusive, and I am not convinced that the reviewer's concerns have been addressed.

CaMKII-signalling:

This was an interesting but not fully supported element of the original manuscript. I agree with the decision to withdraw these data, but it leaves a less impactful paper.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Hollis T Cline as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Didier Stainier as the Senior Editor. The following individual involved in review of your submission have agreed to reveal their identity: Alanna J Watt (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential Revisions:

Please modify the text of the paper to identify caveat, alternate interpretations and open questions for future research, as suggested in the reviewers comments, below.

Reviewer #1 (Recommendations for the authors):

1. It would be more clear to name the splice-blocking morpholino as 'Gjd2b-MO'

2. Control morpholinos were tested separately in experiments using immunolabeling, but were not used in the electrophysiology experiments. This is unusual but acceptable.

3. Figure 1. Supplement 1. Please label columns and rows in panel B.

4. For morphants, please state if experimental and control animals were from the same clutch and how many clutches of embryos were used for each experiment.

5. Figure 4. Panel B, list specific ages in legend, rather than 'mixed ages'. Panel C, state that this is done on the presence of TTX.

6. In the first paragraph of discussion, authors should more clearly state that they think that conductance through Cx36/Gjd2b-mediated gap junctions is not be sufficient to pass dye and they are therefore not able to identify cells coupled to PCs.

7. Given the ambiguity about the sequence of events governing branch extension gap junction formation -> synaptogenesis->branch extension, repeat versus gap junction formation -> branch extension-> synaptogenesis, repeat., the interpretation on line 242 'resulting in fewer synapses' is overstated.

8. The authors state that it is likely that gap junctions promote dendritic arbor growth directly, independent of their actions on chemical synapses. What evidence (citation) supports this statement?

9. On pg 19 the first full paragraph is repeated in the next section.

10. Figure 7, supplement 1. There is a problem with the labeling

Reviewer #2 (Recommendations for the authors):

Electrophysiology data:

1. The data on mEPSC frequency is clear evidence for fewer excitatory synapses. However, interpreting larger amplitudes and faster decays as an explanation for the loss of distal synapses is problematic. The shift towards larger amplitudes can also be explained simply by the absence of weaker synapses rather than arguing for the absence of distal synapses. Moreover, larger amplitudes can also be a homeostatic scaling mechanism to compensate for fewer synapses. Regarding the decay, this evidence would be more compelling if these experiments were done in current clamp where the absence of gap junctions would lead to longer voltage decay as a result of increased input resistance of neurons. In contrast, the faster current decay observed by the authors could also be a result of different subunit composition or other biophysical considerations (Laurence et al., Nature Neuroscience, 2005; Kumar et al., JNeuroscience 2002), and unless that is ruled out explicitly, I would recommend adding this as an important alternative explanation in the discussion.

2. What is the input resistance of mutant neurons versus wildtype Purkinje neurons? Given the smaller dendritic arbors and the absence of gap junctions, one would expect Gjdb2-/- neurons have a higher input resistance. It will be good to see these data.

3. In comparing the WT data in Figure 1D,E,F with Figure 3D,E,F, I am confused by differences in the distribution of WT data between these figures. There is a long tail in the distribution of WT data in Figure 1D,E,F which is absent in Figure 3D,E,F. I am curious why this is the case.

EM data:

1. I agree with the other reviewers that while the EM data are beautiful and hard to collect and analyze, not being able to attribute synapses to PNs significantly limits the conclusions one can reach from this experiment. I do not know much about cerebellar circuitry but is there any estimate for what fraction of excitatory synapses are formed on PNs versus other neurons? Also, the inability to distinguish between excitatory and inhibitory synapses is a major limitation. Between the EM and electrophysiology data, I would argue that the electrophysiology data are far stronger. I understand that the authors see this as converging evidence, but in my opinion, the EM data have substantial caveats and at best provide weak support for the conclusions the authors are trying to reach. If, for instance, the authors have data on spontaneous IPSCs in mutant and WT neurons that are similar in frequency, that can at least help argue that the numbers of inhibitory synapses are similar.

2. I have not seen the PSD95 data, but I agree with the authors that PSD95 staining is not a compelling experiment. Light microscopy resolution is a major challenge. If anything, the authors could have tried expressing PSD95-tagged GFP in individual neurons in mutant and WT fish as was done in Niell et al. (Nature Neuroscience 2004) in the zebrafish optic tectum. I am not suggesting that the authors do this experiment but wanted to just throw in support for their argument that PSD95 staining is inconclusive.

Dendritic elaboration and CaMKII

1. The gain of function experiments are interesting, but perhaps I am missing something here. My understanding is that functional electrical synapses need the assembly of a pore in the presynaptic and post-synaptic neuron. So, how would expressing Gjd2b in one neuron ensure functional electrical connectivity with other neurons? I see this is addressed in the limitations document, but the authors' reliance on pore dead experiments is unconvincing. If anything, the pore dead neurons have dendrites that are comparable to WT neurons and longer than the Gjd2b neurons (Figure 7b). I think the authors need to do a statistical analysis of differences between WT and Gjd2b rescue as well as between WT and Gjd2b pore-dead mutants. I think there is something interesting there that might allude to functional aspects of non-pore forming regions of Gjd2b. In the absence of clear experiments to demonstrate functional electrical synapses, I think this experiment falls significantly short of implicating electrical synapses in dendrite elaboration. If the authors do want to make this claim, in the very least they need to show that the "rescued" neurons have comparable excitatory synapses

2. A lot of the work on dendritic elaboration has parallels with the rich body of work done in Hollis Cline's lab which the authors reference extensively. However, it's not clear to me that the dendritic effects are not simply a downstream consequence of the absence of Gjd2b rather any information transmitted through the electrical synapses. Since Gjd2b knockout reduces the number of AMPAR synapses based on the electrophysiology, isn't a simple explanation for all the dendritic effects simply a consequence of fewer AMPAR synapses as shown in Haas et al. (PNAS, 2006). Moreover, given the lack of direct evidence that the rescue experiments lead to functional electrical synapses, I am not convinced that molecules transmitted through gap junctions are somehow responsible for elevated CaMKII.

In summary, while this paper represents a substantial amount of work and relies on converging lines of evidence to arrive at their conclusions, there are several limitations within each technique and these shortcomings are not addressed by the complimentary experiments.

The authors present good evidence for fewer chemical synapses and shorter dendrites in Purkinje neurons in fish where Gjd2b dependent electrical synapses are knocked down or knocked out. The concerns about electrophysiology data can be addressed in the discussion as an important caveat.

However, my bigger concern is with disambiguating Gjd2b mediated changes in dendritic structure from downstream effects of simply having fewer AMPAR synapses. Previous work has provided compelling evidence that chemical transmission through AMPAR synapses is a key driver of dendritic elaboration. So, if there are fewer AMPAR synapses, is it not unsurprising that the dendrites are smaller and that has nothing to do directly with Gjd2b function? Perhaps I am missing a key piece of the argument here and would be happy to be proven wrong.

Reviewer #3 (Recommendations for the authors):

Having read through the previous response to the reviewers, I think that the authors have addressed them very well. I do not think further experiments or analyses are required.

https://doi.org/10.7554/eLife.68124.sa1

Author response

Essential revisions:

Morphants: The MO experiments are carefully executed and properly controlled, with a clear confirmation of protein knockdown. But the subsequent electrophysiological analysis requires further characterization. For example, a conclusive negative result in the PPR experiment (Figure 2D) is not clearly seen. A consistent separation between the groups across three ISIs from 100-200ms is shown, with an adjusted p value of 0.08 with an n of only 5. This looks like a preliminarily positive result, not a conclusively negative one. The experimental n should be increased to better test the null hypothesis, and/or the alternative interpretation should be presented and discussed.

Also, the interpretation of the peak amplitude needs attention, as the cumulative distributions show different shapes. A greater proportion of control events reach 15pA, but a larger proportion of morphant events reach 40pA. Clear evidence for increased amplitude from these data cannot be deduced.

Given that data from mutant analysis is stronger, the authors could consider to move the morphant analysis to the supplementary part. If the morphant analysis remains in the main body of the manuscript, data in Figure 1 and Figure 3 should be presented with the same scale values to allow for a better comparison between wild type, knock-down and knock-out data.

To conclusively say if the paired pulse ratios in PNs differ after Gjd2b manipulation, we decided to repeat the PPD experiment in WT and gjd2b-/- zebrafish larvae. We recorded climbing fibre evoked responses from PNs in 7dpf larvae, at inter stimulus intervals varying from 30-1000 ms (for details, see Materials and methods). We observe robust paired pulse depression in wildtype PNs, as is seen by the significant difference in paired pulse ratios at 30,35,40,50,70 ms inter-stimulus intervals in comparison to all longer intervals (Figure 4). We also do not observe any significant difference between wildtype and mutant PPRs at any of the ISIs, except at 40ms (p value 0.03). Though average values of the two genotypes show separation, the distributions are largely overlapping. This indicates that Gjd2b manipulation either by knock-out or knock-down approaches does not affect the neurotransmitter release probability at climbing fibre-Purkinje neuron synapses.

The peak amplitude of mEPSCs shows an increase in morphants compared to control, as seen by a shift to the right of the morphant distribution relative to control for about 90% of all events. This can also be seen in the box plots that we have now added to this figure. We disagree that this is not clear evidence for an increase in amplitude.

The scales for plots in Figure 1 and 2 have been matched.

TALEN mutants: Immunohistochemistry against Cx35/36 clearly shows expression in the membrane of PNs (Figure S1B). In homozygous gjd2b-mutants immunoreactivity against Cx35/36 is not absent but reduced in the cerebellum (Figure S3E, to show exemplary images in addition would be helpful). This is explained by the antibody recognizing both Gjd2a and Gjd2b. It remains unclear then, whether Gjd2b is expressed in PNs. Likewise, if Cx35/36 staining remains in homozygous gjd2b-mutants due to parallel Gjd2a expression, then PNs should still contain a number of gap junctions. This needs to be clarified (see also dye-coupling experiments).

A nice approach is the rescue of dendrite outgrowth by PN-specific overexpression of Gjdb-mCherry in gjdb mutants. This is the only cell-autonomous approach presented. Does this approach also rescue the number of glutamatergic synapses? How about a rescue of electrophysiological properties of PNs?

Finally, it would be interesting to know whether any gross functional consequences/behavioral abnormalities have been observed in homozygous gjdb-mutants.

Corroborating our IHC results, the study by Takeuchi et al., 2016 shows that both Gjd2a and Gjd2b transcripts are expressed in zebrafish Purkinje neurons (https://doi.org/10.1002/cne.24114). This study looks at the transcript levels of various genes in 4 cell types of the zebrafish cerebellum: granule cells, Purkinje neurons, eurydendroid cells and Bergmann glia; and also in neurons of the inferior olive. All 3 biological replicates of Purkinje neuron samples show expression of the Cx35b/Gjd2b transcript.

Exemplar images on Cx35 IHC from WT and mutant larvae have been included in figure 2.

We acknowledge that knocking out Gjd2b will not rid PNs of all their gap junctions. Keeping that in mind, we only claim that Gjd2b gap junctions affect glutamatergic synapse formation and dendritic arborisation in PNs. There may be similar roles for other gap junction proteins not studied here.

While it would have been further confirmation of our results to demonstrate rescue of the number of glutamatergic synapses using TEM or electrophysiology, these are technically difficult and low yield experiments. The gjd2b-mCherry construct is expressed in a mosaic fashion in isolated PNs to ascertain cell-autonomous rescue and such neurons are difficult to target specifically with TEM or electrophysiology. In the manuscript, we make specific and limited claims regarding the rescue experiment with respect to the dendritic growth phenotype and do not extrapolate it to the number of glutamatergic synapses. We discuss a possible rescue approach using PSD95 IHC and its caveats in the section below.

Qualitatively, we do not observe any stark changes in the behaviour of gjd2b-/- fish, as compared to wildtype larvae. In-depth, quantitative assays to determine if finer behavioural aspects are affected by loss of Gjd2b is out of scope of this paper.

Electrophysiological recordings: It remains unclear how differences in voltage clamp recordings can be assigned to a reduced number of glutamatergic synapses. Inhibitors should be used to support that the recording inputs are indeed glutamatergic.

An alternative explanation is that the number of synapses could remain the same, but the release is affected; thus, gap junctions can modulate the synaptic strength.

The reported absence of NMDA mediated responses should be supported by data and control experiments. Can the authors record from a PC and apply NMDA in TTX? If there is no increase in the activity, this will be a piece of strong evidence.

Regarding climbing fiber stimulation, the authors should provide further evidence for the specificity of their stimulation to exclude that parallel fibers are not activated as well.

All voltage clamp recordings were performed in the presence of a cocktail of blockers to block any GABA and NMDA mediated minis. This is mentioned in the ‘Electrophysiology’ section of methods, line 518-519, “mEPSCs were recorded after bath application of TTX (1μM), Gabazine (10μM) and APV (40μM)”. The recorded mEPSCs constitute AMPAR-mediated glutamatergic currents.

To check if a change in probability of release can explain the change in mEPSC frequency, we stimulated CFs and measured paired pulse ratios of the CF to PN synapse. We have demonstrated previously that stimulation of CFs evokes synaptic currents in PNs which have markedly higher amplitudes than those evoked by parallel fibers (eLife 2015;4:e09158). In both mutants (new data, figure 4F and G) and morphants, we failed to find significant changes in the paired pulse ratios when compared to control. These findings suggest that the observed increase in mEPSC frequency probably arises due to changes in synapse numbers, which we then confirm using ultrastructural analysis.

As suggested by the reviewers, we have now added new experiments where we pulsed NMDA and measured PN responses in wild type and mutant larvae. With TTX in the bath, brief puffs of NMDA applied to PNs held at -20mV failed to elicit any responses in wild type or mutant larvae (figure 4C).

EM analysis: This is a labor-intensive approach that was nicely performed. Yet, the ultrastructural analysis of synapse density and the synapse maturation index in the molecular layer of the cerebellum is not PN specific. A potential contribution of eurydendroid cells to the reduction of synapse number and mature synapses should be taken into account. Also, it remains unclear whether the reduction in synapses is specific to glutamatergic synapses. An immunohistochemical analysis of excitatory synapses e.g. by PSD95-staining in wildtype versus gjd2b-mutants could reveal both whether the number of excitatory synapses in PNs is reduced and whether remaining synapses are indeed placed closer to the soma. Both images for controls and mutants should be shown.

Understanding the labor intensity of EM, and the 1.2um spacing that avoids resampling the same synapses, these data cannot be analysed as an n of (for example) 637 slices. This is an n of 3 animals. Strictly speaking, this is the only analysis that should be done (average value for each animal as a data point). Minimally, these fish-level data should be shown (as a scatter plot) alongside the cumulative probabilities across the numerous slices, with accompanying statistics.

We present complementary lines of evidence using electrophysiology and EM analysis for the change in synapse number following gjd2b manipulation. The electrophysiology experiment is specific for glutamatergic synaptic inputs and the decrease in mEPSC frequency is suggestive of decrease in synapse numbers. The EM analysis provides definitive counts of synapse numbers in the molecular layer where PNs elaborate dendrites and make synapses. Yet, the EM approach is not specific for PNs or for glutamatergic synapses. Taken together, these two complementary lines of evidence indicate a reduction in glutamatergic synapse density.

We performed immunohistochemical staining of PSD95 in wildtype and mutant larvae as suggested by the reviewers. We identified PNs with Parv7 or EGFP counterstaining. We quantified the numbers of PSD95 puncta colocalized with Parvalbumin-7/EGFP and compared the values between wild type and mutant larvae. We found that the number of PSD95 puncta was increased significantly in mutants compared to wild type (Figure 5 —figure supplement 1). While this result is inconsistent with our EM and electrophysiology results, it also suffers from these caveats:

1.Though we're only counting puncta which are colocalised with Parv7 positive neurons, the resolving limit of light microscopy can cause us to include puncta which are present in adjacent neurons as well. Therefore, this method does not guarantee specificity for synapses on PNs.

2. Presence of a PSD95 punctum does not guarantee the presence of a functional synapse. The immunostaining technique will label synaptic as well as non synaptic clusters of PSD95 that are either at nascent synapses or other intracellular membrane organelles (El-Husseini et al., J. Cell Biol., 2000). We do not know if loss of Gjd2b causes PSD95 to accumulate in these non-synaptic or immature synaptic clusters.

Including a pre-synaptic marker and performing super resolution microscopy will give more reliable results. But such a study is beyond the scope of this paper.

Seeing that this experiment did not recapitulate our EM results, we did not continue with assessing the rescue capabilities of Gjd2b-mCherry on glutamatergic synapses using PSD95 IHC.

Fish-level data for the TEM has been shown as a boxplot with scatter, alongside the cumulative probabilities, in Figure 5.

Dye-coupling experiments: Electroporation of rhodamine dextran and neurobiotin into PNs does not cross to other neurons, although electroporation into non PNs occasionally labels PNs. This is not an especially strong argument for functional gap junctions. In addition, dye coupling in knocked-down and knocked-out animals should be shown to verify that the animals are indeed missing gap junctions in Purkinje cells. In addition, it remains unclear whether gap junctions are unidirectional and which cell type is electrically coupled to PNs.

We used dye-coupling approaches to establish if PNs make functional gap junctions. We have tested this using neurobiotin (323 Da) and serotonin (176 Da). Both approaches gave us the same result, which is that when Purkinje neurons were filled with the dye, we did not observe any coupled neurons. In both experiments, electroporation of other cerebellar neurons resulted in coupled cells taking up the dye, demonstrating the effectiveness of our methodology. In all trials where non-PNs were electroporated, at least a couple of PNs showed uptake of the dye from the electroporated neuron (Figure 7 —figure supplement 1 and Supplementary file 1). While this is not confirmatory proof, it is suggestive of rectifying gap junctions and we present it as such in the manuscript. The identity of the cell type coupled to PNs awaits future experimentation.

The absence of coupling among PNs and the low numbers of coupled PNs observed after electroporating non-PNs preclude attempts to verify their absence or reduction in the mutants and morphants respectively.

CaMKII-signaling: The expression analysis regarding the relationship between gap junctions and CaMKII signalling is intriguing as it suggests that loss of gap junctions increases CaMKII signalling, which is in turn responsible for stabilizing the dendrites in a smaller/stunted form. This analysis though was performed on total RNA extractions of the entire brain, which does not resolve expression in Purkinje neurons, in addition CaMKII activity could also be regulated at the translational and posttranslational level. While the successful KN-93 rescue places CaMKII signaling downstream of Gjd2b, it could occur in a cell-autonomous or non-cell-autonomous fashion. These investigations lack information about direct or indirect functions of CaMKII and should be considered to be left out.

Acting on advice from the reviewers, we have decided to remove these data from this manuscript.

Statistics: All statistical analysis values are missing, e.g., U value, F values and degrees of freedom, and the actual and adjusted P values. The asterisks are just representative of the actual values and cannot replace the actual statistics.

All statistical values have been provided in the Results section. A separate statistical summary document is also provided.

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the second round of review.]

In order to provide orientation about the points of concern and strategies for adding experimental insight, I am listing below a collection of specific concerns raised by individual reviewers:

Immunohistochemistry and TEM data:

The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

We mentioned this in the corresponding figure legend (line 971), and have made the suggested modification in the main text (lines 110, 121 and 125).

The authors report that anti-PSD95 immunohistochemistry analysis results in a higher number of synapse-counts on PNs that is contradictory to their results from TEM analysis and explain this with a limit in resolution by light microscopy (line 231) that does not allow one to unequivocally assign synapses to PNs. But should this erroneously assigning of synapses to PNs be the same in wild type and mutants? Why should this effect occur more often in the gjd2b mutants? This argument remains unclear.

The main issue with the EM data pointed out by the reviewers was that it was neither specific to PNs nor to glutamatergic synapses. However, the experiment suggested by the reviewers does not address either of those issues unequivocally:

The PSD95 puncta observed in the IHC experiment can be of three kinds:

1. Synaptic puncta on PNs

2. Non-synaptic puncta on PNs

3. Synaptic and non-synaptic puncta in non-PNs located very close to PNs.

Nevertheless, in an attempt to address reviewer concerns, we performed PSD95 immunohistochemistry with PN labeling and quantified PSD95 puncta on PNs in wild type and mutant larvae. We saw an increase in the number of PSD95 puncta in gjd2b mutants. While this may be taken to mean an increase in synaptic puncta in PNs in the mutants, the other equally likely possibilities are:

1. An increase in PSD95 accumulation at non-synaptic sites in PNs due to the mutation.

2. An increase in PSD95 accumulation in synaptic and/or non-synaptic sites in nonPNs also due to the mutation in Gjd2b.

Due to the limit in resolution of light microscopy, PSD95 puncta cannot be assigned to PNs with certainty. If the Gjd2b mutation affects PSD95 expression levels in neighbouring cells or promotes its accumulation intracellularly, this error in assignment will be greater in mutants compared to wild type.

Therefore, the results from the PSD95 experiment are inconclusive, while those from the electrophysiology and TEM analysis point clearly to a decrease in PN synapse numbers when Gjd2b is impaired. Since the new PSD95 data do not provide categorical answers and suffer from major caveats, we have removed this experiment from the current version.

The new PSD95 results are concerning. They are among the strongest results in the paper, and while the authors give two caveats associated with interpreting these data, the overall picture is not, on the whole, supportive of their interpretations.

We disagree that the PSD95 results are among the strongest in the paper. In our original submission, reviewers had two concerns in the EM data about the lack of specificity to either PNs or glutamatergic synapses. They suggested quantifying PSD95 puncta on PNs as a way of addressing these concerns. However, for reasons mentioned above, this experiment does not categorically answer either the specificity to PNs or that all puncta are glutamatergic synapses. In our earlier response, we clearly explained how the EM and the mEPSC data have to be taken together as complementary approaches, one ultrastructural and the other functional. We are extremely disappointed that the reviewers have discounted all of the other experiments presented in the manuscript and focused on one experiment with equal number of caveats. We do not believe that the PSD95 results have completely negated all of the other experimental results. We now include a table that shows the main result, interpretation and caveats of all the experiments presented in the manuscript (Supplementary file 4).

I do not see that fish-level data have been added to Figure 5, where box plots continue to show data from hundreds of micrographs drawn from a small number of fish.

As mentioned in the manuscript, the EM data were derived from sectioning the brains of 3 fish each in wild type and mutants. It will not be meaningful to plot 3 data points. We present fish level means and SEMs in the form of a table (Supplementary file 1).

Electrophysiology:

I remain unconvinced that a conclusive negative result is shown for PPR in Figure 4 (D-G). There is a nonsignificant drop in PPR in the morphants and a sometimes-significant rise in PPR at short ISIs in the mutants. I understand that the distributions overlap, and that this is why most results are not significant, but I do not believe that the results and the experimental n are sufficient to support the claim that this is a negative result. I do not believe that this argument has changed appreciably since the initial submission

We respectfully disagree with the reviewer. The n’s we have for the PPR experiments are 14 for wild type PNs and 13 for the mutants. These n’s are typical for PPR experiments and perhaps even higher than what is reported in the literature. The p-value we obtained was 0.17, which is not even borderline. The distributions are overlapping and the statistical testing did not yield significance. We cannot reject the null hypothesis. We do not understand how a visual difference in the mean can override rigorous statistical analysis.

We want to point you to some recent papers in eLife where PPR experiments have been reported for comparisons on n’s used.

DOI: 10.7554/eLife.45920: Figure 2—figure supplement 2, n = 8 and 9 cells, ns, p = 0.08.

DOI: 10.7554/eLife.31755: Figure 4—figure supplement 4, n = 7 and 8 cells, ns, p value not reported.

DOI: 10.7554/eLife.36209: Figure 4: n = 5 and 6 cells, ns, p = 0.97

DOI: 10.7554/eLife.33892: Figure 2B and D: n = 6 cells, ns, p=0.156

The authors suggest that the increase in peak amplitude and faster kinetics of mEPSCs in gjd2b mutants results from excitatory synapses being placed closer to the PN soma, implying a stunted dendritic arbor and loss of distal synapses. This line of arguments is difficult for me to understand. Why should the loss of distal excitatory synapses lead to increases in peak amplitude?

mEPSCs arriving from synapses placed closer to the soma will be filtered less by the cable while those arriving from synapses on distal dendrites will be filtered more, resulting in slower kinetics and attenuated amplitudes. In neurons with stunted arbors, these distally placed synapses will be absent or reduced in number. Therefore, the distribution of mEPSCs moves towards larger amplitudes and faster kinetics.

Or do the authors suggest that the stunted dendrites represent "compressed" dendrites in which synapses are moved closer to the soma? Then synapse density close to the soma should be investigated rather than distal synapses.

No we are not suggesting this at all. There is no need to assume changes in synapse placement or synapse density on individual branches.

Then the analysis reveals a reduced total dendritic branch length in gjd2b mutant PNs and confirms a stunted dendritic arbor from which the authors conclude that fewer synapses are placed proximally to the PN somata in gjd2b mutants (line 260). Is this not contradictory to the argument presented in line 240? For me this was confusing and I suggest to reword this paragraph.

Apologies, this seems to be a problem with the wording. We meant that there are fewer synapses and that they are placed closer to the soma. We have rewritten this sentence to clarify (lines 254-257).

Did the immunohistochemistry analysis against PSD95 reveal an insight into the localization of excitatory synapses closer to PN somata as suggested in the previous review?

Since we are not suggesting that there is change in synapse placement, this point is moot.

Dye-coupling experiments:

These seem inconclusive, and I am not convinced that the reviewer's concerns have been addressed.

We performed dye-coupling experiments with neurobiotin (323 Da) and serotonin

(177 Da). Neither probe could reveal the coupled partners of PNs when injected into PNs. When injected into non-PNs, we were able to label at most 2 coupled PNs. We attribute the low efficiency of dye coupling that we observed to the low unitary conductance of Gjd2b gap junction channels.

Zebrafish Gjd2b channels have a unitary conductance of around 24pS (Valiunas et al., 2004). This is also true of Cx36, the mammalian ortholog of Gjd2b, which have very small unitary conductances of 10-15pS (Srinivas et al., 1999). Cx36 has been shown to not support dye coupling (Teubner et al., 2000; Quesada et al., 2003). Further, when these same cells were made to overexpress Cx32, which has a larger conductance compared to Cx36, dye coupling could be observed (Quesada et al., 2003). Electrical coupling in the absence of dye coupling has been widely observed (Ransom and Kettenmann, 1990; Meda et al., 1991; Pérez-Armendariz et al., 1991; Moser, 1998; Teubner et al., 2000; Quesada et al., 2003). We would like to place these facts before the reviewers and want to suggest that the low conductance of Gjd2b channels might be the reason we are not able to observe dye coupling.

Presentation of Data

The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

Already addressed above.

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

This was due to a misunderstanding on what the reviewer was asking us to do. We assumed that the reviewer was referring to all of the cumulative probability plots shown in Figures1 and 3 and we matched the scales on all the plots shown in Figures1 and 3. Representative traces are usually shown at a scale that best shows the detail in each case. Even so, as you will see from the scale bars, they are roughly of the same scale. The scale bar in 1A is roughly twice the size of 3A with twice the value. The 10ms label for 3A is a typo. We apologize and have now fixed it.

In the discussion (line 365-370) the authors emphasize their findings that glutamatergic synapses are decreased based on structural and physiological data but do not mention their contradictory finding with PSD95 immunohistochemistry. This should be discussed more cautiously.

We have removed the PSD95 data for reasons explained above.

"The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

We have removed the holding potential label in the figure and mention it in the legends.

The authors should try to stimulate the inferior olive to specifically address the input of climbing fibers.

The authors should discuss the discrepancy between their findings and the observation of gap junction coupling between cerebellar neurons recently found in adult zebrafish (Chang et al., 2020). That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

We do not view this as necessarily a discrepancy. Chang et al., in their 2020 PNAS paper (Chang et al., 2020) show that in adult zebrafish, PNs which share a chemical synaptic connection are also likely to be dye coupled and electrically coupled. The authors show Cx35/36 puncta in PNs but it is not known whether other connexins are also expressed in the same PNs. We did not observe dye coupling between PNs in the larva. As mentioned above, Cx35/36 have been shown to be not supportive of dye coupling and the dye coupling observed in the adult may be mediated via other connexins which are expressed by PNs in addition to Cx35. The composition of electrical synapses can change dramatically during development (Bhattacharya et al., 2019).

Reviewer #1:

Reading the revised version of the manuscript "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons" by Sitaraman et al., I am pleased by the new experimental data and changes to the manuscript that have been added by the authors. The data is in most parts presented clearly, but a few points should be addressed to make some of their arguments more clear or to avoid misunderstandings of some of their claims.

1) According to the guidelines of eLife the title of the manuscript should mention the use of zebrafish as model system e. g.: "Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in zebrafish Purkinje neurons"

The policy states that the ‘biological system’ be mentioned where appropriate – which in our case is the Purkinje neuron.

2) The authors carefully state that Gjd2b levels are reduced in morphants and mutants (e.g. line 131 or line 158), but this reduction (and not the complete absence) is difficult to understand if it is not mentioned initially that the anti-mouse Cx35/36 antibody cross reacts with Gjd2a and Gjd2b. This should be made clear from the beginning. Accordingly, line 116 should state "We confirmed that Gjd2a/b puncta..".

Addressed above.

3) Line 221: the reduced number of synapses observed in TEM recordings could suggest but does not indicate a reduced number of formed synapses on PNs as this analysis is not able to distinguish between PNs and e.g. eurydendroid cells. I suggest to reword this sentence into "..these results indicate that Gjd2b is involved in the formation of synapses in the molecular layer of the cerebellum to which PN synapses significantly contribute.."

We have modified this sentence as suggested.

4) The authors report that anti-PSD95 immunohistochemistry analysis results in a higher number of synapse-counts on PNs that is contradictory to their results from TEM analysis and explain this with a limit in resolution by light microscopy (line 231) that does not allow one to unequivocally assign synapses to PNs. But should this erroneously assigning of synapses to PNs be the same in wild type and mutants? Why should this effect occur more often in the gjd2b mutants? This argument remains unclear to me.

Comment has been addressed above.

5) Line 240: the authors suggest that the increase in peak amplitude and faster kinetics of mEPSCs in gjd2b mutants results from excitatory synapses being placed closer to the PN soma, implying a stunted dendritic arbor and loss of distal synapses. This line of arguments is difficult for me to understand. Why should the loss of distal excitatory synapses lead to increases in peak amplitude?

Or do the authors suggest that the stunted dendrites represent "compressed" dendrites in which synapses are moved closer to the soma? Then synapse density close to the soma should be investigated rather than distal synapses.

Then the analysis reveals a reduced total dendritic branch length in gjd2b mutant PNs and confirms a stunted dendritic arbor from which the authors conclude that fewer synapses are placed proximally to the PN somata in gjd2b mutants (line 260). Is this not contradictory to the argument presented in line 240? For me this was confusing and I suggest to reword this paragraph. Did the immunohistochemistry analysis against PSD95 reveal an insight into the localization of excitatory synapses closer to PN somata as suggested in the previous review?

Also addressed above.

6) In the discussion (line 365-370) the authors emphasize their findings that glutamatergic synapses are decreased based on structural and physiological data but do not mention their contradictory finding with PSD95 immunohistochemistry. This should be discussed more cautiously.

We do believe that the combination of the electrophysiology and TEM data clearly point to a decrease in glutamatergic synapses. We present all of the caveats of the data in this paper in the supplementary table (Supplementary file 4).

7) Also, the authors should point out in the discussion that currently besides the cell type specific rescue of PN dendrite outgrowth in gjd2b mutants, they can currently not distinguish between cell-autonomous and non-cell autonomous effects of gjd2b loss on PNs. This should be made clear to the reader.

We have two ways in which we address the locus of action of Gjd2b:

1. Expressing full length Gjd2b in single PNs of mutants

2. Analysing the effect of Gjd2b puncta on individual branch behaviors in wild type PNs

These two experiments showed that signaling via Gjd2b locally can affect dendrite growth. These points are discussed in lines 448-456.

Reviewer #2:

The manuscript is now improved in several aspects. The authors performed additional experiments to verify their observations. They followed most of the reviewers' recommendations, yet the overall impression is that the Authors did not consider some critical comments. Specifically, the ones that aimed to improve the data's presentation.

Specifically:

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

Addressed above.

Regarding the question of the specificity of the CF stimulation, the answer is not convincing. The fact that the authors used this approach in the past is not enough to claim specific. Why did the authors not try to stimulate the inferior olive?

We cannot stimulate the olive directly because it is located on the ventral side and is hard to reach with the stimulating electrode. The CF EPSCs are much larger than PF EPSCs, completely blocked by CNQX and reverse at around +10mV, all pointing to AMPAR type synaptic input. A thorough characterization of these inputs has been done in our 2015 paper (eLife 2015;4:e09158).

Regarding the Dye-coupling experiments where the authors state that they do not observe any dye coupling between PCs. In light of the new paper published in PNAS (Chang et al., 2020), the authors should discuss this discrepancy between their findings and what is observed in adult zebrafish. That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

Addressed above.

Regarding our previous comment: "The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

Addressed above.

Recommendations for the authors:

The manuscript is now improved in some aspects; however, the overall impression that I have is that the Authors did not take into consideration all the comments, and I found somehow quite lousy the revision of the previous manuscript. Specifically:

The authors state: "The scales for plots in Figures 1 and 2 have been matched.". I actually question that. First, I think that the Authors refer to Figures 1 and 3 as requested. The scale bar for Figure 3A, I doubt that is 10ms. I think it is 10 sec. Accordingly, in Figure 1A is 5pA, while in Figure 3A is 10pA. Finally, why is it so difficult to add the scale numbers in Figure3 as they have them in Figure 1 to compare the two figures directly? From this, I think the authors should be more careful regarding their statements and care more about their work presentation.

Regarding the question of the specificity of the CF stimulation, the answer is not convincing. The fact that the authors used this approach in the past is not enough to claim specific. Why did the authors not try to stimulate the inferior olive?

Regarding the Dye-coupling experiments where the authors state that they do not observe any dye coupling between PCs. In light of the new paper published in PNAS (Chang et al., 2020), the authors should discuss this discrepancy between their findings and what is observed in adult zebrafish. That is important as in most neuronal networks; the gap junctions appear earlier, and they are more abundant in earlier developmental stages than the chemical synapses.

Regarding our previous comment: "The authors present the holding potential as the membrane potential, which is confusing to distinguish between EPSPs and EPSCs." I referred to figure 4, where the authors in Panel B write the mV at the beginning of the traces that is a common practice for current-clamp recordings. In voltage-clamp recordings, there is the 0pA. The holding is usually added above, below, or in legend and refer as that (e.g., Holding at -65 mV). To this end, I also noticed that for Figures 1 and 3, any reference to holding potential is missing. Is that again the -65mV or different?

Reviewer #3:

As outlined in my original review of this manuscript, I view it as interesting and potentially impactful, but preliminary in its interpretations and not consistently convincing. I do not view this has having changed in a meaningful way with these revisions. The new data provided do not particularly strengthen conclusions, and in one case conflict with the manuscript's narrative. Most reviewers' comments have been addressed without further experiments, but not in a way that is consistently satisfying. I address these revisions below, broken down by sections of the rebuttal letter.

Morphants:

I remain unconvinced that a conclusive negative result is shown for PPR in Figure 4. There is a nonsignificant drop in PPR in the morphants and a sometimes-significant rise in PPR at short ISIs in the mutants. I understand that the distributions overlap, and that this is why most results are not significant, but I do not believe that the results and the experimental n are sufficient to support the claim that this is a negative result. I do not believe that this argument has changed appreciably since the initial submission.

Addressed above.

I accept the authors' assertion that increased peak amplitude for the mutant is convincingly demonstrated in Figure 3D.

I continue to think that the mutant provides stronger support for the authors' claims than the MOs do.

We also agree, which is why we made the mutants after initial promising results with the morphants.

TALEN Mutants:

I am generally convinced by the authors' responses to these questions and with the added data, although I defer to the reviewer who originally raised the issue of the two orthologs with regard to whether this has been adequately addressed.

Electrophysiological recordings:

As described above, I continue not to be convinced of a negative result for the PPD data shown in Figure 4D-G.

Addressed above.

The arguments presented about blockers and the new NMDA pulse data appear to be valid to me, but I defer to the relevant reviewer.

EM Analysis:

I understand the converging lines of evidence that the authors refer to (ephys and EM), and understand that these could be viewed as complementary, given each line's strengths and caveats. This does not really address the reviewer's concern, although I defer to him/her with regard to whether they are convinced.

Thank you. We are not aware of the concerns regarding the converging lines of evidence. The concerns regarding EM data are addressed above.

The new PSD95 results are concerning. They are among the strongest results in the paper, and while the authors give two caveats associated with interpreting these data, the overall picture is not, on the whole, supportive of their interpretations.

I do not see that fish-level data have been added to Figure 5, where box plots continue to show data from hundreds of micrographs drawn from a small number of fish.

Dye-coupling experiments:

These seem inconclusive, and I am not convinced that the reviewer's concerns have been addressed.

All addressed above.

CaMKII-signalling:

This was an interesting but not fully supported element of the original manuscript. I agree with the decision to withdraw these data, but it leaves a less impactful paper.

We are flummoxed as we removed these data at the suggestion of the reviewers. We hope that suggestion was made after evaluating the rest of the paper on its own strengths. We have now included the CaMKII data in the manuscript.

References:

Bhattacharya A, Aghayeva U, Berghoff EG, Hobert O. 2019. Plasticity of the Electrical Connectome of C. elegans. Cell176:1174-1189.e16.

doi:10.1016/j.cell.2018.12.024

Chang W, Pedroni A, Hohendorf V, Giacomello S, Hibi M, Köster RW, Ampatzis K. 2020. Functionally distinct Purkinje cell types show temporal precision in encoding locomotion. PNAS117:17330–17337.

doi:10.1073/pnas.2005633117

Meda P, Chanson M, Pepper M, Giordano E, Bosco D, Traub O, Willecke K, el Aoumari A, Gros D, Beyer EC. 1991. in vivo modulation of connexin 43 gene expression and junctional coupling of pancreatic B-cells. Exp Cell Res

192:469–480. doi:10.1016/0014-4827(91)90066-4

Moser T. 1998. Low-conductance intercellular coupling between mouse chromaffin cells in situ. The Journal of Physiology506:195–205.

doi:https://doi.org/10.1111/j.1469-7793.1998.195bx.x

Pérez-Armendariz M, Roy C, Spray DC, Bennett MV. 1991. Biophysical properties of gap junctions between freshly dispersed pairs of mouse pancreatic β cells.

Biophys J59:76–92. doi:10.1016/S0006-3495(91)82200-7

Quesada I, Fuentes E, Andreu E, Meda P, Nadal A, Soria B. 2003. On-line analysis of gap junctions reveals more efficient electrical than dye coupling between islet cells. American Journal of Physiology-Endocrinology and Metabolism

284:E980–E987. doi:10.1152/ajpendo.00473.2002

Ransom BR, Kettenmann H. 1990. Electrical coupling, without dye coupling, between mammalian astrocytes and oligodendrocytes in cell culture. Glia

3:258–266. doi:10.1002/glia.440030405

Srinivas M, Rozental R, Kojima T, Dermietzel R, Mehler M, Condorelli DF, Kessler JA, Spray DC. 1999. Functional properties of channels formed by the neuronal gap junction protein connexin36. J Neurosci19:9848–9855.

Teubner B, Degen J, Söhl G, Güldenagel M, Bukauskas FF, Trexler EB, Verselis VK, De Zeeuw CI, Lee CG, Kozak CA, Petrasch-Parwez E, Dermietzel R, Willecke K. 2000. Functional expression of the murine connexin 36 gene coding for a neuron-specific gap junctional protein. J Membr Biol176:249–262.

doi:10.1007/s00232001094

Valiunas V, Mui R, McLachlan E, Valdimarsson G, Brink PR, White TW. 2004. Biophysical characterization of zebrafish connexin35 hemichannels. Am J

Physiol Cell Physiol287:C1596-1604. doi:10.1152/ajpcell.00225.2004

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Essential Revisions:

Please modify the text of the paper to identify caveat, alternate interpretations and open questions for future research, as suggested in the reviewers comments, below.

Thank you for the suggestions below. We have modified the text and included new analysis as suggested by the reviewers.

Reviewer #1 (Recommendations for the authors):

1. It would be more clear to name the splice-blocking morpholino as 'Gjd2b-MO'

All references to the splice blocking morpholino are now mentioned as Gjd2b-MO in the manuscript.

2. Control morpholinos were tested separately in experiments using immunolabeling, but were not used in the electrophysiology experiments. This is unusual but acceptable.

Since we showed that the control morpholino did not alter protein levels as observed with immunolabeling (figure 1—figure supplement 2E), we did not perform further electrophysiology experiments.

3. Figure 1. Supplement 1. Please label columns and rows in panel B.

We have added the labels to panel B of Figure 1—figure supplement 1.

4. For morphants, please state if experimental and control animals were from the same clutch and how many clutches of embryos were used for each experiment.

For immunohistochemistry, morphant and control animals were from the same clutch and the processing of control and morphant embryos was done in parallel. Data were collected from two such clutches of embryos.

For the mEPSC data, since parallel processing was not possible, control and morphant embryos were from different clutches. Control data are from 7 cells recorded from 5 clutches and morphant data are from 12 cells from 8 clutches. This information is now added to the respective figure legends.

5. Figure 4. Panel B, list specific ages in legend, rather than 'mixed ages'. Panel C, state that this is done on the presence of TTX.

The following was added to the figure legend:

EPSCs recorded at a hyperpolarized holding potential (bottom row traces) and at a depolarized holding potential (top traces) in normal saline (left side traces) and in saline containing the AMPAR blocker CNQX (right side traces). No EPSCs were detected in the presence of CNQX at -65 or +60mV at 7dpf (N = 4 cells) or at 19dpf (N = 2 cells). Data from 19 dpf shown above.

We mentioned in the methods that the NMDAR currents were recorded in the presence of TTX. We have now added this sentence to figure 4 legend as well.

6. In the first paragraph of discussion, authors should more clearly state that they think that conductance through Cx36/Gjd2b-mediated gap junctions is not be sufficient to pass dye and they are therefore not able to identify cells coupled to PCs.

We have addressed it thus (line 361):

“We found Gjd2b puncta localized to PN cell membrane but when dye was injected into PNs, we failed to observe any dye-coupled cells. This may be attributable to the low conductance of zebrafish Gjd2b channels, which have a unitary conductance of around 24pS (Valiunas et al., 2004).

7. Given the ambiguity about the sequence of events governing branch extension gap junction formation -> synaptogenesis->branch extension, repeat versus gap junction formation -> branch extension-> synaptogenesis, repeat., the interpretation on line 242 'resulting in fewer synapses' is overstated.

We have replaced “resulting in” with “having”. (line 254)

8. The authors state that it is likely that gap junctions promote dendritic arbor growth directly, independent of their actions on chemical synapses. What evidence (citation) supports this statement?

This statement is based on our data in Figure 7C where we saw increased elongations of branches containing at least one Gjd2b punctum compared to those that didn’t have any. These differences were observed even within the 5-minute imaging windows. We have moved this sentence ahead (line 442) to explain the context better.

9. On pg 19 the first full paragraph is repeated in the next section.

This was an oversight on our part. It has been rectified.

10. Figure 7, supplement 1. There is a problem with the labeling

This has been fixed.

Reviewer #2 (Recommendations for the authors):

Electrophysiology data:

1. The data on mEPSC frequency is clear evidence for fewer excitatory synapses. However, interpreting larger amplitudes and faster decays as an explanation for the loss of distal synapses is problematic. The shift towards larger amplitudes can also be explained simply by the absence of weaker synapses rather than arguing for the absence of distal synapses. Moreover, larger amplitudes can also be a homeostatic scaling mechanism to compensate for fewer synapses. Regarding the decay, this evidence would be more compelling if these experiments were done in current clamp where the absence of gap junctions would lead to longer voltage decay as a result of increased input resistance of neurons. In contrast, the faster current decay observed by the authors could also be a result of different subunit composition or other biophysical considerations (Laurence et al., Nature Neuroscience, 2005; Kumar et al., JNeuroscience 2002), and unless that is ruled out explicitly, I would recommend adding this as an important alternative explanation in the discussion.

We thank the reviewer for their appreciation of the mEPSC frequency data. Regarding the interpretation of the amplitude and kinetics data, we did not interpret them as indicative of loss of distal synapses. We only offer these observations as a motivation for looking into the dendritic arbor structure of mutant PNs (line 226-234). Nevertheless, we have added the following sentences to the discussion (line: 397):

“The amplitude and kinetics of mEPSCs of mutant PNs followed a trend that was consistent with smaller dendritic arbors. However, alternate explanations such as changes in receptor numbers and subunit types are also possible. “

2. What is the input resistance of mutant neurons versus wildtype Purkinje neurons? Given the smaller dendritic arbors and the absence of gap junctions, one would expect Gjdb2-/- neurons have a higher input resistance. It will be good to see these data.

We thank the reviewer for bringing up this interesting point. We compared the input resistance of wild type and mutant Purkinje neurons. Interestingly, the input resistances were not different when recorded with potassium gluconate internal solution but when recorded with cesium gluconate internal solution, the input resistance was significantly higher in the mutants. This observation is also consistent with the smaller arbor size. With potassium gluconate, probably only the proximal dendrites contribute to the leak, while with cesium gluconate, a larger extent of dendritic arbors contribute to the leak and therefore to the input resistance. A smaller dendritic arbor in the mutants would mean smaller dendritic leak, and hence larger input resistance. We have now included these data in Figure 3—figure supplement1.

3. In comparing the WT data in Figure 1D,E,F with Figure 3D,E,F, I am confused by differences in the distribution of WT data between these figures. There is a long tail in the distribution of WT data in Figure 1D,E,F which is absent in Figure 3D,E,F. I am curious why this is the case.

The long tail in Figure 1 D, E, and F is due to a few events, which are on the right end of the distribution. This can be seen in the inset box plots as well. We looked at whether these large values were contributed by one or two cells and this was not the case. The large values were typically from few events (2-3) in almost every cell recorded from. The morphant and mutant data sets were also acquired with a time gap of several months in between. The corresponding wild type data sets were also recorded several months apart. So, the few large outlier values seen only in Figure 1 may also be related to fish population level variability.

EM data:

1. I agree with the other reviewers that while the EM data are beautiful and hard to collect and analyze, not being able to attribute synapses to PNs significantly limits the conclusions one can reach from this experiment. I do not know much about cerebellar circuitry but is there any estimate for what fraction of excitatory synapses are formed on PNs versus other neurons? Also, the inability to distinguish between excitatory and inhibitory synapses is a major limitation. Between the EM and electrophysiology data, I would argue that the electrophysiology data are far stronger. I understand that the authors see this as converging evidence, but in my opinion, the EM data have substantial caveats and at best provide weak support for the conclusions the authors are trying to reach. If, for instance, the authors have data on spontaneous IPSCs in mutant and WT neurons that are similar in frequency, that can at least help argue that the numbers of inhibitory synapses are similar.

We agree with the reviewer regarding these caveats and have stated them as such in the Supplementary file 4 document. Unfortunately estimates of synapse numbers within the molecular layer of zebrafish cerebellum are not available. We have not recorded mIPSCs from these neurons as this would entail substantial new effort. We plan to investigate the effect of gjd2b on inhibition in the future.

2. I have not seen the PSD95 data, but I agree with the authors that PSD95 staining is not a compelling experiment. Light microscopy resolution is a major challenge. If anything, the authors could have tried expressing PSD95-tagged GFP in individual neurons in mutant and WT fish as was done in Niell et al. (Nature Neuroscience 2004) in the zebrafish optic tectum. I am not suggesting that the authors do this experiment but wanted to just throw in support for their argument that PSD95 staining is inconclusive.

We performed PSD95 immunohistochemistry to tag the endogenous PSD95. We did not try overexpressing PSD95-EGFP as multiple studies indicate that the overexpression of PSD95 in neurons affects synaptogenesis and dendrite growth (Graf et al., 2004; Charych et al., 2006; Nikonenko et al., 2008) to name a few. Thank you for agreeing with us regarding this experiment.

Dendritic elaboration and CaMKII

1. The gain of function experiments are interesting, but perhaps I am missing something here. My understanding is that functional electrical synapses need the assembly of a pore in the presynaptic and post-synaptic neuron. So, how would expressing Gjd2b in one neuron ensure functional electrical connectivity with other neurons? I see this is addressed in the limitations document, but the authors' reliance on pore dead experiments is unconvincing. If anything, the pore dead neurons have dendrites that are comparable to WT neurons and longer than the Gjd2b neurons (Figure 7b). I think the authors need to do a statistical analysis of differences between WT and Gjd2b rescue as well as between WT and Gjd2b pore-dead mutants. I think there is something interesting there that might allude to functional aspects of non-pore forming regions of Gjd2b. In the absence of clear experiments to demonstrate functional electrical synapses, I think this experiment falls significantly short of implicating electrical synapses in dendrite elaboration. If the authors do want to make this claim, in the very least they need to show that the "rescued" neurons have comparable excitatory synapses

We had technical difficulty in performing mEPSC recordings on rescued neurons as it was difficult to ascertain Gjd2b-mCherry expression at the wide-field microscope in our electrophysiology rig and the expressing neurons also tended to be deeper than could be accessed by our recording pipette. The dendrites of pore-dead expressing neurons are not significantly longer than Gjd2b neurons (p=0.44). The point about functional electrical synapses in the rescue experiment is addressed in point #3 above. We also show that in wild type neurons, when Gjd2b is expressed, branches containing one or more Gjd2b puncta elongate more than those that don’t have any within 5 minute imaging windows. This is another piece of evidence in support of Gjd2b promoting dendritic growth. However, as stated above, we cannot completely disambiguate dendrite growth promoting versus synaptogenic roles as the primary effect of Gjd2b. We have included this point in the discussion (line 440-448).

2. A lot of the work on dendritic elaboration has parallels with the rich body of work done in Hollis Cline's lab which the authors reference extensively. However, it's not clear to me that the dendritic effects are not simply a downstream consequence of the absence of Gjd2b rather any information transmitted through the electrical synapses. Since Gjd2b knockout reduces the number of AMPAR synapses based on the electrophysiology, isn't a simple explanation for all the dendritic effects simply a consequence of fewer AMPAR synapses as shown in Haas et al. (PNAS, 2006). Moreover, given the lack of direct evidence that the rescue experiments lead to functional electrical synapses, I am not convinced that molecules transmitted through gap junctions are somehow responsible for elevated CaMKII.

These points have also been brought up by Reviewer #1. We include in the discussion the two possible scenarios (Gjd2b→ AMPAR synapses → dendritic growth and Gjd2b→ dendritic growth → AMPAR synapses) mentioned by Reviewers 1 and 2. In addition we have removed strong wordings of causality with respect to CaMKII as suggested by both the reviewers.

In summary, while this paper represents a substantial amount of work and relies on converging lines of evidence to arrive at their conclusions, there are several limitations within each technique and these shortcomings are not addressed by the complimentary experiments.

The authors present good evidence for fewer chemical synapses and shorter dendrites in Purkinje neurons in fish where Gjd2b dependent electrical synapses are knocked down or knocked out. The concerns about electrophysiology data can be addressed in the discussion as an important caveat.

We have now stated these caveats in the discussion (line 397-401).

However, my bigger concern is with disambiguating Gjd2b mediated changes in dendritic structure from downstream effects of simply having fewer AMPAR synapses. Previous work has provided compelling evidence that chemical transmission through AMPAR synapses is a key driver of dendritic elaboration. So, if there are fewer AMPAR synapses, is it not unsurprising that the dendrites are smaller and that has nothing to do directly with Gjd2b function? Perhaps I am missing a key piece of the argument here and would be happy to be proven wrong.

This point is addressed above.

Reviewer #3 (Recommendations for the authors):

Having read through the previous response to the reviewers, I think that the authors have addressed them very well. I do not think further experiments or analyses are required.

Reference:

Charych EI, Akum BF, Goldberg JS, Jörnsten RJ, Rongo C, Zheng JQ, Firestein BL. 2006. Activity-independent regulation of dendrite patterning by postsynaptic density protein PSD-95. J Neurosci 26:10164–10176. doi:10.1523/JNEUROSCI.2379-06.2006

Graf ER, Zhang X, Jin S-X, Linhoff MW, Craig AM. 2004. Neurexins Induce Differentiation of GABA and Glutamate Postsynaptic Specializations via Neuroligins. Cell 119:1013–1026. doi:10.1016/j.cell.2004.11.035

Haas K, Li J, Cline HT. 2006. AMPA receptors regulate experience-dependent dendritic arbor growth in vivo. Proc Natl Acad Sci USA 103:12127–12131. doi:10.1073/pnas.0602670103

Miller AC, Whitebirch AC, Shah AN, Marsden KC, Granato M, O’Brien J, Moens CB. 2017. A genetic basis for molecular asymmetry at vertebrate electrical synapses. eLife Sciences 6:e25364. doi:10.7554/eLife.25364

Nikonenko I, Boda B, Steen S, Knott G, Welker E, Muller D. 2008. PSD-95 promotes synaptogenesis and multiinnervated spine formation through nitric oxide signaling. J Cell Biol 183:1115–1127. doi:10.1083/jcb.200805132

https://doi.org/10.7554/eLife.68124.sa2

Article and author information

Author details

  1. Sahana Sitaraman

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Gnaneshwar Yadav
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4093-0619

  2. Gnaneshwar Yadav

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Data curation, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Sahana Sitaraman
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1723-2831

  3. Vandana Agarwal

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared

  4. Shaista Jabeen

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1888-2836

  5. Shivangi Verma

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5245-9425

  6. Meha Jadhav

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8898-1434

  7. Vatsala Thirumalai

    National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    vatsala@ncbs.res.in
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2318-5023

Funding

Wellcome Trust DBT India Alliance (500040/Z/09/Z and IA/S/17/2/503297)

  • Vatsala Thirumalai

Department of Biotechnology, Ministry of Science and Technology, India (BT/PR4983/MED/30/790/2012)

  • Vatsala Thirumalai

Science and Engineering Research Board (EMR/2015/000595)

  • Vatsala Thirumalai

Department of Atomic Energy, Government of India (12-R&DTFR-5.04-0800)

  • Vatsala Thirumalai

CSIR-UGC-NET (UGC Fellowship)

  • Shaista Jabeen

Science and Engineering Research Board (YSS/2015/000908)

  • Gnaneshwar Yadav

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the following sources of funding support: Wellcome Trust-DBT India Alliance Intermediate and Senior fellowships (VT; 500040/Z/09/Z and IA/S/17/2/503297), Department of Biotechnology (VT; BT/PR4983/MED/30/790/2012), Science and Engineering Research Board, Department of Science and Technology (VT; EMR/2015/000595), Department of Atomic Energy (VT; 12R and D-TFR-5.04-0800), CSIR-UGC fellowship (SJ), NCBS-TIFR graduate student fellowship (SS, VA, SV, MJ, and SJ), and SERB Young Scientist Scheme (GY). We would also like to thank Prof. Masahiko Hibi for the aldoca construct and Dr. Hideaki Matsui for the Ca8 enhancer construct. Further thanks are also due to Dr. Igor Kondrychyn, Mr. Sriram Narayanan for technical assistance, and Mr. PT Jagadeesh for the maintenance of our fish lines. In addition, we would like to thank the NCBS-TIFR Genomics facility and the Central Imaging and Flow Facility for support.

Ethics

Animal experimentation: Institutional Animal Ethics and Biosafety committee approvals were obtained for all procedures adopted in this study (NCB/IAEC/VT-1/2011 and TFR/NCBS/14-IBSC/VT-1/2011). Larvae and adults were reared using standard procedures (Westerfield, 2000).

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Hollis T Cline, The Scripps Research Institute, United States

Reviewers

  1. Hollis T Cline, The Scripps Research Institute, United States
  2. Alanna J Watt, McGill University, Canada

Version history

  1. Preprint posted: January 31, 2020 (view preprint)
  2. Received: March 7, 2021
  3. Accepted: August 3, 2021
  4. Accepted Manuscript published: August 4, 2021 (version 1)
  5. Version of Record published: August 23, 2021 (version 2)

Copyright

© 2021, Sitaraman et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,770
    Page views
  • 197
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Sahana Sitaraman
  2. Gnaneshwar Yadav
  3. Vandana Agarwal
  4. Shaista Jabeen
  5. Shivangi Verma
  6. Meha Jadhav
  7. Vatsala Thirumalai
(2021)
Gjd2b-mediated gap junctions promote glutamatergic synapse formation and dendritic elaboration in Purkinje neurons
eLife 10:e68124.
https://doi.org/10.7554/eLife.68124

Categories and tags

Research organism

Further reading

    1. Genetics and Genomics
    2. Neuroscience

    Translation of dipeptide repeat proteins in C9ORF72 ALS/FTD through unique and redundant AUG initiation codons

    Yoshifumi Sonobe, Soojin Lee ... Paschalis Kratsios
    Research Article Updated

    A hexanucleotide repeat expansion in C9ORF72 is the most common genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). A hallmark of ALS/FTD pathology is the presence of dipeptide repeat (DPR) proteins, produced from both sense GGGGCC (poly-GA, poly-GP, poly-GR) and antisense CCCCGG (poly-PR, poly-PG, poly-PA) transcripts. Translation of sense DPRs, such as poly-GA and poly-GR, depends on non-canonical (non-AUG) initiation codons. Here, we provide evidence for canonical AUG-dependent translation of two antisense DPRs, poly-PR and poly-PG. A single AUG is required for synthesis of poly-PR, one of the most toxic DPRs. Unexpectedly, we found redundancy between three AUG codons necessary for poly-PG translation. Further, the eukaryotic translation initiation factor 2D (EIF2D), which was previously implicated in sense DPR synthesis, is not required for AUG-dependent poly-PR or poly-PG translation, suggesting that distinct translation initiation factors control DPR synthesis from sense and antisense transcripts. Our findings on DPR synthesis from the C9ORF72 locus may be broadly applicable to many other nucleotide repeat expansion disorders.

    1. Cell Biology
    2. Neuroscience

    Removal of extracellular human amyloid beta aggregates by extracellular proteases in C. elegans

    Elisabeth Jongsma, Anita Goyala ... Collin Yvès Ewald
    Research Article Updated

    The amyloid beta (Aβ) plaques found in Alzheimer’s disease (AD) patients’ brains contain collagens and are embedded extracellularly. Several collagens have been proposed to influence Aβ aggregate formation, yet their role in clearance is unknown. To investigate the potential role of collagens in forming and clearance of extracellular aggregates in vivo, we created a transgenic Caenorhabditis elegans strain that expresses and secretes human Aβ1-42. This secreted Aβ forms aggregates in two distinct places within the extracellular matrix. In a screen for extracellular human Aβ aggregation regulators, we identified different collagens to ameliorate or potentiate Aβ aggregation. We show that a disintegrin and metalloprotease a disintegrin and metalloprotease 2 (ADM-2), an ortholog of ADAM9, reduces the load of extracellular Aβ aggregates. ADM-2 is required and sufficient to remove the extracellular Aβ aggregates. Thus, we provide in vivo evidence of collagens essential for aggregate formation and metalloprotease participating in extracellular Aβ aggregate removal.

    1. Computational and Systems Biology
    2. Neuroscience

    Task-dependent optimal representations for cerebellar learning

    Marjorie Xie, Samuel P Muscinelli ... Ashok Litwin-Kumar
    Research Article Updated

    The cerebellar granule cell layer has inspired numerous theoretical models of neural representations that support learned behaviors, beginning with the work of Marr and Albus. In these models, granule cells form a sparse, combinatorial encoding of diverse sensorimotor inputs. Such sparse representations are optimal for learning to discriminate random stimuli. However, recent observations of dense, low-dimensional activity across granule cells have called into question the role of sparse coding in these neurons. Here, we generalize theories of cerebellar learning to determine the optimal granule cell representation for tasks beyond random stimulus discrimination, including continuous input-output transformations as required for smooth motor control. We show that for such tasks, the optimal granule cell representation is substantially denser than predicted by classical theories. Our results provide a general theory of learning in cerebellum-like systems and suggest that optimal cerebellar representations are task-dependent.