Drosophila spastin regulates synaptic microtubule networks and is required for normal motor function

Abstract

The most common form of human autosomal dominant hereditary spastic paraplegia (AD-HSP) is caused by mutations in the SPG4 (spastin) gene, which encodes an AAA ATPase closely related in sequence to the microtubule-severing protein Katanin. Patients with AD-HSP exhibit degeneration of the distal regions of the longest axons in the spinal cord. Loss-of-function mutations in the Drosophila spastin gene produce larval neuromuscular junction (NMJ) phenotypes. NMJ synaptic boutons in spastin mutants are more numerous and more clustered than in wild-type, and transmitter release is impaired. spastin-null adult flies have severe movement defects. They do not fly or jump, they climb poorly, and they have short lifespans. spastin hypomorphs have weaker behavioral phenotypes. Overexpression of Spastin erases the muscle microtubule network. This gain-of-function phenotype is consistent with the hypothesis that Spastin has microtubule-severing activity, and implies that spastin loss-of-function mutants should have an increased number of microtubules. Surprisingly, however, we observed the opposite phenotype: in spastin-null mutants, there are fewer microtubule bundles within the NMJ, especially in its distal boutons. The Drosophila NMJ is a glutamatergic synapse that resembles excitatory synapses in the mammalian spinal cord, so the reduction of organized presynaptic microtubules that we observe in spastin mutants may be relevant to an understanding of human Spastin's role in maintenance of axon terminals in the spinal cord.

Figure 1. Drosophila Spastin: Sequence Alignment and Gene Map

(A) Clustal alignment of complete D. melanogaster and H. sapiens Spastin amino acid sequences and the AAA region of Drosophila Katanin-60. Identical and similar residues are highlighted in dark and light gray, respectively. (B) Map of the Drosophila spastin gene, including exons (black boxes) and introns, the position of the T32 EP insertion (nucleotide 58 of the 5′ UTR), and the regions deleted by imprecise excision in lines 10-12, 17-7, and 5.75. The 3′ end of the adjacent Rox8 gene is also shown. Arrows indicate direction of transcription. (C) Unrooted phylogenetic tree generated by the “neighbor” algorithm, showing relationships between the AAA domains of Spastins and their close relatives in human and fly. Dm CG3326 is the counterpart of the human fidgetin/fidlik gene pair, while CG1193 probably encodes a second fly ortholog of human Katanin-60. In the mouse, fidgetin mutations produce inner ear defects that cause head-shaking and circling behaviors (Cox et al. 2000).

Figure 2. Spastin Protein Localizes to the Cytoplasm

(A) In embryo “fillets” in which Spastin overexpression is driven by the engrailed-GAL4 driver, a polyclonal antibody, pAb1239, generated against the C-terminal half of Spastin (aa 380–758) recognizes the characteristic striped pattern of Engrailed cells. Anterior is up; the CNS is the structure in the center, and the lateral epithelial stripes extend to either side. (B) An enlarged view of the CNS shows Spastin protein in these embryos localizing to neuronal cell bodies (arrow indicates the ventral unpaired midline [VUM] neurons), as well as in commissural and longitudinal axons (arrowheads). (C) A high-magnification view of the Spastin-positive epithelial cells shows that the protein fills the cytoplasm (arrow), and is excluded from the nucleus (arrowhead). Scale bar: (A) 25, (B) 10, and (C) 6.5 μm.

Figure 7. Spastin Overexpression in Muscles Erases the Microtubule Network

(A) An antibody against β3-tubulin stains body wall muscles and chordotonal cap cells in stage 16 wild-type embryos. Two abdominal hemisegments are shown; muscle fiber numbers are labeled in one. The cap cells (brackets) are difficult to distinguish in this panel because of high levels of muscle tubulin staining. They extend diagonally from about the middle of muscle 18 to muscle 22. Anterior is to the left, and dorsal is up. (B) When Spastin is overexpressed in muscles (genotype: G14-GAL4/+; T32/+), β3-tubulin staining is very weak and has a disorganized pattern in most muscle fibers, but an intact microtubule network is still present in the cap cells, which do not express this driver (brackets). The muscle fibers are misshapen and partially (arrowhead) or completely (arrow) detached from their insertion sites. (C–H) Similarly, the microtubule network (recognized by antibodies to α-tubulin) is almost eliminated by high-level Spastin expression in third instar larval muscles. Larvae of genotype UAS-spastin/MHC-GS-GAL4^213-3; spastin^5.75/TM3Ser-ActGFP overexpress Spastin protein specifically in muscles to varying degrees. Wild-type larval muscles had undetectable levels of Spastin using pAb 1239 (C) and displayed a dense network of microtubule bundles in the muscle (D and E), as well as in trachea (D, arrow) and neurons (D, arrowhead denotes a terminal arbor). In contrast, larval muscles expressing high levels of Spastin (F) show only faint muscle microtubule staining (G and H), while tracheal (G, arrow) and neuronal (G, arrowhead) staining remain robust.

Figure 3. Neuronal Overexpression of Spastin Causes Midline Convergence of Embryonic CNS Axons

(A) Anti-Fasciclin II (mAb 1D4) staining of filleted late stage 16 control embryos reveals three longitudinal axon bundles (arrow) on each side of the midline. Anterior is up. (B) In sca-GAL4/+; T32/+ embryos raised at room temperature, overexpression of Spastin in neurons causes the ladder to constrict toward the midline (e.g., arrow). (C) Increased Spastin expression at 29 °C causes collapse of the CNS onto the midline (arrow). Longitudinal axon tracts are thin or absent (arrowhead). (D) A phenotype similar to that in (C) is produced by sca-GAL4-driven expression of the UAS-spastin cDNA insertion at 23 °C. Arrow and arrowhead indicate same as in (C).

Figure 4. Synaptic Boutons Are Smaller, More Numerous, and Clustered in spastin LOF Mutants

(A–F) Representative A3 NMJs on muscles 6/7 (A–C) or muscle 4 (D–F) stained with antibodies against Dlg (green) and Syt (magenta) are shown for control larvae (WCS; A and D), spastin^5.75 larvae (B and E), and larvae expressing Spastin from the spin-GAL4 driver in a spastin^5.75 mutant background (Rescue; C and F). Boutons are arranged in a linear pattern in WCS larvae, whereas in spastin^5.75 larvae their distribution is more clustered and individual boutons are smaller. These phenotypes are rescued by Spastin expression via the spin-GAL4 driver. Scale bars, 10 μm. (G) Quantitation of bouton numbers in spastin mutants relative to wild-type and rescued larvae demonstrates complete rescue of the null phenotype by spin- or Elav-GS-GAL4-driven expression of Spastin. spastin-null mutants have on average 1.6-fold more type Ib boutons on muscle 4 compared to WCS control larvae. Similarly, spastin-null mutants (of genotype spin-GAL4/CyOKr-GFP; spastin^5.75) have 1.7-fold more boutons compared to their sibling rescued larvae (genotype spin-GAL4/UAS-spastin; spastin^5.75). Boutons are also 1.6-fold more numerous in spastin-null larvae from a neuronal rescue cross (genotype +/CyOKr-GFP; Elav-GS-GAL4,spastin^5.75/spastin^5.75) compared to their siblings in which UAS-spastin is expressed in neurons postembryonically (genotype UAS-spastin /CyOKr-GFP; Elav-GS-GAL4, spastin^5.75/spastin^5.75).

Figure 5. NMJs in spastin Mutant Larvae Display Reduced QC

(A) Representative EJP (upper) and mEJP (lower) traces are shown for control (WCS), spastin-null mutant (spastin^5.75), and spin-GAL4/UAS-spastin; spastin^5.75 (Rescue) larvae. All recordings were from the A3 or A4 muscle 6 NMJ. (B) The average EJP amplitude is decreased by about 20% in spastin-null mutants (37 ± 2.0 mV, n = 26) relative to control (48 ± 2.4 mV, n = 14) and Rescue (42.4 ± 1.2 mV, n = 28) larvae, and is intermediate between control and null levels in hypomorphic spastin^5.75/17-7 transheterozygotes (41.9 ± 1.5 mV, n = 22). (C) The average amplitude of spontaneous events (mEJPs) is increased slightly in spastin nulls relative to control and Rescue larvae. (D) The average frequency of spontaneous events is not affected in spastin mutants compared to control. Rescue larvae had a slightly higher mEJP frequency. (E) Average QC, a measure of the amount of neurotransmitter released per action potential, is significantly lower in transheterozygotes (28 ± 1.3) versus control (35 ± 1.7), and reduced even further in spastin nulls (23 ± 1.2). This decrease is completely rescued by spin-GAL4-driven rescue (30 ± 1.9, p = 0.1 compared to WCS). (F) Average QC is temperature dependent in spastin^5.75/spastin^17-7 transheterozygous larvae, but not in homozygous spastin-null or control larvae. QC measured in transheterozygotes raised at 18 °C (light gray bars) is intermediate between that of control and nulls. At room temperature (dark gray) and 29 °C (black bars), similar QC values are measured in transheterozygotes and null mutants. *, p < 0.05; **, p < 0.005.

Figure 6. spastin Mutant Flies Have Compromised Motor Behavior and Reduced Lifespans

(A) In a flight test assay, over twice as many adult spastin hypomorphs (spastin^10-12 and spastin^17-7) fail to fly before falling to the bottom of a cylinder, in comparison to w¹¹¹⁸ and T32 homozygous controls. (B) Although fewer than half of the hypomorphs fly, compared to more than 70% of controls, the distribution of collision sites of the fliers along the height of the cylinder parallels that of the controls, suggesting that these spastin mutations affect flying ability in some animals but not in others. (C and D) spastin mutants are compromised in their climbing ability. (C) All control (WCS) and nearly all spastin hypomorphs climb to the top of a vial in 30 s, but only 40% of spastin nulls do so. (D) Climbing velocity (measured for those flies that reach the top in 30 s) is 3.8 ± 0.2 cm/s in WCS (n = 45), but only 1.8 ± 0.2 and 1.4 ± 1.1 cm/s in spastin^10-12 (n = 28) and spastin^17-7 (n = 38) flies, respectively, and 0.3 ± 0.1 cm/s in spastin^5.75 null mutants (n = 17; p < 1 × 10⁻⁸ for all relative to WCS). (E) Lifespan curves. The curve inflection point at which WCS and hypomorph flies begin to die off at a rapid rate occurs at 30–35 d after eclosion, and more than 70% of hypomorph flies and 95% of wild-type flies are still alive at 30 d. In contrast, approximately 45% of spastin^5.75 null mutant flies die prior to 4 d after eclosion. However, the majority of the remaining null flies survive more than 25 d, so that the curve inflection point for nulls occurs only a few days before that for controls and hypomorphs. (F) Mean lifespan is 46 ± 2.7 d in WCS controls (n = 39) compared to 35 ± 3.2 and 35 ± 3.4 d, respectively, in spastin^10-12 (n = 24, p < 0.02) and spastin^17-7 (n = 32, p < 0.006), and 7.6 ± 0.6 d in nulls (n = 62, p < 10⁻³¹). (G) Only about 10% of spastin^5.75 flies eclosing at room temperature are males, while 40%–45% are males for controls and hypomorphs.

Figure 8. The Distribution of Stable NMJ Microtubule Bundles Marked by the MAP1B-like Protein Futsch Is Altered in spastin Mutant Larvae

(A–C) Anti-Futsch labels stable microtubule bundles in axons, NMJ boutons, and interbouton regions. The muscle 4 NMJs in segment A3 of third instar wild-type (A), spastin^5.75 (B), and spin-GAL4/UAS-spastin; spastin^5.75 (Rescue) (C) larvae were immunostained with anti-HRP (A and B) or anti-Syt antibodies (C) to label presynaptic boutons (magenta), and mAb 22C10 to label Futsch protein (A–C, green). Arrows and arrowheads mark the terminal boutons (those at the ends of synaptic branches); boutons marked by arrows in (A–C) are enlarged in insets in the middle and right panels to show examples of Futsch patterns. (A) In control (WCS) larvae, terminal boutons have both looped (arrowheads) and diffuse, punctate (right panel, arrows and inset) patterns of Futsch staining. (B) In spastin mutants, Futsch staining appears similarly strong in axon bundles (not shown) and along the main branches of the bouton arbor. More distal and terminal boutons, however, have diffuse or no Futsch staining (arrows and arrowheads). Note the absence of green staining in insets. (C) The distribution of Futsch staining is restored to the control pattern by spin-GAL4-driven expression of Spastin in the mutant background (arrows and arrowheads indicate loops). Scale bar, 5 μm. (D and E) Quantitative assessment of Futsch staining data. Futsch staining at A2 and A3 muscle 4 NMJs was classified as continuous (bundles or splayed bundles), looped, or diffuse or undetectable (none) for each bouton. (D) The percentage of boutons exhibiting continuous or looped Futsch staining (relative to the total number of boutons for each NMJ) is decreased in spastin mutants relative to controls, while the percentage of boutons having diffuse or no staining is increased. In total, 58% ± 4.2% of boutons in controls have a continuous pattern of Futsch staining, while only 42% ± 1.5% do in mutants. Boutons in this class are predominantly along the major (more proximal) branches of the axon arbor. Similarly, 11% ± 1.7% of wild-type boutons have Futsch loops, but only 6.0% ± 1.1% do in mutants. Most mutant boutons show only diffuse or no Futsch staining (52% ± 2.2%, versus 32% ± 5.2% in controls). Futsch distribution is restored to the control pattern by spin-GAL4- or Elav-GS-GAL4-driven expression of Spastin. (E) The difference in Futsch distribution is most pronounced at terminal boutons. There is no detectable Futsch staining in the majority of terminal boutons (65% ± 4.5%) in spastin mutants, compared to only 7.8% ± 5.8% of terminal boutons in wild-type larvae (p < 2 × 10⁻⁶). Futsch staining is restored in most terminal boutons of spin-GAL4- or Elav-GS-GAL4-rescued larvae, with only 20% ± 3.7% and 19% ± 5.9% of boutons, respectively, showing no staining (p = 0.09 compared to WCS). Terminal bouton staining in larvae overexpressing Spastin in neurons was unaffected relative to controls (p = 0.12). **, p < 0.005; *, p < 0.03 relative to WCS; n >8 NMJs scored in all cases.

Figure 9. The Microtubule Network in NMJ Boutons Is Altered or Absent in spastin Mutant Larvae

(A) In wild-type (WCS) larvae, an antibody against α-tubulin (green) reveals the distribution of the network of microtubule bundles within the A3 muscle 4 NMJ bouton arbor. Presynaptic bouton membranes are labeled by anti-HRP antibody (magenta). The microtubule network has a complex structure and extends into the terminal boutons (arrowheads and inset). Many proximal boutons have loops (arrows). Microtubules are also observed outside of the boundaries of the NMJ; these are within the muscle fiber, which also expresses α-tubulin. Staining of these muscle microtubules is minimized by the use of Bouin's fix. (B) In spastin^5.75 mutants, the microtubule network is much sparser than in controls, particularly in the distal boutons at the edges of the bouton clumps that are characteristic of spastin mutant NMJs (arrowheads and inset). Many of these distal boutons have little or no detectable α-tubulin staining. More proximal boutons still have tubulin loops, however (arrows). Scale bar, 5 μm.