Spatacsin is present in the endoplasmic reticulum and promotes ER–lysosomes contacts

We first investigated the subcellular localization of endogenous spatacsin. Spatacsin has been proposed to contain transmembrane domains that would allow its tight association with membranes [16]. We tested this hypothesis on membrane samples obtained from brains of Spg11 knockout mice (Spg11^−/−) devoid of spatacsin and compared them to samples of wild-type mice (Spg11^+/+). We subjected membranes of Spg11^+/+ and Spg11^−/− mouse brains to various extraction conditions (i.e., high salt, low pH, high pH, and detergents) [31]. Spatacsin and the integral ER–membrane protein STIM1 were not released from membranes by high salt concentration or low or high pH buffer but were solubilized by the detergent deoxycholate (S1A Fig). Conversely, the membrane-associated protein calreticulin was released from the membranes by high and low pH buffers (S1A Fig). Overall, these data showed that endogenous spatacsin was likely associated to membranes by transmembrane domains.

We then evaluated in which membranes endogenous spatacsin might be present by fractionating mouse brain samples by differential centrifugation (Fig 1A). Spatacsin was present in all but the S3 fraction corresponding to the soluble fraction of cytosolic proteins confirming its anchoring in membranes. Spatacsin was enriched in the P3 microsomal fraction along with the endolysosomal membrane protein Lamp1 and ER proteins STIM1 and calreticulin, whereas the lysosomal peptidase cathepsin D and outer mitochondrial channel VDAC were enriched in the denser P2 fraction and almost absent from the P3 fraction (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Spatacsin is present in the endoplasmic reticulum and promotes ER–lysosomes contacts.

(A) Western blot analysis of fractions obtained from Spg11^+/+ and Spg11^−/− mouse brains and separated by differential centrifugation. Immunoblots with antibodies raised against spatacsin, the ER proteins STIM1 and calreticulin, the lysosomal proteins cathepsin D (Cath D) and Lamp1, the mitochondrial protein VDAC. Light membrane fractions where spatacsin is enriched are encircled by the red rectangle. (B) Western blot analysis of ER- and lysosome-enriched fractions obtained from Spg11^+/+ and Spg11^−/− mouse brains. Immunoblots with antibodies raised against spatacsin, the ER proteins STIM1 and REEP5, lysosomal hydrolase cathepsin D and lysosomal membrane protein Lamp2. ER fractions are encircled by the red rectangle. The asterisk indicates a nonspecific signal. (C) Confocal images of MEFs expressing the ER marker GFP-Sec61β and V5-tagged spatacsin. Cells were immunostained with anti-V5 antibody and the lysosome marker Lamp1. Scale bar: 5 μm. (D) STED images of MEFs expressing the ER marker GFP-Sec61β and V5-tagged spatacsin. Cells were immunostained with anti-V5, anti Lamp1, and anti-GFP antibodies. Scale bar: 1 μm. Arrowheads point spatacsin-V5 colocalized with the ER marker GFP-Sec61β. Spatacsin V5 occasionnaly colocalized with the lysosome marker Lamp1 and the ER marker GFP-Sec61β (arrows). (E) Quantification of the proportion of V5-tagged spatacsin spots from STED images that are colocalizing either with Lamp1 staining or with GFP-Sec61β expressed in MEFs. Superplot: means and SEM, N = 18 cells from 3 independent experiments. Paired t test on the means. (F) Transmission electron microscopy images of ER-lysosomes contacts in Spg11^+/+ and Spg11^−/− MEFs. Arrows point toward areas where ER and lysosomes are in contact; asterisks indicate ER. Scale bar: 200 nm. (G) Proportion of lysosomes that are less than 30 nm from ER membrane analyzed on transmission electron microscopy images in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N > 17 cells from 3 independent experiments. Paired t test on the means. (H) Quantification of the proportion of lysosomal membrane that is less than 30 nm from ER membrane analyzed on transmission electron microscopy images in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N > 17 cells from 3 independent experiments. * P = 0.032, paired t test on the means. (I) Binarized image of fibroblasts expressing GFP-Sec61β and Lamp1-mCherry. Note the overlap (white) between the ER (magenta) and lysosome (green) masks. Scale bar: 5 μm. (J) Quantification of the proportion of lysosomes that have an area overlapping with the ER > 30% in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N = 13 cells from 4 independent experiments. Paired t test on the means. The raw data underlying panels E, G, H, and J can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g001

Next, we prepared lysosomes and ER-enriched fractions from Spg11^+/+ and Spg11^−/− mouse brains using density gradients. Spatacsin was present in the ER-enriched fraction that contained the ER proteins REEP5 and STIM1 but also contained the lysosomal membrane protein Lamp2 but no cathepsin D (Fig 1B). In contrast, the lysosomal fraction that contained both Lamp2 and cathepsin D was free of ER markers and contained almost no spatacsin, suggesting that spatacsin is not abundant in degradative lysosomes (Fig 1B). Together, these data indicate that endogenous spatacsin is a transmembrane protein present either at the membrane of nondegradative late endosomes or lysosomes, at the ER membrane, or both.

We then explored subcellular localization of spatacsin by immunostaining. Since there are no specific antibodies to spatacsin for immunostaining, we transfected mouse embryonic fibroblasts (MEFs) with a vector allowing the expression of spatacsin with a C-terminal V5 tag or an N-terminal GFP tag. We investigated the localization of spatacsin-V5 or GFP-spatacsin by immunostaining and confocal imaging. Both constructs showed a diffuse distribution of spatacsin that poorly colocalized with late endosomes and lysosomes stained with Lamp1 (Figs 1C and S1B). By contrast, spatacsin-V5 colocalized better with the ER labelled by the expression of GFP-Sec61β. Consistently, STimulated Emission Depletion (STED) imaging showed that spatacsin-V5 was mainly associated with the ER labelled by GFP-Sec61β (Fig 1D, quantification in E) or by REEP5 immunostaining (S1C Fig). Occasionally, spatacsin-V5 colocalized with lysosomes labelled by Lamp1 (Fig 1D and 1E). Of note, lysosomes that were colocalized with spatacsin were in contact with the ER, supporting that spatacsin is localized in the ER.

The absence of spatacsin had no visible impact on the morphology of the ER, as observed by live confocal imaging (S1D Fig). As we have observed that spatacsin is present in the ER and that its loss of function is known to affect lysosomes [20,21], we investigated by transmission electron microscopy if contacts between the ER and lysosomes were altered in absence of spatacsin (Fig 1F). Analysis of MEFs showed that more lysosomes were in contact with the ER in Spg11^+/+ than in Spg11^−/− cells (Fig 1F and 1G). Moreover, the proportion of lysosomal membrane that was in contact with the ER was decreased in absence of spatacsin (Fig 1H). The lower proximity between the ER and lysosomes in absence of spatacsin was also detected by confocal microscopy. We transfected Spg11^+/+ and Spg11^−/− MEFs with vectors expressing GFP-Sec61β, to label the ER, and Lamp1-mCherry, a marker of late endosomes and lysosomes (henceforth referred to as lysosomes). While lysosomes had the same average size and abundance (S1E and S1F Fig), there was a decrease in the proportion of lysosomes with their area overlapping by more than 30% with the ER staining in absence of spatacsin (Figs 1I, 1J and S1G). Overall, these data show that spatacsin is a protein localized in the ER that promotes contacts between the ER and lysosomes.

Spatacsin regulates the dynamics of lysosomes that are moving along the ER

Since spatacsin promotes the formation of contacts between the ER and lysosomes that are important to modulate lysosome function [4,14], we evaluated how the loss of spatacsin may affect the lysosomes and their interaction with the ER.

Using MEFs transfected with a vector expressing Lamp1-mCherry, we observed by live imaging a higher number of lysosomes with tubular shape in Spg11^+/+ compared to Spg11^−/− cells (Fig 2A and 2B), suggesting that spatacsin is required for the formation of tubular lysosomes. We also observed a higher number of tubular lysosomes in Spg11^+/+ than in Spg11^−/− MEFs when they were labelled with the acidic marker Lysotracker, with a pulse of Dextran-Texas Red followed by a long chase, or with DQ-BSA, which fluoresces upon degradation by lysosomal hydrolases [32] (S2A and S2B Fig). These results suggests that tubular lysosomes that are fewer in absence of spatacsin represent a population of catalytically active lysosomes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Spatacsin regulates the dynamics of tubular lysosomes that are moving along the ER.

(A) Lamp1-mCherry expression in Spg11^+/+ and Spg11^−/− MEFs imaged by spinning disk confocal microscopy. Note the presence of tubular lysosomes in Spg11^+/+ MEFs (white arrowheads). Scale bar: 5 μm. (B) Quantification of the number of tubular lysosomes in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N = 45 cells from 3 independent experiments. Paired t test on the means. (C) Quantification of the proportion of lysosomes that have an area overlapping with the ER > 30% based on their shape. Superplot: means and SEM, N = 14 cells from 4 independent experiments. Paired t test on the means. (D) STED image of a tubular lysosome stained with Lamp1 antibody and its close interaction with the ER tubular network stained by anti-V5 antibody targeting Reticulon2-V5 (Rtn2-V5) expressed in wild-type MEFs. Scale bar: 1 μm. (E) Snapshot images of live imaging of a wild-type MEF transfected with Lamp1-mCherry (Magenta) and GFP-Sec-61β (Green). A tubular lysosome trafficking along the ER tubule network is indicated by an arrow. Scale bar: 5 μm. Time is expressed in min:s. (F) Average speed of tubular and round lysosomes in wild-type MEFs. Superplot: means and SEM, N > 97 cells from 4 independent experiments. Paired t test on the means. (G) Quantification of the average speed of tubular lysosomes in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N > 77 cells from 4 independent experiments. Paired t test on the means. (H) Quantification of the instant speed of lysosomes in Spg11^+/+ during their transition from round to tubular or tubular to round shape. Mean and SEM, N = 23 pairs of lysosomes from 12 independent MEFs. Paired t tests. (I) Quantification of the instant speed of lysosomes in Spg11^−/− during their transition from round to tubular or tubular to round shape. Mean and SEM, N = 19 pairs of lysosomes from 5 independent MEFs. Paired t tests. The raw data underlying panels B, C, F, G, H, and I can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g002

To further investigate the properties of tubular lysosomes, we analyzed live images of MEFs expressing Lamp1-mCherry and GFP-Sec61β. This showed that the overlap of lysosome and ER staining was greater for tubular lysosomes than round lysosomes, suggesting that tubular lysosomes were in closer proximity to the ER (Fig 2C). Accordingly, STED microscopy showed that the tubular lysosomes were entangled in a network of ER tubules (Fig 2D). Live imaging also showed that tubular lysosomes moved along the ER network (Fig 2E and S1 Video), suggesting that tubular lysosomes are closely associated with the ER.

Since spatacsin promotes contacts between ER and lysosomes and contributes to the formation of tubular lysosomes, we evaluated the properties of tubular lysosomes. Tracking individual lysosomes in wild-type and Spg11^−/− MEFs by live imaging showed tubular lysosomes to move, on average, faster than round lysosomes (Fig 2F). The proportion of lysosomes with a speed >0.3 μm/s, corresponding to microtubule-dependent movement for these organelles [33], was higher among the tubular than round lysosomes (S2C Fig), suggesting that tubular lysosomes are highly motile and dynamic. Comparison of the speed of lysosomes in Spg11^+/+ and Spg11^−/− MEFs showed that tubular lysosomes moved faster in Spg11^+/+ than in Spg11^−/− MEFs (Fig 2G and S2 and S3 Videos), whereas the dynamics of round lysosomes was not affected (S2D Fig). Finally, tracking of individual lysosomes showed that lysosomes can change morphology from round to tubular and from tubular to round. The transition between the 2 states was associated with a strong change in the speed of displacement that was greater in Spg11^+/+ than in Spg11^−/− MEFs (Fig 2H and 2I) This showed that tubular lysosomes were more motile than round lysosomes, but their average speed was lower in absence of spatacsin.

Overall, these data suggest that tubular lysosomes represent a population of catalytically active and highly dynamic lysosomes moving in close contact with the ER. As spatacsin plays a key role in the formation and motility of tubular lysosomes, we next aimed to investigate how it could regulate their dynamics.

Spatacsin regulates the function of lysosomes by interacting with proteins involved in degradation pathways

We used MEFs derived from a mouse line in which exons 32 to 34 of Spg11 are spliced out (S3A–S3C Fig) to elucidate the molecular mechanisms by which spatacsin controls these lysosomal phenotypes. Such splicing retained the reading frame and led to the expression of a protein called spatacsin^Δ32–34, which lacks a domain of 170 amino acids, partially deleting the conserved Spatacsin_C domain (Fig 3A). Western blot analysis of brains obtained from Spg11^+/+, Spg11^−/−, and Spg11^{Δ32–34/Δ32–34} mice showed the latter strain to express a spatacsin protein that is smaller than the wild-type protein (S3D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Spatacsin regulates lysosome function by interacting with partners involved in protein degradation.

(A) Representation of the Spatacsin_C domain and the region truncated in the Spg11^{∆32–34/∆32–34} mouse model. (B) Quantification of the number tubular lysosomes in MEFs devoid of spatacsin or expressing truncated spatacsin. Superplot: means and SEM, N > 44 cells from 3 independent experiments. RM one-way ANOVA on the means followed by Dunnett’s multiple comparisons test. (C) Quantification of the proportion of tubular lysosomes moving > 1.2 μm over 1 minute in Spg11^+/+, Spg11^−/−, and Spg11^{∆32–34/∆32–34} MEFs. Superplot: means and SEM, N > 28 cells from 3 independent experiments. RM one-way ANOVA on the means followed by Dunnett’s multiple comparisons test. (D) Scheme showing the strategy to identify interactors of the domain of spatacsin encoded by exons 32 to 34 of SPG11. Yeast two-hybrid screens were performed with the C-terminal domain of human spatacsin (aa 1,943 to 2,443) or the same domain missing exons 32 to 34 as bait. Interactors specifically interacting with the spatacsin domain encoded by exons 32 to 34 (orange) were selected. (E) Design of the screening process for interactors of the spatacsin domain encoded by exons 32 to 34 of SPG11. Each interactor was down-regulated by siRNA in wild-type MEFs, and lysosomes were imaged by spinning disk confocal microscopy. The effect of the siRNAs was analyzed using an unbiased method (trained neural network) or directed analysis to quantify the presence of tubular lysosomes. (F) Table of the spatacsin interactors identified by the trained neural network and the directed analysis as modifiers of lysosomal function. (G) Table summarizing the pathways identified by gene ontology analysis as being significantly enriched in the list of genes identified in common by neural network analysis and directed analysis. FDR, false discovery rate. The raw data underlying panels B and C can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g003

To evaluate the importance of the domain deleted in Spg11^{Δ32–34/Δ32–34} mice, we performed behavioral and immunohistochemical evaluations of these animals and compared them to Spg11^−/− mice that present cognitive impairment and motor dysfunction [22]. To evaluate motor dysfunction, Spg11^{Δ32–34/Δ32–34} mice were challenged by an accelerating rotarod. They remained on the apparatus for significantly less time than wild-type mice but remained on the apparatus longer than Spg11^−/− mice (S3E Fig). The alternation in the Y-maze test was used to monitor cognitive function [22]. Spg11^+/+ mice explored a new environment (alternation) in 75% of the trials. In contrast, Spg11^{Δ32–34/Δ32–34} and Spg11^−/− mice showed a lower alternation than the Spg11^+/+ mice from the age of 4 months (S3F Fig). Together, these data demonstrate that the domain encoded by exons 32 to 34 of Spg11 plays a critical role as expression of a protein lacking this domain is sufficient to impair the cognitive and motor functions. We then investigated whether the absence of the domain encoded by exons 32 to 34 contributed to accumulation of autophagic material, as observed in Spg11 knockout mice [22,23]. We observed accumulation of the autophagic marker p62 in the motor cortex in 8-month-old Spg11^−/− and Spg11^{Δ32–34/Δ32–34} mice, whereas Spg11^+/+ mice presented almost no p62 accumulation (S3G Fig). Together, these data demonstrated that the domain encoded by exons 32 to 34 are required for the proper function of spatacsin. We therefore investigated the role of spatacsin^Δ32–34 on lysosomes.

Overexpression of spatacsin^Δ32–34 with a C-terminal V5 tag in MEFs showed diffuse and ER-associated localization like that of full-length spatacsin (S3H Fig). Like Spg11^−/− MEFs, Spg11^{Δ32–34/Δ32–34} fibroblasts had fewer tubular lysosomes than wild-type cells and had altered tubular lysosomes dynamics (Fig 3B and 3C). The difference in the dynamics of tubular lysosomes was validated by automated tracking, which showed tubular lysosomes to travel longer distance during 1 minute in Spg11^+/+ than in Spg11^−/− and Spg11^{Δ32–34/Δ32–34} MEFs (Fig 3C). We used this method to analyze lysosomal dynamics in subsequent experiments. Overall, these results suggest that the spatacsin domain encoded by exons 32 to 34 of Spg11 plays an important role in the formation and dynamics of tubular lysosomes.

We next aimed to define the molecular action of spatacsin in the formation of tubular lysosomes. We thus sought to identify proteins that bind to the domain encoded by exons 32 to 34 of Spg11. We performed a two-hybrid screen with the C-terminal region of human spatacsin (aa 1,943 to 2,443, containing the Spatacsin_C domain) and a second screen with the same C-terminal fragment in which the amino acids encoded by exons 32 to 34 were deleted (S1 and S2 Tables). Comparison of the 2 screens identified several proteins that potentially bind directly to the domain encoded by exons 32 to 34 (Fig 3D).

Among the proteins that could potentially bind to the domain encoded by exons 32 to 34, we aimed to identify those important for the regulation of lysosome function. Thus, we down-regulated each identified partner in wild-type MEFs using siRNA and analyzed the consequences on lysosomes, which were imaged by spinning disk confocal microscopy after staining with Texas Red–conjugated Dextran.

We used 2 methods to quantify the effect of siRNA on lysosomes (Fig 3E). First, we developed an unbiased classification method to discriminate between lysosomal staining in Spg11^+/+ and Spg11^−/− MEFs, which are expected to represent extreme phenotypes linked to function of spatacsin. This was performed using large series of images to train a neural network that exploited all parameters of the lysosomal staining in images in Spg11^+/+ and Spg11^−/− MEFs. The neural network classification was validated on an independent set of images of lysosomes obtained from Spg11^+/+ and Spg11^−/− MEFs. The trained neural network was then used to predict the probability of the cell to be considered as a Spg11^−/− fibroblast for each image of fibroblasts transfected with siRNA. This prediction was based on the parameters the network identified during its training phase as significant to discriminate between Spg11^+/+ and Spg11^−/− lysosomal stainings. In parallel, we performed a directed analysis that automatically detected tubular lysosomes. For both methods, we evaluated how well the down-regulation of each candidate using siRNA in wild-type MEFs phenocopied the lysosomal phenotype of Spg11^−/− MEFs. We compared the effect of each siRNA with that of 3 independent siRNAs down-regulating Spg11 (S4A and S4B Fig). The neural network approach identified 28 genes and directed analysis identified 11 genes that, upon down-regulation by siRNA, were at least as effective as Spg11 siRNA to phenocopy Spg11^−/− MEFs (Fig 3E and S3 and S4 Tables). Eight genes were identified in common by both analyses (Figs 3F, S4C and S4D), indicating that the products of these genes may interact with spatacsin to regulate its function in lysosomes.

Gene ontology analysis of the list of identified candidates pointed toward a role of the ubiquitin-dependent protein catabolic process and proteolysis in modulation of the lysosomal phenotype (Fig 3G), suggesting that the action of spatacsin on lysosomes may be linked to protein degradation pathways.

Spatacsin promotes degradation of AP5Z1 by lysosomes

Our screening approach led us to investigate how the control of tubular lysosome formation and dynamics by spatacsin relied on a degradation pathway. However, exploration of the interactome of the spatacsin domain encoded by exons 32 to 34 revealed no binding partners with endolysosomal localization. We therefore hypothesized that the degradation-dependent regulation may act on proteins present in the endolysosomal system. Importantly, 2 partners of spatacsin, spastizin and AP5Z1, colocalize with lysosomes [18].

We first investigated whether spastizin or AP5Z1 might be degraded in a spatacsin-dependent manner. We observed that overexpression of spatacsin-GFP in wild-type MEFs lowered levels of AP5Z1, while levels of spastizin were unaffected (Fig 4A). Of note, overexpression of spatacsin-GFP lowered the levels of another subunit of the AP5 complex, AP5M1 (Figs 4A and S5A). Overexpression of spatacsin strongly increased the amount of ubiquitinated AP5Z1 (Fig 4B), which may contribute to its degradation. Interestingly, overexpressed spatacsin itself was also ubiquitinated (Fig 4B), suggesting its levels may also be regulated in an ubiquitin-dependent manner. To test whether endogenous spatacsin may promote the degradation of AP5Z1 or spastizin, we overexpressed AP5Z1-His and Spastizin-HA in Spg11^+/+ and Spg11^−/− MEFs. We observed higher expression of AP5Z1-His in Spg11^−/− than in Spg11^+/+ MEFs, whereas there was no significant difference in the levels of spastizin-HA (S5B Fig), suggesting that endogenous spatacsin promotes degradation of AP5Z1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Spatacsin promotes degradation of AP5Z1 by lysosomes.

(A) Western blots showing levels of spatacsin-GFP, spastizin, AP5Z1, AP5M1, and actin in MEFs overexpressing GFP, spatacsin-GFP, or spatacsin^Δ32–34-GFP. Quantification of AP5Z1 levels by western blot: means and SEM, N = 5 independent experiments. Kruskal–Wallis test. (B) Western blots showing levels of AP5Z1 and spatacsin-GFP in inputs and in Ni-NTA beads pulldown performed in MEFs transfected with vectors expressing GFP, 6-His-Ubiquitin or 6-His-Ubiquitin together with spatacsin-GFP. Pulldown of 6-His-tagged proteins with Ni-NTA beads shows polyubiquitination of spatacsin-GFP, as well as modification of AP5Z1 with 6-His Ubiquitin when spatacsin-GFP is overexpressed. Note that ubiquitinated AP5Z1 appears with a slightly lower apparent molecular weight. (C) Western blot monitoring expression levels of AP5Z1 in wild-type MEFs overexpressing spatacsin-GFP and treated for 16 hours with either MG132 (15 μM) or bafilomycin (Baf., 100 nM). Actin immunoblot was used as loading control. Ubiqutin immunoblot showed accumulation of ubiquitylated proteins upon MG132 treatment. Quantification of AP5Z1 levels by western blot: means and SEM, N = 3 independent experiments. Kruskal–Wallis test. (D) Live images of Spg11^+/+ and Spg11^−/− MEFs expressing GFP-mCherry-AP5Z1 and stained with Lysotracker Blue. Inset shows the mCherry fluorescence colocalized with acidic lysosomes labelled by lysotracker while the GFP signal is located on the surface of lysosomes. Scale bar: 10 μm. (E) Quantification of the ratio of GFP and mCherry fluorescence in lysosomes compared to fluorescence in the cytosol in Spg11^+/+ and Spg11^−/− MEFs transfected with a vector expressing GFP-mCherry-AP5Z1. Superplot: means and SEM, N > 50 cells in 4 independent experiments. Paired t test on the means. The raw data underlying panels A, C, E, and I can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g004

The decrease in AP5Z1 levels upon spatacsin-GFP overexpression was blocked when MEFs were treated with the inhibitor of lysosomal acidification, bafilomycin, but not with the proteasome inhibitor MG132 (Fig 4C), suggesting that AP5Z1 was degraded by lysosomes in a spatacsin-dependent manner. To test this hypothesis, we expressed AP5Z1 fused to both GFP and mCherry and analyzed by live imaging the colocalization of fluorescence with the acidic lysosome marker, lysotracker (Figs 4D and S5C). In wild-type MEFs, mCherry was mainly colocalized with lysosomes, and a significant amount of signal was also present in the cytosol. In contrast, GFP that is sensitive to pH was poorly colocalized with lysosomes, indicating that GFP-mCherry-AP5Z1 colocalized with lysosome was mainly inside the acidic subcellular compartment (Fig 4D). Most GFP signal was in cytosol, suggesting that AP5Z1 is a cytosolic protein that transiently associates with lysosomes. However, GFP was slightly enriched around lysosomes (inset Fig 4D), and some AP5Z1 may thus be localized at the lysosomal surface. In Spg11^−/− MEFs, the amount of GFP colocalized with lysosomes was higher than in wild-type MEFs, suggesting that AP5Z1 accumulated at the cytosolic surface of lysosomes in absence of spatacsin (Fig 4D and 4E). Of note, GFP signal was recovered and fully colocalized by mCherry signal when MEFs were treated with bafilomycin to neutralize lysosomal pH, confirming that AP5Z1 was internalized into acidic lysosomes (S5C Fig).

Together, these results indicated that spatacsin contributes the degradation of AP5Z1 by promoting its translocation inside lysosomes.

UBR4 contributes to spatacsin stability and AP5Z1 degradation

We then tested whether the domain encoded by exons 32 to 34 is important for the degradation of AP5Z1, by overexpressing spatacsin^Δ32–34-GFP. The lower levels of AP5Z1 were not observed upon overexpression of spatacsin^Δ32–34-GFP, suggesting that the domain encoded by exons 32 to 34 is important to promote the degradation of AP5Z1 (Fig 4A). The domain encoded by exons 32 to 34 notably interacts with UBR4, which was identified in our screen as a mediator of lysosomal function (Fig 3F). We confirmed by co-immunoprecipitation that UBR4 interacts with the C-terminal domain of spatacsin, but not the domain lacking the fragment encoded by exons 32 to 34 of Spg11 (Fig 5A). Down-regulation of UBR4 partially prevented the degradation of AP5Z1 mediated by spatacsin-GFP overexpression (Fig 5B) and also reduced the levels of overexpressed spatacsin-GFP (Fig 5B), suggesting that UBR4 may modulate the levels of both proteins. Of note, down-regulation of UBR4 also led to higher levels of AP5Z1 in control MEFs (Fig 5B), suggesting that UBR4 may regulate degradation of AP5Z1 by endogenous spatacsin. To further address the role of UBR4, we down-regulated UBR4 and monitored the levels of endogenous AP5Z1 and spatacsin after inhibition of protein translation with cyloheximide. Levels of endogenous AP5Z1 decreased slightly slower upon UBR4 down-regulation (Fig 5C), consistent with a role of UBR4 to control degradation of AP5Z1. Interestingly, down-regulation of UBR4 enhanced the decrease in endogenous spatacsin levels (Fig 5C), suggesting that the presence of UBR4 is important for spatacsin stability.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. UBR4 contributes to spatacsin stability and AP5Z1 degradation.

(A) Western blots showing co-immunoprecipitation of UBR4 with the C-terminal domain of spatacsin (aa 1,943–2,443, GFP-spatacsin-Cter) but not the construct lacking amino acids encoded by exons 32 to 34 (GFP-spatacsin-Cter^Δ32–34). Input represents 5% of lysate added to the immunoprecipitation assay. (B) Western blot showing levels of AP5Z1 in wild-type MEFs transfected with vectors expressing GFP, or spatacsin-GFP with control siRNA or siRNA down-regulating UBR4. Means and SEM, N = 6 independent experiments. Kruskal–Wallis test (multiple comparison) and Mann–Whitney (comparison of 2 groups). (C) Western blots showing levels of AP5Z1 and spatactin after inhibiting protein translation with 10 μg/ml cycloheximide (CHX) in MEFs transfected with control siRNA or siRNA down-regulating UBR4. Means and SEM, N = 4 independent experiments. Two way-ANOVA. The raw data underlying panels B, C, and I can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g005

Together, these data suggested that UBR4 is important for spatacsin stability, and its down-regulation likely prevented the degradation of AP5Z1 mediated by spatacsin, leading to change in relative levels of both proteins.

Spatacsin promotes spastizin recruitment to lysosomes

We then investigated whether spatacsin may also regulate the function of its other interacting partner spastizin. We observed weaker colocalization of spastizin-GFP with Lamp1-mCherry in Spg11^−/− than Spg11^+/+ MEFs by live imaging (Fig 6A and 6B), consistent with previous observation [27]. To test whether spatacsin interaction with spastizin was required for spastizin localization to lysosomes, we performed a proximity ligation assay in MEFs transfected with spatacsin-V5 and spastizin-HA. This assay showed that the sites where spatacsin interacted with spastizin colocalized with the ER labeled by GFP-Sec61β and lysosomes labelled by Lamp1 immunostaining (Figs 6C and S6A). This suggests that the spatacsin–spastizin interaction occurs at contact sites between the ER and lysosomes to allow spastizin recruitment to lysosomes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Spatacsin-mediated degradation of AP5Z1 promotes spastizin recruitment to lysosomes.

(A) Expression of spastizin-GFP and Lamp1-mCherry in Spg11^+/+ and Spg11^−/− MEFs. Note the localization of spastizin-GFP along the tubular lysosomes in Spg11^+/+ MEFs (arrowheads in insert), and the weaker colocalization of spastizin-GFP with lysosomes in Spg11^−/− MEFs. Scale bar 10 μm. (B) Quantification of the proportion of spastizin-GFP colocalized with Lamp1-mCherry in Spg11^+/+ and Spg11^−/− MEFs. Superplot: means and SEM, N > 50 cells from 3 independent experiments. Paired t test on the means. (C) Proximity ligation assay (PLA) showing the interaction between V5-tagged spatacsin and HA-tagged spastizin in wild-type MEFs. The PLA signal (magenta) is detected at the level of the ER labelled by GFP-Sec61β and lysosomes immunostained with Lamp1 (arrowheads). Scale bar: 10 μm. Right: quantification of the proportion of PLA spots colocalized with only the ER, lysosomes, or both the ER and lysosomes. Superplot: means and SEM. N = 3 independent experiments. (D) Co-immunoprecipitation of spastizin and AP5Z1 with spatacsin-GFP upon down-regulation of AP5Z1 using siRNA (siAP5Z1). Input represents 5% of lysate added to the immunoprecipitation assay. Note that spatacsin and spastizin looked slightly different in input and co-immunoprecipitation, likely due to the high amount of both proteins in immunoprecipitates. Right: quantification of the relative amount of AP5Z1 or spastizin co-immunoprecipitated with Spatacsin-GFP. Means and SEM, N = 4 independent experiments. Mann–Whitney test. (E) Quantification of the proportion of spastizin-GFP colocalized with Lamp1-mCherry in wild-type MEFs transfected with siRNA down-regulating AP5Z1 (siAP5Z1). Superplot: means and SEM, N > 52 cells from 3 independent experiments. Paired t test on the means. The raw data underlying panels B, C, D, and E can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g006

We then investigated whether the degradation of AP5Z1 may have an impact on the interaction of spatacsin with spastizin by co-immunoprecipitation. To mimic the consequence of its degradation, we down-regulated AP5Z1 using siRNA (Fig 6D). Down-regulation of AP5Z1 favored the co-immunoprecipitation of spatacsin with spastizin (Fig 6D). Conversely, overexpression of AP5Z1 lowered the amount of spastizin co-immunoprecipitated with spatacsin (S6B Fig), suggesting that interaction of spatacsin with spastizin relied on the relative levels of AP5Z1 and that AP5Z1 may compete with spastizin to bind spatacsin. Furthermore, down-regulation of AP5Z1 promoted spastizin colocalization with lysosomes (Fig 6E), whereas overexpression of AP5Z1 partially prevented association of spastizin with lysosomes (S6C Fig).

Overall, these data suggested that AP5Z1 competes with spastizin to interact with spatacsin and that excess of AP5Z1 prevents association of spastizin with lysosomes that is mediated by spatacsin.

Spastizin and AP5Z1 are required for the formation of tubular lysosomes and interact with motor proteins

We then investigated whether AP5Z1 and spastizin regulated the formation and dynamics of tubular lysosomes. To evaluate whether AP5Z1 levels impacted lysosomes dynamics, we cotransfected wild-type MEFs with Lamp1-mCherry and either a vector expressing GFP-AP5Z1 or siRNA down-regulating AP5Z1 (Figs 7A and S7A–S7C). Both overexpression and down-regulation of AP5Z1 decreased the number of tubular lysosomes (Figs 7A and S7B) and impaired their dynamics in wild-type MEFs (Figs 7C and S7C). This observation suggested that AP5Z1 was required to regulate tubular lysosome motility, but its level must be tightly regulated.

To evaluate the role of spastizin at the lysosomes, we down-regulated spastizin using siRNA (S7D Fig). This condition decreased the number of tubular lysosomes and their dynamics (Fig 7B and 7C). Similar results were obtained when we treated wild-type MEFs with the PI3 kinase inhibitor wortmannin that prevented the lysosomal localization of spastizin (S7E and S7F Fig). Therefore, both AP5Z1 and spastizin contributed to regulate tubular lysosome motility.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Spastizin and AP5Z1 regulate association of lysosomes with motor proteins.

(A) Quantification of the number of tubular lysosomes in wild-type MEFs transfected with a control siRNA or an siRNA down-regulating AP5Z1. Superplot: means and SEM, N = 74 cells from 3 independent experiments. Paired t test on the means. (B) Quantification of the number of tubular lysosomes in wild-type MEFs transfected with a control siRNA and an siRNA down-regulating Spg15. Superplot: means and SEM, N > 92 cells from 4 independent experiments. Paired t test on the means. (C) Proportion of tubular lysosomes moving > 1.2 μm over 1 minute in wild-type MEFs transfected with a control siRNA, an siRNA down-regulating AP5Z1 or Spg15. Superplot: means and SEM, N > 60 cells from 3 different independent experiments. RM one-way ANOVA on the means, Dunnett’s multiple comparisons test. (D) Western blots showing co-immunoprecipitation of spastizin-HA with KIF13A-YFP or mutant KIF13A-ST-YFP (devoid of the motor domain). Input represents 5% of lysate added to the immunoprecipitation assay. (E) Quantification of the number of tubular lysosomes in wild-type MEFs transfected with wild-type KIF13A or mutant KIF13A-ST. Superplot: means and SEM, N > 69 cells from 4 independent experiments. RM one-way ANOVA on the means, Dunnett’s multiple comparisons test. (F) Proportion of tubular lysosomes moving > 1.2 μm over 1 minute in wild-type MEFs transfected with wild-type KIF13A or mutant KIF13A-ST. Superplot: means and SEM, N > 19 cells from 3 different independent experiments. RM one-way ANOVA on the means, Dunnett’s multiple comparisons test. (G) Western blots showing co-immunoprecipitation of p150^Glued with GFP-AP5Z1. Input represents 5% of lysate added to the immunoprecipitation assay. (H) Quantification of the number of tubular lysosomes in wild-type MEFs transfected with a vector overexpressing the dominant negative construct p150^Glued-CC1-mCherry. Superplot: means and SEM, N > 77 cells from 3 independent experiments. Paired t test on the means. (I) Proportion of tubular lysosomes moving > 1.2 μm over 1 minute in wild-type MEFs transfected with the dominant negative construct p150^Glued-CC1-mCherry. Superplot: means and SEM, N > 65 cells from 3 different independent experiments. Paired t test on the means. The raw data underlying panels A, B, C, E, F, H, and I can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g007

Spastizin has been shown to interact with the motor protein KIF13A, with an interaction domain mapped close to the motor domain according to two-hybrid analysis [34]. We confirmed by co-immunoprecipitation that KIF13A-YFP and its mutant form devoid of the motor domain KIF13A-ST-YFP [35] were capable of interacting with spastizin (Fig 7D). Of note, the mutant form of KIF13A interacted more with spastizin than wild-type KIF13A. The absence of motor domain in the KIF13A-ST-YFP protein may facilitate access of spastizin to its binding domain. Overexpression of the mutant KIF13A-ST-YFP prevented the formation of tubular lysosomes and altered their dynamics (Fig 7E and 7F), suggesting that KIF13A was important to control the formation of tubular lysosomes and their motility. To test whether interaction of KIF13A with spastizin was important for the formation and motility of tubular lysosomes, we generated a mutant human spastizin lacking the C-terminal domain (aa 2,120 to 2,539) mapped as the interacting domain with KIF13A by two-hybrid analysis [34]. Upon down-regulation of spastizin using siRNA, expression of human wild-type spastizin restored the number of tubular lysosomes and the lysosomal dynamics (S7H and S7I Fig). However, expression of human spastizin lacking its C-terminal domain and unable to interact with KIF13A (S7G Fig) did not restore normal tubular lysosome functions (S7H and S7I Fig). These data showed that the C-terminal domain of spastizin was required to promote the formation and motility of tubular lysosomes, likely by interacting with the motor protein KIF13A.

The formation of tubular membrane organelles requires the coordination of numerous effectors [36]. We therefore investigated whether AP5Z1 might also interact with some motor protein. The endocytic adaptor protein complex AP2 was shown to interact with p150^Glued, a subunit of dynein/dynactin complex implicated in retrograde transport [37,38]. Co-immunoprecipitation showed that AP5Z1 also interacted with p150^Glued (Fig 7G). To test whether this protein contributed to lysosome dynamics, we expressed a dominant negative construct p150^Glued-CC1 [39], which prevented the formation of tubular lysosomes and altered their dynamics (Fig 7H and 7I). These results suggested that AP5Z1 interaction with p150^Glued contributed to the formation of tubular lysosomes and regulates their motility.

Together, these data show that both spastizin and AP5Z1 contributed to the formation of tubular lysosomes and regulated their trafficking. Both effects on morphology and trafficking of lysosomes were likely mediated by the interactions of spastizin and AP5Z1 with anterograde and retrograde motors proteins, respectively. Since spatacsin controlled the levels of AP5Z1 as well as the association of spastizin with lysosomes, the balance between spastizin and AP5Z1 may regulate the directionality of lysosome movements.

Spatacsin regulates directionality of lysosome trafficking in axons

We then investigated highly polarized neurons that are degenerating in the absence of spatacsin, spastizin, or AP5Z1 [22,23,29,30] to test whether spatacsin may regulate direction of lysosome trafficking. Primary cultures of wild-type or Spg11^−/− cortical neurons were transfected with Lamp1-GFP and mCherry-TRIM46 and analyzed after 7 days in vitro. Since the uniform polarity of microtubules in axons facilitates the analysis of transport [40], the analysis of lysosomal trafficking was focused on axons, which were identified by the presence of the axon initial segment protein TRIM46 (S8A Fig). As observed in MEFs, tubular lysosomes in axons were more motile than the general population of lysosomes (Fig 8A and 8B). Furthermore, as in MEFs, neurons devoid of spatacsin presented a lower number of tubular lysosomes along the axon (Fig 8C), and a lower proportion of lysosomes were engaged in a dynamic movement in absence of spatacsin (Fig 8D and 8E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Spatacsin regulates directionality of lysosome trafficking in axons.

(A) Live images of Spg11^+/+ and Spg11^−/− primary mouse neurons at DIV7 expressing Lamp1-GFP. Axons are highlighted in yellow. White arrows point tubular lysosomes. Note the loss of tubular lysosomes in Spg11^−/− neurons. Scale bar: 10 μm, inset: 1 μm. (B) Quantification of the proportion of motile lysosomes among the total population of lysosomes or tubular lysosomes in wild-type neurons. Superplot: means and SEM, N > 18 neurons from 3 independent experiments. Paired t test on the means. (C) Quantification of the number of tubular lysosomes along the axon in Spg11^+/+ and Spg11^−/− mouse neurons. Superplot: means and SEM, N > 80 neurons from 5 independent experiments. Paired t test on the means. (D) Kymographs representing lysosomes movement over 30 seconds along the axon of Spg11^+/+ and Spg11^−/− primary mouse neurons. Left side of the kymographs is toward the soma, and right is toward the axon. Length of the axon segment: 80 μm. (E) Quantification of the proportion of lysosomes that are motile along the axon of Spg11^+/+ and Spg11^−/− primary mouse neurons. Superplot: means and SEM, N > 79 neurons from 5 independent experiments. Paired t test on the means. (F) Quantification of the proportion of lysosomes that are moving anterogradely along the axon of Spg11^+/+ and Spg11^−/− primary mouse neurons. Superplot: means and SEM, N > 79 neurons from 5 independent experiments. Paired t test on the means. (G) Left: Scheme representing the distribution of lysosomes along the axon of mouse neurons; the maximum distance considered is from the farthest detected axonal lysosome from the soma to the beginning of the axon. Right: Quantification of the proportion of lysosomes that are found within 50% of the maximum distance along the axon in Spg11^+/+ and Spg11^−/− primary mouse neurons. Superplot: means and SEM, N > 80 neurons from 5 independent experiments. Paired t test on the means. (H) Quantification of the proportion of lysosomes that are moving anterogradely along the axon of Spg11^+/+ primary mouse neurons upon transfection with Lamp1-GFP and control siRNA or siRNA down-regulating AP5Z1. Superplot: means and SEM, N >50 neurons from 3 independent experiments. Paired t test on the means. (I) Quantification of the proportion of lysosomes that are moving retrogradely along the axon of Spg11^+/+ primary mouse neurons upon transfection with Lamp1-GFP and control siRNA or siRNA down-regulating AP5Z1. Superplot: means and SEM, N > 50 neurons from 3 independent experiments. Paired t test on the means. (J) Scheme showing that spatacsin regulates the levels of AP5Z1 by promoting its degradation using a pathway implicating UBR4. This regulates the amount of AP5Z1 and spastizin at the lysosome surface. Since AP5Z1 and spastizin interact with the retrograde p150^Glued and anterograde KIF13A motor proteins, respectively, the regulation of AP5Z1 and spastizin levels at lysosome surface is a mechanism regulating directionality of lysosome trafficking. The raw data underlying panels B, C, E, F, G, H, and I can be found in S1 Data file.

https://doi.org/10.1371/journal.pbio.3002337.g008

Furthermore, investigation of lysosome trafficking in axons allowed us to monitor the direction of their displacement. Analysis of the directionality of movement showed a lower proportion of lysosomes with anterograde movement in Spg11^−/− axons compared to wild-type axons, whereas the proportion of lysosomes with retrograde movement was not significantly modified (Figs 8F and S8B). Importantly, axons of wild-type but not Spg11^−/− neurons presented similar proportions of lysosomes with anterograde and retrograde movement (Figs 8F and S8B). The imbalance between anterograde and retrograde movement observed in Spg11^−/− axons led to a concentration of lysosomes in the proximal part of their axons (Fig 8G). Together, these data suggest that spatacsin regulates the equilibrium between anterograde and retrograde trafficking of lysosomes and therefore affects the distribution of lysosomes along the axon.

We then investigated whether the action of spatacsin on lysosome trafficking in axons may be affected by the relative levels of spatacsin and AP5Z1. Overexpression of GFP-AP5Z1 in neurons expressing Lamp1-mCherry increased the proportion of lysosomes with retrograde movement in axons (S8C and S8D Fig). Conversely, down-regulation of AP5Z1 levels using an siRNA promoted anterograde movement (Fig 8H) but did not modify retrograde movement (Fig 8I). Overall, these results showed that the levels of AP5Z1 may have an action on the directionality of lysosome trafficking in axons. As spatacsin regulates the levels of AP5Z1 by promoting its degradation by lysosomes, this mechanism likely explained the change in the directionality of lysosome trafficking and repartition of lysosomes in the axon of Spg11^−/− neurons (Fig 8J).

Experimental procedures

Mouse models

The Spg11 knockout (Spg11^−/−) model has been previously described [22]. It was generated by inserting 2 stop codons in exon 32, leading to the loss of expression of spatacsin, and can thus be considered as a functional Spg11 knockout. To obtain this mouse model, we created an intermediate model in which floxed exons 32 to 34 bearing the stop codons were inserted in the reverse orientation in intron 34 [22] (S3A Fig). Reverse transcription PCR (RT-PCR) of the transcripts followed by sequencing of brain and spleen samples showed this intermediate model to express the floxed allele, with the splicing of exons 32 to 34 and conservation of the reading frame between exon 31 and exon 35 (S3B and S3C Fig). It was thus equivalent to a functional deletion of exons 32 to 34 and was named Spg11^{Δ32–34/Δ 32–34}. Behavioral evaluation of the mice and immunohistochemical analysis of the mouse brains were performed as previously described [22].

Antibodies

The antibodies used for immunofluorescence and immunoblotting were rat anti-Lamp1 (clone 1D4B, Development Studies Hybridoma Bank, University of Iowa, USA, deposited by JT August), rabbit anti-V5 (Cat#8137, Sigma), mouse anti-V5 (Cat#Ab27671, Abcam), rat anti-HA (clone 3F10, Cat#11867423001, Merck), rabbit anti-HA (Cat#ab9110, Abcam), rabbit anti-GFP (Cat#6556, Abcam), mouse anti-Myc (clone 9E10, Development Studies Hybridoma Bank), rabbit anti-Stim1 (Cat#5668, Cell Signaling Technology), rabbit anti-cathepsin D (Cat#Ab75852, Abcam), rabbit anti-spatacsin (Cat#16555-1-AP, ProteinTech), rabbit anti-spastizin (Cat#5023, ProSci), rabbit anti-AP5Z1 (Cat#HPA035693, Sigma), rabbit anti AP5M1 (provided by Dr. J Hirst; [18]), mouse anti-clathrin heavy chain (clone 23, Cat#610500, BD Biosciences), rabbit anti-UBR4 (Cat#Ab86738, Abcam), mouse anti-p150^Glued (Cat#610474, BD Biosciences), rabbit anti-calreticulin (Cat#SPA600F, Enzo Life Sciences), mouse anti-VDAC1 (Cat#Ab16814, Abcam), rabbit anti-REEP5 (14643-1-AP, ProteinTech), mouse anti-p62 (Cat#Ab 56416, Abcam), rat anti Lamp2 (Cat#Ab13524, Abcam), mouse anti-α-tubulin (Clone DM1A, Cat#Ab7291, Abcam), mouse anti-actin (Clone C4, Cat#Ab3280, Abcam), and mouse anti-ubiquitine (Clone P4D1, Cat #3936S, Cell Signaling Technology).

The secondary antibodies used for immunofluorescence were purchased from Thermofisher: donkey anti-mouse IgG Alexa 488 (Cat#A21202), goat anti-rabbit IgG Alexa 555 (Cat#A21429), and goat anti-rat IgG Alexa 647 (Cat#A21247). For STED microcopy, the secondary antibodies were anti-rabbit Atto-488 (Cat#A11008), anti-mouse IgG2a Alexa-594 (Cat#A21135), and goat anti-rat Alexa-647 (Cat#A21247). The secondary antibodies coupled to horseradish peroxidase used for immunoblotting were purchased from Jackson ImmunoResearch (Ely, UK): donkey anti-mouse IgG (Cat#JIR715-035-151) and donkey anti-rabbit IgG (Cat#711-035-152).

Plasmids

The human spastizin-GFP and spastizin-V5 vectors have been previously described [25]. Spastizin-HA was obtained by replacing GFP with an HA tag in the spastizin-GFP vector. To generate spastizinΔCT-GFP and spastizinΔCT-V5 vectors, we amplified by PCR (using Platinum SuperFi II PCR Master Mix, Thermofisher) the fragment of human Spg15 encoding aa 1 to 2,120 and used Gibson cloning kit (New England Biolabs) to replace Spg15 cDNA in the spastizin-GFP or spastizin-V5 vectors. A codon-optimized vector expressing human spatacsin was generated (Baseclear, Leiden, the Netherlands) in a gateway compatible system (Thermofisher). The cDNA was transferred by LR clonase into the pDest-47 vector or pDest-53 (Thermofisher), leading to a vector expressing spatacsin in fusion with a C-terminal or N-terminal GFP, respectively. Deletion of nucleotides 6,013 to 6,477 of optimized spatacsin cDNA resulted in spatacsin^Δ32–34. C-terminal fragments of spatacsin (aa 1,943 to 2,433) and spatacsin^Δ32–34 (aa 1,943 to 2,226) were amplified by PCR and inserted in the pDest-53 vector (Thermofisher), leading to vectors that expressed GFP-spatacsin-Cter and GFP-spatacsin-Cter^Δ32–34. Vector expressing AP5Z1-His was obtained from M. Slabicki [26]. To generate GFP-AP5Z1 or GFP-mCherry-AP5Z1 constructs, a stop codon was inserted after the final codon of AP5Z1 in the AP5Z1-His vector, and AP5Z1 cDNA was transferred by LR clonase into the pDest-53 or the pDEST-CMV-N-Tandem-mCherry-EGFP vector (Addgene #123216), respectively. The other plasmids used in the study were obtained from other laboratories or Addgene. GFP-Sec61β was obtained from G. Voeltz [59], reticulon2-V5 was from E. Reid [60], KIF13A-YFP and KIF13A-ST-YFP were from C. Delevoye [35], p150^Glued-CC1-mCherry was from T. Schroer [39], Ubiquitin-6His was obtained from R. Baer [61], Lamp1-GFP (#16290), Lamp1-mCherry (#45147), and mCherry TRIM46 (#176401) were from Addgene.

siRNA

The siRNAs used to down-regulate Spg11 were either On-target plus siRNAs (Dharmacon), with the sequences CAGCAGAGAGUUACGCCAA (#J-047107-09-0002) and CAGUAUGUGCCGGGAGAUA (#J-047107-12-0002), or from Thermofisher, with the sequence GGUUCUACCAGGCUUCUAUtt (#s103130). The siRNAs used to down-regulate AP5Z1 and Spg15 were Silencer Select siRNAs form Thermofisher: GGAGCAGAGUAACCGGAGAtt (#s106997, AP5Z1) and UCUGCUCCCGGGUCACUAAtt (#s106999, AP5Z1), CUUCAACUCCUGCAACGAAtt (#s102537, Spg15) and GAGCGAUACCAAGAGGUAAtt (#s102536, Spg15). The siRNAs used to test the role of spatacsin interactors identified by the two-hybrid screen were Silencer Select siRNAs from ThermoFisher and are listed in S5 Table.

Subcellular fractionation of brain tissue

Mice were killed using CO₂ and the brains immediately dissected and rinsed twice in PBS at 4°C. Subcellular fractionation was performed according to a previously described procedure [31]. Dissected brains were homogenized in 0.32 M sucrose and 10 mM HEPES (pH 7.4) using a PFTE (polytetrafluoroethylen) pestle attached to a stirrer (Heidolph, Germany) rotating at 500 rpm. Lysates were centrifuged at 1,330 × g for 3 minutes, generating a pellet (P1) and a supernatant (S1). The S1 supernatant was centrifuged at 21,200 × g for 10 minutes, producing a pellet (P2) and a supernatant (S2). The S2 supernatant was then centrifuged at 200,000 × g for 1 hour, generating a pellet (P3) and a supernatant (S3).

Membrane association assay

To determine the membrane association of spatacsin, mouse brain were extracted and homogenized as described above. After a first centrifugation at 1,330 × g for 3 minutes, the post-nuclear supernatant was centrifuged at 200,000 × g for 1 hour to collect a pellet corresponding to membranes. To evaluate spatacsin association with the membrane fraction, the latter was processed as previously described [31]. The membrane fraction was treated with one of the following solutions: 1 M NaCl and 25 mM phosphate buffer (pH 7.4); 100 mM glycine buffer (pH 2.8); 100 mM carbonate buffer (pH 11.0); or 1.0% sodium deoxycholate and centrifuged at 200,000 × g for 60 minutes. The final pellet and the supernatants were analyzed by western blotting.

Isolation of the ER- and lysosome-enriched fractions

Isolation of the ER- and lysosome-enriched fractions was performed according to previously described protocols with several modifications [62,63]. After killing mice using CO₂, the brains were immediately extracted and washed with PBS at 4°C. Brains were mechanically dissociated in 250 mM sucrose, 1 mM EDTA, 10 mM HEPES (pH 7.4), 1 mM DTT, and 25 mM KCl supplemented with a protease inhibitor cocktail (Thermofisher), using a PFTE pestle attached to a stirrer (Heidolph, Germany) rotating at 500 rpm. Lysates were centrifuged at 800 × g for 5 minutes and the pellets discarded. The supernatant was centrifuged at 20,000 × g for 10 minutes and the resulting pellet A retained for lysosome isolation and the supernatant A for ER isolation. For lysosome isolation, pellet A was resuspended in 2 mL of the initial buffer and deposited on 10 ml of 27% Percoll solution in a 15-mL tube. After 90 minutes of centrifugation at 20,000 × g, the lysosomal fraction was visible close to the bottom of the tube and collected by pipetting. It was then resuspended in the initial buffer and centrifuged at 20,000 × g for 10 minutes. The pellet was resuspended in sample buffer and analyzed by western blotting.

To isolate the ER, supernatant A was deposited on a gradient of several sucrose solutions prepared in 10 mM Tris (pH 7.4) and 0.1 mM EDTA. The sucrose concentrations of the 3 solutions were from the bottom up: 2 M, 1.5 M, and 1.3 M. The preparation was centrifuged for 70 minutes at 152,000 × g. After centrifugation, the ER-enriched fraction was found at the phase limit between the 1.3 M sucrose solution and the 1.5 M sucrose solution. The fraction was collected and resuspended in the initial buffer and centrifuged for 45 minutes at 152,000 × g. The pellet was resuspended in sample buffer and analyzed by western blotting.

Mouse embryonic fibroblast cultures

Mouse embryonic fibroblasts were prepared using 14.5-day-old embryos obtained from the breeding of heterozygous (Spg11^+/− or Spg11⁺/^Δ32–34) mice as previously described [19]. Comparisons between mutant and wild-type fibroblasts were always performed using fibroblasts originating from embryos of the same breeding. All experiments were performed with fibroblasts between passages 4 and 6.

Transfection of fibroblasts

Fibroblasts were transfected using the NEON transfection system (Thermofisher) with 1 pulse of 30 ms at 1,350 V, according to manufacturer instructions. Cells (5 × 10⁵) were transfected with 5 μg plasmid and analyzed 24 hours later. When we cotransfected a vector expressing a fluorescent protein together with a vector expressing a nonfluorescent protein for live imaging, we imaged cells expressing the fluorescent protein and then fixed the cells afterwards to verify that >95% of cells expressing the fluorescent marker were also positive for the nonfluorescent protein by immunostaining. For transfection with siRNA, 50 × 10³ cells were transfected with 1 pmol siRNA and analyzed after 48 hours in culture.

Primary cultures of neurons and transfections

Mouse cortical neurons were prepared using cortices of 14.5-day-old embryos obtained from the breeding of heterozygous (Spg11^+/−) mice. After a chemical dissociation for 15 minutes at 37°C (Trypsin 0.05% in Hibernate medium, Thermofisher) and mechanical dissociation, cortices were passed through a 70-μm filter and plated at a density of 100,000 neurons per cm² on precoated with Poly-L-Lysine (100 mg/L) 8-well coverslip IBIDI chambers. Neurons were grown in Neurobasal medium supplemented with B27 (ThermoFisher). Neurons were transfected on day 4 of culture. For each well of the IBIDI chamber, we prepared a mix containing 0.6 μl of Lipofectamine 2000 in 15 μl Opti-Mem medium (Thermofisher), with 500 ng of DNA or 1 pmol of siRNA that was added to cultured neurons for 3 hours.

Chemicals

Lysotracker Blue, Green, and Red (Thermofisher) were used at 50 nM for 30 minutes to stain acidic lysosomes in fibroblasts. DQ-Red-BSA and DQ-Green-BSA (Thermofisher) were added to the culture medium at 2 μg/ml 1 hour before imaging and then washed once with culture medium. Texas Red–conjugated dextran (10,000 MW, Thermofisher) was added to the culture medium at 100 μg/ml and the cells incubated for 4 hours to allow its internalization by endocytosis and chased for 24 hours to stain the lysosomal compartment. The PI3 kinase inhibitor wortmannin (Sigma) was used at 100 nM for 1 hour. The proteasome inhibitor MG132 was purchased from Tocris and was added to cells for 16 hours at 15 μM. Bafilomycin (Tocris) was added to cells for 16 hours at 100 nM.

Immunofluorescence

Cells were fixed in 4% PFA in PBS for 20 minutes and then permeabilized for 5 minutes in PBS containing 0.2% v/v Triton X-100. Cells were then blocked for 45 minutes in PBS with 5% w/v BSA (PBS-BSA) and incubated with primary antibodies in PBS-BSA overnight at 4°C. Cells were washed 3 times with PBS and incubated with secondary antibodies coupled to fluorophores. After 3 washes with PBS, glass coverslips were then mounted on glass slides using Prolong Gold antifade reagent (Thermofisher).

Confocal microscopy

Images of immunofluorescence were acquired using an inverted laser scanning Leica SP8 confocal microscope (Mannheim, Germany) with a 63× objective N.A. 1.40. STED images were acquired on SMD detectors with a confocal laser scanning microscope LEICA SP8 STED 3DX equipped with a 93×/1.3 NA glycerol immersion objective, in a thermostated chamber maintained at 22°C. The specimens were imaged with a white light laser and a pulsed 775 nm depletion laser to acquire nanoscale imaging. Typically, images of 1,024 × 1,024 pixels were acquired with a magnification above 3, resulting in a pixel size in the range of 25 to 45 nm.

For live imaging, cells were imaged at 37°C and 5% CO₂ using a Leica DMi8 inverted spinning disk confocal microscope equipped with 63× objective N.A. 1.40 and a Hamamatsu Orca flash 4.0 camera. Timelapses of MEFs were acquired to analyze the trajectories of the lysosomes with 1 image taken every 1 second for 1 minute. For timelapses capturing the trajectories of lysosomes in neurons, 1 image was taken every 500 ms for 30 seconds. Axons were identified by the accumulation of the axon initial segment protein TRIM46 upon transfection of the vector expressing mCherry-TRIM46.

Electron microscopy

Cultured MEFs were fixed with 2.5% glutaraldehyde in PBS for 2 hours at 22°C. The cells were then postfixed in 1% osmium tetroxide for 20 minutes, rinsed in distilled H₂O, dehydrated in 50% and 70% ethanol, incubated in 1% uranyl acetate for 30 minutes, processed in graded dilutions of ethanol (95% to 100%, 5 minutes each), and embedded in Epon. Ultrathin (70 nm) sections were cut, stained with uranyl acetate and lead citrate, and analyzed with a JEOL 1200EX II electron microscope at 80 kV.

Two-hybrid screen

The yeast two-hybrid screen was performed by Hybrigenics (Paris, France) using an adult human cDNA brain library. The bait was either the complete 1,943 to 2,443 domain of human spatacsin or the same domain lacking the amino acids encoded by exons 32 to 34.

Image analysis

Image classification by a neural network

Lysosomes of Spg11^+/+ and Spg11^−/− fibroblasts, as well as Spg11^+/+ fibroblasts transfected with siRNAs, were stained using Texas Red–conjugated dextran. Images were acquired using a spinning disk confocal microscopy, generating an image library.

To train the Spg11 classification model, Tensorflow (https://www.tensorflow.org/?hl=fr) and Scikit-learn (https://scikit-learn.org/stable/) Python libraries were used. Images of lysosomes of Spg11^−/− (n = 742 cells) and Spg11^+/+ (n = 735 cells) MEFs were used as a database. Training and test sets were generated randomly with a test set size of 15% (111 Spg11^+/+ fibroblast images and 112 Spg11^−/− fibroblast images). Initial images of 921 × 1,024 pixels were resized to 224 × 224 pixels to reduce input size while retaining consistent information. Data augmentation was performed using flip Tensorflow functions to improve training and artificially increase the number of images. Finally, the image pixel values were normalized between 0 and 1. The transfer learning approach was used to avoid model training from scratch. The VGG16 [64] neural network structure was used and downloaded using the Tensorflow_hub library (https://www.tensorflow.org/hub?hl=fr). VGG16 is a convolutional neural network model trained on ImageNet, which is a dataset of over 14 million images belonging to 1,000 classes. The top 3 layers were excluded and replaced with 3 other layers: one dense layer of 512 neurons, a dropout layer, and a 64-neuron layer. The final layer was the soft-max layer. The neural network was trained using a NVIDIA GeForce GTX 1050 Ti for 150 epochs, with a starting learning rate at 0.0001 and a batch size at 32. Model evaluation resulted in 79.5% total accuracy on the test set.

The trained model was used to predict the probability of the cell to be considered as a Spg11^−/− fibroblast for each image of fibroblast transfected with siRNA. For each siRNA, the arithmetic mean of the probability was calculated.

Protein extraction from cells

MEFs were washed twice with PBS and lysed in 100 mM NaCl, 20 mM Tris (pH 7.4), 2 mM MgCl₂, 1% SDS, and 0.1% Benzonase (Sigma). Samples were centrifuged at 17,000 × g for 15 minutes and the supernatants recovered as solubilized proteins. The protein concentration was determined using the BCA assay kit (Thermofisher). To evaluate stability of proteins, MEFs were treated with the protein translation inhibitor cylcoheximide (10 μg/ml) for 2 or 4 hours before cell lysis.

Protein ubiquitination

To detect ubiquitinated proteins, MEFs were transfected with a vector expressing Ubiquitin tagged with a 6-His tag as previously described [65]. Cells were wahed in PBS, lysed in 8 M Urea, 300 mM NaCl, 50 mM Na₂HPO₄, 50 mM Tris (pH 7.4), 1 mM MgCl₂, 0.5% NP40, 0.1% Benzonase, and 5 mM imidazole. 6-His ubiquitin modified proteins were purified by incubation with NiNTA agarose beads for 1 hour at 22°C. Beads were washed 5 times with lysis buffer containing 15 mM imidazole. Ubiquitin-conjugated proteins were eluted in sample buffer supplemented with 100 mM imidazole and analyzed by western blot.

Western blotting

Protein lysates supplemented with sample buffer (final concentration 80 mM Tris-HCl (pH 6.8), 10 mM DTT, 2% SDS, 10% glycerol) were separated on 3% to 8% Tris-acetate or 4 to 12 Bis-Tris gels (Thermofisher). Proteins were then transferred to PVDF membranes (Merck). Membranes were then incubated in Ponceau red for 5 minutes and blocked in PBS-0.05% Tween (PBST) with 5% milk for 45 minutes. The membranes were incubated with primary antibodies in PBST-5% milk overnight at 4°C. Secondary antibodies were conjugated with HRP (Jackson Lab) and the signals visualized using chemiluminescent substrates (SuperSignal West Dura/Femto; Thermofisher). The chemiluminescent signal was then acquired on Amersham Hyperfilm ECL. Images shown in figures are representative of at least 3 independent experiments.

Co-immunoprecipitation

Cells were lysed on ice in 100 mM NaCl, 20 mM Tris (pH 7.4), 1 mM MgCl₂, and 0.1% NP40 supplemented with a protease inhibitor cocktail. Samples were centrifuged at 17,000 × g for 15 minutes at 4°C. Approximately 5% of the supernatant was retained, supplemented with sample buffer, and was used to monitor protein quantity for the inputs. The remaining 95% of the supernatants was incubated with 10 μL of GFP-trap beads (Chromotek, Germany) for 90 minutes using a rotating wheel at 4°C. Beads were washed 4 times in lysis buffer and supplemented with sample buffer with DDT. For co-imunoprecipitation of spastizin with spatacsin, cells were transfected with a vector expressing tagged spastizin, as endogenous spastizin was not detected with the chosen lysis buffer. For co-immunoprecipitation of p150^Glued by AP5Z1 constructs, beads were washed 4 times in 300 mM NaCl, 20 mM Tris (pH 7.4), 1 mM MgCl₂, and 0.1% NP40. Beads and inputs were then analyzed by western blotting.

Proximity ligation assay

MEFs were fixed in PBS containing 4% PFA for 15 minutes. The Duolink Proximity Ligation Assay (PLA DuoLink in situ Starter Kit red, Sigma) was then performed according to the manufacturer’s instructions. After performing the PLA reaction, we immunostained the cells with fluorescent secondary antibodies to detect lysosomes labelled by LAMP1 antibody. As a negative control, we omitted the anti-V5 antibody. Coverslips were mounted using the Prolong Gold antifade mounting medium (Thermofisher) instead of the Duolink in situ mounting medium provided with the kit. Images were acquired with a confocal microscope.

Statistics

Data were analyzed using GraphPad Prism version 9 software. Comparisons of the means of replicate experiments were performed by paired t tests or ANOVA followed by multiple comparisons test following recommendations [66].