|Home | About | Journals | Submit | Contact Us | Français|
The actin binding protein plastin 3 (PLS3) has been identified as a modifier of the human motoneuron disease spinal muscular atrophy (SMA). SMA is caused by decreased levels of the survival motor neuron protein (SMN) and in its most severe form causes death in infants and young children. To understand the mechanism of PLS3 in SMA, we have analyzed pls3 RNA and protein in zebrafish smn mutants. We show that Pls3 protein levels are severely decreased in smn−/− mutants without a reduction in pls3 mRNA levels. Moreover, we show that both pls3 mRNA and protein stability are unaffected when Smn is reduced. This indicates that SMN affects PLS3 protein production. We had previously shown that in smn mutants, the presynaptic protein SV2 is decreased at neuromuscular junctions (NMJs). Transgenically driving human PLS3 in motoneurons rescues the decrease in SV2 expression. To determine whether PLS3 could also rescue function, we performed behavioral analysis on smn mutants and found that they had a significant decrease in spontaneous swimming and turning. Driving PLS3 transgenically in motoneurons rescued both of these defects. These data show that PLS3 protein levels are dependent on SMN and that PLS3 is able to rescue the neuromuscular defects and corresponding movement phenotypes caused by low levels of Smn suggesting that decreased PLS3 contributes to SMA motor phenotypes.
Modifiers of human disease lend insight into disease processes. Disease modifiers can be identified by a number of approaches including genome association analysis and subsequent identification of proteins, analysis of discordant families, and genetics screens in animal models. However, the identification of modifiers is still relatively rare for human diseases. Spinal muscular atrophy (SMA) is primarily a motoneuron disease that in its most severe form is fatal in infancy or childhood. The survival motoneuron (SMN) gene is genetically linked to SMA (Lefebvre et al., 1995). In the human genome there are two copies of the SMN gene, SMN1 and SMN2; a situation unique to humans since all other species only have SMN1 (Rochette et al., 2001). SMN2, however, undergoes splicing so that only ~10% of transcripts encode full-length RNA (Lorson et al., 1999; Monani et al., 1999). Thus, patients missing or carrying a mutation in their SMN1 gene have insufficient SMN protein levels and develop SMA. SMN functions in the assembly of RNA and protein complexes involved in pre-mRNA splicing and as such is an essential protein (Paushkin et al., 2002; Schrank et al., 1997). SMN has also been shown to interact with a number of other proteins suggesting alternative functions (Rossoll and Bassell, 2009). Thus, although SMN is genetically linked to SMA, the function of SMN critical to SMA remains unclear (Burghes and Beattie, 2009).
Whereas SMN2 copy number is linked to disease severity (Prior et al., 2004; Prior et al., 2009), other genetic factors appear to also modify the SMA phenotype. Knowledge of these modifiers has the potential to identify key functions of SMN that contribute to disease severity. To reveal modifiers of SMA, Wirth and colleagues examined gene expression in cell lines derived from siblings with the same genetics but a different disease severity (i.e. discordant siblings) (Oprea et al., 2008). It was found that plastin 3 (PLS3) was expressed at higher levels in cell lines derived from the less affected siblings. Further studies showed that over expressing PLS3 in both cultured neurons and in zebrafish could rescue the axon defects caused by low Smn levels (Oprea et al., 2008). Actin dynamics is an essential mechanism in axon outgrowth and synapse formation (Dent et al., 2011; Cingolani and Goda, 2008) and PLS3 belongs to a family of actin binding proteins and bundles and binds actin (Glenney et al., 1981). A number of studies have suggested a role for actin dynamics in SMA. For example, in a moderate model of SMA (SMN2B/−) the actin binding protein profilin IIA is increased, PLS3 is decreased, and rhoA kinase, an negative regulator of actin dynamics, is overactive (Bowerman et al., 2009, 2010). Interestingly, when rhoA kinase, is chemically inactivated in this model, survival is prolonged (Bowerman et al., 2010). Thus, a further understanding of how actin dynamics becomes altered by low SMN levels is warrented.
Here we show using zebrafish that Smn influences Pls3 protein levels by a post-transcriptional mechanism. Moreover, we show that PLS3 driven only in motoneurons rescues pre-synaptic NMJ defects and movement defects in smn mutants. These data indicate that a reduction of Pls3, caused by decreased Smn, contributes to key aspects of SMA.
Zebrafish used in this study were on the *AB/ TL (Tupfel long fin) background. smnY262stop mutants have been previously described (Boon et al., 2009). Adult zebrafish and embryos were kept at temperatures between 27 and 29°C and maintained by standard protocols (Westerfield, 1995).
Gene and protein nomenclature conform to species conventions (Genetic Nomenclature Guide, 1998). Zebrafish gene or RNA is denoted as lower case italics smn, pls3 and protein as Smn, Pls3. The human gene or RNA is upper case italics SMN, PLS3, and protein SMN, PLS3. When speaking of the protein or RNA in the general sense, the human convention is used.
The RFP-hSMN (Zhang et al., 2007) in pCS2 was digested with Bam HI and Not I and cloned into pBluescript vector that contained 2 SceI sites (RFP-hSMN/pBluescripSceI plasmid). The 1.5 kb heat shock 70 promoter (hsp70) was released from the pHSP70/4 EGFP-1 plasmid (Halloran et al., 2000) using the Bam HI site and cloned into the RFP-hSMN/pBluescript plasmid containing I-SceI sites to enhance transgenesis (Thermes et al., 2002). The orientation of the hsp70 promoter was determined by injecting DNA into one-cell embryos following heat shock at day 1 and screening for RFP-SMN at day 2. The hsp70: RFP-hSMN/pBluescript-ISceI that gave an RFP signal was used to make transgenic line Tg(hsp70: RFP-hSMN)os36 as described below.
This construct was made as described (Dalgin et al., 2011). Briefly, the zebrafish mnx1 (hb9) promoter (Flanagan-Steet et al., 2005) was cloned into a pBSK plasmid containing a 0.6 kb fragment of the zebrafish hsp70 promoter and GFP generated from (Halloran et al., 2000; Shoji and Sato-Maeda, 2008).
The human PLS3 cDNA fragment was released from pcDNA3.1 vector (Oprea et al., 2008) using Bam HI and XhoI and cloned into the pCS2 vector to obtain the polyA tail. The human PLS3-poly A fragment was then cloned into the mnx1:0.6hsp70:GFP/pBSK-ISceI plasmid described above using Bam HI and SacII. The final DNA construct mnx1:0.6hsp70:GFP; mnx1:PLS3/pBSK-ISceI was used to make mnx1:PLS3;0.6hsp70:GFP transgenic line Tg(mnx1:PLS3;0.6hsp70:GFP)os26.
Once we generated Tg(mnx1:PLS3;0.6hsp70:GFP) fish we found that even without heat shocking GFP was expressed in motoneurons and their axons starting at 19 hours post fertilization (hpf) suggesting that the mnx1 promoter was also acting to drive GFP. Therefore, we used the GFP in motoneurons as a marker for screening transgenic fish. We also found that the 0.6 kb fragment of hsp70 acted as an enhancer to increase the expression from the mnx1 promoter. This was confirmed by comparing fish with the 0.6 kb hsp70 to those made that lacked this construct.
The 1.5 kb hsp70 promoter was released from the hsp70/4 EGFP-1 (Halloran et al., 2000; Shoji and Sato-Maeda, 2008) plasmid using Bam BI site and cloned into mnx1:PLS3;0.6hsp70GFP/pBSK-ISceI. The orientation of the hsp70 promoter was determined by injecting DNA into one-cell stage embryos followed by heat shock induction at 1 day post fertilization (dpf) followed by a PLS3 western blot at 3 dpf. The final DNA construct was used to make the transgenic line Tg(mnx1:1.5hsp70:PLS3;0.6hsp70:GFP). Although this line has both the mnx1 and the 1.5 kb hsp70 promoters, it was used just for the heat shock induction experiments.
DNA injections were performed as described (Rembold et al., 2006). DNA plasmid samples were prepared using Qiagen Plasmid Mini kit and diluted to ~50 ng/l in I-SceI buffer (10 mM Tris-HCl, 1 mM Dithiothreitol, 10 mM MgCl2, pH 8.8), 5 Units of I-SceI enzyme (New England Biolab) and 0.1% phenol red. DNA samples were freshly prepared before each injection as this increased the rate of transgenesis. DNA was injected into embryos at the early one-cell stage (~0.5 nl). Injected embryos were transferred into fish water containing 50 units penicillin and 5 g streptomycin (Invitrogen). Injected fish (F0s) were heat shocked at 1 dpf at 37 for 30 minutes in a PCR machine and screened at 2 dpf. Only fish expressing RFP or GFP (depending on the line) in all/most cells were selected for growing to adulthood. F0s were outcrossed to wild types and the resulting F1s were screened for RFP or GFP fluorescence signal after heat shock induction. Fluorescent F1s were kept as F1 founders for the transgenic lines. Upon reaching adulthood, they were crossed to smnY262stop+/− fish to establish the transgenic line on the mutant background. Each transgenic line is given a lab (os, ohio state) designation. The smnY262stop+/− mutants were genotyped as described (Boon et al., 2009).
RNA was isolated using Trizol reagent (Invitrogen). qRT-PCR was performed using the two step method. First, cDNA was made using iScript cDNA systhesis kit (BioRad). qPCR was performed on the iCycler (Bio-rad) with IqSYBR Green Supermix (BioRad) containing 5 ng of cDNA and 10 pmol of each primer. Primers for zebrafish pls3 (forward-5’-GGTGATGGTATAGTCTTGTG-3’ and reverse-5’-TCTGGAGACAGTTTCATCAG-3’) gave a 320 bp PCR product. Primers for human PLS3 (forward-5’-GTTGGTGATGGAATTGTGCT-3’ and reverse-5’-GAAGCTCTTCTGGAGACAAT-3’) gave a 231 bp PCR product. Primers for zebrafish β-actin (forward- 5’-GATGACATGGAGAAGATCTG-3’ and reverse- 5’-ACCAGAGTCCATCACAATAC-3’) gave a 231 bp PCR product. qPCR was performed under the following conditions: one cycle at 94°C for 5 minutes; 45 cycles of 94°C for 30 seconds, 60°C for 30 seconds and 72°C for 30 seconds; and then a step down from 94°C to 4°C, by 0.5°C every 10 seconds. RNA expression levels were determined as described (Yin et al., 2008).
To examine PLS3 RNA decay, RNA was isolated from 3 embryos at each time point. This was done 3 times on embryos from different clutches and each sample was run in triplicate for both β-actin and pls3.
To examine endogenous zebrafish pls3 RNA levels, smnY262stop heterozygotes (smn+/) fish were incrossed and grown until 8 dpf then genotyped as described (Boon et al., 2009). Tg(hsp70:RFP-SMN)+/+;smn+/− were outcrossed with smn+/−. At 5 and 7 dpf, fish were heat shocked at 37°C for one hour. On 8 dpf, fish in both the heat shocked and non-heat shocked groups were genotyped for smn−/−. RNA was isolated from 3 embryos. This was done 3 times on embryos from different clutches and each sample was run in triplicate for both β-actin and pls3.
To analyze zebrafish pls3 splicing, total RNA from 11 dpf zebrafish wild-type or mutant embryos were isolated using Trizol reagent (Invitrogen) as manufacturer’s protocol. RT-PCR was performed using Qiagen one-step RT-PCR kit. Full-length zebrafish pls3 RNA was amplified by primers Fish-pls3-FL-F 5’-ATGGCTGGTAAGATATCCAAAGAAGAGCTG and Fish-pls3-FL-R 5’-TTAAGCCCTTTTCATACCCTCTGCCCATTAG. PCR products were run on 8% polyacralamide gel to aid in resolving any splice products.
The half-life (T1/2) of human PLS3 RNA and protein in control and smn morphants was determined by measuring RNA (by qRT-PCR) or protein (by westerns) at subsequent hours or days post heat shock. This was plotted on a log scale and the formula T1/2=ln2/X (where X is the slope of the line) was used to calculate the half-life.
The antisense morpholinos (MO; GeneTools, Eugene OR) used for Smn knock down and the control MO were used as previously described (McWhorter et al., 2003). Each embryo was injected with a dose of MO that resulted in approximately 80% knockdown of Smn protein. This ranged from 3–9 ng of smn MO depending on the efficacy of the batch purchased. The concentration of MO delivered per embryo was determined as previously described (Carrel et al., 2006). For pls3 knock down, two different MOs were used: a spice blocking MO 5’-AAAGTAAAATCAAACCTGTATTGTG-3’ (Oprea et al., 2008) and a translation blocking MO 5’-ATATCTTACCAGCCATCTCCCAAAG -3’. A standard MO (5’-CCTCTTACCTCAGTTACAATTTATA-3’) was used for control injections. Different doses of the splice pls3 MO (9–12 ng) were injected into 1-cell stage embryos. Consistent with previous result, 9 ng of pls3 MO resulted in only 30–40% knockdown, without a phenotypic effect on motor axons, a higher dose of pls3 MO (12 ng) resulted in 100% lethality (Oprea et al., 2008). 11 ng of pls3 MO resulted in deformed embryos and 10 ng resulted in slow development but normal morphology and very low levels of Pls3 thus this dose was used for the splice blocking pls3 MO. Since the splice blocking pls3 MO was so difficult to titrate, we decided to use the translation blocking pls3 MO. A dose response revealed that at 9 ng all embryos were deformed, at 6 ng all embryos developed slowly and at 4 ng embryos developed normally and western blot confirmed low levels of Pls3 protein (see text). Thus, we chose to use 4 ng of the translation blocking MO for our experiments. Note, all experiments with pls3 MO were confirmed with both MOs, but data shown here were obtained with the translation blocking pls3 MO. For the human PLS3 RNA rescue experiments, 200 pg of PLS3 RNA was co-injected with the translation blocking pls3 MO. Neither pls3 MO recognizes human PLS3 RNA. The tdp43 MO sequence and concentration used have been previously described (Kabashi et al., 2010). Motor axons were visualized and classified as previously described (Carrel et al., 2006). Approximately 25 embryos were scored from at least three separate injections.
Three zebrafish embryos or larvae were dissolved in 10 µl of blending buffer (50 mM Tris pH6.8, 10 mM EDTA and 10% SDS) and boiled for 10 minutes. An equal volume of loading buffer was then added to the samples that were then boiled for 2 additional minutes and resolved on a 12.5% polyacrylamide gel. The gel was electrotransferred to the Protran BA 83 Nitrocellulose membrane (Whatman). The membranes were probed with primary antibodies. Signals were detected with horseradish peroxidase-conjugated goat anti-mouse antibody 1/5000 (Jackson ImmunoResearch Laboratories, Inc) or horseradish peroxidase-conjugated goat anti-rabbit IgG 1/5000 (Santa Cruz Biotechnology), ECL reagents and Amersham Hyperfilm ECL (Amersham Bioscience). Antibodies were used for western blot in this study: SMN monoclonal antibody MANSMA12 (1/500) (gift from Glenn Morris; Young et al., 2000), PLS3 polyclonal antibody 1/5000 (gift from Dr. Brunhilde Wirth; Oprea et al., 2008); profilin IIb monoclonal antibody PF2A3 (1/100) (gift from Glenn Morris; Sharma et al., 2005), gemin 2 monoclonal antibody (1/10) (gift from Dr. Livio Pellizzoni; McWhorter et al., 2008); gemin 5 monoclonal antibody GEM5E (1/100) (gift from Dr. Glenn Morris; Hao et al., 2007) and β-actin monoclonal antibody 1/5000, Santa Cruz Biotechology. Under these conditions, proteins run as follows: zebrafish gemin 5, 170 kD; zebrafish gemin 2, 32 kD; zebrafish Pls3, 65 kD; human PLS3, 75 kD; zebrafish Smn, 42 kD; human SMN, 38 kD; zebrafish Sod, 16 kD; zebrafish profilin, 16 kD; zebrafish β-actin 42 kD. Pls3 and Smn were always run on the same gel, then the blot cut and probed separately with anti-PLS3 antibody (top half) or anti-Smn antibody (lower half). The lower half was subsequently stripped and reprobed with β-actin. Western blots were quantitated as previously described (Hao et al., 2011).
Zebrafish larvae were anesthetized with tricaine (Sigma, A-5040) and their heads removed for genotyping. Trunks were fixed in 4% paraformaldehyde/PBS and 1% DMSO overnight at 4C. Larvae were then washed in 1XPBS for 10 minutes then distilled water for 10 minutes. This was followed by a 15 minute incubation at room temperature with −20 acetone, then a final incubation in distilled water for 20 minutes. Samples were incubated overnight at 4C with presynaptic antibody SV2 (Developmental Studies Hybridoma Bank, Iowa City, IA), diluted 1/100 in PBDT buffer (1XPBS, 1% DMSO, 1% BSA, 0.5% TritonX-100) and 2.5% normal goat serum. Samples were washed for 10 minutes 5 times with PBST at room temperature and incubated overnight at 4°C with Alexa Fluor 633 goat-anti mouse IgG (Invitrogen, Grand Island, NY) diluted 1/400 in PBDT and 2.5% normal goat serum. Samples were washed for 10 minutes 5 times in PBST, mounted on slides using vectashield (Vector Labs, Burlingame, CA, USA) and images were captured with the Leica TCS-SL scanning confocal microscope system. The intensity of the SV2 immonostaining was calculated using Metamorph software (Molecular Devices Inc, Sunnyvale, CA. USA). Single-plane gray-scale images were first corrected by arithmetically subtracting the background from the original images and intensity thresholds were adjusted. Six regions per fish were chosen throughout the ventral hemisegments excluding lateral hemisegment boundaries. The number of pixels per fish was then calculated and reported as mean ± s.d.
For pls3 morphants, acetylated tubulin (1/500, Sigma; Alexa Fluor 488 anti mouse 1/500, Invitrogen) labeling was performed on 34 hpf fish and analyzed by confocal microscopy (Leica TCS-SL scanning confocal microscope).
To examine spontaneous larval motor behaviors, we captured video recordings using a Motionpro high speed camera (Redlake, Tucson AZ) at 1000 frames per second with 512 × 512 pixel resolution, using a 50 mm macro lens. Using a 4 × 4 grid, we simultaneously analyzed the movement of 16 individually housed 9 dpf larvae and recorded 400 millisecond bouts at 3 second intervals for a total of 60 trials. Prior to testing for spontaneous movement, larvae were pre-adapted to the intensity of light in the testing arena for at least 2 hours. Video recordings were initiated 5 minutes after moving larvae to the testing grid in order to allow sufficient time for locomotor activity to stabilize. Behavioral experiments were carried out at 26–27 °C. All behavioral measurements were made with the FLOTE software package (Burgess and Granato, 2007a, b). This software performs tracking of individual larvae and then performs automated analysis of body curvature on each larva to extract kinematic details of swimming and turning movements. The software then classifies each movement according to previously defined kinematic parameters that distinguish discrete motor behaviors. Based on these kinematic details, the software assigns a discrete motor behavior using previously defined kinematic parameters. For each larvae, the movement frequency was calculated by dividing the # of turns (or swims) by 60. Larvae were genotyped following behavioral testing, thus all behavioral analysis was done blinded to the genotype,
To better understand the relationship between SMN and PLS3, we analyzed Pls3 protein levels in zebrafish smn mutants. We have previously shown that Smn protein in the smnY262stop−/− zebrafish larvae (hereafter referred to as smn−/−) was present at high levels until ~ 4 days post fertilization (dpf) due to maternal contribution of RNA and protein in the yolk. These levels gradually decreased reaching low levels by ~7 dpf until the mutants died at ~12 dpf (Boon et al., 2009) (Fig. 1A). Analysis of smn−/− revealed a similar decrease in Pls3 protein expression. Pls3 was initially present but decreased to very low levels as Smn decreased (Fig. 1A). To determine whether Pls3 reduction was unique among actin modifying proteins, we examined profilin II levels. Profilin II is a regulator of cytoskeletal and actin dynamics in the nervous system and is known to bind to SMN (Giesemann et al., 1999). Western blot analysis revealed that profilin II (a and b) levels were not altered when Smn levels decreased. Furthermore, we examined levels of other SMN binding partners, gemin 2 and gemin 5. Gemins are proteins that form a complex with SMN necessary for snRNP assembly and splicing (Paushkin et al., 2002; Kolb et al., 2007). Gemin 2 is an SMN binding protein that decreases in cultured cells and in mice when SMN is decreased (Wang and Dreyfuss, 2001; Feng et al., 2005; Gabanella et al., 2007). Consistent with these data, we found that gemin 2 was also decreased in smn−/− zebrafish. Gemin 5 has been show in mouse SMA models and cell culture data to not be affected by SMN levels (Feng et al., 2005; Hao et al., 2007) and our analysis of gemin 5 was consistent with these results as well. Taken together, we show that Pls3 levels correlate with Smn levels. Moreover, our data in zebrafish are consistent with previous data in mouse and cell culture indicating that the relationship between Smn and these proteins is conserved among species.
Before ~4 dpf, Smn levels in smn−/− larvae are present at wild-type levels due to maternally deposited RNA. Protein from these maternal transcripts can be transiently reduced using a translation blocking antisense morpholino (MO) against zebrafish smn (smn MO) as previously described (McWhorter et al., 2003; Carrel et al., 2006). Therefore, we asked whether Pls3 levels were decreased when Smn was reduced transiently in embryonic development. Either smn or control MO was injected into wild-type zebrafish (MO injected embryos are referred to as morphants) and protein for western blots collected at 1 dpf. As in the smn−/− mutants, Pls3 levels were also reduced in smn morphants during embryogenesis (Fig.1B). To address the reciprocal question, we asked whether Pls3 affects Smn protein levels. Using a morpholino against pls3, we decreased Pls3 levels but saw no change in Smn levels in 1 dpf larvae (Fig. 1C). Thus, decreasing Smn, results in decreased Pls3, but decreasing Pls3 does not affect Smn.
To determine whether decreased Pls3 was a common phenotype of motoneuron disease in general, we examine Pls3 expression in two zebrafish models of ALS. We had previously generated and characterized a transgenic line, Tg(sodG93R) over expressing mutant zebrafish Sod1 as a genetic model of ALS (Ramesh et al., 2010). Subtle neuromuscular junction (NMJ) changes were observed at 11 dpf in these animals, thus we examined Pls3 expression at this time. Western blot analysis revealed an increase in Sod protein consistent with over expressing the mutant zebrafish Sod protein, SodG93R, via the transgene. However, no change was observed in Pls3 expression levels in Tg(sodG93R) zebrafish compared to wild-type controls (Fig. 1D). Since we do not see motor axon defects in this line, we wanted to examine a zebrafish motoneuron disease model that like smn morphants, also showed motor axon defects. TDP43 has been genetically linked to ALS (Kwiatkowski et al., 2009) and knock down experiments in zebrafish have shown that tdp43 morphants exhibit motor axon defects (Kabashi et al., 2010; data not shown). Thus, we examined Pls3 expression in tdp43 morphants. Western blot analysis on 3 dpf larvae showed that neither Smn nor Pls3 were reduced in tdp43 morphants (Fig. 1E). These data indicate that decreased Pls3 levels are not a common occurrence across motoneuron diseases, but may be unique to SMA.
If Pls3 levels were dependent on Smn levels, then we would expect that restoring Smn expression in an Smn-depleted animal would rescue Pls3 protein levels. To test this, we created a transgenic zebrafish line over expressing RFP tagged human SMN (RFP-hSMN) (Rossoll et al., 2003) under the control of the zebrafish heat shock 70 promoter (hsp70), Tg(hsp70:RFP-hSMN) (Table 1; Fig. 2A). Upon heat shock for one hour at 37C, RFP-hSMN is expressed throughout the whole embryo (Fig. 2A). To analyze the RFP-hSMN level after heat shock induction, we performed western blot on protein obtained from embryos/larvae 0.5 to 120 hours after heat shock. We found that RFP-hSMN was detectable 0.5 hours after induction and was constant between 1 and 6 hours post induction (data not shown). We then examined protein turnover after induction. We heat shocked at 1 dpf then looked at 1–5 days after induction (2–6 dpf). The levels were highest 1 day post induction, reduced at 3 days and present at only very low levels at 5 days post induction (Fig. 2B, lanes 2–6). We then repeated the heat shock induction at 6 and 8 dpf, and in so doing kept the RFP-hSMN at levels comparable to endogenous Smn (Fig. 2B. lanes 6–9). Therefore, by heat shocking for one hour every other day, we could maintain RFP-hSMN in these transgenic fish.
To determine whether restoring SMN to fish with low Smn effects Pls3 protein levels, we crossed this transgenic line onto the smn−/− background. Embryos were first heat shocked at 1 dpf to select transgenic embryos then those carrying the transgene were heat shocked again at 5 and 7 dpf. At 8 dpf all of the larvae were genotyped and protein extracted. Western blots revealed that when RFP-SMN was added back to smn−/− larvae, Pls3 levels increased to levels comparable to those seen in wild types (Fig. 2C–D). These data indicate that SMN levels modulate Pls3 protein levels.
To determine if reduced Pls3 expression was the result of low levels of pls3 RNA, we examined pls3 RNA in smn−/− larvae. Eight day old larvae from an smn+/− incross were genotyped and RNA prepared and quantitated by qRT-PCR. We found that pls3 RNA levels in wild type and smn−/− were not statistically different (Fig. 3A). To ensure that the Tg(hsp70:RFP-SMN) was not altering pls3 RNA levels, we analyzed pls3 RNA in Tg(hsp70:RFP-SMN);smn−/− larvae with and without heat shock. Again, we found that the pls3 RNA levels were not affected (Fig. 3A). SMN is known to function in the assembly of RNAs and proteins important for RNA splicing (Meister et al., 2002; Paushkin et al., 2002). Thus, we examined whether pls3 RNA was mis-spliced in smn mutants. RNA was extracted from 11 dpf smn−/− larvae and full-length zebrafish pls3 RNA was amplified by PCR. PCR products were run on 8% polyacrylamide gel to aid in resolving any splice products. Using this approach, we did not find any alterations in splicing of endogenous pls3 (data not shown). These data indicate that pls3 RNA is not affected by Smn.
To address whether Smn could be affecting pls3 RNA stability, we generated a transgenic line driving human PLS3 under the hsp70 promoter, Tg(hsp70:PLS3) (Table 1). This transgene was used to induce a pulse of PLS3 expression throughout the animal upon heat shock induction. By measuring PLS3 turnover in wild-type and Smn depleted larvae, we could calculate PLS3 RNA half-life. We injected 1–2 cell Tg(hsp70:PLS3) embryos with control or smn MO and at 1 dpf heat shocked the embryos to induce PLS3. RNA was collected at various time points and levels quantitated using qRT-PCR (Fig. 3B). RNA decay was then calculated from the slope of the log plot (see methods) (Fig. 3C). Using this approach, we found no difference in the amount of PLS3 RNA generated in control versus smn morphants (Fig. 3B). Moreover, we found that the half-life of PLS3 RNA in control and smn morphants was not statistically different (1.31± 0.02 hrs versus 1.33 ± 0.01 hrs, p=0.42) indicating that PLS3 RNA stability was not affected by Smn levels (Fig. 3C).
Since PLS3 RNA is unaffected by Smn levels and thus could not explain why Pls3 levels decrease in smn−/−, we next examined the effect of Smn on PLS3 protein induction and stability. We performed the same experiment as above for analyzing RNA, but this time analyzed PLS3 protein levels. We heat shocked control and smn morphant Tg(hsp70:PLS3) embryos at 1 dpf and collected protein 0.5 hrs post heat shock. Surprisingly, we found at this early time point that control morphants had a 4-fold greater amount of induced PLS3 than smn morphants (Fig. 4A, B). To determine whether this was due to protein stability, we analyzed protein turnover by collecting larvae and extracting protein at 3–6 days post induction. At all times examined there was significantly less PLS3 protein in smn compared to control morphants as determined by western blot (Fig. 4C). We also found that the half-life of PLS3 in control and smn morphants was not different (1.92 ± 0.09 versus 1.72 ± 0.15 days, p=0.14) (Fig. 4D). Together these data rule out a role for Smn in PLS3 transcription, splicing, RNA and protein stability and suggest that Smn affects PLS3 translation.
Although we could not induce high levels of PLS3 in smn−/−, we wanted to know whether even this slight increase could rescue survival on a low SMN background. For these experiments we used the transgenic line discussed above, Tg(hsp70:PLS3), to express PLS3 ubiquitously upon heat shock induction (Table 1). To elucidate the role of PLS3 solely in motoneurons, we also generated a transgenic line expressing PLS3 just in motoneurons by using the zebrafish mnx1 (hb9) promoter driving expressing of PLS3, Tg(mnx1:PLS3) (Table 1). We crossed both the Tg(hsp70:PLS3) and the Tg(mnx1:PLS3) onto the smn−/− background. The transgenic fish were selected based on GFP expression (see methods and Table 1) at 1 dpf then grown until 20 dpf and survival determined as described (Boon et al., 2009). Fish carrying Tg(hsp70:PLS3) were heat shocked every other day to induce ubiquitous expression. Analyzing these lines, we found no difference in the survival of smn−/−, Tg(hsp70:PLS3);smn−/− , and Tg(mnx1:PLS3);smn−/− larvae (Fig. 5). These data indicate that low levels of PLS3 cannot rescue survival of smn−/− larvae. We are not able to test higher levels of PLS3 since PLS3 is not produced well when Smn levels are low (see Fig. 1, ,22 and and44).
Our previous data showed that PLS3 could rescue the motor axon defects caused by low Smn levels in zebrafish (Oprea et al., 2008). We also reported in this paper that decreasing PLS3 using a MO was difficult as the embryos were either deformed or died. We have since performed a large number of injections using two different pls3 MOs and found a concentration of the translation blocking pls3 MO that decreased Pls3 protein (Fig. 6A) but did not affect embryonic development. This was confirmed by morphological analysis (Fig. 6B, C) and primordia (prim) staging at ~28 hpf with control morphants at prim stage of 10.0 ± 0.5 (n=9) and pls3 morphants at prim stage 9.8 ± 0.9 (mean ± s.d) ( Kimmel et al., 1995). Analysis of pls3 morphants revealed that these embryos displayed motor axon defects such as aberrant branching and large growth cones (Fig. 7A, C, G). These defects were similar, but not identical, to the motor axon defects in smn morphants and show that decreasing Pls3 can lead to motor axon defects. We did not observe defects in the trigeminal peripheral sensory axons, Rohon-Beard peripheral sensory axons, or the lateral line nerve in pls3 morphants suggesting that axial motor axons are more sensitive to decreased Pls3 levels (Fig. 6D–G). To show that these motor axon defects were indeed caused by a decrease in Pls3, we co-injected pls3 MO and PLS3 RNA and found that we could rescue the motor axon defects compared to pls3 morphants (Fig 7A–D, F. p<0.0001). We next asked whether decreased Pls3 in motoneurons caused these defects. To test this, we injected pls3 MO into Tg(mnx1:PLS3) embryos that express PLS3 in their motoneurons and observed statistically fewer motor axon defects compared to pls3 morphants (p<0.0001; Fig. 7 C, E, F). These data indicate that motor axon defects are indeed caused by decreasing Pls3 in motoneurons since they can be statistically rescued by driving PLS3 expression solely in motoneurons.
We next wanted to test whether PLS3, even at low levels, could rescue other defects in the neuromuscular system caused by Smn deficiency. We previously showed that smn−/− larvae had decreased SV2 neuromuscular junction (NMJ) expression at 11 dpf (Boon et al., 2009). This indicates that the NMJs are altered when Smn levels are depleted. Furthermore, expressing human SMN protein exclusively in motoneurons in these mutants rescued the SV2 phenotype (Boon et al., 2009). To test whether PLS3 could rescue this defect, we analyzed NMJs in Tg(mnx1:PLS3);smn−/− fish. At 11 dpf, fish were genotyped and stained with SV2 antibody (Fig. 8A). Quantification of SV2 immunohistochemistry confirmed earlier data showing that smn−/− had decreased SV2 at 11 dpf (Fig. 8) (Boon et al., 2009). Importantly, when PLS3 was expressed in motoneurons, the SV2 defect was rescued (p<0.0001, Fig. 8). These data reveal that PLS3 expressed in motoneurons, even at low levels, rescues the presynaptic NMJ defect caused by low Smn.
To determine if rescue of the SV2 NMJ defect resulted in enhanced motor function, we examined movement behavior in smn−/− fish in the presence of PLS3. Zebrafish have characteristic movements that can be measured by analysis of motor output (Wolman and Granato, 2011). To capture larval motor behaviors, we examined the initiation of spontaneous turning and swimming bouts at 9 dpf by recording video at 1000 frames/sec for 400 milliseconds at 3-second intervals. Using the automated tracking analysis FLOTE (Burgess and Granato, 2007b, a), we determined whether each larva performed a ‘swim’, ‘routine turn’, or did not move for 60 trials of 400 millisecond videos. Interestingly, smn−/− larvae initiated turning and swimming bouts significantly less frequently than their wild-type and smn+/− siblings (Fig. 9, p<0.001), suggesting that the presynaptic motoneuron deficits in smn−/− mutants reduced motor activity. To test whether PLS3 could rescue these motor defects, we analyzed motor behavior in Tg(mnx1:PLS3);smn−/− larvae. We found that driving PLS3 in motoneurons could significantly rescue the behavioral deficits in smn−/− larvae. Indeed, Tg(mnx1:PLS3);smn−/− larvae initiated turning and swimming bouts statistically better than smn−/− mutants (p< 0.001) and similar to wild types (p=0.16). These data show that PLS3 expression in motoneurons rescues the motor behavioral deficits caused by low levels of Smn indicating that loss of Pls3 significantly contributes to SMA phenotypes.
PLS3 is an actin binding protein that decreases the disassembly rate of actin and controls actin turnover and the length of actin filaments (Giganti et al., 2005). Because actin dynamics are important for both axonal outgrowth and synaptic regulation (Dent et al., 2011; Cingolani and Goda, 2008), alterations in actin dynamics can greatly affect these processes. Our data show that when Smn is decreased, Pls3 levels are also decreased and this is likely controlled at the level of translation. This causes pre-synaptic changes at the NMJ and aberrant motor behaviors that can be significantly rescued by adding low levels of PLS3 back to motoneurons. Decreasing Pls3 independent of Smn supports that motor axons are sensitive to low levels of Pls3 even more so than other axons. Therefore, when SMN levels are decreased, as in SMA, PLS3 decreases thus compromising axons and synapses leading to motoneuron dysfunction. This is a cell autonomous effect that can be rescued by adding PLS3 back to motoneurons. These data support that changes in the actin cytoskeleton contribute to SMA phenotypes.
Work by others has also shown a link between SMA phenotypes and actin dynamics (Bowerman et al., 2007, 2009, 2010; Nölle et al., 2011). RhoA, a small GTPase that promotes growth cone collapse and axon stalling, increases when SMN is depleted from cultured PC12 cells (Bowerman et al., 2007). In addition rho-kinase is upregulated via a mechanism that includes profilin (Nölle et al., 2011), a protein that binds SMN (Giesemann et al., 1999). Blocking rho-kinase improved NMJs and survival in an intermediate SMA mouse model (Bowerman et al., 2010) supporting that changes in the actin cytoskeleton are an important component of SMA phenotypes. Data from an SMA mouse model also supports a link between SMN and PLS3 levels. Bowerman et al., (2009) showed in a mild model of SMA, that PLS3 is decreased approximately 40% in spinal cord. The less dramatic decrease likely derives from the fact that these animals have higher levels of SMN than smn−/− zebrafish and are a milder model of SMA. Two studies in humans have shown a correlation between PLS3 expression in blood and SMA severity in females (Oprea et al., 2008; Stratigopoulos et al., 2010) while another failed to show a strong correlation (Bernal et al., 2011). It is perhaps not surprising that these studies are somewhat variable since a modifier may not be present in all patient populations. It may be that PLS3 levels are higher in some individuals and that this affords protection when SMN levels are low. Lastly, Genetic modifier screens in C. elegans and Drosophila have also identified pls3 as a gene that can modify phenotypes caused by low Smn levels (Dimitriadi et al., 2010). Thus, PLS3 as a modifier spans human, zebrafish, Drosophila and C. elegans. SMA models across species have indicated defects at synapses, which depend on endocytosis for synaptic vesicle recycling. Interestingly the yeast homologue of pls3, sac6, effects endocytosis (Adams et al., 1989; Kubler and Riezman, 1993) and a report has shown a link between huntingtin, ataxin 2, the endocytic protein endophilin and Pls3 (Ralser et al., 2005). Thus, Pls3 pathways may play an important role in synaptic maintenance and function in disease.
We were surprised to find such a dramatic decrease in Pls3 in smn−/− larvae. To investigate the mechanism, we analyzed both RNA and protein levels. Decreased levels of SMN lead to defects in RNA splicing (Gabanella et al., 2007; Zhang et al., 2008), however, we did not find any splicing defects in smn−/− pls3 or changes in pls3 RNA levels or RNA decay. We therefore reasoned that Smn could be impacting either Pls3 protein translation or stability. To look at these parameters, we analyzed production and stability of PLS3 via an inducible transgene. As early as 0.5 hours post induction, there was dramatically less (~ 4-fold) PLS3 protein under conditions of low Smn compared to wild type. This was not due to delayed production, since at all subsequent time points PLS3 levels were significantly lower compared to PLS3 levels in wild types. We reasoned that this could be caused by decreased PLS3 stability. It has been suggested that the decrease in gemin 2 observed upon SMN reduction is due to decreased gemin 2 protein stability (Feng et al., 2005). Thus, we examined PLS3 protein half-life in the context of low Smn, but found no significant difference when compared to PLS3 half-live in wild types. These data strongly suggest that PLS3 translation is severely compromised when SMN is decreased. The mechanism and whether SMN is acting directly or indirectly in this process remains to be determined. Whereas the endogenous zebrafish pls3 has introns and a 3'UTR, the human PLS3 cDNA used to make the transgenics did not contain either yet both the endogenous and transgenic PLS3 were affected similarly when Smn was decreased. This does not, however, rule out mechanisms such as microRNAs (miRNAs) as some have been shown to bind to sequences in open reading frames (Forman et al., 2008; Huang et al., 2010). There is also a suggested link between SMN and miRNAs. Gemin3, which binds to SMN, is also found in a complex with eIF2C2 and miRNAs (Mourelatos et al., 2002). Since SMN binding to gemin 3 is decreased when SMN levels are decreased (Charroux et al., 1999), this could affect gemin 3 function. Another possible mechanism is that decreased Smn levels could cause a protein change (perhaps because it is splicing improperly) in a protein important for translation initiation (Sonenberg and Hinnebusch, 2009). Although the mechanism of how SMN is affecting translation is unknown at this time, it raises the question of whether other proteins are also not effectively translated under conditions of low SMN.
Because PLS3 could not be expressed well in an Smn deficient background, we really could not test its impact on survival. Therefore, it is rather remarkable that PLS3 from the Tg(mnx1:PLS3) transgene could rescue both the SV2 pre-synaptic defect and the motor behavior. This indicates that even small increases in PLS3 can have an effect on neuromuscular phenotypes. This is a characteristic consistent with a modifier in that small changes in a modifier protein would be expected to alter the phenotype. We had previously shown that driving SMN in motoneurons on a smn−/− background could rescue the SV2 NMJ defect but could not rescue survival (Boon et al., 2009) much like driving PLS3 in motoneurons could rescue the SV2 NMJ defect and motor behavior but not survival. These data indicate that SMN levels need to reach a certain threshold in all cells, not just motoneurons, for survival of the animal. This is consistent with data from a reciprocal experiment in SMA mice showing that decreasing SMN just in motoneurons yields a more mild disease phenotype compared to when SMN is decreased to low levels in all tissues (Park et al., 2010).
Our data support a role for SMN in PLS3 translation and indicate that decreased PLS3 levels lead to phenotypes seen in SMA such as presynaptic defects and movement defects. Thus, PLS3 is not merely an informational suppressor but has a direct link to SMN and SMA. We speculate that decreased PLS3 contributes to SMA phenotypes, along with other cellular defects caused by low SMN levels such as splicing (Gabanella et al., 2007; Zhang et al., 2008) and changes in RNA localization in motoneurons (Akten et al., 2011; Fallini et al., 2011; Rossoll and Bassell, 2009). The accumulation of these defects together could lead to SMA. These data also raise the issue of whether translation control could be a downstream affect of low SMN.
We thank the OSU fish facility staff for fish care and Dr. Brunhilde Wirth for the PLS3 antibody, Dr. Livio Pellizzoni for the gemin 2 antibody, and Dr. Glenn Morris for the SMN, gemin 5, and profillin 2 antibodies. The SV2 monoclonal antibody developed by K. M. Buckley was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. This work was funded by National Institute of Health grant NS050414 (C. E. B), The SMA Foundation (C. E. B.) with additional support from OSU Neuroscience Center Core Grant P30 NS045758, by a NRSA postdoctoral fellowship (5F32EY019434-03) to M.A.W. and a National Institute of Health grant (MH092257) to M.G.
Any Conflict of Interest: The authors declare no conflicts of interest
Author contributions: L. T. H. helped design the experiments and performed all of the experiments except the behavior. M. W and M. G. performed the behavior. C. E. B. designed the experiments, provided oversight of the study, and wrote the manuscript with L. T. H.