Additional results

Previously, we showed that LNP-1 plays an important role in the synaptic vesicle cargo distribution and thus regulates the synaptic strength in adult C. elegans. We decided to focus on the following questions:

• Does LNP-1 defects influence C. elegans development, and how? We could obtain viable lnp-1-deletion adults, but is their embryonic development affected?

• Does the vesicle cargo misplacement affect the normal life span of the mutant animals?

Are these defects observed in adults becoming more severe with age?

• What is the molecular pathway in which LNP-1 is acting? Which are its partners?

• Is the post-synaptic element affected in lnp-1-deletion mutants? Or is the function of LNP-1 restricted pre-synaptically? If not, does LNP-1 act in similar or different ways to regulate pre- and postsynaptic proteins?

• Is the function of C. elegans LNP-1 conserved in mammals?

Answering these questions might improve the comprehension of synaptic proteins regulation and synaptic strength regulation.

4.2.1 LNP-1 IS EXPRESSED VERY EARLY IN EMBRYOGENESIS

To study whether LNP-1 is influences the embryonic development of C. elegans, we first mapped when LNP-1 expression is upregulated during development. Immunofluorescence assays on different stages of C. elegans embryos showed that LNP-1 protein is present in the oocytes and throughout embryogenesis (Figure 1A). In very early embryos, LNP-1 exhibits a punctuated staining, which is present in the cytoplasm of all cells. Later, starting with 2-fold stage (plum stage), LNP-1 becomes enriched in the anterior part of the worm and later, in larval stages, it is enriched in the nervous system. A time line of C. elegans embryogenesis with the outline of the important stages is presented in Figure 1B. To compare the development of wild-type worms with the development of lnp-1(tm1247) mutant, we filmed the embryonic development and separated it in four steps:

• step 1 – from 2-cells embryo until the beginning of gastrulation (26-cells),

• step 2 – from 26-cells until bean stage,

• step 3 – from bean stage to 1.5-fold stage (tadpole),

• step 4 – represents the moment in which the embryo starts twitching.

The amount of time spent by the developing embryo in each of these stages is illustrated graphically in Figure 1C. There is no statistical difference in the embryonic

Figure 1. A) Immunofluorescence on different stages of worm embryos with DAPI, anti-LNP-1 antibody and anti-MH27 antibody (epithelial marker); B) Time line of C. elegans embryogenesis (from WormAtlas); C) The time spent by wild-type (in dark red) and lnp-1(tm1247) mutant (in blue) in different stages of development is similar.

development between wild-type and lnp-1(tm1247) mutant embryos, although lnp-1(tm1247) mutant seems to spend slightly more time in each step. Additional tests, like the moment of laying the first egg, confirmed that lnp-1(tm1247) development is normal.

The lnp-1 gene is expressed very early and throughout embryogenesis, as well as in the adult worm stage. The embryonic development of C. elegans was not affected by mutations in lnp-1 gene.

4.2.2 MUTATIONS IN LNP-1 INDUCE A REDUCTION OF THE C. ELEGANS LIFE SPAN AND EARLY NEURODEGENERATION

To determine if the vesicle cargo mislocalization affected the normal life span of the lnp-1 mutants, we measured the life span of lnp-1 mutants versus wild-type animals at 20°C. Wild-type adults live 20.67±1.53 days (mean±SD), whereas lnp-1(tm1247) and lnp-1(tm733) adults have an average life span of 14.67±1.16 days and 12.67±1.16 days, respectively (Figure 2A).

We then examined the RAB-3::Venus localization in the DA 1-9 cholinergic motorneurons using the transgenic strain nuIs168 in 1-day old adults and compared with 5-days old animals in order to establish whether the lnp-1-deletion related defects were influenced by the age of the animals. In 1-day old wild-type adults, RAB-3::Venus is evenly distributed in punctuated structures along the dorsal nerve cord (Figure 2B, left panel), while in lnp-1(tm733) and lnp-1(tm1247) animals, RAB-3::Venus is more diffusely distributed (Figure 2B, middle and right panels), the positive fluorescent puncta being more numerous (Figure 2C, left graph), but smaller in diameter (Figure 2C, right graph). In 5-days old adult lnp-1(tm733) and lnp-1(tm1247) animals, the RAB-3::Venus pattern was more disorganized compared with age-matched controls (Figure 2B, lower panels). We also observed an increase in puncta size, but the puncta density was drastically reduced (Figure 2C). The increase in size might represent cargo accumulation and the unusually shaped and widely spaced puncta might indicate both structural changes in synapse morphology and possible synapse retraction or loss. The increases in size and number of misaccumulations, combined with reduced life span, have been observed in certain neurodegenerative diseases (Cleveland and Rothstein, 2001). For example, four principal factors have been put forward to account for motor neuron death in ALS (amyotrophic lateral sclerosis): oxidative damage, axonal strangulation from the disorganization of neurofilaments, toxicity from intracellular aggregates and/or a failure of protein folding, and excitotoxic death arising from the mishandling of glutamate. The involvement of axonal strangulation is supported by the abnormal accumulation of neurofilaments in cases of sporadic and familial ALS. Deficits in slow axonal transport arise early on in ALS, and neurofilament mutations can be direct causes of motor neuron death in mice. The presence of intracellular aggregates is also common feature of familial ALS. It has been proposed that the intracellular aggregates might be toxic, and this toxicity might arise from protein misfolding mediated by

mutant SOD1 (superoxide dismutase 1). The mutant SOD1 chronically ties up chaperones, which are needed for catalysing the folding of other proteins, whereas ubiquitin-mediated protein degradation might be congested by these aggregates (Cleveland and Rothstein, 2001).

Recently, Evason et al. proposed that neural activity regulates lifespan in C. elegans (Evason et al., 2005) and elaborated two models (Kornfeld and Evason, 2006). In the first model neuronal activity sustains life by mediating essential sensory or motor activities, for example, breathing in vertebrates. As age-related degenerative changes impair neuronal function, the activity of these life support systems decreases, culminating in death and thereby limiting life span. According to this model, interventions that delay neuronal aging and sustain neuronal activity might extend the life span. In the second model neural activity promotes age-related degenerative changes in nonneuronal cells, which are essential for life support. Neurons might Figure 2. A) Life-span curve for wild-type and lnp-1 mutants; B) RAB-3::Venus expressed in DA 1-9 neurons in wild-type and lnp-1 mutants from 1-day adults compared with 5-days adults; C) Puncta density per 50#m remain constant in wild-type, while in lnp-1 mutants is severely decreased. Puncta diameter is decreased with age in wild-type worms, but increasing in lnp-1 mutants (probably due to cluster formation).

affect the nonneuronal cells relatively directly, for example, by forming a synapse on a muscle cell or, more indirectly, by stimulating an endocrine organ to release a hormone. According to this model, interventions that decrease neural activity might delay the aging of nonneuronal cells and extend the life span (Kornfeld and Evason, 2006). The worms treated with anticonvulsant drugs like ethosuximide and/or trimethadione display a significant extension of their lifespan and delay age-related degenerative changes. The life span extension caused by the anticonvulsant drugs is independent of the IGF pathway, in contrast to many of the genetic and cell ablation studies, which suggests that neural activity affects aging by means of multiple mechanisms. These drug-treated worms are also hypersensitive to aldicarb-mediated paralysis suggesting that the anticonvulsant stimulates synaptic neurotransmission in the neuromuscular system (Evason et al., 2005). Conversely, our results showed that lnp-1 mutants, resistant to aldicarb, displayed also a significant reduction in their lifespan. These results suggest once more a link between the neural activity and lifespan.

4.2.3 SIAH-1: A PUTATIVE PARTNER FOR LNP-1

Yeast two-hybrid screening which allowed successful identification of protein-protein interactions at small and large scale in C. elegans, has been previously described (Crowe and Candido, 2004; Li et al., 2004; Malone et al., 2003; Walhout et al., 2000). Using this approach, we screened a C. elegans cDNA library, using as bait a deleted form of LNP-1, which does not contain the two predicted transmembrane domains. The expression of full length LNP-1 in E.

coli system leads to the production of an insoluble protein, which accumulates in inclusion bodies. In two independent yeast two-hybrid screens, we obtained eight positive clones, which were sequenced and further analysed. All these clones code for the Y37E11AR.2 gene (Figure 3A). Y37E11AR.2 gene is located on chromosome IV of C. elegans and encodes for a putative Ring finger/E3 Ubiquitin ligase sharing about 60% amino acid identity with the Seven In Absentia protein (SINA; Figure 3C). We renamed the Y37E11AR.2 gene as siah-1 (sina homolog; accepted and registered by the Caenorhabditis Genetics Center). This interaction was further confirmed by immunoprecipitation (IP) experiments, which were performed by Patrizia Latorre. The %lnp-1 and siah-1 cDNAs were inserted in mammalian expression vectors and the resulting pCIHA::%LNP-1 and pCDNA3FLAG::SIAH-1 constructs were coexpressed in HEK293T cells. IPs were performed from transfected cell lysates using an anti-HA antibody. Reaction products analyzed by Western blot with an anti-FLAG antibody showed that HA antibody could specifically immunoprecipitate a complex formed by &LNP-1 and SIAH-1, confirming that these two proteins are physically interacting (Figure 3B).

The nematode siah-1 gene encodes a 419 amino acid protein containing several domains conserved among all orthologs identified so far (Figure 3C): an N-terminal RING finger domain (C3HC4) involved in ubiquitination as ubiquitin ligase, followed by a conserved

cysteine/histidine region (C2HC4H3) in the C-terminal region (Figure 3D) involved in the interaction with protein targets (Hu and Fearon, 1999). SINA was originally discovered as a

Figure 3. A) !-galactosidase activity was detected only when competent yeast cells were cotransformed with &LNP-1 and the SIAH-1 candidates; B) IP experiments confirming direct interaction between &LNP-1 and SIAH-1. HEK293T cells were cotransfected with the indicated constructs in the figure. The cell lysates were immunoprecipitated with anti-HA antibody and the presence of SIAH-1 was detected with anti-FLAG antibody. C) Alignments of SIAH of H. sapiens, D. rerio, X. laevis, Drosophila, C. elegans, C.

briggsae. Identities are highlighted in dark grey; D) Description of C. elegans siah-1 and siah-1-deleted genes. SIAH-1 protein has 420 amino acids and two distinct domains. siah-1 mutants contain a truncated SIAH-1 protein, which does not contain the protein-protein interaction domain.

RING finger-containing protein that is critically involved in neuronal development of the R7 photoreceptor cell in Drosophila (Carthew and Rubin, 1990). SINA, together with PHYLLOPOD, promotes the ubiquitin proteasome-dependent degradation of TRAMTRACK, a negative regulator of neuronal differentiation (Tang et al., 1997). In mammals, Siah proteins have been shown to target to degradation different proteins including Deleted in Colorectal Cancer (DCC), synaptophysin, synphylin-1, numb, glutamate receptors, etc. (Hu et al., 1997; Moriyoshi et al., 2004; Nagano et al., 2003; Susini et al., 2001; Wheeler et al., 2002). These studies suggest that Siah proteins may act to regulate neuronal development and function by mediating the ubiquitin-dependent degradation of a number of neuronal target proteins.

We obtained a siah-1 mutant, siah-1(tm1968), containing a frame-shift deletion, from the NBRP consortium. The resulting protein contains the Ring finger domain and the last residues of the C-terminal part of the protein; however, the conserved cysteine/histidine region (C2HC4H3) from the C-terminal region involved in the interaction with protein targets is missing (Figure 3D). The homozygous strain is viable and does not show any striking behavioural and/or developmental defects. The 1(tm1968)-deletion strain was used in further analyses of siah-1 function and lnp-siah-1/siah-siah-1 interaction in C. elegans.

4.2.4 E3/UBIQUITIN LIGASE ACTIVITY OF SIAH-1 IS CONSERVED IN C. ELEGANS

To investigate whether the E3/ubiquitin ligase activity of SIAH-1 is conserved in C.

elegans, we performed in vitro ubiquitination assays. As anticipated, polyubiquitinated products were only detected when GST::SIAH-1 was present; moreover, this effect was dose-dependent (Figure 4A, lanes 1-6) and ATP/E1/E2-dependent (Figure 4A, lanes 10-12). A putative LNP-1 intrinsic E3/ubiquitin ligase activity was tested by incubating increasing amounts of purified GST::%LNP-1 protein in the absence of GST::SIAH-1. The formation of polyubiquitinated conjugates was detected in the presence of GST::%LNP-1 (Figure 4A, lanes 7-9) but not when either ATP, or ubiquitin, or E1, or E2 were omitted from the reaction (Figure 4A, lane 10-12).

However, the concentration of polyubiquitinated products detected in the presence of GST::%LNP-1 alone was significantly lower compared to that obtained in the presence of comparable concentration of GST::SIAH-1 alone (Figure 4A, compare lanes 6 and 7).

Moreover, no GST::%LNP-1 dose dependency was observed in the formation of polyubiquitinated products in the absence of GST::SIAH-1 (Figure 4A, compare lanes 7 to 9).

We showed that SIAH-1 E3-ligase activity is conserved in C. elegans and that LNP-1 seems not to have an E3-ligase activity. One possible explanation of LNP-1 – SIAH-1 interaction is that LNP-1 could be either a new ubiquitination target of SIAH-1 or its co-factor.

LNP-1 is a SUMOylated protein

Small ubiquitin-related modifier (SUMO) addition on proteins has several functions, among the most frequent and best studied are protein stability, nuclear-cytosolic transport and transcriptional regulation. As opposed to ubiquitin modification, which targets proteins for degradation, SUMOylation increases a protein's lifetime (reviewed in (Wilson and Heaton, 2008). The SUMOylation reaction is very similar to the ubiquitination pathway, being also ATP-dependant and using E1-activating, E2-conjugating and E3-ligating enzymes. One LKSP-peptide, which is required for SUMO ligases recognition was predicted bioinformatically in the LNP sequence (see Results, Chapter 4.3). To test if LNP is SUMOylated, we performed in vitro SUMOylation assays. The SUMOylation reaction is ATP-dependent (Figure 4B, lane 1 and 3).

All three SUMO ligases acted on LNP to form SUMOylated conjugates, however the best

Figure 4. A) In vitro ubiquitination experiments. Lanes (1-6): Increasing amounts of GST::SIAH-1 protein (respectively, 0.01;0.02;0.05;0.1 and 0.25 µg) were incubated with ATP, E1, GST::E2 and His::Ubiquitin and in the absence of GST::&LNP-1. Lanes (7-9): Increasing amounts (0.2; 1 and 2 µg) of purified GST::&LNP-1 protein were incubated in presence of E1, GST::E2 and His::Ubiquitin and in the absence of GST::SIAH-1. Lanes (10-12): Polyubiquitinated conjugates are not detected when E1/E2 or SIAH-1/LNP-1 are missing from the reaction mix. Polyubiquitinated conjugates were detected with an anti-Ubiquitin antibody. B) In vitro SUMOylation experiments. Lane (1,3): ATP is required for the SUMOylation reaction. Lane (2): Positive control: RANGAP1 is known to be SUMOylated by

results are obtained by SUMO 2/3 ligases (Figure 4B, lane 4-6). The SUMOylated LNP-1 could represent the active form, protected against the ubiquitination activity of SIAH-1 partner.

4.2.5 SIAH-1 AND LNP-1 COLOCALIZE IN C. ELEGANS NERVOUS SYSTEM

To provide further support in favour of LNP-1/SIAH-1 interaction in vivo, we constructed a gene fusion between the reporter gene gfp (or rfp) and a 3-kb region upstream the siah-1 coding sequence. Transgenic and double transgenic worms expressing either siah-1 alone or siah-1 and lnp-1 were generated by microinjection (see Figure 3D). Similar with plnp-1::GFP transgene expression pattern, numerous neuronal cell bodies along the ventral cord, around the pharynx and tail expressed the psiah-1::RFP transgene (Figure 5A,B). Furthermore, worms expressing both psiah-1::RFP and plnp-1::GFP transgenes displayed high colocalization of the reporters in the nervous system (Figure 5C-H). As no antibodies against C. elegans SIAH-1 were available, we used a commercially available anti-human SIAH 1/2 antibody raised against the highly conserved C-terminal part of the human protein, which turned out to be positive in C.

elegans. Immunostaining in whole worms showed that endogenous SIAH-1 colocalized with LNP-1 (Figure 5I-K). The colocalization of LNP-1 with SIAH-1 in both transgenic animals and immunohistochemistry assays confirms in vivo the interaction between these two proteins.

Recently, we requested the production and obtained an anti-SIAH-1 antibody specific

Figure 5. SIAH-1 colocalizes with LNP-1 in the nervous system. A) psiah-1::RFP is detected in the commissures (asterisk) and cell bodies (arrow) of motor neurons in the ventral cord. B) In the head, RFP is found in the nerve ring and localized to sensory neurons (as demonstrated by the presence of RFP in the dendritic processes, arrowhead). (C-H) Transgenic animals coexpressing psiah-1::RFP (C,F) and plnp-1::GFP (D,G). Neurons containing both RFP and GFP are found in the ventral nerve cord (F, arrow), in the nerve ring and in the tail (respectively arrowhead and asterisk in E,). (I,K) Endogenous SIAH-1 was visualized by immunostaining with human anti-SIAH antibody. SIAH-1 is colocalizing with LNP-1 (J,K respectively). Localization of SIAH-1 protein in cell bodies and in neuronal processes of neurons of the ventral cord is shown (respectively arrows and arrowheads in K). L) Western blot on total protein extracts from N2 (wt)(lane 1), lnp-1(tm733) mutant (lane 2), siah-1(tm1968) mutant (lane 3) and lnp-1(tm733);siah-1(tm1968) double mutant (lane 4) showing SIAH-1 levels (upper panel), LNP-1 levels (middle panel). As loading control "-tubulin was utilized (lower panel).

for C. elegans protein. This antibody was generated against the N-terminal part of SIAH-1 (1-207 peptide) and tested in western blot obtaining positive bands at expected size (Figure 5L).

Using this antibody on total protein extracts from siah-1(tm1968) mutant, we observed a lower band (around 25kDa), probably the N-terminal part of SIAH-1 which is still produced in this deletion mutant. We also observed an increase in the SIAH-1 protein levels in lnp-1(tm733) mutant and lnp-1(tm733);siah-1(tm1968) double mutant (Figure 5L). Using this antibody in immunohistochemistry assays, the localization pattern of endogenous SIAH-1 and co-localization with endogenous LNP-1 should confirm the results obtained with the commercial anti-SIAH1/2 antibody.

4.2.6 E3/UBIQUITIN LIGASE SIAH-1 INVOLVEMENT IN SYNAPTIC FUNCTION

To further demonstrate the functional importance of the LNP-1/SIAH-1 interaction in vivo, we performed full-time course aldicarb assays. As already showed in Chapter 4.1, the lnp-1 mutant displayed an increased resistance to aldicarb. Similarly, siah-lnp-1 mutant presented an increased resistance to aldicarb paralysis, but less stronger compared with lnp-1 mutant.

Interestingly, the double mutant lnp-1(tm733);siah-1(tm1968) showed normal behaviour in presence of aldicarb drug (Figure 6A). To study if $SIAH-1 is responsible for the rescued behaviour or if the total absence of SIAH-1 protein will induce the same outcome, we examined the behaviour of animals fed with bacteria expressing lnp-1 or siah-1 dsRNAs in the presence of the aldicarb drug (Figure 6B). As expected, both lnp-1(dsRNAs) and siah-1(dsRNAs) treated animals displayed a significantly increased resistance to aldicarb, compared to control RNAi animals (Figure 6B, left panel). As expected, siah-1(tm1968) mutants fed with lnp-1(dsRNAs) behaved as wild-type in presence of aldicarb (Figure 6B, right panel). When siah-1(dsRNAs) expressing bacteria was fed to lnp-1(tm733) no rescue was observed (Figure 6B, middle panel) suggesting that the presence of $SIAH-1 is required for the normal behaviour in presence of aldicarb of lnp-1 mutant. Moreover, the thrashing rate test showed a mild but significant 19%

reduction of the thrashing rate of lnp-1(RNAi) and siah-1(RNAi) animals, compared to control animals (Figure 6C). To prove that RNAi feedings indeed produced a reduction in lnp-1 and siah-1 RNA levels, we performed RT-PCR on animals fed with respective dsRNAs (Figure 6D).

4.2.7 LNP-1 AND E3/UBIQUITIN LIGASE SIAH-1 REGULATE THE NORMAL DISTRIBUTION OF PRESYNAPTIC PROTEIN SNB-1 IN THE VENTRAL NERVE CORD

To investigate potential functions of the SIAH-1 in neurons, we examined C. elegans siah-1 mutants for changes in SNB-1 containing synapses. To visualize these synapses, we analyzed the distribution of GFP-tagged SNB-1 (SNB-1::GFP). We previously showed that SNB-1::GFP distribution was significantly altered in lnp-1 mutants (Ghila and Gomez, 2008).

GFP fluorescence was no longer restricted to the pre-synaptic sites, but was detected in both cellular bodies and neuronal processes. Likewise, SNB-1::GFP distribution was affected, even Figure 6. A) Full-time course aldicarb assay showing that lnp-1 mutant and siah-1 mutant are more resistant to the paralysis induced by the drug compared to wt. B) RNAi feeding experiments with lnp-1(dsRNA) and siah-lnp-1(dsRNA) confirms the results from mutant analysis; C) Thrashing assay on dsRNA feed animals shows that lnp-1 and siah-1 depleted animals are less active in liquid; D) RT-PCR on wt and lnp-1(tm733) mutant confirms a decrease in RNA levels after RNAi feedings.

though less strongly, in siah-1 mutant (compare Figure 7B with 7C). Similar defects in SNB-1::GFP distribution were observed in lnp-1(dsRNAs) and siah-1(dsRNAs) treated animals (Figure 7 D-F). These results suggest that both SIAH-1 and LNP-1 are required for normal distribution of SNB-1::GFP at pre-synaptic sites.

4.2.8 LNP-1 AND E3/UBIQUITIN LIGASE SIAH-1 REGULATE THE ABUNDANCE OF GLR-1 IN THE VENTRAL NERVE CORD

To test if LNP-1 function is restricted at the presynaptic sites or it is involved in general synaptic regulation, we analyzed the localization of a postsynaptic protein, namely glutamate receptor 1 (GLR-1). The nuIs24 strain expresses the GLR-1::GFP transgene driven by its endogenous promoter (Rongo et al., 1998). Expression of GLR-1::GFP in ventral-cord interneurons rescues the behavioural defects caused by glr-1 null mutations (Rongo and

To test if LNP-1 function is restricted at the presynaptic sites or it is involved in general synaptic regulation, we analyzed the localization of a postsynaptic protein, namely glutamate receptor 1 (GLR-1). The nuIs24 strain expresses the GLR-1::GFP transgene driven by its endogenous promoter (Rongo et al., 1998). Expression of GLR-1::GFP in ventral-cord interneurons rescues the behavioural defects caused by glr-1 null mutations (Rongo and