Carbonic Anhydrase I modifi es SOD1- induced motor neuron toxicity in Drosophila via ER stress pathway

AAZ: Acetazolamide; ALS: Amyotrophic Lateral Sclerosis; FALS: Familial Amyotrophic Lateral Sclerosis; SALS: Sporadic Amyotrophic Lateral Sclerosis; ASK1: Apoptosis Signal Regulating Kinase 1; ATF4: Activating Transcription Factor 4; BSA: Bovine Serum Albumin; CA: Carbonic Anhydrase; CA1: Carbonic Anhydrase I; CAH1: Drosophila Carbonic Anhydrase form 1; CAH2: Drosophila Carbonic Anhydrase form 2; hCA1: Human Carbonic Anhydrase 1; CMT1A: Charcot-Marie-Tooth type 1A; DDS: Direct Digital Synthesis; Derlin-1: Degradation in Endoplasmic Reticulum Protein 1; dDerlin-1: Drosophila degradation in Endoplasmic Reticulum Protein 1; hDerlin-1: human Degradation in Endoplasmic Reticulum Protein 1; DMSO: Dimethyl Sulfoxide; dsod: Drosophila copper/zinc superoxide dismutase gene; E106I: Human CA1 mutant with a mutation in the amino acid at position 106 from glutamic acid (E) to isoleucine (I); eIF2: eukaryotic translation Initiation Factor 2 subunit alpha; p-eIF2: phosphorylated eukaryotic translation Initiation Factor 2 subunit alpha; ER: Endoplasmic Reticulum; FUS: Fused in Sarcoma, an RNA-binding protein; G93A: Human copper/zinc superoxide dismutase mutant with a mutation in the amino acid at the position 93 from glycine (G) to alanine (A); GFP: Green Fluorescent Protein; HNPP: Hereditary Neuropathy with liability to Pressure Palsies; Hsp70: The 70 kDa heat shock protein; JNK: c-Jun N-terminal Kinases; p-JNK: phosphorylated c-Jun N-terminal Kinases; PFN1: Pro ilin 1; PMP22: Peripheral Myelin Protein-22; SOD1: Copper/Zinc superoxide Dismutase; SOD1G85R: Human copper/zinc superoxide dismutase mutant with a mutation in the amino acid at the position 85 from glycine (G) to arginine (R); dSOD1: Drosophila Copper/Zinc Superoxide Dismutase; hSOD1: human Copper/Zinc Superoxide Dismutase; TBST: Tris-Buffered Saline, pH 7.5, 0.05% Tween 20; TDP-43: TAR DNA-Binding Protein; TRAF2: TNF Receptor-Associated Factor 2; XBP1: X-box Binding Protein 1; CA1 lies: Transgenic lies expressing human CA1 in motor neurons; E106I lies: Transgenic lies expressing human CA1 mutant E106I in motor Abstract


Introduction
Amyotrophic lateral sclerosis (ALS) is an adult-onset motor neuron disease characterized by the predominant loss of motor neurons in the spinal cord, brain stem, and motor cortex [1]. More than 90% of ALS is sporadic (SALS) while about 10% is inherited. According to the most recent review article, a total of 27 genes with diverse functions have been listed to be linked to familial ALS (FALS) [2]. The mechanisms of ALS pathogenesis are complex involving many players and pathways [3]. Recently, it was reported that human carbonic anhydrase I (hCA1) might be a player in ALS pathology as its levels were increased in spinal cord motor neurons of ALS patients including SALS and FALS [4]. Carbonic anhydrases (CAs) catalyze the reversible conversion from carbon dioxide and water to bicarbonate and proton which is the most essential and fundamental process in cellular metabolisms [5]. CAs are responsible for dynamically regulating and maintaining the intracellular and extracellular pH [6]. CA1 is known to be one of the cytosolic isoforms of 17 mammalian CAs which are present in several subcellular compartments [5]. In addition to its expression in erythrocytes, several types of epithelial cells, synovium, and cardiac capillary endothelial cells, CA1 expression was also detected in the motor neurons of human spinal cord [4]. In Drosophila, there are two forms of CAs, CAH1 and CAH2 [7]. Phylogenetic analysis indicates that CAH1 and mammalian cytosolic forms of CAs share a common ancestor while CAH2 shares a common ancestor with the extracellular form of CAs [7]. Drosophila CAH1 (dCAH1) shares 44.6% identity with hCA1 in the overall amino acid sequences ( Figure S1). Fly CAH1 expression can be detected in the salivary gland, hindgut and midgut, eye as well as the brain [7]. Knockdown of ly CAH1 in the midgut resulted in a less acid environment [8]. More importantly, changes in CA1 levels have been detected in several pathological conditions including the in lammation, cancer and brain injury [9][10][11][12][13][14][15][16][17][18]. However, the roles CA1 might play in motor neurons in the contexts of both physiology and pathology are completely unknown.
The copper/zinc superoxide dismutase (SOD1) was the irst gene discovered in 1993 that is responsible for 20% of FALS [19]. Tremendous amount of information has been obtained in rodent models of SOD1-linked ALS as well as from in vitro mammalian cells. It has been established that mutant SOD1s cause ALS in a dominant fashion by gain of toxic function rather than loss of enzyme activity [20,21]. One piece of the overwhelming data to support this gain of toxicity was the fact that more than 130 mutant SOD1s selectively interact with human Derlin-1 (hDerlin-1) while the wild-type SOD1 does not [22]. Furthermore, the mechanism of the gained toxicity by the SOD1 mutants is the activation of endoplasmic reticulum (ER) stress response resulting in trapping apoptosis signal-regulating kinase 1 (ASK1) in the aggregates followed by downstream activation of the c-Jun N-terminal kinases (JNK) pathway [23]. The Drosophila model of SOD1-linked toxicity in motor neurons was also established [24]. Even though the SOD1 gene is highly conserved among many species, Drosophila SOD1 protein (dSOD1) shares 61% identity with human SOD1 (hSOD1) [24]. In addition, overexpression of wild-type or ALS-linked mutant hSOD1s in ly motor neurons was toxic, while overexpression of the Drosophila SOD1 gene (dsod) was not [24]. Stress response of Hsp70 activation in glia was reported to be associated with the toxicity [24]. Similar to what's discovered in mouse [20], dsod null lies had defect but did not readily exhibit ALS-like phenotype [25]. Interestingly, the Drosophila knock-in model of hSOD1 mutations was reported [25]. Flies homozygous for the knock-in dsod which contained the equivalent human ALS mutation were lethal, while lies heterozygous for the same knock-in mutant gene exhibited ALS-like pathology in a recessive manner [25]. Because mutant SOD1s act in a dominant fashion in humans, we decided to use the ly model of ALS which overexpresses the mutant form of ALS (SOD1 G85R ) as originally described by Watson et al. [24] for our study. In order to understand the role of CA1 in ALS pathology, we decided to examine whether SOD1-induced toxicity in ly motor neurons can be modi ied by the presence of hCA1. We also make the following hypotheses: 1) the molecular mechanism of SOD1-induced damage in ly motor neurons might be through Derlin-1, ASK1 and JNK pathways; 2) if CA1 can modify the process, it can also imply for the irst time a new role CA1 might play in motor neurons.

Plasmid constructs
The cDNA for human CA1 was purchased from Genscript (Clone ID: OHu24513, Piscataway, NJ, USA). The E106I mutant form of CA1 was generated via site-directed mutagenesis via two rounds of PCR as described by Mohanty et al. [26]. Brie ly, a total of 4 primers were designed and used in the PCRs. The sequences of the primers are: F1: The nucleotide sequences in lower cases in Primers F2 and R2 indicate the mutated nucleotides that will result in the altered amino acid from glutamic acid (E) to isoleucine (I) at the amino acid poistion 106 in the CA1 peptide. The inal PCR fragment was puri ied and cloned into pcDNA3.1 (+) and the mutations were con irmed by sequencing. The cDNA of CA1 or E106I was cloned into the Xba I and Kpn I site of the pUASTattB plasmid for making transgenic lies.

The crawling test
Third instar larvae were used for the crawling test conducted at 25 o C. A dissecting microscope equipped with a camera, DDS function generator and computer were used to image the larval locomotion movements on the surface of 1% agarose gel in a 10-cm dish against a black background together with a reference ruler. One larva was taken out from the vial and put on the agarose gel to allow it to acclimate to the environment for 2 min after which time its movement was captured by the camera using a customized macro command in ImagePro software. The video images were collected by taking a snap picture every 0.5 second for a total period of 100 seconds. The larva's position on the plate was manually tracked and the total crawling distance of its movements was calibrated using the reference calibration curve.

The climbing assay
Male lies of F 1 progeny from the correct genotypes including GFP, CA1, SOD1 and CA1+SOD1 lies were collected within 24 hrs after eclosion and designated as Day 1 lies. A total of 10 male lies were collected in one vial. The climbing assay was conducted according to the method described by Romero et al. [27]. Brie ly, on Day 2, the lies in each vial were transferred to a clean vial on which a 5 cm line was marked from the bottom. After being acclimated in the testing vial for 5 min, the lies were tapped down to the bottom of the vial and the number of lies which climbed above the 5 cm line at least once within 18 seconds were recorded. This performance was repeated twice for each vial. The climbing index is calculated as the average of the three recorded numbers divided by the total number of lies in each vial. When no lies were alive in a vial, the climbing index for that vial no longer exists for that time point and all the time points afterwards. All lies were tested every three days, from Day 2 to Day 38, the data from all vials from different independent experiments were combined and analyzed. After the test, the lies were placed in a new vial with fresh growth medium.

Measurement of the rate of survival
The rate of survival was also measured for the lies used in the climbing test every three days and the total number of alive lies on the test day were recorded. The rate of survival was calculated by the recorded ly number on the testing day divided by the total number of lies on Day 1 for each vial. Once no lies were alive in a vial, the survival rate for that vial reached zero and remained zero for all the time points afterwards. Data from all vials in different independent experiments were combined and used for making the survival curves.

Western blot analysis
Determination of transgene expression in lies was conducted by extracting proteins from an individual ly using the 1X sample loading buffer (50 mM Tris-HCl pH 6.8/2% SDS/10% glycerol/1% β-mercaptoethanol/0.02 % bromophenol blue). A total of 3 randomly collected lies were test for each cross. Examination of changes in the levels for proteins involved in pathology was performed using the dissected head and trunk materials from the ly. The tissues were homogenized in the lysis buffer containing 20 mM Tris-HCl (pH 7.5)/12 mM β-glycerophosphate/150 mM NaCl/5 mM EGTA/10 mM NaF/1% Triton X-100/1% deoxycholate/1 mM DTT/1 mM sodium orthovanadate, and 1X proteinase inhibitor cocktail (ThermoFisher Scienti ic, Waltham, MA, USA). An equal number of lies was used for each time point. Protein samples are separated on the SDS/PAGE gel, transferred onto the nitrocellulose membrane and probed with the desired antibodies. The membrane was blocked in either milk or BSA in TBST, incubated with the primary antibody, washed, incubated with the secondary antibody and washed again. The images were acquired on Odyssey and the data were analyzed using Image Studio Lite Ver 5.2 software.

Co-immunoprecipitation
The head and thorax of the ly was dissected out and used for co-immunoprecipitation (co-IP). Proteins were extracted in the same lysis buffer as used for Western Blot analysis by homogenizing the tissues and taking the supernatant after microcentrifugation at the maximal speed for 30 min at 4 o C. Protein concentrations were determined using the BCA method (ThermoFisher Scienti ic, Waltham, MA, USA). Co-IP was performed in the IP buffer (20 mM Tris/150 mM NaCl/1% Triton, pH 7.5) with a ratio of 5g: 1 mg for antibody and protein, respectively. The IP volume is 350 l and proteins were pre-cleaned with Protein G beads (GE Healthcare, Pittsburgh, PA, USA) for 2hr at 4 o C, followed by incubation with the IP antibody overnight at 4 o C. The beads were washed with IP buffer three times followed by one last wash in 1X PBS. Both a fraction of the supernatant and all proteins on the beads were subjected to Western Blot analysis.

Antibodies
The GFP antibody was from ProSci (XR-9001, Poway, CA, USA). The SOD1 antibody used in Western Blot analysis was from Enzo Life Sciences (ADI-SOD1-100, Farmingdale, NY, USA), the one used in co-IP was from Leica Biosystems (NCL-SOD1, Newcastle, UK). The actin antibody was from Sigma (A2066, St Louis, MO, USA). The antibodies against CA1 (ab124976, used in Western Blot; ab54912, used in co-IP), tubulin (ab52866) were from Abcam (Burlingame, CA, USA). The antibodies that detect only the phosphorylated eIF2 (CST#9721) and JNK (CST#9251) were from Cell Signaling Technology (Davers, MA, USA). The antibody speci ic for detecting Drosophila Derlin-1 but not human Derlin-1 was generated by making a polyclonal antibody against the epitope sequence of RAPPRQATESPWG produced by GenScript (Nanjing, China). The speci icity of the antibody was con irmed by epitope peptide competition experiment using proteins from ly and human samples.

Statistical analysis
Statistical analysis was performed using either the Student T-test or GraphPad Prism 8 software.

CA1 causes motor neuron toxicity at the larval stage and in adults
In order to understand whether CA1 can modulate mutant SOD1-induced toxicity, we irst examined locomotion function in transgenic lies that express hCA1 in the motor neurons by the binary expression system of the D42-GAL4 [28] and UAS-hCA1 (see also Materials and Methods). These lies will be described as CA1 lies for this study. The locomotion function was measured by the crawling distance and climbing index at the larval stage as well as in adults, respectively. In addition, the rate of survival was also determined. Western Blot analysis of proteins from the individual ly demonstrated that the respective transgene expression of GFP and CA1 in ly motor neurons was readily detectable by the antibodies (Figure 1a, GFP & CA1). Using the 3 rd instar larvae grown in the regular medium for the crawling test, it was discovered that the average distance CA1 larvae crawled in 100 seconds was signi icantly shorter than that by GFP larvae (Figure 1c), indicating that there is a locomotion defect caused by expression of hCA1 in ly motor neurons during the larval development. Locomotion behavior was also tested in adult CA1 lies via the classic antigeotaxis test every 3 days starting Day 2 after eclosion until Day 38. The climbing index was de ined as the percentage of the numbers of lies reached the marked 5 cm distance at least once in 18 seconds in a vial, which is used as an indication of locomotion activity. It was discovered that the climbing index exhibited by CA1 lies was signi icantly lower than that of GFP lies as early as Day 8 (Figure 2a). The climbing index continued to decline in CA1 lies to be about 30.5% while GFP lies maintained an average climbing index value of 57.5% at Day 38 (Figure 2a). In other words, the locomotion activity in CA1 lies was about half of that of GFP lies between week 5 and week 6 in adult life. Consequently, CA1 lies had also shortened lifespan demonstrated by the reduced survival rate in the same period of time ( Figure 2d). Therefore, we conclude that for the irst time we have demonstrated that hCA1 when expressed in ly motor neuron can cause toxicity as re lected in the locomotion defect as well as the reduced rate of survival in adult lies.

CA1-induced motor neuron toxicity is activity-dependent
The next question we were interested in asking was whether the observed CA1-induced toxicity requires the enzymatic activity. To address this, transgenic lies expressing a mutant form of CA1 (E106I) with a mutation in the conserved amino acid residue Glu106 in the motor neurons were established. Mutations in Glu106 were known to inactivate CA [29] and  4-9). The CA1 antibody used here also detected E106I mutant. MW standards are indicated by the numbers in kDa on the right side of the images. b) Quantitation of the levels of CA1 and E106I mutant expression in transgenic fl ies. c) The scatter plot of the crawling distances measured in 3rd instar GFP and CA1 larvae grown in regular media. The number of larvae used for each group is indicated by "n". The p value was determined by one-way ANOVA analysis using non-parametric and Mann-Whitney tests. The dashed black lines represent Mean ± SE. "*" is used to indicate p < 0.05. the loss of enzymatic activity in E106I was also con irmed by the carbonic anhydrase activity assay ( Figure S2). The mutant E106I expression was detected by the same CA1 antibody in individual lies (Figure 1a, lanes 6-9) and the level of E106I mutant protein is somewhat lower (about 70% of that of the wild-type CA1) when compared to CA1 expression ( Figure  1b). In addition, the CA inhibitor acetazolamide (AAZ) was also included in the ly growth medium as an alternative way of inactivating the anhydrase activity in 3rd instar larvae for the crawling test (Figure 3a-3c). DMSO was included in the growth media as AAZ is only soluble in DMSO. While DMSO did not have any effect on the reduced locomotion activity in 3rd instar CA1 larvae compared with GFP lies (Compare Figure 1c with Figure 3a), AAZ at two different concentrations of either 1 M or 10 M eliminated such defect in CA1 larvae ( Figure  3b and 3c, respectively). Similarly, larvae expressing E106I performed no differently from GFP larvae in the crawling distance, reversing the defect seen in CA1 larvae (Figure 3d, compare the purple with green). The anhydrase activitydependent toxicity was also tested in the climbing assay and survival rate in adult lies. The E106I lies, virtually behaved the same as GFP lies with regard to both the climbing index ( Figure 2c) and survival rate (Figure 2f) when compared with CA1 lies (Figure 2, compared 2a with 2b for climbing index; and compare 2d with 2e for the survival rate). Therefore, we conclude that CA1-induced toxicity in ly motor neurons requires its anhydrase activity.

CA1 can ameliorate SOD1-induced toxicity at the early stage of adult fl ies
In order to understand the relationship between SOD1and CA1-induced locomotion defects in adult lies which is relevant to the mechanisms of FALS-linked motor neuron toxicity, we also created transgenic lies expressing both SOD1 and CA1 in the motor neurons (described as CA1+SOD1 lies). Both the climbing and survival tests were conducted in a total of four transgenic ly lines expressing GFP, CA1, SOD1 and CA1+SOD1. The expression of the above transgene of interests in each transgenic ly line was con irmed by the Western Blot analysis (Figure 4). Expression of the mutant form of SOD1 (SOD1 G85R ) caused toxicity in the climbing test up to Day 14 ( Figure 5a). However, this SOD1-induced toxicity plateaued and reached a signi icant bene icial effect compared with GFP lies at Day 38 ( Figure 5a). In contrast and as described earlier, CA1 caused the continued decline in motor function compared with GFP lies (Figure 2a). Interestingly, when CA1 was expressed in addition to SOD1 in ly motor neurons, the locomotion activity in CA1+SOD1 lies was restored to that of the control GFP lies (Figure 5d and Figure S3C). This CA1rescuing SOD1 toxicity and "returning to normalcy" was most unambiguously and clearly observed within the irst 2 weeks or up to Day 14 (Figure 5b and Figure S3A). The rescue effect was two-edged in a sense that SOD1 also signi icantly rescued CA1-induced toxicity, as demonstrated by p=0.003, between CA1 and CA1+SOD1 lies in the same period of time ( Figure  5c and Figure S3B). At the later stage of the adult ly life, particularly after Day 26, SOD1-induced toxicity plateaued while CA1 continued to be toxic (Figure 5e), with the former being dominant to the latter eliminating the CA1-induced toxicity completely to be no different from that of GFP lies (Figure 5c and 5d). This above statement is consistent with the enhanced p value (p<0.0001) between SOD1 and CA1+SOD1 lies within 14 days ( Figure S3A) than p=0.0402 within 38 days (Figure 5b), while the p values for CA1 and CA1+SOD1 lies exhibited a reverse trend: p=0.003 and p<0.0001 within 14 and 38 days, respectively ( Figure S3B and Figure 5c).

The ER stress pathway is involved in SOD1-and CA1induced toxicity
Due to the fact that there obviously is a crosstalk between SOD1-and CA1-induced motor neuron toxicity in adult lies, we irst tested the possibility of whether there was either a direct or indirect interaction between SOD1 and CA1 by coimmunoprecipitation using proteins from ly tissues. We did not detect a potential interaction between the two proteins using this method ( Figure S4). Studies with mammalian cells have established that almost all mutant SOD1s can gain toxicity by abnormally interacting with Derlin-1 which induced ER The data for each graph were analyzed by two-way ANOVA followed by Bonferroni's multiple comparisons test using the GraphPad Prism software. The signifi cance of the difference between the two groups was indicated by the p value next to "N" in the graph area where "*" is used to indicate p < 0.05. The signifi cance between two data points at a given time was also labeled by "*" in which: *: p < 0.05; **: p < 0.01, ***: p < 0.001; and ****: p < 0.0001, situated above each data point in the graph line. The data values and error bars represent Mean ± SEM.  Data for each graph were analyzed by two-way ANOVA followed by Bonferroni's multiple comparisons test using the GraphPad Prism software. The signifi cance of the difference between the two groups was indicated by the p value next to "N" in the graph area where "*" is used to indicate p < 0.05. The signifi cance between two data points at a given time was also labeled by "*" in which: *: p < 0.05; **: p < 0.01, ***: p < 0.001; and ****: p < 0.0001, situated above each data point in the graph line. The data values and error bars represent Mean ± SEM. stresses leading to aggregates formation and activation of ASK1 [23]. In addition, the ER stress response is one of the earliest pathological features in mutant SOD1-linked FALS models in rodents [30]. To determine whether SOD1-induced motor neuron toxicity in lies may involve similar mechanisms, we hypothesized that the mutant SOD1 might also be able to interact with Drosophila Derlin-1 (dDerlin-1), as the sequence identi ied in hDerlin-1 which interacts with mutant SOD1s shares homology with the corresponding sequence in dDerlin-1 ( Figure S5A). However, co-immunoprecipitation experiment did not yield positive interaction between the two proteins using either puri ied proteins (data not shown) or proteins from ly tissues ( Figure S4, bottom panel; Figure  S5B and S5C). Nevertheless, changes in the levels of molecules that are known to be upregulated in ER stress response were examined (Figure 6a). One of such molecule is eIF2α and its activation by phosphorylation (p-eIF2α). The temporal regulation of levels of p-eIF2α were determined in transgenic lines at Day 3, Day 14, and DaWWy 26. For the irst time, we documented an early and peak activation of eIF2α at Day 3 in SOD1 lies (Figure 6b, orange bars) which subsided to the level not different from that in GFP lies at Day 14 ( Figure  S6A, Day 14, green and orange bars). This is supportive of the notion that the ER stress pathway known to be engaged in mammalian cells [23], is also involved in mutant SOD1induced toxicity in Drosophila, the latter of which has not been demonstrated before. Differently, a delayed activation of eIF2α was observed in CA1 lies which started to increase at Day 3 but peaked at Day 14 ( Figure 6b, Day 14, red bars; Figure S6A, Day 14, green and red bars). In contrast, there is no detectable activation of eIF2α in CA1+SOD1 lies compared with GFP lies (Figure 6b, blue bars), suggesting that eIF2α activation can serve as an indication of the underlying mechanism that results in the counteractive effect of CA1-and SOD1-induced toxicity observed in early ly adulthood ( Figure 5).

CA1-and SOD1-induced toxicity involves JNK activation
To understand the toxicity at the level of apoptosis, the changes in phosphorylated JNK (p-JNK) were also examined. Similar activation of JNK at Day 14 were observed in both CA1 and SOD1 lies (Figure 6c, red and orange bars) which again was not seen in CA1+SOD1 lies (Figure 6c, blue bars). This is consistent with the normal locomotion function observed in the climbing test in CA1+SOD1 lies until Day 14 ( Figure 5). The activation of JNK in CA1 lies at Day 26 was not signi icantly different from Day 14 (Figure 6c, no signi icant difference for the red bars between Day 14 and Day 26), while in SOD1 lies, the activation of JNK was signi icantly decreased at Day 26 compared with Day 14 (Figure 6c, orange bars, p=0.03 between Day 14 and Day 26). These changes correlated with the sustained locomotion defect in CA1 throughout the time course while the defective behavior stabilized and ultimately improved in SOD1 lies when compared to GFP lies after Day 14 (Figure 2a and Figure 5a).

Conclusion and Discussion
Our study for the irst time characterized a novel ly model of motor neuron toxicity that might be relevant to ALS. Previously discovered human CA1 whose levels were shown to be increased in the spinal cord of ALS patients [4], has now been demonstrated that when overexpressed in ly motor neurons is also toxic (Figure 2a & 2d). Among the established Drosophila models of ALS-linked genes, 5 molecules including SOD1, TDP-43, FUS and PFN1, when their wild-type human proteins were overexpressed in ly neurons caused toxicity [31,32,24]. Apparently, CA1 behaved similarly to these ive molecules. Our approach in the effort to understand how CA1 might modulate motor neuron function is to see how it might affect a known ALS-related phenotype which human mutant SOD1-induced toxicity in ly motor neurons. It was noted that this SOD1-induced toxicity in our study was observed earlier (Figure 5a) than the published data [24]. This difference can be attributed to the different control used in the experiments. While lies overexpressing the endogenous ly SOD1 were use in the climbing test by Watson et al. [24], we used lies overexpressing GFP in our study. There de initely is a crosstalk between CA1-induced toxicity and SOD1-mediated pathway as demonstrated by the data in igure 5. The net result of this crosstalk is the reversal of two toxicities to normal, implying the mechanism of the individual toxicity can nullify each other. Such rescuing effect is of the similar nature to the published study in which the defect in axonal transport caused by a mutation in dynein completely rescued SOD1 mutant G93Ainduced de icit in axonal transport seen in the transgenic mouse model of ALS, even though the molecular mechanism was not entirely clear [33]. In theory, there are at least two potential possibilities for such interaction. One is that there might be a shared and rate-limiting factor in the CA1-and SOD1-induced pathways. When both proteins are present, the amount of this factor is rendered to be insuf icient for either pathway resulting in the disappearance of the toxicity. Second, there is a common factor or pathway affected by CA1 and SOD1 and for the normal cell function, the level of this factor or the active state of this pathway has to be maintained at an optimal value. An example of such factor is peripheral myelin protein-22 (PMP22) whose levels are tightly regulated. PMP22 is a component of myelin which covers the nerves and is responsible for the ef icient transmission of neuronal impulses. Increased levels of PMP2 due to an additional copy of the PMP22 gene results in a neurological disorder known as Charcot-Marie-Tooth type 1A (CMT1A), while decreased levels PMP22 due to heterozygous deletion of the gene causes a different neurological disorder known as hereditary neuropathy with liability to pressure palsies (HNPP). If CA1 caused a change in the level of the factor or the degree of the active pathway in one direction, while SOD1 in the opposite direction, this can lead to disruption of the normal function. When both proteins are present, opposing effects on the changes of the factor /pathway resumes the optimal level of the condition and normal cellular function. Further investigations are needed to elucidate the mechanisms of this complex and interesting phenomenon of crosstalk between CA1 and SOD1-induced toxicity. The fact that CA1 can modify SOD1-induced toxicity helped to test the hypothesis whether there might be a shared signaling pathway leading to motor neuron dysfunction. Although the physical interaction between SOD1 and dDerlin-1 was not detected in our study, we demonstrated that the ER stress response pathway was involved by the activation of eIF2α in SOD1 lies. This is consistent with the data in mammalian cells that ER stress-induced ASK1 activation was responsible for mutant SOD1-linked toxicity [23]. It is entirely possible that mutant SOD1 can physically engage with dDerlin-1 in a transient and/or regulated way (that was not detectable by the co-immunoprecipitation method) which leads to toxicity, or there is dDerlin-1-independent signaling pathway. On the other hand, it is equally possible that SOD1-induced toxicity in ly motor neurons is dDerlin-1-independent. Furthermore, the same activation of eIF2α was also detected in CA1-induced defect, but which peaked at a later time compared with that by SOD1 (Figure 6b, compare the red and yellow bars). In other words, an earlier rather than later ER stress response might be critical in determining whether the neurons are going to be able to overcome or succumb to the stresses. Indeed, it appeared that the quicker activation of eIF2α helped SOD1 neurons in the long run survive the stress (Figure 5a) while the relatively delayed eIF2α activation in CA1 lies resulted in persistent toxicity in motor neurons (Figure 2a). This type of correlation between the onset of ER stress response and cell survival was also reported by Walter et al., with regard to the kinetics of XBP1 splicing and ATF4 translation and their correlations with cell death [34]. The JNK pathway is known to be activated in response to stress and in Drosophila, activation of JNK is normally associated with apoptosis [35]. It is also known that ER stress can cause activation of JNK via TRAF2 and kinases complex including ASK1. Increased p-JNK levels were observed within CA1 and SOD1 lies at Day14, which can be potentially associated with eIF2a activation. However, no signi icant JNK activation was detected in CA1+SOD1 lies, which is consistent with the lack of locomotion defect in these lies, suggesting the toxicity seen in lies are potentially due to ER stress response and downstream activation of JNK signaling pathway. It was noted that CA1 and SOD1 lies exhibited lower levels of activation of JNK at Day 3 and Day 26 compared with GFP lies ( Figure S6B). It is not clear what contributes to this lower level of p-JNK in CA1 or SOD1 lies. One might speculate that initial CA1-and SOD1-induced stress might pre-conditioned cells to lower the p-JNK levels to better cope with further activation. If so, the fact that the level of p-JNK in CA1+SOD1 lies was signi icant lower than either CA1 or SOD1 lies at Day 26 ( Figure S6B, compare the blue to red or orange bars) would be consistent with the counteractive crosstalk between the two pathways resulting in even better coping mechanisms. These pure speculations warrant future investigations.
To summarize, our study for the irst time con irmed that mutant SOD1 can induce ER stress response and JNK activation in ly motor neurons, which is consistent with the mechanisms elucidated in mammalian cells [36][37][38][39][40]. We also used the transgenic ly model to further characterize a recently reported molecule CA1 whose levels were changed in ALS patients and discovered that CA1 can induce ER stress response in ly motor neurons. Most importantly, CA1 can cancel out SOD1-linked ALS pathology in lies within the irst two weeks in adulthood indicating that the potential crosstalk and time-sensitive interactions between signaling pathways induced by apparently different toxic proteins can bene it cells or organisms depending on the nature and kinetics of the associated cellular pathways. Such information is important and potentially useful for discovering drugs in therapeutic development.