MT-JIP3 induces JNK signaling–dependent apoptosis in HeLa cells, reducing cell growth, and JIP4 interacts with JIP3. Since endogenous JIP3 protein was almost undetectable in HeLa cells (Figure 1C and Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.187199DS1), we first transiently transfected human WT-JIP3 and MT-JIP3 into HeLa cells. Considering the compensatory role of JIP4 (11), siRNAs were used to reduce the expression of JIP4 (Supplemental Figure 1A). Overexpression of WT-JIP3 or MT-JIP3 resulted in cell death and apoptosis (Figure 1, A and B). Importantly, when JIP4 expression was reduced, WT-JIP3 no longer induced cell death and apoptosis, in contrast to the persistent increase in cell death after expression of MT-JIP3 (Figure 1, A and B, and Supplemental Figure 1B), even when endogenous JIP4 was reduced. JIP3 has been reported to activate JNK signaling to induce apoptosis (9, 17). Overexpression of MT-JIP3 continued to activate JNK signaling when JIP4 was reduced (Figure 1C). The protective effect of JIP4 against JIP3’s cytotoxicity prompted us to explore whether they interact. Our findings revealed JIP4 in Flag immunoprecipitants of both Flag-tagged WT-JIP3 and MT-JIP3 in HeLa cells (Figure 1D). However, MT-JIP3 coprecipitated significantly less with JIP4 (Figure 1D and Supplemental Figure 1C) than with WT-JIP3. The apparent interaction of JIP3 and JIP4 suggests that heterodimers may be formed. Our data also suggest that excess JIP3/JIP4 activity may be cytotoxic and that MT-JIP3 may independently acquire cytotoxicity and result in less heterodimer formation.

MT-JIP3 induces JNK signaling and disrupts cargo-laden LE mobility, thereby inhibiting HeLa cell growth. HeLa cells were transfected with siRNAs (siCtrl or siJIP4) recovered in fresh culture medium for 24 hours and then transfected overnight with plasmids encoding either Flag-tagged WT-JIP3 or MT-JIP3, followed by another 24-hour recovery before experiments. *P < 0.05, **P < 0.01. (A) Cell proliferation was measured by MTT assay. Data were normalized to cells transfected with empty vectors. n = 3. (B) Apoptosis was measured by caspase-3/7 activity. Data were normalized to cells transfected with empty vectors. n = 3. (C) Levels of p-JNK, p-c-Jun, JIP3, and JIP4 proteins were measured by Western blot. (D) JIP3 and JIP4 interaction was detected by co-IP. Briefly, Flag antibody was used for pulling down cell lysates, and JIP3 and JIP4 proteins were measured by Western blot. (E) Cell imaging of different siRNA/plasmid-treated HeLa cells incubated with 1 μM Cy3-labeled PS-ASO 446654 for different time points. The colocalization of Cy3-ASOs and Rab7 (LE [LE] marker) were observed by Keyence X800 microscope. Nuclei were marked with white dashed lines. Zoomed-in views were provided for the regions within the red frames, with white arrows indicating representative colocalization sites. Scale bar: 10 μm. (F) Trafficking of Cy3-ASOs along with microtubules in cells was detected using a Keyence BZ-X800 microscope. Microtubules were stained by anti-tubulin antibody. mTOCs were marked with white dashed lines. (G) Live-cell imaging of HeLa cells expressing GFP-labeled microtubules incubated with Cy3-ASO for 15 minutes. Images were recorded at different times, and snapshots are shown. The movement of a particular Cy3-ASO foci is marked by yellow cycles. Quantification of the speed and distances of Cy3-ASO foci to the original position, as measured from approximately 15 cells. P values were calculated with unpaired t test using Prism.

MT-JIP3 disrupts endosomal mobility following microtubule trafficking in HeLa cells and alters the binding affinity with associated proteins. JIP3 has been reported to be an adaptor protein for endosomal trafficking (18). Endosomal processes are primarily responsible for cellular uptake and subcellular distribution of PS-ASOs and the process has been thoroughly studied (19). We examined PS-ASO as a cargo to follow endosomal movement after transient transfection of WT-JIP3/MT-JIP3 in HeLa cells with or without siRNA-mediated reduction of JIP4. Cy3-labeled PS-ASOs were incubated in HeLa cells for 15 minutes to visualize EE localization or for 120 minutes to visualize LE localization, as previous studies showed for ASO endosome trafficking (20). After the indicated time points, cells were stained for Rab7 (LE), which is a well-established LE marker. Colocalization analysis with Rab7 (LE marker) revealed similar patterns in WT-JIP3/MT-JIP3–expressing cells after 15 minutes (Figure 1E). However, at 2 hours, MT-JIP3 cells, particularly with reduction of JIP4, displayed reduced colocalization with Rab7 (LE marker) compared with WT-JIP3 cells (Figure 1E, Supplemental Figure 1D, and videos in Supplemental Figure 1F). Furthermore, Cy3-ASOs moved relatively rapidly along microtubules at the periphery of the cytoplasm toward the minus end (microtubule organizing center [mTOC] in the perinuclear area) in WT-JIP3 cells in 2 hours, as opposed to MT-JIP3 cells, in which reduced mTOC accumulation was observed (Figure 1F and Supplemental Figure 1E). This was further supported by the quantification of the mean trafficking speed and total trafficking distance of PS-ASO foci (Figure 1G). In addition, slower speed and shorter trafficking distance of total cellular PS-ASO foci were observed at 15 minutes after PS-ASO incubation in MT-JIP3 cells (Figure 1G), consistent with the observations that at 2 hours fewer PS-ASO foci accumulated at perinuclear areas (Figure 1F). Collectively, these results demonstrate that MT-JIP3 disrupts endosomal mobility along microtubules in HeLa cells.

MT-JIP3 reduces cell growth and disrupts endosomal mobility in patient-derived fibroblasts. To further substantiate our results in a patient-relevant cell system, we obtained skin fibroblasts from a patient carrying the MAPK8IP3 R578C mutation. The patient-derived fibroblasts express MT-JIP3 at detectable levels (Figure 2B). Importantly, patient-derived fibroblasts showed less cell proliferation and higher induction of apoptosis than any of the other 3 human dermal control fibroblasts (control samples from healthy individuals were from Coriell Institute for Medical Research, GM02936, GM03529, and GM03440) (Figure 2A). Furthermore, LPS induced cell growth and reduced apoptosis in control cells, but patient cells did not respond to LPS (Figure 2A). The activation of p-JNK was elevated in patient fibroblasts compared with control fibroblasts in both the vehicle- and LPS-treated groups. Conversely, the basal level of JIP4 was lower in patient fibroblasts than in control fibroblasts (Figure 2B), which may explain why in this cell model reduction of JIP4 was not necessary to demonstrate the effects of MT-JIP3, as opposed to the HeLa cell model.

MT-JIP3 induces JNK signaling and disrupts cargo endosome mobility, thereby inhibiting cell growth in patient-derived fibroblasts. (A) Cells (2,000 cells/well) were seeded in 96 wells for 3 days. Cell proliferation was measured by MTT assay (left). Apoptosis induction was measured by caspase-3/7 activity (right). All of results were normalized to healthy patient-derived fibroblasts (control cells, 2,936). n = 5, *P < 0.05, **P < 0.01. (B) JIP3, JIP3, p-JNK, and total JNK proteins levels were measured by Western blot and normalized to GAPDH. (C) Trafficking of Cy3-transferrin along with microtubules in cells was detected with a Keyence microscope. Microtubules were stained by anti-tubulin antibody. mTOCs were marked with white dashed lines. Zoomed-in views were provided for the regions within the frames, with white arrows indicating representative colocalization sites. Scale bar: 50 μm; original magnification, ×63. (D) Cell imaging of either patient-derived fibroblasts or control fibroblasts incubated with 1 μM Cy3-transferrin for different time points. Colocalization of Cy3-transferrin and Rab7 (late endosome [LE] marker) was observed with a Keyence microscope. Zoomed-in views were provided for the regions within the frames, with white arrows indicating representative colocalization sites. Scale bar: 50 μm. (E) Cell imaging of either YFP-TLR4–transfected patient fibroblasts or healthy control fibroblasts incubated with 10 μg/mL LPS for different time points. Colocalization of activated YFP-TLR4 and Rab7 observed with a Keyence BZ-X800 microscope. Nuclei were marked with white dashed lines. Scale bar: 50 μm. (F) Cell imaging of lysosomes (LAMP1, a lysosome marker) either in patient-derived fibroblasts or healthy control fibroblasts observed by a Keyence microscope. Average distance between lysosome cluster to the center of nucleus in cells was measured by BZ-X800 analyzer software with the 3D model analysis function. Scale bar: 10 μm. Control cells, n = 85; patient cells, n =109. Unpaired t test.

To further assess effects of MT-JIP3 on endosomal trafficking, we assessed Cy3-labeled transferrin endosomal trafficking and we observed less accumulation of transferrin in the mTOC region in patient fibroblasts after 120-minute incubation with labeled transferrin than that in control fibroblasts (Figure 2C). Moreover, we observed less colocalization of Cy3-labeled transferrin with Rab7 in patient fibroblasts than in control cells at 120-minute incubation (Figure 2D). To further confirm these observations with intracellular cargos, we transiently transfected cells with YFP-labeled TLR4 (21). YFP-labeled TLR4s were extensively colocalized with Rab7 in control fibroblasts upon LPS treatment for 2 hours, but very few colocalization foci were detected in patient fibroblasts (Figure 2E and Supplemental Figure 2B). However, the colocalization of TLR4 and Rab5 at 20 minutes showed no difference between control and patient-derived fibroblasts (Figure 2E and Supplemental Figure 2B). Furthermore, more colocalization of JIP3 with Rab5, but less colocalization with Rab7, was observed in the patient fibroblasts as compared with controls (Supplemental Figure 2, C and D). Furthermore, MT-JIP3 coprecipitated significantly less with JIP4 (Supplemental Figure 2E) than with WT-JIP3, and this was consistent with our observation in HeLa cells. It is important to note that we did not observe any changes in lysosomal location (Figure 2F), suggesting that the R578C mutation in MAPK8IP3 does not result in effects on lysosomes. This result is consistent with our data in the HeLa cell model, in which MT-JIP3 disrupted endosome mobility, pointing toward an established effect of MT-JIP3 on endosomal mobility, which is independent of the cell type. Together, these results demonstrate that MT-JIP3 reduces cell growth, enhances JNK signaling, and disrupts endosome mobility in patient fibroblasts. Thus, the R578C mutations appear to be cytotoxic via at least 2 mechanisms, increased JNK signaling and disruption of endosomal trafficking in two distinctly different types of cells.

MT-JIP3 induces JNK signaling, reducing cell viability, and disrupts axonal trafficking in patient iPSC–derived neurons. To further dissect the role of MT-JIP3, we differentiated the patient’s induced pluripotent stem cells (iPSCs) cells into neuronal cells. The expression of MAPK8IP3 (JIP3) RNA increased markedly as differentiation progressed. However, that was not the case for SPAG9 (JIP4) RNA, and SPAG9 (JIP4) expression only increased during differentiation from initial iPSC cells into neural cells (Figure 3A). This is consistent with our prediction that JIP4 expression is lower in neuronal cells and there is not enough compensation to maintain the normal functions of JIP3/JIP4 in the patient neurons. We also showed that the expression of JIP3 and JIP4 proteins was consistent with RNA expression. In patient-derived neuronal cells, therefore, JIP3 was well expressed, but JIP4 expression was relatively low (Figure 3B). We then assessed JNK signaling in patient neuronal cells. Our data show that JNK signaling was enhanced in patient-derived neuronal cells compared with control neurons (Figure 3B). Next, we assessed cell viability. Consistent with results in other cell types, the patient’s neuronal cells were less viable than control neurons (Figure 3C).

MT-JIP3 induces JNK signaling, disrupts cargo endosome mobility, and causes accumulation of viscosities, thereby increasing risk of death in patient iPSC–derived neurons. (A) mRNA levels of MAPK8IP3 (JIP3) and SPAG9 (JIP4) either in patient iPSCs or iPSC derived neurons were detected by qPCR. n =3. (B) JIP3, JIP4, p-JNK, total JNK, and βIII-tubulin protein levels either in CRISPR control or patient iPSC–derived neurons measured by Western blot. GAPDH was used as a loading control. (C) CRISPR control and patient iPSC–derived neurons cultured in 96 wells for 6 days after DIV14 (14 days in vitro). Cell culture medium was replaced every 2 days. Cell viability was measured by trypan blue cell count assay. n = 3, ***P < 0.001. (D) CRISPR control and patient iPSC–derived neurons were fixed and stained with anti-JIP3 antibody. Viscosities in long axons were detected by STORM with different magnifications. Representative viscosities were pointed out by white arrows. Scale bar: 10 μm (left); 1 μm (right). (E) Cell imaging of iPSC-derived neurons incubated with 1 μM Cy3-PS-ASO for 2 hours. Colocalization of Cy3-PS-ASO (light blue) and Rab7 (late endosome [LE] marker) (red) axons were observed by STORM on ONI Nanoimager. Representative colocalization sites (white) were pointed out by yellow arrows. Scale bar: 10 μm (left); 1 μm (right). (F) Colocalization of JIP3 and KIF5B either in CRISPR control or patient iPSC–derived neurons observed by STORM. Representative colocalization sites (yellow) are indicated by yellow arrows. Scale bar: 10 μm (left); 1 μm (right). (G) Cell imaging of lysosomes (LAMP1, a lysosome marker) either in CRISPR control or patient iPSC–derived neurons observed with a Keyence BZ-X800 microscope. The average distance between lysosome clusters to the center of nucleus in cells was measured by BZ-X800 analyzer software with the 3D model analysis function. Scale bar: 10 μm. CRISPR iPSC–derived neurons, n = 14; patient iPSC–derived neurons, n = 13. Unpaired t test.

Varicosities refer to enlarged presynaptic boutons along the length of axon, which have a key function in neurotransmitter release and indicate the location of synapses under normal physiological conditions (22). We observed that more varicosities accumulated in axons of patient iPSC–derived neuronal cells by either stochastic optical reconstruction microscopy (STORM), which measures single molecule colocalization (Figure 3D and Supplemental Figure 3A), or Keyence microscopy (Supplemental Figure 3B). We also observed less of the colocalization of Cy3-PS-ASO (light blue) and Rab7 (red) in the axon and soma of the patient neuronal cells (Figure 3E). Interestingly, more JIP3 and Rab5 colocalization was shown in patient iPSC–derived neurons compared with that in CRISPR control cells (Supplemental Figure 3C). This suggests that patient iPSC–derived cells exhibit delayed endosomal movement. Furthermore, greater JIP3 (green) and KIF5b (red) colocalization was observed in patient iPSC–derived neurons compared with that in CRISPR controls (Figure 3F). This is consistent with our co-IP results showing altered binding of JIP3 and KIF5b in iPSC-derived neurons (Supplemental Figure 3D). Similar to that in patient-derived fibroblasts, we also did not observe any changes in lysosome distribution in the patient-derived neuronal cells (Figure 3G), further substantiating that the R578C mutation was not affecting lysosomes but late endosomes (LEs). These results indicate that MT-JIP3 impairs axonal trafficking, primarily of LEs in patient neuronal cells, and are consistent with our observations in other cell models that MT-JIP3 is a severely TGOF mutation that causes increased cell death via several mechanisms.

Patient-derived fibroblasts and iPSC-derived neurons showed impaired D1 signaling. NLR family pyrin domain containing 2 (NLRP2) has been reported to be upregulated in iPSCs in bipolar disorder–associated (BD-associated) inflammation (23). Furthermore, iPSC-derived BD cortical neural stem cells harbor multiple abnormalities in the dopamine and GABA receptor pathways. Much higher transcript levels of NLRP2 were observed in patient iPSC–derived neurons than in CRISPR controls (Supplemental Figure 4A). This prompted us to consider whether the mutation affects dopamine signaling, which is closely associated with Parkinson’s disease and might contribute to the limited movement observed in the patient (24). Given that we have employed patient skin fibroblasts for our experiments, and they proved a very important cell system in our efforts to understand the effects of the R578C mutation and develop potential therapeutic approaches, we examined the expression of dopamine receptors in fibroblasts. Dopamine receptors were abundantly expressed in both healthy and patient-derived fibroblasts indeed (Figure 4A). This is in line with previous studies that have highlighted the expression of these receptors in fibroblasts (25). Furthermore, the downstream signal, p-AKT, of dopamine receptor 2 (D2) receptors was activated (Figure 4A). This result demonstrates that fibroblasts can be used as a cell model to detect the effects of mutations on dopamine signaling. Next, we used D1- and D2-selective agonists, SKF81297 (SKF) and ropinirole (ROP), to activate D1 and D2, respectively (26, 27). It is evident that cAMP in control cells increased markedly after 30 minutes of SKF treatment (Figure 4B). In contrast, there was no significant increase in cAMP in patient-derived fibroblasts (Figure 4B). However, upon ROP treatment, cAMP was greatly reduced in healthy control- and patient-derived fibroblasts, and this was time dependent (Figure 4C). D1 and D2 signaling in control cells was a dose dependent (Figure 4, D and E), while only dose-dependent D2 signaling was observed in patient-derived fibroblasts (Figure 4E). These results indicate that D1 is coupled to cAMP signaling in control cells, as expected, but not in patient-derived fibroblasts. In contrast, D2 coupling was apparent in both control- and patient-derived fibroblasts. To further confirm this finding, we used the D2-selective antagonist, domperidone (Dom), to inhibit the effect of D2 and then used SKF to activate D1 in both cells. We found that only control cells responded to D1 activation (Figure 4F). This further suggests that D1 signaling is uncoupled in patient-derived fibroblasts. Although the basal level of cAMP in patient cells was higher than that in control cells (Supplemental Figure 4B), the cAMP in patient cells was unchanged when D2 activation was blocked by the D2-selective antagonist (Dom) (Figure 4G). These results suggest that the higher basal level in patient cells may be driven by other surface G proteins. To further address this question, β1 and β2 adrenergic receptor–selective (AR-selective) antagonists (bisoprolol and ICH118551) were used to block AR-induced cAMP production in both cells (28–30). Interestingly, the basal level of cAMP in patient cells was dramatically decreased and restored to the level of control cells, but D1 antagonist (SCH23390) only inhibited the cAMP induction in control cells (Figure 4H). This further demonstrates that the higher basal level of cAMP in patient cells is likely driven by AR receptors.

Dopamine signaling was impaired in patient fibroblasts (GM03440). (A) Dopamine receptor 1 and 2 (D1 and D2) and p-AKT proteins levels in fibroblasts measured by Western blot. GAPDH was used as a loading control. Either healthy fibroblasts or patient fibroblasts were seeded in 96 wells (10,000 cells/well) and treated with 1 μM (B) D1-selective agonist SKF 81297 and (C) D2-selective agonist ropinirole for different time points. cAMP levels were measured with a cAMP-Glo Assay (Promega). Data were normalized to untreated control (UTC) of cells at time 0. n = 3. Either healthy individual–derived fibroblasts or patient-derived fibroblasts were seeded in 96 wells (8,000 cells/well) and treated by different concentrations of (D) D1-selective agonist SKF 81297 and (E) D2-selective agonist ropinirole for 30 minutes. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. (F) Fibroblasts were treated with 1 μM D2-selective antagonist domperidone (Dom) for 5 minutes before addition of 1 μM D1-selective agonist SKF 81297 (SKF) for 30 minutes. Data were normalized to UTC of cells. n = 3. *P < 0.05. (G) Fibroblasts were treated with different concentrations of D2-selective antagonist domperidone (Dom) for 30 minutes. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. (H) Fibroblasts were treated with 1 μM β1-AR–selective antagonist bisoprolol, β2-AR–selective antagonist ICH118551, and D1-selective antagonist SCH23390 for 30 minutes. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. *P < 0.05, **P < 0.01.

To further confirm the defect in D1 signaling in the most disease-relevant patient cells, we repeated the experiments in patient iPSC–derived neuronal cells. Although D1 and D2 receptors were more highly expressed in patient iPSC–derived neuronal cells as compared with the CRISPR control neurons (Figure 5A), cAMP in patient cells was not induced by SKF, irrespective of the presence or absence of D2 activation (Figure 5B). To further confirm this observation, we used β1 and β2 pan-AR antagonists (timolol and dapiprazole) to inhibit the effects of ARs on cAMP to selectively observe dopamine receptors activation (29, 31). As shown, it is evident that SKF still only activates D1 in control cells (Figure 5C). Next, to further substantiate the connection of JIP3 and D1 and understand the underlying mechanism of MT-JIP3–induced D1 uncoupling in patient cells, we employed co-IP to investigate the direct interaction between JIP3 and the dopamine receptor. As shown, D1 interacted with JIP3 and MT-JIP3 increased this interaction (Figure 5D). This result indicates that MT-JIP3 may overinteract with D1 to interrupt D1 signaling activation in patient cells. In summary, these results demonstrate that MT-JIP3 abnormally interacts with D1 to disrupt D1 signaling.

Dopamine signaling was uncoupled in patient iPSC–derived neurons. (A) mRNA levels of MAPK8IP3 (JIP3), DRD1, and DRD2 either in CRISPR control or patient iPSC–derived neurons were measured by qPCR. n = 3. *P < 0.05. (B) iPSC-derived neurons were treated with 1 μM D1-selective agonist SKF 81297, D2-selective agonist ropinirole, and 1 μM D2 selective antagonist domperidone for 30 minutes. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. *P < 0.05. (C) iPSC-derived neurons were treated with 1 μM D2-selective antagonist domperidone and/or pan–β-AR antagonist timolol and/or pan–α-AR antagonist dapiprazole for 5 minutes before addition of 1 μM D1-selective agonist SKF 81297 or D2-selective agonist ropinirole for 30 minutes. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. *P < 0.05. (D) The interaction of JIP3 and D1 was detected by co-IP. Briefly, JIP3 antibody was used for pulling down cell lysates and D1 and JIP3 proteins were measured by Western blot with D1 and JIP3 antibodies, respectively.

Non-allele-selective PS 2′-methoxyethyl gapmer ASOs activate RNase H1–mediated reduction of both MT- and WT-JIP3 mRNA and protein, restore endosome mobility, and rescue patient fibroblasts from mutation-induced cell death. While JIP4 can compensate for the loss of JIP3, our studies have shown that JIP4 levels were significantly reduced in patient-derived neuronal cells, limiting its compensatory potential. This suggests that an excessive reduction of WT-JIP3 might be toxic in neuronal cells. To assess the risk of reducing WT-JIP3 and determine the necessary degree of allele selectivity for ASOs designed to target only MT-JIP3, we synthesized non-allele-selective PS 2′-methoxyethyl gapmer ASOs (PS-2′MOE ASOs) that target both alleles of MAPK8IP3. PS-2’MOE ASOs are chemically modified oligonucleotides that bind to complementary mRNA sequences, forming DNA-RNA hybrids. This triggers RNase H1 to degrade the RNA strand, reducing the target mRNA levels. The PS-2′MOE modifications enhance stability and binding affinity, making them effective for gene silencing. These ASOs target both WT and mutant alleles, offering potential therapeutic benefits in conditions involving TGOF mutations. However, their non-allele selectivity can also pose risks by reducing essential proteins from WT alleles. Key advantages include enhanced stability and specificity, but challenges like off-target effects and effective delivery remain critical considerations. We developed and tested 3 PS-2′MOE ASOs (ASO1, ASO2, and ASO3, as listed in Supplemental Table 2) to target MAPK8IP3 (JIP3). These PS-ASOs successfully rescued cell proliferation in patient fibroblasts and suppressed MAPK8IP3 expression without affecting SPAG9 (JIP4) expression (Supplemental Figure 5, A and B). To confirm the TGOF of MT-JIP3, we used the non-allele-selective ASO3 to inhibit MT-JIP3 expression. This inhibition was dose dependent and significantly reduced MT-JIP3 expression in patient cells (Figure 6B). WT-JIP3 expression was also reduced in both control and patient cells, as expected due to the non-allele-selective nature of the ASO (Figure 6C). However, SPAG9 (JIP4) expression was not affected (Figure 6D), indicating the specificity of the ASO for MAPK8IP3 (JIP3).

Non-allele-selective PS-ASOs rescue patient fibroblasts from MT-JIP3–induced cell toxicity. Healthy individual–derived (control) fibroblasts from GM03440 and patient-derived fibroblasts (5,000 cells/well) were electroporated with different concentrations of non-allele-selective ASO3 targeting MAPK8IP3 (JIP3) and incubated in 96 wells with normal culture medium for different time points for different experiments. Scrabble PS-ASO 676630 (5 μM) was used as a control ASO. (A) After 72-hour incubation, cell proliferations were measured by MTT assay and normalized to UTC of control cells. n = 3, *P < 0.05, **P < 0.01. (B) After 48-hour incubation, RNA was extracted by GITC assay, and MT-MAPK8IP3 mRNA levels were measured using q-PCR. Data were normalized to UTC of control cells. n = 3, **P < 0.01. (C) WT MAPK8IP3 mRNA levels were measured using q-PCR. Data were normalized to UTC of control cells. n = 3, **P < 0.01. (D) SPAG9 (JIP4) mRNA levels were measured using q-PCR. Data were normalized to UTC of control cells. n = 3, **P < 0.01. (E) Cell proteins were harvested after 72-hour incubation of ASOs with different concentrations. JIP3, JIP4, p-JNK, and total JNK were measured by Western blot. GAPDH was used as a loading control. (F) 10 μM non-allele-selective ASO3 was used to treat mutations in patient iPSC neurons for 72 hours, and then 1 μM D1 agonist (SKF 81297) was added to each test group with different incubation time points. cAMP levels were measured by cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3, *P < 0.05. (G) After 48-hour ASO treatment, cell imaging of either patient fibroblasts or healthy control fibroblasts incubated with 1 μM Cy3-transferrin for different time points. Colocalization of Cy3-transferrin and Rab7 (late endosome [LE] marker) was observed with a Keyence BZ-X800 microscope. Scale bar: 20 μm.

When MT-JIP3 expression was inhibited, patient cell proliferation increased in a dose-dependent manner (Figure 6A), a result corroborated by applying ASO1 and ASO2 (Supplemental Figure 5C). This confirmed that MT-JIP3 contributes to reduced cell proliferation. Additionally, JNK signaling activation was attenuated by the ASO, further demonstrating the TGOF nature of MT-JIP3 (Figure 6E). We also observed that ASO treatment normalized endosomal movement; Cy3-transferrin in patient cells accumulated in the late endosomal area at 2 hours and colocalized with Rab7, similar to control cells (Figure 6G and Supplemental Figure 5D). Moreover, ASO3-treated patient-derived fibroblasts showed increased cAMP levels upon stimulation with the D1 agonist SKF (Figure 6F). These findings highlight the therapeutic potential of non-allele-selective ASOs in restoring normal cellular function in patient cells.

Suppression of MT-JIP3 by ASOs rescues patient iPSC–derived neurons from mutation-induced cytotoxicity. Next, to substantiate the therapeutic potential of ASOs, we chose to employ the same 3 ASOs for the rescue experiments. All ASOs restored cell viability of patient iPSC–derived neurons to the level of CRISPR control neurons (Figure 7A). Furthermore, all 3 ASOs could effectively reduce the expression of MT-JIP3 mRNA, but no significant JIP4 reduction was observed (Figure 7, C and D), demonstrating their specificity. Among them, ASO2 was the most efficacious in reducing the expression of MT-JIP3 mRNA over 80%. It is worth noting that these ASOs are non-allele selective, thus the expression of WT-JIP3 mRNA was also reduced (Figure 7B). Importantly, WT-JIP3 was also reduced, but this was tolerated by the cells. In addition, when the expression of MT-JIP3 was inhibited by ASOs, cell viability increased significantly (Figure 7A). These data will provide substantial information toward a therapeutic avenue and subsequent ASO drug discovery and development, as they show that a lower allele selectivity can be tolerated for this gene target.

Non-allele-selective PS-ASOs rescue patient iPSC–derived neurons from MT-JIP3–induced cell toxicity. CRISPR control and patient iPSC–derived neurons were treated with 3 different non-allele-selective ASOs, including ASO1 (1713794), ASO2 (1713687), and ASO3 (1713510) targeting MAPK8IP3 (JIP3) for 4 days. Scrabble PS-ASO 676630 was used as a control ASO. (A) Cell viability was measured by trypan blue cell count assay and normalized to UTC of CRISPR control cells. n = 3, *P < 0.05. (B) RNA was extracted by GITC assay, and MT-MAPK8IP3 mRNA level was measured using q-PCR. Data were normalized to UTC of control cells. n = 3, *P < 0.05. (C) WT-MAPK8IP3 mRNA level was measured using q-PCR. Data were normalized to UTC of control cells. n = 3, *P < 0.05. (D) SPAG9 (JIP4) mRNA levels were measured using q-PCR. Data were normalized to UTC of control cells. n = 3. (E) CRISPR control and patient iPSC–derived neurons were treated with different concentrations of non-allele-selective ASO2 targeting MAPK8IP3 (JIP3) for 4 days. Scrabble PS-ASO 676630 was used as a control ASO. Cell viability was measured by trypan blue cell count assay and normalized to UTC of CRISPR control cells. n = 3, *P < 0.05. (F) Cell proteins were harvested after 4-day incubation of ASO2. JIP3, JIP4, and p-JNK were measured by Western blot. GAPDH was used as a loading control. (G) Expression of the MT-JIP3 in patient iPSC–derived neurons leads to increased risk of death compared with CRISPR control iPSC-derived neurons. ASO2 (5 μM) was used to treat mutation. Cells were screened using a Keyence microscope every 3 or 4 days. Viable cell numbers were manually counted in 3 representative areas from each well (center, upper center, and lower center). Cumulative risk was calculated by Cox proportional hazard regression in Prism 9.5. ****P < 0.0001. (H) 5 μM non-allele-selective ASO2 was used to treat mutations in patient iPSC neurons for 72 hours, and then 1 μM D1 agonist (SKF 81297) was added to each test group with different incubation time points. cAMP levels were measured using a cAMP-Glo Assay (Promega). Data were normalized to UTC of cells. n = 3. *P < 0.05.

Next, we selected ASO2, which had the best efficacy, to conduct the dose-dependent experiment. The results showed that the ability of ASO2 to restore cell viability was dose dependent (Figure 7E). Furthermore, Western blot results showed that protein expression of JIP3 was significantly reduced upon cell treatment with ASO2, and the activation of JNK was also markedly weakened (Figure 7F). Cox proportional hazards analysis has been widely used to study the toxicological properties of a variety of substances associated with neurodegenerative diseases (32, 33); it generates a hazard ratio to represent relative and estimates risk of death for each group of iPSC-derived neurons. MT-JIP3 increased the hazard ratio in patient iPSC–derived neurons compared with WT-JIP3 in CRISPR control neurons, but ASO2 significantly reduced the MT-JIP3–induced hazard ratio (Figure 7G and Supplemental Figure 6). Additionally, ASO2 restored D1 coupling on patient iPSC–derived neurons in response to D1 agonist SKF stimulation (Figure 7H). These results collectively demonstrate that a non-allele-selective ASO rescues the patient iPSC–derived neurons from cytotoxicity.