Zfra Inhibits the TRAPPC6AΔ-Initiated Pathway of Neurodegeneration

When WWOX is downregulated in middle age, aggregation of a protein cascade, including TRAPPC6AΔ (TPC6AΔ), TIAF1, and SH3GLB2, may start to occur, and the event lasts more than 30 years, which results in amyloid precursor protein (APP) degradation, amyloid beta (Aβ) generation, and neurodegeneration, as shown in Alzheimer’s disease (AD). Here, by treating neuroblastoma SK-N-SH cells with neurotoxin MPP+, upregulation and aggregation of TPC6AΔ, along with aggregation of TIAF1, SH3GLB2, Aβ, and tau, occurred. MPP+ is an inducer of Parkinson’s disease (PD), suggesting that TPC6AΔ is a common initiator for AD and PD pathogenesis. Zfra, a 31-amino-acid zinc finger-like WWOX-binding protein, is known to restore memory deficits in 9-month-old triple-transgenic (3xTg) mice by blocking the aggregation of TPC6AΔ, SH3GLB2, tau, and amyloid β, as well as inflammatory NF-κB activation. The Zfra4-10 peptide exerted a strong potency in preventing memory loss during the aging of 3-month-old 3xTg mice up to 9 months, as determined by a novel object recognition task (ORT) and Morris water maize analysis. Compared to age-matched wild type mice, 11-month-old Wwox heterozygous mice exhibited memory loss, and this correlates with pT12-WWOX aggregation in the cortex. Together, aggregation of pT12-WWOX may link to TPC6AΔ aggregation for AD progression, with TPC6AΔ aggregation being a common initiator for AD and PD progression.


Introduction
Neurodegeneration and related dementia normally occur in humans after 65 years of age [1][2][3][4][5]. Many neurodegenerative diseases such as Parkinson's disease (PD), Alzheimer's disease (AD), and Huntington's disease possess neuronal protein aggregates or inclusion bodies as one of the pathological hallmarks of the diseases [1][2][3][4][5]. Indeed, the proteins may start to aggregate and accumulate in the brain of middle-aged normal humans beginning at 40 years old [1][2][3][4][5]. AD is a chronic neurodegenerative disease that usually starts slowly from an asymptomatic stage and becomes worse as one grows older. The most common early symptom is difficulty in remembering recent events (short-term memory loss) [1-5].

Protein Aggregation in Knockout Wwox −/− Cells and Transforming Growth Factor Beta (TGF-β) Increased the Aggregation
We used our established Wwox knockout mice [38,44], as well as wild-type Wwox +/+ , knockout Wwox −/− , and heterozygous Wwox +/− mouse embryonic fibroblasts (MEF) [21,22]. The knockout MEF cells exhibited constitutive phosphorylation of Smad3 of the SMAD pathway and p38 in the MAPK pathway [31]. Compared to the wild-type MEF cells, the Wwox knockout cells exhibited intracellularly elevated levels of aggregated TIAF1, whereas the levels of wild-type TPC6A and JNK1 remained relatively unchanged ( Figure 1A). When these knockout cells were treated with TGF-β for 1 h, rapid increases in the accumulation of TPC6A, TIAF1, and JNK1 were observed ( Figure 1A). The levels of neurofibrillary tangles (NFT) were also elevated in the Wwox knockout MEF cells, and tumor necrosis factor alpha (TNF-α) did not increase the NFT expression ( Figure 1B). Wild-type TPC6A underwent polymerization in both the cytoplasm and nucleus of the Wwox knockout MEF cells, and TGF-β2 increased the polymerization during treatment for 1 h ( Figure 1C). 1A). When these knockout cells were treated with TGF-β for 1 h, rapid increases in the accumulation of TPC6A, TIAF1, and JNK1 were observed ( Figure 1A). The levels of neurofibrillary tangles (NFT) were also elevated in the Wwox knockout MEF cells, and tumor necrosis factor alpha (TNF-α) did not increase the NFT expression ( Figure 1B). Wild-type TPC6A underwent polymerization in both the cytoplasm and nucleus of the Wwox knockout MEF cells, and TGF-β2 increased the polymerization during treatment for 1 h ( Figure  1C).

Neurotoxin MPP+ Upregulated TPC6AΔ for Aggregation in Neuroblastoma Cells
Downregulation of WWOX in middle age leads to aggregation of a cascade of proteins, and TPC6AΔ is the first protein to become polymerized or aggregated, which may lead to AD at old age of greater than 70 to 110 [43,44,46]. Under physiological conditions, The expression of endogenous TPC6A, TIAF1, and JNK1 was determined in the wild-type and knockout Wwox MEF cells (with gel run under reducing SDS-PAGE and then Western blotting). These cells were treated with TGF-β1 (5 ng/mL) for 1 h. (B) The presence of NFT was shown in the heterozygous and knockout Wwox MEF cells. TNF-α (100 ng/mL) did not increase NFT expression in both cells during treatment for 1 h. (C) TGF-β1 (5 ng/mL) increased polymerization of wild-type TPC6A in the knockout Wwox MEF cells during treatment for 1 h at 37 • C.

Neurotoxin MPP+ Upregulated TPC6A∆ for Aggregation in Neuroblastoma Cells
Downregulation of WWOX in middle age leads to aggregation of a cascade of proteins, and TPC6A∆ is the first protein to become polymerized or aggregated, which may lead to AD at old age of greater than 70 to 110 [43,44,46]. Under physiological conditions, WWOX binds TPC6A∆ via its C-terminal D3 tail, so as to prevent TPC6A∆ selfaggregation [44,46,47]. When WWOX becomes Tyr33 phosphorylated, the self-folded WWOX, due to the binding of intramolecular first WW domain with SDR domain, opens up, so that TPC6A∆ interacts with the fist WW domain [21]. This further prevents TPC6A∆ aggregation. Nonetheless, the higher the MPP+ concentration, the more TPC6A∆ becomes aggregated.
Neuroblastoma SK-N-SH cells were treated with various doses of MPP+ in vitro for 24 h at 37 • C. By non-reducing SDS-PAGE and then Western blot analysis, protein dimerization and further polymerization were observed. The SH3GLB2 dimer showed a dose-dependent increment (Figure 2A). Similarly, TIAF1 polymerization increased upon MPP+ treatment from 125 to 250 µM ( Figure 2B). TPC6A∆ showed polymerization at 62.5 µM MPP+ treatment, followed by gradual increase in polymerization at 125 and 250 µM MPP+ treatment ( Figure 2C). TPC6A∆ phosphorylated at Ser35 upon treatment with 125 and 250 µM MPP+ ( Figure 2D). The level of α-tubulin as a loading control is shown ( Figure 2E). In addition, both time-course and dose-response experiments were carried out ( Figures S1 and S2). MPP+ had no apparent effect on APP expression ( Figures S1 and S2). However, MPP+ significantly suppressed the expression of pY33-WWOX, and yet significantly increased the expression of Aβ ( Figures S1 and S2), Overall, our data revealed that TPC6A∆ is most responsive to polymerization caused by MPP+.
due to the binding of intramolecular first WW domain with SDR domain, opens up, so that TPC6AΔ interacts with the fist WW domain [21]. This further prevents TPC6AΔ aggregation. Nonetheless, the higher the MPP+ concentration, the more TPC6AΔ becomes aggregated.
Neuroblastoma SK-N-SH cells were treated with various doses of MPP+ in vitro for 24 h at 37 °C. By non-reducing SDS-PAGE and then Western blot analysis, protein dimerization and further polymerization were observed. The SH3GLB2 dimer showed a dosedependent increment (Figure 2A). Similarly, TIAF1 polymerization increased upon MPP+ treatment from 125 to 250 µM (Figures 2B). TPC6A∆ showed polymerization at 62.5 µM MPP+ treatment, followed by gradual increase in polymerization at 125 and 250 µ M MPP+ treatment ( Figure 2C). TPC6A∆ phosphorylated at Ser35 upon treatment with 125 and 250 µ M MPP+ ( Figure 2D). The level of α-tubulin as a loading control is shown ( Figure 2E). In addition, both time-course and dose-response experiments were carried out (Figures S1 and S2). MPP+ had no apparent effect on APP expression ( Figures S1 and S2). However, MPP+ significantly suppressed the expression of pY33-WWOX, and yet significantly increased the expression of Aβ ( Figures S1 and S2), Overall, our data revealed that TPC6A∆ is most responsive to polymerization caused by MPP+.   Next, we examined the effect of the Zfra4-10 peptide in restoring memory loss in aging 3xTg mice. Ten-month-old triple-transgenic mice (3xTg) for Alzheimer's disease (AD), possessing mutant genes Psenl (M146V), APPSwe, and Tau (P301L), were used [5]. 3xTg mice received tail vein injections of PBS (sham control) or Zfra4-10 peptide (2 mM in 100 µL PBS) once per week for 4 consecutive weeks, followed by resting for another week. In agreement with our previous report [5], Zfra4-10-treated mice exhibited significant improvement in the novel object recognition task (ORT) ( Figure 3A), which is supported by the hippocampus-dependent nonspatial learning and memory.

Zfra Restored the Learning and Memory Capabilities in 10-Month-Old 3xTg AD Mice
Next, we examined the effect of the Zfra4-10 peptide in restoring memory loss in aging 3xTg mice. Ten-month-old triple-transgenic mice (3xTg) for Alzheimer's disease (AD), possessing mutant genes Psenl (M146V), APPSwe, and Tau (P301L), were used [5]. 3xTg mice received tail vein injections of PBS (sham control) or Zfra4-10 peptide (2 mM in 100 µL PBS) once per week for 4 consecutive weeks, followed by resting for another week. In agreement with our previous report [5], Zfra4-10-treated mice exhibited significant improvement in the novel object recognition task (ORT) ( Figure 3A), which is supported by the hippocampus-dependent nonspatial learning and memory. 3xTg mice, aged 10 months old, received Zfra4-10 tail vein injections, followed by examination of their learning and memory capabilities by a novel object recognition task (A) and a Morris water maze (B) at 1 month after the injections. The percentages of novel object exploring time (time spent on novel object/time spent on both objects) of acquisition, short-term (2 h) delay, and long-term (24 h) delay are shown (control group: sham, n = 10; Zfra injection group: Zfra4-10, n = 10). In the Morris water maze, the learning curve represents the latency of finding the hidden platform during the training session. The probe test was quantified as time spent in the target quadrant (Bonferroni's post hoc test and ANOVA: * 0.01< p < 0.05, ** 0.001< p < 0.01, *** p < 0.001, experimental group vs. respective control group. Sham: n = 10; control: n = 10). 3xTg mice, aged 10 months old, received Zfra4-10 tail vein injections, followed by examination of their learning and memory capabilities by a novel object recognition task (A) and a Morris water maze (B) at 1 month after the injections. The percentages of novel object exploring time (time spent on novel object/time spent on both objects) of acquisition, short-term (2 h) delay, and long-term (24 h) delay are shown (control group: sham, n = 10; Zfra injection group: Zfra4-10, n = 10). In the Morris water maze, the learning curve represents the latency of finding the hidden platform during the training session. The probe test was quantified as time spent in the target quadrant (Bonferroni's post hoc test and ANOVA: * 0.01< p < 0.05, ** 0.001< p < 0.01, *** p < 0.001, experimental group vs. respective control group. Sham: n = 10; control: n = 10).
The abilities of hippocampus-dependent nonspatial memory are expressed as the percentages of novel object exploring time. That is, mice spent their time exploring both objects (time spent on novel object/time spent on both objects), defined as acquisition. During the 5 min acquisition phase, both mouse groups spent equal exploring times on each object ( Figure 3A), suggesting that the task environment did not provide preferences for the mice.
In the short-term memory task (2 h after the acquisition phase), Zfra-treated mice exhibited significant improvement in expanding the exploring times of a replaced new or novel object ( Figure 3A). Similarly, in the long-term memory task (24 h after the acquisition phase), the novel object exploring times for the Zfra-treated group were significantly higher than the sham control group ( Figure 3A). These observations suggest that the hippocampusdependent nonspatial memory of 3xTg mice was significantly improved by Zfra treatment.
Two days after the novel object recognition test, all mice were subjected to the Morris water maze for determining the hippocampus-dependent spatial learning and memory. Both control and Zfra groups showed a time-dependent decrease in escape latency ( Figure 4B). However, Zfra-treated mice had a significantly reduced time in escaping from the water and spent more time in the target quadrant ( Figure 3B; p < 0.05, ANOVA). That is, in the probe test with the platform removed ( Figure 3B), the Zfra mice spent more time in the targeted quadrant than the PBS controls (Bonferroni's post hoc test and ANOVA: ** p < 0.01, *** p <0.001; Zfra vs. control experiments; sham: n = 10; control: n = 10). Together, in agreement with our previous observations [5], Zfra was potent in restoring memory loss in aging 3xTg mice.
In the short-term memory task (2 h after the acquisition phase), Zfra-treated mice exhibited significant improvement in expanding the exploring times of a replaced new or novel object ( Figure 3A). Similarly, in the long-term memory task (24 h after the acquisition phase), the novel object exploring times for the Zfra-treated group were significantly higher than the sham control group ( Figure 3A). These observations suggest that the hippocampus-dependent nonspatial memory of 3xTg mice was significantly improved by Zfra treatment.
Two days after the novel object recognition test, all mice were subjected to the Morris water maze for determining the hippocampus-dependent spatial learning and memory. Both control and Zfra groups showed a time-dependent decrease in escape latency ( Figure  4B). However, Zfra-treated mice had a significantly reduced time in escaping from the water and spent more time in the target quadrant ( Figure 3B; p < 0.05, ANOVA). That is, in the probe test with the platform removed ( Figure 3B), the Zfra mice spent more time in the targeted quadrant than the PBS controls (Bonferroni's post hoc test and ANOVA: ** p < 0.01, *** p <0.001; Zfra vs. control experiments; sham: n = 10; control: n = 10). Together, in agreement with our previous observations [5], Zfra was potent in restoring memory loss in aging 3xTg mice.  Three-month-old 3xTg mice received Zfra4-10 peptide via tail vein injections once per week for 3 consecutive weeks, followed by one-week rest, and then they were subjected to novel ORT (A) and Morris water maze analysis (B). The animal behavior experiments were repeated when mice reached 6 and 9 months old, respectively. The percentages of novel object exploring time of acquisition, short-term (2 h) delay, and long-term (24 h) delay are shown (n = 5 for each group). The learning curve represents the escape latency to find the hidden platform during a training session. The probe test was quantified as time spent in the target quadrant (Bonferroni's post hoc test and ANOVA: * p < 0.05, ** p < 0.01, *** p < 0.001 vs. the respective control group; n = 5).

Zfra4-10 Potently Prevented Age-Dependent Memory Loss in 3xTg Mice
Next, we investigated whether Zfra4-10 rescued aging-induced neurodegeneration in 3xTg mice. To start with, 3-month-old 3xTg mice received Zfra4-10 peptides via tail vein injections once per week for 4 consecutive weeks. After resting for one week, mice were subjected to examination of their capabilities in memory and learning, as determined by novel ORT. The experiment was repeated every 3 months until the mice reached 10 months old. Our data showed that in the acquisition phase, both control and Zfra mouse groups spent approximately equal exploring times on each object at month 3, 6, and 9 of age ( Figure 4A). That is, mice had no specific preferences in the task environment.
In the short-term memory task, the control mice tended to have age-related decline in novel ORT ( Figure 5A). However, Zfra-treated mice did not show memory decline in novel ORT ( Figure 4A, p = 0.06 ANOVA). Moreover, in the long-term memory task, control groups showed increased decline in novel ORT, whereas the Zfra group had increased memory in handling novel ORT ( Figure 4A).

Zfra4-10 Potently Prevented Age-Dependent Memory Loss in 3xTg Mice
Next, we investigated whether Zfra4-10 rescued aging-induced neurodegeneration in 3xTg mice. To start with, 3-month-old 3xTg mice received Zfra4-10 peptides via tail vein injections once per week for 4 consecutive weeks. After resting for one week, mice were subjected to examination of their capabilities in memory and learning, as determined by novel ORT. The experiment was repeated every 3 months until the mice reached 10 months old. Our data showed that in the acquisition phase, both control and Zfra mouse groups spent approximately equal exploring times on each object at month 3, 6, and 9 of age ( Figure 4A). That is, mice had no specific preferences in the task environment.
In the short-term memory task, the control mice tended to have age-related decline in novel ORT ( Figure 5A). However, Zfra-treated mice did not show memory decline in novel ORT ( Figure 4A, p = 0.06 ANOVA). Moreover, in the long-term memory task, control groups showed increased decline in novel ORT, whereas the Zfra group had increased memory in handling novel ORT ( Figure 4A).  In the Morris water maze task, the spatial learning and memory in a hippocampusdependent manner was investigated. Compared to the control group, the Zfra-treated mouse group showed significant restoration of memory loss, as reflected by a significant decrease in escape latency and significantly increased time spent in the target quadrant ( Figure 4B).

WWOX Deficiency Led to Protein Aggregation and Memory Loss in Mice
Previously, we reported that Wwox heterozygous mice exhibited an age-related faster decline in both short-and long-term memories than those in 3xTg mice, as determined by novel object recognition tests [5]. To further validate the role of WWOX in memory maintenance, 10-month-old wild-type and heterozygous Wwox mice were subjected to behavioral tests. In the novel object recognition task, the non-spatial learning and memory were significantly better in the wild-type than the heterozygous Wwox mice ( Figure 5A). Again, there was no preference on the acquisition of the designated objects. In both short-and long-term memory analyses, wild-type mice performed significantly better than heterozygous Wwox mice, suggesting that WWOX plays a critical role in the maintenance of non-spatial and spatial memory ( Figure 5A,B), that is, lacking one allele of Wwox gene in the gnome leads to memory deficiency. In parallel with the behavioral analyses, the 10-monthold wild-type mice had significantly reduced amounts of TPC6A∆ in the hippocampus compared to the heterozygous Wwox mice, as determined by immunohistochemistry ( Figure 5C). We reported the presence of TPC6A∆ and TIAF1 protein aggregates in the brain of middle-aged normal individuals due, in part, to downregulation of WWOX [5,42,44,46]. The protein aggregates continue to stay and accumulate in the human brain without apparent degradation from middle age. These may account for the problematic neurodegeneration when one reaches ages greater than 75 years old. Additionally, the presence of both TPC6A∆ and TIAF1 aggregates has been identified in 3-week-old Wwox knockout mice [5], indicating that lack of WWOX rapidly causes protein aggregation in the brain.
Isoform TPC6A∆ protein aggregated significantly greater in the hippocampus and cortex of 11-month-old heterozygous Wwox mice than wild-type mice ( Figure 5C). Shown in Figure 6 is the indicated protein aggregation in 11-month-old heterozygous Wwox mice and wild-type mice. Significant increases in protein aggregates such as pS35-TPC6A∆ ( Figure 6A) and SH3GLB2 ( Figure 6C) were observed in the cortex of 11-month-old heterozygous mice. The digitally enlarged Figure 6 are shown in Figure S3. Aggregation of pS37-TIAF1, TIAF1, wild-type TPC6A, and pT181-Tau were not significantly increased ( Figure 6B-F). The failure in significant increases was probably due to the presence of a functional Wwox allele in the heterozygous mice, which retards protein aggregation. Wild-type TPC6A expression was expressed intracellularly in the neurons of the cortex of heterozygous Wwox mice and wild-type mice ( Figure 6B).

Identification of pT12-WWOX as Protein Aggregates in the Cortex of Old Heterozygous Wwox Mice
Increased phosphorylation of WWOX at Y33 and Y287 was observed in the cortex of heterozygous Wwox mice, as reflected by increased immunogenicity for these proteins ( Figure 7A,B). pY287-WWOX appeared to be located in the nuclei, and pY33-WWOX was released into the extracellular matrix. Aggregation for both proteins was poor. Y287 phosphorylation is needed for WWOX to undergo ubiquitination and degradation [20], whereas Y33-phosphorylated WWOX maintains mitochondrial functional and induces apoptosis when overexpressed [9,12,15].
In stark contrast, pT12-WWOX aggregates or plaques were shown in the cortex of 11-month-old heterozygous Wwox mice, but not in the wild-type mice ( Figure 7C). The data suggest that increased phosphorylation at T12 is sufficient to cause WWOX to undergo aggregation. Under similar conditions, the expression levels of pS14-WWOX were low in both wild-type and heterozygous mice ( Figure 7D). Without using a specific primary antibody, negative control staining was shown ( Figure 7E). The quality of our newly generated pT12-WWOX antibody, as well as pS14-WWOX antibody [19], is shown ( Figure  S5). The immunoblot was from immunodeficient NOD-SCID mouse brains. The levels of pS14-WWOX were lower than that of pT12-WWOX.  , and pT181-Tau (F) was examined in the brain cortex of 11-month-old heterozygous Wwox mice and age-matched wild-type mice. The extent of protein aggregation was examined (n = 5 to 10; Student's t-test). Negative controls without staining with primary antibodies are shown. The digitally enlarged pictures are shown in Figure S3.
heterozygous Wwox mice, as reflected by increased immunogenicity for these proteins ( Figure 7A,B). pY287-WWOX appeared to be located in the nuclei, and pY33-WWOX was released into the extracellular matrix. Aggregation for both proteins was poor. Y287 phosphorylation is needed for WWOX to undergo ubiquitination and degradation [20], whereas Y33-phosphorylated WWOX maintains mitochondrial functional and induces apoptosis when overexpressed [9,12,15].

Discussion
In this study, we determined that Zfra4-10 peptide is potent in preventing the progression of AD-like symptoms in 3xTg mice. We previously reported that Zfra4-10 peptide restores the memory deficit and mitigates the AD-like symptoms in 3xTg mice [5]. The un-derlying mechanism is that Zfra blocks the aggregation of pS35-TPC6A∆, TIAF1, SH3GLB2, tau, and amyloid β and inhibits NF-κB-activation-mediated inflammation [5]. Zfra accelerates protein degradation with a ubiquitin/proteasome-independent mechanism [5]. Zfra4-10 peptide can be used for both therapy and prevention of age-related AD. Indeed, a clinical trial for Zfra in has been planned in the near future. This will cover both AD and cancer.
TPC6A∆ is the first one among the known three initiators in the protein aggregation cascade [44][45][46][47][48][49][50]. When intracellular WWOX is low or becomes dysfunctional, aggregation of TPC6A∆ starts to occur, followed by polymerization of TIAF1 and SH3GLB2 [44][45][46][47][48][49][50]. The aggregation complex repeats the polymerization or aggregation process and thereby allows accumulation of more protein aggregates in the brain, so as to cause caspase activation, cytochrome c release, and APP degradation for generation of Aβ [44,46,47]. The protein aggregation cascade starts in middle age and may last more than 30 years. By treating neuroblastoma SK-N-SH cells with neurotoxin MPP+, upregulation and aggregation of TPC6A∆, along with aggregation of TIAF1, SH3GLB2, Aβ, and tau occurred. MPP+ is an inducer of PD, suggesting TPC6A∆ is a common initiator for AD and PD pathogenesis.
Zfra, a 31-amino-acid zinc finger-like WWOX-binding protein, restores memory deficits in 9-month-old 3xTg mice by blocking the aggregation of TPC6A∆, SH3GLB2, tau, and amyloid β; accelerating intracellular protein degradation; and inhibiting inflammatory NF-κB activation [5]. Whether Zfra-mediated activation of Hyal-2/WWOX/Smad4 signaling is needed for retardation of AD and PD progression remains to be established. Presumably, the signaling may enhance intracellular protein degradation and thereby facilitate neuronal survival. The Hyal-2/WWOX/Smad4 signaling is also crucial for the activation of Hyal-2+ Z cells [52]. Whether activated Z cells block neurodegeneration via secreted cytokines is unknown and remains to be established. Activated Z cells kill cancer cells via unknown cytokines or through direct contact with cancer cells [53,54].
Zfra binds WWOX to the Nand C-terminal domains of WWOX, and this may lead to acceleration of the degradation of WWOX [5,47]. Zfra4-10 or WWOX7-21 peptide alone blocks cancer growth in mice [52]. In contrast, when both peptides are combined in treating cancer, cancer growth cannot be blocked in vivo. Functional nullification is probably due to covalent binding between Zfra4-10 and WWOX7-21. Notably, the stronger the binding of WWOX with intracellular protein partners, the better the cancer suppression [53]. By the same token, neurodegeneration is likely to be retarded when WWOX has strong physical binding interactions with protein partners in the neurons. This likely scenario is very promising and worth pursuing for further investigation.
A recent report showed that high glucose induces the activation of WWOX for causing mitochondrial apoptosis [29]. Zfra blocks the effect of high glucose due to its interaction with WWOX [29]. Zfra can be of therapeutic efficacy in treating hyperglycemia/diabetesassociated neurodegeneration [29]. This finding is important and yet not surprising for Zfra. We believe that Zfra can be a universal drug for targeting many diseases via Hyal-2/WWOX/Smad4 signaling.
The secondary WW domain (WW2) exhibits three antiparallel β-sheets in the solution [55]. However, the tertiary structure of the first WW domain (WW1) is unknown. It is not clear whether these β-sheets are able to stack up together to generate protein aggregates. We determined that WW domains or SDR domain undergo domain self-binding or inter-domain binding [21]. Whether inter-WW domain binding is Y33 phosphorylation dependent remains to be established [21].
Our data from immunohistochemistry showed that there was no apparent aggregate formation for pS14-, pY33-, and pY287-WWOX. The presence of pT12-WWOX aggregates in the cortex of heterozygous Wwox mice, but not in wild-type mice, suggests that reduced amounts of WWOX in the cytoplasm leads to T12 phosphorylation and increased WWOX protein instability. While pS14-WWOX supports the pathogenic progression toward AD severity and cancer growth and metastasis [5,52,55], the relationship between pT12 and pS14 in WWOX needs further elucidation, so as to increase our understanding of protein aggregation and the downstream TPC6A∆ in the protein aggregation cascade. That is, whether pT12-WWOX aggregates induce TPC6A∆ polymerization remains to be established. Another intriguing issue is that pY33-WWOX is apparently released from the nuclei of neurons to the extracellular matrix in the cortex. pY33-WWOX is critical in maintaining cellular functions and stabilizing the normal physiology of mitochondria with p53 [9]. If this holds true, the way in which accumulation of pY33-WWOX in the extracellular space occurs remains to be established.
Under physiological conditions, WWOX binds TPC6A∆ via its C-terminal D3 tail, so as to prevent self-aggregation of TPC6A∆ [46]. When WWOX becomes Tyr33 phosphorylated, the self-folded WWOX from the binding of intramolecular WW domains with SDR domain opens up, so that TPC6A∆ interacts with the WW domains. This further prevents TPC6A∆ aggregation. Nonetheless, the higher the MPP+ concentration, the more TPC6A∆ becomes aggregated.
An additional mechanism is that the Zfra peptide regulates the expression of WWOX. WWOX is a potent inhibitor for GSK3β, due to its binding of GSK3β via the SDR domain. Importantly, WWOX inhibits GSK3β-stimulated S396 and S404 phosphorylation within the microtubule-binding domains of Tau, indicating that WWOX limits GSK3β activity in cells [60,62].

Synthetic Peptides
The Zfra4-10 (RRSSSCK) peptide was synthesized by Genemed Synthesis (San Antonio, TX, USA). The peptide was prepared at 10 mM in degassed sterile Milli-Q water as stocks. Each tube was flushed with nitrogen and stored in a −80 • C freezer. For tail vein injections, the peptide stock solution was freshly diluted to 1-4 mM in 100 µL degassed phosphate-buffered saline (PBS).

Animals
We followed the National Institutes of Health Guidelines for animal research (Guide for the Care and Use of Laboratory Animal) and the approved protocol by the National Cheng Kung University Institutional Animal Care and Use Committee. The 3xTg mice (B6; 129-Psen1tm1Mpm Tg[APPSwe, tauP301L]1Lfa/Mmjax) were from the Jackson Laboratory (Bar Harbor, ME, USA) [5,63,64]. The mice were housed in automatically conditioned rooms (temperature 23 ± 1 • C, humidity 55 ± 5%, 12 h light/12 h dark cycle, light cycle beginning at 06:00), which are located in the Laboratory Animal Center of National Cheng Kung University (Tainan, Taiwan). These mice had unrestricted access to food and water [5]. Genotyping was carried out for all mice using a protocol provided by the Jackson Laboratory.
For Zfra-mediated restoration of memory loss and learning deficit, five 10-month-old 3xTg mice (body weight: 31~35 g) received four consecutive injections of Zfra4-10 solution (2 mM in PBS, 100 µL each injection) via the tail veins on a weekly basis. Under similar conditions, the sham mouse group received PBS injections only. One week later, behavioral tests were carried out, and then mice were sacrificed by CO 2 anesthesia (75% CO 2 /25% O 2 ) and perfused with 10 mL PBS (to reduce background in tissue section staining with specific antibodies). Brains were then rapidly harvested.
To investigate whether Zfra prevented age-related neurodegeneration, 3-month-old 3xTg mice received Zfra4-10 peptide or PBS via tail vein injections once per week for 3 consecutive weeks. One week later, mice were subjected to behavioral tests. Zfra treatment was repeated twice when mice reached 6 and 9 months old. Again, behavioral tests were then carried out.

Novel Object Recognition Test
Normally, the experiment was carried out from 6 PM each day as previously described [5,63]. After Zfra or PBS treatment, mice were allowed to rest for a week. Mice were initially habituated to a polycarbonate box (47.5 cm × 25.8 cm × 21 cm) by staying in there for 10 min per day for 3 consecutive days. Twenty-four hours later, each mouse faced two identical objects (glass bottle, 4 cm diameter and 6.5 cm high) positioned 7 cm apart away from a wall. The durations that the mouse explored each object were recorded over 5 min. For the short-term memory (STM) test, the mouse was placed into the box 2 h later, facing an old and a new object (white Lego bricks, 3 × 3 × 6 cm). For the long-term memory (LTM) test, the mouse was allowed in the box 24 h later to face an old and a new object (white plastic bottle, 3.5 cm diameter and 7.2 cm high). The times spent in exploring each object during the 5 min period were recorded in either the STM or the LTM test. All of the objects were cleaned by 70% alcohol between trials to reduce olfactory cues.

Morris Water Maze Test
The Morris water maze test was intended to measure the hippocampus-dependent spatial learning and memory, which was carried out 1 day after the novel object recognition test [5]. The Morris water maze was performed in a custom-made circular pool with a diameter of 120 cm and a wall height of 31 cm. The pool was filled with clear tap water at a temperature of 24 ± 1 • C and depth of 25 cm. The 4-day training session started at 6 pm. Each session consisted of four swim trials (maximum 120 s per trial) with different quadrant starting positions for each trial. On the morning of the fifth day, mice were subjected to a probe test. The mouse was placed in the pool in the southwest position, the longest distance away from the previous platform position (northeast). The mouse was allowed to swim for 60 s without a platform present. A CCD camera was used to record the whole process, and the escape latency (i.e., time to reach the platform, in seconds), path length, and swim speed (cm/s) were analyzed by a EthoVision video tracking system (Noldus Information Technology, Wageningen, The Netherlands).

Brain Sections and Immunohistochemistry
After the experiments, mice were sacrificed. Brains were harvested and fixed with 4% buffered paraformaldehyde. Paraffin-embedded blocks were made. Tissue sectioning and immunohistochemistry were carried out as described previously [5,7,12]. Frontal sections (6 µm in thickness) were made and then de-paraffinized by xylene, followed by stepwise hydration using decreasing concentrations of ethanol. Antigen retrieval was performed using Tris-EDTA (pH 8.0) with 0.5% Tween 20 by boiling for 15 min using a microwave. The tissue section was blocked with 1% H 2 O 2 in Tris-buffered saline (TBS) for 15 min to remove the peroxidase background, followed by washing 3 times for 5 min each using TBS with 0.5% Tween 20. Sections were then blocked with 0.5% Tween 20 and 5% BSA at room temperature (RT) for at least 1 h. Immunohistochemistry was carried out using an assay kit (HRP-conjugated secondary antibody (DAKO)). Specific primary antibodies are listed in the aforementioned Section 4.3. Microscopy was carried out, and 10 pictures per microscopic field were taken (OLYMPUS BX51 Microscope). Images were digitally analyzed and quantified using ImageJ software.

Statistical Analysi
Where indicated, all data (mean ± standard error) were analyzed by a two-tailed Mann-Whitney test (e.g., a novel object recognition test and a probe test of the Morris water maze) and a two-tailed unpaired t-test (protein expression in immunohistochemistry) [5]. Bonferroni post hoc tests were performed if significant (p < 0.05) main effects were found [5].

Conclusions
The first critical discovery in this study is that the Zfra4-10 peptide effectively retards age-dependent neurodegeneration in 3xTg mice, as determined by an novel object recognition task and Morris water maize analysis, that is, prevention of neurodegeneration at old age can be achieved by receiving a Zfa4-10 peptide at a young age. Zfra suppresses the aggregation of TPC6A∆, TIAF1, SH3GLB2, and Aβ and significantly downregulates the aggregation of pT181-Tau [5]. Zfra also enhances the degradation of intracellular proteins. Zfra covalently crosslinks with target proteins for proteolytic degradations-the so-called zfration-and thereby prevents the Aβ generation and neurodegeneration shown in AD [5].
The second important finding is that as an inducer of PD, neurotoxin MPP+ rapidly increases WWOX phosphorylation at Tyr33 [29] and upregulates the expression of TPC6A∆, followed by TIAF1, SH3GLB2, Aβ, and tau in neuroblastoma SK-N-SH cells. Upon MPP+ induction, TPC6A∆ is the first protein to polymerize and lead others to aggregation, suggesting that TPC6A∆ is a common initiator for AD and PD pathogenesis. Finally, pT12-WWOX becomes aggregated in the cortex of 11-month-old Wwox heterozygous mice, compared to age-matched wild-type mice. Whether aggregation of pT12-WWOX links to TPC6A∆ polymerization is unknown.

Patents
Nan-Shan Chang and Yu-Min Kuo. Method for treating Alzheimer's disease and a method for downregulating protein aggregation in the brain. U.S. patent #10344065. 6 March 2018.