Concurrent suppression of Aβ aggregation and NLRP3 inflammasome activation for treating Alzheimer's disease

Alzheimer's disease (AD) is a neurodegenerative illness accompanied by severe memory loss, cognitive disorders and impaired behavioral ability. Amyloid β-peptide (Aβ) aggregation and nucleotide-binding oligomerization domain (NOD)-like receptor protein 3 (NLRP3) inflammasome play crucial roles in the pathogenesis of AD. Aβ plaques not only induce oxidative stress and impair neurons, but also activate the NLRP3 inflammasome, which releases inflammatory cytokine IL-1β to trigger neuroinflammation. A bifunctional molecule, 2-[2-(benzo[d]thiazol-2-yl)phenylamino]benzoic acid (BPBA), with both Aβ-targeting and inflammasome-inhibiting capabilities was designed and synthesized. BPBA inhibited self- and Cu2+- or Zn2+-induced Aβ aggregation, disaggregated the already formed Aβ aggregates, and reduced the neurotoxicity of Aβ aggregates; it also inhibited the activation of the NLRP3 inflammasome and reduced the release of IL-1β in vitro and vivo. Moreover, BPBA decreased the production of reactive oxygen species (ROS) and alleviated Aβ-induced paralysis in transgenic C. elegans with the human Aβ42 gene. BPBA exerts an anti-AD effect mainly through dissolving Aβ aggregates and inhibiting NLRP3 inflammasome activation synergistically.


Introduction
Alzheimer's disease (AD) is a common form of dementia characterized by the accumulation of extracellular amyloid bpeptide (Ab) plaques, neuroinammation and neuronal cell death in the brain. 1 About 50 million people are living with AD globally in 2019, which put an enormous economic and mental burden on the society and families. 2 Although great effort has been made, the pathogenesis and pathogenic factors of AD are not yet fully elucidated. 3 Existing anti-AD drugs merely delay the symptoms to some extent but cannot stop the progression of the disease and have various side effects. 4 The Ab cascade hypothesis is the most prevalent supposition about the pathogenesis of AD, which suggests that Ab deposits play a vital role in initiating the disease. 5,6 In the past few decades numerous studies focused on cellular Ab deposits as a pathological hallmark and target of therapeutic drugs. 7 The newly FDA approved aducanumab is the rst anti-AD drug based on the Ab cascade hypothesis, though its efficacy is inconclusive. 8 Recently, increasing evidence supported that innate immunity-mediated neuroinammation plays a crucial role in the pathogenesis and progression of AD. 9 Inammasome plays an important role in neuroinammation and neurodegenerative diseases. 10,11 Particularly, the microglia-specic nucleotidebinding oligomerization domain (NOD)-like receptor protein 3 (NLRP3) inammasome mediates the pathogenesis of AD. 12,13 The NLRP3 inammasome is an intracellular multimeric protein complex composed of the receptor protein NLRP3, the effector protein cysteine protease-1 (caspase-1), and the adaptor protein called apoptosis-associated speck-like protein containing a caspase activation and recruitment domain (ASC). 14 It affects a variety of physiological functions, including the innate immune process and caspase-1-dependent response. 15,16 Ab plaques activate the NLRP3 inammasome, which releases inammatory factors such as cytokines interleukin-1b (IL-1b), interleukin-18 (IL- 18), and ASC. 17 The released inammatory cytokines and ASC trigger chronic neuroinammation and lead to cognitive impairment. 12,18 In reverse, neuroinammation aggravates the formation of Ab bers and plaques, and worse, boosts tau phosphorylation, thus leading to their aggregation to promote the pathogenesis of AD. 16 The activation of the NLRP3 inammasome/caspase-1 axis contributes much to AD in vivo, 10 while the deciency of NLRP3 or caspase-1 markedly reduces the Ab burden and cognitive impairment in amyloid precursor protein/presenilin-1 (APP/PS1) mice. 19 The elevation of IL-1b in the brain has been associated with the progression and onset of AD. 20,21 The inhibition of IL-1b could signicantly diminish brain nerve inammation, alleviate cognitive impairment, and partially reduce Ab deposition in 3xTg-AD mice. 22 Various inhibitors of the inammasome have been reported, 23 such as OLT1177, 19,24 CY-09, 25 tranilast, 26 oridonin, 27 benzenesulfonamide analogues, 28 sulphonamides (CRIDI, MCC950), [29][30][31] and non-steroidal anti-inammatory drugs (NSAIDs). 12,32 More inhibitors are targeted to Ab aggregation; [33][34][35] however, inhibitors that emphasize both Ab aggregation and inammasome are rare.
Since the synergism between Ab oligomers or plaques and pro-inammatory factors could increase the neural damage to the brain, 17 a combination therapy involving the inhibition of Ab aggregation and NLRP3 inammasome activation may enhance the therapeutic effect on AD. Herein, we integrate benzothiazole, an Ab-targeting group, with o-aminobenzoic acid, an analogue of the NLRP3 inammasome inhibitor mefenamic acid, 32 into a single molecule BPBA (Fig. 1), which may lead a dual inhibition of NLRP3 inammasome activation and Ab aggregation simultaneously. A series of experiments demonstrate that BPBA remarkably inhibits the self-and metalinduced Ab aggregation, reduces the level of inammatory cytokine IL-1b, restrains the activation of caspase-1 in vitro, and alleviates the formation of Ab oligomers and plaques as well as the Ab-associated toxicity in vivo.

Design, synthesis and physicochemical properties
The design of BPBA is based on the structures of benzothiazole and o-aminobenzoic acid; the former is a potential Ab-targeting group that has a specic affinity for Ab aggregates rich in bsheet structures, and the latter is an analogue of mefenamic acid, which is a known inhibitor of the NLRP3 inammasome. 32 In addition, the O, N, and S atoms in BPBA could chelate metal ions, which may inhibit the metal-induced Ab aggregation. We suppose that BPBA has the capability to prevent the formation of Ab plaques and restrain the activation of the NLRP3 inammasome in the brain of AD sufferers.
The synthesis and characterization of BPBA are shown in Scheme 1 and Fig. S1. † The synthetic intermediate BP was synthesized by reacting 2-aminobenzoic acid with 2-aminothiophenol in polyphosphoric acid (PPA) as reported in the literature. 36 BPBA was prepared by a modied literature method, 37 which produced a yellow solid with a yield of 40%.
BPBA is soluble in acetonitrile, methanol, and dimethyl sulfoxide (DMSO), but is insoluble in water. In the acetonitrile solution, two absorption peaks around 285 and 380 nm were detected by UV-vis spectroscopy, representing the existence of the benzene ring and the whole BPBA, respectively (Fig. S2A †). The maximum emission peak of BPBA is at 505 nm (l ex ¼ 380 nm) in the emission spectrum (Fig. S2B †).
The blood-brain barrier (BBB) is the main obstruction for developing anti-AD agents. 38 The BBB-penetrating ability of BPBA was rstly evaluated on the basis of Lipinski's "rule of ve", which predicts that a compound would cross the BBB when logBB is larger than 0.3. 39 The calculated logBB of BPBA is 0.18 (Table S1 †), and all other indices meet Lipinski's rule except Clog P, implying that it may potentially cross the BBB. The log P octanol/water partition coefficient was further determined by the shaking ask method and UV spectroscopy. The lipophilicity (log P o/w ) was calculated to be 1.24 AE 0.08 (Table  S2 †), suggesting that BPBA could penetrate the BBB.

Inhibition of Ab aggregation
Excessive Zn 2+ is associated with the generation of Ab aggregates, and Cu 2+ may lead to the aggregation of Ab and production of reactive oxygen species (ROS), 40,41 which would result in damage to neurons and synapses in AD patients. 42,43 Since the O, N, and S atoms in BPBA could coordinate to metal ions, the chelating ability of BPBA to Zn 2+ or Cu 2+ was investigated by uorescence titration in Tris-HCl buffer. As shown in Fig. 2, with the addition of Zn 2+ or Cu 2+ , the maximum uorescence intensity of BPBA decreased gradually and approached equilibrium when the ratio of [Zn 2+ ]/[BPBA] reached 2.0, whereas the uorescence intensity kept decreasing even when the ratio of The results indicate that BPBA can bind to Zn 2+ or Cu 2+ . However, since BPBA contains at least 4 coordination atoms and could form different chelates with Zn 2+ or Cu 2+ , that is, the product is not unique, it is hard to identify these species in a complex mixture. Thus we only tentatively obtained an apparent binding constant of BPBA to Zn 2+ at 1 : 1 stoichiometry, which was calculated to be 0.254 mM À1 according to the reported method. 44,45 The chelation of BPBA to Zn 2+ or Cu 2+ was further conrmed by the HR-MS spectra in the Tris-HCl buffer, where the chelate cations of Zn 2+ or Cu 2+ with BPBA were observed (Fig. S3 †).
The neurotoxicity of Ab aggregates largely originates from the b-sheet conformers. 46 The inhibition effect of BPBA on the bsheet formation of Ab was studied by the ThT assay, which is widely used to detect the content of b-sheets in Ab aggregates. 47 As presented in Fig. 3A and C, the uorescence intensity increased obviously when Ab 40 was incubated with Zn 2+ or Cu 2+ as compared with Ab 40 alone, especially for Zn 2+ , suggesting that these metal cations can promote the formation of b-sheet aggregates and the effect of Zn 2+ is greater than that of Cu 2+ . The uorescence intensity of Ab 40 decreased when BPBA was added into the solution of Ab 40 , Zn 2+ -, or Cu 2+ -Ab 40 aggregates, indicating that BPBA can inhibit the self-and metal-induced Ab 40 aggregation. Interestingly, the effect of Zn 2+ or Cu 2+ on Ab 42 is different from that on Ab 40 . As shown in Fig. 3B and D, compared with the uorescence intensity of Ab 42 , in the presence of Zn 2+ or Cu 2+ , the uorescence intensity decreased apparently. This is consistent with the ndings reported by Mirica, et al., that is, Cu 2+ stabilizes soluble Ab 42 oligomers, and Zn 2+ leads to the formation of insoluble amorphous, non-brillar aggregates. 48 Therefore, the aggregates of Ab 42 in the presence of Cu 2+ are mainly formed by the self-aggregation of bsheet conformers rather than by the induction of metal ions. Comfortingly, BPBA can also inhibit the self-aggregation of Ab 42 (Fig. S4 †). The association constant of BPBA to Ab 42 was calculated to be 8.727 AE 4.023 mM À1 . Although BPBA contains a ThT core, it did not interfere with the uorescence of ThT in the presence of Ab 42 (Fig. S5 †).

Morphological alteration of Ab
The morphology of Ab 42 and Zn 2+ -or Cu 2+ -induced Ab 42 aggregates in the absence or presence of BPBA was further visualized by transmission electron microscopy (TEM). As shown in Fig. 4, some brils were formed in the solution of Ab 42 , and large amounts of mature aggregates were observed aer the addition of Zn 2+ or Cu 2+ to the Ab 42 solution, which are consistent with the literature; 49 however, in the presence of BPBA, the Ab 42 -induced brils and Zn 2+ -or Cu 2+ -induced Ab 42 aggregates changed into granule-like species or short fragments. The morphological changes indicate that BPBA can effectively inhibit the self-formed Ab 42 brils and metal-induced Ab 42 aggregates. The chelation of BPBA with metal ions is the primary reason for the disaggregation of Cu 2+and Zn 2+ -Ab 42 aggregates.

Inuence on the hydrophobicity of Ab
Ab brils are highly hydrophobic, which can be detected by using the uorescent probe 8-aniline-1-naphthalenesulfonic acid (ANS). 50 The uorescence of ANS is enhanced signicantly aer binding to the hydrophobic structure on the surface of proteins. 51 Therefore, the inuence of BPBA on the hydrophobicity of Ab can be reected by the uorescence changes of ANS. As shown in Fig. 5, the uorescence intensity of ANS decreased and the maximum emission wavelength red-shied once Cu 2+ was added to the solution of Ab 42 , suggesting that Cu 2+ reduced the number of surface hydrophobic structures or the hydrophobicity of Ab 42 brils. In contrast, the uorescence intensity of ANS increased aer Zn 2+ was incubated with Ab 42 , indicating that Zn 2+ increased the hydrophobicity of Ab 42 brils, and its impact on the hydrophobicity is different from that of Cu 2+ . Once BPBA was added into the above solutions, the uorescence intensity of ANS decreased signicantly, suggesting that BPBA has a remarkable inhibitory effect on the surface hydrophobic structures of self-and metal-induced Ab 42 aggregates. These results verify that BPBA can reduce the hydrophobicity or increase the hydrophilicity of Ab 42 aggregates. The uorescence interference of BPBA with ANS is negligible on this occasion (Fig. S6 †).

Effect of BPBA on nerve cells and Ab toxicity
Before assessing the effect of BPBA on the Ab toxicity to neuron cells, we rst examined the possible neurotoxicity of BPBA. Mouse neuroblastoma N2a (or Neuro-2a) cells are oen used to study the pathological mechanism of AD. 52 Therefore, the survival of N2a cells aer incubation with BPBA for 24 h was tested by the MTT assay. As shown in Fig. 6A, the survival rate of N2a cells is around 100% even when the concentration of BPBA reached 140 mM, indicating that it is almost non-toxic to the neuron cells. The extremely low neurotoxicity suggests that BPBA per se is safe for neuron cells in the following experiments. The neurotoxicity of Ab 42 , Cu 2+and Zn 2+ -Ab 42 aggregates in the absence and presence of BPBA toward neuronal model cell PC12 from rat pheochromocytoma was tested using the MTT assay. As shown in Fig. 6B, Ab 42 , Cu 2+and Zn 2+ -Ab 42 aggregates were markedly toxic to the PC12 cells, with the cell viability declining by more than 30%. The toxicity of Ab 42 , Cu 2+and Zn 2+ -Ab 42 aggregates was signicantly attenuated in the presence of BPBA, with the cell viability rising to more than 80%. The results show that BPBA can suppress the neurotoxicity of Ab 42 , Cu 2+and Zn 2+ -Ab 42 aggregates and enhance the viability of neuron cells.

Inhibition on inammasome
The NLRP3 inammasome is an intracellular complex that activates caspase-1, which processes the IL-1b precursors into  active molecules, and the mature IL-1b is released to the extracellular uid. 13 Human acute monocytic leukemia THP-1 cells are similar to the phenotype and functional characteristics of human primary monocytes/macrophages, therefore are commonly used as their model cells. 53 THP-1 cells become mononuclear macrophages under the differentiation induced by phorbol ester (PMA). When the THP-1 macrophages were costimulated with lipopolysaccharide (LPS) and ATP, the NLRP3 inammasome/caspase-1 was activated and a variety of cytokines like IL-1b were synthesized and released. 54 Therefore, the effect of BPBA on the NLRP3 inammasome was evaluated using the THP-1 macrophages.
The NLRP3 inammasome requires the adapter protein ASC to activate caspase-1. Aer inammasome activation, ASC assembles into a large protein complex called "speck", which can be detected by immunocytochemistry as the size reaches around 1 mm. Therefore, the formation of ASC specks is regarded as a simple upstream indicator of inammasome activation. 25,55 As shown in Fig. 7, red ASC specks were formed in the cytoplasm of THP-1 macrophages aer stimulation with LPS and ATP; when the stimulated cells were treated with BPBA subsequently, the formation of ASC specks was dramatically inhibited, suggesting that BPBA can inhibit the activation of the NLRP3 inammasome. Theoretically, the inhibition of BPBA on ASC specks should depend on its concentration; however, a quantitative relationship is unavailable because the exact position and amount of ASC specks are difficult to determine. According to the inhibition effect, 20 mM BPBA seems to be an appropriate concentration for the inhibition.
As a result of ASC decline, the activation of caspase-1 was inhibited accordingly. As shown in Fig. 8, caspase-1 mainly exists as inactive pro-caspase-1 with or without stimulation; the activated caspase-1 p12 and caspase-1 p10 only increased aer THP-1 macrophages were co-stimulated with LPS and ATP. However, in the presence of BPBA, the expression of activated caspase-1 decreased to the unstimulated level.
IL-1b is a main inducer of inammation and one of the main mediators of innate immune response, which is produced as an inactive precursor (pro-IL-1b) that requires cleavage by caspase-1 for activation and secretion. 56 Its level is elevated in the brain of AD patients and is associated with the progression and onset of AD. 20 As shown in Fig. 9, a large amount of IL-1b was secreted when THP-1 macrophages were stimulated with LPS and ATP, whereas the secretion of IL-1b was dramatically suppressed by  BPBA, hence manifesting that BPBA can inhibit the release of IL-1b. All these results suggest that BPBA could inhibit the activation of the NLRP3 inammasome and suppress the maturation and release of IL-1bthe nal product of the inammasome, thus predicting that BPBA could reduce the inammatory response in AD patients.

Production of ROS and alleviation of paralysis in C. elegans
The transgenic C. elegans expressing Ab gene in muscles or neurons is widely used as an in vivo model of AD to test the effect of compounds on Ab aggregation and toxicity. 57 Ab plaques can induce ROS accumulation and oxidative stress, thus aggravating the pathological progress of AD ulteriorly. 58 The changes of ROS in C. elegans CL4176 strains with or without BPBA were detected with a 2 0 ,7 0 -dichlorodihydrouorescein diacetate (DCFH-DA) probe. The ROS level was reected by the DCF uorescence that is positively dependent on the oxidization of DCFH by intracellular ROS. 59 As shown in Fig. 10A, the green uorescence intensity of BPBA-treated worms is signicantly weakened as compared with that of the control, suggesting that BPBA can inhibit the production of ROS in the worms. The quantitative data of the uorescence intensity are shown in Fig. 10B. The relative uorescence intensity of the BPBA-treated group markedly decreased as compared to that of the control, indicating that BPBA can inhibit the level of ROS in C. elegans.
CL4176 specically expresses the Ab 42 gene in the SMG-1 mRNA temperature induction system, which results in the time-dependent aggregation of Ab and paralytic phenotypes. 60 The activated SMG-1 system recognizes the Ab gene and degrades it at 15 C, causing the worms to produce a low level of Ab and move as rollers; when the temperature is raised to 25 C, the SMG-1 system is inactivated and Ab aggregates are formed in the muscle cells, which paralyze the nematode gradually. The effects of BPBA on the Ab-induced paralysis in CL4176 are shown in Fig. 10C. The BPBA-treated CL4176 nematodes were not paralyzed until 42 h aer the temperature upshi; however, the paralysis rate of the untreated nematodes began to rise quickly thereaer, reaching more than 50% at 44 h. The paralysis rate of the worms treated with BPBA (68 mM) is only 7% at 48 h, while that of the control group is 73%. Evidently, BPBA can inhibit the Ab-induced paralysis in transgenic CL4176 strains; in other words, it can downregulate the expression of the Ab 42 gene in CL4176 or alleviate the Ab toxicity to neurons of C. elegans.

Reduction of Ab and IL-1b in AD mice
The potential of BPBA to eliminate the pathogenic factors of AD was further veried in mice. APP/PS1 double-transgenic mice that produce elevated levels of Ab by expressing human APP and PS1 mutants from 6 months of age were used as AD models. 61 To select an appropriate dose for the assay, the acute toxicity of BPBA to wild type (WT) C57BL/6J mice was rst evaluated. No obvious toxicity was observed at a dose of 10 mg kg À1 BPBA based on the mortality in 2 weeks (Fig. S7 †). The APP/PS1 mice   were treated with BPBA at a dose of 5 mg kg À1 every 3 days for 3 months. The brain tissues of the BPBA-treated and untreated AD mice were taken out and the expression of Ab was analyzed by western blotting. As shown in Fig. 11A and S8, † the Ab species with a molecular weight #55 kDa (oligomeric species) decreased signicantly, thus conrming that BPBA can inhibit the formation of Ab oligomers and plaques in the brain of AD mice. The results also imply that BPBA could pass through the BBB of the mice.
The effect of BPBA on pro-inammatory IL-1b in the brain of AD mice was also investigated. As shown in Fig. 11B, the content of IL-1b in the brain of saline-treated wide-type (WT) and AD mice is 125 and 250 pg mL À1 , respectively; however, that in the brain of BPBA-treated mice is 100 pg mL À1 , which is even lower than that in the WT mice. The results demonstrate that BPBA can reduce the level of IL-1b in vivo and thereby douse the inammatory responses. Since IL-1b is an immunomodulatory cytokine, the decrease of IL-1b production may further affect the innate immune cells in the brain.

Efficacy on APP/PS1 mice
Finally, the in vivo therapeutic effect of BPBA on the learning and cognitive abilities of APP/PS1 mice was assessed by the Morris water maze (MWM) test according to reported procedures. 61 As shown in Fig. 12A, the time required for the BPBAtreated mice to nd the survival platform is similar to that for the saline-treated WT mice, but is shorter than that for the AD  control mice. On the 6th day, the survival platform was removed. The frequency of presence in the survival platform area (quadrant II) for the BPBA-treated mice increased within 60 s as compared with that for the saline-treated AD mice (Fig. 12B). Preliminary behavioral experiments show that BPBA can slightly alleviate the memory impairment of AD mice. The benecial effect of BPBA on the memory of AD mice seems not so effective as expected, because the recovery of memory involves ceasing of neuron damage and regeneration of damaged neurons. 62 BPBA can eliminate the pathogenic factors of neuron damage but cannot restore the damaged neurons. It is very possible that when the treatment began, the neuron damage had occurred, so the improvement of memory is not evident. This may be the reason why so many AD drug candidates failed in the clinic. Fortunately, the AD symptom did not get worse under the treatment of BPBA; in other words, BPBA can stop the progression of the disease, which is superior to most AD drugs.

Conclusion
Ab aggregation is believed to be a key factor in the pathogenesis of AD; likewise, chronic neuroinammation mediated by the activation of the NLRP3 inammasome also plays a crucial role in the AD pathogenesis. 10,63 Ab deposits activate the NLRP3 inammasome, leading to an overproduction of IL-1b and neuroinammation. Therefore, dissociating Ab aggregates not only directly reduces the Ab-induced neurotoxicity to neurons, but also inhibits the activation of the NLRP3 inammasome, curbs inammatory responses, and decreases the release of neuro-destructive inammatory cytokines. 17,31 BPBA possesses the basic structural characters of both ThT and NSAID mefenamic acid, thus showing Ab-targeting and anti-inammatory abilities. Its function includes inhibiting Ab aggregation, reducing ROS production, alleviating Ab toxicity, deactivating the NLRP3 inammasome, and restraining IL-1b release. In vivo studies on transgenic C. elegans show that BPBA can allay the Ab-associated paralysis or Ab toxicity to the neural system of C. elegans. Although BPBA cannot effectively recover the learning and cognitive abilities of AD mice, it can terminate the progression or deterioration of AD. The synergistic impact of BPBA on Ab aggregation and neuroinammation may bring about a new inspiration for the design of anti-AD drugs. Nevertheless, the exact mechanism of regulating the NLRP3 inammasome as well as the interactions between NLRP3 inammasome activation and other signaling pathways in AD remain to be claried.

Author contributions
T. Y. and L. Z. prepared the compounds and performed the experiments under the supervision of X .Y. W. and Z. J. G, Y. C. S. and Z. Z. Z. performed the western blotting and animal assays, S. X. J. and T. Y. analyzed the data and wrote the original dra, X. Y. W. edited the manuscript.

Ethical statement
All animal experiments were performed in accord with the institutional animal use and care regulations approved by MARC and GDMLAC.

Conflicts of interest
There are no conicts to declare.