*3.2. A*β *PF-Selective Antibody*

PFs have been identified in the human brain and the APP transgenic mouse brain [48,56]. mAb158 is a murine monoclonal antibody developed to selectively target HMW-Aβ1-42 assemblies [56]. Using an enzyme-linked immunosorbent assay (ELISA), it has been elucidated that mAb158 has an at least 1000-fold higher selectivity for PFs than monomeric Aβ and 10-15 times better binding affinity to PFs than to mature fibrils, thereby targeting the more toxic species of the peptide [57]. In immunohistochemistry, mAb158 also detects Aβ in plaques and the vasculature of AD brains because of the massive amount of Aβ in these structures [58]. In addition, Lord et al. reported that mAb158 inhibits in vitro Aβ1-42 fibril formation and protected cells from Aβ PF-induced cytotoxicity [59]. A co-culture study of astrocytes, neurons, and oligodendrocytes exposed to Aβ1-42 PFs in the presence or absence of mAb158 demonstrated that the presence of mAb158 almost entirely abolished Aβ accumulation in astrocytes, indicating an effect towards Aβ PF degradation. Consequently, mAb158 treatment was shown to rescue neurons from Aβ-induced cell death [60].

The treatment of tg-APPArcSwe mice with mAb158 resulted in the prevention of plaque formation if the antibody was administered before the appearance of plaques in young mice. If the treatment was started later in this mouse model, levels of insoluble Aβ were unaffected in the brains of plaque-bearing older mice. However, in both cases, soluble Aβ PF levels were diminished, supporting the notion that mAb158 can selectively reduce PF levels [59]. Similarly, the authors found that PF levels were elevated in young tg-APPArcSwe mice compared with several transgenic models lacking the Arctic mutation. In older tg-APPArcSwe mice with plaque deposition, the levels of Aβ PFs were approximately 50% higher

than in younger mice, whereas levels of total Aβ were exponentially increased. Young tg-APPArcSwe mice showed deficits in spatial learning, and individual performances in the Morris water maze were inversely correlated with levels of Aβ PF, but not with total Aβ levels. These findings indicated that Aβ PFs accumulated in an age-dependent manner, and increased levels of Aβ PFs may result in spatial learning impairments in tg-APPArcSwe mice [47]. Lannfelt et al. reported that the murine version of mAb158 reached the brain and reduced brain PF levels by 42% in an exposure-dependent manner both after long-term (13 weeks) and short-term (4 weeks) treatment in tg-APPArcSwe mice [14]. Notably, a 53% reduction in PFs/oligomers in the cerebrospinal fluid (CSF), found to be correlated with reduced brain PF levels, was observed after long-term treatment, suggesting that CSF PFs/oligomers may be used as potential biomarkers of AD [14].

Recently, Sehlin's group succeeded in facilitating the brain uptake of mAb158 by using transferrin receptor-mediated transcytosis across the blood–brain barrier in tg-APPArcSwe mice [61]. ELISA analysis of the brain extracts demonstrated a 40% reduction in soluble Aβ PFs in both ten-fold lower-dose modified mAb158 and high-dose mAb158-treated mice, whereas there was no Aβ PF reduction in mice treated with a low dose of mAb158 [61]. Furthermore, ex vivo autoradiography and PET imaging have revealed different brain distribution patterns of modified mAb158 (brain parenchyma) and mAb158 (central periventricular areas), suggesting that these antibodies may affect Aβ levels by different mechanisms. This strategy may allow for decreased antibody doses, thereby reducing the side effects and treatment costs [61].

#### *3.3. Clinical Application of mAb158*

BAN2401, a humanized IgG1 monoclonal form of mAb158, exhibits a strong binding preference for soluble Aβ PFs compared with monomers [14]. In addition, it has been confirmed that both mAb158 and BAN2401 efficiently immunoprecipitate soluble Aβ aggregates in human AD brain extracts.

The first clinical study of BAN2401 demonstrated that the compound was safe and well tolerated in mild to moderate AD [62]. The incidence of amyloid-related imaging abnormalities (ARIA-E for edema /H for hemorrhage) on brain MRI scans was comparable to that of the placebo. BAN2401 exposure was approximately dose-proportional, with a serum terminal elimination half-life of approximately seven days. Only a slight increase in plasma Aβ1-40 was observed, but there were no measurable effects of BAN2401 on CSF biomarkers such as Aβ1-42, total-tau, and phosphorylayed-tau (p-tau) [62]. A recent phase 2 randomized trial reported that BAN2401's highest dose (10 mg/kg) significantly slowed cognitive decline in early AD, with a concomitant reduction in amyloid plaques, as measured by amyloid PET compared with placebo at 18 months [21]. BAN2401 significantly reduced amyloid plaques in the brain at all five treatment doses used in the trial, which involved 856 patients with mild cognitive impairment. The 30% slowing of cognitive decline at 18 months was based on the Alzheimer's Disease Composite Score (ADCOMS) created by Eisai. On the more widely used Alzheimer's Disease Assessment Scale cognitive subscale (ADAS-Cog), the highest dose of BAN2401 slowed a cognitive decline of 47% compared with placebo. However, the trial was not large enough to definitively demonstrate efficacy in improving cognitive function according to an overall optimistic statement from the Alzheimer Association. The drug also did not achieve its primary efficacy endpoint, namely, a change from baseline on the ADCOMS at 12 months [21]. Currently, BAN2401 is a part of an ongoing phase 3 clinical trial. In contrast, other clinical trials of monoclonal antibodies targeting fibrillar Aβ, such as bapineuzumab [63], or soluble monomeric Aβ, such as solanezumab [20], have failed to produce clinical effects.

In the fall of 2019, after trials of the drug EMERGE (aducanumab; BIIB037) were previously discontinued following a phase III futility analysis, Biogen, the company that developed the drug, announced that subsequent analysis of a larger dataset instead showed that EMERGE had met its primary endpoint. Patients on the highest dose, 10 mg/kg, had a significant reduction in decline in terms of the primary endpoint using the Clinical Dementia Rating Scale-Sum of Boxes (CDR-SB). This group also declined less in terms of secondary endpoints, including the Mini-Mental State Examination

(MMSE), ADAS-Cog, and the Alzheimer's Disease Cooperative Study/Activities of Daily Living scale adapted for patients with mild cognitive impairment (ADCS-ADL-MCI). In a parallel clinical trial of aducanumab, termed the ENGAGE trial, aducanumab did not meet the primary endpoint; however, an exploratory analysis suggested that a subgroup of people who had received 10 or more 10 mg/kg doses declined more slowly, which is consistent with the EMERGE participants. In both trials, aducanumab caused a dose-dependent reduction in brain Aβ and CSF p-tau. Based on the updated data analysis, Biogen announced plans to apply for regulatory approval of aducanumab in the US in early 2020 [64]. Since aducanumab may also bind aggregates such as oligomers of Aβ [65], these results may be important for interpreting data from the phase 3 clinical trial of BAN2401.

#### **4. PFs Are Present in Other Neurodegenerative Diseases**

PFs are formed from proteins implicated in other neurodegenerative diseases, including tauopathy [66], Parkinson's disease [67,68], familial amyloid polyneuropathy [69], and Huntington's disease [70], indicating a common mechanism. Similar to Aβ, tau and α-synuclein (αS) also form PFs with annular, pore-like structures, thereby exerting membrane permeabilization activity [66,67]. Analyses of annular tau PFs in brain tissue from patients with progressive supranuclear palsy, as well as that from the P301L mouse model, indicated that the annular PFs of tau are preceded by tau oligomers and do not go on to form neurofibrillarly tangles (mature fibrils) [66]. In addition, it was recently reported that the αS oligomer and PFs interconvert during polymerization reactions, using the thioflavin T assay combined with SEC and EM [68]. Similarly, Groenning et al. described a dynamic transthyretin (TTR) protofibril structure that exchanges protomers with highly unfolded monomers in solution, using a combination of primarily small-angle X-ray scattering and hydrogen exchange mass spectrometry analysis. The TTR PFs were shown to only grow to an approximate final size of 2900 kDa and a length of 70 nm [69]. In a recent micro electron diffraction study at 0.75Å resolution, ultrahigh-resolution cryo-EM revealed that prion PFs are stabilized by a dense three-dimensional network of stabilizing hydrogen bonds that link residues between and within its β strands through polar clasps [71].

#### **5. Conclusions and Future Perspectives**

Unlike current therapies limited to the treatment of AD symptoms, research on Aβ aggregation has rapidly advanced, with growing evidence that soluble pre-fibrillar aggregates (i.e., oligomers of Aβ) are proximate neurotoxins. Indeed, recent data from both in vitro and in vivo studies have suggested that HMW oligomers as PFs induce neuronal injury and cognitive deficits via multiple mechanisms, including not only increasing Aβ plaque accumulation but also increasing direct membrane and synaptic damage. Furthermore, additional projects to fully characterize the PFs actually present in the human brain have been undertaken. Aβ PFs may be the primary pathogenic species of Aβ-related cognitive deficits, particularly in the early stage of AD, although it remains to be established how Aβ PFs, alone or together with other soluble oligomeric Aβ species, cause the neurodegeneration leading to AD. Disease-modifying therapies targeting toxic PFs will reach the clinical stage in the near future, and may have the potential to delay or even halt the further progression of AD. Further clarification of the toxic PFs of brain Aβ should aid in the development of more effective and safe drugs, as well as in novel diagnostic assays.

**Author Contributions:** K.O. and M.T. wrote the paper. Authorship must be limited to those who have contributed substantially to the work reported. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a Grant-in-Aid for Scientific Research (C) (26461266) (K.O.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and a Grant for Research and Development Grants for Dementia from the Japan Agency for Medical Research and Development (16dk0207021h0001) (K.O.).

**Conflicts of Interest:** The authors declare no conflicts of interest.
