Discovery of Compounds That Selectively Repress the Amyloidogenic Processing of the Amyloid Precursor Protein: Design, Synthesis and Pharmacological Evaluation of Diphenylpyrazoles

The rationale to define the biological and molecular parameters derived from structure–activity relationships (SAR) is mandatory for the lead selection of small drug compounds. Several series of small molecules have been synthesized based on a computer-assisted pharmacophore design derived from two series of compounds whose scaffold originates from chloroquine or amodiaquine. All compounds share similar biological activities. In vivo, Alzheimer’s disease-related pathological lesions are reduced, consisting of amyloid deposition and neurofibrillary degeneration, which restore and reduce cognitive-associated impairments and neuroinflammation, respectively. Screening election was performed using a cell-based assay to measure the repression of Aβ1–x peptide production, the increased stability of APP metabolites, and modulation of the ratio of autophagy markers. These screening parameters enabled us to select compounds as potent non-competitive β-secretase modulators, associated with various levels of lysosomotropic or autophagy modulatory activities. Structure–activity relationship analyses enabled us to define that (1) selectively reducing the production of Aβ1–x, and (2) little Aβx–40/42 modification together with (3) a decreased ratio of p62/(LC3-I/LC3-II) enabled the selection of non-competitive β-secretase modulators. Increased stability of CTFα and AICD precluded the selection of compounds with lysosomotropic activity whereas cell toxicity was associated with the sole p62 enhanced expression shown to be driven by the loss of nitrogen moieties. These SAR parameters are herein proposed with thresholds that enable the selection of potent anti-Alzheimer drugs for which further investigation is necessary to determine the basic mechanism underlying their mode of action.


Introduction
Alzheimer's disease (AD), the most frequent form of dementia worldwide, is a brain disease associated with two neuropathological processes, leading to a slow decline in cognitive and behavioral disabilities. Neuropathological lesions include neurofibrillary tangles (NFTs) and amyloid deposits, together with astrogliosis, neuroinflammation, and neuronal death [1]. NFTs are made up of the intraneuronal accumulation and aggregation of abnormally modified isoforms of the microtubule-associated Tau. Parenchymal amyloid deposits are composed of amyloid-β (Aβ) peptides originating from complex and sequential proteolytic cleavages of a precursor protein, namely, the amyloid precursor protein (APP) (for a comprehensive review see [2]). The non-amyloidogenic pathway, initiated by the α-secretase cleavage of APP, is opposed to the amyloidogenic pathway that is initiated of abnormally modified isoforms of the microtubule-associated Tau. Parenchymal amyloid deposits are composed of amyloid-β (Aβ) peptides originating from complex and sequential proteolytic cleavages of a precursor protein, namely, the amyloid precursor protein (APP) (for a comprehensive review see [2]). The non-amyloidogenic pathway, initiated by the α-secretase cleavage of APP, is opposed to the amyloidogenic pathway that is initiated by a primary β-secretase cleavage at the first amino acid of the Aβ peptide sequence ( Figure 1) [3,4]. Both secretase-mediated steps shed soluble ectodomains of APP (sAPPα and sAPPβ) and membrane-bound carboxyl-terminal fragments (APP-CTFs), referred to as αCTF and βCTF. These αCTFs or βCTFs are further cleaved by the γsecretase to give rise to p3 and Aβ peptides, respectively, along with the APP intracellular domain (AICD). Along the cell-secretory pathway, APP can also be cleaved by β-secretase at the β′-site at position 11 of the Aβ peptide sequence, which is suggested to be protective [5][6][7][8]. The definite diagnosis relies on the presence of both NFTs and amyloid deposits, which are therefore considered to stem pathophysiological processes leading to AD, and which posit that therapeutic development should therefore target both of these processes. Figure 1. Summary scheme of the major APP processing pathways. The amyloid precursor protein is a type I transmembrane protein subject to sequential and alternative proteolytic cleavage. The βsecretase cleaves APP at the N-terminus of the Aβ peptide sequence to generate a soluble fragment sAPPβ and a carboxy-terminal fragment CTFβ. This β-secretase can alternatively cleave APP at position 11 and generate sAPPβ' and CTFβ'. The α-secretase cleavage at position 17 precludes the Aβ formation and generates sAPPα and CTFα. All CTFs can be further processed by the γ-secretase to generate the Aβ1-x, P9, and amyloid intracellular domain (AICD). AD treatment remains symptomatic and currently, disease-modifying treatment remains ill-defined except for aducanumab immunotherapy, which reduces the amyloid load and Tau PET imaging in clinical trials [9]. The definite diagnosis relies on these pathological processes, which are therefore all related to AD pathophysiology. Therefore, alternative disease-modifying therapeutic options are urgently needed, and over the past few years, much effort has been dedicated to the development of such disease-modifying drugs [10]. One such strategy would be the development of drugs that efficiently modify both amyloid and Tau pathological processes.
Redirection of APP processing can be achieved either by blocking the amyloidogenic pathway or by promoting the non-amyloidogenic pathway. The results of these activities would be a decrease in Aβ secretion and an increase in αCTF. We previously showed that CQ inhibits Aβ production, whereas levels of other APP metabolites such as APP-CTFs and AICD are maintained and even increased, whereas the γ-secretase cleavage of Notch Figure 1. Summary scheme of the major APP processing pathways. The amyloid precursor protein is a type I transmembrane protein subject to sequential and alternative proteolytic cleavage. The β-secretase cleaves APP at the N-terminus of the Aβ peptide sequence to generate a soluble fragment sAPPβ and a carboxy-terminal fragment CTFβ. This β-secretase can alternatively cleave APP at position 11 and generate sAPPβ' and CTFβ'. The α-secretase cleavage at position 17 precludes the Aβ formation and generates sAPPα and CTFα. All CTFs can be further processed by the γ-secretase to generate the Aβ 1-x , Aβ 11-x , P9, and amyloid intracellular domain (AICD).
We have recently developed two families of compounds derived from either CQ (family A) or amodiaquine (AQ, family B) (Figure 2) that demonstrated a strong inhibitory effect on both Aβ  and Aβ  secretions [13][14][15][16][17][18][19]. These effects were associated with a strong increase in αCTFs and AICD levels. These families of molecules interact with the autophagic/endolysosomal systems, some of which have been shown to be effective against both amyloid and Tau pathologies in vivo [14,17]. The lead compound of family A is currently in a phase II clinical trial for the treatment of tauopathy progressive supranuclear palsy [20]. To try to identify new and more efficient compounds in the absence of a defined molecular target of the previously developed compounds, a ligand-based computer-assisted pharmacophore modeling approach [21], coupled with a de novo design, was implemented. In the first paper, we described a new family of compounds based on a biaryl scaffold decorated with amino side chains (family C, Figure 2) [22]. Among these compounds, several were shown to inhibit Aβ 1-x peptide secretion and to promote αCTFs and AICD stabilities as the parent compounds (families A and B). In this previous structure-activity relationship based on the analysis of APP metabolism and markers of autophagy, non-competitive β-secretase inhibitors with or without a lysosomotropic activity were identified [22]. Interestingly, compound ) with n = 3, m = 2, X = CH, a phenyl ring, Ar, and a dimethylamino group, NR 1 R 2 , provided the most interesting pharmacological properties. In vivo, this compound, with reduced lysosomotropic activity, reduced neurofibrillary degeneration, astrogliosis, neuroinflammation, and short-term spatial memory impairments in the Thy-Tau22 transgenic mouse model of hippocampal tauopathy [23]. Our results suggest that the non-competitive β-secretase inhibition is necessary for the pharmacological effect in vivo whereas the lysosomotropic activity should be reduced and likely dispensable for drug efficacy. remains unmodified, precluding the contribution of the γ-secretase in this sequential proteolytic process of APP [11,12].
We have recently developed two families of compounds derived from either CQ (family A) or amodiaquine (AQ, family B) (Figure 2) that demonstrated a strong inhibitory effect on both  and  secretions [13][14][15][16][17][18][19]. These effects were associated with a strong increase in αCTFs and AICD levels. These families of molecules interact with the autophagic/endolysosomal systems, some of which have been shown to be effective against both amyloid and Tau pathologies in vivo [14,17]. The lead compound of family A is currently in a phase II clinical trial for the treatment of tauopathy progressive supranuclear palsy [20]. To try to identify new and more efficient compounds in the absence of a defined molecular target of the previously developed compounds, a ligandbased computer-assisted pharmacophore modeling approach [21], coupled with a de novo design, was implemented. In the first paper, we described a new family of compounds based on a biaryl scaffold decorated with amino side chains (family C, Figure  2) [22]. Among these compounds, several were shown to inhibit Aβ1-x peptide secretion and to promote αCTFs and AICD stabilities as the parent compounds (families A and B). In this previous structure-activity relationship based on the analysis of APP metabolism and markers of autophagy, non-competitive β-secretase inhibitors with or without a lysosomotropic activity were identified [22]. Interestingly, compound ) with n = 3, m = 2, X = CH, a phenyl ring, Ar, and a dimethylamino group, NR1R2, provided the most interesting pharmacological properties. In vivo, this compound, with reduced lysosomotropic activity, reduced neurofibrillary degeneration, astrogliosis, neuroinflammation, and short-term spatial memory impairments in the Thy-Tau22 transgenic mouse model of hippocampal tauopathy [23]. Our results suggest that the non-competitive β-secretase inhibition is necessary for the pharmacological effect in vivo whereas the lysosomotropic activity should be reduced and likely dispensable for drug efficacy. Design of a biphenylpyrazole series based on families A and B. Structure of compounds A1 and B1, representative compounds of families A and B, respectively. Ligand-based approach leading to biaryl family C [17,22] and diphenylpyrazole family D. Compounds 24-35 and 61-72 are those studied in this paper.
In this paper, we describe a new family of compounds derived from a diphenylpyrazole scaffold decorated with amino side chains (family D, Figure 2). In the Design of a biphenylpyrazole series based on families A and B. Structure of compounds A1 and B1, representative compounds of families A and B, respectively. Ligand-based approach leading to biaryl family C [17,22] and diphenylpyrazole family D. Compounds 24-35 and 61-72 are those studied in this paper.
In this paper, we describe a new family of compounds derived from a diphenylpyrazole scaffold decorated with amino side chains (family D, Figure 2). In the screening process, the assessment of APP metabolites including soluble APP fragments arising from the βor α-secretase endoprotease activity, the so-called sAPPβ and sAPPα, together with the carboxy-terminal APP fragments (APP-CTFs) and Aβ 1-X peptides, provide outcomes on non-competitive β-secretase inhibition [22]. The lysosomotropic activity of our compounds was evaluated by a lower degradation rate of CTFα, AICD, and LC3-I/LC3-II, and p62/SQSTM1(Sequestresome protein 1) markers of autophagy whose expression and ratio increase as the result of autophagy to lysosome flux inhibition.

Ligand-Based Pharmacophore Modeling
We previously described a computer-assisted ligand-based approach and applied this method to determine a common pharmacophoric model between the two families of compounds A and B considering the sole inhibitory effect on Aβ production ( Figure 1) [22]. In brief, a set of 51 compounds of family A [14] and 35 of family B [18,19] were chosen regarding the ability of the compounds to inhibit Aβ 1-x secretion. Compounds A1 and B1 are representative compounds of families A and B, respectively. The best 3D pharmacophoric model, describing four spatial points Positive/Positive/Positive/Aromatic (PPPA), was used to design new APP modulators. Several structures were then conceptualized to spatially organize pharmacophoric elements in agreement with this model. We then chose chemical structures based on a scaffold enabling the orientation of amino side chains in three different orientations. Two different scaffolds were first obtained: (1) a biaryl scaffold (family C) with amino side chains previously described [22], and (2) a diphenylpyrazole scaffold (family D) decorated with amino side chains.
This family of compounds was designed around a pyrazole scaffold substituted at the 1-and the 3-positions by phenyl rings, each of which being optionally further substituted by R 1 and R 2 groups, respectively. Two series of compounds were designed with different substituents at the 4-position of the pyrazole moiety, one with a methyl substituent and the other one with an (N, N-dimethylamino)alkyl group substituent wherein the alkyl is methyl or propyl. In both series, the optional R 1 and R 2 substituents of the phenyl rings were independently chosen from (dialkylamino)alkyl and (heterocyclylamino)alkyl. The nature of the amino side chains on the pyrazole scaffold was chosen based on previous biological readouts. Different modulations then allowed us to select and validate the importance of adding two or three different amino side chains: monoamine, linear, or cyclic diamine.
The compounds were then tested to evaluate the impact of the structural modifications on cytotoxicity and metabolism of APP (APP-CTF quantification, secretion of Aβ 1-x peptides, ...) using SY5Y-APP wt cells, a well-established model mastered in our laboratory and used to evaluate and compare compounds of all family of drugs synthesized thus far. sAPPα and sAPPβ were also quantified. In addition to Aβ  and Aβ  , which arise from the canonical β-secretase and γ-secretase, Aβ x-38/40/42 peptides were also measured since truncated Aβ species are also produced such as Aβ 11-x , which are generated by the β-secretase along the secretory pathway ( Figure 1) [5].

Synthetic Chemistry
For the synthesis of the first series with (N, N-dimethylamino)alkyl groups in the 4-position of the pyrazole scaffold of compounds 24-35, we first describe the preparation of intermediates 10-11, 14-15 (Scheme 1).

Scheme 1.
Synthesis of intermediates 10-11, [14][15] Arylhydrazines 3 and 4 were prepared from anilines 1 and 2 by diazotization in hydrochloric acid (37%), followed by the reduction of the diazonium salts using tin(II) chloride according to the reaction protocols known in the literature [24][25][26]. Compound 5 was prepared to start from 1-[4-(3-bromopropyl)phenyl]ethanone and dimethylamine (2 M in methanol). Compound 5 was then condensed with 3-4 to provide hydrazone derivatives 6-7. By the double addition of 28], the desired cyclized aldehydes 8-9 were obtained. Starting from derivatives 8-9, two routes were used to afford the final compounds 10-11, 14-15 with [(N, N-dimethyl)amino]methyl or [(N, Ndimethyl)amino]propyl substituents at the 4-position of the pyrazole scaffold. The first route was a reductive amination of 8-9 with sodium triacetoxyborohydride, dimethylamine, and acetic acid in DCE to give compounds [10][11]. The second route, in two steps, was a Horner-Wadsworth-Emmons olefination of 8-9 with sodium hydride, diethylcyanomethylphosphonate in THF, followed by catalytic hydrogenation of the intermediates in the presence of Raney nickel and palladium on carbon in methanol saturated with ammonia under a hydrogen atmosphere, to afford compounds 12-13. The last step was the reductive methylation of derivatives in methanol with formaldehyde, and sodium triacetoxyborohydride to afford the corresponding compounds 14-15. Starting from compounds 10-11 and 14-15, final compounds 24-35 were prepared in three steps (Scheme 2). Scheme 1. Synthesis of intermediates 10-11, [14][15]. Reagents and conditions: (a) i: NaNO 2 , 37% HCl, 0 • C; ii: SnCl 2 , 0 • C to rt; (b) dimethylamine (2 M in  Arylhydrazines 3 and 4 were prepared from anilines 1 and 2 by diazotization in hydrochloric acid (37%), followed by the reduction of the diazonium salts using tin(II) chloride according to the reaction protocols known in the literature [24][25][26]. Compound 5 was prepared to start from 1-[4-(3-bromopropyl)phenyl]ethanone and dimethylamine (2 M in methanol). Compound 5 was then condensed with 3-4 to provide hydrazone derivatives 6-7. By the double addition of 28], the desired cyclized aldehydes [8][9] were obtained. Starting from derivatives 8-9, two routes were used to afford the final compounds 10-11, 14-15 with [(N,N-dimethyl)amino]methyl or [(N,N-dimethyl)amino]propyl substituents at the 4-position of the pyrazole scaffold. The first route was a reductive amination of 8-9 with sodium triacetoxyborohydride, dimethylamine, and acetic acid in DCE to give compounds [10][11]. The second route, in two steps, was a Horner-Wadsworth-Emmons olefination of 8-9 with sodium hydride, diethylcyanomethylphosphonate in THF, followed by catalytic hydrogenation of the intermediates in the presence of Raney nickel and palladium on carbon in methanol saturated with ammonia under a hydrogen atmosphere, to afford compounds 12-13. The last step was the reductive methylation of derivatives in methanol with formaldehyde, and sodium triacetoxyborohydride to afford the corresponding compounds 14-15. Starting from compounds 10-11 and 14-15, final compounds 24-35 were prepared in three steps (Scheme 2).
For the synthesis of the second series with a methyl group in the 4-position of the pyrazole scaffold (compounds 61-72), we first describe the preparation of intermediates 44-45 (Scheme 3). Reduction of the ester group of compounds 10-11, [14][15] with lithium aluminum hydride in THF gave derivatives [16][17][18][19], which were then oxidized with manganese oxide in chloroform to the corresponding compounds 20-23 (Scheme 2). Subsequently, target compounds 24-35 were prepared by reductive amination of 20-23 with the appropriate amine in the presence of sodium triacetoxyborohydride and acetic acid in DCE.
For the synthesis of the second series with a methyl group in the 4-position of the pyrazole scaffold (compounds 61-72), we first describe the preparation of intermediates 44-45 (Scheme 3). Compound 4 and 4-or 3-cyanoacetophenone (36-37) were condensed in methanol to provide hydrazone derivatives 38-39, which were then treated with Vilsmeier-Haack reagent with POCl3 in DMF to afford pyrazoles 40-41. Selective reduction of the aldehyde functions (40-41) with triethylsilane and trifluoroacetic acid gave nitrile derivatives 42-43, which were reduced with the borane-THF complex in THF to afford compounds 44-45. Reduction of the ester group of compounds 10-11, [14][15] with lithium aluminum hydride in THF gave derivatives [16][17][18][19], which were then oxidized with manganese oxide in chloroform to the corresponding compounds 20-23 (Scheme 2). Subsequently, target compounds 24-35 were prepared by reductive amination of 20-23 with the appropriate amine in the presence of sodium triacetoxyborohydride and acetic acid in DCE.
Compounds 50-51, with a [(N,N-dimethyl)amino]methyl group, R 2 , were synthesized in three steps starting from derivatives 44-45, which were engaged in the reductive methylation with formaldehyde, sodium triacetoxyborohydride in methanol to furnish compounds 46-47. Using the previously described method in Scheme 2, the reduction of the ester group of compounds 46-47 with lithium aluminum hydride in THF gave derivatives 48-49, which were then oxidized with manganese oxide in chloroform to afford aldehydes 50-51. fragments (sAPPα and sAPPβ).  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38,   [24][25][26][27][28][29][30][31][32][33][34][35] fragments (sAPPα and sAPPβ).  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, .
compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, .

16,17,24-35
to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, .

R1
Position of the R2 Group αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, .  αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The  results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . αCTFs and AICD (C20) when compared to the control condition was ca compounds were also evaluated for their ability to modulate the secretio fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and ) after tr reference and tested compounds were measured in the cell media by ELI results are expressed as IC50 values, which correspond to the concent compound that inhibits  concentration by 50% (either  or Aβ1to the concentration of  in non-treated SY5Y-APP WT cells. Selecte were then further evaluated for their ability to decrease the level of peptides (Aβx-38, .

16,17,24-35
2.4 ± 0.2 3.2 ± 0.8 2.0 3.5 11 63 --αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ).  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The  results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, .  αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . αCTFs and AICD (C20) when compared to the control condition was calculated. Selected compounds were also evaluated for their ability to modulate the secretion of soluble APP fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits  concentration by 50% (either  or  in comparison to the concentration of  in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38,  fragments (sAPPα and sAPPβ). Aβ1-x levels (Aβ1-40 and  after treatment with the reference and tested compounds were measured in the cell media by ELISA ( Table 1). The results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38, . results are expressed as IC50 values, which correspond to the concentration of a given compound that inhibits Aβ1-x concentration by 50% (either  or  in comparison to the concentration of Aβ1-x in non-treated SY5Y-APP WT cells. Selected hit compounds were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38,   --were then further evaluated for their ability to decrease the level of N-truncated Aβ peptides (Aβx-38,      or Aβ  peptide secretion in SY5Y cells. IC 50 values are expressed as the mean ± SD of at least three experiments performed in triplicate. b Mean values calculated based on at least three independent experiments with less than 10% deviation. c C 20 indicates the concentration of compound necessary to increase by 20-fold the quantity of APP fragments compared to the untreated control conditions. d Compound concentration causing 50% of cell death after 72 h treatment. e [14]. f [22]. nd: not determined. For the first series of compounds [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35], different modulations were applied with the R 1 group, in the meta-or para-position of the phenyl ring, as hydroxymethyl, [(N,Ndimethyl)amino]methyl or [(4-methyl)piperazin1 -yl]methyl or [3-(dimethylamino) propylmethyl-amino]methyl. Intermediates 16 and 17, with a hydroxyl substituent as the R 1 group, were tested to evaluate the influence of the presence of an additional positive charge or not. Substitution with an R 1 hydroxyl group led to an increase in cell toxicity (Table 1), whereas substitution with a [(N, N-dimethyl)amino]methyl or [(4-methyl)piperazin1-yl]methyl or [3-(dimethylamino) propyl-methyl-amino]methyl group reduced the cell toxicity to concentrations above 100 µM. First, when considering the effect on the Aβ 1-X levels, the variation in the position of the R 1 group gave a micromolar IC 50 close to reference A1 and lower than reference B1 and CQ for compounds 24-31 (Table 1). The compounds with [(N,N-dimethyl)amino]methyl or [(4-methyl)piperazin1 -yl]methyl, as the R 1 group, showed the best efficacy toward Aβ  and Aβ  (Aβ  : IC 50 = 1. .0 µM and Aβ  : IC 50 = 2. 4-5.8 µM) compared to derivatives with [3-(dimethylamino)propylmethyl-amino]methyl substitution 32-35 and benzylic alcohols 16 and 17 (Aβ  : IC 50 = 4. 2-12.4 µM and Aβ  : IC 50 = 5. 4-13.2 µM). (Table 1). Second, αCTFs and AICD C 20 were evaluated and shown to be lower for compounds 24-32 than for 33-35 and 16-17, showing on average a C 20 also lower than the reference A1, B1, and CQ (Table 1), together suggesting that the increased stability of CTFα and AICD is indicative of a gain in lysosomotropic activity, similar to CQ (C 20 for CTFα and AICD over 10 µM). In general, the para position yields a better reduction of Aβ 1-x concentrations in contrast to the little impact of CTFα and AICD C 20 . Considering the pyrazole ring, a shorter chain (n = 1) yielded lower IC 50 and C 20 than a longer chain (n = 3), especially when R 1 is the [3-(dimethylamino)propyl-methyl-amino]methyl group. Compounds 24-35 present very low cytotoxicity (CC 50 > 100 µM). In this first series of compounds, the replacement of the R 1 amino chain with an uncharged CH 2 OH group yielded a significant increase in cytotoxicity and a low increase in IC 50 for Aβ secretion, while αCTF and AICD C 20 remained in the same range. This result underlines the importance of the presence of a positive charge in this N1 phenyl ring.
For the second series of compounds 61-72, different modulations were performed within the R 1 group at the meta-position of the N1 phenyl ring as well as the variation in the position (meta and para) of the R 2 group as [(N,N-dimethyl)amino]methyl or [3-(dimethylamino) propyl-methyl-amino]methyl, and the introduction of a methyl group in the 4-position of the pyrazole scaffold, leading to very different results in terms of the cytotoxicity (CC 50 between 2 and >100 µM). In general, compounds of this second series are more cytotoxic than those of the first, except for compounds 64-65 and 57, which had a cytotoxicity CC 50 above 100 µM, which is less toxic than the CQ CC 50 in our conditions (CQ CC 50 = 30 µM). On the other hand, the IC 50 of Aβ secretion was homogeneous regardless of whether it was R 1 or R 2 and its position (Aβ  : IC 50 = 1. 2-3.9 µM and Aβ  : IC 50 = 1. 6-4.1 µM). C 20 values for αCTF and AICD were homogeneous or even slightly lower than the first series of compounds, with compounds 69, 71, and 72 showing the best C 20 values between 600 nM and 1 µM. In this series, the nature of the R 1 group did not lead to any significant difference in APP metabolism. In contrast, the replacement of the R 1 alkylamino chain with an uncharged CH 2 OH group yielded a significant decrease in activity (IC 50 for Aβ secretion higher than 10 µM for 48, αCTF higher than 10 µM for 57). The presence of a diamino group such as R 2 led to slightly improved values even if insignificantly different.
A selection of compounds with several selected activities was performed including sAPPα and sAPPβ, which were further explored. We selected compounds 67, 69, and 71 as the most efficient compounds decreasing Aβ production (IC 50 around 1 µM) and increasing CTFα and AICD (C 20 around 1 µM) but with various cytotoxic CC 50 . As a comparison, we also selected compounds 33 and 34 from the first series with lower efficacy in repressing Aβ 1-x production (IC 50 comprised between 6 to 9 µM) and various effects on the CTFα and AICD expression (C 20 around between 1 and 6 µM). The decrease in Aβ could result from a decrease in either βor γ-secretase cleavage or an increase in the α-secretase cleavage of APP. Thus, medium concentrations of sAPPβ and sAPPα, which are the βand α-secretase extramembrane shed APP fragments, were determined after cell drug treatments. All compounds were shown to decrease sAPPβ efficiently (Table 2). Interestingly, the greater the effect of the compounds on the repression of Aβ 1-x , the lower the media concentrations of sAPPβ. The IC 50 values of sAPPβ were thus greater for compounds 33 and 34 than for 67, 69, and 71 (Table 2). Conversely, compounds 67 and 71 were able to significantly increase the sAPPα concentrations (C 1.5 around 1 µM), whereas the three other compounds did not show an increase in APPsα up to 10 µM (5 µM for compound 69). These results strongly suggest that these compounds, in particular compounds 67 and 71, reduce Aβ 1-x peptide secretion by repressing the β-secretase cleavage of APP while increasing the secretion of sAPPα, as suggested for other compounds from a previous family [22].

Effect of Compounds on Autophagic Flux and Lysosomal Degradation Pathways
Previous work dedicated to compounds of family C [22] has underlined the effect of our compounds on autophagic flux and the lysosomal degradation pathways. More precisely, the role of the amino side chains and their number was highlighted. Therefore, we assessed whether the activity of compounds involved the modulation of lysosomal activities toward the accumulation of APP-CTFs and AICD. Thus, we studied the effect of a selection of compounds on the autophagic flux, more particularly, we evaluated the effect of compounds 24, 33, 34, 57, 65, 67, 70, and 71 on two markers associated with autophagy: p62 and LC3-I lipidation into LC3-II. To the previous selection of efficient compounds decreasing Aβ 1-x production and increasing CTFα and AICD, we added compound 65 with no cytotoxicity at 100 µM. As a comparison, we also selected compound 57 with a hydroxymethyl substituent as a control. In addition to CQ, bafilomycin A, a well-known inhibitor of autophagic flux, was used as a control. Treatment with compounds 67 and 71 induced an increase in p62 expression by 2.5 and 4-fold, respectively, compared to the control condition ( Figure 3). LC3-I lipidation into LC3-II was also increased by 5.2 and 6. 5-fold, respectively. An increase in the two markers was also observed with compound 57, but to a smaller extent (1.7-fold for p62 and 2-fold for LC3-I/LC3-II). In contrast, compound 34 showed no significant effect on these two autophagy markers.  --were then further evaluated for their ability to decrease the level of N-trunc peptides (Aβx-38, .   --were then further evaluated for their ability to decrease the level of N-trunc peptides (Aβx-38, . Pyrazole compounds are derived from a pharmacophore of structural superposition of families A and B of molecules themselves derived from CQ and AQ. Both CQ and AQ share alkalizing and lysosomotropic activities related to their accumulation in cell vesicular acidic compartments and polynitrogen protonability, making them weak bases. This latter chemical property could contribute to the understanding of structure-activity differences between compounds 34, 57, 67, and 71, associated with APP metabolism and autophagy flux modulatory function. We addressed this question by calculating the pKa of each nitrogen of compounds 34, 57, 67, and 71 [29]. Compound 71 has four protonable nitrogens, although the second nitrogen of the R 1 piperazine moiety has a low pKa. Three protonable nitrogens were found in compounds 34 and 67, whereas compound 57 only had two. No relationship was observed between the pKa of compounds with either APP metabolism, the autophagy flux, or cell cytotoxicity.
We next assessed the potential relationship between the autophagy flux and Aβ 1-x production or the compound's cytotoxicity.
While considering the compounds' cytotoxicity, p62 expression was best correlated with this cytotoxicity while LC3 or the p62/LC3 ratio was not. Therefore, for the selection of lead compounds, these parameters of p62/LC3 ratio and p62 expression should be considered with regard to the Aβ 1-x production inhibitory activity and cell cytotoxicity. Accordingly, compound 24 from the first series and compounds 65 or 70 from the second appeared as having the selected pharmacological properties. Noticeably, these selected molecules had similar CC 20 (µM) activities toward APP metabolism and αCTF and AICD expression. , and β-actin. Statistical significance was set at * p < 0.05, ** p < 0.001, *** p < 0.001, **** p < 0.0001, and ns as not significant.
We next assessed the potential relationship between the autophagy flux and Aβ1-x production or the compound's cytotoxicity.
While considering the compounds' cytotoxicity, p62 expression was best correlated with this cytotoxicity while LC3 or the p62/LC3 ratio was not. Therefore, for the selection of lead compounds, these parameters of p62/LC3 ratio and p62 expression should be considered with regard to the Aβ1-x production inhibitory activity and cell cytotoxicity. Accordingly, compound 24 from the first series and compounds 65 or 70 from the second appeared as having the selected pharmacological properties. Noticeably, these selected molecules had similar CC20 (µM) activities toward APP metabolism and αCTF and AICD expression.
Compounds from both series were selected to first determine the potential relationship between the repression of Aβ1-x production and the increased expression of LC3 and p62. Noticeably, LC3 and p62 were not modulated within similar amplitudes ( Figure 3). For instance, p62 was increased by 1.5-fold while LC3 remained unchanged for compound 33, whereas LC3 was increased by 6-fold and p62 by 3-fold by compound 70. We then also considered the ratio between the p62 and LC3 expression levels. While considering LC3 or p62 separately, the repression of Aβ1-x was neither associated with LC3 nor the sole expression of p62. In sharp contrast, when considering the p62/LC3 ratio, the lower the ratio, the higher the repression of Aβ1-x. Both compounds 34 and 35, with a ratio close to 1, had the lowest inhibitory activity (Table 3). Compounds 70 and 71, having a p62/LC3 ratio of 0.30 and 0.34 had the strongest Aβ1-x production inhibition (Table 3). These ratios are comparable with those of chloroquine and bafilomycin A1, both of which  (20 µM), and bafilomycin A1 (BafA1) at 100 nM as the control. Cell lysates were immunolabeled with the following antibodies: p62 (a), LC3 (b), and β-actin. Statistical significance was set at * p < 0.05, ** p < 0.001, *** p < 0.001, **** p < 0.0001, and ns as not significant.
Compounds from both series were selected to first determine the potential relationship between the repression of Aβ 1-x production and the increased expression of LC3 and p62. Noticeably, LC3 and p62 were not modulated within similar amplitudes ( Figure 3). For instance, p62 was increased by 1.5-fold while LC3 remained unchanged for compound 33, whereas LC3 was increased by 6-fold and p62 by 3-fold by compound 70. We then also considered the ratio between the p62 and LC3 expression levels. While considering LC3 or p62 separately, the repression of Aβ 1-x was neither associated with LC3 nor the sole expression of p62. In sharp contrast, when considering the p62/LC3 ratio, the lower the ratio, the higher the repression of Aβ 1-x . Both compounds 34 and 35, with a ratio close to 1, had the lowest inhibitory activity (Table 3). Compounds 70 and 71, having a p62/LC3 ratio of 0.30 and 0.34 had the strongest Aβ 1-x production inhibition (Table 3). These ratios are comparable with those of chloroquine and bafilomycin A1, both of which are well-known to repress Aβ production. This effect appeared to be independent of the p62 or LC3 expression levels, for instance, compound 24 had a p62/LC3 ratio of 0.49 whereas the LC3 and p62 expression levels were both lower when compared to chloroquine, bafilomycin A1, or compounds 70 and 71.  Fold-change. c Compound concentration inhibiting 50% (IC 50 ) of Aβ  or Aβ

Discussion
Previous studies in our laboratory underlined the effect of compounds derived from chloroquine CQ and amodiaquine AQ, two antimalarial compounds with lysosomotropic activities that are able to modulate the amyloid and Tau pathologies, both of which are the two major pathophysiological processes of AD. Two families of compounds (A and B, Figure 2) have been developed and used for a ligand-based approach. In a first study dedicated to the development of biaryl compounds (family C, Figure 2), we identified ), a compound able to inhibit Aβ 1-x peptide production without modifying Aβ x-40/42 and with little modulatory activity of the expression of αCTF and AICD when compared to the effect of chloroquine. This selective modulatory effect toward Aβ 1-x reduced the production, and maintenance of the global Aβ x-40/42 levels has been reported consequently to BACE1 overexpression or rare inherited mutations of APP, suggested by a modification of the enzyme/substrate recognition [30,31]. In vivo, , having reduced lysosomotropic activity, diminished the neurofibrillary degenerating process together with the short-term spatial memory in the Thy-Tau22 transgenic mouse model [23], and reduced amyloid burden in APPxPS1 transgenic mice model (to be published).
In the present study, a second family of compounds based on the same computer-assisted pharmacophoric design was synthesized, and 30 compounds around a diphenylpyrazole scaffold substituted with amino side chains were evaluated (D, Figure 2). One or two (dialkylamino)alkyl side chains were introduced at the 4-position of the pyrazole ring and/or the phenyl rings. Two series of compounds were then obtained with different (dialkylamino)alkyl or (heterocyclylamino)alkyl groups. The first series contained between two and four amino groups while the second had from one to four amino groups. Compounds were assumed to repress Aβ 1-x production, stabilize αCTF and AICD expression, and modulate autophagy through an increased expression of LC3-I/LC3-II and p62. Accordingly, all of these parameters as well as the pKa of compounds were therefore assessed. These parameters are shared by lysosomotropic drugs such as CQ, a weak di-base, or bafilomycin A1 (BafA1), which is an H + -ATPase proton-pump vacuolar inhibitor. Both CQ and bafilomycin have lysomotropic activity mediated by the alkalinization of cell-acidic compartments, leading to lysosome and autophagy-flux inhibition, and are, in addition, both toxic at µM concentrations [11,12]. APP metabolism is modified by this lysosomotropic activity with two consequences: (1) the repression of Aβ peptide production through indirect or non-competitive inhibition of BACE1, the β-secretase, and (2) the stabilization of both αCTF and AICD, where proteolysis occurs in lysosomes. Interestingly, in the previous series,  was selected because Aβ 1-x production was repressed without significant modifications in the CTFα or AICD quantities [22,23]. However,  was the sole compound showing both properties, suggesting that the non-competitive β-secretase inhibition differs from the lysosomotropic activities of CQ and BafA1 and that  activity occurs through a yet undetermined mechanism. Importantly, this unknown mechanism is unrelated to the alkalizing or lysosomotropic property of CQ or BafA1.
The lysosomotropic activity of CQ is mostly related to the internalization and accumulation of CQ in endolysosomal vesicles, in which this weak base alkalinizes the luminal content of endolysosomes. This intravesicular pH modification is then likely to be responsible for the non-competitive and indirect inhibition of β-secretase aspartyl protease activity since BACE1 activity is pH-dependent and optimal at acidic pH [32]. In the previous series of compounds, the loss of a unique basic group of the triamino compound  (X = CH instead of X = N, Figure 2) modified the pKas of the compound, possibly associated with the loss of the alkalizing property of the corresponding diamino compound. These compounds both enhanced αCTF and AICD expression to a greater extent than CQ [22]. In sharp contrast, all compounds of the present series had lower C 20 than CQ, although this was non-related to their pKa and therefore, not related to the weak base property of these compounds. All pyrazole-derived compounds were more efficient to repress Aβ 1-x production with an IC 50 inhibitory activity comprised between 1.2 and 12.4 µM and 1.7 to 13.2 µM for Aβ  or Aβ  , respectively. Within these same cell-based assays, CQ has an IC 50 of 7 and 12.7 µM for Aβ  or Aβ  , respectively. Together, these results suggest that this inhibitory activity is different from that of CQ or even BafA1 [11,12], and is most likely not mediated by the weak base property. Moreover, several compounds including those having the greatest Aβ 1-x inhibitory activity also repressed the expression of sAPPβ, which is produced following APP cleavage by the β-secretase, suggesting that these pyrazole-derived compounds are also potent indirect or non-competitive inhibitors of β-secretase.
From the previous families of compounds,  and RPEL [17] were shown to reduce both amyloid and neurofibrillary degeneration in vivo, suggesting that our compounds are also effective. Although modulation of the β-secretase activity has been suggested to also modulate Tau protein expression [33,34], the proteinaceous component of neurofibrillary degeneration, there is no direct evidence of the contribution of either BACE1 or BACE2 to Tau metabolism. Tau protein is a long half-life protein whose degradation is principally mediated by chaperone-mediated autophagy and more recently, intracellular Tau aggregate clearance was shown to occur through aggrephagy [35]. Compounds from previous families including  were shown to repress Tau aggregation and to modulate p62 and LC3-I/LC3-II, two master regulators of autophagy, also implicated in the aggrephagy. Increased expression of p62 and LC3-I/LC3-II following CQ or BafA1 treatment induced the accumulation of these proteins due to the blockade of the autophagy flux and lysosome activity. Herein, we showed that the inhibitory effect of our compounds toward Aβ 1-x expression was neither associated with p62 nor LC3-I/LC3-II. The noncompetitive β-secretase effect is more related to the expression ratio of p62/(LCI3)/LC3II), in which individual expression is modified similarly between compounds of the present series. More precisely, p62 expression was less stimulated by compounds of family A than by compounds of family B. However, as for CQ and BafA1, the accumulation of p62 over 1.5-fold enhanced the cytotoxicity of the compounds similar to the loss of a protonable nitrogen in family A. Taking into account that all families are derived from a unique pharmacophore, itself derived from the structure-activity relationship of the most efficient compounds of family A [13,14] and family B [18,19], suggests a common mechanism for which the criteria of selection relies on (1) the repression of Aβ 1-x production without affecting the Aβ x-38/40/42 release; (2) a decrease ratio of p62/(LC3-I/LC3-II) expression below 0.7; and (3) an accumulation of p62 below 2.0-fold. The accumulation of APP metabolites αCTF and AICD by our compounds appears more likely to be unrelated to an alkalizing or lysosomotropic activity.
In conclusion, according to the present parameters of lead compound selection, compounds 24, 65, and 70 are those with comparable non-competitive β-secretase activity, low toxicity, and reduced lysosomotropic activity. Moreover, these common properties for several of our families of compounds suggest a similar mechanism of action that is likely to be mediated through a limited number, if not a single target interaction, and modulatory activity. However, target identification and the mechanism of action remain to be elucidated. This study is also bringing simple biochemical and biological parameters that could be useful for selecting drugs, which we showed to be active against both the pathophysiological processes of AD. The development of quantification methods of these parameters for high throughput screening would then enable the testing of already existing FDA-approved drugs and potentially help to decipher the mechanism of action.

Chemistry
All commercial reagents and solvents were used without further purification. Organic layers obtained after the extraction of aqueous solutions were dried over MgSO 4 and filtered before evaporation. Reaction yields were not optimized. Column chromatography was performed using Macherey-Nagel silica gel (230-400 mesh). 1 H and 13 C NMR spectra were obtained using a Bruker DRX 300 spectrometer (Division BioSpin, Wissembourg, France) operating at 300.13 MHz for proton, operating at 300 MHz for 1 H and 75 MHz for 13 C, equipped with a BBFO 5 mm probe and a sample XpressLite. The data were processed using software TOPSPIN 4. Chemical shifts (δ) were expressed in ppm relative to either TMS or the residual proton signal in deuterated solvents. Mass spectra were recorded with an LC-MS (Waters Alliance Micromass ZQ 2000, Waters Corporation, Milford, MA, USA) using electrospray ionization. The purity of the final compounds was verified by two types of high-pressure liquid chromatography (HPLC) columns: C18 Interchrom UPTISPHERE and C4 Interchrom UPTISPHERE. Analytical HPLC was performed on a Shimadzu LC-2010AHT system equipped with a UV detector set at 254 nm and 215 nm. The following eluent systems were used: buffer A (H 2 O/TFA, 100:0.1) and buffer B (CH 3 CN/H 2 O/TFA, 80:20:0.1). Compounds were dissolved in 50 µL of buffer B and 950 µL of buffer A and injected into the system. HPLC retention times (HPLC t R ) were obtained at a flow rate of 0.2 mL/min using a gradient run from 100% of buffer A to 100% of buffer B over 30 min. The spectra and chromatograms of final compounds can be found in Supplementary Materials. 4.1.1. General Procedure A Dimethylformamide (81.5 mmol) was cooled to 0 • C with a salt/ice bath. Phosphorus oxychloride (22.5 mmol) was added dropwise with the temperature maintained below 0 • C. The mixture was then stirred for 40 min at 0 • C. Hydrazone (5.39 mmol) was added and the reaction mixture was allowed to warm to room temperature. After 2 h, the temperature was increased to 50 • C. The reaction was stirred at this temperature for 4 h. The mixture was then added to crushed ice and stirred for 1 h. Potassium carbonate was added until pH = 8 and the mixture was extracted twice with methylene chloride. The combined organic layers were washed with brine, dried, and evaporated. The residue was purified by column chromatography (DCM/MeOH/NH 4 OH = 9:1:0.1).

General Procedure B
To a solution of alkylamine (3.73 mmol), 37% formaldehyde in water (22.4 mmol) and acetic acid (22.4 mmol) in methanol (20 mL) was slowly added sodium triacetoxyborohydride (18.7 mmol) over 30 min. The mixture was stirred until the completion of the reaction. The solution was an aqueous carbonate potassium (10%) and ethyl acetate was added to the residue and the mixture stirred for 10 min. The layers were separated, and the aqueous layer was extracted twice with ethyl acetate. The combined organic layers were washed with brine, dried, and evaporated. The residue was purified by column chromatography (DCM/MeOH-NH 3 sat = 95:5).

General Procedure C
LiAlH 4 (1 M in THF,2.82 mmol) was added to 20 mL of anhydrous THF under nitrogen. The solution was cooled to 0 • C with an ice bath and a solution of ester (1.88 mmol) in 20 mL of anhydrous THF was added dropwise. The reaction was stirred at 0 • C for 20 min and rt for 1 h. The mixture was then cooled with an ice bath and 0.11 mL of H 2 O was added, followed by 0.11 mL of 15% NaOH and 0.33 mL of H 2 O. The solid was isolated by filtration and washed with THF. The filtrate was evaporated. The residue was purified by column chromatography (DCM/MeOH/NH 4 OH = 9:1:0.1).

General Procedure D
To a solution of the desired compound (1.42 mmol) in chloroform (25 mL) was added manganese(IV) oxide (14.2 mmol). This was stirred at rt for 24 h. More manganese(IV) oxide (14.2 mmol) was added, and the mixture was stirred for 24 h. The solid was filtered off and washed with methylene chloride. The filtrate was washed with 10% K 2 CO 3 and with brine. The organic layer was dried and evaporated to give the aldehyde that was used in the next step without further purification.