Orally Bioavailable Prodrugs of ψ-GSH: A Potential Treatment for Alzheimer’s Disease

Abstract

Addressing glycation-induced oxidative stress in Alzheimer’s disease (AD) is an emerging pharmacotherapeutic strategy. Restoration of the brain glyoxalase enzyme system that neutralizes reactive dicarbonyls is one such approach. Toward this end, we designed, synthesized, and evaluated a γ-glutamyl transpeptidase-resistant glyoxalase substrate, ψ-GSH. Although mechanistically successful, the oral efficacy of ψ-GSH appeared as an area in need of improvement. Herein, we describe our rationale for the creation of prodrugs that mask the labile sulfhydryl group. In vitro and in vivo stability studies identified promising prodrugs that could deliver pharmacologically relevant brain levels of ψ-GSH. When administered orally to a mouse model generated by the intracerebroventricular injection of Aβ1–42, the compounds conferred cognitive benefits. Biochemical and histological examination confirmed their effects on neuroinflammation and oxidative stress. Collectively, we have identified orally efficacious prodrugs of ψ-GSH that are able to restore brain glyoxalase activity and mitigate inflammatory and oxidative pathology associated with AD.

Introduction

Advanced glycation endproducts (AGEs) are implicated in the mal-oxidative and inflammatory components of metabolic disorders such as diabetes, atherosclerosis, and a host of neurodegenerative conditions.^1,2 These products are the result of nonenzymatic reactions of glucose and other sugars with protein amino acids forming reversible Schiff bases.^3,4 Specifically, glucose metabolism-derived products such as reactive dicarbonyls react with the amino groups of biomolecules. The chemical rearrangement of these intermediates forms stable post-translational modifications that are termed AGEs.⁵ In addition to amino acids, specific nucleotides are also susceptible to AGE formation.⁶ These modifications are known to alter the secondary structure, and therefore, the physiological function of these biomolecules. Their pathology notwithstanding, AGEs have at least one known concrete physiological function: their association with the RAGE (Receptor for Advanced Glycation Endproducts) serves to trigger proinflammatory processes (which include oxidative stress components) that form a component of immune cell recruitment through cytokine release.^7,8 However, an overabundance of this process drives disease pathology.

Chronic hyperglycemia elevates the level of reactive dicarbonyls and consequentially the levels of intra- and extracellular AGEs.⁹ It is associated with poor cognition. Empirically, it is a risk factor for Alzheimer’s and related dementias.^10,11 Insufficient ability to regulate blood glucose (glucose intolerance) is associated with the accumulation of pathologic β-amyloid plaques, neuroinflammation, oxidative stress, mitochondrial dysfunction, and ultimately neurodegeneration. Immunohistochemistry of senile plaques and neurofibrillary tangles show overabundance of AGEs in human and animal brain.^12,13 Glycation of Aβ1–42 enhances its toxicity and tendency to aggregate.¹⁴ Glycation of tau impairs the formation of paired helical tau filaments.¹⁵ AGE levels do increase with aging owing to decreased homeostatic efficiency.¹ Under normostasis, however, this is a slow process in comparison to the abject pathology seen in neurodegenerative disorders. It follows then that the entities able to impede AGE formation or can reverse their formation would have tremendous potential for the management of AD.

Among the dicarbonyls responsible for formation of AGEs, methylglyoxal (MEG) is a key highly reactive dicarbonyl. MEG is a product of an internal redox–dephosphorylation reaction sequence on dihydroxyacetone phosphate, which may either be spontaneous or enzymatic.¹⁶ It forms irreversible adducts with amino acids arginine and lysine, resulting in the formation of hydroimidazolone (MG-H1) and Nε-carboxyethyl-lysine (CEL), respectively.¹⁷ Under physiological conditions, MEG is detoxified by a conserved enzyme system, the ubiquitous glutathione (GSH)-dependent glyoxalase system (GLO), consisting of enzymes GLO1 and GLO2.¹⁸ Reduced brain GLO1 function has been associated with the buildup of AGEs and cause of neurodegeneration in AD.¹⁹ Expression of GLO1 is upregulated in the early and middle stages of AD,^20,21 hence the dysfunction of this enzyme system is likely due to the unavailability of the cofactor GSH. An imbalance between oxidized and reduced GSH, a function of the redox potential of the tissue, is further responsible for the progression of disease pathology. Although utilization of GSH or its precursor could seem a viable option, poor bioavailability of GSH and dysfunctional GSH catabolism in aging/diseased brain precludes such an approach. Our answer is to restore brain GLO1 activity by supplementation of a γ-glutamyl transpeptidase (GGT) stable analogue, ψ-GSH.²² Treatment of the transgenic APP/PS1 Alzheimer’s disease animal model with ψ-GSH before²³ and after initiation of AD pathology¹⁹ prevented and mitigated biochemical and cognitive damage caused by the disease.

Although the design of ψ-GSH addressed the metabolic instability of GSH around the γ-glutamyl peptide amide bond, the aliphatic thiol still remained an unaddressed liability. The short plasma half-life of ψ-GSH (0.56 h)²³ displayed oxidized ψ-GSH as the major metabolite. Even with such limitations, an acute high dose of oral ψ-GSH is able to alleviate hepatotoxicity induced by acetaminophen (Tylenol).²⁴ However, this model relies on the peripheral action of oral ψ-GSH, and thus, might not be translatable to the treatment of chronic conditions of the central nervous system such as Alzheimer’s disease. When tested in an AD transgenic mouse model at a dose and frequency found efficacious for intraperitoneal ψ-GSH treatment,²³ oral ψ-GSH resulted in modest improvement in cognitive function and biochemical markers of oxidative stress. In the current study, we describe the design and synthesis of prodrugs of ψ-GSH, the ultimate aim being to improve stability of the parent compound and deliver ψ-GSH at the target tissue, the brain. We have conducted cognitive behavioral analysis and biochemical characterization of prodrug-treated groups in an intracerebroventricular (i.c.v.) Aβ1–42-injected mouse model. We have identified ψ-GSH prodrugs that (a) present enhanced stability in biological tissues and plasma, (b) are able to convert to the parent compound ψ-GSH, and (c) are able to mitigate inflammatory and cognitive phenotypes presented by the AD mouse model.

Results and Discussion

Efficacy of Oral ψ-GSH in APP/PS1 Mice

In previous studies, we evaluated the efficacy of intraperitoneal ψ-GSH in 3–4 month²³ and 12-month-old¹⁹ cohorts of APP/PS1 mice. However, oral administration of ψ-GSH at 500 mg/kg dose in a preventative study (initiation of treatment in APP/PS1 mice at 3–4 months) was unsuccessful in conferring definitive cognitive benefits. The Morris water maze test was used to evaluate spatial working memory and recall abilities in these mice. In the case of the APP-PS1 mice treated orally with ψ-GSH, there was a trend toward higher spatial memory development and lower escape latencies when contrasted with the saline-treated group, with the trend reaching significance (one-way ANOVA, p < 0.05, Figure 1A) on day 2; however, this difference diminished in the following days. The path lengths traversed by the compound-treated and untreated APP-PS1 mice correlated well with the corresponding escape latencies (Figure 1B). The difference between path lengths traversed by APP-PS1 mice treated with oral ψ-GSH and the saline-treated controls was insignificant. A probe trial conducted 24 h after the last training trial showed limited improvement in the retention of learned tasks in the oral ψ-GSH-treated group (Figure 1E). Time spent in the target quadrant by untreated versus treated APP/PS1 mice was 16.39 ± 3.39 and 21.76 ± 5.36%, respectively, while the nontransgenic vehicle controls showed retention of the learned task, spending 36.67 ± 5.30% of the total time in the target quadrant. In the visible platform trial, the escape latency of the oral treatment group was similar to the saline-treated APP/PS1 mice (APP/PS1/saline: APP/PS1/ψ-GSH oral: WT/saline; 6.64 ± 1.58, 6.90 ± 1.33, 6.53 ± 0.75 s, Figure 1C). Path lengths (Figure 1D) and swimming speeds (data not shown) also followed this trend, indicating no motor defects or transgene-related visual impairment in these mice.

Figure 1.

Behavioral assessment and analysis of brain oxidative stress of APP/PS1 mice treated with saline (A/P saline) or oral ψ-GSH (A/P ψ-GSH) and age-matched nontransgenic mice (NTG) for 12 weeks. (A–E) Cognitive assessment was conducted using the Morris water maze test. Performance during the hidden platform training period (four trials per day for 4 consecutive days) was measured by mean escape latency (A) and the path length (B) required to locate the submerged platform. (C, D) Escape latency and the path length results of visible platform trials conducted at the end of the hidden platform training. (E) Retention of memory was assessed by a probe trial 24 h after the hidden platform training. Significant impairment was observed in A/P saline mice compared to the NTG saline group, while oral ψ-GSH displayed only moderate improvement in memory retention (expressed as percent time spent in the area that previously contained the platform). (F–J) Analysis of amyloid burden expressed as soluble (F) and insoluble (G) Aβ1–42 and oxidative stress markers such as protein carbonyls (H), TBARS (I), and reactive oxygen species (ROS, J). Marginal alleviation of oxidative stress and reduction in amyloid burden was observed in the oral ψ-GSH-treated group. Data are expressed as the mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.001).

Biochemical analysis followed a trend similar to that observed in the cognitive behavioral assessment of these mice. Amyloid deposition in the APP/PS1 mice is detected by 6 months of age. When administered intraperitoneally, ψ-GSH treatment was able to reduce Aβ1–42 in the brain cortex.²³ Quantitation of the brain amyloid load using the Aβ1–42-specific ELISA assay showed a modest decrease in insoluble Aβ1–42 in the APP/PS1 mice treated orally with ψ-GSH (90.80 ± 6.72% of vehicle-treated APP/PS1 mice; p > 0.05, Figure 1F). Immunohistochemical analysis of Aβ aggregates confirmed the findings of the ELISA assay and no significant reduction in amyloid plaque burden was detected (Supporting Information, Figure S1). Additionally, differences between levels of PBS-soluble Aβ were statistically insignificant between the untreated and ψ-GSH-treated mice (Figure 1G). Reduced GSH levels and redox potential are reported in the APP/PS1 mouse brain.^19,25 Evaluation of additional oxidative stress in the brain tissue was conducted by quantitation of the levels of protein carbonyls (Figure 1H), lipid peroxidation (Figure 1I), and total ROS species (Figure 1J). Contrary to the i.p. ψ-GSH treatment,²³ oral ψ-GSH was found ineffective in affecting these hallmark indicators of oxidative damage. These data indicate that the brain levels of active ψ-GSH attained after oral administration are ineffective in affecting the behavioral and biochemical consequences of AD pathology. Thus, the oral ψ-GSH treatment regimen needs to be either altered or a higher oral dose equivalent in brain accumulation to the i.p. ψ-GSH treatment should be administered. Given the high dose of ψ-GSH utilized in this study (500 mg/kg) and practical challenges in synthesizing quantities sufficient for long-term (∼3–4 month) efficacy studies, we focused our efforts on designing prodrugs of ψ-GSH that would be orally efficacious at an equivalent or lower dose.

Design and Synthesis

A major hurdle in designing effective GSH analogues is their poor bioavailability due to the rapid cleavage of the pseudoamide bond between γ-glutamic acid and cysteine.²⁶ We have addressed this metabolic liability by incorporation of a stable ureide linkage, without compromising its recognition by GSH-dependent biological systems.²⁷ However, the naked sulfhydryl group is susceptible to oxidative modifications, resulting in poor plasma stability. The classical approach to stabilize this thiol would be its acylation or carbamoylation (Figure 2). Compounds 1 and 2 contained acyl and ethyl thiocarbonate modifications. Another attempt to stabilize the thiol was its cyclization with one of the urea nitrogens through a thiocarbamoyl link, essentially forming an analogue of the 5-oxoproline residue. The latter is known also as pyroglutamic acid, an intermediate in GSH metabolism.²⁸ Analogue 3 could potentially get recognized by the ubiquitous enzyme 5-oxoprolinase liberating the parent compound ψ-GSH. Similarly, a reduced 5-membered thiazolidine analogue compound 4 was introduced, which could get hydrolyzed in an acidic or alkaline environment of the stomach or the intestines, liberating the active compound.²⁹ Another classical approach involved dicyclopentyl esterification of ψ-GSH, since this ester previously produced promising results in biological testing.³⁰ Simultaneous protection of thiol and carboxylic acids or the free amino functional group was also attempted by the design of compounds 6 and 7. Similar strategies have been employed for the formulation of prodrugs of GSH.³⁰ S-Acylation and carboxylic acid esterification, yielding compounds 1, 2, 5, and 6, as well as the inclusion of cysteine prodrugs such as l-thiazolidine-4-carboxylic acid (as in prodrug 4) or l-2-oxothiazolidine-4-carboxylate (as in prodrug 3), have been described as promising approaches³⁰⁻³² for the design of GSH prodrugs. The symmetrical disulfide of ψ-GSH or unsymmetrical disulfide prodrugs could be useful. These were not considered in our prodrug design due to increased oxidative stress existent in the AD-afflicted brain, limiting the availability of endogenous reductants that can render the prodrugs into active species.

Figure 2.

Chemical structures of ψ-GSH and its prodrugs.

Chemical reactions for the synthesis of ψ-GSH prodrugs are displayed in Scheme 1. The synthesis of the parent compound ψ-GSH is described in our previous publication.²² We opted for S-trityl (STrt)-protected cysteine in place of the SStBu protecting group to prevent liberation of large quantities of the infamous stench of butyl mercaptan. Deprotection of the trityl group was achieved under acidic conditions (TIPS, TFA) to obtain N-Boc and t-butyl ester-protected ψ-GSH with free thiol (8, Supporting Information). Acetylation of thiol 8 using acetic anhydride, followed by TFA-mediated global deprotection of N-Boc and tBu groups afforded S-acetyl-ψ-GSH (1). Similarly, the reaction of 8 with ethyl chloroformate and global deprotection resulted in the thiocarbonate prodrug 2. Thermal cyclization of 8 using carbonyl diimidazole, followed by acidic deprotection resulted in the formation of the cyclic thiocarbamate prodrug 3. Traditional thionyl chloride-mediated diesterification afforded the dicyclopentyl ester prodrug 5, which upon reaction with ethyl chloroformate and acidic deprotection resulted in compound 6. Global acidic deprotection and the reaction of the resulting ψ-GSH with ethyl chloroformate resulted in the N- and S-protected prodrug 7.

Scheme 1. Synthesis of Prodrugs of ψ-GSH Compounds 1–3 and 5–7.

Synthesis of prodrug 4 began with commercially available l-4-thiazolidinecarboxylic acid (12, Scheme 2). After N-Fmoc protection, compound 12 was subjected to coupling with glycine-t-butyl ester affording the dipeptide 13. Deprotection of the Fmoc group and formation of the ureide linkage of the resulting amine 14 with N-Boc-Dap-OAll (19) resulted in the tripeptide 15. Synthesis of the protected 2,3-diaminopropionic acid (Dap) intermediate was performed using a known procedure beginning with N-Boc-asparagine.³³ Sequential deprotection of the allyl ester and N-Boc, t-butyl ester groups of the tripeptide 15 afforded the desired thiazolidine-containing prodrug 4.

Scheme 2. Synthesis of the ψ-GSH Prodrug 4.

Stability Evaluation of ψ-GSH Prodrugs and Conversion to ψ-GSH

The key factors that we considered for the selection of an optimal prodrug are (a) conversion into the parent compound ψ-GSH in target tissues; (b) better stability than ψ-GSH along the gastrointestinal tract for better absorption; and (c) improved plasma stability to enhance distribution to the target organ (brain) via systemic circulation.

We studied the conversion of synthesized ψ-GSH prodrugs into ψ-GSH in freshly prepared brain homogenates containing proteinase inhibitors. Except for 3 and 4, all of the prodrugs resulted in the unmasking of thiol protection liberating the parent compound (Figure 3A). Similar results were obtained from the incubation of these prodrugs with liver homogenates (Figure 3B). Thus, prodrugs 3 and 4 were not considered further. Conversion rates were higher in the liver homogenate compared to the brain homogenate, which significantly impacted the tissue stability of the prodrugs (Table 1). However, the resultant highest levels of ψ-GSH attained from incubation of prodrugs were similar to that by ψ-GSH itself in brain and liver homogenates (Figure 3A and B). Compounds 5, 6, and 7 displayed a relatively higher and sustained release of ψ-GSH in both tissue homogenates. The diester prodrug 5 exhibited marginally better stability in the brain homogenate over ψ-GSH itself; however, it fared similar to compound 7 in the liver homogenate. Prodrug 6 fared similar to compound 5 for stability in the brain homogenate and displayed relatively poor stability in the liver homogenate. Analysis of the liver homogenate incubates of compound 6 identified the corresponding thiocarbonate cleavage product as the major metabolite. However, the compound was still far superior to ψ-GSH in in vitro stability.

Figure 3.

(A, B) Conversion of ψ-GSH prodrugs into ψ-GSH examined by incubation with tissue homogenates. The presence of ψ-GSH was detected in the brain (A) and liver (B) incubates of prodrugs 1, 5, and 7. Data are normalized to the highest concentration of ψ-GSH detected in each incubation reaction. No significant differences in the highest ψ-GSH levels were detected among all prodrugs tested. (C, D) In vivo conversion of the lead prodrugs 5 and 7 into ψ-GSH after oral administration in CD1 mice at 250 mg/kg dose in comparison with a therapeutically effective dose of ψ-GSH (500 mg/kg, i.p.). Plasma levels of prodrugs (C) are consistently higher than those achieved by ψ-GSH itself at the time points tested. Prodrug treatment resulted in higher plasma concentrations of ψ-GSH when compared to the levels achieved by i.p. ψ-GSH. A similar increase in the levels of prodrugs and ψ-GSH was also observed in the brain tissues of these mice (D).

Table 1. Half-Life (h) of ψ-GSH and Synthetic Prodrugs in InVitro Biological Systems.

	GI tract			tissue homogenate
prodrug	gastric phase	intestinal phase	plasma	brain	liver
ψ-GSH	28.75	1.28	0.23	1.51	0.02
1	28.88	5.02	5.80	0.74	0.02
2	27.70	38.50	47.14	2.72	0.58
3			25.57		177.69
4			30.64		115.5
5	4.87	5.24	72.20	7.74	2.20
6			61.32	5.64	0.47
7	17.96	24.24	113.6	9.79	1.30

It is imperative to determine the stability of these prodrugs in gastric and intestinal fluids prior to intestinal membrane permeation (Table 1). Stability in the gastrointestinal tract may be confirmed by incubating the prodrug in gastric and intestinal fluids representative of in vivo drug exposure to these fluids. An ideal prodrug should be stable within the transmit period for the stomach and intestine, which is 1 h in simulated gastric fluids (SGF) and 3 h in simulated intestinal fluids (SIF), respectively.³⁴Table 1 shows the results of the SGF and SIF experiments. The half-life of thiocabonate prodrugs 2, 7, and the S-acyl prodrug 1 was ≈24 h in SGF, while the acidic hydrolysis of the ester in 5 led to reduced stability of this prodrug. A marked improvement in intestinal stability was observed for the S-carbonate prodrugs over other synthetic prodrugs and over the parent compound ψ-GSH. Similar observations were made from the plasma stability studies, and compound 7 was determined to be an optimal candidate to advance for efficacy evaluation in an i.c.v. Aβ1–42-injected acute AD model. It is important to note that compound 7 is converted to prodrug 2 as an intermediate and that has an impact on the half-life calculated here. Because of the similarities in compounds 2 and 7, and the ease of synthesis of prodrug 7, it was advanced further for stability characterization using tissue homogenates. Compound 6 showed marginal improvement over the structurally similar analogue 5 and did not offer any advantage in plasma and liver homogenate experiments, and thus was not considered for further evaluation. To gauge the advantage of prodrug 7 and improved plasma stability conferred by the diester prodrug 5, both prodrugs were evaluated in our efficacy study. It would also be beneficial to study prodrug 6 given that it incorporates beneficial chemical properties of both, compounds 5 and 7. In this study, however, we focused on distinct chemical characteristics offered by prodrugs 5 and 7 to gauge their preliminary pharmacokinetic profiles and potential adverse effects of these chemical substitutions.

Evaluation of the lead prodrugs 5 and 7 was conducted in an intact animal to validate the findings from the in vitro experiments. When administered orally to CD1 mice at 250 mg/kg, prodrugs 5 and 7 displayed increased C_max, suggesting enhanced stability in plasma (Figure 3C). Levels of the tripeptide ψ-GSH released from both the prodrugs were higher than that achieved by ψ-GSH itself over the time course of this experiment. The dose of ψ-GSH that was found to be efficacious in our previous studies with transgenic AD mice (500 mg/kg, i.p.) was selected for the purpose of comparison. Analysis of the brain tissues of these mice offered similar changes in the concentrations of prodrugs and ψ-GSH (Figure 3D). Prodrug 7 was detected in the brain tissue, and levels of ψ-GSH achieved after prodrug 7 treatment were comparable to or higher than those achieved by administration of ψ-GSH itself. Brain concentrations of ψ-GSH detected after prodrug 5 administration were lower than those achieved after prodrug 7 and ψ-GSH treatment. The extents of brain exposure to prodrugs and ψ-GSH were comparable as deemed from their relative brain AUC (8.06, 8.266, and 1.22 nmol·h/mg tissue) for ψ-GSH and prodrugs 7 and 5, respectively. Beneficial effects of prodrugs were more apparent in the plasma exposure with 16–26 fold higher AUC than that of ψ-GSH itself. This translates into a lower brain/plasma (B/P) ratio for prodrugs in comparison to ψ-GSH, indicating poor brain permeation of the prodrugs. However, in the absence of early time points in the pharmacokinetic curve, protein binding estimation, and calculation of unbound compound concentrations, this observation should be interpreted with caution.³⁵ The improved peripheral stability offered by the prodrugs could result in pharmacologically relevant brain exposure to the active compound ψ-GSH upon their oral administration. A dose–response study with compound 7 showed a significant increase in plasma and the corresponding brain concentrations of the parent prodrug and liberated ψ-GSH at 250 mg/kg dose. The increase at the 500 mg/kg dose was only marginal over that of the 250 mg/kg dose (Supporting Information; Figure S2). We selected the 250 mg/kg dose for efficacy studies with these prodrugs.

Others and we have demonstrated carrier-mediated transport of GSH and ψ-GSH across the BBB.^22,36 It is expected that organic cation transporters such as OCT1/2/3 could enable transport of prodrug 5.³⁷ Additionally, there are mono- and low-capacity dicarboxylic acid transporters (depending on the intra- and intermolecular hydrogen bonding status³⁸) located at the BBB that could explain permeation of prodrug 7.^39,40 Intact prodrug 7 concentrations were 2–3 fold lower than the concentration of released ψ-GSH. This coupled with a low B/P ratio of the prodrugs could indicate that the levels of ψ-GSH achieved could be due to the congregate effect of conversion of the prodrugs to ψ-GSH inside the brain and permeation of externally formed ψ-GSH due to the enhanced peripheral stability of the prodrugs.

Efficacy Testing of ψ-GSH Prodrugs in Intracerebroventricular Aβ-Injected Mice

Prior to extensive and chronic efficacy evaluation in a transgenic mouse model, we evaluated the prodrugs in an acute nontransgenic AD mouse model that has been previously utilized in our lab.⁴¹ This model is established by an intracerebroventricular (i.c.v.) injection of Aβ1–42 aggregates and presents a behavioral and biochemical AD-like phenotype.⁴² Evaluation of the effect of compound treatment on working memory impairment observed in this model has been used by us and others for the initial screening of AD therapeutics.^43,44 Here, we studied the effect of oral prodrug treatment 10 days after the i.c.v. Aβ administration on spontaneous alternation in the T-maze cognitive test (Figure 4A). The choice of the dose (250 mg/kg) was based on dose–response studies with prodrug 7 (Supporting Information; Figure S2). The Aβ-only treated mice exhibited a lower alternation rate (47.14 ± 1.95%) when compared to the saline-treated group (70.54 ± 2.05%, p < 0.001). Significant improvement in the alternation rate (67.86 ± 2.70%) was observed after i.p. treatment of ψ-GSH; however, oral ψ-GSH was only marginally effective (51.79 ± 4.43%) in this test (Figure 4B). Both prodrugs 5 and 7 when treated orally were effective in improving the alternation behavior. Similarly, i.p. ψ-GSH and oral treatment of prodrugs 5 and 7 resulted in a significant reduction in repetitive arm entries (0.50 ± 0.19, 1.80 ± 0.20, 1.00 ± 0.31, respectively; Figure 4C) in comparison with the Aβ-only treated group (3.22 ± 0.27). There were no significant intergroup differences in the preference between the two arms of the T-maze (i.e., ratio of the right to left arm entries) or the time taken to complete the task (Figure 4D,E). These controls indicate that the effect of the prodrugs is not likely due to motor impairment or spatial bias of the mice during the T-maze test. The results present convincing evidence of the efficacy of these prodrugs, specifically prodrug 7, on working memory in this acute AD model.

Figure 4.

Oral ψ-GSH prodrug treatment restored cognitive impairment induced by i.c.v. Aβ1–42. (A) Timeline of the experimental procedure. Three days after pretreatment with the prodrugs, i.c.v. injection of Aβ1–42 was performed and then followed by continued treatment with the prodrugs for an additional 8 days. The T-maze spontaneous alternation test was used to assess cognitive function. (B) Decreased alternation observed in saline-treated Aβ-injected mice was restored to levels comparable to vehicle control mice after prodrug 7 treatment. Treatment with ψ-GSH (i.p), but not oral ψ-GSH, showed significant improvement in alternation. (C) Reciprocal reduction in the number of re-entries was observed in the prodrug-treated mice compared to saline-treated Aβ-injected mice. No spatial bias and motor impairment were evident in any of the treatment groups as assessed by the ratio of arm entries (D) and the total time spent to complete the task (E), respectively. Data are shown as the mean ± SEM. Statistical significance was assessed by a one-way ANOVA with Tukey’s post hoc test (*p  < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.001). The timeline (A) was created with BioRender.com.

The results of this experiment indicate the beneficial effects of oral ψ-GSH prodrug treatment on cognition and the advantages of these prodrugs over that of oral ψ-GSH itself. However, the relationship between the pharmacokinetics and efficacy of these compounds is rather complex. This is compounded by several factors—the major metabolism of ψ-GSH is its oxidation to a homo or mixed disulfide. Conversion of such oxidized metabolites of ψ-GSH (homo or mixed disulfides) into active reduced forms would be limited in transgenic AD mice exhibiting increased oxidative stress and compromised GSH biosynthetic machinery. The mouse model employed in this study is generated using relatively young (8–10-week old) wild-type mice. The intact metabolic machinery in this model could potentially reduce the oxidized products to the active moiety by redox recycling, making it difficult to gauge the effect of the metabolic difference on efficacy. Additionally, the mechanism of action of ψ-GSH is a composite effect of its action on the glyoxalase enzyme system, on other GSH-dependent enzymes, and its antioxidant potential. The release kinetics of ψ-GSH can be expected to influence the efficacy of prodrugs in this Aβ-induced AD mouse model. In contrast, the parent compound, ψ-GSH, is readily available to engage other secondary mechanistic pathways. Although evaluation of lower doses of prodrugs could be helpful in understanding the pharmacokinetic-efficacy relationship, this study establishes the advantage of oral administration of prodrugs over that of oral ψ-GSH.

Immunohistochemical and Biochemical Evaluation of Mouse Brain for Oxidative Stress and Neuroinflammatory Markers

Induction of amyloidopathy by i.c.v. injection of oligomeric Aβ1–42 in nontransgenic mice allowed us to study cognitive consequences of oral ψ-GSH prodrug treatment. This rather efficient method allows control over the onset of the pathogenic phenotype induced by Aβ aggregation, in turn facilitating cost-effective drug candidate screening prior to the undertaking of the rather involved experimentation in transgenic AD animal models. Although this model does not replicate all of the behavioral and physiological pathology observed in human AD, it reliably mirrors the neuronal inflammation and toxicity of induced amyloidosis.^45,46 We studied neuroinflammation in the mouse brain using glial fibrillary acidic protein (GFAP) and the ionized calcium-binding adapter molecule 1 (Iba1) as markers (Figure 5). Consistent with the behavioral data, compound 7 caused a significant reduction in dorsal hippocampal GFAP staining with a similar trend also exhibited by the treatment of compound 5 (Figure 5A,B, and Supporting Information, Figure S3). The extent of microglial activation in the dorsal hippocampus determined using the Iba1 antibody corresponded with the extent of astrocyte activation. Increased Iba1 staining apparent in the i.c.v. Aβ and saline-treated controls was significantly reduced by prodrug 7 as well as prodrug 5 treatment (Figure 5C,D, and Supporting Information, Figure S3). Mitigation of Aβ1–42-induced astrocytic and microglial reactivity by treatment of ψ-GSH prodrug 7 was also confirmed in the dentate gyrus region (Supporting Information, Figure S4).

Figure 5.

Treatment with the ψ-GSH prodrug 7 reduced reactive astrogliosis and microglial reactivity in i.c.v. Aβ-injected mice. (A) Representative images of reactive astrocytes in the dorsal hippocampus visualized by the GFAP antibody. (B) Quantification of GFAP staining in the dorsal hippocampus of saline and prodrug-treated cohorts. (C) Representative images of activated microglia in the dorsal hippocampus visualized by the Iba1 antibody. (D) Quantification of Iba1 staining in the dorsal hippocampus of the saline and prodrug-treated cohorts. For comparisons between saline, Aβ, and Aβ+ prodrug 7 treated groups, a one-way ANOVA with Tukey’s post hoc test was used for data analysis. (**p < 0.01, ***p < 0.005).

Oxidative stress plays an important role in the pathophysiology of Alzheimer’s disease, and increased brain oxidative stress is reported in this Aβ-induced AD model.^47,48 Hence, we examined the brain tissue for biochemical indicators of oxidative and inflammatory stress. The level of antioxidant GSH was used as a surrogate for a qualitative gauge of the redox potential of tissue. Increased oxidative stress was evident in Aβ-treated control mice from the reduced levels of the total GSH and GSH/GSSG ratio (Figure 6A). Administration of orally available ψ-GSH prodrugs restored the levels of reduced GSH to levels higher than those found in Aβ- and saline-treated groups (1.45–1.7 fold). To determine the effect of multiple dosing of prodrugs on the brain levels of ψ-GSH in comparison with the levels detected after a single injection in the pharmacokinetic study (Figure 3), we measured ψ-GSH concentrations in the brain tissues of i.c.v. Aβ-injected mice. Quantitation of reduced ψ-GSH in brain homogenates offered similar concentrations in the prodrug-treated mice and i.p. ψ-GSH group (3.16 vs 2.64 nmol/mg tissue, respectively). This analysis may have been affected by freeze–thawing of the samples leading to variation in the levels of oxidized and reduced ψ-GSH species. The levels, however, were comparable to the concentrations achieved in the pharmacokinetic study (Figure 3). To establish the engagement of the glyoxalase enzyme pathway after prodrug administration, we measured the levels of AGEs in these tissues (Figure 6B). The GLO1 substrate methylglyoxal, being the main precursor of AGE formation, affects the total AGE content. Changes in levels of AGE would indicate the activation of the GLO1 enzymatic function. An increased AGE content in the Aβ1–42 group, presumably due to inactivation of the GLO1 enzyme function due to unavailability of GSH, was reduced significantly by the oral prodrug treatment. This essentially serves as evidence supporting the release of the active parent compound, ψ-GSH, by these prodrugs in the presence of AD pathology. Quantitation of d-lactate, the product of the glyoxalase pathway, offered inconclusive results (Supporting Information, Figure S5), which could be due to the lower sensitivity of the available ELISA kit or nonideal dose/time points offered by this study for evaluation of such a sensitive marker. Evaluation of d-lactate release after administration of a high dose of prodrug 7 (500 mg/kg) using multiple time points enabled detection of an immediate spike in d-lactate release (Supporting Information, Figure S5). Collectively, these data indicate that increased inflammatory and oxidative stress caused by i.c.v. administration of Aβ1–42 was counteracted successfully by ψ-GSH prodrug treatment, as assessed by the corresponding alleviation of neuroinflammation and toxicity that was mirrored in improved cognitive function.

Figure 6.

(A, B) Restoration of glyoxalase function was confirmed by quantitation of brain GSH levels (A) and advanced glycation endproducts (AGEs, B). Prodrugs 5 and 7 significantly increased the levels of reduced GSH compared to the Aβ-only group. Consequently, the total AGE content, a surrogate for brain methylglyoxal levels, was significantly reduced. Data are presented as the mean ± SEM, and the groups were compared with a one-way ANOVA with Tukey’s post hoc test (*p < 0.05, **p < 0.01).

Conclusions

In summary, we have demonstrated that classical prodrugs of ψ-GSH with masked thiol are able to overcome in part, the metabolic liability of ψ-GSH, rendering it orally bioavailable. The prodrugs displayed an improved in vitro and in vivo stability profile and were able to convert to the parent compound at the site of action. When evaluated in a mouse model subjected to i.c.v. amyloid-induced cognitive dysfunction, the lead prodrug 7, was able to alleviate oxidative and inflammatory changes observed in the brain and reversed the cognitive behavioral phenotype displayed by the saline-treated amyloid-only group. We have also confirmed the restoration of glyoxalase activity by oral administration of these compounds, attesting to the replenishment of the redox potential of the brain tissue and mitigation of glycation-induced oxidative damage. In addition to prodrug 7, analogues 5 and 6 offer a significant pharmacokinetic advantage over the parent compound ψ-GSH. This study opens the tantalizing prospect for further development of ψ-GSH-based orally available prodrugs for effective management of AD. The results presented here form a rational basis for extensive pharmacokinetic and efficacy evaluation of the lead compounds in transgenic animal models of this devastating disease.

Experimental Section

Chemistry

General Procedures

All commercial reagents and solvents were used as provided. Flash chromatography was performed with Ultrapure silica gel or with RediSep R_f silica gel columns on a Teledyne ISCO CombiFlash Rf system using the solvents as indicated. Reverse-phase chromatography was performed with C18-bound silica gel. The general method for purification of these compounds involved the adjustment of the pH of aqueous TFA salt solutions to 8.0 using 28% aqueous ammonia, followed by evaporation. The residues were loaded on a C-18 bound silica gel cartridge, followed by elution with water. Relevant fractions were lyophilized to obtain the final compounds. Nuclear magnetic resonance (NMR) spectra were recorded on a Varian 600 MHz (Agilent Technologies, Santa Clara, CA) or Bruker 400 spectrometer (Bruker, Billerica, MA) with Me₄Si or signals from the residual solvent as the internal standard for ¹H and ¹³C. Chemical shifts are reported in ppm, and signals are described as s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), brs (broad singlet), and dd (double doublet). Values given for coupling constants are first order. High-resolution and low-resolution mass spectra were recorded on an Agilent TOF II TOF/MS instrument equipped with an ESI interface. All compounds were >95% pure by HPLC analysis.

N-Boc-γ-l-Glz[-l-Cys(SAc)-Gly-OtBu]-OtBu (9)

To a stirred solution of 8 (130 mg, 0.25 mmol) in ethyl acetate (5 mL) were added acetic anhydride (29 μL, 0.3 mmol) and potassium carbonate (69 mg, 0.5 mmol), and the resulting mixture was stirred at rt overnight. The reaction mixture was partitioned between ethyl acetate and water. The organic layer was washed with brine and dried over sodium sulfate, filtered, and the residue obtained after evaporation of ethyl acetate was purified by column chromatography using EtOAc/hexane (1:1) to isolate the product as a colorless oil (100 mg, 77% yield). ¹H NMR (600 MHz, CDCl₃) δ 7.30 (s, 1H), 5.98 (d, J = 6.9 Hz, 1H), 5.79 (d, J = 6.7 Hz, 1H), 5.68 (s, 1H), 4.50 (dd, J = 12.9, 7.4 Hz, 1H), 4.12 (s, 1H), 3.87–3.83 (m, 2H), 3.51–3.42 (m, 2H), 3.29 (dd, J = 14.0, 5.2 Hz, 1H), 3.14 (dd, J = 13.6, 7.8 Hz, 1H), 2.28 (s, 3H), 1.39 (s, 9H), 1.38 (s, 9H), 1.36 (s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 196.41, 171.31, 170.16, 168.53, 158.09, 155.77, 109.99, 82.08, 79.68, 55.08, 53.69, 42.14, 31.42, 30.45, 28.32, 28.00, 27.91, ESI-MS m/z (M + H)⁺ 563.30.

NH₂-γ-l-Glz[-l-Cys(SAc)-Gly-OH]-OH (1)

To a stirred solution of 9 (100 mg, 0.18 mmol) in DCM (5 mL) at 0 °C was added TFA (5 mL), and the resulting mixture was stirred at rt for 6 h. The reaction mixture was neutralized and then evaporated to dryness. The residue was loaded on a C-18 bound silica gel column, and the column was eluted with water. Relevant fractions were evaporated to dryness to obtain an oil that was triturated with methanol to afford the title compound as a white solid (70 mg, 100%). ¹H NMR (600 MHz, D₂O) δ 4.37–4.31 (m, 1H), 4.02–3.97 (m, 1H), 3.86 (s, 2H), 3.64–3.50 (m, 2H), 3.27 (dd, J = 14.3, 5.0 Hz, 1H), 3.06 (dd, J = 14.3, 7.5 Hz, 1H), 2.24 (s, 3H). ¹³C NMR (151 MHz, D₂O) δ 200.05, 173.13, 172.91, 170.41, 159.48, 53.99, 53.50, 41.04, 39.63, 30.50, 29.81. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 351.0969; found 351.0971.

N-Boc-γ-l-Glz[-l-Cys(SCOOEt)-Gly-OtBu]-OtBu (10)

To a stirred solution of 8 (130 mg, 0.25 mmol) in DCM (5 mL) was added ethyl chloroformate (29 μL, 0.3 mmol) and Et₃N (70 μL, 0.5 mmol), and the resulting mixture was stirred at rt overnight. The reaction mixture was partitioned between ethyl acetate and water. The ethyl acetate layer was washed with brine and dried over sodium sulfate, filtered, and the residue obtained after evaporation of the solvent was purified by column chromatography using EtOAc/hexane (3:2) to isolate the product as a colorless oil (130 mg, 88%). ¹H NMR (600 MHz, CDCl₃) δ 7.37 (s, 1H), 6.14 (d, J = 6.1 Hz, 1H), 5.84–5.79 (m, 2H), 4.58 (dd, J = 13.3, 7.0 Hz, 1H), 4.22 (brs, J = 4.9 Hz, 2H), 4.14 (s, 1H), 3.86 (brs, 2H), 3.53–3.44 (m, 2H), 3.29 (dd, J = 14.2, 5.3 Hz, 1H), 3.13 (dd, J = 13.3, 6.9 Hz, 1H), 1.41 (s, 9H), 1.40 (s, 9H), 1.38 (s, 9H), 1.23 (t, J = 7.1 Hz, 3H). ¹³C NMR (151 MHz, CDCl₃) δ 171.34, 171.07, 170.21, 168.52, 158.10, 155.79, 82.05, 79.63, 63.92, 55.09, 53.78, 42.17, 33.12, 28.30, 27.97, 27.89, 14.17, ESI-MS m/z (M + H)⁺ 593.31.

NH₂-γ-l-Glz[-l-Cys(SCOOEt)-Gly-OH]-OH (2)

To a stirred solution of 10 (130 mg, 0.22 mmol) in DCM (5 mL) at 0 °C was added TFA (5 mL), and the resulting mixture was stirred at rt for 6 h. The reaction mixture was neutralized and then evaporated to dryness. The residue was loaded on a C-18 bound silica gel column, and the column was eluted with water. Relevant fractions were evaporated to dryness to obtain an oil that was triturated with methanol to afford the title compound as a white solid (70 mg, 83%). ¹H NMR (600 MHz, D₂O) δ 4.60–4.38 (m, 1H), 4.28–4.09 (m, 2H), 3.92–3.73 (m, 2H), 3.59 (dd, J = 15.2, 3.6 Hz, 1H), 3.46 (dd, J = 15.1, 6.5 Hz, 1H), 3.30 (dd, J = 14.5, 4.9 Hz, 1H), 3.07 (dd, J = 14.5, 7.6 Hz, 1H), 2.98-2.87(m,1H), 1.15 (t, J = 7.1 Hz, 3H). ¹³C NMR (151 MHz, D₂O) δ 173.69, 173.61, 173.03, 172.03, 159.74, 65.22, 55.36, 53.78, 41.58, 40.27, 32.51, 13.42. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 381.1075; found 381.1084.

N-Boc-γ-l-Glz[-l-Glp(4-thio)-Gly-OtBu]-OtBu (11)

Under argon, carbonyl diimidazole (112 mg, 0.69 mmol) was added to a solution of 8 (300 mg, 0.58 mmol) in dioxane (4 mL), and the resulting mixture was stirred at rt overnight. The solvent was evaporated and the residue was redissolved in xylene (4 mL), and the resulting mixture was refluxed for 4 h and then was evaporated to dryness. The residue obtained was purified by column chromatography using EtOAc/hexane (1:3) to afford the title compound as a colorless oil (120 mg, 40%). ¹H NMR (600 MHz, CDCl₃) δ 8.29 (t, J = 5.8 Hz, 1H), 6.85 (s, 1H), 5.32 (d, J = 6.6 Hz, 1H), 5.15 (d, 1H, J = 6.0 Hz), 4.27 (s, 1H), 3.93–3.82 (m, 2H), 3.74–3.72 (m, 1H), 3.65–3.62 (m, 1H), 3.53–3.48 (m, 2H), 1.41 (s, 9H), 1.40 (s, 9H), 1.38 (s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 174.37, 169.55, 168.34, 168.32, 155.35, 152.52, 82.88, 82.37, 80.01, 59.44, 53.88, 42.29, 42.19, 28.26, 28.01, 27.92, 27.30, ESI-MS (M + Na)⁺ 569.24.

NH₂-γ-l-Glz[-l-Glp(4-thio)-Gly-OH]-OH (3)

To a stirred solution of 19 (106 mg, 0.22 mmol) in DCM (5 ml) at 0 °C was added TFA (5 mL), and the resulting mixture was stirred at rt for 6 h. The reaction mixture was neutralized and then evaporated to dryness. The residue was loaded on a C-18 bound silica gel column, and the column was eluted with water. Relevant fractions were evaporated to dryness to obtain an oil that was triturated with methanol to afford the title compound as a white solid (61 mg, 94%). ¹H NMR (600 MHz, D₂O) δ 5.11 (dd, J = 9.3, 2.2 Hz, 1H), 4.09–4.03 (m, 1H), 3.92 (dd, J = 47.3, 18.0 Hz, 2H), 3.79–3.64 (m, 3H), 3.26 (dd, J = 11.9, 2.2 Hz, 1H). ¹³C NMR (151 MHz, D₂O) δ 176.51, 172.68, 172.07, 170.15, 153.02, 59.68, 53.33, 41.12, 39.68, 27.65. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 335.0656; found 335.0662.

HCl.NH₂-γ-l-Glz[-l-Cys(SH)-Gly-OcPe]-OcPe (5)

To a stirred solution of 8 (149 mg, 0.29 mmol) in TFA (4 mL) at 0 °C was added TIPS (200 μL, 0.97 mmol), and the resulting mixture was stirred at rt overnight. The reaction mixture was then evaporated to dryness and the residue was loaded on a C-18 bound silica gel column, and the column was eluted with water. Relevant fractions were evaporated to dryness to obtain an oil that was triturated with methanol to afford the title compound as a white solid (88 mg, 80%). ¹H NMR (600 MHz, D₂O) δ 4.34–4.29 (m, 1H), 3.92–3.83 (m, 3H), 3.62 (dd, J = 15.2, 3.8 Hz, 1H), 3.51 (dd, J = 15.2, 6.2 Hz, 1H), 2.88–2.77 (m, 2H). ¹³C NMR (151 MHz, D₂O) δ 173.59, 173.17, 171.47, 159.72, 55.74, 54.82, 41.15, 40.03, 25.98. ESI-HRMS calcd m/z (M + H)⁺ 309.0863; found 309.0854.

Thionyl chloride (60 μL, 0.5 mmol) was slowly added to an ice-cold solution of deprotected tripeptide (ψ-GSH) (100 mg, 0.33 mmol) in cyclopentanol (5 mL). The reaction mixture was allowed to warm to rt and then stirred at 50 °C for 5 h. The solvent was evaporated under vacuum to afford diester 8 as a white solid. The crude product was recrystallized from ethyl acetate resulting in the title compound (140 mg, 90%). ¹H NMR (400 MHz, DMSO-d6) δ 8.75–8.47 (m, 4H), 6.96–6.41 (m, 2H), 5.34–4.88 (m, 1H), 4.34 (m, 1H), 4.08–3.96 (m, 2H), 3.86–3.71 (m, 2H), 3.68–3.35 (m, 3H), 2.89–2.64 (m, 1H), 1.94–1.31 (m, 16H). ¹³C NMR (101 MHz, DMSO-d6) δ 171.23, 169.84, 168.13, 158.18, 79.28, 77.62, 72.20, 60.22, 55.32, 53.39, 41.49, 35.47, 32.57, 23.76, 23.40. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 445.2039; found 445.2068.

NH₂-γ-l-Glz[-l-Cys(SCOOEt)-Gly-OcPe]-OcPe (6)

Compound 5 (100 mg, 0.33 mmol) was dissolved in 200 mM aqueous potassium carbonate (2 mL). The solution was cooled in an ice bath and then covered with THF (10 ml). Ethyl chloroformate (35 mg, 0.33 mmol) was added to the solution in one portion with vigorous stirring for 1 h at 0 °C. The temperature was allowed to increase to 10 °C and stirred for an additional 2 h. After evaporation of the organic solvent, the mixture was partitioned between ethyl acetate and the aqueous layer, followed by washing with brine. The residue obtained after evaporation of EtOAc was triturated with hexane to provide the product as a white solid (92 mg, 86%). ¹H NMR (400 MHz, CD₃OD) δ 5.38 (m, 1H), 5.25(m,1H), 5.12 (s, 1H),4.38-4.23 (m,2H) 4.20–4.13 (m, 1H), 4.08–3.91 (m, 2H), 3.81-3.55 (m, 2H), 3.32–3.17(m, 1H), 1.95–1.39 (m, 18H), 1.29–0.92 (m, 3H). ¹³C NMR (101 MHz, CD₃OD) δ 171.76, 170.70, 169.45, 167.18, 159.28, 80.09, 78.22, 63.77, 54.12, 53.36, 40.99, 39.85, 32.65, 32.16, 32.12, 23.27, 23.24, 23.22, 13.17. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 539.2149; found 539.2179.

N-Ethyloxycarbonyl-γ-l-Glz[-l-Cys(SCOOEt)-Gly-OH]-OH (7)

ψ-GSH synthesized as described in the synthesis of compound 5 (123 mg, 0.4 mmol) was dissolved in a mixture of THF/H₂O (2:5, 4.2 mL), followed by addition of sodium bicarbonate (148 mg, 1.76 mmol). The solution was cooled in an ice bath before addition of ethyl chloroformate (153 μL, 1.61 mmol) was added to the solution in one portion with vigorous stirring for 1 h at 0 °C. The temperature was allowed to increase to rt and stirred for an additional 3 h. The solvents were then evaporated in vacuo and the resulting mixture was partitioned between diethyl ether and aqueous sodium bicarbonate. The aqueous layer was acidified with sodium hydrogen sulfate and extracted with ethyl acetate thrice. Combined ethyl acetate layers were washed with brine, dried over sodium sulfate, and concentrated to afford the product as a white solid. The solid was further purified by reverse-phase column chromatography (153 mg, 85%). 1H NMR (400 MHz, D₂O) δ 4.52–4.49 (m, 1H), 4.29–4.26 (m, 2H), 4.16–4.11 (m, 1H), 4.08–4.01 (m, 2H), 3.89-3.82(m, 2H), 3.59–3.53 (m, 1H), 3.42–3.36 (m, 2H), 3.16–3.09 (m, 1H), 1.27-1.07 (m, 6H). ¹³C NMR (101 MHz, D₂O) δ 175.91, 174.66, 172.88, 172.16, 159.47, 158.25, 65.15, 62.02, 55.77, 53.69, 42.25, 41.27, 32.76, 13.83, 13.45. ESI-HRMS calcd m/z (M – H)⁻ 451.1149; found 451.1164.

N-Fmoc-l-Thz-Gly-OtBu (13)

To a stirred solution of 12 (1.33 g, 10 mmol) in 50 mL of 0.4N NaOH solution at 0 °C was added dropwise Fmoc-Cl (2.84g, 11mmol) in 50 mL of dioxane, after 2h of stirring at 0 °C, and the reaction mixture was stirred at rt overnight. The dioxane layer was evaporated, and the aqueous solution was extracted with ethyl acetate two times. The aqueous layer was then acidified to pH 2 and extracted with ethyl acetate two times. The combined organic layer was dried over sodium sulfate to afford the Fmoc-protected acid as white foam (1.94 g, 60%). ¹H NMR (600 MHz, cdcl₃) δ 11.05 (brs, 1H), 7.74–7.70 (m, 2H), 7.58–7.53 (m, 2H), 7.38–7.24 (m, 4H), 4.68 (s, 1H), 4.57–4.38 (m, 3H), 4.24–4.17 (m, 1H), 3.71 (s, 1H), 3.25 (s, 2H).

EDC (1.26 g, 6.55 mmol), HOBt (0.88 g, 6.55 mmol), and NMM (1.8 mL, 16.38 mmol) were added to an ice-cold solution of Fmoc-acid (1.94 g, 5.46 mmol) and HCl·Gly-OtBu (1.13 g, 6.76 mmol) in CH₂Cl₂ (20 mL). The solution was allowed to stir for 12 h before being poured into water (20 mL). The dichloromethane layer was washed with 10% citric acid solution (25 mL), sodium bicarbonate solution (2 × 20 mL), and brine and dried over Na₂SO₄. The residue obtained after evaporation of CH₂Cl₂ was purified by column chromatography using EtOAc/hexane (1:5) to isolate the product as white foam (1.8 g, 72% yield). ¹H NMR (600 MHz, CDCl₃) δ 7.74–7.70 (m, 3H), 7.58–7.53 (m, 2H), 7.38–7.24 (m, 4H), 4.80–4.37 (m, 5H), 4.21 (s, 1H), 3.90 (s, 2H), 3.39 (s, 1H), 3.14 (s, 1H), 1.43 (s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 169.77, 168.43, 154.68, 143.48, 141.28, 127.82, 127.13, 124.91, 124.87, 120.04, 82.24, 68.19, 63.03, 49.07, 47.07, 42.11, 32.76, 28.03.

NH₂-l-Thz-Gly-OtBu (13)

(R)-tert-Butyl 2-(thiazolidine-4-carboxamido)acetate (14). Diethylamine (15 mL) was added to a solution of 13 (1.8 g, 3.84 mmol) in acetonitrile (15 mL), and the resulting mixture was stirred at rt for 30 min to ensure complete removal of the Fmoc group. The solvent was evaporated out and the residue was purified by column chromatography (CH₃OH/CH₂Cl₂ 1:9) to obtain the product as a light yellow oil (0.79 g, 83%). ¹H NMR (600 MHz, CDCl₃) δ 7.70 (s, 1H), 4.25 (d, J = 9.7 Hz, 1H), 4.19 (dd, J = 6.5, 3.7 Hz, 1H), 4.06 (d, J = 9.6 Hz, 1H), 3.96–3.87 (m, 2H), 3.37 (dd, J = 10.5, 4.1 Hz, 1H), 3.08 (dd, J = 10.3, 7.7 Hz, 1H), 2.80 (s, 1H), 1.47 (s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 171.20, 168.75, 82.01, 65.95, 53.54, 41.78, 35.18, 28.00, ESI-MS (M + H)⁺ 247.13.

N-Boc-γ-l-Glz[-l-Thz-Gly-OtBu]-OAll (15)

Triphosgene (61 mg, 0.21 mmol) and TEA (100 μL, 0.72 mmol) were added to a solution of 13 (118 mg, 0.48 mmol) in dry THF (10 mL). After the mixture was stirred at room temperature for 10 min, a solution in dry THF of 19 (131 mg, 0.54 mmol) was added. The reaction mixture was then stirred at room temperature for 3 h. The solvent was evaporated, and the residue was dissolved in dichloromethane. The organic solution was washed with water, dried over Na₂SO₄, and evaporated in vacuo. The residue was purified by column chromatography (EtOAc/hexane 2:3) to obtain the product as a light yellow oil (70 mg, 55%). ¹H NMR (600 MHz, CDCl₃) δ 7.19 (s, 1H), 6.92 (s, 1H), 5.92–5.88 (m, 1H), 5.72 (s, 1H), 5.33 (d, J = 17.2 Hz, 1H), 5.24 (d, J = 6.0 Hz, 1H), 5.09 (t, 1H, J = 6.0 Hz), 4.77–4.69 (m, 1H), 4.62–4.57 (m, 2H), 4.39– 4.37 (m, 1H), 3.98–3.82 (m, 3H), 3.72–3.67 (m, 1H), 3.58–3.56 (m, 1H), 3.45–3.41 (m, 1H), 3.26–3.16 (m, 1H), 1.45(s, 9H), 1.42(s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 170.71, 169.81, 168.76, 156.85, 155.96, 131.50, 118.92, 82.30, 80.20, 66.27, 63.98, 63.06, 49.22, 43.22, 42.11, 33.48, 28.27, 28.01, ESI-MS m/z (M + H)⁺ 517.23 (M + Na)⁺ 539.22.

N-Boc-γ-l-Glz[-l-Thz-Gly-OtBu]-OH (16)

(Ph₃P)₄Pd (16 mg, 10 mol%) was added under nitrogen to a solution of allyl ester 15 (70 mg, 0.14 mmol) in CH₂Cl₂ (5 mL) at room temperature, and the resulting mixture was treated dropwise with (117 μL, 1.36 mmol) morpholine. After completion of the reaction (1 h) as judged by TLC, the solvent was removed in vacuo. The residue was redissolved in CH₂Cl₂ (20 mL), and the resulting solution was washed with 1 N HCl (15 mL) and water (15 mL). The organic layer was dried over Na₂SO₄ and evaporated in vacuo. The crude product (light yellow foam, 65 mg, 100%) was used for the next step without further purification (M – H)⁻ 475.21.

NH₂-γ-l-Glz[-l-Thz-Gly-OH]-OH (4)

To a stirred solution of 16 (65 mg, 0.14 mmol) in DCM (5 mL) at 0 °C was added TFA (5 mL), and the resulting mixture was stirred at rt for 6 h. The reaction mixture was neutralized and then evaporated to dryness. The residue was loaded on a C-18 bound silica gel column, and the column was eluted with water. Relevant fractions were evaporated to dryness to obtain an oil that was triturated with methanol to afford the title compound as white foam (23 mg, 80%). ¹H NMR (600 MHz, D₂O) δ 4.70–4.67 (m, 1H), 4.42 (d, J = 7.9 Hz, 1H), 4.39–4.22 (m, 1H), 3.84 (s, 2H), 3.79–3.78 (m, 1H), 3.62 (d, J = 15.0 Hz, 1H), 3.50–3.46 (m, 1H), 3.25–3.22 (m, 1H), 3.07 (d, J = 12.0 Hz, 1H). ¹³C NMR (151 MHz, CD₃OD) δ 173.56, 173.29, 171.73, 157.98, 63.16, 55.27, 48.53, 41.36, 40.70, 33.45. ESI-HRMS calcd m/z (M + H)⁺ calcd m/z 321.0863; found 321.0877.

N-Boc-l-Dap(N-Fmoc)-OH (18)

Using a literature-reported procedure,⁴⁹ a slurry of Boc-l-asparagine (17, 5.0 g, 21.5 mmol) in ethyl acetate (24 mL), acetonitrile (24 mL), water (12 mL), and iodosobenzene diacetate (8.32 g, 25.8 mmol) was cooled and stirred at 16 °C for 30 min. The temperature was increased to 20 °C, and the reaction was stirred for 4 h. The reaction mixture was cooled to 0 °C and filtered under vacuum. The filter cake was washed with ethyl acetate and dried in a vacuum to obtain N-Boc-protected 2,3-diaminopropionic acid (Dap) (3.2 g, 76%) as a white solid.

To a stirred solution of the isolated amine (2.35 g, 11.52 mmol) in 25 mL of dioxane and 25 mL of water were added NaHCO₃ (2.42 g, 28.8 mmol) and Fmoc-Cl (2.98 g, 11.52 mmol), and the reaction mixture was stirred overnight at rt. The organic phase was evaporated, and the aqueous solution was extracted with ethyl acetate two times and then the aqueous layer was acidified to pH 2 and extracted with ethyl acetate two times. The combined organic layer was dried to afford the title compound as white foam (5.54 g, 100%) which was used in the next step without further purification.

N-Boc-l-Dap(NH₂)-OAll (19)

Compound 18 (5.5 g, 12.90 mmol) was dissolved in a mixture of DIPEA (14.19 mmol, 2.47 mL) and acetonitrile (30 mL). Allyl bromide (10 mL) was added and the solution was stirred for 16 h at rt. The reaction mixture was concentrated, and the residue was diluted with ethyl acetate, washed with sat. NaHCO₃ and brine, and dried over sodium sulfate. The solvent was evaporated and the residue was purified by column chromatography (EtOAc/hexane 1:2) to obtain the product as a white solid (4.18 g, 70%). ¹H NMR (600 MHz, CDCl₃) δ 7.70 (d, J = 7.5 Hz, 2H), 7.54 (d, J = 7.2 Hz, 2H), 7.34 (t, J = 7.4 Hz, 2H), 7.25 (t, J = 7.2 Hz, 2H), 5.85 (s, 1H), 5.74 (s, 1H), 5.66 (s, 1H), 5.28 (d, J = 17.2 Hz, 1H), 5.17 (d, J = 10.3 Hz, 1H), 4.60 (s, 2H), 4.43 (s, 1H), 4.32 (s, 2H), 4.14 (s, 1H), 3.60 (s, 2H), 1.43 (s, 9H). ¹³C NMR (151 MHz, CDCl₃) δ 170.47, 156.83, 155.61, 143.80, 143.74, 141.18, 131.46, 127.63, 126.99, 125.03, 119.90, 118.82, 80.09, 66.97, 66.20, 54.23, 47.04, 42.84, 28.26, ESI-MS m/z (M + Na)⁺ 489.20.

Diethylamine (20 mL) was added to a solution of isolated allyl ester (4.18 g, 8.96 mmol) in acetonitrile (20 mL), and the resulting mixture was stirred at rt for 30 min to ensure complete removal of the Fmoc group. The solvent was evaporated out and the residue was purified by column chromatography (CH₃OH/CH₂Cl₂ 1:5) to obtain the product as a light yellow oil (740 mg, 64%). ¹H NMR (600 MHz, CDCl₃) δ 5.86–5.80 (m, 1H), 5.58 (d, J = 6.5 Hz, 1H), 5.26 (d, J = 17.2 Hz, 1H), 5.17 (d, J = 10.4 Hz, 1H), 4.57 (t, J = 6.5 Hz, 2H), 4.24 (s, 1H), 2.98 (d, J = 4.3 Hz, 2H), 1.36 (s, 9H), 1.15 (s, 2H). ¹³C NMR (151 MHz, CDCl₃) δ 171.35, 155.63, 131.65, 118.73, 79.81, 65.85, 56.10, 43.91, 28.33, ESI-MS m/z (M + H)⁺ 245.15.

In Vitro Stability Analysis of ψ-GSH Prodrugs

Simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) were prepared according to the procedure of the USP National formulary.⁵⁰ The in vitro stability of the ψ-GSH and its prodrugs was investigated in mouse plasma (CD-1), mouse brain and liver homogenates, SGF, and SIF. Briefly, the prodrugs (final concentration 1 μM) were incubated with plasma, brain, liver homogenates, SGF, or SIF at 37 °C. The incubation aliquots were collected at 0, 5, 15, 30, 60, 120, and 240 minutes and quenched with three volumes of acetonitrile containing 1% formic acid. After centrifugation of samples, the supernatants were analyzed by HPLC and/or LC-MS/MS to determine the concentrations of prodrugs, ψ-GSH, and calculate the half-life (t1/2) of each prodrug. An Agilent 1260 Infinity HPLC with a UV detector was used for HPLC quantitation of compounds after derivatization with Fmoc-chloride using a Phenomenex Gemini C18 column (150 mm × 4.6 mm, 5 μm). The chromatography conditions were: detector wavelength 254 nm; mobile phase of 0.1% formic acid in water/acetonitrile (99:1; A) and 0.1% formic acid in acetonitrile/water (99:1; B) at a flow rate of 1 mL/min using gradient separation (from 0 to 2 min, 10% (v/v) B; from 2 to 10 min, 10 to 100% (v/v) B; from 10 to 12 min, 100 to 10% (v/v) B; and from 12 to 15 min, 10% (v/v) B). Samples were detected. Retention times of ψ-GSH and compounds 1–6 were 11.9, 9.1, 9.54, 9.35, 9.43, 11.63, and 10.56.

The LC-MS/MS quantitation was performed using an Agilent 1260 Infinity HPLC (Agilent Technologies, Santa Clara, CA) coupled to an AB Sciex QTrap 5500 mass spectrometer (AB Sciex LLC, Toronto, Canada). The samples were separated using a Thermo Aquasil C18 column (150 mm × 2.1 mm, 3 μm), and the analytes were eluted with a mobile phase of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) at a flow rate of 0.3 mL/min using the following gradient: from 0 to 2 min, 0 to 45% (v/v) B; from 2 to 3.3 min, 45% (v/v) B; from 3.3 to 3.5, 45 to 0% (v/v) B; and from 3.5 to 10 min, 0% (v/v) B. Only desired fractions of eluates (2 to 3 min, 5.5 to 8 min) were analyzed with an electrospray ionization source operated in positive mode. Multiple reaction monitoring (MRM) was conducted by monitoring the following transitions: m/z 309.1 → 105.0 (ψ-GSH); m/z 381.1 → 148.1 (compound 2); m/z 445.2 → 179.1 (compound 5, MRM 1); m/z 445.2→302.2 (compound 5, MRM 2); and m/z 453.1 → 148.1 (compound 7). The following gradient was used for the analysis of compound 6: from 0 to 1 min, 20 to 95% (v/v) B; from 1.0 to 2.8 min, 95% (v/v) B; from 2.8 to 3.0, 95 to 20% (v/v) B; and from 3 to 6 min, 20% (v/v) B, and mass transition from m/z 517.2 → 319.2 (compound 6) was monitored.

Animals

The transgenic mice expressing APPswe/PS1ΔE9 (APP/PS1) were obtained from the Jackson Laboratory (Bar Harbor, ME). The wild-type C57BL/6 and CD-1 mice for the Aβ1–42-induced model and pharmacokinetic studies, respectively, were obtained from Charles River Laboratories (Wilmington, MA). All mice were housed in a specific pathogen-free (SPF) facility, which maintained a constant temperature (23 ± 1 °C) and humidity (55 ± 5%) under a 12 h light/dark cycle. Four mice per cage had free access to water and food ad libitum. All experimental protocols involving mice were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Minnesota, Minneapolis, MN. All experimental procedures and animal handling were performed with strict adherence to the national ethics guidelines.

Brain and Plasma Distribution Study in Mice

Eight-week-old male CD-1 mice were used for evaluating the brain-to-plasma distribution of the lead ψ-GSH prodrugs via oral administration in two separate experiments (N = 4 per time point or dose). In the first study, the mice were randomly assigned to receive an oral dose of prodrug 5 or 7 (250 mg/kg of each prodrug). The mice were euthanized, and blood and brain samples were collected at 2, 4, and 6 h after administration. In the second study, all mice were orally administered with varying doses of prodrug 7 (100, 250, and 500 mg/kg). Blood and brain samples were obtained at 6 h after prodrug treatment.

For analysis of compound levels, the brain tissue was homogenized with 1:5 (w/v) of DPBS with a mechanical homogenizer. Plasma and brain homogenates were mixed with 100 mM N-ethylmaleimide and the mixtures were incubated at ambient temperature for 10 min. This was followed by deproteinization with 3% sulfosalicylic acid before the samples were analyzed with LC-MS/MS. The samples were injected in an Agilent 1260 HPLC coupled to an AB Sciex QTRAP 5500 mass spectrometer and separated using a Thermo Aquasil C18 column (150 mm × 2.1 mm, 3 μm) with a mobile phase of 0.1% formic acid in water (mobile phase A) and 0.1% formic acid in acetonitrile (mobile phase B) at a flow rate of 0.3 ml/min. The analytes were eluted with a gradient as follows: from 0 to 2 min, 0–50% B (v/v); from 2 to 4 min, 50% B (v/v); from 4 to 4.5 min, 50–0% B (v/v); and from 4.5 to 10 min, 0% B (v/v). All samples were analyzed by an electrospray ionization source in positive mode. The optimized source and gas parameters were as follows: curtain gas, 25 psi; CAD gas, medium; ion spray voltage, 5000 V; temperature, 650 °C; gas 1, 50 psi; gas 2, 50 psi. Multiple reaction monitoring (MRM 1) was conducted by monitoring the following transition: m/z 434.1→105 (ψ-GSH-NEM); m/z 445.2→302.2 (compound 5); m/z 570.2→304.1 (compound 5-NEM, MRM 1); m/z 570.2→427.2 (compound 5-NEM, MRM 2); m/z 453.1→ 148.1 (compound 7).

ψ-GSH Administration to Transgenic Mice

Following the study design described previously, 3–4-month-old APP/PS1 mice were treated with either saline or ψ-GSH (500 mg/kg) by oral gavage, three times a week for 12 weeks. Age-matched nontransgenic wild-type mice treated with saline were used as controls. Efficacy evaluation was conducted 11 weeks after the initiation of the treatment and animals were euthanized a week after for collection of tissues for histological and biochemical analysis.

Behavioral Assessment of Transgenic Mice Using the Morris Water Maze

This testing involved a training period and a probe trial at the end for the evaluation of memory retention. The training phase consisted of four trials per mouse per day for 4 consecutive days. Retention of memory was tested in a probe trial conducted 24 h after the last training trial. Time spent in the quadrant, which previously housed the escape platform, was calculated as a gauge of performance. A visible platform trial was also performed to assess any problems with visual acuity or motor performance in these mice.

Amyloid-β Peptide and Prodrug Administration to Nontransgenic Mice

The oligomeric Aβ1–42 peptide solution was prepared, as described previously.^41,42 HFIP-treated Aβ1–42 peptide was used to prepare 1 mM stock solution. After dilution with DPBS (10% DMSO and 90% DPBS), the resulting Aβ1–42 solution (100μM) was incubated at 37 °C for 3 days before the injection. Wild-type C57BL/6 mice (8–10-week old) were injected intracerebroventricularly with either Aβ1–42 oligomers or DPBS. Another group of mice was injected with the same volume of methylene blue solution to confirm the accuracy of Aβ1–42 injection.

Animals were randomly distributed into the following groups (8–10 mice per group): (1) vehicle (saline, p.o.), (2) Aβ1–42 (saline, p.o.), (3) Aβ1–42 + ψ-GSH (ψ-GSH, i.p.), (4) Aβ1–42 + ψ-GSH (ψ-GSH, p.o.), (5) Aβ1–42 + prodrug 5 (prodrug 5, p.o.), and (6) Aβ1–42 + prodrug 7 group (prodrug 7, p.o.). Compounds ψ-GSH, prodrug 5, and prodrug 7 were dissolved in saline. The compound solution was administered to mice at a dose of 250 mg/kg for ψ-GSH and its prodrugs, once a day for 12 days by oral or intraperitoneal routes. The i.c.v. administration of Aβ1–42 oligomers was conducted on day 4 after initiation of compound treatment. The mice were assessed for cognitive function on days 10 and 11. After completion of the treatment, the mice were euthanized to dissect brains for further biological assays and histopathological analysis.

Spontaneous Alternation T-Maze Test

Working memory in this study was accessed using spontaneous alternation T-maze, following established protocols.⁴¹ The T-maze itself was constructed out of black Plexiglas, as described by Deacon and Rawlins.⁵¹ Mice were acclimatized to the testing room for 1 h prior to evaluation. Mice were individually placed at the end of the start arm of the T-maze. The subject mouse was allowed to roam around the entire maze for free exploration. Once the mouse entered the start arm after the free exploration, a guillotine door was closed, and the mouse was confined in the start arm for 30 sec. At the end of confinement, a central partition was gently placed, and the guillotine door was removed. The mouse was then allowed to make 15 free-choice trials. Mice spending more than 30 min to complete the 15 trials were excluded from the analysis. The percentage of alternation as an index of working memory was calculated using the formula: (alternation/14 × 100). Re-entries three times in a row were considered the repetitive arm entry. The side preference was evaluated by calculation of the ratio of right and left arm entries.

Brain Tissue Preparation

After euthanization, the mice’s brain tissues were removed and half brains were postfixed with fresh 4% paraformaldehyde solution for 24 h. The PFA-fixed brain was transferred to 20% sucrose in PBS until the tissue sank (∼16 h) and then treated with the 30% sucrose solution in PBS until sectioning. The other half brain for biochemical studies was rapidly frozen and stored at −80 °C.

Analysis of Aβ Load, Lipid Peroxidation, and Protein Carbonyls in Mouse Brain

These assays were performed using protocols described previously.²³ Levels of the hallmark Aβ1–42 in transgenic mouse brain was determined using a sandwich ELISA assay according to the manufacturer’s instructions (Thermo Fisher, Waltham, MA). Oxidative stress markers, lipid peroxidation and protein carbonyls, were determined by colorimetric assay for malondialdehyde (MDA) (TBARS assay) and protein-hydrazone products (DNPH assay) using commercial kits (Cayman Chemical Co., MI) per instructions.

Quantitation of GSH in Mouse Brain

The GSH levels in the mouse brain homogenate were measured using a GSH assay kit (Cayman Chemical, #703002, Ann Arbor, MI). Briefly, the mouse brain was homogenized per the manufacturer’s protocol followed by deproteinization with metaphosphoric acid. After neutralization, the samples were diluted with 50 mM MES buffer (pH = 6.0, containing 1 mM EDTA) before being analyzed in this enzymatic GSH recycling assay using DTNB (5,5′-dithiobis-2-nitrobenzoic acid). The yellow-colored 5-thio-2-nitrobenzoic acid (TNB) product formed was measured colorimetrically at 412 nm using a Molecular Devices SpectraMax M5e Microplate reader.

Immunohistochemistry and Stereological Analyses of Neuroinflammatory Markers

The cryopreserved brain tissues were cut into 40 μm frozen coronal sections using a freezing sliding microtome (Leica, Wetzlar, Germany). Antigens of interest were detected by incubation with the corresponding primary antibodies followed by the ABC method (Vector Laboratories, Burlingame, CA) using the chromogen, 3,3′-diaminobenzidine (DAB; Sigma Aldrich, St. Louis, MO) for visualization. Antigen retrieval was performed using a Rodent DeCloaker (BioLegend, San Diego, CA) for all samples. Primary antibodies used are the GFAP anti-rabbit polyclonal antibody (Dako, Glostrup Kommune, Denmark), Iba1 anti-rabbit polyclonal antibody (Wako, Japan), and anti-rabbit polyclonal antibody (Millipore, Burlington, MA).

Stereo Investigator software (Micro Brightfield; Colchester, VT) was used for stereological analyses.^52,53 A region of interest (ROI) was defined using The Mouse Brain Stereotaxic Coordinates⁵⁴ as the reference, and the extent of brain area covered by reactive astrocytes (GFAP) and microglia (Iba1) was quantitated using an area fraction fractionator probe.⁵⁵ The ROIs include the barrel field region of the primary somatosensory cortex (S1BF; sections between bregma −0.10 and −1.22 mm, posterior to the anterior commissure and anterior to the hippocampus) and dorsal hippocampus (dentate, CA1, CA2/3; sections between bregma −1.46 and −2.18 mm); the S1 barrel cortex (S1BF) and dorsal hippocampal regions. Every 12th coronal section (4–6 sections) was immunostained and analyzed using a ×40 objective. Details of stereological calculations are described in our recent publication.¹⁹

Glossary

Abbreviations Used

AD

Alzheimer’s disease

AGE

advanced glycation endproducts

Dap

2,3-diaminopropionic acid

DCM

dichloromethane

ELISA

enzyme-linked immunosorbent assay

EtOAc

ethyl acetate

GFAP

glial fibrillary acidic protein

Glz

l-γ-azaglutamic acid

GGT

γ-glutamyl transpeptidase

GLO

glyoxalase

Glp

pyroglutamic acid

GSH

glutathione

HPLC

high-performance liquid chromatography

i.c.v.

intracerebroventricular

Iba1

ionized calcium-binding adapter molecule 1

LC-MS/MS

liquid chromatography with tandem mass spectrometry

OcPe

cyclopentylester

MEG

methylglyoxal

rt

room temperature

SGF

simulated gastric fluids

SIF

simulated intestinal fluids

TFA

trifluoroacetic acid

THF

tetrahydrofuran

Thz

thiazolidine-4-carboxylic acid

TNFα

tumor necrosis factor

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.2c00779.

• Additional synthetic scheme; experimental details and compound characterization; NMR and HPLC data of compounds tested in animals; biological data comprising immunohistochemical Aβ staining; dose–response pharmacokinetic evaluation; and quantitation of GFAP and Iba1 stains and d-lactate measurement (PDF)

• Molecular formula strings (CSV)

Author Present Address

^∥ XtalPi Inc., Simbay Park, Pudong District, Shanghai, China

Author Present Address

^⊥ Nitto Denko Avecia Inc., Cincinnati, Ohio 45215, United States.

Author Present Address

^# Eisai Center for Genetics Guided Dementia Discovery, Eisai Inc., Cambridge, Massachusetts 02140, United States.

Author Contributions

^∇ W.X. and B.C. contributed equally to this work. The original draft of the manuscript was written by W.X. and S.S.M. and was reviewed and edited by all authors. S.S.M., R.V., and M.K.L. conceptualized and guided the project; B.C., H.Z., and A.R. synthesized the prodrugs described in this manuscript; W.X., R.J., and N.J. conducted biological data collection and analysis; and J.X. conducted the LC-MS/MS analysis of the biological samples. All authors have given approval to the final version of the manuscript.

This research was funded by National Institutes of Health Grants to S.S.M. (R01-AG062469, 1912-37730A) and M.K.L. (RF1-AG062135, R01-NS108686, 1902-36759A) and by funding from the Center for Drug Design (CDD), University of Minnesota.

The authors declare no competing financial interest.