**The Insulin Receptor: A Potential Target of Amarogentin Isolated from** *Gentiana rigescens* **Franch That Induces Neurogenesis in PC12 Cells**

**Lihong Cheng 1, Hiroyuki Osada 2, Tianyan Xing 3, Minoru Yoshida 4,5, Lan Xiang 1,\* and Jianhua Qi 1,\***


**Abstract:** Amarogentin (AMA) is a secoiridoid glycoside isolated from the traditional Chinese medicine, *Gentiana rigescens* Franch. AMA exhibits nerve growth factor (NGF)-mimicking and NGFenhancing activities in PC12 cells and in primary cortical neuron cells. In this study, a possible mechanism was found showing the remarkable induction of phosphorylation of the insulin receptor (INSR) and protein kinase B (AKT). The potential target of AMA was predicted by using a smallinterfering RNA (siRNA) and the cellular thermal shift assay (CETSA). The AMA-induced neurite outgrowth was reduced by the siRNA against the INSR and the results of the CETSA suggested that the INSR showed a significant thermal stability-shifted effect upon AMA treatment. Other neurotrophic signaling pathways in PC12 cells were investigated using specific inhibitors, Western blotting and PC12(rasN17) and PC12(mtGAP) mutants. The inhibitors of the glucocorticoid receptor (GR), phospholipase C (PLC) and protein kinase C (PKC), Ras, Raf and mitogen-activated protein kinase (MEK) significantly reduced the neurite outgrowth induced by AMA in PC12 cells. Furthermore, the phosphorylation reactions of GR, PLC, PKC and an extracellular signal-regulated kinase (ERK) were significantly increased after inducing AMA and markedly decreased after treatment with the corresponding inhibitors. Collectively, these results suggested that AMA-induced neuritogenic activity in PC12 cells potentially depended on targeting the INSR and activating the downstream Ras/Raf/ERK and PI3K/AKT signaling pathways. In addition, the GR/PLC/PKC signaling pathway was found to be involved in the neurogenesis effect of AMA.

**Keywords:** neurodegenerative disease; aging; Alzheimer's disease; insulin receptor; target identification

#### **1. Introduction**

Alzheimer's disease (AD) is a type of progressive neurodegenerative disease that accounts for 60–70% of dementia cases and its symptoms include an initial memory loss, later visual, language and cognitive disorders and a decline in the executive capacity in daily life [1]. The World Alzheimer Report 2019 states that over 50 million people are estimated to live with dementia worldwide and the number of patients will increase to 152 million by 2050. Additionally, the current yearly expenditure of dementia is estimated to reach USD 1 trillion, which will double by 2030 [2]. Currently, several drugs on the market such as tacrine, rivastigmine, huperzine A, donepezil, galantamine and memantine are used to treat AD. However, only the symptoms are mitigated and the efficacy of the drugs is not ideal, implying that a new strategy is needed for an effective AD treatment [3].

**Citation:** Cheng, L.; Osada, H.; Xing, T.; Yoshida, M.; Xiang, L.; Qi, J. The Insulin Receptor: A Potential Target of Amarogentin Isolated from *Gentiana rigescens* Franch That Induces Neurogenesis in PC12 Cells. *Biomedicines* **2021**, *9*, 581. https:// doi.org/10.3390/biomedicines9050581

Academic Editor: Lorenzo Falsetti

Received: 7 April 2021 Accepted: 11 May 2021 Published: 20 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The nerve growth factor (NGF), the first recognized neurotrophic factor that plays a very important role in the survival, growth and maintenance of neuron cells, has become a drug candidate [4]. Nevertheless, with its high polarity and large molecule weight, the NGF cannot pass through the blood-brain barrier (BBB) and is difficult to apply as a drug [5]. This finding indicates that discovering a small molecule with an NGF-mimicking activity may be a potential alternative for AD treatment.

Given the characteristic of exhibiting sympathetic neuron-like phenotypes under the stimulation of the NGF, the PC12 cell line, which is derived from rat pheochromocytoma cells, is widely used as a model to screen small molecules with NGF-mimicking activities [6]. In previous studies, under the guidance of a PC12 cell bioassay system, several small molecules with NGF-mimicking activities were isolated from traditional Chinese medicines (TCMs) such as *Gentiana rigescens* Franch, *Lindernia crustacean* and *Desmodium sambuense* and the mechanism of the action studies was also identified [7–12].

The genus *Gentiana* is a major group in the Gentianaceae family and its major constituents include iridoids and secoiridoids, which are responsible for various biological activities, and other important molecules such as essential oils, xanthones and terpenoids [13]. *G. rigescens* Franch (Jian Long Dan in Chinese), a well-known TCM that is widely distributed in the Yunnan Province, southwest China, is generally utilized for hepatitis, rheumatism, cholecystitis and inflammation treatment [14]. This TCM is praised with its anti-aging activity and cognition-improving effect in 'Sheng Nong's Herbal Classic', a classic book on TCM material medica. In previous studies, gentisides A–K, which are 11 novel neuritogenic benzoate-type molecules, were isolated from *G. rigescens* and their mixture was confirmed to alleviate the impaired memory of an AD model [7,8,15].

In the present study, a secoiridoid-type compound was isolated from *G. rigescens* Franch. The chemical structure was determined as amarogentin (AMA) (Figure 1a). AMA was previously reported by our group to be a molecule with anti-aging and neuroprotection effects by an anti-oxidative stress activity [16]. Herein, the NGF-mimicking and NGFenhancing activities of AMA were revealed and the mechanism of the action of the neurite outgrowth induced by AMA was investigated by using specific inhibitors in combination with Western blotting assays. Furthermore, the potential target was predicted using the cellular thermal shift assay (CETSA) and a small-interfering RNA (siRNA) analysis. The results indicated that AMA potentially targeted the insulin receptor and activated the PI3K/AKT and Ras/Raf/MEK/ERK signaling pathways. In addition, the GR/PLC/PKC was also involved in the neuritogenic activity of AMA in PC12 cells.

**Figure 1.** Neurogenesis effect of AMA in PC12 cells. (**a**) Chemical structure of AMA. (**b**) Percentage of PC12 cells with neurite outgrowth after treatment with AMA at different doses or AMA combined with a low dose of the NGF. (**c**) Morphological changes in PC12 cells under an inverted optical microscope at 48 h after treatment with (**i**) control (0.5% DMSO); (**ii**) NGF (40 ng/mL); (**iii**) NGF (1 ng/mL); (**iv**) AMA (3 μM); (**v**) AMA (3 μM) + NGF (1 ng/mL). (**d**) Cell viability analysis results of PC12 cells after treatment with various doses of AMA or AMA combined with the NGF. Each experiment was repeated three times. The data were expressed as a mean ± SEM. \*\*\* indicates significant differences at *p* < 0.001 compared with the negative control and ### indicates a significant difference at *p* < 0.001 compared with the 3 μM AMA group.

#### **2. Experimental Section**

#### *2.1. Chemicals and Reagents*

TrkA (k252a), GR (RU486), PI3K (LY294002), MEK/ERK (U0126) and PKC (Go6983) inhibitors, DMSO and NGF were purchased from Sigma—Aldrich Co. (St. Louis, MO, USA). The Ras inhibitor (farnesylthiosalicylic acid) was purchased from Cayman Chemical (Ann Arbor, MI, USA). The INSR (HNMPA-[AM]3), PLC (U73343) and Raf (AZ628) inhibitors were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). The TrkB inhibitor (ANA-12) was purchased from Selleck (Shanghai, China) (see the details in Supplementary Table S1). Insulin and demethylasterriquinone B1 were purchased from YEASEN Biotech Co. Ltd. (Shanghai, China) and GlpBio Technology (Shanghai, China), respectively.

#### *2.2. Preparation of the AMA*

AMA was isolated from the roots of *G. rigescens* and the chemical structure was determined by comparing the 1H NMR and 13C NMR spectra with the reported literature (Figure 1a). The detailed separation and structure elucidation steps were reported in a previous study [16].

#### *2.3. Evaluation of the Neuritogenic Activity*

The neuritogenic activity was evaluated as described in our previous paper [12]. Briefly, in each well of a 24-well microplate, around 50,000 PC12 cells were seeded and

cultured under humidified conditions with 5% CO2 at 37 °C for 24 h. After 24 h, 1 mL of serum-free Dulbecco's modified eagle medium (DMEM) containing a test sample or DMSO (0.5%) was used to replace the previous medium in each well. An NGF (40 ng/mL) was used as the positive control. Approximately 100 cells were counted thrice from a randomly selected area. Cells with a neurite outgrowth longer than the diameter of its body were counted as positive cells. The percentage of the positive cells in the selected area was regarded as the activities and the results were expressed as a mean ± SEM.

In the inhibitor test, the cells in each well of a 24-well microplate were first pretreated with 500 μL of the culture medium containing the specific inhibitor for 30 min. After this, 500 μL of the culture medium containing the sample or DMSO (0.5%) was added. The morphological changes in the cells were observed after 48 h.

In addition, the wide-type of the PC12 cell lines and corresponding mutants (PC12(ras-N17), PC12(mtGAP)) were provided by Prof. Hiroyuki Osada (RIKEN Center for Sustainable Resource Science, Japan).

#### *2.4. Analysis of the Cell Viability by Using the MTT Assay*

The cell viability was determined in accordance with the mitochondria-dependent reduction of MTT to purple formazan. Briefly, cells with AMA at concentrations of 0, 0.03, 0.3, 3 and 10 μM or AMA (3 μM) combined with a low dose of NGF were incubated for 48 h. The medium was removed carefully by aspiration. Afterward, 0.5 mL of fresh medium containing MTT (200 μg/mL) was added to each well and plates were incubated at 37 ◦C for 2 h. The medium in each well was then completely replaced with 0.2 mL DMSO to solubilize the formazan crystals. The resultant formazan was detected using a plate reader at 570 nm. All experiments were repeated at least three times.

#### *2.5. Primary Culture of Mouse Cortical Neuron Cells*

According to a previous study, primary cortical neuron cultures were prepared from the brains of C57BL/6J mice at embryonic day 17 [12]. Briefly, the cortex was digested in 0.5% trypsin in a 5% CO2 incubator at 37 ◦C for 20 min. Around 6×10<sup>4</sup> neurons were seeded into the poly-L-lysine-coated 24-well plates in a serum-free neurobasal medium (Gibco, Grand Island, NY, USA). Samples with different concentrations (AMA at 0.1, 0.3, 1 and 3 μM; 0.1 μM AMA together with 1 ng/mL NGF) were added to each well and incubated for 24 h and 0.5% DMSO and NGF were used as negative and positive control samples, respectively. After 72 h of treatment with samples, 500 nM of NeuO were added to the cultures for 1 h. Fluorescence microscopy at an excitation/emission wavelength of 430/560 nm was then used to image the neurons. Image J software (National Institutes of Health, Bethesda, MD, USA) was used to measure the relative length of the neurite outgrowth of the neurons. The results were expressed as a mean ± SEM.

#### *2.6. Western Blot Analysis*

A Western blot analysis was performed in accordance with previous studies [12]. Briefly, in each 60 mm culture dish containing 5 mL DMEM, approximately 2 × 106 PC12 cells were seeded and incubated for 24 h. For the time-dependent study of AMA, AMA (3 μM) was supplemented to the dishes, which were incubated for specific time periods. For the study of the inhibitors, AMA (3 μM) or AMA (3 μM) with a low dose of the NGF (1 ng/mL) were added to the dishes, which were then incubated for a certain period (2 h for GR, p-GR, PLC and p-PLC; 8 h for the INSR and p-INSR; 24 h for AKT and p-AKT; 48 h for ERK1/2, p-ERK1/2, PKC and p-PKC). Sodium dodecyl sulphate polyacrylamide gel electrophoresis was used to separate the proteins (15 μg) and transfer them onto a PVDF membrane. The membranes were incubated with primary antibodies and secondary antibodies (see the details in Supplementary Table S2). The antigens were visualized using a high sensitivity chemiluminescence detection kit (Beijing Cowin Biotech Company, Beijing, China). The primary antibodies used for immunoblotting were as follows: anti-insulin receptor antibody, anti-phospho-insulin receptor (Tyr1150/1151) antibody, anti-phosphoAKT (Ser473), anti-AKT antibody, anti-phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204) antibody, anti-44/42 MAPK (ERK1/2) antibody, anti-phospho-PLC γ antibody, anti-PLC antibody, anti-phospho-PKC antibody, anti-PKC antibody (Cell Signaling Technology, Boston, MA, USA), anti-phospho-GR antibody (Affinity BioReagents, OH, USA) and anti-GR antibody (Santa Cruz, CA, USA) and GAPDH antibody (Beijing Cowin Biotech Company, Beijing, China). The secondary antibodies used in this study were as follows: horseradish peroxidase-linked anti-rabbit and anti-mouse IgGs (Beijing Cowin Biotech Company, Beijing, China). The bands were quantitatively measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

#### *2.7. Cellular Thermal Shift Assay*

A CETSA was performed as described in other reports [17]. First, in 60 mm dishes containing 5 mL DMEM, 2 × 106 cells were separately added and incubated for 24 h. In each plate, AMA was added at a final concentration of 3 μM. After a continuous incubation for 8 h, cells were collected and heated at temperatures ranging from 46 ◦C to 66 ◦C. Finally, a Western blot analysis was used to detect the changes in the INSR protein and GR protein.

#### *2.8. RNA Interference*

PC12 cells were transfected with different concentrations of FAM-labelled siRNA to evaluate the transfection efficiency. Finally, 150 nM was decided as the final concentration to perform the experiment at which 90% of the transfection efficiency was obtained. The following primer sequences were used to generate siRNAs that knocked down the INSR and the negative control (Sangon Biotech Co. Ltd., Shanghai, China): for INSR-4295, sense: 5 -GUG AAG AGC UGG AGA UGG ATT-3 , anti-sense: 5 -UCC AUC UCC AGC UCU UCA CTT-3 ; for the negative control, sense: 5 -UUC UCC GAA CGU GUC ACG UTT-3 , anti-sense: 5 -ACG UGA CAC GUU CGG AGA ATT-3 .

The transfection of PC12 cells with an siRNA was performed on the basis of the manufacturer's instructions. Briefly, in each well of 24-well plates, 5 × <sup>10</sup><sup>4</sup> cells were seeded and allowed to reach 70–90% confluence in a growth medium without antibiotics one day before the transfection. SiRNA against the INSR or the negative control siRNA were then used at a concentration of 150 nM with Lipofectamine 2000 (Invitrogen) as the transfection agent. After 6 h of transfection, the fresh medium containing 3 μM AMA or AMA combined with a low dose of the NGF was used to replace the previous medium in the plates and the plate was then incubated for another 24 h. The cell morphological features were observed and recorded using an inverted microscope fitted with a camera. The percentage of the cells with a neurite outgrowth was expressed as the mean ± SEM. Finally, a Western blot analysis was used to detect the changes in the INSR protein.

#### *2.9. Statistical Analysis*

Data were presented as a mean ± SEM of three independent experiments in triplicate. Data were subjected to a one-way ANOVA and a Tukey's post hoc analysis by using the GraphPad Prism software. *p* < 0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. AMA-Induced Neuritogenic Effect in PC12 Cells and in Primary Cortical Neuron Cells*

The neuritogenic activity of AMA was first detected in PC12 cells. PC12 cells were treated with different concentrations of AMA (0.03, 0.3 and 3 μM) for 48 h. The results showed that AMA induced neurite outgrowth in PC12 cells in a dose-dependent manner. The percentages of the cells with a neurite outgrowth after treatment with 0, 0.03, 0.3 and 3 μM of AMA were 6.0% ± 0.6%, 11.3% ± 1.2%, 36.7% ± 2.3% (*p* < 0.001) and 53.0% ± 2.1% (*p* < 0.001), respectively (Figure 1b). Interestingly, AMA with a low dose of the NGF (1 ng/mL) significantly increased the percentage of PC12 cells with neurite outgrowth from 53.0% ± 2.1% to 77.3% ± 1.3% (*p* < 0.001, Figure 1b). The morphological changes in PC12 cells after treatment with 3 μM of AMA and AMA combined with 1 ng/mL of NGF are

displayed in Figure 1c. These results indicated that AMA exhibited NGF-mimicking and NGF-enhancing activities in PC12 cells. The effect of AMA in PC12 cell viability was then determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) analysis. The viabilities of PC12 cells were 96.9% ± 2.3%, 97.3% ± 2.4%, 106.1% ± 4.4% and 94.0% ± 4.9% after treatment with AMA at doses of 0.03, 0.3, 3 and 10 μM, respectively (Figure 1d). None of these concentrations produced considerable cytotoxicity as detected by the MTT assay. Furthermore, the viability of PC12 cells in the 3 μM AMA-treated group was significantly increased to 135.7% ± 13.4% after adding a low dose of the NGF (1 ng/mL, *p* < 0.001, Figure 1d). These results suggested that AMA showed no cytotoxicity at a dose of 10 μM and that the low-dose NGF could increase the cell viability of AMA in PC12 cells.

In addition, the neuritogenic effect of AMA was further estimated in the primary cortical neuron cells. As shown in Figure 2, the neurite outgrowth was increased significantly after treatment with different concentrations of AMA and AMA with the NGF. The morphological changes in the primary cortical neurons are shown in Figure 2a. The average of the neurite length and primary dendrite number are displayed in Figure 2b,c, respectively. Treatment with AMA at 0.1, 0.3 and 1 μM significantly increased the neurite length from 41.7 ± 1.2 μm to 61.7 ± 3.2 μm (*p* < 0.05), 69.3 ± 2.2 μm (*p* < 0.01) and 80.9 ± 5.7 μm (*p* < 0.001), respectively. Moreover, 0.1 μM AMA combined with 1 ng/mL of NGF increased the neurite length to a level that was comparable with the effect of the NGF at 10 ng/mL (*p* < 0.01). Collectively, these results demonstrated that AMA exhibited significant neuritogenic activity in PC12 cells and in primary cortical neuron cells.

**Figure 2.** Neurogenesis effect of AMA in primary cortical neuron cells. (**a**) Micrographs of primary cortical neuron cells at 48 h after treatment with (**i**) control (0.5% DMSO); (**ii**) NGF (10 ng/mL); (**iii**) NGF (1 ng/mL); (**iv**) AMA (0.1 μM); (**v**) AMA (0.3 μM); (**vi**) AMA (1 μM); (**vii**) AMA (3 μM); (**viii**) AMA (0.1 μM) + NGF (1 ng/mL). (**b**) Average length of neurite outgrowth of the indicated groups in the primary cortical neuron cells. (**c**) Average primary dendrite number in each group. Each experiment was repeated three times. The data were expressed as a mean ± SEM. \*, \*\* and \*\*\* indicate significant differences at *p* < 0.05, *p* < 0.01 and *p* < 0.001 compared with the negative control; #, ## indicate a significant difference at *p* < 0.05 and *p* < 0.01 compared with the 0.1 μM AMA group.

#### *3.2. Effect of AMA on the Ras/Raf/MEK/ERK Signaling Pathway*

Different neurotrophic factors such as NGF and BDNF specifically bind to the transmembrane receptors TrkA and TrkB and activate several kinases to stimulate the function of differentiation and survival in neuron cells [18,19]. Therefore, the mechanism of the action of AMA was first investigated using the inhibitors of TrkA and TrkB. However, the neurite outgrowth induced by AMA or AMA combined with the NGF did not change after the treatment with the inhibitor of TrkA, K252a (Figure 3a). Similarly, the inhibitor of TrkB, ANA-12, did not affect the NGF-mimicking or NGF-enhancing effect of AMA in PC12 cells (Figure 3b).

**Figure 3.** Effect of AMA on the Ras/Raf/MEK/ERK signaling pathway in PC12 cells. (**a**,**b**) Effect of TrkA inhibitor K252a and TrkB inhibitor ANA-12 on the neurite outgrowth induced by AMA and AMA combined with the NGF. (**c**–**e**) Effects of Ras, Raf and MEK inhibitors on the neurogenesis activity of AMA and AMA combined with the NGF. (**f**) Percentage of the neurite outgrowth induced by AMA and AMA combined with the NGF for 48 h in wide-type or Ras mutant PC12 cells. (**g**) Phosphorylation of ERK at different time points induced by AMA. The ERK phosphorylation was reduced by the inhibitor of MEK and quantified using Western blots through ImageJ software. Each experiment was repeated three times. \*\* and \*\*\* indicate significant differences at *p* < 0.01 and *p* < 0.001 compared with the negative control; ##, ### indicate a significant difference at *p* < 0.01 and *p* < 0.001 compared with the 3 μM AMA group; \$\$\$ indicates a significant difference at *p* < 0.001 compared with the AMA-combined NGF group.

Ras/Raf//MEK/ERK was believed to be the major cascade for the NGF-stimulated differentiation in PC12 cells [20]. Therefore, the effect of these signaling pathways was investigated using specific inhibitors, mutants and a Western blot analysis. As displayed in Figure 3c–e, after adding the inhibitors of Ras (farnesylthiosalicylic acid, FTA), Raf (AZ628) and MEK (U0126), the neurite outgrowth induced by AMA was significantly reduced from 53.0% ± 2.1% to 24.0% ± 1.2% (*p* < 0.001), 22.3% ± 1.2% (*p* < 0.001) and 21.0% ± 1.0% (*p* < 0.001), respectively. Similarly, the neuritogenic activity of AMA combined with the NGF was decreased by these above mentioned inhibitors from 75.3% ± 1.9% to 31.3% ± 2.4% (*p* < 0.001), 28.3% ± 0.9% (*p* < 0.001) and 29.3% ± 1.7% (*p* < 0.001), respectively (Figure 3c–e).

Furthermore, the Ras mutant types of PC12 cells including the membrane-targeted PC12(mtGAP) or the dominant inhibitory mutant PC12(rasN17) were used to detect the effect of AMA on the Ras protein. AMA or AMA combined with the NGF failed to induce the neurite outgrowth on the Ras mutant cell lines due to the inhibition of the Ras function. This finding suggested that the Ras signaling was involved in the effect of AMA (Figure 3f, Supplementary Figure S1).

The effect of AMA on ERK phosphorylation at the protein level was studied. The phosphorylation of ERK was increased from 4 h and peaked at 48 h (Figure 3g, Supplementary Figure S2). Meanwhile, the ERK phosphorylation in the AMA-treated group or the AMA with a low dose of NGF-treated group was diminished by the inhibitor of MEK, U0126 (Figure 3g). These results indicated that TrkA and TrkB were not involved in the neurogenesis effect of AMA. However, the Ras/Raf/MEK/ERK signaling pathway took an important role in the neurogenesis effect of AMA.

#### *3.3. Effect of AMA on the INSR/PI3K/AKT Signaling Pathway*

Growing evidence shows that insulin plays an important role in brain functions such as cognitive and memory improvement. Insulin binds to the INSR and activates the PI3K/AKT pathway, thereby enhancing the cell growth and survival [21]. Therefore, the inhibitor of the INSR, HNMPA-(AM)3, was used to study the mechanism of the action of AMA. After treatment with HNMPA-(AM)3, the AMA-induced neurite outgrowth was significantly decreased from 53.0% ± 2.1% to 11.7% ± 0.9% (*p* < 0.001) (Figure 4a). Moreover, the neurite outgrowth induced by AMA was reduced after treatment with the inhibitor of PI3K, LY294002, from 53.0% ± 2.1% to 22.3% ± 1.2% (*p* < 0.001) (Figure 4b). The neurite outgrowth of PC12 cells induced by AMA combined with the NGF was also decreased by HNMPA-(AM)3 and LY294002 (Figure 4a,b).

Subsequently, the phosphorylation of the INSR and AKT induced by AMA were investigated in a time-dependent manner. The INSR phosphorylation increased at 1 h and peaked at 8 h after treatment with AMA (Figure 4c, Supplementary Figure S3). Furthermore, the phosphorylation of AKT after the treatment with AMA was increased at 2 h and peaked at 24 h (Figure 4c). The increase in the phosphorylation of the INSR and downstream protein AKT and ERK in the AMA with or without 1 ng/mL of NGF were significantly decreased by HNMPA-(AM)3 (Figure 4d, Supplementary Figure S3). In addition, the phosphorylation of AKT induced by AMA and AMA combined with the NGF were also reduced by the inhibitor of PI3K, LY294002 (Figure 4e, Supplementary Figure S3). These results suggested that the INSR/PI3K/AKT signaling pathway exerted an important effect on the AMA-induced neurite outgrowth in PC12 cells.

**Figure 4.** Effect of amarogentin on the insulin receptor/PI3K/AKT signaling pathway in PC12 cells. (**a**,**b**) Effect of the insulin receptor inhibitor HNMPA-(AM)3 and PI3K inhibitor LY294002 on the neurite outgrowth induced by AMA and AMA combined with the NGF. (**c**) AMA-induced phosphorylation of the insulin receptor and AKT in a time-dependent manner and quantification of the Western blots by using ImageJ software. (**d**) Phosphorylation of the insulin receptor, AKT and ERK induced by AMA or AMA combined with the NGF was decreased by the inhibitor HNMPA-(AM)3. (**e**) Phosphorylation of AKT induced by AMA or AMA combined with the NGF was decreased by the inhibitor LY294002. Each experiment was repeated three times. \*\*\* indicates significant differences at *p* < 0.001 compared with the negative control; ### indicates a significant difference at *p* < 0.001 compared with the 3 μM AMA group and \$\$\$ indicates a significant difference at *p* < 0.001 compared with the AMA-combined NGF group.

#### *3.4. Effect of AMA on the GR/PLC/PKC Signaling Pathway*

GR has been reported to regulate a series of genes important for neuronal structure and plasticity and is involved in the neuritogenic activity in PC12 cells [22,23]. Therefore, the inhibitor of GR, RU486, was used to elucidate the mechanism of the action of AMA. The percentage of cells with a neurite outgrowth was significantly decreased from 53.0% ± 2.1% to 28.3% ± 1.2% (*p* < 0.001) after treatment with RU486 (Figure 5a). Given that the PLC/PKC signaling pathway is located at the downstream of GR and plays an important role in cell survival and differentiation [24], the inhibitors of PLC (U73343) and PKC (Go6983) were used to examine the effect of AMA. The neurite outgrowth of AMA was diminished from 53.0% ± 2.1% to 16.3% ± 2.0% and 17.7% ± 1.6% (*p* < 0.001) after the addition of U73343 and Go6983, respectively (Figure 5b,c). Similarly, the effect of AMA combined with a low dose of the NGF was also inhibited by RU486, U73343 and Go6983 (Figure 5a–c).

**Figure 5.** Effect of AMA on the GR/PLC/PKC signaling pathway in PC12 cells. (**a**–**c**) Effect of GR (RU486), PLC (U73343) and PKC (Go6983) inhibitors on the neurite outgrowth induced by AMA and AMA combined with the NGF. (**d**) AMAstimulated phosphorylation of GR, PLC and PKC proteins in a time-dependent manner and the quantification of Western blots by using ImageJ software. (**e**) Phosphorylation of GR, PLC and PKC induced by AMA or AMA combined with the NGF reduced by the corresponding inhibitors and quantified using Western blots through ImageJ software. Each experiment was repeated three times. \*\*\* indicates significant differences at *p* < 0.001 compared with the negative control; ### indicates a significant difference at *p* < 0.001 compared with the 3 μM AMA group and \$\$\$ indicates a significant difference at *p* < 0.001 compared with the AMA-combined NGF group.

The phosphorylation of GR/PLC/PKC was then determined at the protein level by using a Western blot analysis. The GR phosphorylation peaked at 2 h and was reduced by RU486 (Figure 5d,e, Supplementary Figure S4). The PLC phosphorylation was increased from 1 h, peaked at 2 h and decreased by the inhibitor of PLC, U73343 (Figure 5d,e). Furthermore, the phosphorylation of PKC peaked at 48 h and was reduced by Go6983, the inhibitor of PKC (Figure 5d,e). The AMA combined with the NGF group changed in a similar way (Figure 5d,e). These results demonstrated that the AMA-induced neuritogenic activity in PC12 cells was related to the GR/PLC/PKC signaling pathway.

#### *3.5. Identification of the Target Protein for AMA by Using siRNA Analysis and CETSA*

Considering that TrkA and TrkB are not involved in the neurogenesis effect of AMA, we predicted that the INSR or GR protein might be the potential target of AMA. Given that the inhibition effect of the INSR for AMA was stronger than that of GR, the INSR was first considered as the potential target of AMA. The 5-carboxyfluorescein (FAM)-labelled siRNA was initially used to confirm the optimal transfection concentration of siRNA and whether AMA targeted the INSR. Approximately 90% of the PC12 cells produced fluorescence after treatment with 150 nM of the FAM-labelled siRNA and 150 nM of the INSR siRNA was used to perform the transfection (Supplementary Figure S5). The INSR siRNA was transfected into PC12 cells for 6 h and treated with 3 μM AMA or AMA combined with the NGF. After the treatment of the PC12 cells with the INSR siRNA, the percentage of cells with a neurite outgrowth induced by AMA with or without a low dose of the NGF for 48 h was significantly decreased (Figure 6a,b). In addition, the total and the phosphorylation protein levels of the INSR were significantly decreased by the treatment with the INSR siRNA regardless of the AMA treatment (*p* < 0.001, Figure 6c, Supplementary Figure S6). Hence, these results indicated that the INSR might be the target protein of AMA.

A CETSA was used to discover the target protein of molecules on the basis of the thermal stabilization of proteins upon ligand binding [17]. Therefore, a CETSA was used to detect the binding correlations between the INSR and AMA to further confirm the potential target of AMA. After treating the PC12 cells with dimethyl sulfoxide (DMSO) or AMA and heating at temperature ranging from 46 ◦C to 66 ◦C, the immunoblotting analysis was conducted using a specific antibody for the INSR. The results suggested a significant thermal stabilization of the INSR protein upon AMA treatment (Figure 6d, Supplementary Figure S6). At the same time, the change of GR at the protein level was detected using the same method. As expected, the GR protein did not show the thermal stability-shifted effect after the AMA treatment (Figure 6e, Supplementary Figure S6). Furthermore, other known insulin agonists such as insulin and demethylasterriquinone B1 (DB1) were selected to detect whether they exhibited a similar neurogenesis as AMA in PC12 cells [25]. The results indicated that both showed a significant NGF-mimicking and NGF-enhancing activity in the PC12 cells (Figure 6f, Supplementary Figure S7). These results indicated that AMA might target the INSR to produce the NGF-mimicking activity.

**Figure 6.** Target prediction of AMA in PC12 cells by using siRNA and a CETSA assay. (**a**) Microphotographs of PC12 cells after treatment with siRNA and AMA or AMA combined with the NGF: (**i**) negative control siRNA, control (0.5% DMSO); (**ii**) negative control siRNA, AMA (3 μM); (**iii**) negative control siRNA, AMA (3 μM) + NGF (1 ng/mL); (**iv**) insulin receptor siRNA, control (0.5% DMSO); (**v**) insulin receptor siRNA, AMA (3 μM); (**vi**) insulin receptor siRNA, AMA (3 μM) + NGF (1 ng/mL). (**b**) Percentage of cells with a neurite outgrowth after treatment with siRNA and AMA or AMA combined with the NGF. (**c**) Western blot analysis for the insulin receptor after transfection with negative siRNA or insulin receptor siRNA and treatment with AMA or AMA combined with the NGF. Cells were transfected with Lipofectamine 2000 and 150 nM siRNA for 6 h and treated with AMA or AMA combined with the NGF. (**d**,**e**) CETSA of PC12 cells on the insulin receptor or GR protein and corresponding fitting curves. (**f**) Neuritogenic activity of insulin and demethylasterriquinone B1 in PC12 cells. \*\*\*, ### and \$\$\$ indicate significant differences at *p* < 0.001 compared with the corresponding groups.

#### **4. Discussion**

Aging is a major risk factor for age-related diseases such as Parkinson's disease and AD [26]. We speculated that if we prevent or delay aging, we can prevent the occurrence of AD or cure AD. Our laboratory began to screen small anti-aging molecules from food and TCMs ten years ago to verify this hypothesis. To date, we have found more than 30 anti-aging compounds with different types of chemical structures such as sterols, benzoquinones, phenols and terpenes [27–30]. Furthermore, we have indicated that cucurbitacin B with an anti-aging effect can improve the memory of APP/PS1 mice via the target cofilin and the regulation of GR signaling pathways [23,31]. These results indicate that anti-aging substances may prevent and treat AD.

*G. rigescens* Franch is a TCM used to treat hepatitis, rheumatism, cholecystitis and inflammation in China [14]. In our previous study, we discovered gentisides A–K with a novel NGF-mimicking effect from the nonpolar extract of this plant and indicated that a mixture of benzoates could alleviate the impaired memory of AD model mice induced by scopolamine [15]. We have also focused on the water layer of *G. rigescens* Franch to isolate active molecules under the guidance of PC12 cells and a yeast replicative lifespan assay to understand whether the small molecules of the polar part have the same function. We have found that AMA produces anti-aging effects on yeasts and neuron protection in PC12 cells via anti-oxidative stress [16]. In the present study, we used PC12 cells and primary cortical neuron cells to investigate the neurogenesis effect of AMA. The morphological changes of PC12 cells and primary cortical neuron cells after AMA treatment suggested that AMA had a neurogenesis effect on PC12 cells and primary cortical neuron cells (Figures 1 and 2). These results were consistent with those of our previous reports [12,23].

The target protein identification has an important role in drug development and can provide strong evidence for the elucidation of the mechanism of the action, safety evaluation and targeted treatment of a disease [32]. Therefore, we first focused on the target protein discovery of AMA to perform deep research with specific inhibitors, siRNA, a CETSA and a Western blot analysis. The results of the specific inhibitors for TrkA, TrkB, INSR, GR, PI3K, PLC, PKC and MEK and the Western blot analysis in Figures 3–6 indicated that AMA induced neuritogenic activity in PC12 cells by activating the INSR and regulating the PI3K/AKT/Ras/Raf/ERK and GR/PLC/PKC signaling pathways. Interestingly, the mechanism of the action of AMA for its NGF-mimicking effect was different from that of previously reported compounds (such as ABG-001, lindersin B, 3beta,23,28-trihydroxy-12-oleanene 3beta-caffeate and CuB). Tetradecyl 2,3-dihydroxybenzoate (ABG-001) was designed and synthesized as a lead compound in accordance with the gentiside series to induce neurogenesis in PC12 cells by the IGF-1R/PI3K/MAPK signaling pathway [9,10]. Lindersin B from *L. crustacea* induced neuritogenic activity through the activation of the TrkA/PI3K/ERK signaling pathway [11]. 3beta,23,28-trihydroxy-12-oleanene 3betacaffeate from *D. sambuense* induced neurogenesis in PC12 cells mediated by the ER stress and BDNF-TrkB signaling pathways [12] and CuB induced neuritogenic activity by targeting cofilin and regulating the GR TrkA signaling pathways [23]. These molecules possess different structures but exhibit neurogenesis effects by activating various related signaling pathways. AMA was the first compound we discovered to target the INSR for the NGFmimicking activity in PC12 cells. These results provided insights into the combination that the use of these molecules may have in increasing the therapy effect for AD.

TrkA and TrkB are specific transmembrane receptors that bind to neurotrophic factors such as NGF and BDNF [18,19]. Therefore, the effects of these two proteins were investigated. We found that TrkA and TrkB were not involved in the neurogenesis effect of AMA (Figure 3b). We focused on the INSR and GR to determine the target protein. The results of the INSR knockdown experiment, a CETSA and a Western blot analysis for the INSR and GR in Figure 6 revealed that the INSR was the potential target protein of AMA. Furthermore, known insulin agonists including insulin and DB1 showed similar neurogenesis effects as AMA in PC12 cells, which confirmed the INSR as the potential target of AMA (Figure 6). It was different from the target proteins of CuB, cofilin and

3beta,23,28-trihydroxy-12-oleanene 3beta-caffeate and ER stress [11,23]. AMA may have effects for diabetes and inflammation because of the involvement of insulin and GR signaling pathways [33,34].

In conclusion, AMA from *Gentiana rigescens* Franch showed significant neuritogenic activity in PC12 cells and in primary cortical neuron cells. The neuritogenic activity induced by AMA in PC12 cells was through the targeting of the INSR and the regulation of the PI3K/AKT/Ras/Raf/ERK and GR/PLC/PKC signaling pathways (Figure 7). This study indicated the potential applications of AMA for its neurogenesis effect and provided evidence for the treatment of neurodegenerative diseases and anti-aging. Furthermore, the structure-activity relationship of AMA should be studied to discover the novel leading compounds and elucidate the underlying mechanism in animal levels and also applied to clinical trials.

**Figure 7.** Proposed mechanism of the action of AMA in the neuritogenic activity in PC12 cells.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/biomedicines9050581/s1, Figure S1: Differentiation of PC12 cells expressing dominant negative Ras (PC12(rasN17)) or membrane-targeted GAP (PC12(mtGAP)) induced by AMA or AMA combined with NGF, Figure S2: Origin data of Western blot analysis in Figure 3g, Figure S3: Origin data of Western blot analysis in Figure 4c–e, Figure S4: Origin data of Western blot analysis in Figure 5d,e, Figure S5: Microphotograph of PC12 cells after transfection with different concentrations of FAMsiRNA (50 nM, 100 nM and 150 nM), Figure S6: Origin data of Western blot analysis in Figure 6c–e, Figure S7: Microphotograph of PC12 cells after treatment with insulin and Demethylasterriquinone B1, Table S1: List of inhibitors used in this study, Table S2: List of antibodies used in this study.

**Author Contributions:** L.C. performed the bioassay, mechanism study and data analysis as well as writing the original draft; T.X. provided assistance with editing; H.O., M.Y., L.X. and J.Q. contributed to designing the overall research strategy, supervision and revision of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** Publication of this paper. This research was funded by the National Key R&D Program of China (Grant No. 2017YFE0117200, Grant No. 2019YFE0100700) and the National Natural Science Foundation of China (Grant No. 21877098, 21661140001).

**Institutional Review Board Statement:** This study did not involve humans or animals experiments.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All figures and data used to support this study are included within this article.

**Acknowledgments:** This work was financially supported by the National Key R&D Program of China (Grant No. 2017YFE0117200, Grant No. 2019YFE0100700) and the National Natural Science Foundation of China (Grant No. 21877098, 21661140001). This work was inspired by the JSPS Asian Chemical Biology Initiative. The authors thank Young-Tae Chang (Pohang University of Science and Technology) for providing the NeuO reagent. We thank Julius Adam V. Lopez and Makoto Muroi for valuable comments on the manuscript.

**Conflicts of Interest:** The authors declare that there is no conflict of interest regarding publication of this paper.

#### **Abbreviations**

Alzheimer's disease (AD); amarogentin (AMA); blood-brain barrier (BBB); cellular thermal shift assay (CETSA); demethylasterriquinone B1 (DB1); dimethyl sulfoxide (DMSO); extracellular signalregulated kinase (ERK); glucocorticoid receptor (GR); insulin receptor (INSR); mitogen-activated protein kinase (MEK); nerve growth factor (NGF); phospholipase C (PLC); protein kinase B (AKT); protein kinase C (PKC); small-interfering RNA (siRNA); traditional Chinese medicines (TCMs).

#### **References**


## *Article* **Brain-Specific Gene Expression and Quantitative Traits Association Analysis for Mild Cognitive Impairment**

**Shao-Xun Yuan, Hai-Tao Li, Yu Gu and Xiao Sun \***

State Key Laboratory of Bioelectronics, School of Biological Science and Medical Engineering, Southeast University, Nanjing 210096, China; 230159460@seu.edu.cn (S.-X.Y.); 230169443@seu.edu.cn (H.-T.L.); 230198583@seu.edu.cn (Y.G.)

**\*** Correspondence: xsun@seu.edu.cn

**Abstract:** Transcriptome–wide association studies (TWAS) have identified several genes that are associated with qualitative traits. In this work, we performed TWAS using quantitative traits and predicted gene expressions in six brain subcortical structures in 286 mild cognitive impairment (MCI) samples from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. The six brain subcortical structures were in the limbic region, basal ganglia region, and cerebellum region. We identified 9, 15, and 6 genes that were stably correlated longitudinally with quantitative traits in these three regions, of which 3, 8, and 6 genes have not been reported in previous Alzheimer's disease (AD) or MCI studies. These genes are potential drug targets for the treatment of early–stage AD. Single–Nucleotide Polymorphism (SNP) analysis results indicated that cis–expression Quantitative Trait Loci (cis–eQTL) SNPs with gene expression predictive abilities may affect the expression of their corresponding genes by specific binding to transcription factors or by modulating promoter and enhancer activities. Further, baseline structure volumes and cis–eQTL SNPs from correlated genes in each region were used to predict the conversion risk of MCI patients. Our results showed that limbic volumes and cis–eQTL SNPs of correlated genes in the limbic region have effective predictive abilities.

**Keywords:** subcortical structure; quantitative trait; longitudinal stably correlated; mild cognitive impairment; conversion

#### **1. Introduction**

Alzheimer's disease (AD) is a progressive and irreversible neurodegenerative disorder, accounting for more than 75% of all dementia events worldwide [1]. Approximately 35% of individuals over 80 years of age suffer from AD around the world [2]. Mild Cognitive Impairment (MCI) is the preclinical stage of AD and is clinically heterogeneous [3]. Genome–wide association studies (GWAS) have identified several susceptible single nucleotide polymorphisms (SNPs) for AD [4–7] and MCI [7]. However, GWAS can be used to understand which SNPs are associated with traits but cannot explain how the SNPs affect the traits. SNPs are likely to influence traits by regulating gene expression [8,9]. On the other hand, gene expression may be regulated by causal SNPs but not by the SNP with the lowest *p*-value within a linkage disequilibrium block.

Transcriptome sequencing can be used to study associations between whole transcription levels and traits in a specific tissue. Howevr, sampling for transcriptome sequencing is costly and difficult. Gusev et al. [10] proposed a new strategy, leveraging expression prediction to perform a transcriptome–wide association study (TWAS) to identify significant trait–expression associations. TWAS first fits tissue–specific models using reference data with both SNP genotype data and gene expression data available. Then, these models are used to predict gene expression in a new dataset with genotype data available. Finally, the predicted gene expression in each tissue is associated with corresponding traits. TWAS has been proved as an effective method to identify gene associations between gene expression and traits in specific tissues [11].

**Citation:** Yuan, S.-X.; Li, H.-T.; Gu, Y.; Sun, X. Brain-Specific Gene Expression and Quantitative Traits Association Analysis for Mild Cognitive Impairment. *Biomedicines* **2021**, *9*, 658. https://doi.org/ 10.3390/biomedicines9060658

Academic Editor: Lorenzo Falsetti

Received: 24 May 2021 Accepted: 4 June 2021 Published: 8 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Several TWAS studies have identified multiple novel susceptibility genes for AD by combining Genotype–Tissue Expression Project (GTEx) gene expression models and genotype data of AD. Raj et al. [12] identified 21 genes with significant associations with AD in two cohorts, 8 of which were were novel. Hao et al. [13] combined TWAS and data from the International Genomics of Alzheimer's Project (IGAP) cohort and identified 29 potential disease–causing genes, 21 of which were new. Jung et al. [14] combined tissue specifically predicted gene expression levels and polygenic risk score from 207 AD cases and 239 cognitively normal controls and found that the inclusion of polygenic risk score and gene expression features provided better performance in AD classification. Gerring et al. [15] performed a multi–tissue TWAS of AD and observed associated genes in brain and skin tissue.

The aim of our study was to identify genes potentially related with specific brain structure quantitative traits in MCI samples, reveal possible relationships with biological mechanisms, and use them for conversion analyses. We performed TWAS between predicted gene expression and longitudinal quantitative traits in six brain subcortical structures to identify longitudinally stable correlated genes for MCI. First, gene expression prediction models provided by GTEx [16] were used to predict gene expression in amygdala, hippocampus, accumbens area, caudate, putamen, and cerebellum using 286 MCI samples from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. Second, the expression of genes in the above six structures was correlated with baseline and 12–month follow–up quantitative traits in the corresponding structures. Overlapping genes in baseline and 12–month follow–up were considered as longitudinally stable correlated genes in each structure. Third, fine–mapping analyses were performed on these longitudinally stable correlated genes and corresponding cis–eQTL SNPs to identify the potential regulation mechanisms. Finally, we further investigated the potentials of baseline quantitative traits and gene expression–determined cis–eQTL SNPs of longitudinally stable correlated genes for conversion analysis of MCI samples.

#### **2. Materials and Methods**

Data used in the preparation of this article were obtained from the ADNI database (adni.loni.usc.edu). ADNI was launched in 2003 as a public–private partnership, led by the Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI is to test whether findings from serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of MCI and early AD.

#### *2.1. Ethics Statement*

We used the ADNI subject data collected from 50 clinic sites. The ADNI study was conducted according to Good Clinical Practice guidelines, US 21CFR Part 50—Protection of Human Subjects, and Part 56—Institutional Review Boards (IRBs)/Research Ethics Boards (REBs)—and pursuant to state and federal HIPAA regulations. Written informed consent was obtained from all participants after they had received a complete description before protocol–specific procedures were carried out based on the 1975 Declaration of Helsinki. IRBs were constituted according to applicable State and Federal requirements for each participating location. The protocols were submitted to appropriate Boards, and their written unconditional approval obtained and submitted to Regulatory Affairs at the Alzheimer's disease Neuroimaging Initiative Coordinating Center (ADNICC) prior to commencement of the study. We have obtained permission to use data from ADNI, and the approval date is 25 November 2019.

#### *2.2. Samples*

A total of 819 samples of European ancestry were recruited by the ADNI cohort, and 757 of them were run on the Human610–Quad BeadChip (Illumina Inc., San Diego, CA, USA) for genotyping. Among these 757 samples, 286 MCI samples were MPRAGE N3–Scaled

sMRI data available at both baseline and 12–month follow–up. MRI images marked with "N3" and "scaled" in the file name were downloaded from the ADNI dataset; these files underwent B1 bias field correction and N3 intensity nonuniformity correction [17]. The following information was also collected from the the ADNI dataset for 286 selected samples: gender, age, education years, Clinical Dementia Rating Sum of Boxes (CDR–SB) score, Mini– Mental State Examination (MMSE) score, Functional Assessment Questionnaire (FAQ) and Alzheimer Disease Assessment Scale scores (ADAS, version 11, 13 and Q4).

#### *2.3. Genotype and Image Data Pre–Processing*

PLINK 1.9 software [18] (Boston, MA, USA) was used for quality control of genotype data for 286 MCI samples. SNPs with a call rate smaller than 90%, Minor Allele Frequency (MAF) smaller than 10%, or deviations from the Hardy–Weinberg Equilibrium (5 × <sup>10</sup><sup>−</sup>7) were removed from the original genotype data. After quality control, imputation was performed using impute2 software [19]. After quality control and imputation, 28,571,732 SNPs were retained from the 286 MCI samples.

Freesurfer 6.0 software (Boston, MA, USA) was applied for automated segmentation and volume measurement of subcortical structures and total intracranial volume (ICV) for all selected MCI samples from MRI image data at baseline and 12–month follow–up. Left and right volumes from the same structure were summed. Adjustments were performed for subcortical structure volumes using gender, age, and ICV, using the following formulas:

$$QT = a \ast AGE + b \ast GENDER + c \ast ICV + d \tag{1}$$

$$\text{C}\,\text{Q}T\_{\text{adj}} = a\*A\,\text{G}E\_{\text{mean}} + b\*\,\text{G}N\,\text{D}ER\_{\text{mean}} + c\*I\,\text{C}\,\text{V}\_{\text{mean}} + d + r \tag{2}$$

*QT* and *QTadj* represent raw quantitative trait volumes extracted using Freesurfer and adjusted quantitative trait volumes of a subcortical structure across the 286 MCI samples. *AGE*, *GENDER*, and *ICV* represent age, gender, and ICV of all MCI samples, while *AGEmean*, *GENDERmean*, and *ICVmean* represent mean age, mean gender, and mean ICV across all MCI samples; *d* represents error, while *r* represents residual. We first calculated coefficients of age (*a*), gender (*b*), and ICV (*c*) from a mixed linear regression model (Equation (1)). Then, adjusted volumes were calculated using Equation (2). Adjusted volumes of each subcortical structure were used as quantitative traits.

#### *2.4. Correspondences among GTEx Models, Anatomical Regions, and Freesurfer–Defined Structures*

We defined correspondences the GTEx models, anatomical regions, and freesurfer– defined structures. The PredictDB Data Repository provides 49 gene–predicted models based on GTEx data (www.gtexportal.org, accessed on 5 September 2020), of which 13 are brain–related gene expression predictive models. Freesurfer software provides 35 brain subcortical structures according to the Desikan–Killiany (DK) atlas template. In our study, 6 one–to–one corresponding gene expression predictive model–subcortical structure pairs were selected and assigned to three regions (Table 1).

**Table 1.** Corresponence of GTEx models, anatomical regions, and subcortical structures.


GTEx models were downloaded from http://predictdb.org/ (accessed on 5 September 2020); Subcortical structures were segmented by Freesurfer software according to the Desikan–Killiany (DK) atlas template.

#### *2.5. Correlation between Predictive Gene Expression and Quantitative Traits*

We utilized the PrediXcan software to predict gene expression based on the genotype data of all MCI samples. PrediXcan establishes a linear prediction model of gene expression in a dataset with both SNP genotype data and gene expression available (GTEx version 8) using a multivariate adaptive shrinkage regression (mashr) approach. Brain–specific gene expressions in 6 structures were predicted by combined prediction models and MCI genotype data. Brain–specific gene expression was determined by corresponding cis–eQTL SNPs from the LD reference files for the corresponding model in PredictDB Data Repository (http://predictdb.org/) (accessed on 5 September 2020).

We annotated the chromosomal locations of cis–eQTL SNPs in the corresponding genes using SNPnexus database [20] (accessed on 15 May 2021). Regulatory information for cis–eQTL SNPs were annotated using HaploReg database [21] (accessed on 15 May 2021) and RegulomeDB database [22] (accessed on 15 May 2021). HaploReg is a web–based tool for annotating SNPs, including chromosome number, protein binding, motif change. RegulomeDB can be used to predict whether an SNP affects transcription factor binding and gene expression. RegulomeDB provides a rank score of SNP, with a low score representing strong evidence of regulatory function. We used VARAdb database [23] to annotate the location of cis–eQTL SNPs in promoter or enhancer regions of corresponding genes (accessed on 15 May 2021). VARAdb determines promoters based on the basic gene annotation file release 33 from GENCODE (2 kb upstream of transcription start site) and determines super enhancers from 542 H3K27ac ChIP–seq samples from the human super– enhancer database [24].

Pearson correlation coefficients were used to calculate correlations between predicted gene expression and adjusted subcortical structure volumes in Table 1. The correlation matrix heatmaps were constructed using the *pheatmap* package (version 1.0.12) in R.

#### *2.6. Conversion Analysis Based on Quantitative Traits and SNPs*

The performances of quantitative traits and cis–eQTL SNPs were further evaluated in terms of their ability to determine the "time to progression" from MCI to AD via Kaplan–Meier analysis. For this evaluation of MCI samples in the ADNI dataset, the midpoint between the first follow–up with an AD diagnosis and the last follow–up without an AD diagnosis was considered as the conversion time point for MCI samples. The longest follow–up time was collected for samples who did not convert to AD, and these samples were regarded as non–conversion MCI samples [25]. First, quantitative trait volumes or genotypes of cis–eQTL SNPs were used as feature vectors to represent MCI samples and to calculate distances across all MCI samples through Euclidean distance. Hierarchical clustering was completed using stats package in R to cluster MCI samples into two subgroups. Then, we applied the "survfit" function in the *survival* package (version 3.2–7) in R and plotted Kaplan–Meier curves for the two subgroups. The median conversion time of MCI samples in the two subgroups was calculated; the group with a high medium time was regarded as a low–risk group, while the group with a low medium time was regarded as a high–risk group. A log rank test with a *p*-value less than 0.05 was considered statistically significant for median conversion time between risk groups [26].

#### **3. Results**

#### *3.1. Sample Characteristics*

The baseline characteristics of 286 MCI samples and their association with AD are shown in Table 2. The samples were obtained from patients with a mean (SD) age of 74.85 (6.97) years; 33.9% were female, 18.5% had less than 12 years of education. In accordance with their MCI diagnosis, the average scores of most neuropsychological tests were in the normal–to–low range. A total of 167 (58.4%) study participants converted to probable AD over a mean (SD) follow–up period of 25.05 (21.76) months. Of the 119 who did not convert, 45 had less than 36 months of follow–up data, whereas 71 were followed for more than 36 months. Three samples had only one follow–up visit.


**Table 2.** Baseline characteristics of 286 MCI samples.

MCI, Mild Cognitive Impairment; CDRSB, Clinical Dementia Rating Sum of Boxes; MMSE, Mini–Mental State Examination; FAQ, Functional Assessment Questionnaire; ADAS, Alzheimer Disease Assessment Scale scores.

#### *3.2. Identification of Quantitative Traits–Related Genes*

PrediXcan software was applied to predict gene expression by integrating GTEx gene expression prediction models and ADNI genotype data. Correlations between quantitative traits and predicted gene expressions were computed by Pearson correlation across all selected samples at baseline and 12–month follow–up. The correlation heatmaps for all six structures at baseline and 12–month follow–up are shown in Figure 1. Gene–quantitative traits pairs with a correlation coefficient greater than 0.2 and lower than −0.2 are displayed in the heatmaps. Genes associated with quantitative traits were distinct across all structures at baseline (Figure 1A) and 12–month follow–up (Figure 1B).

We evaluated the overlapping correlated genes at baseline and 12–month follow– up. Table 3 shows overlapping genes associated with structure volumes at baseline and after 12 months across all MCI samples. In the limbic region, 10 and 8 amygdala–specific expressed genes were correlated with baseline and 12–month amygdala volume, while 9 and 10 hippocampal–specific expressed genes were correlated with baseline and 12–month hippocampal volume. Four amygdala–specific expressed genes were overlapping between baseline and 12–month follow–up, while five hippocampal–specific expressed genes were overlapping between baseline and 12–month follow–up. In addition, we identified 15 overlapping genes with basal ganglia structures, including accumbens area, caudate and putamen, and 9 overlapping genes with the cerebellum. We considered these overlapping genes as stably correlated longitudinally with the corresponding quantitative traits. We used GeneCards database to annotate these genes, to define whether they were related to AD or MCI. We found that six, seven, and three genes were related to AD or MCI, while three (*NOXRED1*, *MYL6B*, and *FAM162B*), eight (*RELCH*, *IRX3*, *RELL1*, *TMEM50A*, *SETD4*, *TMEM253*, *HPS3*, *SLC26A10*), and six (*SLC6A16*, *SLC10A5*, *ENSG00000272542*, *LINC00958*, *FCGRT*, *TRPM4*) genes were potentially correlated to AD or MCI in limbic region, basal ganglia region, and cerebellum region, respectively. We summarized the potential biologic mechanisms of all these longitudinally stable correlated genes (Table S1). Genes in the limbic region are involved in energy metabolism, regulation of cell growth, apoptosis, migration and invasion, and synaptic plasticity. Genes in the basal ganglia region are involved in the inflammatory response and signal transduction. Genes in the cerebellum region are involved in signal transduction, material transport, lipid metabolism, neuronal migration, and neuritic plaques.

**Figure 1.** Heatmaps of correlations between predicted gene expressions and quantitative traits at baseline (**A**) and 12–month follow−up (**B**). Correlations with coefficient *r* greater than 0.2 and less than −0.2 are displayed in the heatmaps. The red color represents positive correlations, while the blue color indicates negative correlations in heatmaps. Column annotations represent brain structures for correlation analyses. For annotations, limbic region, basal ganglia region, and cerebellum region are displayed in green, sky blue, and orange, respectively.

#### *3.3. Fine-Mapping Analyses of Gene Expression-Determined Cis-eQTL SNPs*

We annotated the 56 gene expression–determined cis–eQTL SNPs of all longitudinally stable correlated genes (Table 3) using SNPnexus, HaploReg, RegulomeDB, and VARAdb databases. In this study, 12, 26, and 18 SNPs were found in to 9, 15, and 9 longitudinally stable correlated genes in the limbic region, basal ganglia region, and cerebellum region, respectively. We annotated the locations of these SNPs in the corresponding genes using SNPnexus (Table S2). Among these 56 cis–eQTL SNPs, 54 SNPs (54/56, 96.4%) were in the intronic or untranslated regions of the various transcript isoforms of the genes. According to the annotation from the HaploReg database (Table S3), a total of 49 SNPs (49/56, 87.5%) can affect the corresponding genes through motifs changes, while 25 can affect the corresponding genes through proteins binding (25/56, 44.6%). According to the annotation from RegulomeDB (Table S3), 41 SNPs (41/56, 73.2%) had RegulomeDB rank scores smaller than 4, indicating transcription factor binding and location within a region of DNase hypersensitivity. We used the VARAdb database to annotate whether these cis–eQTL SNPs were located in promoters or enhancers of the corresponding genes. We found that 32 SNPs (32/56, 57.1%) were in the promoters of their corresponding genes (Table S4), while 22 SNPs were located in the forward strand, and 10 in the reverse strand. In addition, 25 SNPs (25/56, 44.6%) were enriched in super enhancers, with the corresponding genes being the closest genes (distance between the gene and the SNP was less than 1000 kb), while 13 SNPs (13/56, 23.2%) were enriched in super enhancers with the corresponding genes being the proximal genes (distance between the gene and the SNP was less than 50 kb) (Table S5). We inferred that cis–eQTL SNPs regulate the expression of the corresponding genes by affecting promoters or enhancers.


**Table 3.** Overlapping quantitative traits-correlated gene sets between baseline and 12-month follow-up in six subcortical structures.

N, number of correlated genes at baseline and 12-month follow-up; n, number of overlapping genes between baseline and 12-month follow-up (positive/negative correlation); Overlapping genes, overlapping genes between baseline and 12-month follow-up; SNPs, gene expression-determined cis-eQTL SNPs; Ranks, ranks of overlapping genes at baseline and 12-month follow-up; Annotations, annotations were performed using https://www.genecards.org/ (accessed on 20 March 2021). The lists of cis-eQTL SNPs of the corresponding genes were download from the LD reference file in PredictDB Data Repository (http://predictdb.org/) (accessed on 5 September 2020); SNPs with superscripts "a" and "b" indicate that these SNPs are in the promoters and enhancers of the corresponding genes, respectively.

> To evaluate whether these 56 SNPs were associated with the volume of the corresponding subcortical structures, we performed quantitative traits–based GWAS analysis using SNPs directly, instead of using predicted gene expression (Figure 2). Among five cis–eQTL SNPs for longitudinally stable correlated genes in the amygdala, four SNPs (80.0%) were significantly associated only with amygdala volume at baseline and 12–month follow–up. Among seven cis–eQTL SNPs (71.4%) for longitudinally stable correlated genes in the

hippocampus, five SNPs were significantly associated only with hippocampus volume at baseline and 12–month follow–up. In the basal ganglia region and cerebellum region, 58.3% and 71.4% of SNPs were significantly associated only with corresponding quantitative traits (Figures S1 and S2). The results indicated that the correlations between quantitative traits and predicted gene expression were reasonable. On the basis of our results, we speculated that these cis–eQTL SNPs can affect both promoters and enhancers, as well as the binding of transcription factors, which may alter the expression of their target genes.

**Figure 2.** Bar plots of associations between 12 SNPs in the limbic region and 6 subcortical structures. (**A**) Five SNPs gene expression-determined SNPs in the amygdala. (**B**) Seven SNPs gene expression-determined SNPs in the hippocampus. The *X*-axis reports six subcortical structures (amygdala, hippocampus, accumbens area, caudate, putamen, and cerebellum cortex) at baseline and 12-month follow-up. The *Y*-axis presents the *p*-value (−log10) of the association based on quantitativetrait GWAS. The blue horizontal line represents <sup>−</sup>log10 (0.05), while the red horizontal line represents <sup>−</sup>log10 (5 <sup>×</sup> <sup>10</sup><sup>−</sup>4).

#### *3.4. Conversion Analysis Based on Quantitative Traits and SNPs*

We used the baseline volumes of limbic region, basal ganglia region, and cerebellum region as quantitative traits and gene expression–determined cis–eQTL SNPs of longitudinal stably correlated genes in each region to perform a conversion analysis for the MCI samples. First, the MCI samples were clustered into two subgroups using quantitative traits or SNPs. Hierarchical clustering was applied based on the Euclidean distance in the *stats* R package (v4.0.4). Then, we compared the conversion times and performed Kaplan–Meier analyses between the two MCI subgroups. Figure 3 shows the Kaplan–Meier plots for the two groups using quantitative traits and SNPs. The volumes of the structures in the limbic region and cis–eQTL SNPs of longitudinally stable correlated genes in the limbic region showed effective predictive abilities (Figure 3A,B), while this was not true for basal ganglia and cerebellum (Figure 3C–F).

We calculated the percent of conversion and non–conversion of MCI samples in risk groups defined by quantitative traits and SNPs in the limbic region. Chi–square tests were used to determine between–group differences in the conversion and non–conversion of MCI samples. As shown in Figure 4, when using quantitative traits and SNPs, the high–risk groups and low–risk groups had significantly different proportions of conversion and non–

conversion, with the high–risk groups showing significantly higher percentages of conversion than the low–risk groups (quantitative traits, 66.7% vs. 38.2%; SNPs: 64.9% vs. 44.4%).

**Figure 3.** Survival curves of the two mild cognitive impairment (MCI) subgroups based on baseline volumes and cis-eQTL SNPs of limbic region (**A**,**B**), basal ganglia (**C**,**D**), cerebellum (**E**,**F**). Confidence intervals are indicated by shaded regions. The blue line represents the low-risk group, while the yellow line represents the-high risk group. Median means the median time (months) of conversion of MCI samples in the two subgroups.

**Figure 4.** Percent of conversion mild cognitive impairment (MCI) (cMCI) and non-conversion MCI (ncMCI) samples in the high-risk group and low-risk group using quantitative traits (**A**) and SNPs derived from longitudinally stable correlated genes (**B**) in the limbic region. P, *p*-value of the chi-square test.

#### **4. Discussion**

In this study, we performed transcriptome–wide association analyses between gene expressions and longitudinal quantitative traits in specific brain subcortical structures to identify longitudinally stable correlated genes for MCI. Combining gene expression prediction models generated from GTEx data and quantitative traits extracted from T1– MRI data, we identified 9, 15, and 6 genes correlated with limbic region, basal ganglia region, and cerebellum region, of which 3, 8, and 6, respectively, have not been reported in previous studies. We also performed quantitative traits–based GWAS analysis using SNPs. Most SNPs derived from previously correlated genes were directly associated with the corresponding quantitative traits, indicating that those correlations between quantitative traits and predicted gene expressions were reasonable. Furthermore, quantitative traits and gene expression–determined cis–eQTL SNPs of longitudinally stable correlated genes were used for conversion analysis of the MCI samples. We found that limbic region structure volumes and cis–eQTL SNPs derived from longitudinally stable correlated genes in the limbic region showed effective conversion predictive ability.

Several studies performed transcriptome–wide association analyses using qualitative traits in Alzheimer's disease. To our knowledge, this is the first research using quantitative traits in transcriptome–wide association analyses. We found that genes associated with quantitative traits of different brain structures were specific. In the limbic region, we found nine longitudinally stable correlated genes, including four for amygdala volume and five for hippocampus volume. Within these nine genes, six genes have been reported to be associated with AD or MCI based on GeneCards. For example, we found that the expression of *EPHA4* was positively correlated with hippocampus volume in baseline and 12–month follow–up. Gene expression of *EPHA4* was predicted by rs149636195 in a hippocampal predictive model. Rs149636195 is located in the 5'–untranslated region of *EPHA4* and regulates *EPHA4* expression by modulating promoter activity and enhancer activity in the hippocampus [21]. A low level of EphA4 is likely to lead to synaptic dysfunction in early AD [27], EphA4 is responsible for amyloid β–protein production regulation, and *EPHA4* mRNA levels were significantly reduced in AD brains [28]. We speculate that rs149636195 is an eQTL of *EPHA4*, and the low expression of *EPHA4* results in a decrease in hippocampal volume, which may cause synaptic dysfunction in MCI. Additionally, we identified three genes in the limbic region which have not been reported in previous AD/MCI studies, including *NOXRED1*, *MYL6B*, and *FAM162B*. *NOXRED1* (NADP–Dependent Oxidoreductase Domain–Containing 1 protein) is a key gene in oxidoreductase activity (Gene Ontology: 0016491). Oxidative stress may play a role in neuron degeneration and, thus, in AD. We suspect that *NOXRED1* may influence the pathogenesis of AD/MCI through oxidative stress. *MYL6B* encodes myosin light–chain 6B protein and is a key component of myosin. *MYL6B* contributes to memory consolidation in the amygdala [29,30]. Myosin is essential for synapse remodeling [31]. We suspect that dysregulation of *MYL6B* may affect the integrity and function of myosin, leading to the impairment of synaptic function in the pathogenesis of early–stage AD. *FAM162B* (Family with Sequence Similarity 162 Member B) is a key gene in the membrane (Gene Ontology: 0016020) and an integral component of

the membrane (Gene Ontology: 0016021). *FAM162B* plays an important role in endothelial cells in the blood–brain barrier (Lifemap discovery database). We propose that *FAM162B* is important to the maintenance of the blood–brain barrier, which is required for proper synaptic and neuronal functioning. Dysregulation of *FAM162B* may cause a breakdown of the blood–brain barrier, leading to increased susceptibility to AD [32].

We investigated the potential regulation patterns of gene expression–determined cis– eQTL SNPs affecting the expression of the corresponding genes. Due to the fact that gene expression prediction models are based on fine–mapped variants that may occasionally be absent in a typical GWAS and frequently absent in older GWAS [11], we explored the annotations of SNPs for longitudinally stable correlated genes using four databases, including SNPnexus, HaploReg, RegulomeDB, and VARAdb. First, these cis–eQTL SNPs appeared to be related to specific transcription factor binding sites. Transcription factors increase or decrease the transcription levels of genes by binding to super enhancers or promoters in specific DNA regions [33]. Second, we found more that than 57% and more than 44% cis–eQTL SNPs are in the promoters and enhancers of the corresponding genes, respectively. Promoters and enhancers are responsible for the initiation and reinforcement of transcription, respectively. SNPs within enhancers can alter transcription factor binding and alter enhancer–promoter interactions, leading to dysregulation of gene expression and diseases [34], such as AD [35,36]. Based on the above observations, we inferred that gene expression–determined cis–eQTL SNPs can affect the expression of corresponding genes by altering the binding ability of some transcription factors and/or by affecting promoter and enhancer activities. We also verified the possibility of SNPs affecting corresponding gene expression. We performed association analyses using these SNPs and all quantitative traits directly. We found that most SNPs in correlated genes were also correlated to corresponding quantitative traits, indicating that the correlations between quantitative traits and gene expressions were reasonable. SNPs appeared to be associated with quantitative traits by regulating the expression of their corresponding genes.

The identified longitudinally stable correlated genes could be drug candidates for AD or MCI. *EPHA4* encodes a tyrosine protein kinase receptor, and several studies have discussed the therapeutic potential to target EphA4 for AD [37,38]. *AHSA1* encodes an activator of heat shock protein 90 (Hsp90) ATPase. Small–molecule inhibitors of Hsp90 have been successful at ameliorating amyloid beta–protein and tau protein burden in AD [39]. *MYL6B* and *VAPA* have been reported to be related to synapse formation and remodeling [40,41]. The breakdown of synaptic connections can lead to a loss of cognitive ability, and synaptic repair is a disease–modifying strategy for neurodegenerative diseases, such as AD [42]. Mitochondrial dysfunction and oxidative stress are important pathogenetic mechanism of AD [43]. Antioxidants are often used in the clinical treatment of central nervous system diseases, such as AD. Antioxidants could improve mitochondrial energy metabolism, eliminate free radicals, reduce the damage of oxidative stress to the nervous system [44]. Targeted antioxidant drugs for the treatment of AD have been developed, such as idebenone [45]. We identified four genes related to mitochondrial dysfunction and oxidative stress in the limbic region, including *NDUFAF3*, *NOXRED1*, *ME3*, and *AGK*, and these genes may be used as drug targets in early–stage AD. Meanwhile, genes in the basal ganglia region and cerebellum region are related to the inflammatory response, signal transduction, and material transport, and could also be new targets for drug development.

We investigated and compared the potential of baseline quantitative traits and cis–eQTL of longitudinally stable correlated genes in each region in predicting conversion of MCI samples. Structure volumes in the limbic region, basal ganglia region, cerebellum region and corresponding cis–eQTL SNPs in each region were used for conversion analyses. Limbic region structure volumes and 12 SNPs in from longitudinally stable correlated genes in the limbic region showed effective predictive abilities. Our results support previous MRI studies of limbic region volumes in MCI progress prediction and found that SNPs obtained by gene– quantitative trait association also showed conversion prediction value [46–48]. We developed an SNP panel with 12 SNPs that can be used for conversion prediction for MCI patients. Based

on conversion analyses using quantitative traits and SNPs, we estimated that about 65% of MCI patients in the high–risk group will convert to AD within the established follow–up in ADNI, compared with about 40% of those in the low–risk group.

#### **5. Conclusions**

In summary, our study revealed several genes which appeared to be stably correlated longitudinally with brain quantitative traits in the limbic region, basal ganglia region, and cerebellum region. These genes can be used as potential drug targets for the treatment of early–stage AD. Gene expression–determined cis–eQTL SNPs influence the expression of their corresponding genes by affecting transcription factor binding or the activities of promoters and enhancers. Quantitative traits and cis–eQTL SNPs in the limbic region can effectively predict the conversion risk of MCI patients.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/biomedicines9060658/s1, Table S1: Function Annotations of Selected Correlated Genes, Table S2: Genomic locations of cis-eQTL SNPs, Table S3: Annotations from HaploReg and RegulomeDB database, Table S4: Annotations of promoters of cis-eQTL SNPs, Table S5: Annotations of super enhancers of cis-eQTL SNPs, Figure S1: Bar plots of associations between 26 SNPs in the basal ganglia region and 6 subcortical structures, Figure S2: Bar plots of associations between 14 SNPs in the cerebellum region and 6 subcortical structures.

**Author Contributions:** Conceptualization, S.-X.Y.; methodology, S.-X.Y.; validation, S.-X.Y., H.-T.L. and X.S.; formal analysis, S.-X.Y. and X.S.; investigation, S.-X.Y.; writing—original draft preparation, S.-X.Y. and X.S.; writing—review and editing, S.-X.Y., Y.G. and X.S.; visualization, S.-X.Y.; supervision, X.S.; project administration, X.S.; funding acquisition, X.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was sponsored by the National Natural Science Foundation of China (81830053, 61972084) and the Key Research and Development Program of Jiangsu province (BE2016002-3).

**Institutional Review Board Statement:** This study did not involve patients. The data collection procedures were approved by the institutional review boards of all participating centers to the Alzheimer's Disease Neuroimaging Initiative.

**Informed Consent Statement:** Not applicable for this study. Participating centers to the Alzheimer's Disease Neuroimaging Initiative obtained written informed consent from all participants or their authorized representatives for data collection.

**Data Availability Statement:** Data used in this study are available through the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu) (accessed on 25 November 2019).

**Acknowledgments:** Data used in the preparation of this paper were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (http://www.loni.ucla.edu/ADNI) (accessed on 25 November 2019). As such, investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in the analysis or writing of this report. A complete list of ADNI investigators is available at http://www.loni.ucla.edu/ADNI/ Collaboration/ADNI\_Auth-orship\_list.pdf.

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

#### **References**


## *Article* **Increased YKL-40 but Not C-Reactive Protein Levels in Patients with Alzheimer's Disease**

**Víctor Antonio Blanco-Palmero 1,2,3,†, Marcos Rubio-Fernández 1,2,†, Desireé Antequera 1,2, Alberto Villarejo-Galende 1,2,3, José Antonio Molina 1,2,3, Isidro Ferrer 1,4,5,6, Fernando Bartolome 1,2,\* and Eva Carro 1,2,\***


**Abstract:** Neuroinflammation is a common feature in Alzheimer's (AD) and Parkinson's (PD) disease. In the last few decades, a testable hypothesis was proposed that protein-unfolding events might occur due to neuroinflammatory cascades involving alterations in the crosstalk between glial cells and neurons. Here, we tried to clarify the pattern of two of the most promising biomarkers of neuroinflammation in cerebrospinal fluid (CSF) in AD and PD. This study included cognitively unimpaired elderly patients, patients with mild cognitive impairment, patients with AD dementia, and patients with PD. CSF samples were analyzed for YKL-40 and C-reactive protein (CRP). We found that CSF YKL-40 levels were significantly increased only in dementia stages of AD. Additionally, increased YKL-40 levels were found in the cerebral orbitofrontal cortex from AD patients in agreement with augmented astrogliosis. Our study confirms that these biomarkers of neuroinflammation are differently detected in CSF from AD and PD patients.

**Keywords:** Alzheimer's disease; Parkinson's disease; YKL-40; C-reactive protein; CSF and plasma biomarkers; inflammation; astrogliosis

#### **1. Introduction**

Neuroinflammation is now widely accepted as a pathological hallmark of Alzheimer's (AD) [1,2] and Parkinson's (PD) [3–5] disease. Several damage signals appear to induce neuroinflammation, including β-amyloid (Aβ) oligomers, tau, and α-synuclein (α-syn), mediated by the progressive astrocyte and microglial cell activation with the consequent overproduction of proinflammatory agents that may leak toward cerebrospinal fluid (CSF) [6]. Despite the analysis of these agents in CSF being a tempting topic to study, levels of inflammatory markers in CSF from AD and PD patients have not been sufficiently investigated. A standard clinical application of inflammatory markers in the clinical diagnosis of these neurodegenerative disorders is lacking, likely owing to contradictory and heterogeneous findings of numerous studies [7,8].

Among these neuroinflammatory markers found in biological samples is YKL-40 (also named Chitinase 3-like I). This marker has been largely associated with the pathogenesis of a variety of human diseases, many of them sharing chronic inflammatory features and high cellular activity, including rheumatoid arthritis, hepatic fibrosis, and asthma,

**Citation:** Blanco-Palmero, V.A.; Rubio-Fernández, M.; Antequera, D.; Villarejo-Galende, A.; Molina, J.A.; Ferrer, I.; Bartolome, F.; Carro, E. Increased YKL-40 but Not C-Reactive Protein Levels in Patients with Alzheimer's Disease. *Biomedicines* **2021**, *9*, 1094. https://doi.org/ 10.3390/biomedicines9091094

Academic Editor: Lorenzo Falsetti

Received: 21 July 2021 Accepted: 23 August 2021 Published: 27 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

121

where YKL-40 levels were found elevated in patient peripheral blood [9–11]. YKL-40 is a secreted glycoprotein with functions including tissue remodeling during inflammation and angiogenic processes, which make YKL-40 a good marker of inflammation and endothelial dysfunction [12–14]. YKL-40 was found elevated in CSF from several acute and chronic neuroinflammatory conditions [15], as well as in preclinical and prodromal AD/mild cognitive impairment (MCI) [16–18]. This is consistent with the potential role of astrocytosis in early AD pathogenesis [19] and with the fact that *YKL-40* expression and YKL-40 protein levels are abundant in reactive astrocytes and residual in microglial cells [15,20,21]. Additionally, YKL-40 was found close to amyloid plaques and neurofibrillary tangles in AD [16]. Contrarily, other works reported different results showing no significant differences in YKL-40 levels in CSF from MCI and AD patients compared with cognitively normal subjects [22]. Other works indicated increased CSF YKL-40 levels only in AD but not in MCI subjects compared with healthy controls [23,24]. Regarding PD, YKL-40 concentrations in CSF were found either decreased or unchanged [25,26].

Although YKL-40 can be considered one of the most promising neuroinflammatory biomarkers in AD, the abovementioned works indicate that brain YKL-40 levels patterns in different neurodegenerative diseases and the potential correlation between brain and CSF levels is largely unknown, indicating that more research regarding YKL-40 expression pattern is required.

On the other hand, C-reactive protein (CRP), a kind of acute-phase protein regulated by proinflammatory cytokines, is the most studied biomarker of systemic inflammation [27]. CRP was linked to chronic inflammatory and neurodegenerative diseases, such as AD and PD [28]. Elevated CRP peripheral blood levels have been frequently associated with increased risk of dementia and cognitive decline. Studies carried out investigating the association between markers of inflammation and risk of dementia showed conflicting results. A systematic review and meta-analysis found that elevation of peripheral CRP levels was associated with increased risk of developing dementia [29]. Nevertheless, another meta-analysis found no significant differences in serum CRP levels between patients with AD and healthy subjects [30]. Epidemiological studies have also explored the relationship between CRP levels and AD risk, describing lower CRP levels in CSF from AD patients [31,32]. Regarding PD and CRP levels, results in the literature are still contradictory. A significant increase in blood CRP levels was reported in subjects suffering from PD compared with healthy controls [33,34], while other works did not identify such a tendency, instead reporting no differences [35]. Furthermore, the CRP levels in CSF remained unchanged in PD patients when compared with healthy subjects [26,32]. Despite these differences, CRP is considered a prominent "risk factor" for PD [36].

Growing evidence indicates that blood-borne CRP can cross the blood–brain and blood–spinal cord barriers; thus, CRP can be found in the CSF and deposited in the diseased central nervous system (CNS). The source of CRP might also be local. However, CRP production may occur in multiple CNS-resident cells including neurons, microglia, and astrocytes [37–39]. Regardless of its origin (hepatic versus local), the presence of CRP in the CNS is associated with numerous diseases including AD [40]. CRP levels were also found increased in brain parenchyma tissue after intracerebral hemorrhage [41]. Additionally, large amounts of the protein were present in perihematomal regions and within neurons and glia of patients who died within 12 h of spontaneous intracerebral hemorrhage [41,42].

Despite these accumulative data supporting a role of neuroinflammation, particularly YKL-40 and CRP in AD and PD, there is no definitive evidence reflecting the peripheral (blood) and central (CSF) concentration changes of YKL-40 and CRP in AD and/or PD patients. We think that further research is needed to elucidate the variable pattern of these inflammatory biomarkers in the CSF and blood from AD and PD patients. In this work, we aimed at clarify YKL-40 and CRP concentrations measured in CSF and plasma and to determine their specificity in AD and PD. To address this issue, we analyzed YKL-40 and

CRP levels in CSF and plasma from a well-characterized cohort of patients with MCI, AD, and PD, using sensitive enzyme-linked immunosorbent assays (ELISAs).

#### **2. Material and Methods**

#### *2.1. Human Donors*

A total of 123 subjects were included in this study: (1) elderly nondemented subjects without any evidence of any neurodegenerative disease (healthy controls) classified as controls (*n* = 37); (2) MCI due to AD (MCI) patients (*n* = 22); (3) probable mild/moderate– severe sporadic AD patients (*n* = 34); (4) PD patients (*n* = 30). Study participants were enrolled from the Memory Clinic (controls, MCI and AD subjects) and Movement Disorders Unit (PD participants) of Hospital Universitario 12 de Octubre (Madrid, Spain). Subject demographic and clinical characteristics are listed in Table 1.


**Table 1.** Demographic and clinical data of participants.

AD: Alzheimer's disease; MCI: mild cognitive impairment. PD: Parkinson's disease; n: number; F: female; ns: non-significant; y: year; M: male; SD: standard deviation; NA, not applicable; CDR: Clinical Dementia Rating. *p* value indicates statistical difference within the cohort1; -: not obtained data; <sup>a</sup> *p* < 0.05 vs PD; <sup>b</sup> *p* < 0.01 vs. PD; <sup>c</sup> *p* < 0.0001 vs. AD; <sup>d</sup> *p* < 0.0001 vs. PD.

> All participants were classified using established diagnostic criteria into those with MCI or probable AD dementia [43–45]. Diagnosis was based on detailed clinical assessment, neuropsychological evaluation, and neuroimaging (MRI). Functional impairment was measured via the Clinical Dementia Rating (CDR) score [46]. PD patients were diagnosed following the Movement Disorder Society (MDS) clinical diagnostic criteria [47], and all fulfilled criteria for clinically established PD. PD patients did not refer cognitive complaints and did not exhibit symptoms of dementia. The control group was constituted by cognitively normal individuals aged 50 years or older, without clinical signs of cognitive impairment and without neurological or psychiatric disease history. Exclusion criteria for every participant were concomitant significant cerebrovascular disease and evidence of any neurological, psychiatric, medication, or non-neurological medical comorbidity that could affect cognition or motor function.

> Approval of the study was obtained from the Research Ethics Committee of Hospital Universitario 12 de Octubre, and all participants provided written informed consent.

#### *2.2. Fluid Sample Collection*

CSF samples were collected from all subjects (including healthy patients and MCI, AD, and PD subjects) and processed according to standardized procedures by lumbar puncture in 15 mL sterile polypropylene tubes. Samples were then centrifuged at 3000 rpm at 4 ◦C for 10 min. Supernatant aliquots were stored at −80 ◦C into 0.5 mL polypropylene cryogenic tubes with Protease Inhibitor Cocktail (Roche, Basel, Switzerland).

Blood samples were obtained through antecubital vein puncture from patients and healthy subjects. Plasma was isolated from whole blood collected in 7 mL EDTA-2Na tubes. Whole blood was centrifuged at 2000 rpm for 10 min at room temperature. Supernatants were then collected and aliquoted in polypropylene cryogenic tubes with Protease Inhibitor Cocktail (Roche, Basel, Switzerland) and stored at −80 ◦C.

#### *2.3. Tissue Samples*

Postmortem cerebral orbitofrontal cortex tissue was obtained from brain donors diagnosed with AD and control individuals. Frozen samples were supplied by the Institute of Neuropathology Brain Bank IDIBELL-Hospital Universitari de Bellvitge (Hospitalet de Llobregat, Spain). Subject consent was obtained according to the Declaration of Helsinki, and approval came from the Research Ethics Committee of the responsible institution. For all cases, written informed consent was available. Subjects were selected on the basis of postmortem diagnosis of AD according to neurofibrillary tangle pathology and Aβ plaques [48]. AD cases showed high AD neuropathologic change (Braak stage V/VI and moderate to frequent neuritic plaque score). Control participants were considered those with/without neurological symptoms or a low grade of AD neuropathologic change. A total of 24 samples were categorized into AD and controls, as presented in Table 2.


**Table 2.** Demographic and clinical data of brain tissue donors.

AD: Alzheimer's disease; n: number; F: female. M: male; SD: standard deviation.

#### *2.4. DNA Purification and Apolipoprotein E (APOE) Genotyping*

Genomic DNA was extracted from peripheral blood using QIAmp DNA Blood Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Human APOE C112R and R158C polymorphisms were detected to identify the APOE ε2, ε3, and ε4 alleles, using the LightCycler 480 II Instruments Kit (Roche Diagnostics, Basel, Switzerland) following manufacturer instructions.

Braak II: 2

#### *2.5. Protein Analysis*

CSF and plasma concentrations of the neuroinflammatory biomarkers (YKL-40 and CRP) were analyzed using ELISA kits (Human Chitinase 3-like 1 Quantikine ELISA kit (DC3L10), R&D; Human CRP Quantikine ELISA kit (DCRP00), R&D) according to the manufacturer's instructions.

Brain YKL-40 and GFAP protein levels were also examined by Western blotting. Postmortem cerebral orbitofrontal cortex tissue was obtained from brain donors diagnosed with AD and control individuals. Briefly, human cerebral orbitofrontal cortex samples were incubated and homogenized in lysis buffer (50 mM Tris/HCl buffer, pH 7.4 containing 2 mM EDTA, 0.2% Nonidet P-40, 1 mM PMSF, Protease and Phosphatase Inhibitor Cocktails; Roche, Basel, Switzerland) and centrifuged for 10 min at 14,000 rpm at 4 ◦C. Supernatants were recovered and stored at −80 ◦C. Protein content was determined using the BCA method (Thermo Fisher Scientific, MA, USA). Equal amounts of protein (20 μg for YKL-40 and 5 μg for GFAP) were mixed with Laemmli sample buffer supplemented with β-mercaptoethanol, heated to 95 ◦C for 5 min, resolved by 10% NuPAGE Bis-Tris Gels (Thermo Fisher Scientific, MA, USA), and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, MA, USA). Afterward, membranes were blocked and

incubated overnight at 4 ◦C with primary antibodies: a recombinant rabbit monoclonal anti-YKL-40 antibody (ab255297, 1:500, Abcam) and a mouse monoclonal anti-GFAP antibody (G3893, 1:0000, Sigma Aldrich). Membranes were then incubated for 1 h with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (G-21234, 1:5000, Thermo Fisher Scientific, MA, USA; ab97023, 1:40000, Abcam). Protein loading was monitored using mouse monoclonal HRP-conjugated antibodies against α-tubulin (ab40742, 1:5000, Abcam) for YKL-40 or against β-actin (A1978, Sigma Aldrich) for GFAP detection. Immunocomplexes were revealed by an enhanced chemiluminescence reagent (ECL Clarity; Bio Rad, CA, USA). Densitometric quantification was carried out with Image Studio Lite 5.0 software (Li-COR Biosciences, NE, USA). Protein bands were normalized to loading controls and expressed as a percentage of the control group.

#### *2.6. Statistical Analysis*

Statistical analysis and graphs were performed using Stata/IC software (Stata 16.1, StataCorp LLC, College Station, TX, USA) and Prism (GraphPad Software version 8.00, La Jolla, CA, USA). After assessing the normality of the distribution, differences in CSF and plasma YKL-40 and CRP levels between groups were analyzed using the nonparametric Kruskal–Wallis rank test. The *p*-value for pairwise comparisons is displayed with Bonferroni correction. A descriptive multiple linear regression model was performed to account for confounding variables (age, sex, and APOE ε4) in CSF YKL-40 association analysis. Interactions of confounding variables with the clinical diagnosis were excluded from the model (significance for the whole set of interactions: *p* > 0.10). The regression coefficient is displayed as "b". Differences in sex distribution, age of participants, age at onset, and years since the onset of the disease between groups were evaluated with Pearson chi-squared and ANOVA tests, where appropriate. Associations between biomarkers and demographic characteristics were examined with Pearson correlation tests, Student *t*-tests and Mann–Whitney U tests, where appropriate. A nonparametric trend test (Jonckheere trend test) was performed to evaluate the existence of a trend when the exposition showed ordinal categories. ROC curves were constructed after modeling the presence or absence of a given clinical diagnosis with a regression logistic analysis. YKL-40 and GFAP Western blot expression levels were normalized to their respective loading controls (α-tubulin and β-actin) and compared with the mean of the control ratio with the nonparametric Mann–Whitney U test. In graphs, CSF and plasma YKL-40 and CRP levels are shown as median and interquartile range. The brain expression of YKL-40 is shown as the mean ± standard error of the mean (SEM). In all cases, statistical significance was set at *p* < 0.05.

#### **3. Results**

#### *3.1. Associations with Demographic Data*

Demographic and clinical data are shown in Table 1 for further characterization of the study cohort. A total of 34 subjects were clinically diagnosed with AD, 22 subjects were grouped as MCI, and 30 subjects were diagnosed with PD. Individuals diagnosed with AD were slightly older than the rest of the cohort, including PD, MCI, and healthy subjects. Female sex was overrepresented in the AD and MCI groups, while males represented around 60% of controls and PD subjects. APOE ε4 carriers were more prevalent in the MCI/AD group than in controls, according to previous publications [49]. Most AD patients had clinically mild dementia (74% scored 1 in CDR scale), and none of the PD patients reached the dementia stage. Furthermore, the majority of individuals diagnosed with PD exhibited mild motor impairment (73% of them were in Hoehn & Yahr stage 1 or 2).

#### *3.2. YKL-40 and CRP Levels in Different Diagnostic Groups*

YKL-40 and CRP levels across all clinical groups are illustrated in Figure 1. In CSF, YKL-40 levels were different among groups and were found to increase in AD dementia subjects compared with healthy controls (Figure 1A). No differences were found in YKL-40 levels between healthy controls and MCI or PD groups in CSF (Figure 1A). Nevertheless,

a trend toward reduced levels was observed in PD patients, which were significantly lower compared to AD and MCI patient groups (Figure 1A). In plasma, YKL-40 levels remained unchanged across all clinical groups (Figure 1B).

**Figure 1.** YKL-40 and CRP levels in CSF and plasma in different diagnostic groups. Box-and-whisker plots showing (**A**,**B**) YKL-40 and CRP levels (**C**,**D**) in CSF and plasma, respectively, across the diagnostic groups. Differences between groups were assessed using Kruskal–Wallis test followed by Bonferroni correction. \* *p* < 0.05; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001. MCI, mild cognitive impairment; AD, Alzheimer's disease dementia; PD, Parkinson's disease. ns: non-significant.

A nonparametric trend test did not show any statistically significant rising tendency of CSF (*p* = 0.48) or plasma (*p* = 0.053) YKL-40 levels along with MCI or mild and moderate AD. When adjusting for age, sex, and APOE ε4 status, levels of CSF YKL-40 remained high in AD dementia patients when compared with controls (b = 125.5 ng/mL, 95% CI = 19.1 to 232.0 ng/mL, *p* < 0.05).

Regarding CRP levels in CSF and plasma, we did not find significant differences between healthy subjects and AD, MCI, and PD patients (Figure 1C,D). Our results are consistent with previous studies indicating no differences in CRP levels from CSF comparing healthy subjects and PD patients [26] or in serum CRP levels between patients with AD and healthy subjects [30].

In order to analyze the discriminative ability of both biomarkers for the diagnosis of PD and AD, we performed a logistic regression analysis and calculated the corresponding ROC curve for each CSF biomarker and diagnosis. CSF YKL-40 differentiated AD patients from the rest of the cohort, including PD, MCI, and healthy subjects, with 65.6% sensitivity and 66.3% specificity (AUC = 0.69, 95%CI = 0.58 to 0.80, cutoff point = 316.5 ng/mL) (Figure 2A). The combination with CSF CRP did not improve the performance. Nevertheless, for the diagnosis of PD, the combination of CSF YKL-40 and CRP yielded the best results, showing a moderate discriminative ability (AUC = 0.82, 95% CI =0.73 to 0.89, cutoff point of the model = 0.300), with 79.2% sensitivity and 82.1% specificity (Figure 2B).

**Figure 2.** Receiver operating characteristic (ROC) analysis of YKL-40 and CRP levels in CSF. (**A**) ROC curve and its corresponding area under the curve (AUC) differentiating YKL-40 levels in CSF from AD patients and non-AD subjects including control subjects. (**B**) AUC differentiating the combination of YKL-40 and CRP levels in CSF from PD and non-PD patients. AUC, area under the curve; AD, Alzheimer's disease dementia; PD, Parkinson's disease.

#### *3.3. Correlations between YKL-40 and CRP Levels in Plasma and CSF*

Both CSF YKL-40 (*r* = 0.39, *p* < 0.001; Figure 3A) and CRP (*r* = 0.56, *p* < 0.0001; Figure 3B) correlated significantly with their respective plasma concentrations in the whole cohort. The stronger positive correlation was found in AD patients (YKL-40: *r* = 0.69, CRP: *r* = 0.84).

**Figure 3.** Correlation between YKL-40 and CRP levels in CSF and plasma, and between YKL-40 and age in the study cohort. Correlations between the expression levels of (**A**) YKL-40 and (**B**) CRP in CSF and plasma in the study cohort. Correlation between (**C**) CSF and (**D**) plasma YKL-40 and age within the diagnostic group. Correlations were examined with Pearson correlation test. MCI, mild cognitive impairment; AD, Alzheimer's disease dementia; PD, Parkinson's disease.

In the whole cohort, plasma and CSF YKL-40 levels positively correlated with age (CSF YKL-40: *r* = 0.38, *p* < 0.0001; Figure 3C; plasma YKL-40: *r* = 0.57, *p* < 0.0001; Figure 3D). This correlation was especially stronger for the control group (CSF YKL-40: *r* = 0.46, *p* < 0.01; plasma YKL-40: *r* = 0.84, *p* < 0.0001). No statistically significant correlation with age was found in the plasma and CSF CRP analysis. Furthermore, the time since symptom onset did not correlate with any biomarker level in any group. Plasma and CSF YKL-40 and CRP levels did not differ by sex or by the presence of an APOE ε4 allele.

#### *3.4. YKL-40 Levels in AD Brain*

Upon inflammation, YKL-40 is produced and secreted by many cells including vascular smooth muscle cells and macrophages [50]. In the brain, YKL-40 is mainly expressed in reactive astrocytes [20,25]. Thus, we investigated if the observed increase in YKL-40 levels in CSF from AD patients could be associated with higher YKL-40 levels in cerebral parenchyma. To explore this hypothesis, we examined the YKL-40 cellular levels in human brain tissue from AD patients and healthy subjects. Immunoblotting showed that YKL-40 levels in cerebral orbitofrontal cortex samples were significantly increased in AD patients compared with healthy subjects (Figure 4A). To determine if increased levels of YKL-40 in cerebral orbitofrontal cortex were associated with astrocyte reactivity, the levels of GFAP

were also analyzed. Western blotting showed that GFAP levels were also higher in AD samples compared to those observed in control subjects (Figure 4B) in parallel with the observed rise in YKL-40 levels, proving that AD astrogliosis increases YKL-40 levels.

**Figure 4.** YKL-40 and GFAP levels in cerebral orbitofrontal cortex of AD patients and control group. Western blot analysis showing (**A**) YKL-40 and (**B**) GFAP in the cerebral orbitofrontal cortex of AD and control samples. Representative Western blots (left panels) and histograms with their densitometric analysis (right panels) are shown. Data are represented as the mean ± SEM. Differences between groups were assessed using Mann–Whitney test; \* *p* < 0.05, \*\* *p* < 0.01.

#### **4. Discussion**

In this cross-sectional study, we showed a variable pattern of the inflammatory biomarkers YKL-40 and CRP in AD and PD patients. We confirmed that YKL-40 levels are significantly increased in CSF from AD patients compared to healthy controls, indicating an inflammatory response at the dementia stage. Such an increase was not seen in MCI or PD patients, where CSF YKL-40 levels remained unchanged. These results were also extended to the cerebral orbitofrontal cortex where we found that YKL-40 expression was augmented in AD patients, suggesting glial activation, thus corroborating our hypothesis. Another finding in this study was related to CRP levels in CSF and plasma. We found lower CRP levels in CSF from PD patients compared with other groups (AD, MCI, and

healthy subjects), but this change did not reach statistical significance. Furthermore, we did not find evidence of significant alterations in plasma for YKL-40 or CRP.

Inflammation is increasingly recognized as part of the pathology of neurodegenerative conditions, including AD and PD. Evidence proposes that neurodegeneration occurs in part because the CNS environment is affected by a cascade of events collectively named neuroinflammation [51]. Despite biomarkers of neuroinflammation being useful for monitoring disease diagnosis, progression, and response to therapy, accurate and reliable biomarkers for many neurological diseases are scarce. In recent years, the interest in new neuroinflammatory biomarkers has grown at early and symptomatic stages of these diseases. Blood and CSF are commonly used to monitor biomarkers of neuroinflammation, with many of them being the consequence of the CNS pathology. Some examples are the levels of cytokines and chemokines, the loss of blood–brain barrier integrity, and neuronal damage indicators [52].

Only a few studies have shown the possibility of analyzing YKL-40 levels in CSF and blood from patients with AD and predementia stages. One of these studies found that YKL-40 concentration in CSF from AD patients was significantly elevated compared to cognitively normal subjects, with an AUC = 0.88 pointing to the potential value of YKL-40 levels in CSF for AD diagnosis [53]. Increased YKL-40 levels were observed not only in AD dementia, but also in the prodromal phase of AD when compared to cognitively normal controls [54]. Similar observations were found in patients with AD, where YKL-40 concentration in CSF was increased in very mild and mild dementia subjects in comparison with cognitively normal individuals [16]. In our study, we found a trend of increased YKL-40 levels in CSF from MCI subjects compared with healthy controls, and this increase was evident in AD patients. However, the resulting AUC in our study was lower; thus, we propose that YKL-40 might only be a modest AD biomarker candidate.

Significantly increased *chitinase-3 like 3* (*CHI3L3*) mRNA expression, a mouse homolog of YKL-40, was found in brains of AD mice models when compared to age-matched controls [55]. Similarly, in autopsied human brain samples from pathologically confirmed AD subjects, *YKL-40* mRNA levels were significantly increased in comparison with nondemented controls [55]. Although there is no clear explanation regarding which factors modulate YKL-40 levels in AD, it has been suggested that elevated *YKL-40* expression and protein levels might result from increased astrocytic reactivity and release in brain [21]. It was shown that astrocytes in the close vicinity of amyloid plaques were immunoreactive for YKL-40, which confirms the involvement of this protein in the neuroinflammatory response to Aβ deposition [16]. It is known that insoluble Aβ aggregates may induce inflammatory reactions and activation of microglia, resulting in increased proinflammatory mediator production. The relationship between YKL-40 and amyloid-related pathways in AD development was further discussed [17,25]. It seems that the YKL-40 concentration in CSF may be linked to AD pathology, particularly astrogliosis. Indeed, it has been shown that *YKL-40* is expressed by reactive astrocytes GFAP+ in AD [25]. Thus, increased expression of *YKL-40* and protein levels in reactive astrocytes may be reflected in the CSF, indicating that astrocyte-associated metabolites may be utilized as potential biomarkers. Although data regarding elevated YKL-40 levels in CSF from early stages of AD are contradictory [16,17,22–24,54], our results support the increase in YKL-40 levels in CSF from AD subjects, as well as the increased astrocytic YKL-40 levels associated with astrocytosis.

Interestingly, we found that YKL-40 levels in CSF from PD patients were significantly lower compared with those levels in AD subjects suggesting that YKL-40, a marker of astroglial activation, is downregulated in PD. It was reported that YKL-40 levels were decreased in synucleinopathies when compared with tauopathies, suggesting that glial activation may be lower in brains from PD patients and other synucleinopathies in comparison with patients who have tauopathies or healthy controls [26,56]. These data may suggest that CSF YKL-40, as a marker of astroglial activation, is downregulated in PD. Despite astrocytes exerting protection against the inflammatory response in PD [57,58], astroglial dysfunction due to α-syn inclusions may occur simultaneously. In vitro evidence showed

that astrocytes are able to efficiently degrade the α-syn aggregates from the extracellular space [59]. More recently, it was shown that primary rat astrocytes receive α-syn aggregates from neurons in mixed cell culture and efficiently transfer them from astrocyte to astrocyte [60]. It is possible that the increase in α-syn levels in astrocytes is a consequence of an endocytic mechanism upon high α-syn levels from the extracellular space, leading to the typical α-syn astrocytic inclusions in PD brains [61]. This accumulation could then lead to the dysregulation of other astrocytic functions, including YKL-40 production/secretion.

Our study yielded no significant changes for CRP levels in CSF or in plasma from AD and PD subjects, although others have described contradictory results [30–32,34]. Pathological studies have demonstrated that CRP is present in the senile plaques and neurofibrillary tangles in AD brains, suggesting that this protein may play a role in the neuropathological processes in AD [62–64]. In PD, aggregated α-syn can promote microglial activation and stimulate the secretion of inflammatory molecules, including CRP [65], thus evoking neuroinflammation [66].

CRP is primarily produced in the liver but is also generated in neurons to a lesser extent [41]. Such residual production of CRP in the CNS does not appear to contribute significantly to CSF levels [39].

In summary, our present study revealed a different inflammatory biomarker profile in individuals with AD and PD. CSF YKL-40 levels were significantly elevated in the AD group, and this increment corroborated the analysis of the YKL-40 protein levels in the cerebral orbitofrontal cortex from pathologically confirmed AD subjects. In PD individuals, plasma and CSF CRP and YKL-40 levels remained unchanged. Notwithstanding, we identified a moderate discriminative ability by combining both biomarkers in CSF for PD diagnosis. Together, our data support the involvement of both inflammatory proteins in the pathogenesis of neurodegenerative diseases.

**Author Contributions:** F.B. and E.C. designed the study and wrote the manuscript; V.A.B.-P., M.R.-F., and D.A. carried out and analyzed experiments; A.V.-G., J.A.M. and I.F. provided CSF and blood samples from participants. All authors read and agreed to the published version of the manuscript.

**Funding:** This study was supported by grants from Instituto de Salud Carlos III (FIS18/00118), FEDER, Comunidad de Madrid (S2017/BMD-3700; NEUROMETAB-CM), and CIBERNED (CB07/502). V.A.B.-P. is supported by the Instituto de Salud Carlos III (ISCIII, Spanish Biomedical Research Institute) through a "Río Hortega" contract (CM 18/0095).

**Institutional Review Board Statement:** This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Research Ethics Committee of Hospital Universitario 12 de Octubre (18/459, 27 November 2018).

**Informed Consent Statement:** Informed consent to obtain their samples was obtained from all subjects involved in the study. Written informed consent was obtained from all subjects involved in this study to publish this paper using the results obtained with their biological samples.

**Data Availability Statement:** The data obtained and presented in this study are available upon reasoned request from the corresponding author.

**Acknowledgments:** We are grateful to the patients and donors without whom these studies would not have been possible.

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

#### **References**


## *Review* **Potential Roles of Sestrin2 in Alzheimer's Disease: Antioxidation, Autophagy Promotion, and Beyond**

**Shang-Der Chen 1,2,†, Jenq-Lin Yang 2,†, Yi-Heng Hsieh 3, Tsu-Kung Lin 1,4,5, Yi-Chun Lin 6, A-Ching Chao 7,8,\* and Ding-I Yang 3,9,\***


**Abstract:** Alzheimer's disease (AD) is the most common age-related neurodegenerative disease. It presents with progressive memory loss, worsens cognitive functions to the point of disability, and causes heavy socioeconomic burdens to patients, their families, and society as a whole. The underlying pathogenic mechanisms of AD are complex and may involve excitotoxicity, excessive generation of reactive oxygen species (ROS), aberrant cell cycle reentry, impaired mitochondrial function, and DNA damage. Up to now, there is no effective treatment available for AD, and it is therefore urgent to develop an effective therapeutic regimen for this devastating disease. Sestrin2, belonging to the sestrin family, can counteract oxidative stress, reduce activity of the mammalian/mechanistic target of rapamycin (mTOR), and improve cell survival. It may therefore play a crucial role in neurodegenerative diseases like AD. However, only limited studies of sestrin2 and AD have been conducted up to now. In this article, we discuss current experimental evidence to demonstrate the potential roles of sestrin2 in treating neurodegenerative diseases, focusing specifically on AD. Strategies for augmenting sestrin2 expression may strengthen neurons, adapting them to stressful conditions through counteracting oxidative stress, and may also adjust the autophagy process, these two effects together conferring neuronal resistance in cases of AD.

**Keywords:** Alzheimer's disease; autophagy; mTOR; oxidative stress; sestrin2

#### **1. Introduction**

Patients with age-related neurodegenerative diseases usually present with a relentlessly deteriorating clinical course. Worst of all, the lack of effective treatment results in heavy socioeconomic burdens to patients, family, and the whole of society [1–3]. Alzheimer's disease (AD), a type of dementia with progressive memory loss and declined cognitive functions, is the most common neurodegenerative disease in the elderly. Based on the information from the World Health Organization (WHO), approximately 50 million people suffer from dementia worldwide, and nearly 10 million new cases are added every year, making the disease one of the main causes of disability and dependence. AD may account

**Citation:** Chen, S.-D.; Yang, J.-L.; Hsieh, Y.-H.; Lin, T.-K.; Lin, Y.-C.; Chao, A.-C.; Yang, D.-I. Potential Roles of Sestrin2 in Alzheimer's Disease: Antioxidation, Autophagy Promotion, and Beyond. *Biomedicines* **2021**, *9*, 1308. https://doi.org/ 10.3390/biomedicines9101308

Academic Editor: Lorenzo Falsetti

Received: 29 July 2021 Accepted: 16 September 2021 Published: 24 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

135

for 60–70% of all dementia cases (https://www.who.int/news-room/fact-sheets/detail/ dementia, accessed on 21 September 2020). According to "2021 Alzheimer's disease facts and figures", in the USA [4], approximately 6.2 million senior Americans over 65 years old have AD. By 2060, with a steep projected increase, the number of AD patients may rise to 13.8 million. Data revealed that, from 2000 to 2019, deaths resulting from human immunodeficiency virus (HIV), heart disease, and stroke declined, while deaths from AD increased more than 145% [4]. The total healthcare costs in 2020 are approximated at \$305 billion and are expected to increase to more than \$1 trillion as the population ages [5]. It is crucial to delay, reduce, or prevent the occurrence of disability from AD and lessen the heavy burden it places on society.

The major pathological hallmarks of AD brains are gross atrophy of the brain, as well microscopically observable senile plaques and neurofibrillary tangles (NFTs) [6–8]. Senile plaques are extracellular structures mainly composed of insoluble deposits of amyloid-beta peptide (Aβ), a peptide fragment of 39–43 amino acids derived from sequential cleavage of the transmembrane protein amyloid precursor protein (APP) by β- and γ-secretase [9–12]. Newly synthesized full-length APP is transported from the endoplasmic reticulum (ER) to the Golgi apparatus (GA)/*trans*-Golgi network (TGN) for further protein processing and maturation. The acidic environment (pH = 6.0–6.5) in the TGN or the late GA is optimal for the activity of many processing enzymes, including BACE1. The full-length APP delivered to the plasma membrane may be subjected to non-amyloidogenic cleavage by α- and then γ-secretase to release the soluble APP-alpha (sAPPα), the p3 fragment, and the APP intracellular domain (AICD). Alternatively, a portion of the full-length APP may also be endocytosed into early endosomes and possibly rerouted to the acidic recycling endosomes (REs), where BACE1 resides, to produce Aβ [13]. In addition, extracellular Aβ can also be taken up through receptor binding and subsequently internalized, thereby leading to its accumulation within various intracellular compartments, including endosomes, multivesicular bodies (MVBs), lysosomes, mitochondria, the ER, the TGN, and cytosol [14].

Aβ can induce neurotoxicity through various mechanisms, such as excitotoxicity [15], excessive generation of reactive oxygen species (ROS) [16], aberrant cell cycle reentry [17,18], impaired mitochondrial function [19], and DNA damage [20], all of these mechanisms together contributing to neuronal damage or even death. Moreover, Aβ can also alter gene transcription [19], and thereby affect protein expression, which may influence the survival or death of neuronal cells in AD-related pathophysiology.

Maintenance of neuronal functions depends on axonal transport of proteins, organelles, and vesicles from the soma to the nerve terminals [21]. Going the other way, neurotrophic factors, including the members of the neurotrophin family, secreted from postsynaptic targets must be transmitted retrogradely from nerve terminals via axonal transport back to the soma [22]. Thus, failure of axonal transport may contribute to neuronal death. As a microtubule-binding protein important for microtubule assembly and stabilization, hyperphosphorylation of tau compromises its biological functions and destabilizes the structures of microtubules, and is accompanied by disturbance to axonal transport [23]. Furthermore, increasing evidence suggests that Aβ may also disrupt axonal transport and contribute to AD pathophysiology [21].

It was proposed two decades ago that fibrils may not be the only toxic form of Aβ; small oligomers of Aβ, or Aβ-derived diffusible ligand (ADDL), and Aβ protofibrils may also have potent neurotoxicity [24]. Like Aβ oligomers, tau oligomers formed during the early stages of aggregation are also pathologically relevant to the loss of neurons and behavioral impairments in several neurodegenerative disorders called tauopathies, the most common of which is AD [25]. In addition to the aggregation of extracellular amyloid plaques, emerging evidence has revealed the crucial role of intraneuronal amyloid species (iAβs) which can appear in the membrane or the lumen of late endosomes and precede further aggregation, eventually accumulating inside the endosome or endolysosome [26,27]. It was also noted that, besides the extracellular aggregation of homologous Aβ species, cross-seeding of different amyloid proteins, or even between different misfolded proteins,

such as Aβs and tau, may be biologically significant, and even critical in the progression of AD [28]. Apart from cross-seeding, crosstalk between Aβ and tau may also play a vital role contributing to AD pathogenesis. For example, Aβ has been shown to trigger alternative splicing of tau isoforms via glycogen synthase kinase-3beta (GSK-3β), making tau more susceptible to hyperphosphorylation [29,30]. Overall, these effects could further aggravate aberrant cellular signaling, induce excessive tau phosphorylation, worsen toxic tau accumulation, and lead to synapto/neurotoxic effects [26]. A simplified cartoon summarizing the pathogenic mechanisms of AD is shown in Figure 1, below.

**Figure 1.** The cartoon diagram demonstrates the pathogenic processes of amyloid-beta peptide (Aβ) and tau protein. Through the amyloidogenic pathway, the full-length amyloid precursor protein (APP) is sequentially cleaved by β-secretase (encoded by beta-site amyloid precursor protein cleaving enzyme-1 or BACE1) and γ-secretase to generate Aβ. Newly synthesized APP is transported from the endoplasmic reticulum (ER) to the Golgi apparatus (GA) for protein maturation. The acidic pH in the trans-Golgi network (TGN) or the late GA is optimal for BACE1 activity, with production of secreted Aβ; the sequential amyloidogenic cleavages of full-length APP by β- and γ-secretase also generate soluble APP-beta (sAPPβ) and the APP intracellular domain (AICD), though these are not depicted in the diagram. A portion of the full-length APP reaching the plasma membrane may be subjected to the non-amyloidogenic cleavage by α- and then γ-secretase to release the soluble APP-alpha (sAPPα), the p3 fragment, and the AICD. Another portion of the full-length APP may also be endocytosed into early endosomes and possibly be rerouted to the acidic recycling endosomes (REs; not depicted), where BACE1 resides, for intracellular production of Aβ. Furthermore, extracellular Aβ can also be taken up through receptor binding and subsequent internalization, resulting in its accumulation within various intracellular compartments, including endosomes, multivesicular bodies (MVBs), and mitochondria (not depicted). The extracellular Aβ monomers aggregate into oligomers and then into fibrils, eventually forming senile plaques. Tau protein is a microtubule-binding protein, which is hyperphosphorylated in AD neurons. The phosphor-tau monomer may also aggregate into tau oligomers and, finally, into neurofibrillary tangles (NFTs). The intraneuronal Aβ species also oligomerize or even mix with tau proteins to form mixed aggregates. The extracellular senile plaques,

the extracellular and intraneuronal Aβ oligomers, as well as tau oligomers and NFTs, together lead to excessive production of reactive oxygen species (ROS), Ca2+ overload, mitochondrial dysfunction, and disrupted energy homeostasis, ultimately causing neuronal death. In addition to those pictured above, other pathogenic mechanisms are not demonstrated in this figure due to limited space. For example, loss of tau binding destabilizes microtubules, thus compromising anterograde axonal transport of proteins, mitochondria, and vesicles from soma to the nerve terminals, which may negatively impact nerve transmission. Conversely, neurotrophic factors, especially neurotrophins, secreted from target cells also fail to be retrogradely transported from the nerve terminal back to the soma to nourish the neurons, also leading to neuronal demise. Please see the text for more details.

Sestrins, including sestrin1, sestrin2, and sestrin3, belong to a group of highly evolutionarily conserved proteins in mammalian cells, and may play a crucial role in stressful conditions, such as oxidative stress, hypoxia, and DNA damage [31–34]. While the structures of sestrin1 and sestrin3 await further elucidation, the essential characteristics of sestrin2 have been gradually revealed in recent years [35,36]. Three distinctive functional sites were identified, which are critical for inhibition of ROS production, modulation of the mammalian/mechanistic target of rapamycin (mTOR) complex 1 (mTORC1), and for leucine-binding [35,36]. Inhibiting either ROS for antioxidation or mTORC1 for autophagy promotion may attenuate degenerative processes associated with aging [35]. Therefore, sestrins may possess two beneficial effects that are pivotal for anti-aging [37,38].

Despite the potential effect of sestrins on age-related neurological disorders, only quite limited studies about AD have been reported. We have shown in a previous study that sestrin2 was induced by Aβ in primary rat cortical neurons and an increased expression of sestrin2 was also found in the cortices of 1-year-old AD transgenic mice [39]. We also showed that sestrin2 functions as an endogenous protective mediator against Aβ-induced neurotoxicity, in part through enhancement of autophagy activity [39]. In another recent study, we further demonstrated that Aβ-induced sestrin2 expression contributes to antioxidative activity in neurons; furthermore, Aβ induction of sestrin2 is at least partly mediated by the activation of transcription factors NF-κB and p53 [40]. In this review article, we discuss recent progress in revealing the underlying molecular mechanisms concerning the sestrin2-mediated protective effects against neuronal dysfunction in AD. Better understanding of the potential novel pathway in AD may guide further research into developing effective therapeutic regimens in the future. Finding the way to augmenting sestrin2 expression may have significant clinical implications, especially in treating many devastating neurodegenerative diseases, including AD.

#### **2. The Biological Roles of Sestrin2**

Sestrins, including sestrin1, sestrin2, and sestrin3, belong to a gene family and function as stress-inducible proteins that affect metabolism through perceiving nutrient status and redox level in living organisms. Sestrin1 (also known as PA26) was initially discovered in human Saos-2 osteosarcoma cells as one of the p53-induced transcripts and was mapped to chromosome 6q21 through a differential display screening [34,41]. Sestrin1 is ubiquitously expressed in most tissues, including lung, kidney, pancreas, skeletal muscle, and brain tissues [33], and it can be activated under oxidative stress and irradiation in a p53-dependent fashion [34,42]. Sestrin2 (also known as Hi95), located in chromosome 1p35.3, was first discovered in glioblastoma cells under prolonged hypoxia and its transcription was found to be increased following DNA damage [33]. Later, it was noted that sestrin1 and sestrin2, through activating the AMP-dependent kinase (AMPK) pathway, may affect tuberous sclerosis complex 2 (TSC2) expression to inhibit mTOR-mediated cell over-proliferation [43]. Sestrin3, located in chromosome 11q21, was identified from database mining of the PA26 related gene family [32,33]. mRNA expression of these sestrin genes is presented diffusely during mouse embryogenesis and also in adult tissues at various levels [32]. Sestrin1 is robustly expressed in the brain, heart, liver, and skeletal muscle; sestrin2 is expressed more

in the kidney, leucocytes, lungs, and liver; sestrin3 is expressed at higher levels in the brain, kidney, small intestine, and skeletal muscle [32,34,44].

It has been revealed that the crystal structure of human sestrin2 (hSesn2) has distinct globular subdomains, each possessing separate functions [35]. As shown below in Figure 2A, the N-terminal domain (Sesn-A) diminishes alkyl hydroperoxide radicals through the helix-turn-helix oxidoreductase motif. Mutations of Cys125, His132, and Tyr127, which are, respectively, the catalytic cysteine, the residue critical for the conserved proton relay system, and the residue potentially involved in the catalytic process, reduce this redox activity. The C-terminal domain (Sesn-C) of hSesn2, whose sequence is highly conserved across the sestrin family, has lost its antioxidant activity but acquired another important function in mTORC1 inhibition via physical association with GTPase-activating protein activity toward the Rags-2 (GATOR2) complex, in which process Asp406 and Asp407 (the DD motif) are vital. Furthermore, the DD motif is involved in activation of AMP-dependent protein kinase (AMPK), which is also important for mTORC1 inhibition. Besides GATOR2 binding and AMPK activation for mTOR inhibition, sestrin2 may also carry the guanosine nucleotide dissociation inhibition (GDI) function. However, mutation studies of Arg419/Lys422/Lys426 in Sesn-C suggested that whether these amino acid residues are truly critical for GDI functions is still in question [35].

The availability of amino acids is critical for the regulation of protein synthesis in living organisms. Leucine, one of the essential amino acids, is indispensable for this process and, more importantly, leucine was found to be crucial for mTORC1 activation in cells [45]. Located in the Sesn-C of hSesn2 (Figure 2A), charged residues Glu451 and Arg390, from two sides of a single binding pocket, anchor leucine in place through salt bridges with the free amine and carboxyl groups, respectively, whereas the isopropyl side chain of the bound leucine forms extensive hydrophobic interactions with residues Leu389, Trp444, and Phe447 in the pocket. In addition to contacting the charged sides and hydrophobic base of the pocket, three threonine residues (Thr374, Thr377, and Thr386) are positioned directly above the leucine to form a "lid" that encloses the top of the leucine, thereby locking the ligand in place [36]. As a leucine sensor, sestrin2 inhibits mTORC1 activity through the Rag guanosine triphosphatases (GTPase) and its regulators-GATOR1 and GATOR2. Thus, the binding of leucine with sestrin2 disrupts the connection of sestrin2 with GATOR2, allowing GATOR2 to enhance mTORC1 activity [36]. It has previously been demonstrated that adult sestrin2 gene knockout mice subject to a fasting/refeeding regimen or maintained with a high-fat diet suffered from various metabolic derangements, such as hepatosteatosis, insulin resistance, and glucose intolerance, with increased ROS extent and mTORC1 activity [38,46].

Despite the availability of the crystal structure of hSesn2, the detailed molecular information for sestrin1 and sestrin3 remains to be fully elucidated. However, sequence alignment of the three human sestrins revealed an overall 44.8% amino acid sequence identity [47]. Furthermore, the amino acid residues critical for alkyl hydroperoxidase activity (Cys125, His132, and Tyr127), GATOR2-binding and AMPK activation for mTORC1 inhibition (Asp406 and Asp407), and leucine-binding (Glu451 and Arg390; Leu389, Trp444, and Phe447; Thr374, Thr377, and Thr386) are all evolutionarily conserved in the three human sestrins. It is therefore reasonable to speculate that hSesn1 and hSesn3 may share most, if not all, of the functional roles of hSesn2. However, as compared with sestrin2, the potential involvement of sestrin1 and sestrin3 in nervous systems has been studied much less well. Below, in Figure 2B, is the list of known biological functions of all three sestrins.


**Figure 2.** The structural and functional domains as well as the biological functions of three sestrin members. (**A**) The strip diagram illustrates the three major structural domains (Sesn-A, Sesn-B, and Sesn-C). Cys125/Tyr127/His132, located within the Sesn-A domain, is critical for alkyl hydroperoxidase activity. The Asp406/Asp407 residues, the so-called "DD motif", located within Sesn-C are vital for GATOR2 binding and AMPK activation, both contributing to mTORC1 suppression. The leucine binding pocket spanning from Thr374 to Glu451 in the Sesn-C is also important for amino acid sensing and mTOR regulation. The guanosine nucleotide dissociation inhibition (GDI) domain containing Arg419/Lys422/Lys426 is also shown in Sesn-C. Based on the crystal structure, however, whether these amino acid residues are critical for GDI functions remains questionable. All the information was based on Kim et al., 2015 [35] and Saxton et al., 2016 [36]. (**B**) Potential biological functions of three sestrins are listed. Information was derived from UniProt (https://www.uniprot.org) for human sestrin1 [UniProtKB-Q9Y6P5 (SESN1\_HUMAN)], human sestrin2 [UniProtKB-P58004 (SESN2\_HUMAN), human sestrin3 [UniProtKB-P58005 (SESN3\_HUMAN)], and mouse sestrin3 [UniProtKB- Q9CYP7 (SESN3\_MOUSE)].

Expression of the sestrin2 genes is regulated by several critical transcription factors, enabling the cells to cope with various stressful insults. Initially the crucial role of the p53 tumor suppressor in regulating the expression of sestrin2 under hypoxic and genotoxic stress was revealed [33]. Later, additional studies revealed further transcription factors that are critical for the expression of sestrin2 under a variety of stressful conditions. Oxidative stress can activate the nuclear factor erythroid 2-related factor-2 (Nrf2) to regulate sestrin2

expression [48,49]. Hypoxia may induce sestrin2 expression where hypoxia-inducible factor-1 (HIF-1) may play a certain role [33,50–52], although the detailed mechanism is not well understood. In our earlier study [53], we found that brain-derived neurotrophic factor (BDNF) induced sestrin2 expression, which required dimerization of nuclear factorκB (NF-κB) subunits p65 and p50. Further, BDNF also enhanced production of nitric oxide (NO), formation of 3 ,5 -cyclic guanosine monophosphate (cGMP), and activation of cGMP-dependent protein kinase (PKG). Indeed, BDNF induced nuclear translocation of PKG-1 and its direct interaction with p65/p50 to form a ternary complex, thereby leading to heightened NF-κB binding to the sestrin2 gene promoter with resultant upregulation of its mRNA and proteins [53]. Apart from PKG/NF-κB, BDNF has also been shown to induce sestrin2 in neurons by activating transcription factor-4 (ATF4) [54]. In another recent study [40], we also found that NF-κB and p53 are involved in Aβ-induced sestrin2 expression in primary cortical neurons. Additional regulatory mechanisms responsible for sestrin2 induction under various stressful or physiological conditions may emerge in the near future

Nutrients including amino acids, lipids, and glucose are crucial for the biosynthetic processes in the cell. An inadequate supply of nutrients can seriously modify cellular metabolism. Sestrin2 activation may serve as one of the metabolic accommodations to nutrient deficiency in cells [38]. Glucose starvation, inhibition of glycolysis, and impairment of mitochondrial respiration can disrupt energy production, leading to the activation of two transcription factors, ATF4 and Nrf2, that can bind directly to the consensus sequences within the promoter to induce sestrin2 gene transcription [49,55–57]. ATF4 is also involved in the induction of sestrin2 as a result of a deficiency in amino acid supply in mouse embryonic fibroblasts [58]. The inadequacy of growth factors may result in the expression of sestrin2. It has been demonstrated in cancer cells that serum deprivation can activate the c-Jun N-terminal kinase (JNK) pathway and upregulate sestrin2 expression, which could be abolished by specific siRNAs against JNK1/2 or c-Jun [59]. Various physiological and pathological conditions, such as excessive ROS generation, ischemia, Ca2+ dyshomeostasis, and inflammatory response can all cause an accumulation of misfolded proteins in the endoplasmic reticulum (ER), with resultant ER stress [60]. ER stress may lead to cellular dysfunction and/or cell death and contributes to the progression of many diseases. Modulation of ER stress pathways may represent a potential therapeutic strategy. It was reported that activating transcription factor-6 (ATF6)-dependent sestrin2 induction can lessen the severity of ER stress-mediated liver injury [61]. In another study, it was shown that the hepatoprotective role of sestrin2 against chronic ER stress depends on the regulation of CCAAT-enhancer-binding protein-beta (c/EBPβ) [62]. Together, these previous reports identify the crucial roles played by sestrin2 in dealing with various cellular stresses under diverse physiological and pathological conditions. A simplified diagram (Figure 3) demonstrates that distinct transcription factors are activated under a variety of stressful conditions, thereby leading to induction of sestrin2 expression, which can regulate autophagy and contribute to antioxidation.

**Figure 3.** Brain trauma, stroke, neurological disorders, and aging induce hypoxia, the production of reactive oxygen species (ROS), Ca2+ overload, metabolic dyshomeostasis, and neuronal inflammation. Subsequently, the injury-induced signaling pathways promote sestrin2 expression via the activation of various transcription factors (which particular factors depending on which stressors), such as transcription factor-4 (ATF4), ATF6, hypoxia-inducible factor-1 (HIF-1), nuclear factor erythroid 2-related factor-2 (Nrf2), c-Jun N-terminal kinase (JNK)/c-Jun, and CCAAT-enhancer-binding proteinbeta (C/EBPβ). Sestrin2, as a sensor for essential amino acids with a leucine-binding pocket, also has a binding site for the GTPase-activating protein activity toward Rags-2 (GATOR2). In the presence of sufficient amino acids available for protein synthesis, sestrin2 may bind to leucine and release the bound GATOR2. The freed GATOR2 can then physically associate with GATOR1, which can no longer bind to, and hence inhibit, mTORC1, thereby promoting protein synthesis while inhibiting autophagy. Under the stressful condition in which amino acids are insufficient, binding of GATOR2 to sestrin2 allows GATOR1 to inhibit mTORC1, thereby promoting autophagy while inhibiting protein synthesis. In addition to regulating autophagy and protein synthesis via binding with leucine or GATOR2, the endogenous alkyl hydroperoxidase activity of sestrin2 also exerts direct antioxidative actions.

#### **3. Sestrin2 in Age-Related Clinical Conditions**

Persuasive evidence supports the notion that aging is related to various harmful mechanisms, such as escalation of oxidative stress, instability of genetic materials, declined protein homeostasis, impaired mitochondrial function, increased cellular senescence, and stem cell exhaustion [63]. The accumulation of various cellular damages among tissues in aging organisms leads eventually to functional breakdown, causing disability or death. Therefore, aging is believed to be a risk factor for various disorders, such as cardiovascular diseases, stroke, type II diabetes, cancers, and neurodegenerative diseases [63–65]. Inhibition of either ROS production or mTORC1 activation may counteract aging [35], and as sestrin2 is characterized by both these functions, it may exert such beneficial effects [66,67]. In fact, enhancement of sestrin2 expression reduces aging markers. Conversely, lessening sestrin2 expression accelerates aging processes [68].

Aging is a predetermined time-related deterioration in various physiological conditions, and is a critical risk factor for cancer development. Cancer and aging involve similar processes of progressive time-dependent cellular damage. As sestrin2 is critically involved

in aging [38,67], it may play a pivotal role in cancer progression, and is regarded as a potential tumor suppressor. In non-small cell lung cancer patients, higher sestrin2 expression was a favorable prognostic factor, while lower sestrin2 expression was accompanied by poor tumor cell differentiation, as well as more advanced staging in terms of tumor, node, and metastasis (TNM) [69]. It was shown that colorectal cancer patients with lower expression of sestrin2 showed poor prognostic outcomes [70]. Docosahexaenoic acid (DHA) can increase oxaliplatin-induced autophagic cell death through the ER stress/sestrin2 pathway in colorectal cancer [71], whereas downregulation of sestrin2 can accelerate colon carcinogenesis [72].

Hypernutrition, causing obesity, hepatosteatosis, and insulin resistance, is related to chronic activation of p70S6 kinase and mTORC1 [73]. Activation of sestrin2 can lower the extents of fatty liver and insulin resistance [73]. Sestrin2 can activate AMPK, inhibit mTORC1 activity, and maintain a high AKT level to suppress the extent of gluconeogenesis in the liver, thereby reducing the level of blood sugar. Sestrin2-deficient obese mice were found to present an evident decline of AKT activity, leading to insulin resistance and a higher level of glucose production [73]. In a recent study, serum levels of sestrins are significantly decreased in patients with diabetes and dyslipidemia. It appears that sestrin2 levels are robustly associated with diabetes, dyslipidemia, atherosclerosis, and the atherogenic index [74]. Declined serum sestrin2 levels were also observed in diabetic patients with nephropathy, particularly in those with macroalbuminuria [75].

It was demonstrated previously that loss of dSestrin (the only one sestrin homologue in Drosophila) results in age-associated pathologies, including cardiac dysfunction, muscle degeneration, and triglyceride accumulation. The cardiac dysfunction showed reduced heart rate and compromised heart function. The detrimental effects induced by dSestrin deficiency were generally inhibited by AICAR and rapamycin, the AMPK activator and the mTORC1 inhibitor, respectively [67]. These results indicate that the sestrin family may play crucial roles in the pathophysiology of cardiac regulation [76]. In a recent review article, sestrin2 is considered a rising star among antioxidants, with future therapeutic potential for reducing heart injury induced by oxidative stress, promoting cell survival through the activation of Nrf2/AMPK, and inhibiting mTORC1 to combat various cardiovascular diseases, such as cardiomyopathy, heart failure, and myocardial infarction [77]. Despite these promises, however, the occurrence of major adverse cardiac events is predicted in patients with chronic heart failure who have higher plasma sestrin2 concentrations [78]. The conflicting results as far as the beneficial or detrimental effects of sestrin2 in heart failure are concerned await further clarification.

Stroke is the most common age-related cerebral vascular disease and the chief cause of physical and intellectual disability in adults, as well as the leading cause of mortality in developed countries [79]. Several studies have investigated the roles of sestrin2 in cerebral ischemia [80–83]. It was demonstrated that sestrin2 can activate the Nrf2/heme oxygenase-1 (HO-1) pathway, leading to augmentation of angiogenesis following focal cerebral ischemia [82]. Another study also showed the critical role of sestrin2 in promoting angiogenesis in focal cerebral ischemia by activating the Nrf2/p62 pathway [81]. In contrast, silencing sestrin2 expression may reduce mitochondrial activity, suppress mitochondrial biogenesis, and ultimately exacerbate cerebral ischemia/reperfusion injury by preventing the AMPK/PGC-1α pathway [83]. Although sestrin2 seems to have pro-survival characteristics in the context of ischemic brain injury, the anti-inflammatory role of sestrin2 is unknown. In a recent study, it was demonstrated that sestrin2 exerts neuroprotective effects by changing microglial polarization and mitigating the extent of inflammation in the ischemic mouse brain, which may be due to the inhibition of the mTOR pathway and the restoration of autophagic flux [80]. It is to be expected that knowledge of the mechanisms underlying additional protective effects of sestrin2 may emerge in the not too distant future.

#### **4. Potential Roles of Sestrin2 in Age-Related Neurodegenerative Diseases: Focusing on AD**

As mentioned above, the sequences of the critical amino acid residues important for known biological activities of hSesn2, including alkyl hydroperoxide reductase, mTORC1 inhibition, and leucine binding, are also conserved in hSesn1 and hSesn3. However, the crystal structures of sestrin1 and sestrin3 are still not available. Nevertheless, there are a few studies implicating sestrin1 and sestrin3 in nervous system disorders. For example, sestrin1 may exert protective effects in oxygen-glucose deprivation/reoxygenation (OGD/R) induced neuronal injury, a cellular model for mimicking cerebral ischemia/reperfusion injury in vitro [84]. Furthermore, sestrin3 has been identified as a pro-convulsant gene network in the human epileptic hippocampus [85]. Results derived from sestrin3 knockout rats also suggested that sestrin3 may increase the occurrence and/or severity of seizures [86]. Conversely, silencing rno-miR-155-5p in vivo mitigated the pathophysiological features associated with the status epilepticus, which was accompanied by attenuation of apoptosis in the hippocampus, by enhancing expression of sestrin3 in rats, implying that sestrin3 plays a beneficial role in offsetting temporal lobe epilepsy [87]. Further dissection of the pathophysiological roles of sestrin1 and sestrin3 will require a greater understanding of their molecular structures, as well as the upstream regulatory mechanisms involved in their expression in nervous systems.

Among age-related disorders, chronic neurodegenerative diseases are particularly concerning due to the lack of efficacious treatments, their irremediable clinical course, and their association with substantial social-economic burdens [1–3]. The potential roles of sestrin2 in combatting neurodegenerative diseases, including AD, Parkinson's disease (PD), and Huntington's disease (HD), while still awaiting further evidence, have gradually been recognized in recent years.

It is widely accepted that maintaining proper levels of reactive nitrogen species and ROS are crucial for ensuring regular neuronal function [88]. Yet, excessive ROS generation with heightened levels of oxidation in lipids, proteins, and DNA, or inherent lower antioxidant competence in the brain, may have detrimental effects on the organism and play a role in the pathophysiology of various chronic neurodegenerative diseases, including AD, PD, and HD [89,90]. Numerous mechanisms underlie oxidative stressmediated neurodegeneration; these include calcium overload, glutamate excitotoxicity, inflammation, functional impairment of mitochondria, and apoptotic processes [88]. The ability to lessen these harmful effects may be the key to developing effective treatments for neurodegenerative diseases.

As mentioned above, sestrin2, with its dual functions, can directly reduce oxidative stress through restoring overoxidized peroxiredoxins, and indirectly lessen oxidative stress through regulating mTOR to augment the activity of autophagy, or specifically, mitophagy, to remove the worn-out or damaged mitochondria with higher levels of electron leakage and hence free radical production. The N-terminal domain of sestrin2 decreases oxidative stress by its helix-turn-helix motif, while the C-terminal domain of sestrin2 may physically associate with GATOR2, thereby causing the inhibition of mTORC1 [35]. Apart from the effect of oxidative stress, one more common pathogenic mechanism in chronic neurodegeneration is the deposition of aberrant and/or misfolded proteins, such as Aβ and tau protein in AD, Lewy body (LB) in PD, and mutant huntingtin in HD. Enhancing the activity of autophagy may help to eradicate neuronal dysfunction induced by misfolded proteins, thereby opening an opportunity towards developing a new therapeutic strategy for treating neurodegenerative diseases [91]. The dual biological functions of sestrin2, with increasing antioxidative ability and autophagy-promoting activity to eliminate aggregated proteins and damaged mitochondria, give this molecule a unique position in protecting neurons against degeneration.

PD is the second most common aging-related neurodegenerative disease that mainly presents syndromes with slow movements, tremors, and rigidity. The underlying cause of PD is not well understood but may involve various genetic and environmental factors [92]. The main pathological feature of PD is LB, which is composed of ubiquitin-bound, misfolded α-synuclein protein in the dopamine neurons in the substantia nigra of the midbrain [93,94]. In an in vitro PD model with 1-methyl-4-phenylpyridinium (MPP+), it was revealed that MPP<sup>+</sup> neurotoxicity increases sestrin2 expression, whereas downregulation of sestrin2 with small interference RNA augments MPP+-related neurotoxicity in SH-SY5Y cells [95]. In another in vivo PD model induced by rotenone, sestrin2 exerts a protective effect over dopaminergic neurons against rotenone-induced neurotoxicity by activating an AMPK-dependent autophagy pathway [96]. In a clinical study, serum sestrin2 levels were found to be elevated in PD patients compared to controls [97]. In postmortem human samples, it was found that PD patients had higher expression levels of sestrin2 in the midbrain [95].

No report was available concerning HD and sestrin2 either in the clinical or pre-clinical studies. 3-Nitropropionic acid (3-NP) can inhibit the function of the mitochondrial respiratory complex II (also named succinate dehydrogenase), decrease ATP production, impair cellular energy metabolism, aggravate the extent of oxidative stress, cause mitochondrial DNA damage, and thus impair the function of mitochondria [98,99]. Although genetic models of HD are more popular due to their similarity to the phenotypes observed in HD, 3-NP is still a useful model to study neurotoxic phenomena, mitochondrial alterations, and neuroprotective effects for HD patients [100]. Therefore, 3-NP has been used as a pharmacological model to study neurodegeneration and neuronal death involving mitochondrial dysfunction in HD [101]. Despite the indirect relationship, we have shown that BDNF protects 3-NP-induced oxidative stress through augmenting sestrin2 expression. Furthermore, BDNF induction of sestrin2 implicates the NO/PKG/NF-κB pathway [53]. This study thus highlights the probable beneficial role of sestrin2 in this devastating hereditary neurodegenerative disease. Understanding the potential role of sestrin2 in impeding HD pathogenesis may require further investigation into the genetic models of HD, such as R6/2 or other knock-in mice.

AD is the most common age-related neurodegenerative disease involving various pathogenic mechanisms such as excitotoxicity, excessive generation of ROS, aberrant cell cycle reentry, impaired mitochondrial function, and DNA damage [15–19]. Although emerging roles of sestrin2 in various neurological diseases have been suggested before [102], limited studies concerning sestrin2 and AD have been reported [39,40,103–107]. In a 2003 study, in which human neuroblastoma CHP134 cells were analyzed with cDNA microarray technology with confirmation by semi-quantitative RT-PCR, it was revealed that sestrin2 is overexpressed under treatment of Aβ [107]. Furthermore, in human neuroblastoma SH-SY5Y cells, Aβ1-42 dose-dependently enhanced sestrin2 expression, whereas cotreatment with atorvastatin reversed sestrin2 back to the control level [103]. We have also demonstrated, in primary cortical neurons, that both Aβ25-35 and Aβ1-42 triggered the expression of sestrin2 [39,40], as is discussed in more detail below. In addition to these pre-clinical studies, the first human study reported in 2012 using postmortem brain tissues from advanced AD patients with immunohistochemistry findings showed intense sestrin2 expression in the neuropil, which may suggest a diffuse expression in various components among neurons, glia, and vascular cells. Using double-labeling immunofluorescence microscopy, co-localization between phosphorylated tau and sestrin2 is observed in the neurons and the neurites in neurofibrillary lesions [106]. These findings together implied that sestrin2 is expressed at least in the neurons of AD patients. Another clinical study demonstrated significant overexpression of sestrin2 protein and mRNA in the serum of AD patients as compared to the mild cognitive impairment (MCI) and the age-matched control groups. A difference in serum sestrin2 concentration between MCI and the control groups was also evident. However, no significant difference in sestrin1 levels was observed among the study groups. These results therefore suggested the potential role of sestrin2 as a biomarker in the analysis of peripheral blood in AD patients, and highlighted the importance of sestrin2, as opposed to sestrin1, in the progression of AD [104]. Despite these arguments supporting the important roles of sestrin2 in AD, it should be noted

that, with similar biological functions and significantly conserved amino acid sequences identified across the different members of the sestrin family, although potential involvements of sestrin1 and sestrin3 in AD have not been reported, they certainly cannot be overlooked. Overall, this review has only focused on discussing the potential roles of sestrin2 in neurodegenerative disorders, AD in particular.

We have explored the potential link between sestrin2 and Aβ-induced neurotoxicity [39,40]. In an in vitro study, we demonstrated that sestrin2 was induced by Aβs, including both Aβ25-35 and Aβ1-42, in primary culture of fetal rat cortical neurons. We further showed an in vivo result of increased sestrin2 expression in the aged APPswe/PSEN1dE9 transgenic mice. More importantly, sestrin2 functions as an endogenous protective moderator, through the adjustment of autophagy, against Aβ-induced neurotoxicity [39]. It is well known that sestrin2 has an antioxidant character and plays a critical role in agerelated diseases [66]. In our recent report [40], Aβ-induced sestrin2 expression in primary cortical neurons was found to have an antioxidant effect, resulting in the suppression of Aβ-mediated ROS production, enhancement of lipid peroxidation, and formation of 8-hydroxy-2-deoxyguanosine (8-OH-dG) as an index of oxidative DNA damage. Interestingly, we found that lentivirus-mediated overexpression of the N-terminal domain of sestrin2 in primary cortical neurons completely blocked Aβ25-35-induced ROS production, whereas overexpression of the C-terminal domain partially, but statistically significantly, suppressed ROS formation. Although the sestrin2 C-terminal domain is known to have the capability of inhibiting mTORC1 to promote autophagy [35], we speculated that augmentation of autophagy with enhanced removal of damaged mitochondria, or mitophagy, may also contribute to the antioxidant function of sestrin2. Upstream of sestrin2, we found that the observed Aβ effect on sestrin2 expression is at least partially mediated by p53 and NF-κB. Indeed, apart from regulating sestrin2 induction, p53 and NF-κB subunits p65/p50 also affect the expression of each other [40]. Furthermore, upstream of p53 and NF-κB, we identified at least two signaling pathways, namely nitric oxide synthase/cGMP-dependent protein kinase (NOS/PKG) and phosphatidylinositol 3-kinase (PI3K)/Akt, that may have contributed to the observed Aβ induction of sestrin2 in cortical neurons [40]. A diagram summarizing our findings is shown in Figure 4, below.

The synaptic activity of neurons can affect the homeostasis of Aβ and tau. Both are aggregated and accumulated during the progression of AD and are critical for neuronal function. Furthermore, impairment of synaptic activity is linked with AD [108]. Physiologic synaptic activity, through NMDA receptor signaling, can enhance antioxidant activity and increase sestrin2 expression to exert a protective effect through transcription factor C/EBPbeta [109]. Presenilin proteins are catalytic components of γ-secretase involved in various functions such as proteolytic cleavage of the Notch and APP, adjustment of neurotransmitter release, and are vital for the survival of neurons in aging [110]. Mutations of the presenilin genes are one of the main causes of familial AD [111]. Impairment of presenilin activity may compromise synaptic functions, resulting in neurodegeneration and ultimately dementia [112]. It was demonstrated that cells deficient in presenilin have lower levels of sestrin2 and are accompanied with mTORC1 dysregulation. These findings show that sestrin2, through attenuation of oxidative stress and its nutrient-sensing ability via mTOR, plays a critical role in AD-related conditions [105].

Emerging evidence suggested the potential benefit of sestrin2 in AD. Medications with the capability to alter sestrin2 expression may therefore have the potential to prevent or delay the clinical deterioration of this neurodegenerative disease. It was previously shown that atorvastatin reduces Aβ-induced synaptotoxicity and memory impairment through a p38MAP kinase pathway [113]. Atorvastatin could also activate autophagy through AMPK/mTOR signaling [113,114]. In a recent study, it was demonstrated that sestrin2 and the autophagy marker LC3II were increased with Aβ treatment in human neuroblastoma cells; co-treatment of atorvastatin and Aβ reduced oxidative stress and decreased sestrin2 expression [103]. We have shown before that BDNF can induce sestrin2 expression in rat primary cortical neurons and exert a protective effect against 3-NP neurotoxicity by reducing the production of free radicals [53]. BDNF is known to protect against Aβ-induced neurotoxicity in vitro as well as in rodent and primate models [115,116]. However, whether sestrin2 induction by BDNF contributes to this neuroprotective effect has not been tested. The possibility certainly cannot, however, be excluded.

**Figure 4.** Amyloid-beta peptide (Aβ) enhances calcium dyshomeostasis and the generation of reactive oxygen species (ROS), thereby leading to oxidative stress with damaged mitochondria. Meanwhile, Aβ also induces p53, as well as nuclear factor-kappaB (NF-κB) subunits p65 and p50 via activation of nitric oxide synthase (NOS)/cGMP-dependent protein kinase (PKG) and phosphatidylinositol 3-kinase (PI3K)/Akt. The transcription factors, p50, p65, and p53 translocate into the nucleus of the neuron to promote expression of sestrin2 mRNA, as indicated by the red dashed arrow. The alkyl hydroperoxidase activity of sestrin2 may neutralize excessive ROS generated by Aβ with antioxidative functions. In addition, sestrin2 may trigger autophagy, as is indicated by the conversion of the microtubule-associated protein-1 light-chain 3B-I (LC3B-I) into LC3B-II, and possibly also mitophagy, in order to remove Aβ-damaged mitochondria known to produce more ROS. Sestrin2 thus may function as an endogenous protective mediator inducible by Aβ that contributes to neuronal survival against Aβ neurotoxicity.

In addition to alkyl hydroperoxidase activity and enhanced autophagy to alleviate oxidative stress, sestrin2 may also trigger the Nrf2/ARE pathway to augment antioxidant responses. For example, following photochemical cerebral ischemia in rats, expression of sestrin2, Nrf2, HO-1, and VEGF were significantly increased. Overexpression of sestrin2 by AAV injection further enhanced their expression [82]. In another study of photothrombotic ischemia in rats, sestrin2 may promote angiogenesis by activating Nrf2 via upregulation of p62 with enhanced interaction between p62 and Keap1, thereby improving the neurological function, reducing brain infarction, and alleviating brain edema [81]. Sestrin2 was also a direct target of microRNA miR-148b-3p in the HT22 hippocampal neurons challenged with OGD/R. Furthermore, Nrf2/ARE was a downstream antioxidant signal contributing to the observed protective effects through miR-148b-3p inhibition, and hence sestrin2 induction, in response to OGD/R injury [117]. In the H2O2-stimulated retinal ganglion cells (RGCs), sestrin2 overexpression increased the nuclear translocation of Nrf2, thereby upregulating the Nrf2/ARE target genes, including HO-1 and NAD(P)H quinone oxidoreductase-1 [118]. As mentioned above, sestrin2 itself may be a downstream target of Nrf2 [48,49]. Although

these studies were conducted in non-neuronal cells like mammary epithelial cells and hepatocytes, the possibility that Nrf2 activation may induce sestrin2 expression in the nervous system cannot be excluded. Whether sestrin2 may trigger its own expression, thereby forming a positive feedforward loop, via Nrf2/ARE in neurons, also requires further investigation. The potential role of sestrin2 in age-related neurodegenerative diseases is demonstrated in Figure 5.

**Figure 5.** Multiple pathogenic mechanisms including oxidative stress, with excessive production of reactive oxygen species (ROS), glutamate-induced excitotoxicity, calcium overload, mitochondrial dysfunction, and inflammation contribute to neuronal death in various neurodegenerative disorders like Alzheimer's disease (AD), Parkinson's disease (PD), and Huntington's disease (HD). Brainderived neurotrophic factor (BDNF) enhances sestrin2 expression via signaling pathways involving nitric oxide (NO)/3 ,5 -cyclic guanosine monophosphate (cGMP)-dependent protein kinase-1 (PKG-1)/nuclear factor-kappaB (NF-κB). In addition to the alkyl hydroperoxidase activity and autophagy promotion, sestrin2 may also have antioxidant properties by activating nuclear factor erythroid 2-related factor-2 (Nrf2) with enhanced expression of antioxidant proteins like heme oxygenase-1 (HO-1), vascular endothelial growth factor (VEGF), and NAD(P)H quinone oxidoreductase-1. These antioxidant proteins then mitigate oxidative stress, as indicated by the red arrow, that is commonly observed in various neurodegenerative diseases. The possibility that BDNF may exert its neuroprotective effects, in addition to its well-known neurotrophic actions, via induction of sestrin2 in various neurodegenerative disorders, requires further investigation.

#### **5. Medications or Chemical Compounds Capable of Altering Sestrin2 Expression**

The outcomes of clinical trials using drugs to target amyloid and tau have been unsatisfactory up to now, thereby leading to enthusiasm in targeting alternative mechanisms in AD studies [119,120]. Drug repurposing involves taking the research into an existing, readyto-use drug and assessing its therapeutic potential with respect to another disease [121,122]. Several well-known success stories include aspirin, sildenafil, and thalidomide [123]. This approach may provide a less expensive and quicker method of drug discovery. Several recent review articles emphasize the clinical potential of drug repurposing in the context of AD [120,124–126]. It would be worthwhile to search among medications with neuroprotective effects, as these are likely to have a better chance of achieving clinically meaningful results with neurodegenerative diseases [127]. The potential of certain medications to activate sestrin2 expression requires further investigation.

Several studies revealed that certain drugs capable of activating sestrin2 expression in various disease models may be worth testing in AD as well. It was shown that empagliflozin, which is a sodium-glucose cotransporter 2 (SGLT2) inhibitor useful for treating diabetes mellitus (DM) patients, can regulate sestrin2, the AMPK-mTOR pathway, and ROS homeostasis to improve obesity-related cardiac dysfunction in mice [128]. Another study demonstrated that liraglutide, a glucagon-like peptide 1 (GLP-1) agonist for DM patients, may lessen obesity-related fatty liver disease through regulating the sestrin2-mediated Nrf2/HO-1 pathway [129].

5-Fluorouracil is an antimetabolite widely used for chemotherapeutic treatment of cancers [130,131]. It was shown that 5-fluorouracil increases sestrin2 levels in a p53-dependent pathway and inhibits cancer cell migration in an in vitro colon cancer study [132]. Nelfinavir, an ER stress-inducing agent, and bortezomib, a proteasome inhibitor, can both enhance sestrin2 expression, which may be useful to treat cancers [133]. Interestingly, nelfinavir inhibited endogenous Aβ1-40 production from primary cultured human cortical neurons [134]. Whether these reagents may also carry therapeutic potential for AD requires further investigation.

Other chemical compounds such as resveratrol and melatonin possessing pleiotropic effects like antioxidancy or anti-inflammation were studied based on their capability of upregulating sestrin2 in various disease models [135–137]. Resveratrol is a naturally occurring polyphenol that is abundant in grape seeds and skin [138,139]. It can offer protective effects against various age-related diseases like AD through diverse mechanisms [138,140]. These molecular mechanisms include modulation of NF-κB, regulation of inflammatory cytokines, production of antioxidant enzymes, angiogenesis, apoptosis, lipid metabolism, and mitochondrial biogenesis-all critical for its potential clinical application [141,142]. It was demonstrated before that resveratrol affects sestrin2 gene induction and inhibits liver X receptor-alpha (LXRα)-mediated hepatic lipogenesis [137]. Methylglyoxal is implicated in the formation of advanced glycation end-products associated with diabetes and age-related neurodegenerative diseases [143]. In a previous study using methylglyoxal to induce cell death in HepG2, a human liver cancer cell line, it was found that resveratrol reduces methylglyoxal-induced mitochondrial impairment and apoptosis through sestrin2 induction [136]. Other flavonoid polyphenols or flavone derivatives, such as eupatilin [144,145], pentamethylquercetin [146], and isorhamentin [147], also possess the capability to alter sestrin2 expression and are worth studying further in AD models.

Melatonin, a molecule widely distributed in living organisms, is involved in various physiological and biological functions among diverse tissues and organs. It possesses prominent antioxidant effects, functions as a free radical scavenger, augments antioxidant enzymes, lessens mitochondrial electron leakage, and reduces pro-inflammatory signaling pathways [148]. These properties of melatonin underline the possibility for future clinical use in numerous disorders, including neurodegeneration [149]. It was shown that melatonin can inhibit proliferation and apoptosis in the vascular smooth muscle through upregulation of sestrin2, which may be important in preventing atherosclerosis and restenosis of vessel lumen [135]. It would be interesting to know the effect of sestrin2 expression under melatonin treatment in a stressful condition, such as in Aβ-induced neurotoxicity.

It is believed that a long list of medications, natural products, chemical compounds, or small molecules capable of altering sestrin2 expression may exert beneficial effects over AD-related mechanisms. This awaits further investigation and may lead to more opportunities for treating such devastating neurodegenerative diseases as AD.

#### **6. Conclusions and Future Perspectives**

Being a member of the sestrin family, sestrin2 acts as a crucial intracellular detector capable of regulating various biological processes to maintain the homeostasis of living organisms. Emerging evidence reveals that sestrin2 may have beneficial effects for vulnerable cells, such that they may adapt to numerous pathological situations under diverse stressful conditions, including DNA injury, hypoxic state, metabolic dyshomeostasis, and oxidative stress. In age-related neurodegenerative disorders, excessive generation of ROS and dysfunction of autophagy may play pivotal roles in the pathogenesis among these diseases. Sestrin2, with distinctive dual-functional sites to counteract excessive ROS generation and inhibit mTOR activity for autophagy promotion, is presumed to play a crucial role in AD, although at present only limited information is available to firmly establish this notion. Certain medicinal compounds or natural products, such as flavonoid-related products, can alter the expression levels of sestrin2. It is believed that any means of increasing sestrin2 expression may possess significant clinical implications for the abatement of AD-related neurodegeneration. The possibility awaits further investigation. It is uncertain, however, whether the overactivation of sestrin2 may result in detrimental effects due to autophagic dysfunction. It may be difficult to determine the pros and cons of excessive activation or inhibition of autophagy in terms of neurodegenerative diseases, including AD. This concern further reveals the crucial need for a thorough understanding of both the downstream targets, as well as the upstream regulators, of sestrin2. Fuller elucidation of the signaling pathways of sestrin2 would accelerate the discovery of novel therapies for disease treatment, especially for those diseases with a devastating clinical course, such as AD.

**Funding:** This study was supported by the Ministry of Science and Technology (MOST) in Taiwan (MOST 104-2314-B-010-014-MY2, MOST 107-2314-B-010-020-MY3, and MOST 109-2314-B-010-038-MY3 to Ding-I Yang; MOST 108-2314-B-037-038-MY3 to A-Ching Chao; MOST 109-2314-B-182A-078-MY3 to Shang-Der Chen; MOST 108-2320-B-182A-005-MY3 to Jenq-Lin Yang), Department of Health in Taipei City Government (11001-62-038 to Ding-I Yang), and Chang Gung Medical Foundation (CMRPG8I0051, CMRPG8I0052, and CMRPG8I0053 to Shang-Der Chen; CMRPG8K0652 to Jenq-Lin Yang). This study was also financially supported by Kaohsiung Medical University Hospital (KMUH109-9R72 to A-Ching Chao) and Brain Research Center, National Yang Ming Chiao Tung University, from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan (110BRC-B407 to Ding-I Yang).

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

#### **References**

	- 50. Olson, N.; Hristova, M.; Heintz, N.H.; Lounsbury, K.M.; Van Der Vliet, A. Activation of hypoxia-inducible factor-1 protects airway epithelium against oxidant-induced barrier dysfunction. *Am. J. Physiol. Cell. Mol. Physiol.* **2011**, *301*, L993–L1002. [CrossRef]

*Case Report*
