HDAC Inhibitors Induce BDNF Expression and Promote Neurite Outgrowth in Human Neural Progenitor Cells-Derived Neurons

Besides its key role in neural development, brain-derived neurotrophic factor (BDNF) is important for long-term potentiation and neurogenesis, which makes it a critical factor in learning and memory. Due to the important role of BDNF in synaptic function and plasticity, an in-house epigenetic library was screened against human neural progenitor cells (HNPCs) and WS1 human skin fibroblast cells using Cell-to-Ct assay kit to identify the small compounds capable of modulating the BDNF expression. In addition to two well-known hydroxamic acid-based histone deacetylase inhibitors (hb-HDACis), SAHA and TSA, several structurally similar HDAC inhibitors including SB-939, PCI-24781 and JNJ-26481585 with even higher impact on BDNF expression, were discovered in this study. Furthermore, by using well-developed immunohistochemistry assays, the selected compounds were also proved to have neurogenic potential improving the neurite outgrowth in HNPCs-derived neurons. In conclusion, we proved the neurogenic potential of several hb-HDACis, alongside their ability to enhance BDNF expression, which by modulating the neurogenesis and/or compensating for neuronal loss, could be propitious for treatment of neurological disorders.


Introduction
Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS) are the major neurodegenerative diseases imposing a considerable burden on health care systems worldwide. With improved life expectancy in recent years in many countries, the prevalence and incidence of these age-related diseases is also on the rise [1].
Reduction in histone acetylation is considered to be a common cause of various brain pathologies such as neurodegenerative and neurodevelopmental cognitive disorders [2]. In the wake of studies showing the key role of histone acetylation in memory formation, researchers have investigated the HDACis potential to improve memory formation in diseases characterized by memory loss and Int. J. Mol To maximize the available amount of RNA, Cell-to-Ct assay kit was used to extract the RNA and to make corresponding complementary DNA (cDNA). We found that 67 and 57 compounds were effective on the BDNF expression in HNPCs and fibroblast cells, respectively. Z-factor, as a measure of assay suitability for high-throughput screening, was determined. A Z-factor > 0.5 for primary screening in HNPCs was considered a highly reliable assay. The coefficient of variation was another standard parameter used to measure the suitability of the assay for high-throughput screening (HTS); a coefficient of variation <10% was used as a primary screening for HNPCs experiment suitability for HTS.

HNPCs and Fibroblasts Shared Common Hits
HNPCs and fibroblast cells shared several common small molecules belonging to the same functionally related category (Table 1 and Figure 2). Most (19 of 26) of the shared compounds belonged to HDACis category. Methyltransferase inhibitors and PARP-1 inhibitors, each with two compounds, were two other functionally related categories. HAT inhibitor, histone demethylase inhibitor and natural compound, sharing only one compound in each category, are also contributing in this list. To maximize the available amount of RNA, Cell-to-Ct assay kit was used to extract the RNA and to make corresponding complementary DNA (cDNA). We found that 67 and 57 compounds were effective on the BDNF expression in HNPCs and fibroblast cells, respectively. Z-factor, as a measure of assay suitability for high-throughput screening, was determined. A Z-factor > 0.5 for primary screening in HNPCs was considered a highly reliable assay. The coefficient of variation was another standard parameter used to measure the suitability of the assay for high-throughput screening (HTS); a coefficient of variation <10% was used as a primary screening for HNPCs experiment suitability for HTS.

HNPCs and Fibroblasts Shared Common Hits
HNPCs and fibroblast cells shared several common small molecules belonging to the same functionally related category (Table 1 and Figure 2). Most (19 of 26) of the shared compounds belonged to HDACis category. Methyltransferase inhibitors and PARP-1 inhibitors, each with two compounds, were two other functionally related categories. HAT inhibitor, histone demethylase inhibitor and natural compound, sharing only one compound in each category, are also contributing in this list.

HDACis Increased BDNF in Secondary Assays
To confirm the observed data from primary screenings, the effective compounds were tested in Hek293 to assess their ability to increase the BDNF gene mRNA expression ( Figure 3). Trichostatin A was used as positive control for its validated effect on overexpression of BDNF [26,27]. JNJ-26481585, PCI-24781 and SB-939 were used as putative BDNF expression enhancers shared between HNPCs and fibroblast cells. Ellagic acid, Droxinostat and Entacapone were chosen as neutral controls, and were proved not to be effective in primary screening tests. Moreover, we used another compound, I-Bet 151, which was shown to have the opposite effects on BDNF expression in HNPCs and fibroblast cells-increasing BDNF mRNA expression in HNPCs but decreasing it in fibroblasts. qRT-PCR experiments with TaqMan probes confirmed test compounds to be effective in a dose-dependent manner, enhancing the BDNF mRNA expression with elevated concentrations. In case of I-Bet 151, the BDNF mRNA expression was decreased in Hek293 cells, not confirming the effect observed in HNPCs but replicating the effect observed in fibroblast cells (data not shown). Since the mRNA level in a cell is not always proportional to the amount of the corresponding protein, we decided to perform an ELISA test to measure BDNF protein level upon treating the Hek293 cell line with test compounds that was proven to be effective on BDNF mRNA expression ( Figure 4). All selected compounds significantly elevated the BDNF protein level, most notably for SB-939 and JNJ-26481585 with around 30 pg/mL for both compounds at 1 and 0.1 µM concentrations, respectively. Finally, a set of three compounds, all belonging to hydroxamic acid-based HDACis, were verified from the combination of primary screening and secondary validations for further experiments.

HDACis Increased BDNF in Secondary Assays
To confirm the observed data from primary screenings, the effective compounds were tested in Hek293 to assess their ability to increase the BDNF gene mRNA expression ( Figure 3). Trichostatin A was used as positive control for its validated effect on overexpression of BDNF [26,27]. JNJ-26481585, PCI-24781 and SB-939 were used as putative BDNF expression enhancers shared between HNPCs and fibroblast cells. Ellagic acid, Droxinostat and Entacapone were chosen as neutral controls, and were proved not to be effective in primary screening tests. Moreover, we used another compound, I-Bet 151, which was shown to have the opposite effects on BDNF expression in HNPCs and fibroblast cells-increasing BDNF mRNA expression in HNPCs but decreasing it in fibroblasts. qRT-PCR experiments with TaqMan probes confirmed test compounds to be effective in a dose-dependent manner, enhancing the BDNF mRNA expression with elevated concentrations. In case of I-Bet 151, the BDNF mRNA expression was decreased in Hek293 cells, not confirming the effect observed in HNPCs but replicating the effect observed in fibroblast cells (data not shown). Since the mRNA level in a cell is not always proportional to the amount of the corresponding protein, we decided to perform an ELISA test to measure BDNF protein level upon treating the Hek293 cell line with test compounds that was proven to be effective on BDNF mRNA expression ( Figure 4). All selected compounds significantly elevated the BDNF protein level, most notably for SB-939 and JNJ-26481585 with around 30 pg/mL for both compounds at 1 and 0.1 µM concentrations, respectively. Finally, a set of three compounds, all belonging to hydroxamic acid-based HDACis, were verified from the combination of primary screening and secondary validations for further experiments.

Neurite Length Was Significantly Enhanced upon Treatment with Hydroxamate-Based HDACis
We showed that test compounds could increase the BDNF expression at mRNA and protein levels. We subsequently investigated whether the test compounds would be effective on neurite outgrowth in HNPCs-derived neurons ( Figure 5). Since TSA has been reported to increase the neurite length, it was first checked and upon confirmation, used as a positive control [28]. Neurons were labeled with anti-β-tubulin III antibody and the lengths of neurites were measured 24 h after the treatment. With an average neurite length of 73.5 µm, the cells treated with PCI-24781 at a concentration of 0.1 µM had the longest neurons-37% more than the cells treated with 0.1% DMSO (as control). JNJ-26481585 at a concentration of 0.001 µM resulted in the second highest increase in the length of neurites by 72.3 µm-34% more than that with the control. Despite slight differences observed, all test compounds were demonstrated to be significantly effective on the neurite length.

Neurite Length Was Significantly Enhanced upon Treatment with Hydroxamate-Based HDACis
We showed that test compounds could increase the BDNF expression at mRNA and protein levels. We subsequently investigated whether the test compounds would be effective on neurite outgrowth in HNPCs-derived neurons ( Figure 5). Since TSA has been reported to increase the neurite length, it was first checked and upon confirmation, used as a positive control [28]. Neurons were labeled with anti-β-tubulin III antibody and the lengths of neurites were measured 24 hours after the treatment. With an average neurite length of 73.5 µm, the cells treated with PCI-24781 at a concentration of 0.1 µM had the longest neurons-37% more than the cells treated with 0.1% DMSO (as control). JNJ-26481585 at a concentration of 0.001 µM resulted in the second highest increase in the length of neurites by 72.3 µm-34% more than that with the control. Despite slight differences observed, all test compounds were demonstrated to be significantly effective on the neurite length. As depicted, all test compounds were able to induce neurite outgrowth: PCI24781 (0.1 µM) had the greatest effect with 37% increase compared with that of the 0.1% DMSO. It was followed by JNJ26481585 (0.001 µM) with 34% increase. Trichostatin A was used as a positive control. Quantitative analysis of neurite length was done using ImageJ software. Two fields of view were analyzed per well. Two biological replicates were used per treatment and two independent experiments were done. Error bars represent SEM; *** p < 0.001, **** p < 0.0001.

The Hydroxamate-Based HDACis Showed No Toxicity at the Concentrations Used
Except for JNJ-26481585, which was tested at a concentration of 0.001 µM, other compounds were tested at two different doses-0.01 and 0.1 µM-for cell toxicity. Neuronal cells viability was assessed with Cell-Titer Glo assay. HNPCs were seeded and differentiated to neurons according to the same exact method used for neurite outgrowth assay. The viability did not significantly change relative to the control (0.1% DMSO) in all test compounds ( Figure 6).   ), and trichostatin A, JNJ26481585, SB939, and PCI24781 (as test compounds). Then immunostaining was performed with neuronal-specific β-tubulin III antibody (green) to visualize neuronal processes and quantify neurite lengths and Dapi (blue) to visualize cell bodies. (C) Statistical analysis of neurite length in various groups. As depicted, all test compounds were able to induce neurite outgrowth: PCI24781 (0.1 µM) had the greatest effect with 37% increase compared with that of the 0.1% DMSO. It was followed by JNJ26481585 (0.001 µM) with 34% increase. Trichostatin A was used as a positive control. Quantitative analysis of neurite length was done using ImageJ software. Two fields of view were analyzed per well. Two biological replicates were used per treatment and two independent experiments were done. Error bars represent SEM; *** p < 0.001, **** p < 0.0001.

The Hydroxamate-Based HDACis Showed No Toxicity at the Concentrations Used
Except for JNJ-26481585, which was tested at a concentration of 0.001 µM, other compounds were tested at two different doses-0.01 and 0.1 µM-for cell toxicity. Neuronal cells viability was assessed with Cell-Titer Glo assay. HNPCs were seeded and differentiated to neurons according to the same exact method used for neurite outgrowth assay. The viability did not significantly change relative to the control (0.1% DMSO) in all test compounds ( Figure 6). As depicted, all test compounds were able to induce neurite outgrowth: PCI24781 (0.1 µM) had the greatest effect with 37% increase compared with that of the 0.1% DMSO. It was followed by JNJ26481585 (0.001 µM) with 34% increase. Trichostatin A was used as a positive control. Quantitative analysis of neurite length was done using ImageJ software. Two fields of view were analyzed per well. Two biological replicates were used per treatment and two independent experiments were done. Error bars represent SEM; *** p < 0.001, **** p < 0.0001.

The Hydroxamate-Based HDACis Showed No Toxicity at the Concentrations Used
Except for JNJ-26481585, which was tested at a concentration of 0.001 µM, other compounds were tested at two different doses-0.01 and 0.1 µM-for cell toxicity. Neuronal cells viability was assessed with Cell-Titer Glo assay. HNPCs were seeded and differentiated to neurons according to the same exact method used for neurite outgrowth assay. The viability did not significantly change relative to the control (0.1% DMSO) in all test compounds ( Figure 6).

Discussion
Due to the improved physiological relevance and translational potential, the use of primary human cells as experimental cell model offers significant value in drug discovery assays. While these cells provide an ideal system for screening of compounds, their maintenance and scaling for HTS is challenging [29][30][31][32]. Herein, we provided evidence showing that HNPCs could successfully be adapted and scaled to HTS in 384-well plates. HNPCs are comprised of neurospheres, floating cellular aggregates containing neural progenitor cells (NPCs) and radial glial cells (RGCs) isolated from embryonic mammalian brain that have been cultured in vitro in the presence of mitogens. These cells maintain their differentiation capability in creating different neural lineages providing a desirable in vitro model system to study the CNS function and development [33][34][35]. In addition to the differentiation capability of neurospheres, their undifferentiated state is also previously confirmed by detecting the expression of the typical progenitor cell markers containing BLBP, NESTIN, VIM, and GFAP [26].
In this experiment we showed that several hb-HDACi compounds including SB-939, PCI-24781 and JNJ-26481585 were powerful inducers of BDNF expression in mRNA and protein level and were also able to induce the neurite outgrowth in human neural progenitor cells-derived neurons, demonstrating their neurogenic potential.
In addition to HDAC inhibitors, several other distinct molecular categories were also shown to be potent activators of BDNF gene expression in both HNPCs and fibroblasts in our primary screening experiments (Table 1). MM-102 and 5-Aza-2 -deoxycytidine (Decitabine) are methyltransferase inhibitors acting on histone and DNA, respectively. It is demonstrated that BDNF down-regulation in brain samples taken from patients with schizophrenia and bipolar affective disorder is associated with DNA methyltransferase1 (DNMT1) overexpression [36]. Therefore, using 5-Aza-2 -deoxycytidine (Decitabine) as a DNMT1 inhibitor could be a promising strategy to tackle neurodegenerative disorders. MM-102 is also a small molecule inhibitor of WDR5/MLL1 protein-protein interaction that specifically inhibits cell growth and induces apoptosis in leukemia cells harboring MLL1 fusion proteins [37]. PARP-1 inhibition has been reported to play a neuroprotective role in neurodegeneration induced by acute brain ischemia [38]. Its association with the pathogenesis of several nervous system disorders has also been demonstrated [39][40][41]. Due to the inhibitory effects of nicotinamide and BYK 204165 on PARP-1, these compounds may thus be able to demonstrate promising effects on the models of neurodegenerative disorders. CPTH2 inhibits the HAT activity of Gcn5. Histone acetyl transferase function usually leads to transcriptional activation, but in this study it showed an activation effect on the expression of BDNF, which once more draws attention to the complicated regulatory mechanisms of BDNF [42]. 2,4-pyridinedicarboxylic acid (2,4-PDCA), as a histone demethylase inhibitor, has been shown to be able to inhibit several Jumonji domain-containing lysine demethylases when used at low micromolar concentrations [43][44][45]. It should be noted that there is no documented information about the effect of this compound on BDNF. Therefore, this compound can be considered as a putative BDNF enhancer to investigate its positive effects in neurodegenerative disorders. Resveratrol is a natural nonflavonoid polyphenol found in grapes and red wine. It has anti-neuroinflammatory properties and is known to be neuroprotective in neurodegenerative disorders [46,47]. The enhancing effect of resveratrol on BDNF gene expression has been validated by in vitro and in vivo experiments elsewhere [48][49][50][51].
BDNF has a complex structure, containing nine exons (I-IX) in both human and rodents with each exon harboring its own promoter, resulting in more than 10 different alternatively spliced transcripts [52][53][54][55]. Temporal and spatial regulation of BDNF gene expression occurs by using distinct BDNF mRNA spliced variants and is also dependent on its various promoters, which could finally lead to the modulation of synaptic plasticity and spine development in dendrites [56][57][58]. BDNF gene expression is also controlled at the post-transcriptional level [13].
We demonstrated here that BDNF gene was reactive to a diverse array of epigenetic factors, most notably HDAC inhibitors, more specifically hydroxamic-based HDAC inhibitors. We believe that the information obtained from our screening could pave the way for identification of novel treatments for neurodegenerative disorders. In addition, our findings shed light on the complex regulatory mechanisms of BDNF gene and reanalysis of the data would likely lead to identification of new pathways that are important to the regulation of this gene.
Neurite outgrowth plays a key role in the regeneration of the nervous system following injury. It is also an integral element of extracellular signaling that can induce neuronal regenerative activities to enhance the outcomes of cases with neurodegenerative disorders and neuronal injury [59][60][61][62]. In this experiment, we investigated the effect of hb-HDACis on the neurite outgrowth. The rationale behind this analysis was that since BDNF is an important factor capable of inducing neurite outgrowth, the compounds which we proved to be BDNF activators may likewise affect the neurite outgrowth. In this investigation, hb-HDACis was demonstrated to promote neurite outgrowth even more than that induced by TSA, a documented positive control for this specific experiment. Neurite outgrowth might be the direct result of BDNF production; however, this hypothesis cannot be proved unless TrkB signaling pathway is separately blocked when the identified HDAC inhibitors are applied.

High-Throughput Screening Assay Using an In-House Epigenetics Compound Library
HNPCs and WS1 human skin fibroblast cells (ATCC CRL-1502) were seeded in 384-well plates (25K and 20K cells per well, respectively) and treated with an in-house epigenetic compound library at a concentration of 1 µM in duplicate. To maximize the amount of available RNA, TaqMan Cell-to-Ct assay kit (Invitrogen AM1729) was used to extract the RNA and reverse transcription to make the resulting cDNA per the manufacturer's instruction. In short, after aspiration of the media, lysis solution (30 µL) with diluted DNase I was added; then the plate was shaken for 2 min and incubated for 3 min at room temperature. Subsequently Stop solution (3 µL) was added and the solution was incubated at room temperature for 2 min. After spinning, 4 µL of lysates was transferred to a new PCR plate with 16 µL of reverse transcription enzyme mix previously added to each well. The thermal cycling condition was as follows: 60 min at 37 • C, and 5 min at 95 • C. A 3-µL aliquot of each cDNA reaction was then added to 13 µL of each TaqMan master mix reaction along with TaqMan BDNF probe (Hs02718934_s1 Invitrogen 444889). A QuantStudio 6 Flex Real-Time PCR system (Applied Biosystems) was utilized to determine the Ct values. Relative mRNA expression levels were normalized to β-actin (Invitrogen 4326315E) and analyzed using the comparative delta-delta CT method.

Differentiation and Treatment of HNPCs
To induce differentiation, cultured neurospheres were disaggregated into single cells using StemPro Accutase cell dissociation reagent (Gibco A1110501), counted with Countess Automated Cell Counter (Invitrogen, Carlsbad, CA, USA C10227), then 80K NPCs were plated in each well of 4-well glass chamber slides (Millipore Sigma, Burlington, MA, USA PEZGS0816) previously coated with poly-L-lysine (PLL) (Sigma P5899-5MG) for 1 hour at room temperature and laminin (Sigma

Image Acquisition and Neurite Outgrowth Quantification
Following staining, the images were captured with a LSM 710 confocal microscope equipped with ZEN 2010 B SP1 software (Carl Zeiss, Oberkochen, Germany), using the 20× objective with an image size of 1024 × 1024 pixels and 424.7 × 424.7 µM. At least two fields per well containing stained differentiated neuronal cells were imaged.
Fiji image analysis software (ImageJ 1.51u) was used for quantification. Briefly, regions of interest (ROIs) and the length of the longest neurite from each neuron were selected and then 'Measure' function was used to quantify the length of the selected ROIs. Subsequently, the values per treatment were averaged. Error bars indicate SEM of neurite length calculated from four biological replicates per treatment (at least two images per replicate). Student's t test for independent groups was used to compare means between two groups. A two-tailed p value < 0.05 was considered statistically significant.

RNA Extraction, cDNA Synthesis, and Real-Time PCR
Total RNA was extracted from HEK293 cells using the RNeasy Mini Kit (Qiagen, Hilden, Germany 74106) with on-column performed deoxyribonuclease (DNase) treatment using a Ribonuclease-Free DNase Kit (Qiagen), according to the manufacturer's protocol. The quality of RNAs was checked using NanoDrop 2000 spectrophotometer (Thermo Scientific). RNA (1 µg) was then reverse-transcribed using TaqMan reverse transcription reagents (Invitrogen, Cat No. N8080234). Gene expression was then measured by real-time PCR (RT-PCR) using human BDNF (Invitrogen, cat No. 444889, assay ID Hs02718934_s1) and human ACTB (β-actin) endogenous control (Invitrogen, cat No. 4326315E) TaqMan primer-probe assays and TaqMan gene expression master mix (Invitrogen 4369016). Samples were amplified for 40 cycles using the Applied Biosystem Quantstudio Flex 6 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The relative gene expression presented was calculated based on fold change using the delta C t method. Data were analyzed using two-tailed Student's t test with GraphPad Prism software (San Diego, CA, USA).

Cell Culture Drug Treatment
HEK293 cells, were used to test the effect of the compounds in vitro. Concisely, cells were plated overnight in DMEM supplemented with 10% fetal bovine serum in 24-well plates (200,000 cells per well) in a 37 • C incubator, 5% CO 2 , treated the next day with DMSO (0.1%) or selected compounds (1 µM) for 48 h.

Cell Viability Assay
Cytotoxicity of selected compounds was evaluated using the CellTiter-Glo luminescent cell viability assay (Promega G7570) as per the manufacturer's protocol. For neuronal cells, the method used was the same as described in "Differentiation and treatment of HNPCs" section. Subsequently, ATP level was measured after treatment utilizing EnVision 2104 Multilabel Reader (Perkin Elmer, Waltham, MA, USA). The luminescent signal is proportional to the cellular ATP concentration which itself is directly proportional to the cell number. Error bars represent SD. One-way ANOVA was used for data analysis. A p value < 0.05 was considered statistically significant.

ELISA Experiments
HEK293 cell culture supernatants were collected 48 h after treatment with compounds. Levels of secreted BDNF were measured using BDNF Emax ImmunoAssay System per manufacturers' instructions (Promega G7611). Samples were measured in triplicate, and the mean value was used for analysis. Error bars represent SEM of BDNF protein. Statistical significance was calculated using two-tailed Student's t test for unpaired data. A p value < 0.05 was considered statistically significant.