This study presents an optimized RA-mediated neuronal differentiation protocol for the SH-SY5Y cell line and demonstrates the behavior of the AD central gene, APP, during the transition from the undifferentiated to the terminal differentiated state. SH-SY5Y cells produce a nearly homogeneous neuron-like cell population upon RA-induced differentiation. In contrast, other competitive human neuronal cell lines, such as NT-2, require complex differentiation procedures and yield a heterogeneous population of neural lineage cells[23].
Undifferentiated neural cell lines are frequently used in neuroscience research in various experiments. Although they have certain merits, including ease of culture and reproducibility, the validity of experiments conducted using undifferentiated neuronal cell lines is often disputed as these cell lines fail to reflect the complexity of mature neurons and their stimuli-response mechanisms[2]. Neurons exist in a range of differentiation states, and neuronal progenitors pass through a series of stages during the differentiation process. In neurons, the identification of the exact state of differentiation allied with its morphology is crucial because morphology mostly reflects the physiology, which shapes their functional and structural properties.
The time required to reach the final differentiation state varies greatly among the different procedures describing the differentiation of SH-SY5Y cells. The time taken to reach a specific differentiation state potentially varies slightly depending on the study and procedure. However, multiple studies describing the differentiation of SH-SY5Y cells have demonstrated morphologically inconsistent final differentiation states (Table 1). We closely monitored neuronal morphological alterations on consecutive days and photographed and reserved samples for further analysis. Therefore, the differentiation stages described in this experiment can be compared with those described in other studies.
There is no properly established morphological boundary or benchmark for the terminal differentiation of SH-SY5Y cells. Multiple studies have reported considerably varying definitions of their boundaries (Table 1). Most studies have relied on the expression of certain neural markers to indicate differentiation [10],[7]. Therefore, identifying a broadly accepted standard for determining the exact state of differentiation is difficult. A standardized differentiation state for SH-SY5Y cells would considerably improve the reproducibility and reliability of the experiments. In the present study, we carefully defined the "true maturity" state of SH-SY5Y cells, in conjunction with the corresponding terminal morphological and differentiation states.
The oxidative activity of cells is an alternative indicator for their viability and metabolism[24]. Similar levels of oxidative stress were observed at the end of the experiment in both undifferentiated and differentiated cells; this suggests that they had similar levels of metabolic activity throughout the study, despite exposure to long-term culture and differentiation conditions. This finding is significant as it enables researchers to more precisely define the differentiation state, which in fact improves the accuracy and reliability of future experiments. Processing of APP yields complex protein processing outcomes, complicating the accurate comparison of protein expression levels. In the present study, we relied on qPCR, which measures gene expression accurately and independently from the protein processing barriers. Parallel to morphological observations, we studied APP gene expression at each stage of differentiation.
In vitro and in vivo expression of APP gradually increases during neuronal maturation[25],[26]. To determine the total APP gene expression in each differentiation state, we targeted both the coding region and 3'UTR separately using three primer sets: APPED, APP3'UTR and APP770. Expression values of both APPED and APP3'UTR represent the total APP mRNA expression folds in the cells while APP770 represents only the transcript variant group − 1 which includes the major variant, the transcript variant-1 (Table 2). During differentiation, the expression patterns of APPED and APP3'UTR closely followed each other and resembled earlier findings, showing that the APP level gradually rose towards maturity. Another feature of APP is isoform diversity among different types of cells. Among the main isoforms, APP695 mRNA is highly expressed in the hippocampus, cerebral cortex and amygdala[21], which are the critically affected areas of the brain in AD.
In contrast APP770 is expressed systemically and declines with the neural or brain maturity[20]. In our initial assessment, we compared the expression levels of total APP mRNA and APP770. Predictably, we observed a steady decline in the expression of isoform-1 relative to total APP from differentiation day 16. Conversely, the total APP expression, largely contributed by APP695 has increased. APP contains a Kunitz protease inhibitor (KPI) domain[27],[28] and an orexin 2 (OX2) receptor extracellular domain (Ox-2). The Kunitz protease domain can inhibit a wide range of serine proteases[29], while Ox-2 plays a role in ligand binding and receptor activation[30]. The functional, structural, and spatiotemporal diversity among the three main isoforms, APP700, APP751, and APP695, is mainly attributed to the presence or absence of their KPI domain[31],[32],[33].
To determine the transcript variant expression dynamics, we categorized the APP transcript variants into four groups based on the status (presence or absence) of the KPI and the subsequent Ox-2 domains that reside in exon 7 and 8 (Table 2). In this grouping, the major transcript variants (transcript variants 1, 2, and 3) were confined to groups 1, 2, and 3. Group 4 had a single entity, transcript variant 11. We used four sets of transcript variant-specific exon junction-targeted primers (Table S3). These primers were used for PCR and qPCR to amplify each transcript variant group.
The expression of APP increases during neuron differentiation[34]. In line with previous reports, we observed an increase in total APP expression during the differentiation series and in the 24-day differentiated SH-SY5Y cells. APP transcript variant 1 (APP770) is the dominant form of APP in non-neuronal cells and neural progenitors, and its expression is often downregulated in mature neurons[20],[34]. We noted a declining trend in APP770 expression during the later phases of the differentiation series and in 24-day differentiated SH-SY5Y cells; this proves that the 24-day differentiated cells were well-differentiated into their mature states. Conversely, APP695, the main isoform that lacks the KPI domain and is found mostly in the brain and mature nerve cells, is upregulated during neural maturation[21],[22]. It could also serve as an effective marker-candidate for neuronal maturation. We observed a steady increase in APP695 expression during differentiation. Accordingly, the expression of the APP695 was high in the 24-day differentiated SH-SY5Y cells; this confirms that the 24-day differentiated cells were well-differentiated into their mature state. APP transcript variant 11 (APP714) was also highly upregulated in the differentiated state. Apart from the group 3 entities, APP transcript variant 11 is the only APP transcript variant that does not bear the KPI domain. Based on these outcomes, we propose the expression dynamics of APP as a precise indicator of neuronal maturity.
We confirmed the differentiation of SH-SY5Y cells by analyzing the gene expression profiles of the neuron-specific marker genes ENO2, MAP2, SYP, and MAPT. ENO2, also known as neuro-specific enolase (NSE), is involved in neuronal glycolysis and is highly expressed in mature neurons[20]. he neuron cytoskeletal structural protein MAP2, mainly expressed in dendrites, is used as a common marker for dendritic development and thus represents neuronal maturity [20]. SYP is a synaptic vesicle glycoprotein mainly found in presynaptic terminals of the neuroendocrine system and is highly expressed in mature neurons[20]. MAPT is a microtubule-associated structural protein abundant in the axons of mature neurons. It is involved in axonal transport, and its dysfunction is associated with several neurodegenerative diseases, including AD[20]. All differentiation markers were significantly upregulated in differentiated SH-SY5Y cells compared to those in undifferentiated cells. The upregulation of the above-mentioned mature neuron marker genes in differentiated SH-SY5Y cells indicated that these cells gained mature neuronal properties within our differentiation timeframe.
We then analyzed several genes involved in the maintenance of pluripotency and early neurogenesis. Undifferentiated cells were characterized by the expression of SOX2, ASCL1, and NEUROD1 genes. SOX2 is an essential transcription factor involved in maintaining the pluripotency of undifferentiated stem-like cells, including embryonic and induced pluripotent stem cells[35]. SOX2 is expressed in adult neural stem cells and other lower hierarchy neural progenitors and downregulated in post-mitotic neurons[36]. ASCL1 is a marker for the transient differentiation of neural progenitor cells, and its expression is downregulated in differentiated neurons[37]. NEUROD1 is an early embryonic-expressed gene involved in neural differentiation[38]. As maturity progresses, the expression of NEUROD1 is confined to specific regions and certain types of neurons, such as the pyramidal cells in the brain[39], thus downregulating its expression in most types of neurons, including in the SH-SY5Y cells[16]. Accordingly, the immature neuronal marker genes (SOX2, ASCL1, and NEUROD1) were downregulated in the differentiated state, indicating a shift from a proliferative stem-like state to a mature phenotype. In addition to that, we also focused on the expression of neurotransmitter markers during SH-SY5Y cell differentiation. Our results showed that the cholinergic neuronal marker genes, ACHE and VACHT, were highly upregulated in differentiated cells. Multiple studies have reported that RA-induced SH-SY5Y cell differentiation directs dopaminergic phenotypes[2],[3]. Conversely, we observed highly upregulated cholinergic markers in the differentiated cells, consistent with the recent reports of cholinergic differentiation of SH-SY5Y cells[11]. The upregulation of cholinergic markers suggested that differentiated SH-SY5Y cells acquired cholinergic neuronal phenotypes. Notably, a non-cholinergic marker, GABBR1, was highly upregulated during differentiation; this can be explained by evidence showing GABAB receptor expression in certain cholinergic neurons, particularly those in the habenula region[40]. We observed a significant downregulation of FOXA2, a dopaminergic neuron maturation marker[41],[42], suggesting that the differentiated SH-SY5Y cells may not exhibit dopaminergic phenotypes. This confirms our hypothesis on the cholinergic phenotype of mature SH-SY5Y cells.
Finally, we found that PET1 and GLUL expression remained unchanged during differentiation. The expression of PET1 is associated with serotonergic neurons[43] while that of GLUL is associated with glutamatergic neurons and glial cells, mainly astrocytes[44]. Thus, the differentiated SH-SY5Y cells did not exhibit serotonergic, glutamatergic, or glial phenotypes. The technique we used to differentiate SH-SY5Y cells has successfully directed the cells toward the cholinergic neuron phenotype. Besides, cholinergic neurodegeneration has been widely implicated in AD[45] and is one of the earliest pathological events [46], making differentiated SH-SY5Y cells an excellent model for AD studies.
Transcriptomic analysis of AD brains has revealed differentially expressed genes triggered by AD pathology in neurons and in proliferative and mature SH-SY5Y-derived neurons[47],[48]. To demonstrate how AD-responsive genes are involved in neural differentiation, we analyzed several genes, including BEX1, BEX3, STMN1, MTRNR2L8, and PSEN1. Importantly, the expression of these genes was previously compared solely between diseased and healthy states in mature neurons[47]. However, to the best of our knowledge, no studies have compared the gene expression profiles of proliferating neuronal progenitors and mature neurons. In the present study, we compared the gene expression profiles of undifferentiated SH-SY5Y cells (proliferating state) and differentiated SH-SY5Y cells (mature state) to distinguish the differentiated phenotype from the undifferentiated form. All AD-responsive genes were highly upregulated in differentiated SH-SY5Y cells compared to those in undifferentiated cells, suggesting that the genes mentioned above, which play a vital role in mature neurons or the brain under AD conditions, may be differentially expressed in mature neurons and neuronal progenitors.
Our study emphasizes the importance of a differentiation state-sensitive marker system for efficient and precise characterization of SH-SY5Y differentiation. We found that the total APP and its transcripts are expressed differently in SH-SY5Y-derived neurons, and their fluctuation depends on the corresponding differentiation state. Consequently, we proposed a novel APP splice variant-dependent marker system that enhances the output accuracy of conventional markers. The observed deviation of the canonical APP splice variant pattern under overexpression conditions could be an artifact arising from an abundant epithelial phenotype. Furthermore, we propose an efficient and simplified differentiation method for 2D culturing of SH-SY5Y cells. Neurons labeled with fluorescent proteins have enabled direct visualization in the co-culture context, offering researchers a tool for real-time observation of neuronal morphology and interactions. The outcomes of our study indicate that the differentiated SH-SY5Y cell line is an ideal neuronal model, particularly for the study of APP gene expression in AD.
To the best of our knowledge, this is among the most comprehensive studies on SH-SY5Y differentiation with respect to the expression of the AD central gene, APP. The study findings provide valuable insights into SH-SY5Y differentiation and present new tools for improving the accuracy of future neuronal differentiation experiments.