AD is the most common cause of dementia in the elderly, with progressive neuronal loss in the cerebral cortex and hippocampal formation. Although the underlying etiology of most AD remains unclear, Aβ is thought to play a pivotal role in its pathogenesis. Studies from animal and cellular models have shown that mutations in the APP, PS1, and PS2 genes affected the production of Aβ, which contributes to the formation of amyloid plaques 
. In several strains of mouse models, Aβ levels in brain tissue, cerebrospinal fluid (CSF), and plasma have been associated with AD pathogenesis and cognitive impairment 
. Human samples from clinical AD patients have also been used for pathological and biochemical analyses to understand the etiology of AD. Aβ levels in CSF and plasma have been examined to evaluate their risks for AD 
, but brain tissues are only available postmortem for such analyses. On the other hand, immortalized human cell lines derived from kidney or brain, primary neurons derived from mice and rats, or cells artificially overexpressing APP or presenilin with or without familial AD mutations have been utilized for in vitro
. There is no doubt that these cells are quite different from living neurons in the human body in terms of innate qualities. Although we have had no choice until recently, important advances in technology of iPS cells may now provide the opportunity to use intact human-derived neuronal cells 
We evaluated whether iPS cell-derived neuronal cells could be applied to an in vitro cell-based assay system for AD research. In particular, further investigations into the metabolic mechanisms of Aβ are requisite for drug development to treat the brains of patients afflicted with AD. In this respect, we provide a profile of the molecular components associated with Aβ production in hiPS cell-derived neuronal cells and propose to add an Aβ assay system using these cells to the panel of generalized Aβ-monitoring systems (). Human neuronal cells are considered to provide more accurate human neuronal conditions within which to evaluate drug efficacy or toxicity than other human cell lines (e.g., cancer lines). Furthermore, we would be able to investigate how hiPS cell-derived neuronal cells reflect AD-related physiological and pathological conditions based on Aβ production.
Panel of Aβ monitoring systems.
In the present study, we characterized iPS cell-derived neuronal cells in terms of their expression of neuronal and glial markers by exposing them to Noggin and SB431542 during their differentiation ( and ). We observed increases in GFAP mRNA levels and in synapsin I-positive synaptic puncta at day 52. This was consistent with data showing that the existence of astrocytes promotes synaptic activity in human ES cell-derived neurons 
. When differentiation occurred in the presence of non-morphogens, we obtained mainly glutamatergic neurons (), quite in line with previous reports of concerning hES and hiPS cells 
. Expression of the forebrain marker Foxg1 suggests a default forebrain identity of the 253G4 iPS cells used in this study (). We also observed the expression of the neocortex-specific transcriptional factors Tbr1, Ctip2, Cux1, and Satb2 (). These expression schemes appear to mimic human neocortical development in vitro 
, although further analyses are needed to assist in understanding human neuronal subtype-specific differentiation.
This is the first study to observe the expression of APP, β- and γ-secretase, and the production of Aβ in hiPS cell-derived neuronal cells. APP, sAPPβ, APP-CTFβ and BACE1 protein levels were increased ( and ), but protein levels of γ-secretase components were not significantly different during the period from day 38 to 52 (). Aβ production in hiPS cell 253G4-derived neuronal cells increased with differentiation course (), however that in another hiPS cell 201B7 
- and in hES H9-derived neuronal cells did not increase (Figures S5
) although all cell lines showed development of synapse (Figure S4A
) as Aβ releasing site 
, indicating that besides synaptogenesis, subtle changes in localization and assembly of APP 
, BACE1, γ-secretase components would be critical for Aβ production.
The Aβ42/Aβ40 ratio unexpectedly showed a significant decrease from day 38 to 45 (). Serneels et al.
reported that the γ-secretase complex containing Aph-1B was active and involved in the generation of amyloidogenic Aβ42 
. Our data showed that the Aph-1B/Aph-1A ratio did not change significantly with cell differentiation (); therefore, the Aβ42/Aβ40 ratio may be influenced by other unknown factors interacting directly or indirectly with γ-secretase.
BSI, GSI, and the NSAID sulindac sulfide inhibited Aβ production in this human neuronal cell system (). The inhibitory effect on Aβ production by GSI showed a characteristic difference between days 38 (Aβ surge) and 52 (gradual Aβ rise) (). Aβ surge at day 38 was also observed in another hiPS cell (201B7)-derived neuronal cells (Figure S7
) as well as in hES cell line, H9-derived ones (Figure S5
). At day 38, GSI might promote neuronal differentiation with synaptogenesis via blocking Notch signaling 
rather than inhibition of Aβ production, leading to Aβ surge. Another possible explanation for Aβ surge is that change in conformation or components of the γ-secretase affects the sensitivity of γ-secretase to GSI (total Aβ, Aβ40, Aβ42, and Aβ42/Aβ40), although levels of mRNA and the ration for Aph-1A and Aph-1B do not change between days 38 and 52 () Thus, for precise Aβ monitoring in human stem cell-derived neuronal cells, it is necessary to use neuronal cells with a sufficient substrate level and synaptogenesis, because Aβ is released presynaptically, as mentioned above.
Some NSAIDs are known to preferentially lower Aβ42 
. Our data showed that sulindac sulfide was capable of inhibiting Aβ42 secretion at high concentrations (≥10−5
M) (), although a few NSAIDs do not show therapeutic effects for AD. Negative results might be due to low γ-secretase modulator potency 
. To discover novel effective drugs for modulating β- or γ-secretase activity, the in vitro
hiPS cell-derived neuronal cell assay system might be expected to yield such drugs.
Familial AD patient specific neuronal cells generated by direct conversion (induced neuron, iN) show higher Aβ42/Aβ40 ratio than those of unaffected individuals 
. Based on this report, hiPS/hES cell-derived neurons expressing mutant PS1, PS2, or APP may show higher Aβ42/Aβ40 ratio. Comparing to our results, the levels of Aβs in this assay (Aβ40; ~1.7 ng/ml at day 52) is higher than that using iN cells (Aβ40; ~0.1 ng/ml), although iN cells become functional neurons more quickly. The optimization of neuronal cell condition for comparison of the Aβ42/Aβ40 ratio between multiple iPS cell-derived neuronal cells may be required.
In conclusion, our findings indicate that hiPS cell-derived neuronal cells express functional β- and γ-secretases related to the production of Aβ in the present experimental conditions. In addition, our data provide the proof in principle that hiPS cell-derived neuronal cells can be applied to drug screening and AD patient-specific iPS cell research.