Flexible and Accurate Substrate Processing with Distinct Presenilin/γ-Secretases in Human Cortical Neurons

Expression of genes encoding integral γ-secretase subunits during cortical neuronal differentiation of a healthy individual-derived iPSC line, 201B7. A, Strategy of cortical neuron differentiation from iPSCs. Representative images of iPSC colonies (left) and iPSC-derived cortical neurons (right) are also shown. Scale bars: 100 μm (iPSC) and 20 μm (iPSC-neurons). B, Representative images of iPSC neurons stained with antibodies specific for NeuN, TBR1, and βIII-tubulin are shown. Scale bar: 50 μm. Numbers of NeuN+ or TBR1+ cells among Hoechst 33342-stained cells were calculated. Data represent the mean ± SEM (n = 3 of independent culture batches). C, qRT-PCR analysis of iPSCs (day 0), NPCs (NP; day 14), and iPSC-derived cortical neurons (neuron; day 45). Several genes encoding γ-secretase subunits, such as PS1, PS2, APH-1A, and PEN-2, were upregulated in iPSC neurons compared with iPSCs and NPCs. BACE1 and APP were also upregulated in iPSC neurons. Data represent the mean ± SEM (n = 3–4 of independent culture batches). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001 by Dunnett’s test versus iPSCs.

Generation of PS cKO iPSCs. A, Targeting strategy for PS1 cKO. In the targeting vector, exons 2/3 are flanked by a loxP site (red arrowhead) and a PGK-neo (or PGK-puro) selection cassette followed by a loxP. The neo selection cassette is flanked by two FRT (FLP recognition target) sites (black circles), so that the cassette can be removed by FLP recombinase. Green lines were the genomic sequences of homologous regions. iPSC clones were screened by genomic PCR using the primer pairs (arrows) to discriminate iPSC cells carrying the proper recombination from untargeted ones. The homozygous targeted clones were transfected with a plasmid expressing Flp gene to generate homozygous floxed PS1 iPSC. In the presence of Cre, the floxed allele will be recombined to produce the deleted PS1 allele. B, Genomic PCR for respective PS1 alleles. Genomic PCR was performed using primer pair encompassing two loxP sites, following transfection with plasmid expressing Flp or Cre recombinase. Parental iPSC line (201B7) and transfectant with empty vector (nc) were also shown. C, PS1 protein expression in iPSC clone carrying respective PS1 allele. Quantitative analysis shows nearly eliminated levels of PS1 proteins in total lysates isolated from the iPSC clone carrying deleted allele after Cre transduction, compared with other iPSC clones, using rabbit polyclonal antibodies specific for PS1-NTF and PS1-CTF. Data represent the mean ± SEM (n = 3 of independent culture batches). ns, not significant; ***p < 0.005 by Tukey’s test among the genotypes. D, KO strategy for PS2 gene. sgRNAs were transfected along with Cas9 proteins, to delete the genomic region around the exon 5. E, Sequence of both alleles around the PS2 targeted region from clone #12 (nearly negative for PS2 protein was confirmed by PS2 Western blot analysis of single clones transfected with PS2 sgRNA and Cas9 proteins as shown in Extended Data Fig. 2-1). One allele was completely deleted (lower), while another allele was inverted between two sgRNA sites (upper).

Maintenance deficit in iPSC-derived NPCs lacking PS1 and PS2. A, qRT-PCR analysis of iPSCs and NPCs with the genotypes fPS1/fPS1;PS2^+/+ (WT), PS1^−/−;PS2^+/+ (ΔPS1), and fPS1/fPS1;PS2^−/− (ΔPS2). Quantitative analysis shows the almost complete elimination of PS1 and PS2 in the ΔPS1-derived and ΔPS2-derived cells, respectively. Specific genes encoding components of the Notch signaling pathway (NOTCH1, NOTCH2, HES1, and HES5) were robustly increased in NPCs throughout neural patterning. The levels of both HES1 and HES5 were slightly but not significantly decreased in ΔPS1-NPCs compared with those in WT NPCs, whereas the levels of NOTCH1 and NOTCH2 were comparable among the genotypes. Data represent the mean ± SEM (n = 3–5 of independent culture batches). ns, not significant; ***p < 0.001 by Tukey’s test among the three genotypes in the NPC group. B, Experimental scheme of the colony-forming assay using fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− iPSCs by flow cytometry. C, Quantitative analysis of the first neurospheres showing mCherry fluorescence. The fluorescence percentage was measured by flow cytometry after “partial” infection with lentivirus expressing mCherry-ΔCre or mCherry-Cre in fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− neurospheres (NS; flow cytometry plots were shown in Extended Data Fig. 3-1). Data represent the mean ± SD (n = 3 assays). ns, not significant by Tukey’s test among the genotypes. D, Quantitative analysis of the second neurospheres showing mCherry fluorescence. The fluorescence percentage was measured by flow cytometry and then calculated as the % reduction in mCherry⁺ cells from the first NS to the second NS (flow cytometry plots were shown in Extended Data Fig. 3-1). Data represent the mean ± SD (n = 3 assays); ***p < 0.001 by Tukey’s test among the genotypes.

PS1/PS2 expression in genome-edited iPSCs harboring floxed PS1 alleles following Cre expression. A, Representative images of iPSC neurons (day 45) stained with antibodies specific for TBR1 and MAP2 are shown. Nuclear-localized GFP signals indicate iPSC neurons successfully infected with lentiviruses expressing EGFP-ΔCre or EGFP-Cre. Scale bar: 50 μm. B, Western blot analysis of PS1-NTF, PS2-CTF, APP, and β-actin throughout terminal neuronal differentiation. Lentiviruses were infected at day 20. Because of posttranslational modification such as glycosylation, full-length of APP proteins exhibits two bands ∼100 kDa in size. C, Western blot analysis of PS1-NTF, PS2-CTF, and β-actin. Representative blots are shown for each protein in fPS1/fPS1 iPSC neurons infected with ΔCre or Cre lentivirus. Data represent the mean ± SEM (n = 3 of independent culture batches); **p < 0.01, ***p < 0.001 by Student’s t test between the genotypes. D, Quantification of MAP2+ neurite length at 45 d in fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− iPSC neurons with infected with ΔCre or Cre lentivirus showed no alteration among the genotypes, despite a slight shorter tendency in PS1-deficient neurons. Data represent the mean ± SD (n = 5 of independent culture batches). ns, not significant by Dunnett’s test versus the control. E, qRT-PCR analysis at 45 d in fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− iPSC neurons infected with ΔCre or Cre lentivirus showed the almost complete elimination of PS1 and PS2 in Cre-infected neurons and PS2^−/− neurons, respectively, whereas the levels of neuronal markers such as MAP2, synaptophysin, and PSD95 were comparable among the genotypic groups. Data represent the mean ± SEM (n = 3–4 of independent culture batches). ns, not significant; ***p < 0.001 by Dunnett’s test versus the control.

Cleavage of APP and N-cadherin in the iPSC neurons of fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− infected with ΔCre or Cre lentivirus. A, Specific ELISA measurement of Aβ40 and Aβ42 in the iPSC neurons of fPS1/fPS1;PS2^+/+ and fPS1/fPS1;PS2^−/− infected with ΔCre or Cre lentivirus. Quantitative analysis revealed a reduction in the Aβ levels in the iPSC neurons of fPS1/fPS1;PS2^−/− infected with Cre lentivirus. Data represent the mean ± SEM (n = 4–6 of independent culture batches); *p < 0.05, ***p < 0.001 by Dunnett’s test versus the control. No change of Aβ generation in the absence of PS1 alone was corroborated with the iPSC-neurons derived from another fPS1/fPS1;PS2^+/+ clone (#249–3; Extended Data Fig. 5-1A). DAPT treatment nearly abolished the generation of Aβ40 and Aβ42 (Extended Data Fig. 5-1C). B, Levels of APP-CTFs were quantified by Western blotting. Quantification analysis shows a massive increase in APP-CTFs in the iPSC neurons of fPS1/fPS1;PS2^−/− infected with Cre lentivirus. No accumulation of APP-CTFs in human neurons devoid of either PS1 or PS2 alone was further corroborated with the iPSC neurons from WT, ΔPS1, and ΔPS2 (Extended Data Fig. 5-1D). Data represent the mean ± SEM (n = 4–5 of independent culture batches). ns, not significant; ***p < 0.001 by Dunnett’s test versus the control. C, Levels of the APP-CTFs were quantified by immunocytochemistry. Quantification analysis showed a massive increase of the APP-CTFs in the iPSC-neurons of fPS1/fPS1;PS2^−/− infected with Cre lentivirus. Data represent the mean ± SEM (n = 4–6 of independent culture batches). ns, not significant; **p < 0.01 by Dunnett’s test versus the control. D, Levels of N-cadherin-CTF1s were quantified by Western blotting. In contrast to APP cleavage, N-cadherin was cleaved exclusively by PS1/γ-secretase. Data represent the mean ± SEM (n = 4–5 of independent culture batches). ns, not significant; *p < 0.05, ***p < 0.001 by Dunnett’s test versus the control. E, Levels of phosphorylated (AT8) and total tau (Dako) were quantified by Western blotting. Most tau proteins in iPSC-derived neurons were phosphorylated (black arrowhead) rather than unphosphorylated form (gray arrowhead) using total tau antibody, similar to embryonic brains. The state of tau phosphorylation was not altered by the absence of PS1 and/or PS2. Data represent the mean ± SEM (n = 3–4 of independent culture batches). ns, not significant by Dunnett’s test versus the control.

Comparable levels of Aβ secretion in mouse embryonic fibroblast cultures expressing human or mouse PS2 heterologously. A, Schematic diagram of lentivirus expressing human PS1, human PS2 or mouse PS2, along with Venus. Representative pictures of immunocytochemistry in lentivirus-infected cultures. B, Expression of PS1 and PS2 was analyzed in cell lysates of DKONL cultures expressing the construct by Western blotting. nc, DMSO (0.01%) treatment. C, ELISA measurement specific for Aβ40 and Aβ42 in the DKONL cultures expressing hPS1, hPS2, or mPS2. The levels of Aβ were normalized with protein concentration and Venus protein level of the cell lysate. Data represent the mean ± SEM (n = 4 of independent culture batches). Calculated Aβ42/Aβ40 ratio was also drawn from the data measured by specific ELISA. ns, not significant; ***p < 0.005 by Tukey’s test among the genotypes.