Introducing Cellular Senescence in Human Induced Pluripotent Stem Cells and Differentiated Neural Lineage for Modeling of Age-associated Diseases

Abstract

Reprogramming somatic cells into induced pluripotent stem cells (iPSCs) resets the epigenetic landscapes that mark the aging clock, and consequently cells differentiated from iPSCs resemble fetal cells rather than adult or aged cells. The lack of proper cellular aging in cells differentiated from iPSCs presents a unique challenge in iPSC-based modeling of age-associated diseases such as neurodegeneration. To address this challenge, we seek to introduce cellular senescence, a hallmark of aging, into iPSC-based models in a robust and temporally controlled manner. An inducible CRISPR interference (CRISPRi) is used to suppress the expression of TERF2, a key component of the telomere protecting Shelterin complex. We demonstrate that suppression of TERF2 robustly activates the DNA damage response, p53/p21 signaling and cellular senescence in iPSCs in a highly homogeneous and synchronous manner. Applying this inducible CRISPRi-TERF2 system to differentiation of iPSCs to neural progenitor cells (NPCs), we show efficient activation of senescence-associated phenotypes in NPCs. This inducible cell model allows isogenic comparisons of the same cell populations over the course of differentiation with or without the activation of cellular senescence in a synchronous and homogeneous manner, and has broad applications in investigating the role of cellular senescence in the progression of age-related diseases.

1. Introduction

The advent of induced pluripotent stem cell (iPSC) technology has revolutionized disease modeling in vitro. Disease-specific iPSCs can be generated from an array of patients to capture the genetic background underlying diseases and differentiated into disease-relevant cell types. [1, 2] With ongoing improvement of the in vitro systems, modeling of late-onset neurodegenerative diseases such as Alzheimer’s disease (AD), Parkinson’s disease and amyotrophic lateral sclerosis has provided important insight into the early events of disease pathogenesis. [3–9] However, conventional iPSC-based models are insufficient to recapitulate the late-stage pathologies for studies of disease progression. [10] During reprogramming, epigenetic landscapes including those associated with aging are reset in iPSCs. [11–14] Consequently, cells differentiated from iPSCs resemble fetal cells rather than adult or aged cells. [4, 15–20] The short culture period with iPSC-based models also restricts in vitro aging. The lack of proper cellular aging in iPSC-derived model systems presents a unique challenge in modeling age-associated diseases, as age is the largest risk factor of these diseases such as neurodegenerative diseases. [21–23] In contrast, direct reprogramming such as fibroblasts to neurons (iNeurons) bypasses the iPSC stage and largely retains age-related cellular and transcriptomic signatures of the donor cells, offering a clear advantage in the context of cellular aging over neurons differentiated from iPSCs. However, iNeurons are limited by the age of donors, their relatively poor reprogramming efficiency, low yield of reprogrammed cells and more importantly lack of versatility of iPSC-based differentiation into various cell types including distinct neuronal subtypes. [24–27] Further, 3D organoids can be generated from iPSCs to study the complex interactions among different cell types in disease pathogenesis. As aging is a complex biological process, numerous contributing causes present a significant challenge to model age-related changes for age-associated diseases. [28, 29] Introducing proper cellular aging into iPSC-based system is critical for successful modeling of age-associated diseases and investigating the impact of aging on disease progression. Various strategies have been applied to introduce different aspects of aging, including introducing Progerin or Werner proteins, treating with telomerase inhibitors or inducing senescence with chemical compounds, [15, 30–32] yet the lack of a robust method to introduce cellular aging demands further investigation.

Cellular senescence has been found to play a critical role in aging and is considered as one of the hallmarks of aging. [28, 33–35] Cellular senescence is a biological process where cells stop proliferation, suppress intrinsic cell death pathways, and release pro-inflammatory molecules. [36–38] It has been found that senescence can also occur in non-dividing cells. [39–41] Senescence occurs in response to various stress such as telomere dysfunction, persistent DNA damage, oxidative stress, aberrant oncogene activation, mitochondrial dysfunction, or epigenetic alterations. [35, 42] Recently cellular senescence has been implicated to play an important role in neurodegeneration. Senescent cells including neurons, astrocytes, microglia, oligodendrocyte progenitors, or endothelial cells are detected in the brains of AD patients [24, 43–53] and mouse models of AD that overexpress pathological human APP or tau protein. [44, 46, 54, 55] More importantly, selective elimination of senescent cells either genetically or pharmacologically reduces neurodegenerative pathologies β-amyloid or tauopathy and improves memory and cognition in the AD mouse models. [44, 46, 48, 54, 56] These exciting findings suggest that senescence plays a critical role in neurodegeneration.

In this study, we generated an inducible system to trigger telomere dysfunction and induce cellular senescence in human iPSCs and iPSC-derived neural progenitor cells (NPCs) as a way of introducing cellular aging for modeling of age-associated diseases. Telomere dysfunction is a major inducer of senescence and plays a critical role in aging. [29, 35, 57, 58] Suppression or deletion of telomere repeat factor 2 (TERF2), a major component of the telomere protecting Shelterin complex, [59] causes telomere de-protection without telomere shortening, activating persistent DNA damage response (DDR) at telomeres that consequently induces cellular senescence or apoptosis. [60–62] Using inducible CRISPR interference (CRISPRi), [63] we showed that suppression of TERF2 robustly activated senescence in iPSCs and NPCs differentiated from iPSCs in an inducible and synchronous manner, demonstrating a potentially broad application of inducible senescence for the study of age-associated neurodegenerative diseases.

2. Results

2.1. Establishing inducible CRISPRi in iPSCs to suppress TERF2 expression

We used an inducible CRISPRi construct [63] that expresses a catalytically inactive Cas9 (dCas9) tethered to a Krüpple-associated box (KRAB) repressor domain at the N-terminus under the tetracycline inducible TRE3G promoter (Figure S1A). In the same construct, reverse tetracycline-controlled transactivator (rtTA) is constitutively expressed under the CAG promoter. This CRISPRi construct was transfected into iPSCs for integration into the AAVS1 locus by TALEN-mediated targeting. [64] We isolated and expanded stable iPSC clones from two genetic backgrounds. One is from an individual of familial AD (fAD) with a disease onset at age 45 and an A246E mutation in PSEN1, [65] and the other is from a healthy individual (HC) without a history of dementia or neurodegeneration. [66] Colonies derived from each cell line were screened for successful integration in the AAVS1 locus by junction PCR (Figure S1B). The expression of KRAB-dCas9 was induced by doxycycline (Dox) treatment in CRISPRi-iPSCs, and 0.3 μg/mL of Dox was determined to be the optimal dose to achieve the highest expression level (Figure S2). We used this Dox concentration for all experiments thereafter. All established CRISPRi-iPSC lines expressed pluripotency markers Oct4, Nanog and Sox2 (Figure S3A) and were karyotypically stable (Figure S3B). The identity and purity of the isolated CRISPRi lines were confirmed by STR analysis (Table S5).

To suppress TERF2 expression, we designed four guide RNAs (gRNAs) targeting sequences downstream of the transcription start site (TSS) of TERF2 (Figure S4A) and cloned into a lentiviral expression vector. After lentiviral delivery of gRNA, populations of cells with TERF2 gRNA (CRISPRi-TERF2-iPSCs) were established after puromycin selection and treated with Dox for 2 or 4 days. While 2-day Dox treatment did not significantly change TERF2 level, suppression of TERF2 expression was clearly observed after 4-day Dox treatment (Figure S4B), with 70–92% reduction with gRNA-2, gRNA-3 and gRNA-4. Because gRNA-2 and gRNA-3 overlap in targeting sequence, we continued all experiments with gRNA-2 and gRNA-4.

2.2. Suppression of TERF2 triggered DNA damage response (DDR) and activated p53/p21 signaling in iPSCs

Telomeres without protection by the Shelterin complex are recognized as sites of DNA damage, which activates ATM mediated repair response leading to activation of the p53/p21 pathway. [67, 68] After 4 days of Dox treatment, we observed a significant reduction of TERF2 expression in CRISPRi-TERF2-iPSC lines at the transcript (Figure 1A) and protein levels (85.9–91.7% reduction; Figure1B, Figure S5A). Further, immunofluorescence (IF) staining showed KRAB-dCas9 was expressed in majority (93.5–94.1%) of Dox-treated cells, whereas TERF2 expression showed a significant (p<0.0001; 2.7 to 7.4-fold) reduction (Figure 1C, Figure S5B, C). In CRISPRi-TERF2-iPSCs treated with Dox, a significant increase in DDR markers γ-H2AX and phosphorylated ATM (p-ATM-S1981) foci was observed (Figure 1D, Figure S5D). Phosphorylated ATM activates checkpoint kinase CHK1 and CHK2, which in turn phosphorylates and activates p53. Phosphorylated p53 (p-p53) transactivates cyclin-dependent kinase inhibitors p21 (CDKN1A) which orchestrates the exit from cell cycle. [69] Consistently, we observed significantly elevated level of p-p53 at Serine 15 (Figure 1E) and increased expression of p21 (Figure 1A, Figure S5A) in CRISPRi-TERF2-iPSC after Dox treatment. In contrast, no alteration of p16^INK4a (CDKN2A) was observed (Figure S5E). Collectively, our inducible CRISPRi system robustly suppressed TERF2 expression and activated DDR and p53/p21 signaling.

Figure 1.

CRISPRi-mediated suppression of TERF2 activates DDR and the p53/p21 signaling in human fAD (AG25367) iPSCs (clone B6). (A) Quantitative RT-PCR showing deceased TERF2 mRNA by Dox in three biological replicates. (B) Western blotting and quantitation showing decreased TERF2 and increased p21 protein by Dox in three biological replicates. (C) IF staining and quantitation showing increased KRAB-dCas9 and decreased TERF2 by Dox (6 randomly selected fields each with 793 and 503 cells for −Dox and +Dox, respectively). (D) IF staining and quantitation showing increased DDR markers γ-H2AX and p-ATM by Dox (3 randomly selected fields with 882 cells for −Dox and 5 fields with 375 cells for +Dox). (E) IF staining and quantitation showing increased p-p53 by Dox (6 randomly selected fields with 710 cells for −Dox and 7 fields with 492 cells for +Dox). gRNA-2 was used in A, C-E. Data were presented as mean ± SD. Statistical significance determined using Welch’s t-test is denoted as **: p<0.01, ****: p<0.0001.

2.3. TERF2 suppression activated senescence in iPSCs

Suppression of TERF2 has been shown to activate senescence in human fibroblasts [60]. To determine whether CRISPRi-mediated suppression of TERF2 can induce senescence in iPSCs, we analyzed multiple cellular characteristics often associated with senescence, as there is no “one size fit all” senescence biomarkers. [70] We observed that Dox-treated CRISPRi-TERF2-iPSCs grew significantly slower than untreated cells and they stopped proliferating after 3 days of Dox treatment (Figure 2A, Figure S6A). Consistently, 5-ethynyl-2’-deoxyuridine (EdU) incorporation was significantly decreased after 4 or 6 days of Dox treatment (Figure 2B, Figure S6B). We noticed that Dox-treated CRISPRi-TERF2-iPSCs had enlarged and flattened morphology (Figure 2C, Figure S6C), a characteristic features of senescent cells. [71] In contrast, Dox treatment in CRISPRi-iPSCs without gRNA targeting TERF2 did not lead to suppression of cell proliferation or alteration of cell morphology (Figure S6C), indicating the changes we observed in CRISPRi-TERF2-iPSCs after Dox treatment is due to suppression of TERF2 rather than Dox per se. Majority of Dox-treated cells (>90%) stained positively for the senescence hallmark, senescence-associated β-galactosidase (SA-β-gal), [72] in a SPiDER-βGal assay [73] (Figure 2D, Figure S6D).

Figure 2.

Suppression of TERF2 leads to inhibition of cell proliferation and activation of senescence in human fAD (AG25367) iPSCs (clone B6). (A) Growth curves showing decreased cell numbers with Dox treatment. (B) EdU labeling and quantitation showing decreased EdU incorporation by Dox (4–5 randomly selected fields each with 426 to 1560 cells). (C) Cystal violet staining showing altered cell morphology by Dox. (D) SPiDER-βGal staining and quantitation showing increased SA-β-gal activity by Dox (10 randomly selected fields with 971 cells for −Dox and 19 fields with 635 cells for +Dox). gRNA-2 was used in B-D. Data were presented as mean ± SD. Statistical significance determined using Welch’s t-test is denoted as **: p<0.01, ***: p<0.001, ****: p<0.0001, and ns: not significant.

Furthermore, a significant enlargement (2.92 to 3.04-fold increase) of the nuclei was observed in Dox-treated cells compared to untreated (Figure 3A, Figure S7A). Cytoplasmic chromatin fragments (CCF), which are often observed in senescent cells, [74] were significantly increased (8.9 to 10.2-fold) in Dox-treated cells (Figure 3B, Figure S7A). Previous studies have linked CCF with loss of integrity of nuclear lamina, [74, 75] and decreased Lamin B1 level is often observed in senescent cells. [76] A significant decrease in Lamin B1 protein was observed in Dox-treated cells (Figure 3C, D, Figure S7B). CCF has been found to augment senescence-associated secretory phenotype (SASP). [75] We found that expressions of commonly studied SASP genes including IL6, CXCL8 (IL8), MMP3 and CXCL2 were significantly up-regulated in Dox-treated cells (Figure 3E). Collectively, this inducible system targeting TERF2 robustly activated cellular senescence in iPSCs in a highly homogeneous and synchronous manner. Interestingly, the senescent iPSCs still expressed pluripotency markers Nanog and Oct4 (Figure S7C).

Figure 3.

Suppression of TERF2 leads to changes in characteristic features of senescence in human fAD (AG25367) iPSCs (clone B6). (A) DAPI staining and quantitation showing increased nuclear area by Dox (9 randomly selected fields with 610 cells for −Dox and 18 fields with 310 cells for +Dox). (B) DAPI and CellMask staining for plasma membrane showing increased CCF by Dox (13 randomly selected fields with 2166 cells for −Dox and 11 fields with 991 cells for +Dox). Arrow heads point to CCF in A and B. (C) IF staining and quantitation (13 randomly selected fields with 643 cells for −Dox and 14 fields with 210 cells for +Dox), and (D) Western blotting and quantitation showing decreased Lamin B1 by Dox. (E) Quantitative RT-PCR showing increased expression of SASP genes by Dox. gRNA-2 was used in A-E. Data were presented as mean ± SD. Statistical significance determined using Welch’s t-test is denoted as *: p<0.05, **: p<0.01, ****: p<0.0001.

2.4. Genetic pathways altered by TERF2 suppression in iPSCs

To further delineate molecular changes underlying TERF2 suppression, we performed RNA-sequencing analysis of CRISPRi-TERF2-iPSCs treated with or without Dox for 4 or 6 days. Comparison of Dox-treated iPSCs carrying either gRNA-2 or gRNA-4 with untreated cells in both the AD and healthy control lines identified 3638 differentially expressed genes (DEGs; FDR < 0.05, fold change >1.5), with 1937 genes up-regulated and 1701 genes down-regulated in Dox-treated cells (Table S6). Many senescence and senescence-associated secretory phenotype (SASP) genes such as SERPINE1, FGF1, IL1A, IL6, CXCL8 (IL8), CXCL14, PLAU, IGFBP7, and MMP10 showed significant change in expression (Figure S8A). Pathway enrichment analysis revealed that up-regulated DEGs were enriched for pathways of focal adhesion, p53 signaling, PI3K-Akt signaling, MAPK signaling, cytokine-cytokine receptor interaction, extracellular matrix (ECM) organization, ECM-receptor interaction, ECM degradation and cell-cell communication (Figure 4A and Table S7), while down-regulated DEGs are enriched for pathways of chromatin modifications, DNA methylation, chromosome condensation, assembly of ORC complex, packaging of telomere ends, and DNA damage/telomere stress induced senescence (Table S7). Gene set enrichment analysis (GSEA) was performed to further explore potential biological functions. We found that previously published gene signatures of senescence and SASP [77–82] were significantly enriched in CRISPRi-TERF2-iPSCs after Dox treatment (Figure 4B, Figure S8B), consistent with the observed senescence phenotypes in these cells. Genes that significantly changed the expression in the SenMayo senescence signature [78] were shown in Figure 4C. Using Molecular Signature Database (MSigDB) hallmark gene sets, we found that gene sets representing p53 pathway, epithelial-mesenchymal transition, TNFα signaling via NF-κB, and inflammatory response were the most significantly enriched in Dox-treated cells (Figure 4D, Table S8). Collectively, RNA-seq data further indicated that suppression of TERF2 in iPSCs led to activation of p53 signaling, senescence, SASP and inflammatory response.

Figure 4.

Suppression of TERF2 results in changes in genetic pathways characteristic of senescence and inflammatory response in human iPSCs. (A) KEGG pathway enrichment of up-regulated DEGs after Dox-treatment. (B) GSEA of senescence-related gene sets. (C) Heatmap showing up-regulated genes of the SenMayo gene set in Dox-treated iPSCs. (D) GSEA of MSigDB hallmark gene sets. Samples 1 to 8 are fAD-g2-D4, fAD-g4-D4, HC-g2-D4, HC-g4-D4, fAD-g2-D6, fAD-g4-D6, HC-g2-D6, HC-g4-D6 (fAD: AG25367, clone B6; HC: 35L11, clone 22; g2: gRNA-2; g4: gRNA-4; D4: 4-day treatment; D6: 6-day treatment).

2.5. CRISPRi-mediated suppression of TERF2 elicited DDR and senescence in neural progenitor cells (NPCs) differentiated from iPSCs

To determine if our inducible CRISPRi-TERF2 is applicable over the course of differentiation of iPSCs into neural lineages, we used dual SMAD inhibitors to direct the differentiation of iPSCs into NPCs [83, 84] (Figure 5A). Differentiation was efficient, as indicated by over 90% of the differentiated cells expressing NPC markers Sox1, Sox2, Pax6 and Nestin (Figure S9A). With 4-day Dox treatment, KRAB-dCas9 was induced in majority (>80%) cells, while TERF2 expression was significantly suppressed (Figure 5B, C, D, Figure S9B, C, D). DDR were significantly elevated in NPCs treated with Dox, as indicated by the presence of p-ATM and γ-H2AX foci (Figure 5E, Figure S9E). Further, the p53/21 signaling was activated, as shown by increases in p-p53 and p21 (Figure 5C, F, Figure S9C).

Figure 5.

CRISPRi-mediated suppression of TERF2 activates DDR and the p53/p21 signaling in differentiated NPCs from fAD (AG25367) iPSCs (clone B6). (A) Outline of NPC differentiation protocol with Dox treatment. (B) Quantitative RT-PCR showing decreased TERF2 mRNA by Dox. (C) Western blotting and quantitation showing decreased TERF2, Lamin B1 and increased p21 protein by Dox. (D) IF staining and quantitation showing increased KRAB-dCas9 and decreased TERF2 by Dox (8 randomly selected fields with 447 cells for −Dox and 11 fields for 410 cells for +Dox). (E) IF staining and quantitation showing increased DDR markers γ-H2AX and p-ATM by Dox (15 randomly selected fields with 956 cells for −Dox and 11 fields with 921 cells for +Dox). (F) IF staining and quantitation showing increased p-p53 by Dox (7 randomly selected fields with 628 cells for −Dox and 6 fields with 558 cells for +Dox). gRNA-2 was used in B-F. Data were presented as mean ± SD. Statistical significance determined using Welch’s t-test is denoted as *: p<0.05, **: p<0.01, ****: p<0.0001.

We next examined senescence-associated charact0-[peristics in Dox-treated NPCs. Similar to what we observed in iPSCs, the majority of Dox-treated NPCs (93.5–94.3%) stained positive for SA-β-gal (Figure 6A, Figure S10A). Interestingly, CCF (Figure 6B) and nuclear size (Figure 6C) were increased in Dox-treated NPCs differentiated from AG25367-CRISPRi-TERF2, but not in NPCs differentiated from 35L11-CRISPRi-TERF2 (Figure S10A, B). A significant decrease in Lamin B1 was observed in Dox-treated NPCs (Figure 5B, Figure S9C). Quantitative RT-PCR analysis showed significant up-regulation of SASP genes IL6, CXCL8 (IL8), and IL1A (Figure 6D, Figure S10C). In contrast to Dox-treated iPSCs, CXCL2 and MMP3 were not significantly altered in NPCs (Figure 6D, Figure S10C), consistent with previous findings of cell type-dependent heterogeneity in SASP. [33, 85] Overall, we demonstrated that the inducible system targeting TERF2 can activate cellular senescence in iPSC-derived NPCs.

Figure 6.

Suppression of TERF2 leads to changes in characteristic features of senescence in differentiated NPCs from fAD (AG25367) iPSCs (clone B6). (A) SPiDER-βGal staining and quantitation showing increased SA-β-gal activity by Dox (9 randomly selected fields with 937 cells for −Dox and 12 fields with 455 cells for +Dox). (B) DAPI and CellMask staining for plasma membrane showing increased CCF by Dox (10 randomly selected fields each with 1422 cells for −Dox and 1141 cells for +Dox). Arrow heads point to CCF. (C) Quantitation of nuclear area increased by Dox treatment (13 randomly selected fields with 516 cells for −Dox and 11 fields with 399 cells for +Dox). (D) Quantitative RT-PCR showing increased expression of selective SASP genes by Dox. gRNA-2 was used in A-D. Data were presented as mean ± SD. Statistical significance determined using Welch’s t-test is denoted as **: p<0.01, ***: p<0.001, ****: p<0.0001, ns: not significant.

3. Discussion

By utilizing an inducible CRISPR-interference approach, we established an inducible cell system to activate cellular senescence by targeted suppression of a telomere protecting Shelterin complex protein TERF2. This inducible system robustly activated DDR, p53/p21 signaling, cellular senescence and inflammatory response in iPSCs in a highly homogeneous and synchronous manner. The inducible nature of this system allows temporal control for isogenic comparisons of the same cell populations with or without the activation of senescence over the course of differentiation of iPSCs into desired cell types, which we validated in differentiation of iPSC into NPCs. This strategy of isogenic comparison minimizes experimental heterogeneity associated with differentiated cultures and provides a tighter control for inter-line variations than comparison of isogenic cell lines that have been manipulated differently, which are prone to epigenetic and genetic drift. [86] Further, we found that senescence was efficiently induced in iPSCs and NPCs, offering a model system to study senescence in stem cells. The cell system generated in this study provides a great resource for senescence studies.

Age is the greatest risk factor of many human diseases including neurodegenerative diseases, more than combined effects of genetics and environment. [21–23] Even for familial forms of neurodegeneration, patients develop clinical symptoms only at advanced age despite carrying pathogenic mutations throughout their life, underscoring the critical role of aging in disease manifestation. Introducing proper cellular context of aging into iPSC-based models will vastly improve in vitro modeling of age-associated diseases and enable the investigation of the impact of aging on disease pathogenesis. Aging is a complex process accompanied by the exhaustion of tissue-specific stem cells, increased presence of senescent cells and global inflammation with numerous possible contributing causes, [28, 29] which presents a significant challenge in mimic age-related changes for modelling age-associated diseases. Various studies have tried to trigger oxidative or mitochondrial stress by exposing iPSC-derived cells to toxins. [87–91] These strategies accelerate the accumulation of DNA or other types of molecular damage, but stressing iPSC-derived “young” cells alone is not sufficient to appropriately mimic age-related changes. Expressing premature aging (Werner or Hutchinson Gilford progeria syndrome) causing proteins has been employed to accelerated “aging” in iPSC-based models to improve age-associated and disease-related phenotypes, [15, 32] but it remains to be determined how closely these approaches model physiological aging rather than pathological conditions. An alternative strategy is to manipulate aging hallmarks [28] for in vitro “aging” of iPSC-derived cells. Many aging hallmarks are entwined and aberrations in one hallmark could lead to changes in other hallmarks. Accumulating evidence indicates that telomere dysfunction can drive or amplify other aging hallmarks including cellular senescence, genomic instability, stem cell exhaustion, inflammation, mitochondrial dysfunction, and epigenetic dysregulation. [92] Recent studies have focused on triggering telomere shortening or cellular senescence using small molecule compounds and showed these manipulations result in improved age-associated and disease-related phenotypes in iPSC-based modeling of neurodegeneration. [30, 31] Prolonged treatment of telomerase inhibitors is necessary to cause progressive telomere shortening with successive cell divisions in iPSCs, but such treatment is handicapped by the need to preserve iPSCs’ capability to proliferate before differentiation into the desired cell types. With this restriction, telomeres are not sufficiently shortened to activate cellular senescence in iPSCs or differentiated neurons with telomerase inhibitor treatment. [31] The heterogenous nature of telomere lengths of different chromosomes and cells inevitably leads to varied onsets of phenotypes. Our genetic approach of targeting TERF2 results in telomere dysfunction, persistent DDR and cellular senescence accompanied by the inflammatory response, four hallmarks of aging. [28] This approach complements the pharmacological approach to induce telomere shortening and senescence [30, 31] and offers tight control without obvious cytotoxicity. Further, the versatility of our CRISPRi approach allows simultaneous targeting of multiple proteins involved in different aging hallmarks, such as epigenetic alterations, genomic instability, loss of proteostasis or mitochondrial dysfunction, to understand the complex and pleiotropic nature of aging in disease pathogenesis.

Senescence activation in differentiated NPCs validates that the inducible system retains its efficacy upon differentiation. This raises the promise of application of this inducible cell system in differentiated neurons or other neural lineages such as astrocytes or oligodendrocytes. Cellular senescence is commonly associated with irreversible cell cycle arrest, which is not always a necessary outcome of senescence, as senescence can occur in non-dividing cells. [39–41] Senescence activated by telomere dysfunction without telomere shortening is especially relevant to post-mitotic cells such as neurons, as these cells are unlikely to undergo telomere shortening associated with cell division. Senescent neurons have been detected in the brains of aged humans and mice [93–96] as well as AD patients. [24, 44, 52] Recent studies suggest that neurons in AD patients are prone to senescence [24] and senescent neurons constitute the most senescent cells in AD brains. [52] TERF2 expression has been shown to increase in neurons when differentiated from NPCs to protect mature neurons from telomere damage. [97] Telomere dysfunction induced by TERF2 suppression in neurons activates DDR and p21 expression. [98] Analysis of the precuneus region of the AD brains shows that expression of Shelterin complex components including TERF2, TERF1, TIN2 and POT1 decreases, suggesting a selective vulnerability to telomere dysfunction in AD neurons. [99] With the CRISPRi-TERF2 system generated in this study, various cell types of interest such as neurons and glial cells can be generated and investigated in a disease-context by triggering telomere dysfunction and activation of senescence, which has a broad potential to be applied to other age-associated diseases.

4. Experimental Section

4.1. Cell lines and cell culture

Human iPSC line (fAD, AG25367) reprogrammed from skin fibroblasts of a familiar AD patient with a disease onset at age 45 and an A246E mutation in PSEN1 [65] was purchased from Coriell Institute for Medical Research. Human iPSC line (HC, 35L11) reprogrammed from skin fibroblasts of a healthy individual without dementia was generously provided by Dr. Fen-Biao Gao of University of Massachusetts Chan Medical School. [66] Human embryonic kidney 293T (HEK293T) cells were purchased from ATCC. All iPSC lines and the CRISPRi derivatives were cultured in plates coated with recombinant human Vitronectin (Gibco) and fed daily with TeSR-E8 medium (STEMCELL Technologies) at 37°C with 5% CO₂. When cells reached 70–80% confluency, iPSC colonies were treated with EDTA (0.5 mM; Corning) at room temperature and passaged at 1:4 ratio. For single cell culture, dissociation of iPSC colonies was performed using TrypLE express enzyme (Gibco) at 37°C for 3 minutes. ROCK inhibitor Y-27632 (10 μM; R&D Systems) was added in the media for the first 24 hours to promote cell survival in single cell passaging. All experiments with doxycycline hyclate (Sigma) treatment were carried out starting with single cell culture. HEK293T cells were cultured with Dulbecco’s Modified Eagle medium plus 10% FBS (Gibco) at 37°C with 5% CO₂.

4.2. Nucleofection of iPSCs and junction PCR

Two million live cells were mixed with 5 μg of pAAVS1-NDi-CRISPRi (gen2) plasmid [63] (Addgene) and 2 μg each of plasmids AAVS1-TALEN-L and AAVS1-TALEN-R [64] (Addgene). Nucleofection was performed using Nucleofector kit 2 and Amaxa Nucleofector (Lonza) with Program A-23. 48 hours post-nucleofection, cells were selected with G418 sulfate (Life Technologies). Surviving colonies were individually picked under microscope and expanded with continuous G418 selection. Genomic DNA was extracted from isolated iPSC clones with DNeasy blood and tissue kit (Qiagen) following the manufacturer’s protocol. Genomic DNA (100 ng) was used to amplify 5’ or 3’ junctions around the integration sites as well as the wild-type AAVS locus with GoTaq polymerase master mix (Promega). Primer sequences are listed in Table S1.

4.3. Guide RNA (gRNA) design, cloning, lentivirus production and infection

CRISPick (https://portals.broadinstitute.org/gppx/crispick/public) was used to design gRNA targeting sequences after the transcription start site (TSS) of TERF2. Four gRNAs were selected based on the recommended pick orders (Table S2). Oligos for each gRNA were annealed and individually cloned into the lentiviral vector sgOpti [100] (Addgene). Lentiviral packaging and transduction were carried out as previously described. [101] Briefly, lentiviral vectors were co-transfected with the packaging plasmid pCMV-dR8.74 and plasmid pMD2. VSV-G encoding vesicular stomatitis virus glycoprotein into HEK293T cells. Viral supernatant was collected 48 hours after transfection and used to infect CRISPRi-iPSCs in the presence of Transdux (System Bioscience). 48 hours post infection, cells were selected with puromycin (Tocris Bioscience).

4.4. Karyotyping and short tandem repeat (STR) analysis of iPSCs

Samples of iPSC clones were sent to WiCell Research Institute for cytogenetic analyses. At least 20 metaphases were analyzed for G-band karyotyping. STR analysis of 15 loci was carried out to verify the identity and purity of newly generated iPSC clones.

4.5. Differentiation of iPSCs into NPCs

Neural differentiation was performed as previously described with some modifications. [83, 84] Briefly, iPSCs were dissociated into single cells with TrypLE Express and seeded at 50,000 cells per well in a Vitronectin-coated 24-well plate in TeSR-E8 media supplemented with ROCK inhibitor Y-27632 (10 μM). Media was changed with fresh TeSR-E8 media the next day. Two days after seeding, cells were fed daily with neural differentiation media (NDM) consisting of 50% Knockout DMEM/F12, 50% Neurobasal, 0.5X Glutamax, 1X N-2 Supplement (Life Technologies) and supplemented with DMH1 (2 μM) and SB431542 (2 μM; Tocris Bioscience). After 14 days, cells were broken into clumps with EDTA (0.5 mM) and cultured in suspension in non-treated T75 flasks (USA Scientific) in NDM for 7 days. On day 21 of neural differentiation, neurospheres were dissociated into single cells using TrypLE Express to obtain NPCs. Doxycycline hyclate (Sigma) was added to the culture media every other day in NPCs after breaking up neurospheres.

4.6. Immunofluorescence (IF) staining and imaging

Cells grown on coverslip were fixed in 4% paraformaldehyde in PBS for 15 minutes at room temperature. Permeabilization with 0.25% Triton X-100 in PBS was performed at room temperature for 15 minutes, followed by blocking in buffer containing 10% serum, 22.52 mg mL⁻¹ glycine and 0.1% tween-20 in PBS (PBST). Incubation of primary antibodies in PBS-T with 1% bovine serum albumin was performed at 4°C overnight. After washing with PBST three times, cells were incubated with fluorophore conjugated secondary antibodies at room temperature for 1 hour. After washing with PBST three times, nuclei were counterstained with DAPI (BioLegend) and coverslips were mounted with Vectshield (Vector Laboratories). Detailed primary and secondary antibody information is listed in Table S3. Cell membrane staining was performed using CellMask plasma membrane stains (Invitrogen) following manufacturer’s protocol. Microscopy was performed using Zeiss Axio Observer 7 or Leica DMI6000B. Minimal correction for brightness and contrast were performed using ZEN software to best represent image signals. Nuclei size measurement, intensity of TERF2, p-p53 and LaminB1 signals were measured using CellProfiler image analysis software. [102] For scoring and quantification, we examined more than three randomly selected fields in at least three independent experiments of each cell line. Data were quantified using the number of randomly selected fields and/or cells in selected fields.

4.7. Western blot analysis

Cells were lysed in RIPA buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris-HCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing protease inhibitor cocktails and phosSTOP phosphatase inhibitor (Roche). Lysate (30 μg) was loaded onto 4–12% bis-tris gel and transferred to nitrocellulose membrane (Bio-Rad). Transfer of proteins was confirmed by Ponceau S (Sigma) staining. Membrane was blocked at room temperature for 1 hour with 5% non-fat dry milk in PBST. Primary antibody incubation was done in PBST with 5% milk at 4°C overnight, followed by three washes with PBST. HRP-conjugated secondary antibody was incubated in PBST with 5% milk at room temperature for 1 hour. Blots were imaged using Clarity Western ECL reagents (Bio-Rad) with the Bio-Rad ChemiDoc imaging system. Quantification of protein bands were performed using Bio-Rad Image Lab software. Detailed primary and secondary antibody information is provided in Table S3.

4.8. Quantitative RT-PCR

Total RNA was isolated using RNeasy plus kit (Qiagen) following manufacturer’s instruction. RNA (2 μg) was reverse transcribed using Superscript IV reverse transcriptase (Invitrogen). Quantitative RT-PCR was performed using SYBR green PCR master mix on the StepOne Plus Real-Time PCR System (Applied Biosystems). The median cycle threshold (Ct) of each gene (run in triplicate) was normalized to the mean Ct of the housekeeping gene β-Actin. The delta Ct for each gene was used to calculate the relative mRNA expression changes. Primer sequences for TERF2, IL1A, IL6, CXCL8 (IL8), CXCL2, MMP3, MMP10, and β-Actin are provided in Table S4.

4.9. RNA sequencing analysis

Total RNA was extracted with TRI Reagent (Millipore Sigma) and treated with DNase I using RNeasy MinElute Cleanup Kit (Qiagen). RNA integrity was determined using Agilent 4200 TapeStation. Library preparation was performed with 1 μg of total RNA. Ribosomal RNA was removed using the RiboErase kit (Kapa Biosystems) and mRNA was then fragmented. First and second strand cDNA synthesis was performed using SuperScript III reverse transcriptase (Life Technologies) with random hexamers per manufacturer’s instructions. Double stranded cDNA was ligated to Illumina sequencing adapters and then amplified for 15 cycles using primers incorporating unique dual index tags. Fragments were sequenced on Illumina NovaSeq-6000 using paired end reads extending 150 bases targeting 30M clusters. Gene-level read counts, Fragments Per Kilobase of transcript per Million mapped reads (FPKM), FPKM Upper Quartile (FPKM-UQ), and Transcripts Per Million (TPM) values were acquired using the HTAN Bulk RNA Expression pipeline developed by the Ding lab (https://github.com/ding-lab). This pipeline, implemented with STAR v2.7.10b [103] and SAMtools v1.15.1, [104] generated aligned BAM files from raw FASTQ files. Read counts were obtained using featureCounts in Subread v2.0.3 [105] and then converted to FPKM, FPKM-UQ, and TPM using the formulas outlined in GDC’s expression mRNA Pipeline documentation (https://docs.gdc.cancer.gov/Data/Bioinformatics_Pipelines/Expression_mRNA_Pipeline/). For differential gene expression analysis, DESeq2 version 1.42.0 [106] was used and a cutoff of absolute log₂FoldChange> 0.585 (FC≥1.5) and FDR (adjusted p-value) <0.05 was used to define differentially expressed genes (DEGs). Enrichment of KEGG [107] and Reactome [108] pathways was performed using DAVID functional annotation tool. [109, 110] Gene Set Enrichment Analysis (GSEA) [111] was performed using GSEA software v4.3.2 with 1000 permutations for gene sets and Signal2Noise metric for ranking genes. GSEA was carried out with published senescence or SASP signatures [77–82] as well as the Molecular Signature Database (MSigDB) hallmark gene sets. RNA-seq data were deposited in the Gene Expression Omnibus (GSE251814).

4.10. 5-ethynyl-2’-deoxyuridine (EdU) labeling and cell growth curve assay

Cells were treated with EdU (2 μM) for 6 hours at 37°C, followed by detection of EdU using Click-iT EdU cell proliferation kit (Invitrogen). After counterstaining with DAPI, cells were imaged by microscopy. For growth curve assay, cell counting was performed daily for 5 days after plating using Countess II automated cell counter (Invitrogen).

4.11. Senescence-associated β-galactosidase (SA-β-gal) and crystal violet staining

SA-β-gal staining was performed using SPiDER-βGal senescence cell detection kit (Dojindo laboratories) with incubation for 15 minutes at 37°C [73] after fixation with 4% paraformaldehyde in PBS for 3 minutes. For crystal violet staining, cells were fixed with 100% ice-cold methanol for 10 minutes, followed by staining with 0.5% crystal violet solution in room temperature for 10 minutes.

4.12. Statistical analysis

Graphpad (Prism 10) was used for statistical analysis. Data were presented as mean ± standard deviation (SD). Welch’s t-test was used for pairwise sample comparison. One-way ANOVA followed by a Kruskal-Wallis multiple comparison test for multiple comparison was performed. p<0.05 was considered statistically significant.

Abbreviations

iPSCs

induced pluripotent stem cells

NPCs

neural progenitor cells

TERF2

telomere repeat factor 2

CRISPRi

CRISPR interference

AD

Alzheimer’s disease

DDR

DNA damage response

SA-β-gal

senescence-associated β-galactosidase

SASP

senescence-associated secretory phenotype

CCF

cytoplasmic chromatin fragments

DEGs

differentially expressed genes

GSEA

gene set enrichment analysis

MSigDB

Molecular Signature Database

Dox

doxycycline

IF

immunofluorescence

STR

short tandem repeat

TSS

transcription start site

ECM

extracellular matrix

IF

immunofluorescence

Data Availability Statement

RNA sequence data are deposited in Gene Expression Omnibus and the accession number is GSE251814. All data that support the findings of this study are available upon request from the corresponding author.