Chemical inhibitors of cyclin-dependent kinase (CDKi) improve pancreatic endocrine differentiation of iPS cells

Islet transplantation, including pancreatic beta cells, has become an approved treatment for type I diabetes. To date, the number of donors limits the availability of treatment. Induction of pancreatic endocrine cells from pluripotent stem cells including iPSCs in vitro offers promise as a solution, but continues to face problems including high reagent costs and cumbersome differentiation procedures. In a previous study, we developed a low-cost, simplified differentiation method, but its efficiency for inducing pancreatic endocrine cells was not sufficient: induction of endocrine cells is non-uniform, resulting in colonies containing relatively high ratio of non-pancreatic-related cells. Here, we applied cyclin-dependent kinase inhibitors (CDKi) within a specific time window, which improved the efficiency of pancreatic endocrine cell induction. CDKi treatment reduced the prevalence of multi-layered regions and enhanced expression of the endocrine progenitor–related marker genes PDX1 and NGN3 resulting in enhanced production of both INSULIN and GLUCAGON. These findings support a step forward in the field of regenerative medicine of pancreatic endocrine cells. Supplementary Information The online version contains supplementary material available at 10.1007/s11626-023-00776-0.


Introduction
A well-known characteristic of type I diabetes mellitus is hyperglycemia resulting from the inability to produce insulin due to loss of beta cells. The current mainstream treatment for this disease is direct insulin injection. Several problems are associated with this treatment, including lifelong exogenous insulin dependence and unstable control of blood glucose levels (Gamble et al. 2018;Pathak et al. 2019). In recent years, transplantation of pancreatic islet cells has become a promising treatment to overcome the obstacles mentioned above. However, an insufficient number of donors and transplant rejection remain major challenges.
To overcome these challenges, the production of functional beta cells from induced pluripotent stem cells (iPS cells) for diabetes treatment has become a widely accepted challenge. Several bottlenecks exist in currently established methods for inducing beta cell differentiation from iPSCs. These bottlenecks include high cost due to numerous expensive reagents and lengthy procedures including cumbersome induction steps (D'Amour et al. 2006;Kroon et al. 2008;Kunisada et al. 2012;Pagliuca et al. 2014). To overcome these problems, in our previous study, we developed a new method (Horikawa et al. 2021) that utilizes fewer and less costly reagents and a simpler, shorter induction protocol. However, the differentiation level of the final product was found to be insufficient. For instance, in 2-dimensional culturing environment, differentiation seems not uniformly induced, resulting in scattered multiple layers of colonies containing non-pancreatic-related cells, recognized either as deep-colored areas through optical photomicrography or high DAPI signaling area through immunofluorescence imaging, which were defined as multi-layered regions. It has been reported that pancreatic endocrine cell genesis is enhanced by temporally regulated inhibition of cyclindependent kinase by treatment with CDK inhibitors (CDKi). CDK inhibition lengthens the cell cycle, preventing phosphorylation and subsequent degradation of NEUROG3 (NGN3), a key marker of pancreatic endocrine progenitors mainly utilizing human embryonic stem cells and mouse embryos (Krentz et al. 2017). We hypothesized that this treatment would reduce the prevalence of multi-layered regions and improve efficiency of endocrine differentiation. In this study, we examined the ability of CDKi to improve our protocol.

Materials and methods
Cell culture and differentiation TkDN4-M cell line of human iPSCs (hiPSCs), which was established from human dermal fibroblasts cells, was used in this study (Takayama et al. 2010). hiPSCs were cultured on 6-well plates coated with Matrigel (Corning, Corning, NY) in serum-free mTeSR Plus medium (STEMCELL Technologies, Vancouver, Canada), without mouse embryonic fibroblast feeder cells, at 37 °C under 5% CO 2 in air. hiPSCs were cultured with 10 µM Rho-associated kinase inhibitor (Y-27632; Adooq Bioscience, Irvine, CA) and passaged at a ratio of around 1:100-1:200 when hiPSCs reached about 80% confluency using Accutase (Gibco, Grand Island NY, Waltham, MA). Cells used for experiments had less than ten passages.
Immunohistochemistry Cells were fixed and immunostained with a standard protocol (Ninomiya et al. 2014 Statistical analyses Data are expressed as mean ± standard deviation. For comparisons of discrete data sets, paired t-test or Student's t-test was used, as indicated in the legends. Two-tailed p < 0.05 was considered statistically significant.

Results
We previously established a new protocol that differentiates iPS cells into INSULIN-producing cells with fewer steps and lower cost (Horikawa et al. 2021). This protocol, with slightly modified medium and reagents (see "Materials and methods" for detail), was applied in this research. Briefly, this protocol involves four stages of differentiation ( Fig. 1A). At the end of stage 1, increased expression of the definitive endoderm markers SOX17 and FOXA2, and reduced expression of the cell stemness markers OCT4 and NANOG were observed by qRT-PCR ( Fig. 1B-E). This trend was confirmed by immunohistochemistry (IHC) (Fig. 1F). At stage 3, where the posterior foregut differentiates into endocrine progenitor, cells positive for PDX1 and NGN3 were detected. Although expression of both NGN3 and PDX1 was detected in the mono-layered regions, in which DAPI intensity was relatively low, PDX1 was rarely detected in multi-layered regions, where DAPI intensity was higher (Fig. 1G). This suggests that the cells in the multi-layered region are differentiating into non-endocrine cells because co-expression of both NGN3 and PDX1 is essential for endocrine cell identity (Weng et al. 2020). At the final stage of differentiation (stage 4), a certain number of cells expressing C-peptide and GLUCAGON were consistently detected in mono-layered regions, whereas in multi-layered regions, fewer C-peptide-and GLUCAGON-positive cells were found (Fig. 1H). Thus, the formation of multi-layered regions seems to correlate with low efficiency differentiation to endocrine cells.
To further investigate the development of multi-layered regions, we observed cells over time, and it revealed that cells tend to form multi-layered regions over time (Fig. 2B), especially during stages of endocrine fate determination (days 9-16). In a previous report, combined treatment with CDK4/6 and CDK2 inhibitors (CDK inhibitors, CDKi) enhanced endocrine cell fate determination by stabilizing the NGN3 protein (Krentz et al. 2017). Because the reported time range for this treatment (Krentz et al. 2017) was similar to the period of multi-layered region formation in our induction system (although these researchers mainly used different cell types, embryonic stem cells, or mouse embryos), we predicted that treatment would reduce multi-layered regions by limiting cell proliferation. Although our protocol yielded relatively low NGN3 expression (Fig. 1G), this was also expected to be elevated by CDKi treatment. We thus incorporated CDK4/6 and CDK2 inhibitors into our protocol.
To determine the optimal time window for CDKi treatment in our protocol, four different time points ( Fig. 2A) for CDKi addition were examined. qRT-PCR analysis revealed that samples treated with CDKi throughout stage 3 (days 9-15) displayed upregulated INSULIN and GLUCAGON expression compared with the control group and cells treated during other periods (Fig. 2B, C), although this experiment was performed only once. This tendency was confirmed by IHC analysis on various zoom rates with apparent difference in the size of high DAPI areas (Fig. 2D, E). Based on these results, CDKi treatment throughout stage 3 was used in subsequent experiments (Fig. 3A). It should be noted that, after selecting this time window for CDKi treatment, we faced the problem of cells peeling off on the first day of stage 2. When we applied Y-27632 for an additional 24 h, until the end of the first day of stage 2, cell attachment improved. This modification was applied in the following experiment (Fig. 3A). With this new protocol, we first confirmed the inhibitory effect of CDKi on proliferation rate: the total number of cells decreased in the CDKi group (Fig. 3C), and the ratio of multi-layered regions was significantly reduced (Fig. 3B, D) in three repetitive experiments.
Although culture conditions were altered by extending application of Y-27632 by 24 h, qRT-PCR results of three repeated experiments showed that expression levels of INSULIN and GLUCAGON were significantly increased by nearly 2-fold (Fig. 4A, B), as in the previous protocol (Fig. 4C, D). However, in mono-layered regions, no significant change in the ratio of the area of endocrine cells (C-peptide-and/or GLUCAGONpositive cells) due to CDKi treatment was detected (Fig. 4E, F). We wondered whether the endocrine cell ratio was increased in the multi-layered regions, but not in the mono-layered regions. However, this was not the case: the ratio of endocrine cells to all cells showed no significant difference with or without CDKi in the multi-layered regions (Fig. 4G). However, the ratio of endocrine cells was higher in the mono-layered regions than in the multi-layered regions (Fig. 4G). Thus, we hypothesized that the CDKi treatment-induced expansion of mono-layered region would explain the upregulation of INSULIN and GLUCAGON expression levels detected by qRT-PCR. To examine this, we re-examined the samples with less-magnified images and found that the multi-layered regions, the ratio of which was decreased by CDKi treatment, contained just little INSULIN-or GLUCA-GON-expressing cells at the final stage ( Figure S1A, B).
To acquire deeper insight into CDKi treatment during stage 3, we conducted time-course analyses of CDKitreated cells. Expression of both PDX1 and NGN3 peaked 24 h after entering stage 3 in control and CDKi groups, and then slowly dropped to starting levels throughout  NINg ET Al. 414 the remainder of stage 3 (Fig. 5A, B). Peak levels of both PDX1 and NGN3 in the CDKi-treated group were higher than those in the control group (Fig. 5A, B). In agreement with qRT-PCR results, IHC showed that the ratio of NGN3 positive cells increased, and then gradually decreased after reaching its peak in the mono-layered region, as did the fraction of PDX1/NGN3 co-expressing cells in the monolayered regions. In addition, IHC analysis confirmed that the fraction of NGN3 + cells in mono-layered regions was significantly increased by CDKi treatment throughout stage 3, suggesting increased penetrance in endocrine cell fate determination in iPSCs due to CDKi treatment. Although the PDX1 mRNA level was higher in the CDKi group than in the control group, the area size of PDX1 + cells was slightly lower in the CDKi group than in the control group (Fig. 5D). These differences may be explained by our methods. RNA expression of PDX1 was measured over the whole region by qRT-PCR (including mono-and multilayered regions), while the area covered by PDX1 + cells was measured only in mono-layered regions, which were Horizontal bars indicate statistical analyses: *p < 0.05; ns, not significant; paired t-test. C, D qRT-PCR analysis of expression of INSU-LIN (C) and GLUCAGON (D) without Y27362 addition at day 5 of induction, normalized by the control group at stage 4, n = 6. Horizontal bars indicate statistical analyses: *p < 0.05; paired t-test. E IHC for GLUCAGON (GLU, green) and C-peptide (C-PEP, red) at stage 4. Arrowhead indicates the area between mono-and multi-layered regions with high intensity of GLU/C-PEP. Nuclei were stained with DAPI (blue). Scale bar: 100 μm. F Area percentage of cells expressing C-peptide and/or GLUCAGON, which are defined as the endocrine cells, in the mono-layered region, corresponding to C (n = 15, including three biological and five technical replicates). ns, not significant; Student's t-test. G Ratio of endocrine cells, which express INSULIN and/or GLUCAGON in mono-layered or multi-layered region, corresponding to C (n = 15, including three biological and five technical replicates). ns, not significant; Student's t-test.
increased by CDKi. Taken together, our results suggest that CDKi treatment increases pancreatic endocrine progenitor cells by reducing the total area of multi-layered regions and increasing the number of NGN3/PDX1 co-expressing cells in mono-layered regions. with DAPI (blue). Scale bar: 100 μm. D Area percentage of cells expressing PDX1 (left) and NGN3 (right) in 1.34 mm 2 foci from the mono-layered region of IHC photograph in C. *p < 0.05, **p < 0.01, ***p < 0.001; Student's t-test. E Area percentage of cells co-expressing PDX1 and NGN3 in the mono-layered regions, corresponding to D. *p < 0.05, **p < 0.01; Student's t-test.

Discussion
Cyclin-CDK complexes, such as CDK1/CYCLIN B (M phase), CDK2/CYCLIN A (S phase), CDK2/CYCLIN E1 (late G1 phase), and CDK4/CYCLIN D1 (G1 phase), phosphorylate discrete sites on NGN3 (Azzarelli et al. 2017). Phosphorylation decreases NGN3 stability and DNA binding capability, interfering with expression of its target genes (Krentz et al. 2017), suggesting that CDK activity regulates NGN3-mediated differentiation. Indeed, lengthening G1/S phase by restricting CDK2 and CDK4/6 activity decreases NGN3 phosphorylation, retards cell cycle progression, and enhances endocrine cell fate decisions (Krentz et al. 2017). However, NGN3 induces cyclin-dependent kinase inhibitor 1a (cdkn1a) expression, thereby inhibiting cell proliferation. This seems to be a discrepancy: cell cycle inhibition induces NGN3, but NGN3 inhibits cell proliferation. However, downregulation of NGN3 at a certain time point after cells are specified as endocrine progenitors allows expansion of the mature islet cell population (Miyatsuka et al. 2011). Moreover, CDK4, a G1/S-specific CDK, is highly expressed in pancreatic epithelium, increasing the number of endocrine progenitors (Kim and Rane 2011). These results further suggest the importance of limiting NGN3 levels after determining endocrine progenitor fate. Similarly, we removed CDKi after endocrine progenitor induction and succeed in enhancing endocrine cell differentiation. On the other hand, due to the inhibitory effect of CDKi upon proliferation of cell, the expansion of PDX1/NGN3 co-expressing endocrine progenitor cells might have been limited as well. Thus, it is thought that, although PDX1/NGN3 co-expressing cells significantly increased at early phase of stage 3, the change of the ratio of endocrine cells in the mono-layered region by CDKi treatment was insufficient at stage 4. Given these results, it is important to determine the appropriate stage/time window to inhibit cell cycle progression. In addition, several commonly recognized exogenous physical approaches have been reported interfering cell cycle, including temperature (Falahati et al. 2021), hypoxia (Druker et al. 2021), and mechanical interactions (Gupta and Chaudhuri 2022). Considering the importance of cell cycle regulation in beta cell differentiation, these methods might provide new perspective for enhancing pancreatic endocrine induction as well.
The multi-layered regions are considered to be composed of non-endocrine-related cells because IHC demonstrated low presence of C-peptide/GLU-positive cells ( Figure S1A). As mentioned in the "Introduction," NGN3 level was relatively higher in the multi-layered regions comparing with surrounding mono-layered regions as IHC showed. On the other hand, during the earlier stage of induction, we noticed that small amount of iPS cells did not successfully differentiate into endocrine cells; it is possible that these proportions of cells continued to differentiate into other types of cells under the influence of the reagents we added during induction, including SB431542, LDN193189, and retinoic acid. Considering the role these reagents played in other directed paths of differentiation and the phenotype of NGN3 in our research, there is possibility that the main component of multi-layered regions could be neuron-related cells. However, relatively high densities of C-peptide-positive cells were present in the margins of the multi-layered regions (Fig. 4E). Recent research has demonstrated the positive influence of co-cultivating multiple cell types on differentiation. For instance, osteogenic differentiation of mesenchymal stem cells can be stimulated by co-culture with several different types of cells, including osteocytes and osteoblasts, in a simplified bone niche (Birmingham et al. 2012). Neural differentiation of adipose tissue-derived stem cells can also be improved by co-culture with embryonic stem cells (Bahmani et al. 2014). Taken together, these cited studies and our results suggest that multi-layered regions may exert a positive influence on endocrine differentiation.

Conclusion
In summary, we examined the effects of cyclin-dependent kinase inhibitors (CDKi) on induction of pancreatic endocrine cells and found that this treatment enhances expression of the endocrine progenitor-related genes NGN3 and PDX1, while decreasing population of non-pancreatic-related cells by inhibiting cell proliferation, resulting in enhanced endocrine differentiation.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.