IAPP Marks Mono-hormonal Stem-cell Derived β Cells that Maintain Stable Insulin Production in vitro and in vivo

Islet transplantation for treatment of diabetes is limited by availability of donor islets and requirements for immunosuppression. Stem cell-derived islets might circumvent these issues. SC-islets effectively control glucose metabolism post transplantation, but do not yet achieve full function in vitro with current published differentiation protocols. We aimed to identify markers of mature subpopulations of SC-β cells by studying transcriptional changes associated with in vivo maturation of SC-β cells using RNA-seq and co-expression network analysis. The β cell-specific hormone islet amyloid polypeptide (IAPP) emerged as the top candidate to be such a marker. IAPP+ cells had more mature β cell gene expression and higher cellular insulin content than IAPP− cells in vitro. IAPP+ INS+ cells were more stable in long-term culture than IAPP− INS+ cells and retained insulin expression after transplantation into mice. Finally, we conducted a small molecule screen to identify compounds that enhance IAPP expression. Aconitine up-regulated IAPP and could help to optimize differentiation protocols.


Introduction
Diabetes is a heterogenous disease affecting 451 million people worldwide (Cho et al., 2018).
Pancreatic β cell death and dysfunction are the primary pathophysiological mechanisms in patients with severe autoimmune diabetes and severe insulin-deficient diabetes, comprising 6.4% and 17.5% of the diabetic population, respectively (Ahlqvist et al., 2018).β cell transplantation therapies have the potential to replace insulin injections and offer a cure for diabetic patients.However, islet transplantation is restricted by availability of high-quality donor islets and the requirement for immunosuppression to achieve graft survival (Krentz et al., 2021).Human stem cell-derived β (SC-β) cells are an emerging alternative cell source for transplantation that could overcome both of these obstacles (Krentz et al., 2021).SC-β cells secrete insulin in response to glucose and rescue glycemic control after transplantation into diabetic rodents (Pagliuca et al., 2014;Rezania et al., 2014;Russ et al., 2015;Velazco-Cruz et al., 2019).SC-β cells are not yet fully functional at the end of in vitro differentiation using currently published protocols to generate SC-islets from human pluripotent stem cells (Davis et al., 2020).However, SC-β cells mature when transplanted under the kidney capsule of immunocompromised mice (Pagliuca et al., 2014).Currently, SC-β cells for preclinical transplantation studies are defined by co-expression of the β cell transcription factor NKX6-1 and the C-peptide fragment of processed insulin (Pagliuca et al., 2014;Veres et al., 2019).These markers exclude poly-hormonal cells with inappropriate non-β cell hormone expression and select cells that remain insulin-positive after transplantation.NKX6-1 is traditionally considered an essential and specific β cell marker (Schaffer et al., 2013).However, recent single-cell RNA-seq data indicate that NKX6-1 is also expressed by progenitor (immature) cells and enterochromaffin cells generated during in vitro differentiation (Veres et al., 2019).Alternative β cell enrichment strategies are based on staining for intracellular zinc (Davis et al., 2019;Jindal et al., 1993;Lukowiak et al., 2001) and ITGA1 (Veres et al., 2019).However, ITGA1 is not specific to adult functionally mature human β cells in islets (Baron et al., 2016).Zinc granules may also be contained in immature fetal β cells (Meyramova et al., 2021;Pound et al., 2012).Thus, more specific markers are needed to identify functionally mature β cells among the SC-β population.This study aims to identify such markers.

Identification of IAPP
We first explored how changes in gene expression might correlate with improved glucosestimulated insulin secretion (GSIS) after engraftment.The 1016 iPS GFP cell line constitutively expresses GFP from the CAG promoter, enabling the isolation of iPS derived SC-islets following transplantation into mice.1016 iPS GFP cells have a modest efficiency in generating NKX6-1 + Cpeptide + cells when differentiated in vitro to SC-islets using a 6-stage protocol (Pagliuca et al., 2014).
We isolated the SC-islets using GFP expression, the live cell zinc dye TSQ, and negative selection for propidium iodide uptake (Davis et al., 2019), in the same differentiations before and 6 weeks after transplantation (Figure 1A).Post-transplantation SC-β cells lost their poly-hormonal cell population, resolving into mono-hormonal αand β-like cells (Figure 1B).Basal insulin secretion during fasting remained unchanged (Figure 1C) while GSIS improved following in vivo maturation (Figure 1D).
Levels of most islet endocrine hormones remained unchanged.The β cell hormones, islet amyloid polypeptide (IAPP), and adenylate cyclase activating peptide (ADCYAP1) were up-regulated after transplantation (Figure 1E).Co-expression network analysis in human islets revealed that IAPP and ADCYAP1 belonged to the same module (Figure 1F).This gene community was enriched in differentially expressed genes after transplantation (p 2e-7) and genes with calcium ion binding activity (GO:0005509, p 3e-6).The first principal component (PC1) correlated with other studies that investigated islet gene expression signatures associated with HbA1c (Kendall tau 0.29 p 2.7e-4) and BMI (tau 0.22 p 2.9e-3) but not age (tau -0.09, non-significant) supporting functional importance.
IAPP had the highest contribution to the first principal component (Figure 1F) and was selected for further experiments.

Dual INS and IAPP Reporter Line
We generated a dual INS and IAPP knock-in reporter line in the 1016 iPS cell background.
An Insulin Red Nucleus (IRN) INS mCherry knock-in line was generated by inserting an in-frame mCherry coding sequence downstream of the INS gene, separated by a P2A cleavage sequence (Figure 2A).Specificity was confirmed using real time RT-PCR as well as genomic PCR.A single clone of IRN 1016 iPS cells was chosen after validation and further targeted with a knock-in construct to generate an IAPP eGFP knock-in line, Amylin Nuclear Green in IRN (ANGIs) (Figure 2B).2H).Finally, we compared the sorting strategy based on INS and IAPP to the previously proposed sorting strategy for CD49A (ITGA1) (Veres et al., 2019).We found that INS + CD49A + cells were not enriched in IAPP + cells.However, majority of INS + IAPP + SC-β cells also expressed CD49A (Figure 2I).These observations were consistent with the previous studies suggesting that CD49A is a β cell surface marker but not a selective surface marker of mature SC-β cells expressing IAPP.3B and C).In addition, the IAPP eGFP+ subpopulation of SC-β cells is more stable over time in the final stage of differentiation, as these cells retained INS mCherry expression after 9 days of culture reaggregation (Figure 3D).

Functional characterization of INS
After profiling IAPP-expressing SC-β cells in vitro, we next sought to understand the dynamics of IAPP expression during maturation in vivo.We transplanted 5 million unsorted stage 6 dual reporter ANGIs SC-islets under the kidney capsule of NOD SCID immunocompromised mice and allowed the grafts to mature for 9 weeks.After transplantation, the grafts were dissociated and analyzed by flow cytometry.Cells from the same differentiations were kept in prolonged in vitro culture and analyzed in parallel.The fraction of IAPP + cells among INS + cells increased after transplantation in vivo (Figure 3E).Grafts harvested after transplantation exhibited an increase in IAPP + INS + cells among the INS + population compared with the same differentiations maintained even longer in vitro (Figure 3F).Ultimately, nearly all INS + cells gained IAPP expression at the end of 9 weeks in vivo.

Aconitine induces IAPP Expression
We conducted a small molecule screen to identify compounds that increase the proportion of

Discussion
In summary, the INS + IAPP + SC-β cell subpopulation was mono-hormonal, stable in vitro and in vivo and comprised the population of functional, insulin-producing cells after transplantation under the kidney capsule of immunocompromised mice.Thus, IAPP marks a more stable subpopulation of SC-β cells with higher insulin content that may be the source of insulin production after successful engraftment in vivo.
IAPP is often described as a diabetes-associated protein capable of forming amyloid fibrils that damage islets in type 2 diabetes (Westermark et al., 2011).However, IAPP is co-secreted with insulin in adult β cells (Westermark et al., 2011) and does not form fibrils in a heterodimer with insulin (Shen et al., 2006;Wiltzius et al., 2009).Thus, IAPP has a physiological and not only a pathological role and is a logical marker of maturing SC-β cells in vitro.
IAPP mRNA expressed in the pancreatic islet is β cell-specific and highly transcribed (Baron et al., 2016), which is beneficial for visualization using fluorescently encoded reporter transgenes, cell sorting, and screening purposes.We demonstrated feasibility of the approach by generating the dual reporter cell line ANGIs and conducting a small molecule screen.IAPP expression was induced in INS + SC-β cells by incubation with aconitine at the final differentiation stage.Interestingly, extract from the aconite plant (not isolated aconitine) improves glycemic control and increases percentage of β cells on pancreatic histology slices in rats with streptozotocin-induced diabetes (Liou et al., 2006;Shoaib et al., 2020).By contrast, aconitine does not affect insulin secretion in INS1 and MIN6 cells at low and high glucose (Burns et al., 2015) and does not significantly increase percentage of PDX1 + progenitor cells in differentiating embryonic stem cells (Chen et al., 2009).Thus, our findings are in line with published literature indicating that aconitine increases the stability of β cells.Aconitine opens Na + channels but also induces subsequent intracellular accumulation of calcium ions (Sun et al., 2014).Genes co-expressed with IAPP were calcium-dependent, supporting calcium accumulation as a potential mechanism of action.
IAPP has also recently been identified by other groups as a marker of more mature SC-β cells before and after transplantation (Augsornworawat et al., 2020;Bruin et al., 2015;Rezania et al., 2014;Rezania et al., 2012;Van Hulle et al., 2021).Our work supports these findings and expands on the utility of IAPP as a marker by functionally interrogating the IAPP-expressing subpopulation of SC-β cells.The connection between expression of a maturation-associated islet hormone and manipulation of membrane ion channels also suggests that manipulating membrane voltage and ion balances across the cell membrane may be a promising avenue to pursue further maturation of stem cell-derived islets in vitro.Further studies that improve IAPP induction in vitro and provide mechanistic insights into this desirable phenotype might contribute to generating an in vitro stem cell-derived therapy for diabetic patients.

Generation of 1016 iPS GFP Line
The 1016 iPS cell line was obtained from Columbia University.Constitutive GFP-expressing 1016 iPS cells were generated as described (Hockemeyer et al., 2011).Cells were selected using 2 µM puromycin, expanded, and adapted to 3D suspension culture.All experiments using human embryonic or other pluripotent stem cells were reviewed and approved by the Harvard University Embryonic Stem Cell Research Oversight (ESCRO) Committee.

Generation of ANGIs cell line
1016 iPS cells were grown at low passage and electroporated using the Neon nucleofector system using 2, 1050 V pulses at 30 msec in the presence of px330 plasmid encoding Cas9 and containing guide RNAs targeting the stop codon of the insulin locus and a second donor vector containing homology 5' and 3' from the target site and encoding an in-frame P2A mCherry with antibiotic selection.Cells were allowed to recover for 72 hours followed by addition of antibiotic for positive selection of clones having undergone successful recombination.After approximately 10 days, large colonies were picked and grown in 96 well format, expanded, and tested using PCR primers targeting the insulin locus and the transgene.These insulin red nucleus (IRN) clones were then targeted using a second round of electroporation targeting the IAPP locus.Again, colonies were picked, PCR screened, and validated.Two individual clones were used for this manuscript, maintained as separate banks and individually differentiated for experiments.

Donor islets
Islets were purchased from Prodo Laboratories, NDRI, and the University of Miami.All islets were isolated from healthy donors at the time of death and used within 7 days of shipment.All islet studies were performed in accordance with IRB approved procedures at Harvard University under protocol IRB16-0013.

Directed differentiation of SC-β cells
Pluripotent iPS cells were cultured in suspension culture using mTeSR1 basal medium and passaged using accutase and differentiated using the protocol in (Pagliuca et al., 2014).Efficiency of SC-β cell induction was confirmed as in (Pagliuca et al., 2014).Stage 3 culture media was used for extended culture in stage 6.

Immunofluorescence
Sections of SC-β clusters and cross-sections of pancreas were fixed in 4% paraformaldehyde for approximately 1 hour before washing in PBS and embedded in paraffin blocks for sectioning.Images were taken of 10 µm sections of tissue using a Zeiss Imager.Z2.Antibodies used for immunofluorescence were Rat Anti-C-peptide (Developmental Studies Hybridoma Bank, ID-4) and Goat Anti-GLP-2 (Santa Cruz Antibodies, C-20).

Flow cytometry
Flow cytometry analysis of differentiation efficiencies was performed using the LSRII flow cytometer by BD.FACS purification of pre-and post-transplant SC-β cells was performed using a combination of MoFlo Astrios, MoFlo XDP, and BD FASCARIA III sorting instruments.Cells were first gated on side and forward scatter profiles followed by gating on GFP-positive cells and subsequent TSQ (405 nm)-positive PI-negative populations following the protocol (Davis et al., 2019).Cells were sorted into stage 6 medium and centrifuged/washed before lysis in TriZOL RNA reagent for extraction using trizol/chloroform extraction.
Flow cytometry for ANGIs differentiations were performed using the Attune NXT and sorted using the above-mentioned instruments.Gating strategies for ANGIs analysis was to first plot mCherry vs violet or empty channel to identify single positive cells and the mCherry + subset was then plotted for eGFP vs empty channel or violet.Flow cytometry staining for ANGIs used the above antibodies as well as antibodies against endocrine hormones SST, PPY, GHRL in the far-red channel.These experiments included antibodies to mCherry and eGFP to enhance signal after fixation.

Transplantation
The initial in vivo maturation experiments were carried out using immunocompromised SCID-Beige male mice from Jackson Laboratories (CB17.Cg-Prkdc scid Lyst bg-J /Crl).Surgeries were performed under the Melton protocol 10-18 for Harvard University.Animals were anesthetized using avertin and 5 million total cells of SC-β differentiations were transplanted underneath the left kidney capsule using a 23 gauge butterfly needle in ice cold RPMI 1640.Animals were allowed to recover on a heated pad under supervision until active before individual housing after surgery.

Recovery of Transplanted Cells
Animals were sacrificed using CO 2 asphyxiation and secondary cervical dislocation as per our animal protocol regulations.Individual kidneys were removed from the body cavity after sacrifice.Under a fluorescent dissecting microscope the kidney capsule was peeled from the body of the kidney, taking with it the GFP-positive graft (human cell reporter in the initial experiment) or INS mCherry -positive graft (experiments with the ANGIs cell line).After isolation in PBS, 1/3 of each graft was fixed in 4% PFA overnight and washed before embedding in paraffin wax and sectioned at a thickness of 10 μm for immunohistochemistry.The rest of the graft was incubated in accutase for 10 minutes at 37 degrees followed by passing through progressively higher gauge needles from 10/15/27 gauge.After dissociation cell suspensions were filtered through a 40 µM filter and resuspended in stage 6 medium and immediately sorted.

In vitro GSIS
Pre-transplantation analysis of glucose stimulated insulin was performed using reaggregatres of 5000 cells in V-bottom ultra-low attachment 96 well plates.Clusters were fasted for 2 hours before analysis in 2.8 mM Glucose in Kreb's ringer buffer (KRB) before addition of KRB at different glucose concentrations indicated or with 30 mM KCl in the presence of 2.8 mM glucose.Spent KRB was stored in 96 well plates frozen at -20 until analysis by insulin ELISA (Alpco Ultrasensitive Insulin ELISA) and read on a FluoSTAR Optima Spectrophotometer.Readings were taken at 450 and 650 nm and empty channel (650 nm) values were subtracted from 450 spectra to remove background absorbance.

in vivo GSIS
After transplantation SCID-Beige mice were allowed to recover for 2 weeks followed by in vivo GSIS analysis at 2, 4 and 6 weeks post-transplant.Animals were fasted overnight prior to the GSIS assay.
After overnight fast, blood was collected by a facial puncture, immediately followed by intraperitoneal injection of 2 g/kg glucose.After 30 minutes blood was again collected by the mandibular bleeding method.Blood was spun down and plasma isolated and frozen at -20 degrees Celsius until analysis by insulin ELISA.

RNA Sequencing
Following RNA isolation, RNA quality was first assessed using Bioanalyzer RNA Pico chips.cDNA was then prepared using the Wafergen PrepX SPIA platform.Libraries were barcoded using Illumina indices and sequenced using the Illumina NextSEQ.Samples were sequenced using 75 bp paired-end reads.The reads were aligned to GRCh38 and quantified with RSEM (Li and Dewey, 2011).Read counts were TMM normalized (Robinson and Oshlack, 2010).Log 2 (TMM+0.25)followed an approximately normal distribution based on visual examination of quantile-quantile plots against normal distribution.Differentially expressed genes post-vs pre-transplantation were identified using mixed effect linear model with cell differentiation batch as random intercept.Genes with Benjamini-Hochberg FDR < 0.05 were considered significant.GO term enrichment was analyzed online with PANTHER (http://pantherdb.org/)(Mi et al., 2017) and AmiGO2 (http://amigo.geneontology.org/amigo)(Carbon et al., 2009).

Co-expression network
RNA-sequencing data on bulk human islets GSE50244 (Fadista et al., 2014) was reprocessed from FASTQ files with the bcbio pipeline (Guimera, 2011) with alignment to human genome build hg38 with hisat2 aligner and standard quality format.Counts were TPM-normalized and log2 transformed.
MEGENA co-expression network (Song and Zhang, 2015) was constructed from genes with > 1 TPM in all samples.

Small molecule screen
ANGIs cells were differentiated into week 1 of stage 6, followed by dispersion and re-plating on Matrigel in 384-well plates and incubated in compounds for 96 hours in basal stage 6 medium followed by high content imaging analysis.Each compound was included in two concentrations with RNA-seq experiments: GSE135944 of TSQ + PI -sorted SC-islet cells before and after in vivo maturation as well as human cadaveric islets and GSE263366 of INS mCherry+ , IAPP eGFP+/-SC-β cells after 14 days of culture in stage 6.The co-expression network model is included as supplementary data S1 and can be read into R with the igraph package.
Richardson, Christian Hellriegel, and the Biopolymers Facility at Harvard Medical School for logistics and support.
in vitro.Nine compounds increased IAPP expression in 2D culture (Figure 4A, B).Of these compounds, only aconitine was successfully validated in 3D culture and increased INS + IAPP + cells in a dose-dependent manner (Figure 4C).Up-regulation of IAPP was observed already within 48 hours of incubation with aconitine (Figure 4D, E).Addition of aconitine to the final stage of differentiation resulted in accelerated IAPP induction, with approximately double the number of IAPP + cells after three weeks of culture (Figure 4F).Aconitine induced IAPP expression only in the INS + cells, as no change in IAPP signal was observed in INS -cells.Similarly, increased expression of IAPP was not accompanied by change in the overall INS + population size (data not shown).
four replicates per condition.Small molecule libraries were SCREEN-WELL compound libraries for nuclear receptor ligands, bioactive lipids, and Kinase inhibitors from Enzo Life Sciences.Compounds were fed twice, 48 hours apart and imaged using the Zeiss Cell Discoverer and downstream analyses were performed using the Zeiss Blue image analysis software.Aconitine-mediated upregulation of IAPP positivity was validated over short-term time course using IncuCyte ® live cell imaging to quantify the proportions of IAPP eGFP reporter-positive cells among the INS mCherry reporter-positive population.

Figure 1 .
Figure 1.Identification of IAPP.(A) Outline of the in vivo maturation experiment transplanting SCislets to the kidney capsule in immunodeficient mice.(B) Immunofluorescence of β cell hormone Cpeptide and α cell GLP-2 before and after transplantation (Tx) of SC-islets indicates lack of polyhormonal cells in grafts.Human insulin levels in (C) fasted and (D) glucose-stimulated state at three timepoints during in vivo engraftment.Statistical significance was determined using a 2-way ANOVA test (p < 0.01.) (E) Islet hormone expression in SC-islets before and after transplantation by RNA-seq.n = 5 differentiations pre-transplant, n = 7 mice post-transplant, and n = 4 cadaveric islet donors.(F) Network-based prioritization of IAPP.The schematic graph shows gene community (circles) and its underlying correlation structure (lines connecting the dots).Circle color indicates fold change after vs before transplantation.Circles signed with gene symbol are β-cell enriched genes.Bar graph illustrates top 30 member genes by contribution to the first principal component (PC1).

Figure 2 .
Figure 2. Characterization of the Dual INS and IAPP Knock-in Reporter Line ANGIs.Transgene design (top) and validation of the transgene reporter (bottom) for (A) Insulin Red Nuclear (IRN) and (B) Amylin Nuclear Green in IRN (ANGIs) knock-in.(C) Transgene fluorescence of ANGIs reporter line at day 7 of stage 6 of SC-islet differentiation protocol quantified by flow cytometry.(D) Structured illumination image of ANGIs reporter line after SC-islet differentiation.(E) INS mCherry expression during prolonged in vitro culture during at stage 6 of SC-islet differentiation quantified by flow cytometry.(F) Flow cytometry timecourse of IAPP eGFP expression within the INS mCherry population during prolonged in vitro culture during at stage 6 of SC-islet differentiation.