Presence of immunogenic alternatively spliced insulin gene product in human pancreatic delta cells

Aims/hypothesis Transcriptome analyses revealed insulin-gene-derived transcripts in non-beta endocrine islet cells. We studied alternative splicing of human INS mRNA in pancreatic islets. Methods Alternative splicing of insulin pre-mRNA was determined by PCR analysis performed on human islet RNA and single-cell RNA-seq analysis. Antisera were generated to detect insulin variants in human pancreatic tissue using immunohistochemistry, electron microscopy and single-cell western blot to confirm the expression of insulin variants. Cytotoxic T lymphocyte (CTL) activation was determined by MIP-1β release. Results We identified an alternatively spliced INS product. This variant encodes the complete insulin signal peptide and B chain and an alternative C-terminus that largely overlaps with a previously identified defective ribosomal product of INS. Immunohistochemical analysis revealed that the translation product of this INS-derived splice transcript was detectable in somatostatin-producing delta cells but not in beta cells; this was confirmed by light and electron microscopy. Expression of this alternatively spliced INS product activated preproinsulin-specific CTLs in vitro. The exclusive presence of this alternatively spliced INS product in delta cells may be explained by its clearance from beta cells by insulin-degrading enzyme capturing its insulin B chain fragment and a lack of insulin-degrading enzyme expression in delta cells. Conclusions/interpretation Our data demonstrate that delta cells can express an INS product derived from alternative splicing, containing both the diabetogenic insulin signal peptide and B chain, in their secretory granules. We propose that this alternative INS product may play a role in islet autoimmunity and pathology, as well as endocrine or paracrine function or islet development and endocrine destiny, and transdifferentiation between endocrine cells. INS promoter activity is not confined to beta cells and should be used with care when assigning beta cell identity and selectivity. Data availability The full EM dataset is available via www.nanotomy.org (for review: http://www.nanotomy.org/OA/Tienhoven2021SUB/6126-368/). Single-cell RNA-seq data was made available by Segerstolpe et al [13] and can be found at https://sandberglab.se/pancreas. The RNA and protein sequence of INS-splice was uploaded to GenBank (BankIt2546444 INS-splice OM489474). Graphical abstract Supplementary Information The online version of this article (10.1007/s00125-023-05882-y) contains peer-reviewed but unedited supplementary material..


Introduction
Polyhormonal endocrine cells have been shown to reside in human fetal pancreatic islets and in individuals with type 2 diabetes and chronic pancreatitis but fully differentiated endocrine cells are classically dedicated to produce a single hormone (i.e. glucagon by alpha cells, insulin by beta cells and somatostatin by delta cells) [2][3][4][5]. Under this definition, insulin gene (INS) expression is restricted to pancreatic beta cells. Yet, accumulating data indicate that even mature human beta cells are more plastic than previously assumed [6]. While the differentiated state of beta cells is maintained by reinforcement of specific gene regulatory networks and repression of other transcriptional programmes [7][8][9][10], specific circumstances such as metabolic and mechanical stress have been shown to cause spontaneous dedifferentiation and transdifferentiation of human beta cells [11,12]. Conversion of beta cells into alpha and delta cell-like states observed in individuals with type 2 diabetes has been proposed to contribute to reduced functional beta cell mass and beta cell failure [3]. Furthermore, in vitro disruption of human islet integrity has been reported to cause the spontaneous conversion of some beta cells into glucagon-producing cells. This endocrine plasticity has been proposed to allow dysfunctional beta cells to escape apoptosis due to environmental stress as well as replenish beta cell mass [13]. In situ hybridisation and single-cell transcriptome analysis of human islet cells have confirmed the presence of INS mRNA in alpha and delta cells [1]. Approximately 46% of islet cells were found to express more than one additional hormonal transcript per cell, with a considerable portion containing both insulin and somatostatin transcripts [14].
Alternative splicing increases proteome diversity by generating multiple mRNA transcripts from a single gene that differ in their assembly of exons and introns. Approximately 95% of the human transcriptome is estimated to derive from alternatively spliced transcripts [15]. Tissue-specific splicing patterns allow expression of genes in different cell types to produce protein isoforms that differ in biological composition and activity [16]. Alternative splicing networks are implicated in a broad variety of biological processes, including; maintenance of pluripotency, directing cell differentiation, cell lineage commitment and tissue identity [17]. Splicing patterns are highly dynamic and therefore provide a mechanism to allow swift adaptation to changes in the local microenvironment [18]. Many tumours undergo alternative splicing, potentially generating neoantigens that are prominent targets in cancer immunotherapy [19]. Likewise, splice variants generated in beta cells may contribute to autoimmunity and type 1 diabetes [20,21]. Type 1 diabetes autoimmunity is hallmarked by insulitis in which beta cells are specifically destroyed by autoreactive cytotoxic T lymphocytes (CTLs) that are highly reactive to preproinsulin (PPI) epitopes [22][23][24]. The beta cell transcriptome was shown to be highly impacted by inflammatory and metabolic insults [25]. Long RNA sequencing and ribosomal profiling revealed the extreme diversity of the beta cell transcriptome and proteome [26]. Experiments conducted in HEK293T cells overexpressing INS demonstrated the presence of cryptic splice sites in INS mRNA, as multiple PPIcoding insulin transcript variants were detected [27,28].
We investigated alternative splicing of INS mRNA in human islets and determined the expression and immunogenicity of alternative insulin protein products in endocrine cells.

Methods
Human islets Pancreatic islets were obtained from human cadaveric donor pancreases with consent. The reported investigations were carried out in accordance with the declaration of Helsinki (2008). Islets were isolated as previously described [29]. See electronic supplementary material (ESM) Methods for details. The checklist for reporting human islet preparations is presented in ESM Table 1.
Single-cell transcriptome analysis Single-cell RNA-seq data from non-diabetic donors were acquired online [1]. Data from all delta and beta cells were merged into single BAM files per donor. Reads in the region of interest (chr11:2157102-2163862) were extracted and Sashimi plots were generated using the Integrative Genomics Browser.
Generation of custom polyclonal antisera Custom polyclonal antisera were generated by immunising rabbits with the synthetic peptides MLYQHLLPLPAGEC (DRiP 1-13 ; cysteine s e r v e d a s a n c h o r r e s i d u e f o r t h e c a r r i e r ) a n d LLHRERWNKALEPAK (SPLICE 81-95 ) (Eurogentec, Belgium). The rabbits were repeatedly boosted for 28 days with synthetic peptide and bled before and after immunisation. Immune reactivity to the specific peptides was tested by ELISA performed by the manufacturer.
Recombinant polypeptides Human recombinant polypeptides were synthesised as previously described [31]. Protein encoding cDNA was obtained from human pancreatic islets by PCR and cloned in pDest17 for protein production in Escherichia coli using gateway cloning technology (Invitrogen, Carlsbad, CA, USA). Recombinant proteins were purified by His6 affinity purification tag and freeze-dried. Purified polypeptides were dissolved in 0.05% acetic acid in MQ/PBS to a stock concentration of 1 mg/ml. CTL activation assay HEK293T cells expressing the alternatively spiced INS mRNA were cocultured with CTLs directed against the PPI signal peptide PPI [15][16][17][18][19][20][21][22][23][24] . The supernatant fraction was used for detection of MIP-1β production by the CTLs. See ESM methods for details.
Statistical analysis All data points are presented as mean values (±SD). Statistical calculations were carried out using Graphpad Prism 9 (Graphpad software, San Diego, CA, USA). Statistical tests are indicated in the figure legends. A p value of <0.05 was considered significant.

Evidence of alternative INS RNA splicing in human islets
Analyses performed on RNA isolated from human islets of three different donors identified two major INS RNA variants ( Fig. 1). Nucleotide sequencing of these INS cDNA variants indicated that the larger, more-abundant INS RNA variant represents full-length PPI in which intron 1 and 2 have been fully spliced out (ESM Fig. 1). This full-length INS mRNA has been shown to generate an insulin defective ribosomal product (INS-DRiP), in particular under endoplasmic reticulum (ER) stress, which is a target of islet autoimmunity and associated with type 1 diabetes pathology [32]. The shorter, less-abundant cDNA variant resulted from a cryptic splicing site within exon 3 at position 1338 (ESM Fig. 1), predicted by in silico analysis (not shown). The open reading frame that is formed by this alternative splicing may lead to the translation of a polypeptide in which the signal peptide and B chain of the canonical PPI are intact but the C-terminal end of the molecule differs because of RNA translation into the +2 reading frame (referred to as INS-splice). Coincidently, this C-terminal region is identical to the C-terminus of INS-DRiP except for the first ten-amino-acid immunodominant N-terminus that is unique to INS-DRiP [32] (Fig. 1).
Alternatively spliced INS mRNA is a template for translation in delta cells To investigate these alternative INS-derived proteins, rabbits were immunised with a short polypeptide unique to INS-DRIP (DRiP 1-13 ) and a short polypeptide of the C-terminus shared between INS-DRiP and the predicted polypeptide INS-splice ('SPLICE 81-95 '). The peptides were selected from analysis of the UniProt human protein Knowledgebase using the basic local alignment search tool (BLAST) to avoid cross-reactivity to other known proteins To investigate whether the INS-derived polypeptides are generated in islets, human pancreatic sections were stained with the pre-immunisation or post-immunisation antisera. The localisation of the N-terminal INS-DRiP polypeptide within beta cells is consistent with our previous findings and supports beta cell destruction by CTLs directed against INS-DRiP [32] (Fig. 2a). Yet, the SPLICE 81-95 antiserum raised to the C-terminus shared between INS-DRiP and INS-splice did not co-localise with insulin, implying that SPLICE 81-95 + cells are not beta cells (Fig. 2b). To assess the identity of these SPLICE 81-95 + cells, human pancreatic sections were costained with various endocrine cell markers (insulin, glucagon and somatostatin). Staining of the SPLICE 81-95 epitope proved restricted to delta cells as indicated by its exclusive colocalisation with somatostatin (Fig. 2c).
Since the alternatively spliced insulin isoform product shares an N-terminus with PPI, we tested whether the presence of the signal peptide contributes to post-translational processing and intracellular localisation of INS-splice. Detailed examination of pancreatic slices by high-resolution EM with quantum dot-labelled SPLICE 81-95 antiserum demonstrated that INS-splice was localised to secretory granules of delta cells (Fig. 2d-f), which could be clearly distinguished from insulin secretory granules of beta cells by their unique ultrastructure [33]. This confirms that INS-splice is transported to delta cell granules and implies that it is secreted upon degranulation. Of note, staining of other endocrine tissues demonstrated that the presence of the INS-splice polypeptide is limited to pancreatic islets (ESM Fig. 3). Immunohistochemistry of mouse pancreas sections revealed the presence of INS-splice in delta cells, similar to humans (ESM Fig. 4).

SPLICE 81-95 antiserum does not cross-react with somatostatin
To validate the presence of INS-splice in delta cells and exclude cross-reactivity with somatostatin, HEK293T cells expressing INS were generated. Expression of INS in these cells led to expression of both full-length PPI mRNA and the alternatively spliced insulin mRNA variant, as observed in human islets (ESM Fig. 5a, b). Western blot analysis of lysates of these surrogate beta cells indicated that PPI is expressed, as well as an insulin isoform detected by SPLICE 81-95 antiserum (ESM Fig. 5c Fig. 5d-f). SPLICE 81-95 antiserum did not cross-react with recombinant somatostatin, as assessed by western blot (ESM Fig. 5g). In addition, antibody blocking assays using recombinant somatostatin did not affect detection of SPLICE 81-95 antiserum to recombinant INS-splice, while antibody blocking with recombinant INS-splice markedly reduced INS-splice detection (ESM Fig. 5h). Similarly, blocking of SPLICE 81-95 antiserum using the immunisation peptide reduced the mean fluorescence of the SPLICE 81-95 + islet cell population compared with irrelevant peptide (ESM Fig. 6).
Alternatively spliced INS RNA is expressed in beta and delta cells To further validate the presence of spliced INS mRNA in delta cells, we used a publicly available single islet cell transcriptome dataset and adopted the validated cell type classification of Segerstolpe et al [1], characterised by discrete clusters of endocrine cell types (ESM Fig. 7a). All delta cells  Fig. 7b). We searched for supporting reads of the alternative splice junction in both beta and delta cells. Several insulin transcripts were present in beta cells and delta cells (Fig. 3). Among the alternative INS mRNA splice variants, the one using the cryptic splicing site within exon 3 was detected in a subset of delta cells and beta cells. Splicing of insulin transcripts was studied in beta cells and delta cells of five nondiabetic human donors. Of note, aside from the regular and alternatively spliced insulin transcripts coding for PPI and INS-splice, respectively, we report additional alternatively spliced insulin transcripts detectable in subsets of delta cells and beta cells with alternative splice acceptor sites in exon 2 and exon 3 of the INS RNA (Fig. 3). Furthermore, low levels of alternatively spliced INS mRNA were detected in the alpha, epsilon and gamma cell clusters, although mean transcripts per million (TPM) values were 2.8, 2.8 and 2.4 times lower, respectively, compared with delta cells, and 79, 77, 67 times lower, respectively, compared with beta cells (ESM Fig. 7b). was detected in delta cells, we analysed the presence of the insulin B chain fragment in the human pancreas by immunofluorescence. Co-localisation of insulin B chain, insulin and somatostatin was determined (Fig. 4a-c). Although complete co-localisation between insulin and insulin B chain was expected, we only found 22% co-localisation, indicating that the insulin B chain staining is a gross underestimation, likely due to a low sensitivity of the antibody against insulin B chain compared with the 'gold standard' insulin polyclonal antibody from Dako (Fig. 4a). Importantly, some locations showed colocalisation of somatostatin and insulin B chain in the absence of insulin staining (Fig. 4b,d, grey  reconstruction showed that this insulin B chain staining was indeed inside delta cells ( Fig. 4d and ESM video). Control staining with secondary antibody alone was negative (ESM Fig. 8). These data confirm insulin B chain expression in a subset of delta cells.

Insulin B chain expressed in delta cells Since an alternatively spliced insulin gene product containing the insulin B chain
To validate and enumerate insulin B chain expression in delta cells, a single-cell western blot was performed on dispersed islet cells from two human donors (1000 islet equivalents each), using antibodies against somatostatin, C-peptide and insulin B chain. Only single delta cells (n=554) were included for analysis, and doublets were excluded by measuring a composite of their DNA intensity and hormonal content (Fig. 4e). A subset of delta cells was found that expressed insulin B chain besides somatostatin (n=54, 9.7%), of which some were negative for C-peptide (n=7, 1.3%). Detection of insulin B chain without C-peptide indicates the presence of INS-splice and excludes the presence of PPI in this subset of delta cells. These results provide evidence of the presence of INS products, in particular the immunogenic B chain, in a subset of delta cells.
INS-splice activates PPI-specific CTLs INS-splice includes the complete signal peptide and B chain sequences of PPI that contain highly immunogenic epitopes targeted in individuals with type 1 diabetes [22,24]. To investigate the immunogenicity of INS-splice, HEK293T cells were transfected with a vector encoding for INS-splice-IRES-GFP and GFP expression was validated by RT-PCR 24 h post transfection (Fig. 5a). The activation of PPI 15-24 -specific CTLs was determined by measuring their MIP-1β secretion into the supernatant fraction after co-culture with INS-splice-expressing HEK293T cells. In the absence of INS-splice expression, MIP-1β secretion was basal, while its secretion was highly upregulated in the presence of INS-splice in an effector dosedependent manner (Fig. 5b). These data demonstrate that human cells can generate immunogenic epitopes from the INS-splice polypeptide that are processed and presented to   (Fig. 6a). Co-localisation was quantified by analysing 36 islets from six donors and the mean MCC of IDE with somatostatin (0.13) was significantly lower than that for IDE with insulin (0.82) (Fig. 6b). Next, we tested whether IDE could cleave INS-splice. Recombinant INSsplice was incubated with human IDE and visualised by Coomassie staining. INS-splice was indeed digested by IDE ( Fig. 6c and ESM Fig. 9). This supports a role for IDE in the selective expression of INS-splice protein in delta cells, reconciling the discrepancy between INS-splice RNA and protein expression in beta and delta cells.

Discussion
While insulin has been widely studied for its role in glucose homeostasis and islet autoimmunity in type  [37,38]. The low expression rate of INS-derived products in delta cells (i.e. alternatively spliced insulin RNA and INS-splice protein) is conceivably close to the sensitivity limit of these singlecell-based methods; this could help explain why SPLICE 81-95 antiserum detected INS-splice in all delta cells using immunohistochemistry, while INS-splice mRNA and the insulin B chain were only detected in a delta cell subset. Alternatively, the observation that delta cells stained with SPLICE 81-95 antiserum only partially stain with anti-insulin B chain antibody suggests that additional INS-splice products may be expressed in delta cells, as indicated by our transcriptome analyses as well as by detection of Cpeptide traces in some delta cells containing insulin B chain. Detection of these potential additional alternatively spliced INS products is hampered by variation in protein length (lacking stop codon), rapid degradation of nonconventional proteins, low expression levels and limited availability of specific detection methods. Our data suggests that there are subpopulations of delta cells, as previously demonstrated in beta cells. Delta cell subsets showed heterogeneity regarding RNA expression and INS-derived protein expression.
The presence of the INS-splice protein in human and mouse delta cells is intriguing and may have implications for studies using the insulin promotor as a supposedly specific reporter for beta cells [39][40][41]. Our results imply that use of the insulin promotor activity to specifically target beta cells could lead to off-target effects in delta cells. In addition, the presence of INS-splice protein implies the activity of the insulin signal peptide and B chain in delta cells as confirmed by single-cell western blots in a subpopulation of delta cells. These peptides are major targets for islet autoimmunity [22,23,[42][43][44]. INS-splice-expressing cells activated PPI-specific CTLs, validating the immunogenicity of the INS-splice peptide. Hence, a subset of delta cells may produce and present diabetogenic epitopes in HLA, making them vulnerable to attack by diabetogenic T cells. Delta cells have not yet been thoroughly investigated for their involvement in islet autoimmunity in type 1 diabetes. While impaired delta cell function has been reported [45], data on delta cell destruction by autoreactive CTLs are still lacking. INS-splice shares sequence homology with a previously described INS-IGF2 protein [46] and a 74-amino-acid proinsulin protein [47], although their C-termini differ. Since all three isoforms retained the insulin signal peptide and B chain, the intracellular distribution and function may overlap.
The relevance of INS mRNA expression in non-beta endocrine cells remains unclear. Delta cells have an important role in beta cell development during organogenesis [48,49]. Human beta cells have also been shown to change identity via de-and transdifferentiation [11]. Alternative splicing is involved in maintaining lineage differentiation and tissue identity as well as maintenance of cell pluripotency and is influenced by the microenvironment [17]. It remains unknown whether INS promoter activity in non-beta cell endocrine cells is a remnant of their common progenitor cell or contributes to maintaining endocrine cell plasticity in adolescence. While the function of INS-splice protein is still enigmatic, its presence in secretory granules implies that INS-splice is cosecreted with somatostatin during exocytosis and may have paracrine or endocrine function in the developmental destiny of human islet cells.