Replacing murine insulin 1 with human insulin protects NOD mice from diabetes

Abstract

Type 1, or autoimmune, diabetes is caused by the T-cell mediated destruction of the insulin-producing pancreatic beta cells. Non-obese diabetic (NOD) mice spontaneously develop autoimmune diabetes akin to human type 1 diabetes. For this reason, the NOD mouse has been the preeminent murine model for human type 1 diabetes research for several decades. However, humanized mouse models are highly sought after because they offer both the experimental tractability of a mouse model and the clinical relevance of human-based research. Autoimmune T-cell responses against insulin, and its precursor proinsulin, play central roles in the autoimmune responses against pancreatic beta cells in both humans and NOD mice. As a first step towards developing a murine model of the human autoimmune response against pancreatic beta cells we set out to replace the murine insulin 1 gene (Ins1) with the human insulin gene (Ins) using CRISPR/Cas9. Here we describe a NOD mouse strain that expresses human insulin in place of murine insulin 1, referred to as HuPI. HuPI mice express human insulin, and C-peptide, in their serum and pancreata and have normal glucose tolerance. Compared with wild type NOD mice, the incidence of diabetes is much lower in HuPI mice. Only 15-20% of HuPI mice developed diabetes after 300 days, compared to more than 60% of unmodified NOD mice. Immune-cell infiltration into the pancreatic islets of HuPI mice was not detectable at 100 days but was clearly evident by 300 days. This work highlights the feasibility of using CRISPR/Cas9 to create mouse models of human diseases that express proteins pivotal to the human disease. Furthermore, it reveals that even subtle changes in proinsulin protect NOD mice from diabetes.

Fig 1. Replacement of the murine Ins1 gene with human INS.

(A) Alignment of the amino acid sequence of murine and human insulin. Amino acids in murine insulin that differ from the human sequence are highlighted. (B) A schematic diagram of the replacement of the coding sequence for murine Ins1 with human INS. Murine sequences are shown in blue and boxes represent exons. The translation start and stop codons are indicated by arrows. A homology directed repair (HDR) construct containing the human INS transcript (purple box), flanked by sequence homologous to the regions flanking murine Ins1 was used to introduce INS by CRISPR-Cas9 mediated targeting to the Ins1 locus. (5’ homology: 701bp; 3’ homology: 559bp).

Fig 2. Human insulin and C-peptide can be detected in NOD.HuPI serum and NOD.HuPI mice have normal glucose tolerance.

ELISA was used to measure (A) human insulin and (B) human C-peptide concentrations in serum from fasted NOD.HuPI wildtype (+/+, n = 7), heterozygous knock-in (KI/+, n = 10) and homozygous knock-in (KI/KI, n = 8) mice. Data are expressed as mean ± SEM. NS: p > 0.05, * p < 0.05, ** p < 0.01, **** p < 0.0001; unpaired Student’s two-tailed t test. (C) Detection of cDNA specific for human insulin (INS), murine insulin 1 (Ins1) and murine insulin 2 (Ins2) in pancreatic cells. (D) Human and mouse insulin were measured in serum from homozygous knock-in mice (n = 5). (E) A glucose tolerance test was performed on fasted female NOD.HuPI wildtype (+/+, n = 8), heterozygous knock-in (KI/+, n = 8) and homozygous knock-in (KI/KI, n = 8) mice. p > 0.05; two-way ANOVA.

Fig 3. Human insulin is localised to the islets of NOD.HuPI mice.

Pancreas sections from human (A, E), NOD.HuPI wildtype (B, F), and homozygous knock-in (C, D, G, H) were stained with anti-human proinsulin (GN-ID4) (A–D, top row), or an isotype control antibody (40BI3.2.1-s) (E–H, bottom row). Positive staining is indicated by the brown color. No staining was seen in pancreas of HuPI wildtype mice. Photos taken at 100x magnification. Representative images are shown.

Fig 4. NOD.HuPI mice have reduced incidence of diabetes and delayed insulitis.

(A) Female homozygous NOD.HuPI knock-in (KI/KI, N = 30), heterozygous knock-in (KI/+, n = 27) or wildtype (+/+, n = 34) mice were aged for 300 days. Urinary glucose was tested weekly. Mice were declared diabetic if they had high urinary glucose readings for three consecutive days and a blood glucose reading >15mmol/L. **** p < 0.0001 Log-rank test +/+ vs KI/KI and +/+ vs KI/+. Lymphocytic infiltration of the islets was scored in haematoxylin and eosin-stained pancreas sections of female NOD.HuPI homozygous knock-in (KI/KI, n = 5), heterozygous knock-in (KI/+, n = 5) and wildtype (+/+, n = 3) 100d-old mice and female NOD.HuPI homozygous knock-in (KI/KI, n = 5), heterozygous knock-in (KI/+, n = 5) and wildtype (+/+, n = 4) 300d-old mice (B, C). Islet infiltration scoring: 0 = no infiltration, 1 = peri-insulitis, 2 = < 25% islet infiltrated, 3 = > 25% of the islet infiltrated and 4 = complete infiltration. A weighted average insulitis score was calculated as described in the methods. 100d p < 0.05; 300d p > 0.05; one-way ANOVA.