Dual-agonist molecules combining glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) activity represent an exciting therapeutic strategy for diabetes treatment. Although challenging due to shared downstream signalling pathways, determining the relative activity of dual agonists at each receptor is essential when developing potential novel therapeutics. The challenge is exacerbated in physiologically relevant cell systems expressing both receptors. To this end, either GIP receptors (GIPR) or GLP-1 receptors (GLP-1R) were ablated via RNA-guided clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 endonucleases in the INS-1 pancreatic β-cell line. Multiple clonal cell lines harbouring gene disruptions for each receptor were isolated and assayed for receptor activity to identify functional knockouts (KOs). cAMP production in response to GIPR or GLP-1R activation was abolished and GIP- or GLP-1-induced potentiation of glucose-stimulated insulin secretion (GSIS) was attenuated in the cognate KO cell lines. The contributions of individual receptors derived from cAMP and GSIS assays were confirmed in vivo using GLP-1R KO mice in combination with a monoclonal antibody antagonist of GIPR. We have successfully applied CRISPR/Cas9-engineered cell lines to determining selectivity and relative potency contributions of dual-agonist molecules targeting receptors with overlapping native expression profiles and downstream signalling pathways. Specifically, we have characterised molecules as biased towards GIPR or GLP-1R, or with relatively balanced potency in a physiologically relevant β-cell system. This demonstrates the broad utility of CRISPR/Cas9 when applied to native expression systems for the development of drugs that target multiple receptors, particularly where the balance of receptor activity is critical.

Introduction

The incidence of type 2 diabetes continues to increase in the developed world yet available therapies do not halt disease progression, signifying an unmet medical need. The insulinotropic peptide hormones glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) are secreted by enteroendocrine L and K cells, respectively, following nutrient ingestion, and together these ‘incretins’ decrease blood glucose levels by augmenting glucose-stimulated insulin secretion (GSIS) from β cells in the pancreas. This establishes the receptors, GLP-1R and GIPR, as important drug targets for the treatment of diabetes. Targeting agonists to the individual receptors for these peptide hormones improves glycaemic control by way of stimulating pancreatic insulin release, and GLP-1R activation can also induce satiety and weight loss [1,2].

Historically, GIPR agonists have been disregarded in the development of new therapeutics for diabetes, because the insulinotropic effect of GIP is absent from type 2 diabetics [3] and GIPR antagonism is reported to prevent diet-induced weight gain in rodents [4]. However, the role of the GIP in type 2 diabetes may be more complex, as following the near-normalisation of blood glucose using insulin treatment, the responsiveness of patients to GIP is re-established and its glucose-lowering actions are restored [5,6]. There is compelling evidence that GLP-1 and GIP in combination can have beneficial effects, such as promoting β-cell proliferation and survival in both rodents [7] and in pancreatic islets isolated from type 2 diabetics and control subjects [8]. Additionally, the efficacy of DPP-IV inhibitors could be attributed to a combined GLP-1 and GIP increase as this di-peptidyl peptidase cleaves both incretins. A recent suggestion for a beneficial synergistic application of GIP comes from reports of unimolecular dual agonists that show promising effects with regard to glucose control and weight loss in human and rodent studies compared with selective GLP-1 mono-agonists [9]. Although generally well tolerated in the clinic, GLP-1 is dose limited for the treatment of diabetes and obesity due to gastrointestinal side effects [10]; thus, inclusion of a synergistic GIP component has the potential to reduce side effects by permitting lower doses of GLP-1 in combination with GIP to achieve equivalent or greater efficacy. It is therefore essential to fully understand the contribution of each component during the development of dual GLP-1/GIP molecules.

Although distinct proteins, GLP-1R and GIPR share downstream intracellular signalling pathways via the stimulation of adenylate cyclase that result in cAMP production, insulin synthesis and ultimately insulin secretion from β cells. Clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 knockout (KO) technology opens up the possibility to understand the pharmacology and support the development of dual molecules in cells that natively express receptors with overlapping signalling pathways. This provides a superior alternative to standard systems utilising overexpression of cloned receptors that can over estimate potency and allows comparison in a physiologically relevant setting. For GLP-1R and GIPR, the use of antagonists in native cell lines is not ideal as the available competitive antagonists, although fairly selective, cannot achieve complete receptor blockade and add a layer of complexity to high-throughput agonist assays. Furthermore, the use of siRNA gives partial transient knockdown that is unsuitable for routine screening programmes.

Here, we describe the development of physiologically relevant rat β-cell GIPR and GLP-1R KO cell lines and their application in the evaluation of dual agonists engineered to possess both GLP-1 and GIP activity. Determining the relative activity for these dual agonists at each target receptor in vitro is an essential early step in their development into potential therapeutics. To our knowledge, this is the first instance where CRISPR/Cas9 technology has been used to identify balanced dual agonist activity in agents that engage both GLP-1R and GIPR in vivo.

Materials and methods

Cell lines

Clonal INS-1E cells were kindly supplied by Prof. Claes B. Wollheim (University of Geneva Medical Center, Geneva, Switzerland). Clonal INS-1 832/3 cells have been described [11] and were kindly supplied by Prof. Christopher B. Newgard (Duke University, Durham, NC, USA).

Plasmids

The pD1301-AD vector containing cDNAs encoding Streptococcus pyogenes Cas9 (hSpCas9), green fluorescent protein (GFP) and guide sequences was purchased from DNA 2.0, Inc. (Menlo Park, CA, USA). Vector contained either a 20-bp guide sequence (5′-tctggggattgtcactccag-3′) targeting DNA within the second exon of gipr or a 20-bp guide sequence (5′-ccaggagtggcgcttccgtg-3′) targeting DNA within the second exon of glp-1r.

CRISPR/Cas9 transfection and cell sorting

INS-1 832/3 cells were maintained in RPMI 1640 (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 5% fetal bovine serum (FBS), 10 mM HEPES, 1 mM sodium pyruvate, 50 µM 2-mercaptoethanol and penicillin/streptomycin at 37°C in a CO2 incubator. For transfection, cells were detached from culture flasks using accutase and resuspended at a density of 1 × 106 cells/ml in nucleofector solution (Lonza, Basel, Switzerland) for Nucleofection with Program T-027 following the manufacturer's protocol. In brief, cells were transfected with 2 µg of plasmid (containing Cas9, GFP and guide sequence) and transferred into 6-well plates for incubation at 37°C for 48 h. For sorting based on GFP fluorescence, cells were detached from plates using accutase and resuspended in phosphate-buffered saline (PBS) supplemented with 2% FBS and combined with Hoescht (Sigma-Aldrich, Gillingham, UK) at 1:10 000 dilution. Live, GFP-expressing cells were sorted by fluorescence-activated cell sorting (FACS) into 96-well plates containing pre-warmed medium with an FACSAria II cell sorter (BD BioSciences, San Jose, CA, USA). Single cells were expanded for 3 weeks to form colonies.

Sequencing and karyotyping

INS-1 832/3 cell clones were lysed in cell lysis buffer (Viagen Biotech, Inc., Los Angeles, CA, USA) overnight at 55°C. The genomic region of GIPR exon 2 was amplified using the PCR primers 5′-accatgtgcttctggcag-3′ and 5′-aatgtgtgaagagaaactacctgttt-3′, and the genomic region of GLP-1R exon 2 was amplified using the PCR primers 5′-actggaggttaaaaaaccaagagg-3′ and 5′-gaggacgtcagtgaagtgtagc-3′. PCR products were cloned into pCR Blunt vector (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) and analysed by Sanger sequencing (Source Bioscience, Nottingham, UK).

Karyotyping was performed on a fixed suspension of wild-type INS-1 832/3 cells by Cell Guidance Systems Ltd (Cambridge, UK) to identify numerical chromosomal abnormalities.

cAMP homogeneous time-resolved fluorescence accumulation assay

Cell-based cAMP accumulation assays were used to screen peptides for agonist activity against GLP-1R or GIPR endogenously expressed in INS-1 rat β-cell lines as described recently [12]. Peptide serial dilutions were prepared in assay buffer (Hanks' balanced salt solution (HBSS) supplemented with 25 mM HEPES, 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), 0.1% bovine serum albumin (BSA), pH 7.4) using an ECHO® 550 acoustic liquid handler (Labcyte Inc., Sunnyvale, CA, USA) to obtain an 11 point dose–response curve. INS-1 cells were detached from cell culture flasks using accutase or used directly from cryovials, centrifuged for 5 min at 1200 rpm and resuspended in assay buffer. Five microlitres of cell suspension were combined with serially diluted peptides for incubation at room temperature for 30 min. Cellular cAMP levels were measured using a cAMP dynamic 2 HTRF (homogeneous time-resolved fluorescence) kit (Cisbio, Codolet, France), following the two-step protocol as per the manufacturer's recommendations. In brief, cells were incubated with anti-cAMP cryptate and cAMP-d2 in lysis buffer at room temperature for 1 h and read on an Envision plate reader (PerkinElmer, Waltham, MA, USA). For antagonist experiments, INS-1E cells were pre-incubated with 1.2 µM Gipg013 or 60 nM exendin (9–39) for 30 min prior to combination with serially diluted peptides.

Glucose-stimulated insulin secretion

INS-1 832/3 cells were seeded into 24-well poly-d-lysine-coated plates (Corning, Corning, NY, USA) 72 h prior to experiments at 5 × 105 cells/well. Media were changed 24 h prior to GSIS. On the day of the experiment, cells were washed three times in low-glucose (2.8 mM) Krebs Ringer (2.6 mM CaCl2, 98.5 mM NaCl, 4 mM KCl, 1.2 mM KH2PO4, 1.2 mM MgSO4, 20 mM HEPES, 25.9 mM NaHCO3, 0.2% BSA, pH 7.4) before incubation for 1 h at 37°C in low-glucose (2.8 mM) Krebs Ringer. Supernatant was removed, and cells were incubated for 1 h at 37°C in high- (8.3 mM) glucose Krebs Ringer ± GIP (30 nM), GLP-1 (30 nM) or dual agonist C (1 nM). Supernatant was collected and insulin concentration was measured using the Rat/Mouse Insulin kit (Meso Scale Discovery, Rockville, MD, USA) as per the manufacturer's instructions. Data were normalised to the high-glucose control, and Student's t-tests were used for comparison between pairs of data, where statistical significance is indicated by *P < 0.05 and no significance by NS.

Intraperitoneal glucose tolerance test

All animal care and experimental procedures were approved by the UK Home Office and carried out in accordance with the Animals (Scientific Procedures) Act 1986. Wild-type (Glp1r+/+) and Glp1r KO (Glp1r/) littermates on a C57BL/6 background [12a] were acquired under licence from Prof. Dan Drucker. Mice were fasted for 6 h then given vehicle or 4.4 nmol kg−1 dual agonist C subcutaneously 2 h prior to ip glucose (2 g kg−1) challenge. Twenty-four hours prior to the glucose challenge, mice received a subcutaneous dose of either 0.7 µmol kg−1 Gipg013 or 0.7 µmol kg−1 of an isotype control antibody NIP228. Blood samples were collected from tail vein by tail prick at −120, 0, 15, 30, 45, 60, 90 and 120 min relative to glucose challenge for blood glucose measurements using a CONTOUR™NEXT glucose meter (Bayer, Leverkusen, Germany). Statistical significance (n = 7–8 per group) was assessed using a one-way ANOVA comparing vehicle with compound, with a Dunnett's post hoc test (*P < 0.05) using GraphPad Prism v6.03 (La Jolla, CA, USA).

Data analysis

Data are presented as mean ± SEM and n represents the number of independent experiments. For cAMP accumulation experiments, data were transformed to % ΔF as described in the manufacturer's guidelines and expressed as % Max. response, where a maximum concentration of reference peptide GIP (60 nM) defines maximum effect. The transformed data were analysed by four-parameter logistic fit to determine EC50 values using GraphPad Prism v6.03.

Materials

Commercially available chemicals were purchased from Sigma-Aldrich, unless stated otherwise. GLP-1 (7–36) amide, GIP (1–42) and exendin (9–39) peptides were purchased from Bachem (Bubendorf, Switzerland) and were reconstituted to 1 mg ml−1 in dimethyl sulfoxide (DMSO). GIPR antagonist Gipg013 was produced in-house [13]. All dual-agonist peptides were produced in-house at MedImmune; dual agonists A and B were recombinantly expressed as Fc-fusions and contain the agonist sequences YSEGT FTSDY SKLLE EEAVR LFIEW LLAG and YSEGT FISDY SIAMD KIHQQ DFVNW LLAQK, respectively [14]; dual agonist C (free-Y[AIB]EGTF TSDYS IYLDK QAA[AIB]EF VNWLL AGGPS SGAPP PS[KPALC]-amide) has been described [9]. Peptides were reconstituted to 1 mg ml−1 in DMSO or PBS.

Results

Generation of receptor KO cell lines

The individual G-protein-coupled receptors for GIP and GLP-1 peptides are natively expressed in the rat β-cell line INS1 832/3. Both cAMP- and glucose-stimulated insulin responses to incretins in this cell line are comparable with primary pancreatic islets [11,15] and are predictive of acute in vivo efficacy in lean mice [16]. This makes INS-1 832/3 a valuable cell line to screen agonists against the receptors, and this line is superior for screening purposes to other lines such as INS-1E or INS-1 832/13 due to an enhanced incretin effect in GSIS [15]. Chromosome analysis of a fixed INS-1 832/3 cell suspension was performed to investigate whether the karyotype for this cell line was normal and to determine the copy number of the target genes (Supplementary Figure S1). The ideogram showed an abnormal karyotype in all cells examined; of 21 cells evaluated, 16 cells had 38 chromosomes and 5 cells had 37 chromosomes. All cells contained a single very large chromosome, which is likely to be the result of a structural rearrangement between up to three chromosomes. Despite this, there were no anomalies on chromosomes 1 and 20 where GIPR and GLP-1R are located, respectively, suggesting that CRISPR/Cas9 targeting for these receptors should be relatively straightforward.

GIPR and GLP-1R KOs in the INS-1 832/3 cell line were generated by inducing double-strand breaks at the genomic locus with the CRISPR/Cas9 system from S. pyogenes [17]. Indels were introduced by non-homologous end-joining DNA repair. For both receptors, exon 2 was targeted to minimise the risk of functional truncated mutants. Single-transfected cells were isolated by FACS and screened for downstream cAMP accumulation resulting from Gs-coupled GIPR or GLP-1R activation. Five putative GIPR KO clones were screened for the absence of a cAMP response to GIP; four of the five GIPR KO clones were true functional nulls (Figure 1A and Table 1). The GLP-1 response was similar to that observed in the parental line, suggesting successful modification of the GIPR only (Figure 1B and Table 1). The remaining GIPR KO clone (Clone 1) retained both GIP and GLP-1 cAMP responses, suggesting that it contained wild-type receptors. Five putative GLP-1R KO clones were also screened; the GIP response was similar to that observed in the parental line (Figure 1D and Table 1), while the absence of a cAMP response to GLP-1 suggested successful GLP-1R KO in all clones (Figure 1E and Table 1). It should be noted that saturating GLP-1 concentrations were avoided due to the bell-shaped nature of this response in the INS-1 cell line (Supplementary Figure S3). This was not due to GLP-1 insolubility or toxicity at high concentrations (data not shown), and is most likely the result of GLP-1R internalisation. Western blotting was attempted to demonstrate the absence of protein, but did not yield conclusive data due to the quality of available antibodies from commercial sources [18]. However, given the robust functional data, it was decided that protein quantification was less important.

Functional cAMP screen to identify GIPR or GLP-1R KO clones.

Figure 1.
Functional cAMP screen to identify GIPR or GLP-1R KO clones.

Concentration–response curves for GIP (A and C) and GLP-1 (B and D)-induced increase in cAMP production in INS-1 GIPR KO (A and B) and GLP-1R KO (C and D) clones. Data shown are mean (±SEM) of three independent repeats. In all examples, INS-1 wild-type control responses are indicated by open circles. Sequencing data demonstrating the pattern of genomic editing in GIPR KO clone 4 (C) and GLP-1R KO clone 10 (F). Protospacer adjacent motif (PAM) in grey; gRNA sequence underlined; indels in bold.

Figure 1.
Functional cAMP screen to identify GIPR or GLP-1R KO clones.

Concentration–response curves for GIP (A and C) and GLP-1 (B and D)-induced increase in cAMP production in INS-1 GIPR KO (A and B) and GLP-1R KO (C and D) clones. Data shown are mean (±SEM) of three independent repeats. In all examples, INS-1 wild-type control responses are indicated by open circles. Sequencing data demonstrating the pattern of genomic editing in GIPR KO clone 4 (C) and GLP-1R KO clone 10 (F). Protospacer adjacent motif (PAM) in grey; gRNA sequence underlined; indels in bold.

Table 1
GIP and GLP-1 activity in GIPR and GLP-1R KO INS-1 clones
Clone # GIP EC50 (nM) GLP-1 EC50 (nM) 
INS-1 (WT) 0.46 ± 0.01 0.03 ± 0.01 
Clone 1 0.30 ± 0.06 0.04 ± 0.01 
Clone 2 – 0.04 ± 0.01 
Clone 3 – 0.03 ± 0.01 
Clone 4 – 0.03 ± 0.01 
Clone 5 – 0.03 ± 0.01 
Clone 6 0.77 ± 0.38 – 
Clone 7 0.55 ± 0.24 – 
Clone 8 0.77 ± 0.34 – 
Clone 9 0.64 ± 0.30 – 
Clone 10 0.35 ± 0.09 – 
Clone 11 1.55 ± 0.48 0.04 ± 0.01 
Clone # GIP EC50 (nM) GLP-1 EC50 (nM) 
INS-1 (WT) 0.46 ± 0.01 0.03 ± 0.01 
Clone 1 0.30 ± 0.06 0.04 ± 0.01 
Clone 2 – 0.04 ± 0.01 
Clone 3 – 0.03 ± 0.01 
Clone 4 – 0.03 ± 0.01 
Clone 5 – 0.03 ± 0.01 
Clone 6 0.77 ± 0.38 – 
Clone 7 0.55 ± 0.24 – 
Clone 8 0.77 ± 0.34 – 
Clone 9 0.64 ± 0.30 – 
Clone 10 0.35 ± 0.09 – 
Clone 11 1.55 ± 0.48 0.04 ± 0.01 

Mean ± SEM EC50 values from three independent experiments derived from four-parameter logistic fit.

Following the initial functional screens, all 10 clones were sequenced to define the underlying genetic changes responsible for the null phenotype. Amplicons spanning the targeted regions from clonal lines were cloned into the pCR blunt vector and 10 individual transformants for each were sequenced. For Clone 1, all 10 sequences were wild type consistent with the functional data (data not shown). For GIPR KO Clone 4, of the 10 sequenced transformants, 5 had the same single base deletion and 5 had a larger multiple base deletions also resulting in a frame shift, suggesting this clone was a compound heterozygote (Figure 1C). For GLP-1R KO Clone 10, 5 sequences had a single base insertion and 5 sequences had a single base deletion, again suggesting a heterozygous result leading to frameshift mutations (Figure 1F). Sequences for additional clones from Figure 1 are shown in Supplementary Figure S2.

Dual-agonist peptides are active at both GLP-1R and GIPR

A panel of dual-agonist peptides with both GIP and GLP-1 activity were synthesised in-house and primarily screened in CHO cells artificially expressing human GLP-1R or GIPR [14]. Two molecules representing GLP-1- or GIP-biased dual agonists (dual agonists A and B, respectively) and a balanced dual agonist (dual agonist C) described in the literature [9] were screened in wild-type and KO INS-1 cell lines. Dual agonist A was of comparable potency in wild-type INS-1 and GIPR KO cells, but 20-fold less potent in GLP-1R KO cells (Figure 2A and Table 2), suggesting a larger GLP-1 contribution to this dual agonist. Furthermore, dual agonist A reached ∼80% activation relative to the control GIP response. This is consistent with the partial effect of GLP-1 in this cell line. The bias of dual agonist A towards GLP-1R was confirmed in wild-type cells using the GLP-1R antagonist exendin (9–39) and the GIPR-selective antagonist antibody Gipg013 [13], which were confirmed to inhibit target receptor-mediated responses in the cAMP accumulation assay (Supplementary Figure S3). Dual agonist A was of comparable potency in wild-type INS-1 and GIPR KO cells, but 50-fold less potent in exendin (9–39)-treated cells (Figure 2D and Table 2), confirming the KO cell line data. Again, the maximum response achieved was somewhat partial, consistent with the GLP-1 biased properties of this dual agonist. Exposure to both antagonists completely inhibited all dual agonist activity, confirming effects on cAMP accumulation were mediated by both GLP-1R and GIPR.

Determination of GIPR or GLP-1R component of dual-agonist peptides.

Figure 2.
Determination of GIPR or GLP-1R component of dual-agonist peptides.

Concentration–response curves for dual agonist-induced increase in cAMP production in wild-type INS-1 clone 1 (open circles), GIPR KO clone 4 (squares) or GLP-1R KO clone 10 (triangles) (AC). Concentration–response curves for dual agonist-induced increase in cAMP production in wild-type INS-1 following pre-incubation with vehicle (open circles), 1.2 µM Gipg013 (squares), 60 nM exendin (9–39) (triangles) or both antagonists in combination (closed circles) (D and E). Data shown are mean (±SEM) of three independent repeats.

Figure 2.
Determination of GIPR or GLP-1R component of dual-agonist peptides.

Concentration–response curves for dual agonist-induced increase in cAMP production in wild-type INS-1 clone 1 (open circles), GIPR KO clone 4 (squares) or GLP-1R KO clone 10 (triangles) (AC). Concentration–response curves for dual agonist-induced increase in cAMP production in wild-type INS-1 following pre-incubation with vehicle (open circles), 1.2 µM Gipg013 (squares), 60 nM exendin (9–39) (triangles) or both antagonists in combination (closed circles) (D and E). Data shown are mean (±SEM) of three independent repeats.

Table 2
Dual-agonist activity in wild-type, GIPR and GLP-1R KO INS-1 clones or wild-type INS-1 ± GIPR and GLP-1R antagonist
Dual agonist INS-1 WT
EC50 (nM) 
GIPR KO
EC50 (nM) 
GLP-1R KO
EC50 (nM) 
Control
EC50 (nM) 
Gipg013
EC50 (nM) 
Exendin (9–39)
EC50 (nM) 
Gipg013 + exendin (9–39)
EC50 (nM) 
GLP-1 0.04 ± 0.01 0.05 ± 0.02 No activity 0.06 ± 0.01 0.05 ± 0.01 1.8 ± 0.6 3.4 ± 1.8 
GIP 0.7 ± 0.2 No activity 0.4 ± 0.08 0.7 ± 0.2 17.5 ± 6.5 0.5 ± 0.1 21.5 ± 6.2 
24.7 ± 12.6 23.9 ± 13.7 539.0 ± 229.3 30.7 ± 3.1 29.7 ± 8.8 1612 ± 1340.9 No activity 
96.9 ± 11.5 No activity 60.5 ± 7.4 121.9 ± 54.1 No activity 112.3 ± 42.0 No activity 
0.04 ± 0.01 0.03 ± 0.01 0.07 ± 0.02 0.07 ± 0.02 0.1 ± 0.04 0.1 ± 0.04 8.4 ± 5 
Dual agonist INS-1 WT
EC50 (nM) 
GIPR KO
EC50 (nM) 
GLP-1R KO
EC50 (nM) 
Control
EC50 (nM) 
Gipg013
EC50 (nM) 
Exendin (9–39)
EC50 (nM) 
Gipg013 + exendin (9–39)
EC50 (nM) 
GLP-1 0.04 ± 0.01 0.05 ± 0.02 No activity 0.06 ± 0.01 0.05 ± 0.01 1.8 ± 0.6 3.4 ± 1.8 
GIP 0.7 ± 0.2 No activity 0.4 ± 0.08 0.7 ± 0.2 17.5 ± 6.5 0.5 ± 0.1 21.5 ± 6.2 
24.7 ± 12.6 23.9 ± 13.7 539.0 ± 229.3 30.7 ± 3.1 29.7 ± 8.8 1612 ± 1340.9 No activity 
96.9 ± 11.5 No activity 60.5 ± 7.4 121.9 ± 54.1 No activity 112.3 ± 42.0 No activity 
0.04 ± 0.01 0.03 ± 0.01 0.07 ± 0.02 0.07 ± 0.02 0.1 ± 0.04 0.1 ± 0.04 8.4 ± 5 

Mean ± SEM EC50 values from more than three independent experiments derived from four-parameter logistic fit.

Dual agonist B was inactive in GIPR KO cells, but of comparable potency in wild-type INS-1 and GLP-1R KO cells (Figure 2B and Table 2), suggesting a significant GIP component to this agonist. The GIP bias of dual agonist B was confirmed in wild-type INS-1 cells in the presence of antagonists, showing comparable potency in vehicle- and exendin (9–39)-treated cells and inactive in Gipg013-treated cells (Figure 2E and Table 2). Exposure to both antagonists caused no further shift in EC50 from Gipg013 alone.

Dual agonist C is described in the literature as a balanced compound [9]. Equivalent potencies for dual agonist C were confirmed in wild-type INS-1, GIPR KO and GLP-1R KO cells (Figure 2C and Table 2). Potency was also equivalent in vehicle- or antagonist-treated cells (Figure 2F and Table 2), while exposure to both antagonists caused an 120-fold shift in potency.

Glucose-stimulated insulin secretion in INS-1 KOs

The absence of specific incretin potentiation of GSIS in INS-1 confirms receptor KO. Additionally, the persistence of a response via untargeted receptor, that is, GIP response in GLP-1R KO, demonstrates that other components of the intracellular signalling pathway are intact.

To this effect, GIP failed to elicit a significant increase in GSIS in the GIPR KO cell line; however, the fold increase in insulin response in the GLP-1R KO cell line was comparable to the wild-type response (Figure 3A). Conversely, glucose plus GLP-1 caused a significant increase in insulin secretion in the GIPR KO cell line comparable to the wild-type response, whereas the response in the GLP-1R KO cell line was attenuated (Figure 3B). The balanced dual agonist C was also screened in KO cell lines and showed no attenuation of insulin secretion in either KO cell line, confirming dual agonism and compensation via activity at remaining receptor (Figure 3C).

GSIS remains intact in KOs and demonstrates GIPR or GLP-1R contribution of dual agonists.

Figure 3.
GSIS remains intact in KOs and demonstrates GIPR or GLP-1R contribution of dual agonists.

Mean (±SEM) insulin secretion data for wild-type INS-1 clone 1, GIPR KO clone 4 or GLP-1R KO clone 10 in response to high (8.3 mM) glucose plus 30 nM GIP (A), 30 nM GLP-1 (B) or 1 nM dual agonist C (C). Responses are normalised to glucose-induced insulin secretion in the absence of incretin/dual agonist. Data shown are representative of n = 6 (A and B) or n = 5 (C) independent experiments. Statistical significance is indicated by *P < 0.05 as determined by Student's t-test, and no significance by NS.

Figure 3.
GSIS remains intact in KOs and demonstrates GIPR or GLP-1R contribution of dual agonists.

Mean (±SEM) insulin secretion data for wild-type INS-1 clone 1, GIPR KO clone 4 or GLP-1R KO clone 10 in response to high (8.3 mM) glucose plus 30 nM GIP (A), 30 nM GLP-1 (B) or 1 nM dual agonist C (C). Responses are normalised to glucose-induced insulin secretion in the absence of incretin/dual agonist. Data shown are representative of n = 6 (A and B) or n = 5 (C) independent experiments. Statistical significance is indicated by *P < 0.05 as determined by Student's t-test, and no significance by NS.

Mouse intraperitoneal glucose tolerance tests

Dual agonist C with confirmed balanced dual activity in both cAMP and GSIS in vitro assays was progressed to in vivo intraperitoneal glucose tolerance tests (ipGTTs) in wild-type and GLP-1R KO mice [12a]. Dual agonist C reduced glucose excursion compared with vehicle control in wild-type mice, suggesting an improved glucose disposal (Figure 4A,B). Due to the unavailability of a GIPR KO mouse, mice were treated with the GIPR antagonist Gipg013 24 h prior to ipGTT to pharmacologically inhibit the GIPR-mediated component of dual agonist activity. Gipg013 effectively inhibits GIP-induced reduction in glucose excursion in mice (data not shown). In wild-type mice pre-treated with Gipg013, the reduction in glucose excursion due to dual agonist C was still apparent, suggesting that this molecule can compensate for GIPR blockade in vivo and exert effects via GLP-1R activation (Figure 4A,B). In GLP-1R KO mice, dual agonist C was able to improve glucose tolerance in the absence of the GLP-1R, suggesting that this molecule can exert effects in vivo via GIPR agonism (Figure 4C,D). In GLP-1R KO mice treated with Gipg013 24 h prior to ipGTT, the effects of dual agonist C were completely reversed, suggesting that this peptide selectively causes glucose excursion via activation of GIPR and GLP-1R (Figure 4C,D). Together, this data indicate that dual agonist C has activity in vivo at both GLP-1R and GIPR as predicted, and that these activities can compensate for the absence of one of the receptors.

In vivo determination of dual-agonist activity.

Figure 4.
In vivo determination of dual-agonist activity.

Mean (±SEM) data for ipGTT performed in wild-type (Glp1r+/+) (A and B) or GLP-1R KO (Glp1r/) (C and D) mice (n = 7–8). Mice were given isotype control NIP228 or 0.7 µmol/kg Gipg013 24 h prior to ipGTT where indicated. Vehicle or 4.4 nmol kg−1 dual agonist C was dosed 2 h prior to ipGTT. AUC, area under the curve (0–120 min). Statistical significance is indicated by *P < 0.05 as determined by one-way ANOVA comparing vehicle with compound, with a Dunnett's post hoc test.

Figure 4.
In vivo determination of dual-agonist activity.

Mean (±SEM) data for ipGTT performed in wild-type (Glp1r+/+) (A and B) or GLP-1R KO (Glp1r/) (C and D) mice (n = 7–8). Mice were given isotype control NIP228 or 0.7 µmol/kg Gipg013 24 h prior to ipGTT where indicated. Vehicle or 4.4 nmol kg−1 dual agonist C was dosed 2 h prior to ipGTT. AUC, area under the curve (0–120 min). Statistical significance is indicated by *P < 0.05 as determined by one-way ANOVA comparing vehicle with compound, with a Dunnett's post hoc test.

Discussion and conclusions

Here, we describe the successful deletion of either the GIPR or the GLP-1R from an INS-1 rat β-cell line using the CRISPR/Cas9 method of genomic editing, and the application of resulting cell lines towards the understanding of dual agonist activities. When generating CRISPR/Cas9-engineered KOs, it is necessary to examine more than one clone to rule out off-target effects. As all of the clones generated in the present study shared a similar phenotype, it is extremely unlikely that this resulted from anything other than specific gene targeting. The efficiency of the CRISPR/Cas9 method on global cell populations has been discussed [19]. In our hands, within selected cell populations, editing of the GIPR resulted in 40% successful KO, 7% of clones remained wild-type and 53% of clones had shifted functional responses that suggested a heterozygous pattern of genomic editing. Editing of GLP-1R was more efficient, resulting in 80% successful KO. For GLP-1R KO, we observed a high number of homozygous clones; the likely explanation for this is inter-allelic gene conversion, with the mutation on one allele being transferred to the second [20]. This is more likely than the same mutation occurring independently on each allele. To generate a functional null, CRISPR/Cas9 targeting is required at both alleles and the indel(s) generated must result in a disruption of the reading frame. This is evidenced by Clone 11 in which both alleles were modified, but only one resulted in an out of frame mutation, thus this clone retained functionality (Supplementary Figure S4).

Successful genomic editing resulted in a permanent and irreversible disruption to the genome. This is beneficial over RNA interference methods that result in only partial and temporary protein knockdown. While we show broadly comparable data through the application of antagonists, genomic editing proves a superior method as it identifies the actual potencies of a dual agonist at each natively expressed receptor, providing understanding of how each component of the dual contributes in an endogenous cell system. This is far more predictive of relative potencies at native receptors in vivo than that can be estimated using transfected cell lines expressing non-native levels of receptor. In addition, the complete functional ablation of receptor activity allowed via CRISPR/Cas9 cannot be achieved by the available competitive antagonists for GLP-1R and GIPR, which also exhibit non-specific effects at related incretin receptors at high concentrations (data not shown). Furthermore, in a high-throughput ‘design-make-test-analyse’ environment, the practical and cost-saving advantages of utilising the KO cell lines over combinations of agonists and antagonists become apparent.

Combinations of peptide hormones have been reported to display synergistic effects in rodents, monkeys and man. PEGylated peptides with activity at both GLP-1R and glucagon receptors (GCGR) normalised glucose tolerance in diet-induced obese mice, as well as decreasing body weight by decreasing food intake and increasing energy expenditure [21]. Simultaneous agonism of GLP-1R and the melanocortin-4 receptor in diet-induced obese mice improved glycaemic control and caused body weight loss [22], co-infusion of PYY3-36 and GLP-1 in healthy overweight males reduced energy intake [23] and combined GLP-1R and type 1 cholecystokinin receptor (CCK1R) agonism has a significantly greater weight lowering effect than either agonist alone [24]. A dual agonist for GIPR and GLP-1R was superior to mono-agonists in glycaemic control and body weight loss in chronic rodent studies [9], and a triple agonist for GIPR, GLP-1R and GCGR was superior to mono-agonist and dual agonist in the reduction of body weight and glycaemic control and additionally improved hepatic steatosis in rodents [25]. Our study demonstrates translation from the in vitro research stage of dual-agonist characterisation to an acute in vivo setting, and the technology we describe could be applied to future dual-agonist projects.

Here, we describe the use of CRISPR/Cas9 genomic editing to characterise the pharmacology of compounds with activity at both GIPR and GLP-1R. The INS-1 KO cell lines allowed physiologically relevant expression levels of each receptor in order to determine the contribution of each to the dual-agonist effect. These experiments revealed that balanced dual agonist C engaged both GLP-1R and GIPR in vitro and in vivo as predicted. In the absence of one receptor, compensation via activity at the remaining receptor was evident. This is because the signal observed is not a summation of signalling through the two receptors, and when signalling pathways overlap, both components must be deleted in order to observe a shift in EC50 curves.

Predictions made in cAMP accumulation assays concerning the balance of dual-agonist activity translated to GSIS, which in turn was predictive of acute in vivo efficacy in ipGTT, where the use of the GLP-1R KO mouse and Gipg013 GIPR antagonist confirmed the involvement of activity through each incretin receptor. This suggests that KO cell lines can produce data that translate across the entire screening cascade. To our knowledge, this is the first example of the use of CRISPR/Cas9 for this purpose, and the technology could be applied to many other co-expressed targets in β cells and other relevant cell lines.

In summary, we have developed structure–activity relationship screening cascades with incretin receptor KOs in a physiologically relevant β-cell line. This allows the characterisation of dual agonists with varying balance of activities at the diabetes-relevant targets, GIPR and GLP-1R. A balanced dual agonist was shown to be efficacious in vivo in an acute study, contributing to improved glucose tolerance via effects at both GIPR and GLP-1R. It remains to be seen which balance of activities will be effective in man, but the in vitro cell potency assays required for measuring the contributing pharmacologies are now in place.

Abbreviations

AUC, area under the curve; CCK1R, type 1 cholecystokinin receptor; CRISPR, clustered regularly interspaced short palindromic repeat; DMSO, dimethyl sulfoxide; FACS, fluorescence-activated cell sorting; FBS, fetal bovine serum; GCGR, glucagon receptor; GFP, green fluorescent protein; GIPR, glucose-dependent insulinotropic polypeptide receptor; GLP-1R, glucagon-like peptide-1 receptor; GSIS, glucose-stimulated insulin secretion; IBMX, 3-isobutyl-1-methylxanthine; ipGTT, intraperitoneal glucose tolerance tests; KO, knockout; PBS, phosphate-buffered saline.

Author Contribution

J.N., A.S. and J.L. performed the research. J.N., A.T.S., D.J.B. and D.C.H. designed the research study. R.H., I.S. and P.R. contributed essential reagents. J.N., A.S. and J.L. analysed the data. J.N., A.T.S., M.P.C., M.R.S. and D.C.H. wrote the paper.

Acknowledgments

We thank John Saunders for assistance with Karyotyping and MedImmune Biological Sciences Unit for technical support.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Baggio
,
L.L.
and
Drucker
,
D.J.
(
2007
)
Biology of incretins: GLP-1 and GIP
.
Gastroenterology
132
,
2131
2157
doi:
2
Wadden
,
T.A.
,
Hollander
,
P.
,
Klein
,
S.
,
Niswender
,
K.
,
Woo
,
V.
,
Hale
,
P.M.
et al. 
(
2013
)
Weight maintenance and additional weight loss with liraglutide after low-calorie-diet-induced weight loss: the SCALE maintenance randomized study
.
Int. J. Obes.
37
,
1443
1451
doi:
3
Nauck
,
M.A.
,
Heimesaat
,
M.M.
,
Orskov
,
C.
,
Holst
,
J.J.
,
Ebert
,
R.
and
Creutzfeldt
,
W.
(
1993
)
Preserved incretin activity of glucagon-like peptide 1 [7-36 amide] but not of synthetic human gastric inhibitory polypeptide in patients with type-2 diabetes mellitus
.
J. Clin. Invest.
91
,
301
307
doi:
4
Althage
,
M.C.
,
Ford
,
E.L.
,
Wang
,
S.
,
Tso
,
P.
,
Polonsky
,
K.S.
and
Wice
,
B.M.
(
2008
)
Targeted ablation of glucose-dependent insulinotropic polypeptide-producing cells in transgenic mice reduces obesity and insulin resistance induced by a high fat diet
.
J. Biol. Chem.
283
,
18365
18376
doi:
5
Højberg
,
P.V.
,
Vilsbøll
,
T.
,
Rabøl
,
R.
,
Knop
,
F.K.
,
Bache
,
M.
,
Krarup
,
T.
et al. 
(
2009
)
Four weeks of near-normalisation of blood glucose improves the insulin response to glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide in patients with type 2 diabetes
.
Diabetologia
52
,
199
207
doi:
6
Meneilly
,
G.S.
,
Bryer-Ash
,
M.
and
Elahi
,
D.
(
1993
)
The effect of glyburide on β-cell sensitivity to glucose-dependent insulinotropic polypeptide
.
Diabetes Care
16
,
110
114
doi:
7
Drucker
,
D.J.
(
2013
)
Incretin action in the pancreas: potential promise, possible perils, and pathological pitfalls
.
Diabetes
62
,
3316
3323
doi:
8
Lupi
,
R.
,
Del Guerra
,
S.
,
D'Aleo
,
V.
,
Boggi
,
U.
,
Filipponi
,
F.
and
Marchetti
,
P.
(
2010
)
The direct effects of GLP-1 and GIP, alone or in combination, on human pancreatic islets
.
Regul. Pept.
165
,
129
132
doi:
9
Finan
,
B.
,
Ma
,
T.
,
Ottaway
,
N.
,
Muller
,
T.D.
,
Habegger
,
K.M.
,
Heppner
,
K.M.
et al. 
(
2013
)
Unimolecular dual incretins maximize metabolic benefits in rodents, monkeys, and humans
.
Sci. Transl. Med.
5
,
209ra151
doi:
10
Garber
,
A.J.
(
2011
)
Long-acting glucagon-like peptide 1 receptor agonists: a review of their efficacy and tolerability
.
Diabetes Care
34
(
Suppl. 2
),
S279
S284
doi:
11
Hohmeier
,
H.E.
,
Mulder
,
H.
,
Chen
,
G.
,
Henkel-Rieger
,
R.
,
Prentki
,
M.
and
Newgard
,
C.B.
(
2000
)
Isolation of INS-1-derived cell lines with robust ATP-sensitive K+ channel-dependent and -independent glucose-stimulated insulin secretion
.
Diabetes
49
,
424
430
doi:
12
Naylor
,
J.
,
Rossi
,
A.
and
Hornigold
,
D.C.
(
2016
)
Acoustic dispensing preserves the potency of therapeutic peptides throughout the entire drug discovery workflow
.
J. Lab. Autom.
21
,
90
96
doi:
12a
Ayala
,
J.E.
,
Bracy
,
D.P.
,
James
,
F.D.
,
Burmeister
,
M.A.
,
Wasserman
,
D.H.
and
Drucker
,
D.J.
(
2010
)
Glucagon-like peptide-1 receptor knockout mice are protected from high-fat diet-induced insulin resistance
.
Endocrinol.
151
,
4678
4687
doi:
13
Ravn
,
P.
,
Madhurantakam
,
C.
,
Kunze
,
S.
,
Matthews
,
E.
,
Priest
,
C.
,
O'Brien
,
S.
et al. 
(
2013
)
Structural and pharmacological characterization of novel potent and selective monoclonal antibody antagonists of glucose-dependent insulinotropic polypeptide receptor
.
J. Biol. Chem.
288
,
19760
19772
doi:
14
Bednarek
,
M.
,
Collinson
,
A.
,
Hornigold
,
D.
,
O'Brien
,
S.
,
Papwroth
,
M.
,
Ravn
,
P.
et al. 
(
2015
)
GIP and GLP-1 receptor dual-agonists for the treatment of diabetes
.
WO 2015/022420
.
15
Ronnebaum
,
S.M.
,
Jensen
,
M.V.
,
Hohmeier
,
H.E.
,
Burgess
,
S.C.
,
Zhou
,
Y.-P.
,
Qian
,
S.
et al. 
(
2008
)
Silencing of cytosolic or mitochondrial isoforms of malic enzyme has no effect on glucose-stimulated insulin secretion from rodent islets
.
J. Biol. Chem.
283
,
28909
28917
doi:
16
Hornigold
,
D.C.
,
Naylor
,
J.
,
Suckow
,
A.T.
,
Bhagroo
,
N.
,
Salari
,
H.
and
Baker
,
D.J.
(
2015
)
Translation of predictive in vitro GIP receptor assays to in vivo acute efficacy
.
Diabetologia
58
,
S1
S607
doi:
17
Mali
,
P.
,
Esvelt
,
K.M.
and
Church
,
G.M.
(
2013
)
Cas9 as a versatile tool for engineering biology
.
Nat. Methods
10
,
957
963
doi:
18
Panjwani
,
N.
,
Mulvihill
,
E.E.
,
Longuet
,
C.
,
Yusta
,
B.
,
Campbell
,
J.E.
,
Brown
,
T.J.
et al. 
(
2013
)
GLP-1 receptor activation indirectly reduces hepatic lipid accumulation but does not attenuate development of atherosclerosis in diabetic male ApoE−/−
mice
.
Endocrinology
154
,
127
139
doi:
19
Ren
,
X.
,
Yang
,
Z.
,
Xu
,
J.
,
Sun
,
J.
,
Mao
,
D.
,
Hu
,
Y.
et al. 
(
2014
)
Enhanced specificity and efficiency of the CRISPR/Cas9 system with optimized sgRNA parameters in Drosophila
.
Cell Rep.
9
,
1151
1162
doi:
20
Li
,
K.
,
Wang
,
G.
,
Andersen
,
T.
,
Zhou
,
P.
and
Pu
,
W.T.
(
2014
)
Optimization of genome engineering approaches with the CRISPR/Cas9 system
.
PLoS ONE
9
,
e105779
doi:
21
Day
,
J.W.
,
Ottaway
,
N.
,
Patterson
,
J.T.
,
Gelfanov
,
V.
,
Smiley
,
D.
,
Gidda
,
J.
et al. 
(
2009
)
A new glucagon and GLP-1 co-agonist eliminates obesity in rodents
.
Nat. Chem. Biol.
5
,
749
757
doi:
22
Clemmensen
,
C.
,
Finan
,
B.
,
Fischer
,
K.
,
Tom
,
R.Z.
,
Legutko
,
B.
,
Sehrer
,
L.
et al. 
(
2015
)
Dual melanocortin-4 receptor and GLP-1 receptor agonism amplifies metabolic benefits in diet-induced obese mice
.
EMBO Mol. Med.
7
,
288
298
doi:
23
Schmidt
,
J.B.
,
Gregersen
,
N.T.
,
Pedersen
,
S.D.
,
Arentoft
,
J.L.
,
Ritz
,
C.
,
Schwartz
,
T.W.
et al. 
(
2014
)
Effects of PYY3-36 and GLP-1 on energy intake, energy expenditure, and appetite in overweight men
.
Am. J. Physiol. Endocrinol. Metab.
306
,
E1248
E1256
doi:
24
Trevaskis
,
J.L.
,
Sun
,
C.
,
Athanacio
,
J.
,
D'Souza
,
L.
,
Samant
,
M.
,
Tatarkiewicz
,
K.
et al. 
(
2015
)
Synergistic metabolic benefits of an exenatide analogue and cholecystokinin in diet-induced obese and leptin-deficient rodents
.
Diabetes Obes. Metab.
17
,
61
73
doi:
25
Finan
,
B.
,
Yang
,
B.
,
Ottaway
,
N.
,
Smiley
,
D.L.
,
Ma
,
T.
,
Clemmensen
,
C.
et al. 
(
2015
)
A rationally designed monomeric peptide triagonist corrects obesity and diabetes in rodents
.
Nat. Med.
21
,
27
36
doi:

Supplementary data