Essential roles of aspartate aminotransferase 1 and vesicular glutamate transporters in β-cell glutamate signaling for incretin-induced insulin secretion

Incretins (GLP-1 and GIP) potentiate insulin secretion through cAMP signaling in pancreatic β-cells in a glucose-dependent manner. We recently proposed a mechanistic model of incretin-induced insulin secretion (IIIS) that requires two critical processes: 1) generation of cytosolic glutamate through the malate-aspartate (MA) shuttle in glucose metabolism and 2) glutamate transport into insulin granules by cAMP signaling to promote insulin granule exocytosis. To directly prove the model, we have established and characterized CRISPR/Cas9-engineered clonal mouse β-cell lines deficient for the genes critical in these two processes: aspartate aminotransferase 1 (AST1, gene symbol Got1), a key enzyme in the MA shuttle, which generates cytosolic glutamate, and the vesicular glutamate transporters (VGLUT1, VGLUT2, and VGLUT3, gene symbol Slc17a7, Slc17a6, and Slc17a8, respectively), which participate in glutamate transport into secretory vesicles. Got1 knockout (KO) β-cell lines were defective in cytosolic glutamate production from glucose and showed impaired IIIS. Unexpectedly, different from the previous finding that global Slc17a7 KO mice exhibited impaired IIIS from pancreatic islets, β-cell specific Slc17a7 KO mice showed no significant impairment in IIIS, as assessed by pancreas perfusion experiment. Single Slc17a7 KO β-cell lines also retained IIIS, probably due to compensatory upregulation of Slc17a6. Interestingly, triple KO of Slc17a7, Slc17a6, and Slc17a8 diminished IIIS, which was rescued by exogenously introduced wild-type Slc17a7 or Slc17a6 genes. The present study provides direct evidence for the essential roles of AST1 and VGLUTs in β-cell glutamate signaling for IIIS and also shows the usefulness of the CRISPR/Cas9 system for studying β-cells by simultaneous disruption of multiple genes.

Introduction Insulin, a key hormone regulating glucose metabolism, is secreted from pancreatic β-cells to maintain blood glucose levels within a normal range. Glucose is physiologically the most important substance that regulates insulin secretion. Glucose-induced insulin secretion (GIIS) comprises two pathways: the triggering pathway and the metabolic amplifying pathway [1]. The former involves closure of the ATP-sensitive K + channels by increased ATP concentration due to glucose metabolism, which is followed by Ca 2+ influx through voltage-dependent Ca 2+ channels leading to insulin granule exocytosis. The latter involves amplification of the effects of the triggering pathway by metabolic signals other than ATP evoked by glucose metabolism, but the mechanism remains poorly understood. In addition to glucose, neuro-hormonal amplification is critical in normal regulation of insulin secretion, as reviewed in detail by Yokoi et al. [2]. Hormones and neurotransmitters exert their effects on insulin secretion mainly through G-protein coupled receptor-mediated signals such as cAMP, diacylglycerol, and inositol 1,4,5-triphosphate. The incretins glucagon-like peptide 1 (GLP-1) and glucosedependent insulinotropic polypeptide (GIP), which are secreted from enteroendocrine cells in response to meal ingestion, amplify insulin secretion through cAMP signaling in pancreatic βcells [3,4]. Although the effects of incretins in insulin secretion are glucose concentrationdependent [5][6][7], the interaction between glucose metabolism and incretin/cAMP signaling was not known. We recently found β-cell glutamate acts as a critical signal in incretin-induced insulin secretion (IIIS) and proposed a mechanistic model of β-cell glutamate signaling in IIIS, which requires both production of cytosolic glutamate through the malate-aspartate (MA) shuttle linked to glycolysis and transport of cytosolic glutamate into insulin granules via cAMP signaling [2,8]. However, this model was based primarily on experiments using gene knockdown and/or pharmacological inhibition of aspartate aminotransferase 1 (AST1), an enzyme in the MA shuttle that generates cytosolic glutamate by transferring amino residue from aspartate to α-ketoglutarate, and vesicular glutamate transporter VGLUT1, which mediates glutamate transport into insulin granules [9][10][11][12].
The recently developed RNA-guided CRISPR (clustered regularly interspaced short palindromic repeat)/Cas9 genomic editing system has made it possible to selectively disrupt any gene of interest [13]. This technology is now widely used for simple and rapid establishment of gene deficient animals and cell lines. However, there are few reports on its application to pancreatic β-cell lines [14]. In the present study, we applied CRISPR/Cas9 technology to directly prove our model of IIIS by establishing and characterizing CRISPR/Cas9-engineered clonal mouse β-cell lines deficient for the genes involved in glutamate signaling: AST1 (Got1), VGLUT1 (Slc17a7), VGLUT2 (Slc17a6), and VGLUT3 (Slc17a8).

Materials and methods
Cell lines MIN6-K8, a mouse pancreatic β-cell line that secretes insulin in response to both glucose and incretins, was established from the IT6 mice as described previously [15].

Cell culture
MIN6-K8 and gene deficient β-cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% heat-inactivated fetal bovine serum and maintained in a humidified incubator with 95% air and 5% CO 2 at 37˚C.

CRISPR/Cas9 transfection and clonal isolation of cells
The Cas9 D10A nickase expression vector and sgRNA vectors containing hygromycin resistant markers were purchased from GeneCopoeia (Rockville, MD, USA). MIN6-K8 cells were reverse-transfected with a combination of the Cas9 nickase expression vector and a pair of L/ R-sgRNA vectors (0.133 μg/mL for each) using Lipofectamine 2000 transfection reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instruction. After 48 hours of incubation, 400 μg/mL hygromycin B was added for selection. After 4 days of incubation, cells were re-plated to recover for 11 days. For clonal isolation, cells were serially diluted in 10 cm-dishes to form single colonies, which were then manually picked under microscopy and isolated in a 48 well-plate.

Sequencing
Genomic DNA was extracted from each colony using SimplePrep reagent for DNA (Takara Bio, Otsu, Japan) according to the manufacturer's instruction. Cas9 target regions were amplified by PCR and the PCR products were subjected to agarose gel electrophoresis. The PCR products showing different length from the wildtype allele were cloned into pMD20-T blunt vector (Takara Bio, Otsu, Japan) and analyzed by Sanger sequencing using 3100-Avant Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
Insulin secretion from cultured β-cell lines MIN6-K8 and gene knockout β-cell lines were preincubated for 30 min in HEPES-balanced Krebs-Ringer bicarbonate buffer containing 0.1% BSA (H-KRB) with 2.8 mM glucose, then incubated for 30 min in H-KRB with 2.8 mM glucose, 16.7 mM glucose, 16.7 mM glucose plus GLP-1 or GIP, or 16.7 mM glucose plus dimethyl glutamate (dmGlu). Insulin released in the incubation buffer and cellular insulin content were measured by homogeneous time-resolved fluorescence (HTRF) assay using insulin assay kits from CIS Bio international (Gif sur Yvette, France). The amounts of insulin secretion were normalized by the cellular insulin content determined after extraction by 0.1% Triton-X.

Measurement of glutamate content
MIN6-K8 and Got KO β-cell lines were preincubated for 60 min in H-KRB with 2.8 mM glucose and then incubated for another 30 min in H-KRB with [U-13 C]-glucose (2.8 mM or 16.7 mM). Extraction and quantification of cytosolic glutamate contents was performed as previously described [8]. Glutamate contents were normalized by protein contents determined by BCA protein assay kit (Thermo Scientific,Waltham, MA).

Mice
The Slc17a7-floxed mice with a floxed allele, in which exon 2 of Slc17a7 was sandwiched by loxP, were established. The Slc17a7-floxed mice were then crossbred with Ins-Cre mice that express β-cell specific Cre [16] to produce β-cell specific Slc17a7 knockout (βSlc17a7 −/− ) mice. The Slc17a7 flox/flox mice were used as control. All animals were maintained under specific pathogen free conditions at 23 ± 2˚C and 55 ± 10% relative humidity with a 12-h light-dark cycle, and were provided with water and a commercial diet CE-2 (CLEA Japan, Inc., Tokyo, Japan) at the Animal Facility of Kobe Biotechnology Research and Human Resource Development Center of Kobe University. At the end of the experiments, animals were sacrificed by cervical dislocation or overdose of anesthesia with pentobarbital sodium. All animal experiments were approved by the Committee on Animal Experimentation of Kobe University, and carried out in accordance with the Guidelines for Animal Experimentation at Kobe University. The Slc17a7-floxed mice are available from the RIKEN BRC (http://mus.brc.riken.jp/, RBRC09669).

Pancreas perfusion experiment
Pancreas perfusion experiments were performed as previously described [19]. Overnight (16 h) fasted male mice at 17-18 weeks of age were used. Briefly, the perfusion protocol began with a 20-min equilibration period with the same buffer used in the initial step shown in the figures. The flow rate of the perfusate was 1 mL/min. The insulin levels in the perfusate were measured by using the HTRF assay as described above.

Statistical analysis
The results are presented as mean ± SEM. Differences among the groups were analyzed with Dunnett's method or Welch's t-test as indicated in the legends. p < 0.05 was regarded as statistically significant.

Results and discussion
Establishment and characterization of AST1 (Got1) knockout β-cell lines We first focused on AST1 (gene symbol Got1), the enzyme in the MA shuttle that produces cytosolic glutamate in β-cells. We applied the CRISPR/Cas9 nickase system [20], a modification of the CRISPR/Cas9 system with improved specificity, for disrupting Got1 in MIN6-K8. In the CRISPR/Cas9 nickase system, a combination of Cas9 nickase and a pair of sgRNAs induces nicks on both DNA chains to mediate double-stranded breaks (DSB), which is repaired by nonhomologous end joining (NHEJ) with resultant indels [20]. Targeting the genomic locus by two sgRNAs designed for each DNA chain can minimize off-target cleavage [21]. To avoid functional truncated products, the target sequence was set within the first exon. The MIN6-K8 β-cell line was transfected with three plasmids expressing Cas9 nickase and a pair of sgRNAs. Cells were selected with hygromycin and cultured in a 10 cm plate in a very low concentration to form single colonies, the each of which was then isolated and cultured separately to yield different colonies.
We next confirmed mutations in the target region of isolated colonies. First, we amplified the Cas9 target region of each colony by PCR and examined the length of the PCR product for differences from wild-type (WT) allele. To verify the mutation on each allele separately, the PCR products were cloned into pMD20-T vector, and multiple transformants for each colony were sequenced. As a result, two different Got1 knockout (KO) β-cell lines, A60 and A64, were established, both of which were compound heterozygous mutants having frameshifts on both alleles (Fig 1A). Protein expression of AST1 was completely abolished in both KO cell lines ( Fig 1B). To rule out contamination by WT cell line, the absence of WT mRNA was confirmed by PCR of cDNA fragments harboring the Cas9 target region (S1 Fig). qRT-PCR analysis using a Taqman probe targeting 3' region of the gene showed that the expression level of Got1 was significantly down-regulated in these KO cell lines (S2 Fig).
To learn if cytosolic glutamate production is impaired by Got1 deficiency, we measured cytosolic glutamate contents by metabolic flux analysis using stable isotope-labelled [U-13 C]glucose. M and M+1 glutamate isotopomers are both naturally existing glutamate in cells, while M+2, M+3, M+4, and M+5 glutamate isotopomers are synthesized de novo from [U-13 C]-glucose. We found that M+2 to M+5 glutamate isotopomers were significantly increased by glucose stimulation in WT cell line, whereas they were not increased in Got1 KO cell line (Fig 1C), indicating that AST1 is essential for cytosolic glutamate production from glucose.
We then examined insulin secretory response in Got1 KO cell lines. It is known that insulin secretory ability varies among different β-cell lines. Since we found that the amount of released insulin varied between Got1 KO (A60 and A64) cell lines established (S3 Fig), incretin responsiveness was evaluated by expressing the insulin secretion data as fold increase relative to the amount of insulin secretion at 16.7 mM glucose. Both Got1 KO cell lines exhibited a significant impairment in insulin secretion in response to both GLP-1 and GIP, as compared to WT MIN6-K8 cell line (Fig 1D and S3 Fig). GIIS was only mildly impaired in Got1 KO cell lines (S3 Fig) To verify that AST1 is required for IIIS, we next tried to rescue the diminished AST1 activity by introducing WT Got1 into Got1 KO cell line. Due to low transfection efficiency in MIN6-K8 β-cell line, we performed co-transfection assay with WT Got1 and human preproinsulin cDNA [22]. Proinsulin is converted into insulin and C-peptide during the secretory process. Since antibody against human C-peptide does not cross-react with endogenous mouse Cpeptide, the released human C-peptide can be monitored as an indicator of insulin secretion from the transfected cells. Transfection of Got1 KO (A60) cell line with WT Got1 significantly restored incretin-induced C-peptide secretion (Fig 1E). We also examined the effect of dimethyl glutamate (dmGlu), a membrane permeable glutamate precursor [23] that is converted to glutamate in insulin granules as well as in cytosol [8]. In Got1 KO (A60) cell line, dmGlu amplified insulin secretion in a dose-dependent manner (Fig 1F), indicating that dmGlu complements glutamate deficiency in the Got1 KO cell line and that cellular glutamate acts as an amplifying signal in insulin secretion. A  G  G  --------------------------------------------------------------------------- Our previous study showed that knockdown of VGLUT1 (Slc17a7) or pharmacological inhibition of glutamate transport reduced insulin secretion in response to incretins while knockdown of VGLUT2 (Slc17a6) did not [8]. In addition, there was a significant impairment in IIIS from pancreatic islets of global Slc17a7 KO mice at pre-weaning two weeks of age [8]. Based on these findings, we initially anticipated that single KO of Slc17a7 would be enough to cause a severe defect in IIIS.
To further ascertain whether VGLUT1 is required for IIIS, we generated β-cell specific Slc17a7 knockout (βSlc17a7 −/− ) mice by using the Cre-loxP system (S4 Fig). The mice showed normal development and there was no gross abnormality. Immunohistochemical staining confirmed that expression of VGLUT1 protein was successfully ablated in pancreatic β-cells of βSlc17a7 −/− mice (Fig 2A). There were no significant differences in glucose tolerance, meal tolerance, and insulin sensitivity between βSlc17a7 −/− mice and Slc17a7 flox/flox mice (control) (S1 Method and S5 Fig). Unexpectedly, different from our previous finding [8], we found no significant impairment of insulin secretion in response to GLP-1 in βSlc17a7 −/− mice, as assessed by pancreas perfusion experiment (Fig 2B). RT-PCR analyses of pancreatic islets revealed that although expression level of Slc17a7 in βSlc17a7 −/− mice was significantly decreased, there were no significant differences in expression of both Slc17a6 and VGLUT3 (Slc17a8) between βSlc17a7 −/− mice and Slc17a7 flox/flox mice (Fig 2C). VGLUT3 was reported to be present in membranes of insulin-containing secretory granules [24]. It is possible, therefore, that the expression of Slc17a6 and/or Slc17a8 compensates for function of glutamate transport into granules in βSlc17a7 −/− mice. The discrepancy in the phenotype between global Slc17a7 KO mice and β-cell specific Slc17a7 KO mice might be due to the difference in the age of mice: the former were examined at pre-weaning state since the mice could not survive thereafter [25], while the latter were examined at adult state when Slc17a7 deficiency could be compensated by Slc17a6 and/or Slc17a8.
In parallel with generation of βSlc17a7 −/− mice, we also established Slc17a7 single KO (SKO) β-cell lines. In WT MIN6-K8 cell line, Slc17a7 and Slc17a6 are expressed at similar levels and Slc17a8 is not expressed (S6 Fig). To avoid functional truncated products, exon 2, which encodes the first transmembrane domain of VGLUT1, was Cas9-targeted. Two different SKO cell lines, 34 and 39, were established (Fig 3A). Allele 2 of both clones was not detected by initial PCR-sequencing of genomic DNA, probably due to the large deletions involving the primer sequence. We therefore analyzed the cDNA synthesized from the SKO cell lines 34 and 39. Two different products were amplified by PCR using the cDNA obtained from the cell lines as templates: one corresponding to allele 1 and the other corresponding to allele 2, in which exon 2 is skipped by abnormal splicing due to the large deletions spanning intron 1. The absence of WT Slc17a7 mRNA was confirmed in both cell lines (S7 Fig). These analyses showed that the SKO cell lines 34 and 39 both are compound heterozygous mutants having frameshifts on both alleles of Slc17a7. mRNA expression of Slc17a7 measured by qPCR was significantly decreased in both SKO cell lines (S8 Fig) and expression of VGLUT1 protein was successfully ablated in the SKO cell line 39, as assessed by immunofluorescence staining (Fig  4A). In contrast, we found that Slc17a6 was upregulated in SKO cell lines at mRNA and protein levels (Figs 3B and 4B). Expression of Slc17a8 was not detected. We then examined IIIS in SKO cell lines. Both SKO cell lines retained insulin secretory response to GLP-1 to some extent (Fig 3C and S9 Fig), suggesting that upregulation of VGLUT2 partly compensates for the loss of function of VGLUT1, as was found in βSlc17a7 −/− mice. Taken together, these findings indicate that VGLUT2 and/or VGLUT3 can compensate for the loss of function of VGLUT1 in IIIS from β-cells. Switching of one VGLUT isoform to another by transcriptional and/or translational regulation was reported in developing mouse cerebellum [26]. Such regulation might also exist in β-cells.

A B C
WT Slc17a7 exon 2  Roles of AST1 and VGLUTs in incretin-induced insulin secretion line was further transfected with plasmids expressing Cas9 nickase and two pairs of sgRNA each targeting Slc17a6 and Slc17a8, followed by clonal isolation. After confirmation of deficiency of Slc17a6 in these colonies by sequencing, the Cas9 target region of Slc17a8 of these colonies were examined. Finally, three TKO cell lines, V22, V39, and V61, were established. As for Slc17a6, all of these cell lines were compound heterozygous mutants harboring frameshifts on both alleles (S10 Fig). Absence of WT Slc17a6 mRNA and VGLUT2 protein was confirmed by PCR of cDNA and immunofluorescence staining, respectively (S11 Fig and Fig 4B). As for Slc17a8, both alleles of the TKO cell line V39 and a single allele of the TKO cell line V61 were found to have frameshifts (S10 Fig). Although a single allele of the TKO cell line V22 was found to be in frame, a deletion of 29 amino acids around the first transmembrane domain probably disrupts gene function. Other alleles were not detected by PCR, probably due to the large deletions. We then examined insulin secretory response in these TKO cell lines. GIIS was retained, but IIIS were significantly impaired in all of the three TKO cell lines (Fig 5A and S12  Fig). Exogenous introduction of WT Slc17a7 or Slc17a6 into the TKO cell lines significantly rescued incretin-induced C-peptide secretion (Fig 5B and 5C). This finding indicates that VGLUT1 and VGLUT2 functionally compensate each other. We also found that dmGlu markedly amplified insulin secretion in the TKO cell lines (Fig 5D). These findings provide direct evidence that glutamate transport into insulin granules through VGLUTs is essential for IIIS.

Conclusion
The present study of CRISPR/Cas9-engineered mouse pancreatic β-cell lines provides direct evidence that AST1 and VGLUTs are essential for β-cell glutamate signaling in IIIS. This study is the first to show successful and complete disruption of multiple genes in pancreatic β-cell lines by the CRISPR/Cas9 nickase system. As a family of genes encoding multiple isoforms shares similar functions in general, disruption of one isoform may induce expression of the other to compensate for its function. It is therefore necessary to disrupt multiple target genes simultaneously at cell level to elucidate their functions. The CRISPR/Cas9 system is a useful and straightforward method to disrupt multiple genes simultaneously in pancreatic β-cell lines to elucidate their roles in cellular functions.

S1 Method. Oral glucose tolerance test (OGTT), meal tolerance test (MTT), and intraperitoneal insulin tolerance test (IPITT).
For oral glucose tolerance test, glucose (1.5 g/kg) was administered to mice after 6 hours of fasting. For meal tolerance test, 10 μL/g (1.5 g carbohydrate/kg body weight) Twinline (Otsuka Pharmaceuticals, Tokushima, Japan) was administered orally to mice after 6 hours of fasting. For intraperitoneal insulin tolerance test, insulin Roles of AST1 and VGLUTs in incretin-induced insulin secretion (0.5 U Humulin R/kg body weight, Eli Lilly Japan K.K., Kobe, Japan) was administered to mice after 6 hours of fasting . Blood glucose levels were measured by a portable glucose meter  (ANTSENSE III,