Tissue-Specific Knockout of the Insulin Receptor in Pancreatic b Cells Creates an Insulin Secretory Defect Similar to that in Type 2 Diabetes

Rohit N. Kulkarni, Jens C. Bruning, Jonathon N. Winnay, Catherine Postic, Mark A. Magnuson, and C. Ronald Kahn
Cell, Vol. 96, 329-339, February 5, 1999

    The authors of this paper hope to prove that knockout mice which have had the insulin receptor of their pancreatic b cells genetically removed will have metabolic functions similar to a type 2 diabetic.  This includes the loss of insulin secretion in response to increased levels of glucose but not arginine and the inability to tolerate glucose with increased age.  By examining the data, the figures and tables, I will determine if indeed the authors' results support their claims.
    Figure 1a shows the wildtype allele and the targeting vector used in the homologous recombination.  The gene that was placed into embryonic stem cells is labeled homologous recombination and produced mice that after mating with the C57Bl/6J strain produced IRlox animals.  These mice had exon 4 flanked by loxP sites in one of their two alleles.  IRlox mice were then mated with transgenic mice that had both IRlox alleles and the Cre transgene.  The Cre gene removed exon 4 in a type II deletion with a genetic ratio in the offspring to produce the desired knockout mice known as BIRKO (B cell-specific insulin receptor knockout) mice.  Figure 2b shows pancreas cells from the Cre expressing mice labeled by immunofluorescence to show that indeed Cre was present in the B cells.
    Figure 2 sets out to prove that the knockout did occur in the BIRKO mice and compares them to the other offspring of their litter, wildtype, IRLox mice containing lox sites, and Cre mice that have the Cre transgene.  Figure 2a first compares the IRLox allele and the IRKO allele schematically which shows the loss of exon 4 in the knockout allele and the location of PCR primers.  In the bottom panel of 2a is a gel that has PCR DNA loaded.  Using primers 2 and 3 in PCR produces a 300 bp fragment in the IRLox allele or a 220 bp fragment in IRKO alleles.  They took 5, 10, and 25 islets from BIRKO mice and homozygous IRLox mice.  The BIRKO mice had the expected 220 bp fragment in all quantities of islet cells.  In the 10 and 25 islet samples they also contained the 300 bp fragment.  The authors assert that this is due to the small percentage of non B cells, which will therefore have a normal insulin receptor,  in islets.  They also loaded liver and muscle cells from BIRKO mice as a negative control to prove that the knockout was site specific.  IRLox islets and liver and muscle cells all had only the 300 bp fragment as expected.  Figure 2b uses reverse transcriptase PCR to show the mRNA that is present in the islets in two BIRKO mice and a IRLox mice.  All three show a 480 bp fragment that is normal insulin receptor mRNA.  This can again be explained in the BIRKO mice by the fact that non B cells are present in a small percentage in islets.  Only the BIRKO mice contain the 220 bp fragment that is the mRNA from the knocked out gene.  It appears that B-actin was used as a positive control but the fact that the two gels have been cut and put next to each in this figure is a tiny bit unsettling.  Figure 2c is needed to prove the authors claim that islets contain B cells and non B cells and that it is the non B cells that are producing the normal insulin receptor.  Using flow cytometry the non B cells were separated from the B cells in BIRKO mice anc compared with the B cells of IRLox mice.  The IRLox mice and the BIRKO non B cells from the PCR of their DNA had only the larger, IRLox allele present while the BIRKO B cells had only the smaller, IRKO allele.  This is a very nice figure and was needed to back up the claim that it is the non B cells of BIRKO islets that have the normal insulin receptor and not the B cell.  Figure 2d is a western blot of wildtype, IRLox, Cre, and BIRKO muscle, liver, heart, and brain tissue using a anti-insulin receptor antibody as the probe.  There is a fragment in every lane showing that indeed the knockout is site specific and other cells are not affected.  It also shows that IRLox which has been used as the control so far is indeed the same as wild type and therefore a good control.  We have to trust the authors on this one since there is no means of sizing the bands, but I am comfortable with this.
    Figure 3 shows the effects of overnight fasting in 6 month old male and female mice with the BIRKO mice and the control mice of wildtype, IRLox, and Cre.  BIRKO males and females have normal glucose levels after fasting but BIRKO males have double the insulin levels compared to the control mice.  Female BIRKO mice have only a slightly elevated level of insulin.  This is the first figure to demonstrate that knocking out the insulin receptor has an effect on insulin levels in the mice when an increase in glucose is present.
    Figure 4 shows the effects of injecting glucose (a,b) and arginine (c,d) in 3 to 4 month old male and female mice.  The BIRKO male and female mice show little to no response to the sudden rise in glucose unlike the control mice which have a large jump in insulin during the first few minutes.  Arginine produces the same increase in insulin in all mice.  Figure a and b also demonstrates a loss in normal function in the BIRKO mice.  This time it is a loss of acute phase insulin secretion.
    Figure 5 shows the abilty of the mice to dispose of an increase in glucose as they get older.  Male and female mice were tested at 2, 4, and 6 months.  Both the BIRKO males and females were slower in disposing of glucose compared with the control mice.  It is also evident from Figure 5 that the BIRKO mice became worse at metabolizing glucose as they got older.
    Figure 6 compares islet size and pancreatic insulin levels in BIRKO and control mice.  At 2 months BIRKO and control mice have similar morphology in islets as shown in Figure 2a.  At 4 months however the BIRKO mice islets are noticeably smaller than the control mice islets.  To compare the islet size with the insulin production the BIRKO and control mice had the insulin levels measured also at 2 and 4 months.  Not surprisingly levels were the same at 2 months.  At 4 months the BIRKO mice had similar insulin levels to 2 month olds and were approximately 35% lower than control mice.  This decrease in insulin level can partly be explained by a decrease in islet size but the authors argue that there must also be other reasons to account for this decrease.
    Figure 7 shows that islet size may not only be responsible for decreased insulin levels in 4 month old BIRKO mice.  7a compares the amount of insulin granules in BIRKO and IRLox B cells at 4 months of age.  The amount of insulin and overall morphology is similar in both.  7b are pancreatic tissue sample from IRLox, BIRKO, and Ob/Ob (which is not explained) mice that are immunofluorescently stained to show the presence of glucose in the B cells.  The BIRKO have a little less staining than the IRLox tissue but much more than the speckled Ob/Ob tissue.  This figure demonstrates that the same amount of insulin and glucose is available to the BIRKO cells, which helps support their claim that lower insulin levels can't be only attributed to decreased islet size.
    I thought that this was a very good paper.  Any claims that at first seemed shaky, non B cells were totally responsible for normal insulin receptors in BIRKO islets and decreased insulin levels in the pancreas were not due to only decreased islet size, were subsequently addressed in later figures and shown to follow from the data.  I think that the authors results allow for the conclusion that the removal of normal insulin receptors in B cells produces a loss of insulin secretion towards glucose and a loss of glucose tolerance as these mice age.
    Future experiments that I would be interested in seeing include testing the hypothesis that insulin is responsible for regulating gene transcription of the insulin receptor forming a positive feedback loop that was obviously missing in BIRKO mice.  I would fluorescently label insulin and use it as the probe in FISH.  Since we know where the insulin receptor gene is we can look to see if insulin does in fact bind to this area of the DNA in which would suggest that insulin does play a role in DNA transcription of the insulin receptor.  Another option for identifying DNA-protein binding is using a band shift assay.  If insulin binds to the promoter region of the insulin receptor the band will not travel as far as just the DNA which demonstrates a protein-DNA interaction.  The results I would expect depend on the feasibility of labeling insulin without interfering with a possible insulin-DNA epitope and the normal presence of insulin in DNA.  If this would work then I would expect for FISH to show insulin binding to the insulin receptor gene and possible other loci that could prove to be in the insulin pathway.  The band shift assay would also show a band shift for this region of DNA.  There also might be a difference among tissue types in the binding of insulin with more insulin binding in the DNA of B cells.
    Another experiment I feel would be useful is to determine the reason that male and femal BIRKO mice have differences in their response to glucose loads and insulin excretion (Figures 3e and f, 4a and b, and 5e and f).  I would test the differences in islet morphology and B cell morphology to see if there were any differences that couldn't be accounted for just differences in size.  If any differences are found then it might suggest a difference in receptors or insulin pathway.  I don't think that this will actually show much but it can't hurt.
    I would also like to test the effects of arginine on the insulin pathway.  What is it binding to in BIRKO mice that allows it to produce normal insulin levels?  Simple immunofluorescence labeling of arginine should allow for where it binds in the cells and using western blots would allow for the isolation of the receptor that it binds to and the characterization of this receptor.  I would also like to see the chemical structure of arginine be mutated to show the differences in insulin production.  This would also provide more information on the receptor of the arginine.  Learning more about the arginine receptor could help learn more about the insulin receptor and possible key epitopes that could be modified in diabetics to produce normal functioning.

Return to Bill's Molecular Biology Home Page

© Copyright 2000 Department of Biology, Davidson College, Davidson, NC 28036

Send comments, questions, and suggestions to: bimonroe@davidson.edu