Kulkarni, Bruning, Winnay, Postic, Magnuson, Kahn. (1999) Tissue-SpecificKnockout of the Insulin Receptor in Pancreatic Beta-Cells Creates an InsulinSecretory Defect Similar to that in Type 2 Diabetes.  Cell. 96:  329-339.

Summary
    The paper seeks to determine whether pancreatic ß-cells with inactive insulin receptors are responsible for the pathogenesis of type 2 diabetes.  Specifically, type 2 diabetes is characterized by progressive glucose resistance over time that results in an accumulation of glucose in a patient's body.  Furthermore, it has been shown that type 2 diabetes causes reduced insulin secretion in the presence of glucose.  The researcher's found that deletion of the insulin receptor in pancreatic ß-cells resulted in both an increase in glucose resistance and a decrease in glucose induced insulin secretion, supporting their hypothesis that a malfunctioning insulin receptor plays an important role in type 2 diabetes.
Introduction
    Type 2 diabetes is characterized by a lack of insulin production in muscle, fat, and liver cells, and the relative failure of the pancreatic beta-cell.  In muscle, fat, and liver cells insulin resistance has been shown to occur due to "down regulation of the [insulin] receptor, a decrease in receptor kinase activity, decreases in phosphorylation of receptor substrates, and defects in glucose transporter translocation."  However, the ß-cell is more complex.  Though it fails to release insulin in the presence of glucose, it still releases insulin in the presence of another amino acid, arginine.
    The researchers used the same method from their1998 paper to determine the role insulin receptors play in ß-cells. They found that deletions of the insulin receptor result in a loss of first-phase insulin secretion induced by glucose, but not arginine, and an decrease in tolerance to glucose over a period of 6 months (experimental mice metabolize glucose less effectively than control mice).  Thus, the researchers concluded that impaired ß-cell funtioning could result in a decrease in insulin production like that seen in type 2 diabetes.
Results
    The researchers used the Cre-loxP method to produce mice homozygous for the ß-cell deletion.  Figure 1a. shows, from top to bottom, the wt allele, the targetting vector with the selection cassette and exon 4 flanked by loxP sites, a homologous recombinant with the selection cassette still present, and a homologous recombinant with the selection cassette removed.  The selection cassette is important because it contains the herpes simplex virus-thymidine kinase gene segment that induces death in the presence of ganciclovir, thereby preventing a so-called type III deletion where the selection cassette remains but the gene is deleted.
    The researchers transfected embryonic stem cells with the targetting vector and then exposed the cells to the antibiotic neomycin in order to kill any nonrecombinant cells.  The resulting cells were transiently transfected with a Cre cDNA containing plasmid to remove the selection cassette, and then treated with ganciclovir in order to kill any type III deletions.  The new cells contained either the wt and type II deletion or 2 wt alleles. They were injected into mouse blastocysts to generate a chimeric mouse.  In order to determine which of these mice contained the recombinant gene in their germ cells the mice were bred to produce 100% heterozygous individuals.  Finally, these mice were bred with IRLox mice carrying a Cre transgene in order to make homozygous beta-IRKO mice expressing Cre in pancreatic beta cells. The breeding generated three control groups: homozygous IRlox mice that were used to determine the effect of loxP sites on insulin production in the insulin receptor gene, wt controls, and mice carrying the Cre transgene that were used to determine the effect of Cre expression on ß cell function.  All recombinant mice were fertile and healthy.
      Figure 1b shows an immunoflourescent histochemical analysis of pancreaticB-cells in a mouse with the Rip-Cre transgene. In it, the Cre is stained red and the beta-cells are stained green.
    Figure 2 assesses the efficiency of recombination of the IRlox allele. Figure2a, from top to bottom, shows the non-deleted form of the allele (called IRLox) that contains a segment in exon 4 that primer 1 recognizes as well as the conserved segment that primer 2 recognizes, the ßIRKO allele that has exon 4 deleted and uses primers 3 and 2, and a PCR analysis gel of mice liver, muscle. and pancreatic ß-cells. In the PCR gel, two bands can be seen, a 300 bp band that corresponds to the non-deleted IRLox allele, and a 220bp band that corresponds to the ß-IRKO allele. The 220 bp band was found only in the ßIRKO islets, which is to be expected, and the 300 bp band was found exclusively in the IRLox islets, with one exception. In the 25 islet sample of ßIRKO cells a faint 300 bp band can be seen. The researchers address this by stating that ß cells make up only 80 to 85 % of islet cells. Thus, in the larger sample size there is a greater chance that some of the cells would be non-ß cells, would not produce Cre, and so would contain the non-deleted form of the allele. Also, since the muscle and liver cell lanes in both IRLox and ßIRKO mice contain only the 300 bp band, the muscle and liver cells in both types of mice must only contain the IRLox allele, which is what one would expect if these cells were not producing Cre. Predictably, as the number of islets sampled increases the intensity of the band also increases, which is to be expected as more islets means more allele present.
    In figure 2c., FACS was used to separate non-ß and ß cells from ßIRKO islets, and PCR analysis was performed in order to show that the single 300 bp band in the 25 islet lane of Figure 2a. is due to non-ß cells. The researchers support their claim as three bands are present on the gel, a 220 bp band corresponding to the ßIRKO allele in ß-cell extracts from ßIRKO cells, and two 300 bp bands in non-ß ßIRKO cells and IRLox ß-cells that correspond to the undeleted allele.
    Figure 2b. is a RT-PCR gel that shows that two different islets of ßIRKO expressing both functional and nonfunctional insulin receptor mRNA, while IRLox only expresses functional mRNA. This is evident due to the single, very intense 480 bp band in the IRLox allele, and the two bands, 480 and 220 bp respectively, in both ßIRKO cases. These results are consistent with the authors statements that ß-islets are only 80-85% ß-cells and 15-20% non-ß cell. The bottom gel showing ß-actin expression is a positive control for the total amount of RNA present in each type of mouse, as ß-actin should be present in equal amounts in both ßIRKO and IRLox mice.
    Finally, figure 2d is a western blot that shows that the allele is still expressed in brain, muscle, liver, and heart tissue and in relativley the same amount as in wild type individuals. Thus, neither Cre expression nor the presence of LoxP sites interfered with expression of the insulin receptor in non-ßIRKO mice.
    The four types of mice were separated into two groups: fasted and randomfed. Figures 3a and 3b show the difference in glucose levels of both males and females in the four types of mice. No significant difference in glucose concentration was observed for either males or females in either of the four randomfed or fasted groups.  This means that, even in ßIRKO cells, glucose is somehow exiting the bloodstream.  Otherwise, one would expect to see an increase in the amount of glucose present in ßIRKO cells.  Though there was no significant difference in glucose levels, fasted ßIRKO mice do exhibit mildly elevated insulin levels as compared to the other three groups. This effect is most prominent in males, whose insulin levels were nearly two-fold higher than the corresponding levels of the other three groups. This is probably due to the fact that, since insulin secretion is mediated by the presence of glucose, animals with functioning insulin receptors would have lower levels of insulin in the absence of glucose (after fasting), while animals whose insulin concentration is independent of glucose concentration would show a higher level of insulin present when glucose is absent. In animals with nonfunctional insulin receptors, insulin is not reuptaken in the absence of glucose. Still, the data was not statistically significant.
    Figure 4 shows the insulin response versus time in the four types of mice after being challenged by glucose (figures a and b) or arginine (figures c and d). While the response to arginine was virtually the same for all four types of mouse, both male and female, the ßIRKO mice produced significantly less Insulin than the other three groups. Males produced virtually no insulin in response to a glucose challenge while females produced 85% less insulin than the three other female types. However, after thirty minutes there was no significant difference in the amount of insulin present in any of the groups, leading the researchers to suspect that the primary (acute) response to glucose is damaged while the secondary response to glucose and the primary response to arginine are intact in ßIRKO mice.
    Figure 5 shows that ßIRKO mice have impaired glucose tolerance that increases with age. Animals were fasted overnight, injected with glucose,and then the glucose concentration of their blood was assayed.  ßIRKO mice consistently had higher levels of glucose in their blood with respect to the other three groups of mice, and the discrepancy seemed to widen with age.  Coupled with the data from figure 3, this new data seems to suggest that, although ßIRKO mice can get rid of glucose when it is in the bloodstream, they do so much less efficiently than control mice.
    Figure 6a is an immunostain of pancreatic sections of control (wt and IRLox) and ßIRKO mice using an antibody cocktail. At 2 months there is no difference between the control group and the ßIRKO group in terms of islet size or in the ratio of ß to non-ß-cells. However, at 6 months there is a noticeable difference in the size of islets in the pancreas of the control and ßIRKO mice (roughly 20-40%).
    Figure 6 b shows the difference in insulin content of control and ßIRKO mice at 2 and 6 months, respectively. There is no noticeable difference at two months. However, at 6 months there is 35% more insulin in control mice as compared to ßIRKO mice. This shows that there might be a correlation between the decreased size of islets in ßIRKO mice and a lack of insulin, though the researchers make it a point to state that the decrease in insulin of ßIRKO mice is not the sole result of a decrease in islet size.
    Finally, figure 7 contains two images from an electron microscope and shows no apparent differences in the cell membrane, endoplasmic reticulum,Golgi, or ß-cell granules between IRLox and ßIRKO mice. This shows that ß-cell morphology is consistent in both homozygous IRLox and ßIRKO mice.
    Figure 7b shows an immunoflourescent stain of pancreatic ß-cells in IRLox and ßIRKO mice. They are virtually the same and support the claim that changes in morphology that prevent glucose entry into ß cells are not the cause of decreased insulin production. Islets from ob/ob mice were barely detected by Glut-2 antibodies, which is consistent with the literature.  Furthermore, since ob/ob islets differed so drastically from ßIRKO and IRLox mice, the researchers argue that they have discovered an entirely new mechanism for the onset of type 2 diabetes.
Discussion
    The researchers state that type 2 diabetes is characterized by a loss of acute insulin secretion in ß-cells in response to a glucose challenge and hyperglycemia (high blood sugar), two things the researchers also demonstrated were present in ßIRKO mice. Furthermore, the researchers reiterate the fact that insulin is still being secreted in liver, skeletal, brain, and muscle cells in the ßIRKO mice. Therefore, the ß-cells are malfunctioning because of the deleted gene segment and not because of Cre expression or the presence of LoxP restriction sites. They also state that the exact ß-cell defect in type 2 diabetes patients is unknown, though it has been postulated to be any combination of the following: an abnormal pattern of insulin secretion that follows a bell-shaped curve rather than a curve that increases as glucose concentration increases, an increase in proinsulin product secretion in ß-cells, and a decreased response to challenges from insulin-inducing stimuli such as certain amino acids. Two previous studies cited by the researchers showed that ß cells may positively regulate their own insulin secretion, meaning ß-cell insulin release triggers more insulin release, and the ß-cell protein glut2 may correlate with reduced levels of insulin secretion. However, the Glut2 stain the researchers conducted was normal, so the lack of insulin in ß-cells cannot be attributed to reduced levels of the protein Glut2. Thus, the researchers concluded that there was no development abnormality in ß-cells, rather the lack of insulin secretion is most likely due to an alteration in proinsulin to insulin conversion or a change from constitutive pathways to regulated pathways. Furthermore, the researchers cited a third study that showed in humans with mutations in the insulin receptor gene it was shown that insulin secretion increased, which is the exact opposite of what is seen in the ßIRKO mice. The body of evidence presented in the paper leads the researchers to conclude that they have found a new model for type 2 diabetes in which a non-functional insulin receptor in pancreatic ß-cells and insulin resistance in peripheral cells causes age-dependent glucose intolerance (hyperglycemia) and a decrease in insulin secretion due to a glucose challenge in mice.  However, this animal model is the exact opposite of what is known to occur in humans with type 2 diabetes (in humans, more insulin is secreted when the insulin receptor is non-functional).
    The paper details a very-well controlled, logical experiment in which the researchers provide evidence for all their assertions. However, there is one flaw and one incosistency in the paper. The first flaw is more stylistic than contextual; the paper presupposes the reader already has near-expert knowledge of type 2 diabetes and recent research on the disease. This, combined with the fact that the researchers were vague as to many of their methods (such as the breeding of the mice) and many of the figures had vague or inadequate figure legends (why does PCR of the IRLox allele result in a 300 bp band while PCR of the IRKO allele results in a 220 bp band when the cartoon seems to imply that primers P3 and P2 used for the ßIRKO allele would produce a band so small as to be only afew bp in length? Answer: It seems that the only likely answer is the space between the two flox sites is 80 bp in length and the segment upstream of this site is 220 bp in length. Thus, although the primers imply that the segment is only a few bp in length, the actual PCR results are an allele with the upstream bp coding sequence attached to the segment of code between the two PCR primers, which accounts for the 220 bp increase in length), detracted from the overall beauty of the experiment that the researchers flawlessly conducted. In fact, the only unpredicted result the researchers obtained was in figure 2 when some pancreatic islets in ßIRKO mice were found to contain the normal allele.  However, the researchers deduced the cause of this result and then showed that their deduction was consistent with the results of an experiment they ran to test their deduction (Specifically, pancreatic ß-cells are composed of some ß and some non-ß cells. The non-ß cells contained the normal allele).
    The inconsistency in the paper is that the result the researchers obtained from the experiment, namely that a non-functional insulin receptor causes a decrease in the amount of insulin produced in response to glucose, is not only inconsistent, but the exact opposite of what is seen in humans with non-functional insulin receptors. The researchers claim this is a result of the differences between the insulin receptors in mice and humans. Thus, although the researchers set out to study type 2 diabetes in mice as a model for the disease in humans, the model they obtained is not compatible with at least one aspect of the current  human model. Ethical concerns preclude a study similiar to this study in humans in order to see the effects of a deletion.
    Possible follow up experiments include sequencing the allele responsible for the non-functional insulin receptor in mice with type 2 diabetes in order to determine if this allele shows a deletion similliar to the one induced in this experiment. This would be easy enough, as the wt allele has already been sequenced and a restriction map has been obtained. Thus, all the researcher would have to do is make primers complimentary to one end of the allele, cut the allele with two restriction enzymes, and chromosome walk down the allele using the sanger method to sequence it. One would expect that in mice with type 2 diabetes there would be a similiar deletion in the allele to account for the non-functioning insulin receptor. If there wasn't, then it is very unlikely that, in mice, type 2 diabetes is caused by a non-functioning insulin receptor due to a molecular factor. One would then have to search for other possible causes, either ruling out a non-functioning insulin receptor altogether, or searching for another cause other than a molecular one to explain the non-functioning receptor. Such a cause might be the lack of certain enzymes that participate in the recognition of glucose or the secretion of insulin by the insulin receptor. Also, in a study such as this, it might be helpful to do some sort of affinity chromotography experiment in which the insulin receptor produced by individuals with type 2 diabetes could be isolated and compared to a wt individual's receptor.
    Another follow-up experiment suggested by the researchers in their paper might be to try to determine if insulin secretion is a result of glucose binding to the receptor or insulin acting on its own receptor in an autocrine manner. This sort of a study would be relatively easy as well. All one would need are two sets of isolated healthy insulin receptor cells. To one set one would add glucose and to the other one would add insulin. The glucose would serve as the control to make sure the receptors could produce insulin.  If the receptors produced more insulin in response to the insulin then one would know that glucose merely turns on or off the pathway, while insulin is responsible for initiating insulin production in the receptor. After discovering how a healthy insulin receptor behaves, the experiment could be repeated with a receptor obtained from a diabetic. If insulin was made in response to insulin but not glucose, then one would deduce that there is a problem with identifying the glucose molecule and initiating inuslin production but not in the actual synthesis of insulin in diabetic individuals. This would support the model the researchers are proposing and point to a dysfunctional binding domain. However, if the opposite turned out to be true, then there is something wrong with the insulin production process and the model the researchers propose is unlikely to account for diabetes in humans.  This experiment should be conducted in both humans and mice to determine if type 2 diabetes is due to the same cause in both organisms.

References:
Gu, Marth, Orban, Mossmann, and Rajewsky. (1994) Deletion of a DNA polymerase ß gene segment in T cells using cell type-specific gene targeting. Science 265: 103-106.

Return to Davidson College Biology Department Home Page
Return to Biology Course Materials
To email me, click here.
To return to my homepage, click here.