Targeted Inactivation of Kinesin-1 in Pancreatic β-Cells In Vivo Leads to Insulin Secretory Deficiency

A bacterial artificial chromosome DNA containing the Kif5b gene (bacterial artificial chromosome clone 307D12) was partially digested with XbaI and subcloned into a pBluescript II-KS plasmid. A plasmid containing a 14.5-kb DNA insert, pBS-Kif5b, was identified by Southern blotting using a labeled probe containing exon 2, where the ATP-GTP-A site is located (data not shown). The TK4 gene from pBS-304 was cloned into the SalI site of pBS-Kif5b. The first loxP site was introduced into the first intron of Kif5b gene by the recombinogenic targeting of a loxp-Pgk-Tn5-Neo-loxp cassette (PCR amplified from the modified pGK-loxP plasmid with primer: TCTTGTGACTTAGAGTTTATAAAATAGAGTAATTTTGAAAACACATAGATATCTGCAAACCCTATGCTACTCCGTCG and CGCTCTCCTGAGTAGGACAAATCCGCCGGGACTATAGGTTGTGTAATGTGTATTTCAAGTCAAAAAATGAATTGAAAAAAA) into pBS-Kif5b, followed by removal of the fragment between the two loxP sites via Cre-mediated recombination. The second loxP site accompanying the frt-Pgk-Tn5-Neo-frt cassette was introduced into the second intron of Kif5b gene by targeting a frt-Pgk-Tn5-Neo-frt-loxP cassette (PCR amplified from the modified PBS-246-FRT plasmid by PCR primers primer: GGATGCACGGCTGTGAGCACAGGACTTTCCTGTGTTTGGAGT and primer: AGTTGGATTTAAGGAAGTACTACTAAAACTTCAATTAGTCTTACTAAAAA) into the pBS-Kif5b-loxP to form the final Kif5b knockout construct. The Kif5b-knockout construct, containing 14.5 kb of the Kif5b gene, was linearized by NotI and was introduced into CJ7 mouse embryonic cells in Drs. Jenkins and Copeland's lab in National Cancer Institute (NCI), USA, by electroporation and screened by genomic Southern blotting. The neo of Kif5b targeted allele was removed by crossing the mice with a Fple deleter strain TgN(ACTFLPe)9205Dym/J (23). Kif5b^+/− mice were generated by crossing Kif5b^+/fl mice with actin-Cre mice. Mice with Kif5b^fl/fl and Kif5b^+/− genotypes were maintained by backcrossing to C57BL/6N females under a specific pathogen-free environment.

RIP2-Cre transgenic mice (24) (from Jackson Laboratory) were first crossed with Kif5b^+/− to generate Kif5b^+/−; RIP2-Cre mice. Then Kif5b^+/−; RIP2-Cre mice were bred with Kif5b^fl/fl mice to generate the final mutant mice (Kif5b^fl/−:RIP2-Cre) as well as their littermates (Kif5b^+/fl, Kif5b^−/fl, and Kif5b^+/fl:RIP2-Cre). Genotyping was performed by PCR using corresponding primers. To avoid hormonal effects in physiological assays, only male animals were used in all experiments.

Islet isolation and insulin extraction were carried out as previously described (25). Insulin contents in the islet extracts were analyzed by using the insulin ELISA kit (LINCO Research) and normalized to total protein concentrations determined by Bradford method (Bio-Rad).

Protein levels were determined by blotting with anti-Kif5b primary antibody (1:2,000, against synthesized peptide FDKEKANLEAFTADKDIA), anti-Kif5a (1:1,000, against synthesized peptide NGNATDINDNRSDLPC), anti-Kif5c (1:1,000, against synthesized peptide SAKDQKSLEPC), and anti-actin primary antibody (1:3,000, Sigma). Enhanced chemiluminescence kit (Pierce) was used for detection of the immunoreactive bands.

Paraffin sections were immunostained by primary antibodies against Kif5b (1:400), glucagon (1:1,000, Dako), and insulin (1:2,000, Sigma). Visualization of these proteins was realized by incubating with FITC/Cy3 or HRP-conjugated secondary antibodies.

Blood was obtained from the tail vein of mice. Blood glucose concentration was determined by Glucometer Elite (Bayor Inc.). Blood samples were collected by Microvette CB300 (Sarstedt) and centrifuged at 5,000 rpm for 5 min at room temperature. After centrifugation, supernatant was collected for determining plasma insulin levels by using insulin ELISA kit (LINCO Research).

Paraffin sections of pancreata were processed for hematoxylin and eosin staining and observed by a light microscope (Carl Zeiss). For transmission electron microscopy (TEM) study, isolated islets were fixed by 2.5% glutaraldehyde in carcodylate buffer and embedded in 2% soft agar, followed by further embedding in fresh epoxy resin. Ultrathin sections were examined on a Philips EM208s transmission electron microscope operated at 80 kV. All photos were analyzed by using Image-Pro Plus software.

β-cell division was determined by 5-bromo-2′-deoxyuridine (BrdU) incorporation in 1-month-old mice using BrdU staining kit (ZYMED Laboratory). Pancreatic paraffin sections were further counterstained by hematoxylin and imaged with a Zeiss Axioskop microscope to count BrdU⁺ β-cells and total nuclei within islets.

SigmaStat (Systat Software) was used to analyze all data, which were expressed as mean ± SE. The data were analyzed by one-way ANOVA followed by Tukey test or unpaired two-tailed Student t test.

The ability of RIP2-Cre to induce pancreatic β-cell specific ablation has previously been reported when these founder mice were crossed with mice carrying various floxed genes including glucokinase (26), insulin receptor (27), mitochondrial transcription factor A (28), and hepatocyte growth factor (25). Although RIP2-Cre displays a low level of expression in the hypothalamus, it exhibits high expression level in β-cells within the pancreatic islet (24). Therefore, conditional knockout of Kif5b mainly occurred in pancreatic β-cells. Figure 1,A shows the strategy to generate Kif5b^fl/fl and Kif5b^+/− mice. Ablation of one Kif5b allele was confirmed by Southern blot as well as Western blot analyses (Fig. 1,B and C). Figure 2,A shows the genomic organization of different Kif5b alleles. P1 and P2 primers were used to differentiate the wild-type and knockout alleles. The distance between P1 and P2 was 6.6 kb in length, which was too long to be amplified by PCR genotyping. Because of the deletion of exon 2, a 219-bp PCR product could be amplified from the Kif5b knockout allele accordingly (Fig. 2,B). P1 and P3 primers, used to differentiate the floxed and wild-type Kif5b alleles, result in two bands of 275 and 215 bp, respectively (Fig. 2,B). The RIP-2 Cre transgene was detected using Cre primers (Fig. 2 B).

Kif5b was mainly expressed in pancreatic islets (endocrine portion) with little expression in exocrine acinar glands (Fig. 3,A, left). The positive staining was absent after preincubating the anti-Kif5b antibody with the 18-aa synthesized peptides (Fig. 3,A, right). To test the efficiency of RIP-2-Cre mediated pancreatic β-cell specific deletion of exon 2 of Kif5b allele, we examined the protein expression levels of Kif5a, Kif5b, and Kif5c in hypothalamus and isolated islets from both mutant and wild-type mice (Fig. 3,B). Inactivation of Kif5b by Cre-induced ablation resulted in a reduction of Kif5b protein level in islets, whereas both Kif5a and Kif5c were not detectable in islets although they are highly expressed in hypothalamus, consistent with an earlier finding that their expression is neuronal specific (29). Although RIP2-Cre displayed a low level of expression in the hypothalamus (24), we could not detect significantly decreased Kif5b expression in the hypothalamus due to Cre expression. In contrast, Western blot results showed that Kif5b level in the isolated islets was reduced by at least 80% in mutant mice, compared with the wild-type. Considering the fact that RIP-2 Cre is only expressed in ∼80–90% of β-cells and that β-cells account for 75% of islet cells, the Western blot result indicated that Kif5b was absent in most β-cells in the mutant mice islets. The knockout efficiency was further demonstrated by immunohistochemistry analysis of Kif5b expression in pancreatic sections under identical conditions. A significant decrease of Kif5b expression was observed in the islets from mutant mice compared with those in wild-type mice (Fig. 3 C). Taken together, these results suggested that RIP-2 Cre mediated ablation of exon 2 in Kif5b allele had efficiently occurred, leading to significantly reduced Kif5b expression in the islets of mutant mice.

Although there was no significant difference in body weight among newborn mice of different genotypes, the mutant mice had progressively reduced growth comparing to littermate controls after 3 weeks (Fig. 4,A). Comparing the mice growth curves, the body weight differences between mutant and control littermates were significant (as shown in the representative photo in Fig. 4,B) and there was no significant difference among wild-type and heterozygous controls (Fig. 4 A).

To test the effect of Kif5b conditional knockout on glucose homeostasis, we first analyzed the blood glucose level as well as plasma insulin level in both mutant mice and littermate controls. Mutant mice had progressively increased hyperglycemia compared with control littermates in both random (Fig. 4,C) and 16-h fasting conditions (Fig. 4,D) within the investigated ages. Therefore, 2- to 3-month-old mice were selected for further study. Consistent with an elevated glucose level, mutant mice had lower plasma insulin concentrations in fasting conditions (Fig. 4,F) but not in a random fed state (Fig. 4 E). There were no significant differences in the blood glucose levels as well as plasma insulin concentrations between wild-type and heterozygous controls.

We next assessed the impact of inactivation of Kif5b on glucose-regulated insulin secretion in vivo (Fig. 5,A). The responses of 2- to 3-month-old mutant mice and control littermates to glucose challenge were examined by performing an intraperitoneal glucose tolerance test (GTT) after 16-h overnight fast. Kif5b^fl/+:Cre mice showed slight glucose intolerance compared with wild-type mice, which may be due to Cre expression in β-cells (30) in addition to the ablation of one Kif5b allele. Comparison of the GTT responses between mutant mice and Kif5b^fl/+:Cre mice showed that Kif5b conditional knockout mice had markedly impaired glucose tolerance and this intolerance was not a result of Cre expression. Although the mutant mice had hyperglycemia and glucose intolerance compared with wild-type and Kif5b^fl/+:Cre mice, the blood glucose levels dropped to a similar level as control mice at 15 min after insulin injection (Fig. 5 B).

To evaluate the effect of Kif5b deletion on β-cell function, insulin secretion in response to high glucose challenge was examined in 2- to 3-month old mice (Fig. 5,C). In wild-type mice, insulin secretion was increased by about twofold at 2.5 min after intraperitoneal injection of glucose, reaching a peak at about 5 min, and remained higher than base level for at least 1 h. Glucose tolerance and insulin release were slightly impaired in Kif5b^fl/+:Cre mice compared with wild-type mice in the first 30 min after glucose injection, but the glucose and insulin levels were similar in both genotype mice by 60 min. However, both the early and slow phase of insulin secretion were impaired in Kif5b^fl/−:Cre mice. These results indicated that glucose intolerance in Kif5b conditional knockout mice was not associated with insulin resistance (Fig. 5,B) but was related to an insulin secretory defect in response to glucose challenge (Fig. 5,C). It is possible that impaired insulin secretion could be secondary, particularly when considering the expression of Cre in hypothalamic or pituitary cells. To address this possibility, we studied GSIS in isolated islets from mice with different genotypes (Fig. 5,D). As expected, mutant islets secreted significantly less insulin than control islets under glucose-stimulated conditions. No significant difference in insulin secretion was observed in the absence of glucose. Furthermore, the insulin-secretion defect was progressively enhanced in mutant islets with an increase in glucose concentration. A high level of KCl is known to trigger insulin secretion by inducing plasma membrane depolarization followed by Ca²⁺ influx (1). To further investigate the effect of Kif5b knockout on insulin secretion, KCl-stimulated insulin release from islets was analyzed by 30 mmol/l KCl treatment (Fig. 5 E). The wild-type islets exhibited stronger secretory responses than mutant islets.

To rule out the possibility that insulin is produced at a lower level in the mutant islet than in the wild-type, an immunostaining assay was carried out. The result demonstrated that the insulin level in mutant mouse was not less than that of wild-type control (Fig. 6,A). A quantitative analysis of islet insulin content was carried out by ELISA to confirm the result. Results indicated that more insulin accumulated in mutant islets compared with wild-type, although the difference was not statistically significant (Fig. 5 F). Taken together, it is most likely that the impaired insulin secretion in mutant mice is due to a primary defect in β-cell function.

Islets from mutant mice showed similar cellular arrangements and compositions as wild-type mice, with the majority of cells being β-cells and with glucagon-positive α-cells in the periphery region (Fig. 6,A). There were no significant differences of α- and β-cell ratio between mutant and wild-type islets (Fig. 6,B). Because Kif5b has been reported to transport insulin vesicles in β-cell lines (MIN6 & INS-1) (18) and mitochondria in neurons (31), we further analyzed the distribution of these vesicles/organelles in mutant and wild-type β-cells by TEM studies (Fig. 7). Compared with the control, mitochondria in mutant β-cells were found to be more likely to cluster in the perinuclear region (Fig. 7,A). Further quantitative analysis of the distribution of mitochondria in β-cells confirmed that 70% of mitochondria were closer to the nuclear membrane than to the plasma membrane after Kif5b knockout whereas mitochondria in wild-type β-cells were evenly distributed in the cells (Fig. 7 B). This is consistent with earlier findings that Kif5b is involved in mitochondria translocation (32).

The subcellular localization of insulin vesicles was then analyzed and found to not be affected significantly by the decreased Kif5b level. The cytoplasm of both wild-type and mutant β-cells were filled with insulin vesicles. Insulin vesicle numbers per square μm were determined by counting all insulin vesicles in randomly photographed β-cells (Fig. 7,C). More insulin granules were found in Kif5b knockout β-cells compared with control cells. This phenomenon is consistent with the observation that insulin secretion by β-cells is affected. Interestingly, a number of small islets were observed in mutant mice as shown in the representative microphotographs in Figure 8,A. Quantitation of the islet number and size confirmed that the average size of the islet in mutants was smaller than that of the wild-type mice (Fig. 8,B). However, the total islet mass was not decreased in Kif5b deficient mice (Fig. 8,C), concomitant with an increase of islet number (Fig. 8,D). Histomorphometric analysis of the size and number of islets in these pancreatic specimens showed that the number of islets <4,000 μm² was significantly increased in Kif5b conditional knockout mice. However, the number of islets >4,000 μm² in mutant mice was comparable to that of wild-type control (Fig. 8,E). No reduction of islet cell size was observed in Kif5b deficient mice (Fig. 8,F); therefore, the major factor accounting for reduced islet size was reduced cell number in an islet. We analyzed islet cell proliferation rates in pancreatic sections obtained from 1-month-old mice by BrdU⁺ labeling. As shown in Figure 8 G, islet cell proliferation was reduced by ∼50% relative to the wild-type control. There was no detectable increase in islet β-cell apoptotic rate based on TUNEL⁺ analysis (supplementary Fig. 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db09-1078/DC1).

To study the role of Kinesin-1 in pancreatic β-cells in vivo, we have generated cell-specific Kif5b-insKO mice (Kif5b^fl/−:RIP2-Cre) by crossing Kif5b^fl/fl mice with Kif5b^+/−:RIP-Cre mice, in which Cre recombinase is expressed under the control of rat insulin gene promoter (RIP2). Although RIP2-Cre in pancreatic β-cells may have some effect on glucose tolerance in mice (30), this molecular genetic approach allows in vivo studies of specific gene function and could exclude possible side effects of in vitro introduction of anti-kinesin antibodies (33) and overexpression of dominant-negative-acting kinesin (18), which may interfere with biological processes such as endoplasmic reticulum stress.

This in vivo study reveals that Kif5b plays an important role for β-cell function and development. Kif5b-insKO mice displayed significant glucose intolerance and insulin secretory defects. Change in β-cell function was not coupled with changes in islet cell size, islet morphology, and islet cell composition but was associated with a reduced islet size as a consequence of decreased β-cell proliferation rate.

Kif5b-insKO mice exhibited growth retardation within the indicated period. Physiological examination found that Kif5b-insKO mice displayed hyperglycemia and decreased plasma insulin levels, together with significantly diminished glucose tolerance and defective insulin secretion, suggesting a role of Kif5b in β-cell function. Our results confirm that Kif5b is required for normal GSIS in an in vivo model. Hypoinsulinemia may account for the poor growth of Kif5b-insKO mice. However, RIP2-Cre expression in hypothalamus and pituitary (24,34) may also lead to the knockout of Kif5b in insulin-positive cells, although no significant reduction of Kif5b level due to Cre expression was observed in hypothalamic lysates of Kif5b-insKO mice. The functions of Kif5b in hypothalamic and pituitary endocrine hormone secretion are unknown at the current stage. Therefore, we cannot rule out the possibility that the observed poor growth is also related to reduced growth hormone and/or insulin-like growth factor secretion.

In vitro studies showed that suppression of Kif5b by overexpression of dominant-negative-acting Kif5b inhibited only the sustained phase of GSIS in clonal β-cells (18). Interestingly, our in vivo GTT results (Fig. 5 A and C) indicated that the defective insulin response to glucose in a Kif5b-insKO mouse occurred in both the early phase and the slow phase. The early phase of insulin secretion is accompanied by Ca²⁺ influx induced exocytosis (1), which requires the metabolism of glucose to create ATP in mitochondria, leading to depolarization of the cell due to the closing of ATP-gated potassium channels. Kif5b and Kif5c have been reported to bind to RanBP2 to determine mitochondria localization and function in neurons (31). In the current study, we found that mitochondria significantly accumulate at the perinuclear region in the β-cell of Kif5b-insKO mice. Although disruption of Kif5b in extraembryonic cells upregulated the mRNA expression level of Kif5a and Kif5c (32), there is no apparent upregulation of Kif5a/Kif5c at the protein level in mutant β-cells, indicating that Kif5b is the major motor responsible for the dispersion of mitochondria in pancreatic β-cells. The deficiency of mitochondria at the cell peripheral regions may attenuate glucose-to-ATP conversion and partially account for the defect in the early phase of GSIS in response to glucose challenge.

To examine whether blocked Kif5b-mediated insulin vesicle transportation in Kif5b-insKO mice may lead to a decrease in predocked vesicles ready for exocytosis, we also performed TEM studies. However, we observed no significant differences in the number and size of predocked insulin vesicles in Kif5b-insKO mice compared with control mice at the resolution used. Bi et al. (33) observed that KHC (or Kif5b) antibody SUK4 as well as the stalk-tail fragment of KHC specifically inhibited Ca²⁺-regulated exocytosis in Lytechinus pictus sea urchins. Therefore, exocytosis of insulin vesicles could also be affected in the Kif5b knockout islet. In neurons, it was demonstrated that Kif5b can directly bind to syntabulin and SNAP25 to mediate transportation of syntaxin containing vesicles and synaptosome (35,36). Syntaxin and SNAP25 constitute the t-SNAREs at the plasma membrane and are functionally required for Ca²⁺-triggered insulin exocytosis in β-cells (37). Snap25 mutant mice had impaired insulin granule priming, exocytosis, and recycling in pancreatic β-cells (38). It is possible that knockout of Kif5b in pancreatic β-cells affects the normal distribution and function of the SNARE complex, thereby leading to defects in insulin granule exocytosis. Furthermore, Kif5b functions in several cellular processes including lysosomal distribution and stability in cancer cells and extraembryonic cells (32,39), endoplasmic reticulum to Golgi transportation in Hela cells (40), transportation and axonal targeting of Kv1 channels (41), axonal transport of tubulin heterodimer for microtubule polymerization (42) as well as ribosome translocation (43), etc. Similar cellular functions of Kif5b could occur in pancreatic β-cells. It is tempting to speculate that weakening of both the early and slow phases of insulin response to glucose in Kif5b-insKO mice results from defects in a series of cellular functions after knockout of Kif5b in pancreatic β-cells. Taken together, Kif5b probably exerts a modulatory effect in regulating insulin granule priming and exocytosis in addition to mechanically transporting insulin vesicles. Furthermore, impaired early insulin response is consistent with a defect in glucose tolerance, because the early response is more important for glucose disposal (44). Insulin secretion defects result in a positive feedback for the synthesis of insulin in β-cells, thereby leading to accumulation of insulin in Kif5b-insKO mice islets.

Donelan et al. (45) reported that increased [Ca²⁺]_i can activate calcineurin in β-granules leading to dephosphorylation of KHC, which is required for transport of β-granules from the storage pool to replenish the readily releasable pool of β-granules. Rab3A is mainly located on the cytosolic face of β-granule membranes. The Rab3A–calmodulin interaction on β-granules is required for activation of calcineurin and dephosphorylation of KHC (46). Yaekura et al. (47) found that knockout of Rab3A in pancreatic β-cells leads to blunted first-phase glucose induced insulin release in vitro and in vivo. Our results together with these studies indicate that any defect of synthesis or dephosphorylation process of KHC can lead to impaired insulin exocytosis.

On the basis of histomorphological investigations, we found a notable increase in the number of small islets in Kif5b-insKO mice (Fig. 8,A and E). BrdU⁺ staining of 1-month-old mice confirmed that the decreased islet size in Kif5b-inskO mice was due to a defect in islet cell proliferation (Fig. 8 G). Furthermore, decreased cell proliferation rate was also observed in Kif5b deficient Madin-Darby canine kidney cells (supplementary Fig. 2, available in an online appendix), suggesting that Kif5b plays a conventional role during cell division. Zhu et al. (48) systematically analyzed the functions of all human kinesin/dynein microtubule motor proteins by RNA interference and identified at least 12 human kinesins involved in HeLa cell division. Recently, Haraguchi et al. (49) found that the kinesin-2 complex (KIF3A/3B) is localized with components of the mitotic apparatus such as spindle microtubules and centrosomes and plays an important role not only in interphase but also in mitosis. Both Kif5b and KIF3A/B are microtubule plus end-directed motors for membrane organelle transportation. Therefore, it is of considerable interest to determine whether Kif5b has a similar function or another unique function during cell division.

One seemingly contradictory result found in this study was that islet mass was not affected (Fig. 8,C) even though β-cell proliferation was inhibited in the Kif5b deficiency mice. Further analysis of the pancreatic sections found that islet number was significantly increased (Fig. 8 D). Therefore, it is possible that the impaired β-cell function in Kif5b-insKO mice induce β-cell neogenesis to provide a compensatory effect, a process difficult to be defined and more difficult to be quantified at the molecular level (25,50).

In summary, by using Kif5b conditional knockout mice, we have demonstrated that Kif5b is essential in the maintenance of normal β-cell function. Kif5b ablation leads to decreased insulin secretion and diminished glucose tolerance with decreased islet size in the pancreas. Our results indicate that genetic or epigenetic alteration in Kinesin-1 or other related motor proteins may inherit a predisposition to develop diabetes.

This work was supported by grants from the Hong Kong Research Grants Council (RGC) (HKU 7321/04M, HKU 7636/05M) to J.-D.H., a Hong Kong University Small Project Fund to K.-M.Y., and partial funding by an RGC Group Research Project (HKUST6/CRF/08). No potential conflicts of interest relevant to this article were reported.

J.C. wrote the manuscript and researched data. Z.W., Q.C., R.L., and X.-M.Z. researched data. N.G.C. and N.A.J. reviewed/edited the manuscript and generated Kif5b conditional allele. P.S.L. contributed to discussion and reviewed/edited the manuscript. K.-M.Y. and J.-D.H. designed experiments and wrote/reviewed/edited the manuscript.

We thank Dr. Julian Tanner at the University of Hong Kong for his comments on the manuscript.