Glucose-stimulated insulin secretion from pancreatic β-cells requires the kinesin-1/Kif5B-mediated transport of insulin granules along microtubules. 5′-AMPK (5′-AMP-activated protein kinase) is a heterotrimeric serine/threonine kinase which is activated in β-cells at low glucose concentrations, but inhibited as glucose levels increase. Active AMPK blocks glucose-stimulated insulin secretion and the recruitment of insulin granules to the cell surface, suggesting motor proteins may be targets for this kinase. While both kinesin-1/Kif5B and KLC1 (kinesin light chain-1) contain consensus AMPK phosphorylation sites (Thr693 and Ser520, respectively) only recombinant GST (glutathione transferase)–KLC1 was phosphorylated by purified AMPK in vitro. To test the hypothesis that phosphorylation at this site may modulate kinesin-1-mediated granule movement, we developed an approach to study the dynamics of all the resolvable granules within a cell in three dimensions. This cell-wide approach revealed that the number of longer excursions (>10 μm) increased significantly in response to elevated glucose concentration (30 versus 3 mM) in control MIN6 β-cells. However, similar changes were seen in cells overexpressing wild-type KLC1, phosphomimetic (S517D/S520D) or non-phosphorylatable (S517A/S520A) mutants of KLC1. Thus, changes in the phosphorylation state of KLC1 at Ser517/Ser520 seem unlikely to affect motor function.
The role of kinesin-1 in insulin secretion
In mammals, blood glucose levels are maintained within a narrow range of concentrations due to a balance between glucose storage and consumption, controlled primarily by two antagonistic hormones insulin and glucagon. Insulin is secreted from pancreatic β-cells in response to a rise in blood glucose levels through a complex sequence of events , involving initiation and amplification processes, which result in the biphasic release of the hormone . Initiation is characterized by an increase in intracellular ATP levels [3,4], leading to the closure of ATP-sensitive KATP channels,  depolarization of the plasma membrane, and the opening of voltage gated Ca2+ channels . An influx of Ca2+ stimulates the fusion of insulin-containing granules from a ‘readily releasable’ pool, predocked, at the plasma membrane. These events are thought to drive the ‘first phase’ of insulin secretion, occurring 2–5 min after a rise in glucose concentrations . The ‘second phase’ of insulin secretion is likely to involve both initiating and amplifying events and requires the microtubule-dependent recruitment of granules from a ‘reserve pool’ to the cell surface . Consequently, silencing, or inactivation by a dominant negative form of the anterograde microtubule-dependent motor protein kinesin-1, blocks the sustained phase of glucose-stimulated insulin secretion [9–13].
Kinesin is a heterotetramer composed from two heavy [KHC (kinesin heavy chain)] and two light [KLC (kinesin light chain)] chains. In most non-neuronal vertebrate cells, one heavy isoform, Kif5B, also known as uKHC (ubiquitous KHC) or kinesin-1/Kif5B, is expressed along with KLC1 and KLC2 . We have previously shown that kinesin-1 is associated with, and responsible for, the transport of insulin-containing granules during the sustained phase of insulin secretion .
Kinesin is a phosphoprotein with both KHC and KLC being phosphorylated in cells . Whereas several of the described phosphorylation sites stimulate motor activity, others were found to be inhibitory [16,17]. This suggests that the phosphorylation of KHC and KLC on specific residues by specific kinases may be an important regulator of motor protein function or localization in living cells. The regulation of kinesin function is also likely to depend on the cell type, on the isoform expressed and on the interaction between and conformations of the different subunits [18–23]. The mechanism(s) through which elevated glucose concentrations regulate insulin granule transport via kinesin-1 are still poorly understood.
AMPK (AMP-activated protein kinase) is a heterotrimeric serine/threonine protein kinase that is activated by a decrease in the ATP/ADP ratio, and the consequent increase in AMP, therefore allowing it to sense the energy status of a cell [24–26]. Our laboratory and others have in recent years provided evidence that AMPK plays an important role in the acute stimulation of insulin secretion from pancreatic β-cells [25,27,28]. Thus, at low glucose concentrations, AMPK is active in these cells and suppresses the release of insulin [28–31], in part by blocking the glucose-stimulated recruitment of insulin-containing granules to the plasma membrane. As discussed below, AMPK-dependent phosphorylation of KHC and or KLC might represent a mechanism through which kinesin-1 activity is suppressed at low glucose concentrations, blocking granule movement [30–33].
Can AMPK phosphorylate kinesin
The sequences of both KHC and KLC1 were analysed and contain AMPK consensus sequences [Ω(X,β)XXS/TXXXΩ, where Ω represents a hydrophobic residue and β represents a basic residue] at Thr693 and Ser520 respectively. Work in this laboratory using peptide substrates has recently suggested that KLC1 may be phosphorylated by AMPK (E.V. Hill, I. Leclerc and G.A. Rutter, unpublished results) . Here, we first investigated whether KLC1 was phosphorylated by AMPK in vitro using recombinant GST (glutathione transferase)–CHO (Chinese-hamster ovary)-KLC1B and both a phospho-specific anti-Ser520 KLC1 antibody, raised against a peptide derived from the mouse sequence around Ser520 (Figure 1A), and the incorporation of [32P]ATP (Figure 1B). After treatment with purified rat liver or recombinant bacterially expressed AMPK, the phospho-specific antibody recognized a single band corresponding to the molecular weight of KLC, which was abolished when the equivalent serine in CHO-KLC, Ser446, was mutated to an alanine. Furthermore, autoradiography revealed that 32P had been incorporated into the wild-type protein, which was reduced in the S446A non-phosphorylatable CHO-KLC1 mutant, suggesting that KLC1 is indeed a substrate for AMPK in vitro.
AMPK phosphorylates KLC in cell free assays
Impact of phospho/dephosphomimetic mutations at Ser517 of hKLC (human KLC) on motor function
Since KLC1 was a substrate for AMPK in cell-free phosphorylation assays, phosphomimetic and non-phosphorylatable mutants were generated at Ser517 (the conserved serine) in hKLC1 isoform 2. Previous work in this laboratory showed that a dominant-negative KHC mutant blocked glucose-stimulated granule movement . Thus, if phosphorylation at Ser517 of KLC1 had any effect on kinesin motor function we may expect to observe an effect on glucose-regulated granule dynamics.
To date, granule movement has been studied manually or in two dimensions [12,35]. To investigate the possible effects of Ser517 phosphorylation we developed a novel imaging protocol and automated analysis tool to enable us to monitor the behaviour of essentially all of the optically resolvable granules in three dimensions in real time. This was done using a Nokigawa spinning disc confocal microscope fitted with a Ludl piezo stage. Unlike a point scanning confocal microscope, the spinning disc confocal contains multiple pinholes that allow sample illumination of the entire field of view. Coupled with the use of a piezo stage, very fast image acquisition can be accomplished: an entire image stack of approx. 10 slices (5–10 μm total depth) in <1 s.
Cells expressing the granule-targeted fluorescently tagged protein NPY (neuropeptide Y)–mCherry  were imaged to generate a three-dimensional time-lapse sequence. Volocity™ (Improvision) was then used to monitor the granules within the cell using an analysis protocol designed especially to track the movement of granules before and after the addition of elevated glucose concentrations (30 compared with 3 mM) in cells transfected with hKLC1, phosphomimetic or non-phosphorylatable hKLC mutated at Ser517 to aspartic acid and alanine residues respectively.
Once tracks had been generated for all of the optically resolvable granules within a cell, differences in granule dynamics between cells transfected with wild-type, non-phosphorylatable or phosphomimetic KLC1 were analysed. Analysis of track lengths revealed that elevated glucose concentrations caused a marked increase in the number of granules undertaking longer excursions (>10 μm, Figure 2). While there was a significant increase in the proportion of granules undertaking longer excursions in control and wild-type KLC1 transfected cells, similar increases were also seen with cells overexpressing the mutant proteins. Thus, mutating Ser517 to aspartate or alanine had no effect on this increase (Figure 2). We conclude that although KLC1 possesses a potential AMPK phosphorylation site that can be phosphorylated in cell free assays by purified AMPK, phosphorylation of this site in situ has no effect on motor function or granule movement in response to glucose in live cells.
Analysis of insulin granule dynamics
A possible limitation of the present work is the extent to which the overexpressed KLC1s are able to displace the endogenous proteins from all physiologically relevant binding partners. Future studies will be necessary to analyse the impact of the (de)phosphomimetic mutants in the background of cells in which the endogenous KLC1 expression has been ablated entirely.
Molecular Mechanisms in Exocytosis and Endocytosis: 7th Junior Academics Meeting, an Independent Meeting held at University of Edinburgh, Edinburgh, U.K., 5–7 April 2009. Organized and Edited by Rolly Wiegand (Edinburgh, U.K.).
We thank Professor Anne Stephenson (School of Pharmacy, London, U.K.) for the hKLC1 isoform 2 construct, Dr Viki Allan (University of Manchester, Manchester, U.K.) for the GST–CHO-KLC1 constructs and Ms Nasret Harun, Ms Saharnaz Vakhshouri and Mr Gao Sun for excellent technical assistance.
Funded by grants to G.A.R. from the Wellcome Trust programme [grant numbers 067081/Z/02/Z and 081958/Z/07/Z], Medical Research Council [grant number G0401641], National Institutes of Health [grant number ROI DKO71962-01] and European Union Framework Programme 6 (Savebeta). S.F. was supported by a Ph.D. studentship and D.G.H. by a Programme Grant from the Wellcome Trust.