A number of models have been proposed in

A number of models have been proposed in order to explain the mechanisms of protein transport through nuclear pore complexes. One theory is that the transport is mediated by specific peptide signals known as NLSs. The classical NLSs are short sequences containing a high proportion of the positively charged AGK 2 australia lysine and arginine. Although no NLS has previously been characterized for GK, we found an algorithm (PredictNLS) that predicted the sequence 30LKKVMRR36 to fit the description of a monopartite NLS. This stretch of sequence is common to both the pancreatic and liver isoforms of GK. Indeed, the relevance of this seven residue sequence for nuclear import of human pancreatic GK was confirmed by nuclear import of a GFP-NLS fusion protein. Furthermore, we demonstrated a reduced nuclear staining of transfected full-length GK upon substitution of the positively charged amino acids (KK31,32 and RR35,36) within the NLS; the more conserved KK31,32 being the most reduced. The role of this NLS in hepatocytes, however, is unclear since hepatic GK import to the nucleus has been shown to be fully dependent on GKRP (Shiota et al., 1999). Furthermore, a total dependency on GKRP for hepatic GK activity and stability has been shown in GKRP −/− mice. The GK protein levels were strongly reduced or close to absent in GKRP null mice whereas GK mRNA levels remained intact (Farrelly et al., 1999). A possible explanation for the difference observed in the function of GK NLS in hepatocytes and pancreatic beta-cells could be the existence of additional NLSs within the pancreatic GK, as seen for other nuclear proteins (Imagawa et al., 2000, Theodore et al., 2008). In fact, using an alternative NLS prediction tool (cNLS mapper, http://nls-mapper.iab.keio.ac.jp/), a NLS fitting the description of a bipartite NLS between residues 5–36 was identified, and it is unique for the pancreatic isoform. This could explain the inability to completely abolish the nuclear localization of our NLS mutants KK31,32 and RR35,36. In spite of this observation, it is well documented that the residues within the monopartite NLS are crucial for normal GK function, since substitutions within this seven residue sequence (p.L30P, p.V33A, p.R36W and p.R36Q) are associated with MODY in 25 European families (Osbak et al., 2009, Lukasova et al., 2008, Hager et al., 1994). Functional analyses of the p.R36W mutant, for instance, demonstrated a reduced catalytic activity and a reduced affinity for both glucose and ATP (Hager et al., 1994, Miller et al., 1999). We have previously reported on the presence of high-molecular mass cytosolic forms of GK, which includes both SUMOylated and ubiquitinated variants of the enzyme (Aukrust et al., 2013, Negahdar et al., 2012, Bjørkhaug et al., 2007). Here, similar high-molecular mass forms were observed in both cytosolic and nuclear fractions. In nuclear fractions, high levels of GK was detected with size of ∼60 kDa, similar to a mono-SUMOylated GK form previously isolated from rat INS-1 cell cytosol (Aukrust et al., 2013). We also find that the SUMOylation machinery affects the subcellular distribution of GK. Overexpression of SUMO-1, Ubc9 and RanBP2 in MIN6 cells shifted the predominant localization pattern of GK towards the nucleus. Thus, our findings support a role of SUMOylation in the modulation of nuclear GK level. In hepatocytes, the shuttling of GK between the cytoplasm and the nucleus is mediated by a GKRP-GK association/dissociation, which is regulated by the glucose concentration (Beck and Miller, 2013). In our study, glucose increased the level of GK in both the nuclear and cytosolic fractions in human and mouse pancreatic islets, consistent with previous findings on effect of high glucose on GK mRNA levels (Nakamura et al., 2012). However, high glucose did not change the nuclear/cytosolic ratio of GK, indicating that the nuclear localization is not relevant for GK activation in the post-prandial state. Thus, our findings support previous reports indicating that the level of glucose has a stronger effect on GK subcellular distribution in hepatocytes than in pancreatic beta-cells (Lenzen, 2014, Tiedge and Lenzen, 1995, Tiedge et al., 1999). The strong glucose effect on the availability of hepatic GK in the cytosol is thought advantageous, so that sufficient insulin is released from the beta-cell for stimulation of glycogen synthesis in hepatocytes in the post-prandial state. Moreover, the higher cytosolic-to-nuclear ratio of GK in human and mouse islets, and in MIN6 cells, estimated in our study (compared to what reported for hepatic GK (Mukhtar et al., 1999)), thus ensures that the majority of cellular GK is continuously available in the cell cytoplasm for glucose phosphorylation, and secures sufficient insulin secretion for stimulation of glycogen synthesis in the liver.