Glucose-stimulated KIF5B-driven microtubule sliding organizes microtubule networks in mouse pancreatic β cells

Since the first description of the convoluted MT network in MIN6 cells by the Rutter group (Varadi et al., 2003), our views on the regulation, function, and dynamics of the pancreatic β-cell MT network have been gradually evolving (Bracey et al., 2022). However, the field is still far from understanding the mechanisms underlying the network architecture. Here, we show that MT sliding is a prominent phenomenon in β cells and that it is driven by kinesin KIF5 B. This kinesin-1-dependent MT sliding is a critical mechanism needed for the formation and long-term maintenance of β-cell MT networks. In addition, we show that MT sliding activity is facilitated by glucose stimulation, suggesting that this process is involved in the regulation of GSIS and/or providing β-cell fitness during the response to glucose. Overall, our study establishes MT sliding as an essential regulator of β-cell architecture and function.

As we reviewed in the Introduction, the β-cell MT network consists of an interlocked meshwork and peripheral MT arrays co-aligned with the cell border. We find that blocking kinesin has two major effects on MT organization: (1) receding of MTs at the cell periphery and increased MT density in the cell center, and (2) lack of alignment of remaining peripheral MTs with the cell border. We interpret that the first phenotype arises from the lack of MT sliding from the cell center to the periphery, and the second phenotype arises from the lack of MT sliding along pre-existing peripheral MTs (Figure 7).

In cultured mesenchymal cells, the kinesin motor was reported to populate the cell periphery with MTs via promoting rescues and elongation of radially arranged dynamic MTs (Andreu-Carbó et al., 2022). Our data indicate that this mechanism does not noticeably contribute to kinesin-dependent MT organization in β cells, leaving MT sliding the only known mechanism underlying the phenotype observed here. This difference in kinesin action could be due to basic differences in MT network properties in these cell types: in a mostly non-radial, highly stabilized MT network in β cells (Zhu et al., 2015), modulation of MT plus end rescue efficiency is not likely to be a significant factor.

Interestingly, blocking kinesin results in a striking accumulation of MT in the cell center where they are normally nucleated at MTOCs, which include the centrosome and the Golgi, in differentiated β cells, the latter being the main MTOC. Thus, sliding MTs originate from the MTOC area. At the same time, FIB-SEM analysis did not detect many MTs associated with MTOCs in physiologically normal β cells (Müller et al., 2021). This implies that MTs are typically rapidly dissociated from MTOCs so that they become available for transport by sliding. It is worth mentioning that for long-distance transport by sliding, cargo MTs must be short; otherwise, MT buckling and not long-distance transport will occur (Straube et al., 2006). Interestingly, shorter MTs have been observed in high glucose conditions (Müller et al., 2021) when MTs are nucleated more actively (Trogden et al., 2019) and transported more frequently (this paper). Possibly, nucleated MTs are detached from MTOCs before they achieve a length that would prevent their transport. An intriguing possibility has been proposed, indicating that in high glucose conditions, MTs might undergo severing by katanin (Müller et al., 2021). This process could generate MT fragments, potentially facilitating their role as cargos with increased ease of transport. It is also possible that sliding MT subpopulation has some additional specific features that make them preferred cargos, since it is becoming increasingly clearer in the field that there is immense heterogeneity among MTs. Posttranslational modifications and MT-associated proteins, which vastly alter stability and coordination of motor proteins (Hammond et al., 2008; McKenney et al., 2016; Monroy et al., 2018; Yu et al., 2015), might also influence which MTs serve as cargos versus transportation tracks in β cells.

Importantly, we observe a prominent effect of peripheral MT loss only after a long-term kinesin depletion (3–4 days). This is consistent with our observation that only a minor subset of MT is being moved within each experimental time frame. We postulate that the absence of a peripheral MT array in KIF5B-depleted cells is a consequence of prolonged lack of sliding. We hypothesize that MT sliding must contribute to β-cell-specific peripheral MT bundle formation during β-cell differentiation. We also found that increasing MT sliding does not yield a properly configured MT array: kinesore-treated cells lack aligned peripheral MTs, consistent with kinesore-induced MT looping reported in other cell types (Randall et al., 2017). This indicates that, similar to other parts of β-cell physiology, the dose of MT sliding has to be precisely tuned to achieve physiologically relevant architecture. It was shown before that exaggerated kinesin-dependent MT sliding causes MT bundling and buckling into aberrant configuration (Straube et al., 2006). We predict that a fine-tuning regulatory pathway must exist to restrict the number of MT sliding events to the cell needs.

Consistent with this idea, MT sliding is sensitive to metabolic regulation: our data indicate that MT sliding is activated on a short-term basis after glucose stimulation. While the deep understanding of the role of MT sliding in GSIS requires further studies, it is plausible to suggest two potential functions for this process in glucose-stimulated cells. (1) Given that peripheral MTs are destabilized in high glucose (Ho et al., 2020), we suggest that a long-term function for MT sliding is needed to replace MT population at the cell periphery and restore the pre-stimulus MT organization (Figure 7B). Because the amount of MT polymer on every glucose stimulation changes only slightly (Müller et al., 2021; Zhu et al., 2015) and we detect MT loss from the periphery only after a prolonged blocking of sliding, we reason that this function could be essential to maintain long-term β-cell fitness and prepare cells for repeated rounds of stimulation. (2) As a potential short-term function at each stimulation round, MT sliding within the peripheral bundle itself could rearrange fragmented MTs within this array suppressing its role in IG withdrawal from specific secretion sites (Figure 7B). In this scenario, MT sliding would tilt the balance between positive and negative MT-dependent regulation of GSIS toward enhanced secretion at each stimulation.

Our finding provides an example of a phenomenon where a slight change in MT configuration is physiologically significant. This is not an exception because subtle MT defects often have dramatic consequences. For example, in mitotic spindles, a tiny overgrowth of MT ends during metaphase, which causes them to attach to both kinetochores rather than just one, is very significant for the efficiency of chromosome segregation, causing aneuploidy and cancer. The changes in β-cell MT networks that we are reporting are much stronger: the effect on the peripheral MT network accumulated over 3 days of KIF5B depletion is dramatic (Figure 2B and C). Short-term gross MT network configurations after a single glucose stimulation are harder to detect but are consistent with previous reports that MTs at the cell periphery are destabilized and fragmented upon high glucose stimulus (Ho et al., 2020; Müller et al., 2021), and that preventing this MT rearrangement completely blocks GSIS (Zhu et al., 2015; Ho et al., 2020).

One of the most fascinating features of insulin secretion regulation is that the amount of generated insulin granules significantly exceeds the normal physiological needs for insulin secretion (~100 times more than needed). At the same time, even slightly facilitated glucose depletion can be devastating for the human body. Accordingly, the excessive insulin content of a β cell resulted in the development of multiple levels of control, preventing excessive secretion. Our previous data suggest that the peripheral MT array provides one of those mechanisms. This study indicates that MT sliding is necessary to form the proper peripheral network in the long term. Short-term glucose-induced changes in the peripheral MT array likely need to be subtle to prevent oversecretion. Thus, we are not surprised that a dramatic effect of sliding inhibition is only detectable by our approaches after the changes in the MT network accumulate over time.

On a final note, it is important to evaluate the phenomenon reported here in light of the dual role of KIF5B as IG transporter and MT transporter and the coordination of those two roles in IG transport and availability for secretion. Our results indicate that KIF5B is needed for the formation of the peripheral MT bundle, which we have shown to restrict secretion (Bracey et al., 2020; Ho et al., 2020). At the same time, it is well established that KIF5B transports IGs, and KIF5B loss of function impairs GSIS (Meng et al., 1997, Varadi et al., 2002; Cui et al., 2011). After a prolonged KIF5B inactivation, a loss of peripheral readily releasable IG should be expected due to two factors: because there is no MT bundle to prevent oversecretion and IG depletion, and because there is no new IGs being transported from the Golgi area. In contrast, physiological activation of kinesin by glucose (Donelan et al., 2002; Varadi et al., 2003) would both promote replenishment of IG through nondirectional transport through the cytoplasm and restoration of the peripheral MT array to prevent oversecretion.

In conclusion, here, we add another very important cell type to the list of systems that employ KIF5B-dependent MT sliding to build functional cell-type-specific MT networks. This system is unique because, in this case, MT sliding is metabolically regulated and activated on a single-minute timescale by nutrition triggers.