Primary cilia sensitize insulin receptor-mediated negative feedback in pancreatic β cells

Insulin receptors (IR) can localize to the primary cilia of pancreatic β cells. Because primary cilia are known to sensitize or bias the signaling of cell surface receptors, we investigate how ciliary insulin receptors influence glucose-stimulated insulin secretion (GSIS) in β cells by gauging how cytosolic calcium concentration changes in a mouse insulinoma cell line (MIN6). Purified recombinant insulin suppresses calcium elevation in response to glucose in these cells. Interestingly, ciliated cells show attenuated cytosolic calcium elevation compared to cilium-free cells after glucose stimulation even in the absence of exogenous insulin. We observe that ciliary IR is highly phosphorylated, and the phospho-IR density decreases when cells are either treated with an insulin receptor (IR) inhibitor, BMS536924, or ciliary function is disrupted through either IFT88 or BBS1 knockdown. Consistently, the attenuation of calcium elevation in ciliated cells is abrogated when cells are either treated with IR inhibitor or when primary cilia are impaired. We further demonstrate that ciliary IR signaling hyperpolarizes the plasma membrane but has no apparent impact on glucose-induced ATP production. Thus, our results argue that primary cilia sensitize insulin receptor signaling and mediate negative feedback in GSIS in pancreatic β cells.


Introduction
Pancreatic beta cells (β cells) are responsible for secreting insulin and maintain energy homeostasis in the body. Upon elevation of blood glucose, β cells accelerate glucose metabolism and increase the production of ATP. The primary outcome of such changes is the increase of ATP/ADP ratio in the cytosol, which causes the closure of ATP-sensitive potassium channels and membrane depolarization. The subsequent opening of the voltage-dependent calcium channels leads to the elevation of cytosolic calcium, which triggers insulin exocytosis (MacDonald et al., 2005). How β cells sense and launch suitable responses to glucose change remains an active area of diabetes research.
Insulin feedback are thought to regulate β cell function. Insulin signals through either the insulin receptor (IR) or the insulin-like growth factor 1 receptor (IGF1R) (White, 2003). In addition, there are two isoforms of IR, insulin receptor A (IR-A) and insulin receptor B (IR-B), which differ by 12 amino acids (Seino et al., 1990) but signal differently (Leibiger et al., 2001). All of these receptors are expressed in pancreatic β cells, and their signaling are important for maintaining β cell mass, growth, insulin synthesis, and secretion in the long run (reviewed in Leibiger et al., 2008). However, the acute effects of insulin receptor signaling on secretion are more elusive.
Some studies report that insulin or insulin mimetics suppresses insulin secretion in humans, mice or in isolated islets (Elahi et al., 1982, Argoud et al., 1987, Persaud et al., 2002, Zhang et al., 1999. Phosphoinositide 3-kinase (PI3K)-dependent activation of the ATP-sensitive potassium channels are thought to mediate this effect (Khan et al., 2001). Openings ATPsensitive potassium channels lead to membrane hyperpolarization, thus inhibiting the electrical activity and calcium elevation of β cells. However, other studies report insulin or insulin mimetics either have dual effects on insulin secretion (Jimenez-Feltstrom et al., 2004) or promote exocytosis (Roper et al., 2002, Bouche et al., 2010, Aoyagi et al., 2010, Aspinwall et al., 1999, Aspinwall et al., 2000, Hisanaga et al., 2009. The stimulatory effects of insulin on insulin secretion is also dependent on PI3K activity. IR signaling elevates cytosolic calcium through either inhibiting calcium sequestering into the endoplasmic reticulum (ER) or through the translocation of transient receptor potential cation channel subfamily V member 2 (TRPV2), a calcium-permeable channel (Ramsey et al., 2006), from cytosol to the plasma membrane.
Recently, primary cilium has attracted much interest in diabetes research due to the strong association between type 2 diabetes and ciliopathies. Ciliopathies refer to a collection of several rare disorders linked to mutations in primary cilium proteins, such as Alstrom syndrome and Bardet-Biedl syndrome. Patients of Alstrom or Bardet-Biedl syndrome have frequent (~50 or greater) onsets of type 2 diabetes (diIorio et al., 2014). Studies in mice also linked ciliary dysfunction to abnormal insulin secretion (Gerdes et al., 2014, Zhang et al., 2005. Most if not all pancreatic β cells possess a primary cilium in vivo (Gan et al., 2017). Interestingly, IR-A was found enriched in the primary cilium immediately after insulin stimulation (Gerdes et al., 2014).
Previous studies have shown that primary cilia can potentiate or suppress the signaling of their resident receptors, or modulate receptor signaling by influencing the preference of its downstream effectors (reviewed in Schou et al., 2015, Christensen et al., 2011. This has led us to hypothesize that ciliary insulin receptor may acutely affect the function of pancreatic β-cells.
In this study, we aim to find out the impact of IR signaling on GSIS and the role of primary cilium in modulating IR signaling. We demonstrate that insulin has a negative impact on glucoseinduced calcium elevation in a β cell line in culture. We further demonstrate that primary cilium potentiates this effect by phosphorylating insulin receptor at residue concentration of insulin.

Immunofluorescence
MIN6 cells were seeded on fibronectin-coated cover glasses 3 days prior to experiments.

Imaging setup
All time-lapse imaging experiments were performed on a customized Nikon Ti-E inverted microscope. To improve signal noise ratio, total internal reflection fluorescence microscopy (TIRFM) was performed using a 60x oil TIRF objective (NA 1.49). The penetration depth of the evanescent waves from different laser lines were calibrated as described previously (Ref PMID 19816922). The microscope was modified with a "stage-up" design which enables the insertion of two independent, motorized dichroic mirrors/filter cubes in the microscope infinity space. A dichroic mirror in the bottom cube was used to reflect excitation laser lines at 488, 561, 640 nm (Coherent OBIS) for imaging of Cal520 (AAT Bioquest, Sunnyvake, CA), R-GECO, Di-4-ANEQ(F)PTEA (Potentioetric Probe, Farmington, CT), respectively. The fluorescent emission was captured with an EMCCD camera (iXon Ultra, Andor). Live cell imaging was performed at 37°C in a heated chamber (Bioptechs) with a humidified 5% CO2 supply. Customized vitamin and phenol red-free DMEM medium (US Biological) was used in live cell imaging to reduce fluorescence background and photobleaching. For imaging of immunofluorescence labeled cells, a Nikon TE2000 microscope equipped with a CSU-10 spinning disk confocal unit (Yokogawa) was used. The laser excitation light sources were from Coherent OBIS (405nm) and Melles Griot (488, 568 and 647nm). Metamorph software (Molecular Devices) was used to control both imaging setups.

Measurement and analysis of cytosolic calcium concentration
The calcium fluctuations in MIN6 cells were tracked using either R-GECO (lentivirally tranduced stable cell line) or Cal520 (labeled at 0.2 uM for 30min). The cells were seeded on fibronectin- Fluorescence images of R-GECO or Cal520 were collected and background subtracted. For each trace, fluorescence intensity is normalized to the minimum value before stimulation. Pvalues are calculated using the Wilcoxon-Mann-Whitney test or student t-test. The type of test to be used in each situation is determined by the P-value computed using the Shapiro-Wilk test to see whether the sample follows a normal distribution. All computations are done using Python 3.7 (code available at https://github.com/yuezheli/primary_cilia_sensitize_insulin_receptor-mediated_negative_feedback).

Measurement and analysis of membrane potential
MIN6 cells that express primary cilium marker, Arl13b-mScarlet, were seeded on cover glasses 3 days before experiments. Cells were pre-incubated with DMEM with 3mM glucose for 30 minutes before stained with Di-4-ANEQ(F)PTEA (3uM, Potentiometric Probes, Farmington, CT, USA) for 15 minutes. Fluorescence images were collected under either 488 or 640 excitation.
Then images went through background subtraction, shade correction, and then emission ratio images under 488 and 640 excitations were generated and quantified based on the plasma membrane labeled by the Arl13b-mScarlet. P values were calculated using Student t test.

Insulin alone fails to promote insulin secretion in MIN6 cells
To investigate how insulin feedback specifically affects pancreatic beta cells, we lentivirally transduced mouse insulinoma MIN6 cells with R-GECO (Zhao et al., 2011), a red fluorescence emitting genetically-encoded calcium indicator, to gauge the change of cytosolic calcium in response to glucose. The transduced MIN6 cells were glucose and serum-starved prior to stimulation. When the cells were stimulated with glucose, we observed an increase in R-GECO intensity ( Fig 1A, B), suggesting an increase in cytosolic calcium. Statistical analysis shows that the calcium elevation was statistically significant ( Fig 1C). As a control for perfusion artifact, stimulating MIN6 cells with low glucose solution did not show a statistically significant change in cytosolic calcium (Fig 1F-H). Additionally, we demonstrated insulin secretion by expressing a GFP-tagged insulin in MIN6 cells. When insulin was secreted, insulin-containing granules generated a sudden increase of fluorescence intensity before vanishing due to diffusion ( Fig   1D). The increase of fluorescence intensity was due to a pH change from the acidic secretory granule to the neutral pH environment outside the cell. On average, we observed 5.8 (±1.48, n= 15 cells) insulin granules being secreted per minute ( Fig 1D). Thus, MIN6 cells elevate calcium and trigger insulin secretion in response to glucose.

Figure 1. Glucose induced calcium elevation and insulin secretion in MIN6 cells.
MIN6 cells that stably express the R-GECO sensor were treated with glucose. The example sensor images before and after treatment (A) and a plot of the normalized fluorescence intensities in time (B) are shown. The quantifications of normalized mean fluorescence intensities are shown in (C). As a control, addition of imaging solution was used to address any potential noise induced by perfusion (E-G). MIN6 cells that express insulin-GFP were treated with 16.7mM glucose (D). A cropped set of sequential images at 1.04s time intervals are shown. See main text for quantifications.
We further validated that insulin receptor (IR) signaling was attenuated after glucose and serum starvation, and insulin triggered IR signaling. Using the phosphorylation level of insulin receptors to gauge the strength of IR signaling, we found a ~50% reduction of IR phosphorylation (Supp. Since insulin is known to trigger cytosolic calcium elevation and exocytosis (Aoyagi et al., 2010, Aspinwall et al., 1999, Aspinwall et al., 2000, Hisanaga et al., 2009, we investigated whether exogenous insulin can impact insulin secretion by measuring how cytosolic calcium changes post-insulin stimulation. While cells exhibited different basal states among experiments, we found no statistically significant changes in cytosolic calcium concentration after stimulation with human recombinant insulin at concentrations from 0.1nM to 3.5uM ( Supp Fig 2A-C). To investigate the inconsistencies between previous reports and our results, we validated that the insulin source used in our study can induce IR phosphorylation at a concentration as low as 0.1nM (Supp. Fig. 1E). In the concentration range of 0.1nM to 70nM, we found ~20%-40% IR phosphorylation induction (Supp. Fig. 1F). Increasing the insulin concentration also increased phospho-IR concentration: 65% induction after 175nM insulin stimulation, 137% induction after 3.5uM insulin stimulation (Supp. Fig. 1F). 175nM IGF1 treatment led to dramatic induction (~184%). Conversely, we repeated the published experiments (Aoyagi et al., 2010, Aspinwall et al., 1999, Aspinwall et al., 2000, Hisanaga et al., 2009, using >100 nM (175nM) commercially purified bovine insulin to stimulate MIN6 cells. We detected bovine insulin-induced calcium elevation following treatment (Fig 2A, B, upper two rows; Fig 2C, left and middle panel).This contrasted our observations in cells treated with recombinant human insulin ( Supp Fig 2A, Fig 1C, D). We were also able to validate that inhibiting PI3K attenuates bovine insulin-induced calcium elevation (Fig 2A-

Insulin receptor signaling inhibits glucose-stimulated insulin secretion
Because of the limited brightness of R-GECO sensor, we could only track the calcium changes for no more than 2 minutes after insulin stimulation. Longer or more frequent sensor acquisition often led to increased photobleaching or phototoxicity. Thus, we switched to Cal520, a calcium-sensitive dye, to better track how MIN6 cells may respond to glucose. Consistent with the data obtained from the R-GECO sensor, we observed rapid calcium elevation after glucose addition, reaching its peak values around 2 minutes after glucose addition (Fig 3A, top panel). The calcium elevation gradually attenuated but remained elevated above the initial values for the remainder of the extended 10 minutes of acquisition period (Fig 3B, darker gray bars).
Interestingly, co-stimulation of MIN6 cells with glucose and a high concentration (3.5uM) of insulin drastically reduced glucose-stimulated calcium elevation (Fig 3A, second Fig   3C). Further control experiments showed that BMS536924 alone did not alter the initial cytosolic calcium concentration, regardless cells are in low glucose environment or in high glucose environment (Fig. 3D). Thus, these data argue that, for the observed negative feedback to be physiologically relevant, mechanisms that can sensitize the insulin receptor or enrich the insulin ligand would be required in vivo.

Primary cilia promote IR activation and suppress glucosestimulated calcium elevation
Primary cilia are known to sensitize cell surface receptors. It is reported that most, if not all the pancreatic β cells are ciliated in vivo (Gan et al., 2017). However, in MIN6 cells, only ~20-40% of cells are ciliated (Supp. Fig 3G, top panel; Supp. Fig 3H, left bar). Thus, it is possible that ciliated MIN6 cells might have responded to the endogenous insulin, but we failed to detect the effect in bulk measurement due to their small number. To delineate whether primary cilium has a role in glucose-stimulated calcium elevation, we distinguished ciliated cells from cilium-free cells through the expression of Arl13b-mScarlet, a primary cilium marker, in MIN6 cells. By separating the calcium elevation data of ciliated cells from those of cilium-free cells, we found a near 50% reduction of glucose-stimulated calcium elevation in ciliated cells ( Fig. 4A-C, Video 1).
The cilium-free cells remained sensitive to exogenous insulin and showed suppressed response to co-stimulation of glucose, but no further reduction could be detected in ciliated cells upon insulin stimulation (Fig. 4D-F). Conversely, only ciliated cells responded in calcium elevation to the IR inhibitor, BMS536924, and recovered to the level of cilia-free cells (Fig 4G-I). These data suggested that primary cilia sensitize MIN6 cells to low insulin concentrations and negatively impact cells' response to glucose. Primary cilia are pointed out with white arrowheads. Scale bar = 20um. Error bars are standard errors.

Video 1 Ciliated MIN6 cell showed attenuated glucose-induced calcium elevation. MIN6 cells stably expressing
Arl13b, a primary cilium marker, were labeled with Cal520. The cells were treated with glucose alone. Colormap displays the changes in cytosolic calcium concentration. Ciliated cell is pointed out with white arrowheads.
To examine whether IR is activated in the primary cilia of MIN6 cells, we probed for phosphorylation of the endogenous IR receptor. Cilia-free cells responded to insulin with a significant increase of IR phosphorylation as expected (Fig 5C, left 2 bars). In contrast, we observed strong phosphorylation of IR in primary cilia even in serum-starved cells (Fig 5A, top panel). Adding exogenous insulin had little impact on the ciliary phospho-IR signal, suggesting that the receptor is already highly activated in the primary cilia ( Fig 5B, left 2 bars). To further support this idea, we treated cells with sodium orthovanadate, a phosphatase inhibitor (Fig 5D) to saturate the IR phosphorylation. A significant increase of IR phosphorylation in the cilium-free area (Fig 5E) could be observed. However, ciliary IR phosphorylation showed only a minor increase and was not statistically significant (Fig 5F). In addition to the high levels of IR phosphorylation in the primary cilia, the ciliary IR receptor was resistant to the IR inhibitor, BMS536924 (Fig 5A, bottom panel). We found that phospho-IR signal was reduced but remained prominent in the primary cilia, while the phospho-IR signal was completely eliminated outside of cilia or in cilia-free cells (Fig 5B, first and last bar; Fig 5C, first and last bar). These data demonstrated that ciliary IR is highly phosphorylated and is resistant to receptor dephosphorylation.

Ciliary trafficking is required for IR activation and its suppression of glucose-stimulated calcium elevation
To examine if disrupting ciliary trafficking changes IR phosphorylation, we used siRNA to knockdown IFT88 or BBS1, two proteins known to be involved in ciliary protein trafficking (Starks et al., 2015, Malicki and Avidor-Reiss, 2014, Taschner et al., 2011. A knockdown of these two genes of more than 50% at the protein level was confirmed by western blot analysis (Supp. Fig 3A-D). Consistent with the critical role of IFT88 in cilium assembly (Kim et al., 2010), we observed a reduction of fraction of ciliated cells in IFT88-knockdown cells as demonstrated by co-staining 2 primary cilium markers, acetylated ɑ-tubulin, and Arl13b (Supp. Fig. 3G, H). BBS1-knockdown did not significantly alter the fraction of ciliated cells, but it is known to disrupt ciliary function (Malicki and Avidor-Reiss, 2014). Knockdown of either IFT88 or BBS1, resulted in a decrease in density of phosphorylated IR in primary cilium (Fig 5A, 3rd-5th row, Fig 5B), with no impact on IR phosphorylation outside of primary cilium (Fig 5C). Therefore, defects in IR trafficking may lead to less IR being trafficked into primary cilium, contributing to a decrease in phospho-IR density in the primary cilia.
To determine whether knocking down IFT88 or BBS1 affects IR trafficking in or out of the primary cilia, we ectopically expressed a GFP-tagged constitutively active insulin receptor A (ca IR-A-GFP). Consistent with a previous study (Gerdes et al., 2014), we found that ca IR-A-GFP was localized in the primary cilia in siRNA control cells (Supp. Fig 3I, upper panel; Fig 3J,  To investigate the functional significance of knocking down IFT88 and BBS1, we examined how these knockdown cells elevate calcium in response to glucose stimulation. To reduce cell heterogeneity and siRNA-associated artifact, we continued to rely on the ciliary marker to directly compare between ciliated and cilium-free cells in the same culture. In siRNA control cells, we continued to observe a reduced response to glucose stimulation in ciliated cells compared to cilium-free cells (Fig 6A, B, top panels, Fig 6C). However, knocking down either BBS1 or IFT88 completely abolished the difference in calcium elevation between ciliated and cilium-free cells (Fig 6A, B, middle two panels; Fig 6D, E). In addition to perturbations of ciliary trafficking, we also used siRNA to specifically knockdown IR. Knocking down IR also eliminated the difference in calcium elevation between ciliated and cilium-free cells (Fig 6A, B, bottom panel; Fig. 6F; knockdown validation in Supp. Fig. 3E, F). Thus, these results support that ciliary IR signaling suppresses cells' response to glucose stimulation. Figure 6. Knockdown of either genes related to ciliary trafficking or IR abolished the inhibition on glucoseinduced calcium elevation in ciliated cells. MIN6 cells stably expressing Arl13b-mScarlet were transfected with siRNA control, or siRNA against IFT88, BBS1 or IR. The cells were labeled with Cal520 and treated with glucose as described. The example images (A), and Cal520 intensity trajectories (B), and their quantifications divided between ciliated and cilium-free cells (C-F) are shown. Primary cilia are pointed out with white arrowheads. Scale bar = 20um. Error bars are standard errors.

Ciliary IR signaling hyperpolarizes the plasma membrane of ciliated beta cells
To investigate the molecular mechanism on how primary cilium asserts a negative impact on GSIS, we evaluated two potential drives behind this cilium-dependent suppression on glucosestimulated calcium elevation: glucose metabolism, and membrane potential. We used PercevalHR, a biosensor that fluorescence intensity positively correlates with ATP:ADP ratio (Tantama et al., 2013). Upon glucose stimulation, we failed to observe any difference in the fluorescence intensity change of PercevalHR in ciliated and cilium-free cells (Fig 7A; Fig 7B), suggesting ciliary IR signaling has limited impact on glucose metabolism. Since IR signaling was shown to induce membrane hyperpolarization, and thereby inhibit insulin secretion in β cells (Khan et al., 2001), we used a voltage-sensitive dye (Yan et al., 2012) to compare the membrane potentials between ciliated and cilium-free cells. The emission ratio between 488 and 640 excitations is positively correlated with cells' membrane potential. In control experiments, we stimulated MIN6 cells with tetraethylammonium (TEA), a potassium channel blocker and known to depolarize the plasma membrane (Ni et al., 2010), and observed a ~5% increase in the 488/ 640 ratio (Supp. Fig 4A, B). To avoid interference from electronical coupling between adjacent cells (Speier et al., 2007, Benninger et al., 2008, we only quantified ciliumfree cells that were not adjacent to ciliated cells. We observed that the 488/ 640 ratio in ciliated cells were ~10% lower than cilium-free cells (Fig 7C, D). Furthermore, the difference was abolished by treating the cells with the IR inhibitor BMS536924 (Fig 7D right bars). All these results indicate that ciliary IR signaling at least hyperpolarizes plasma membrane of β cells.

Discussion
In this study, we revisited the historically opposing observations that were made on insulin feedback in pancreatic β cells. We showed that insulin receptor signaling suppresses glucoseinduced calcium elevation in insulinoma MIN6 cells. The negative effect of insulin was reported previously by others (Elahi et al., 1982, Argoud et al., 1987, Persaud et al., 2002, Zhang et al., 1999, Jimenez-Feltstrom et al., 2004. On average, cells stimulated with glucose show a 100% induction calcium elevation, but those that are stimulated with glucose and insulin at the same time have only ~50% of the induction. We also showed that insulin alone has little impact on cytosolic calcium concentration. This is in contrast to what was reported in the literature that bovine insulin increases cytosolic calcium concentration and leads to exocytosis (Roper et al., 2002, Bouche et al., 2010, Aoyagi et al., 2010, Aspinwall et al., 1999, Aspinwall et al., 2000, Hisanaga et al., 2009, Ramsey et al., 2006. We showed these observations are likely due to reagent artifacts from crude bovine insulin. We speculate that the artifacts could be caused by other contaminating molecules secreted by cells in pancreatic islets, such as pituitary adenylate cyclase-activating polypeptide, vasoactive intestinal peptide, and glucagon-like peptide-1 (GLP-1), which are known to potentiate GSIS (Di Cairano et al., 2015, Dickson and Finlayson, 2008, Xin et al., 2016, Segerstolpe et al., 2016, Winzell and Ahren, 2007, Yamada et al., 2004. While the precise concentrations of insulin that β cells expose to in vivo are unknown, we examined in this study a broad range of concentrations of insulin on how they affect receptor phosphorylation. We observed insulin-induced insulin receptor phosphorylation plateaued between 0.1nM and 70nM of exogenous insulin stimulation, but unexpectedly, a further increase in exogenous insulin concentration led to a further increase in IR phosphorylation. IGF1 even imposed a more dramatic induction of insulin receptor phosphorylation. We speculate that this increase in insulin/ IGF-induced receptor phosphorylation could be driven by IR -IGF1R hybrid receptors (Benyoucef et al., 2007). These hybrid receptors have a lower affinity to insulin compared to insulin receptors but have a higher affinity of IGF1. The unexpected increase in phospho-IR signal could be due to our phospho-IR antibody that can pick up phospho-IGF1R as well.
Our observation that ciliated cells have less GSIS appears to be in contrast to the report in the literature that both BBS4 -/-and Tg737 ORPK mice have impaired GSIS (Gerdes et al., 2014, Zhang et al., 2004. We saw similar impacts of disrupting primary cilia or their function in siRNA experiments (Fig 5 C-E), but the focus of the study was on specifically the ciliary IR instead of the primary cilia compartment in general. We speculate that ciliopathy mouse models have impaired GSIS is more likely due to the long-term effect of malfunction of primary cilia. For example, the disruption of cell polarity (Walz, 2017, Boehlke et al., 2010 may disrupt the proper release of insulin, as most of the secretion takes places towards the vasculature (Low et al., 2014) or into the interstitial space (Takahashi et al., 2002). Alternatively, primary cilia may also modulate GSIS by altering the energy homeostasis either through other receptors (Volta et al., 2019) or other parts of an organism that can influence pancreatic β cells (e.g. brain, adipose tissue) (Oh et al., 2015).
We found that primary cilia potentiate insulin receptor phosphorylation and suppress calcium elevation in the absence of exogenous insulin. The sensitized ciliary IR phosphorylation could be due to a lack of phosphatase activity inside the primary cilia because ciliary IR phosphorylation appeared to resist the treatment of IR or phosphatase inhibitor. The elevated IR signaling from primary cilium could also be the result of cilia's high surface-to-volume ratio, an environment that promotes access of the activated IR to concentrated or specific downstream targets. This phenomenon is not new, as Zhu et al., 2009 reported that ciliary IGF1R is sensitized to ligand stimulation. However, it is not clear how ciliary IR signaling propagates from primary cilia to the rest of the cells. One possibility is that ciliary IR signaling increases ATP-sensitive potassium channels' overall conductance through a PI3K-dependent pathway, resulting in suppressed GSIS. This is consistent with our observation, and literature suggests that IR signaling, in general, increases the overall conductance of ATP-sensitive potassium channels in a PI3K-dependent manner. (Khan et al., 2001, Zaika et al., 2016, Xu et al., 2015. This is also consistent with the observation in Gerdes et al., 2014 that disrupting primary cilia resulted in an altered PI3K and Akt phosphorylation. However, the caveat of this idea is that the downstream effectors, IRS-1, does not go into primary cilia. It is reported that IRS-1, phospho-IRS-1, Akt, and phospho-Akt are detected at the basal bodies, adjacent but outside the primary cilia (Zhu et al., 2009). Thus, it is not clear how IR inside primary cilia phosphorylates IRS-1 outside primary cilia. Alternatively, ciliary IR may signal by cross-talks between receptor tyrosine kinases and G-protein coupled receptors (Gavi et al., 2006). This may enable primary cilia to signal through cAMP to modulate β cell physiology.
In vivo, primary cilia are localized at the apical surface of beta cells, exposed to an extracellular lumen that is sealed off by tight junctions and is distant from vasculature (Gan et al., 2017). Thus, the exact insulin concentration that the beta cells are exposed to in vivo is difficult to predict. We speculate this extracellular lumen may have a lower insulin concentration and may be less subject to transient fluctuations due to the presence of surrounding tight junctions. It is likely that the negative feedback effect we uncovered in this study is related to the homeostasis of beta cells.
The attenuation of the calcium elevation may provide a tonic inhibition to prevent beta cells from burnout and promoting β cell survival (Bae et al., 2019, Yoshimura et al., 2010, Luciani et al., 2013, Prentki and Nolan, 2006).

Author's contribution
YW and YL conceive the project. YL and PKS validate the cell line. YL characterizes MIN6 cells' response to different stimuli under different conditions, ciliary insulin receptor signaling and enrichment, and membrane potential difference between ciliated and cilium-free cells. YW and YL write the manuscript.