Kir6.2-deficient mice develop somatosensory dysfunction and axonal loss in the peripheral nerves

Summary Glucose-responsive ATP-sensitive potassium channels (KATP) are expressed in a variety of tissues including nervous systems. The depolarization of the membrane potential induced by glucose may lead to hyperexcitability of neurons and induce excitotoxicity. However, the roles of KATP in the peripheral nervous system (PNS) are poorly understood. Here, we determine the roles of KATP in the PNS using KATP-deficient (Kir6.2-deficient) mice. We demonstrate that neurite outgrowth of dorsal root ganglion (DRG) neurons was reduced by channel closers sulfonylureas. However, a channel opener diazoxide elongated the neurite. KATP subunits were expressed in mouse DRG, and expression of certain subunits including Kir6.2 was increased in diabetic mice. In Kir6.2-deficient mice, the current perception threshold, thermal perception threshold, and sensory nerve conduction velocity were impaired. Electron microscopy revealed a reduction of unmyelinated and small myelinated fibers in the sural nerves. In conclusion, KATP may contribute to the development of peripheral neuropathy.


INTRODUCTION
Among peripheral neuropathies, diabetic polyneuropathy (DPN) is the most frequent. DPN, whose prevalence in diabetic patients is up to 50% or more, is one of the most common diabetic complications and causes non-traumatic amputations of lower limbs (Nomura et al., 2018;Singleton and Smith, 2012). Although anti-hyperglycemic therapies, including insulin therapy and other glucose-lowering medications, have been proven to prevent the onset and progression of DPN (Genuth, 2006), no reliable therapy that has the potential to restore the neurophysiological dysfunction in an advanced stage of DPN has been established (Kobayashi and Zochodne, 2018). In the current study, to suggest a novel pathological mechanism in DPN, we investigated the involvement of ATP-sensitive potassium (K ATP ) channels, which regulate neuron development and neuronal excitability, in the physiology of the peripheral nervous system (PNS).
Voltage-gated potassium channels are known to suppress neuronal excitatory activity in the central and peripheral nervous systems and to be neuroprotective through avoiding excitotoxicity (Deng et al., 2005(Deng et al., , 2011Malin and Nerbonne, 2002). Another type of potassium channel, the inward-rectifying potassium channel, also controls the stability of resting membrane potentials and the excitability of neurons. Among the inward-rectifying potassium channels, the K ATP channels attract attention for their association with metabolic diseases. The K ATP channels are hetero-octamers formed by four pore-forming inward rectifier channel subunits, KIR6.1 (KCNJ8) or KIR6.2 (KCNJ11), and four sulfonylurea receptor subunits, SUR1 (ABCC8) or SUR2 (ABCC9). Of the various types of K ATP channels, the K ATP channel composed of KIR6.2 and SUR1 is most known for its pivotal role in the regulation of glucose-responsive insulin secretion from pancreatic beta cells. The K ATP channels of beta cells set a plasma membrane potential by opening the channels in the resting state. Under increased levels of plasma glucose, the channels close, depolarize the membrane potential, and induce an insulin release following the opening of voltage-gated calcium channels (Lang et al., 2011). In addition to the pancreas, it has been proven that K ATP channels are expressed in a wide variety of tissues including the central nervous system (CNS) (Mourre et al., 1990) in which glucose is the major energy-yielding substrate. As K ATP channels are involved in the peripheral analgesic pathway (Sachs et al., 2004), several analgesic substances including diclofenac, sodium nitroprusside, and morphine activate K ATP channels to induce their systemic antinociceptive effect (Alves et al., 2004;Cunha et al., 2010;Ortiz et al., 2012). The depolarization of the membrane potential induced by glucose leads directly to the hyperexcitability of neurons (Himeno and Nakamura, 2017). Those neurons, so-called glucose-sensing neurons, have recently been investigated as a modulator of glucose homeostasis (Khodai et al., 2018;Labouebe et al., 2016;Stanley et al., 2019). Furthermore, in recent reports, some types of K ATP channels exist on mitochondrial membranes and have critical physiological roles in maintaining intact mitochondrial functions (Coetzee, 2013;Foster et al., 2012). In this context, K ATP channel openers, diazoxide and nicorandil, have been proven to prevent neurodegeneration (Kong et al., 2013;Virgili et al., 2013;Watanabe et al., 2008) and reduce neuronal death in cerebral ischemia of rats (Farkas et al., 2006). It has been indicated that these K ATP channel openers mimic ischemic preconditioning in cardiomyocytes (Coetzee, 2013) and neurons in the CNS (Kis et al., 2003) and provide beneficial effects to these cells.
Neuronal damage in the CNS has been reported after ischemic stress increased in Kir6.2 knockout mice (Sun et al., 2006), in which a plasma membrane potential was depolarized . It has also been proven in the PNS that, after 24 h, hyperglycemia induces depolarization of the resting membrane potential in neurons through closing the K ATP channel and through mechanical hyperalgesia (de Campos Lima et al., 2019). KCNJ11 activating mutations in humans are associated with developmental delay, epilepsy, neonatal diabetes, and other neurological features (Gloyn et al., 2006). Regarding the other potassium inward rectifier channel KCNJ10, patients suffering from KCNJ10 mutations develop EAST syndrome, which is characterized by epilepsy, ataxia, sensorineural deafness, and renal tubulopathy (Reichold et al., 2010). In brief, K ATP channels may have the potential to change in activity and maintain the integrity of both CNS and PNS neurons.
Here, to clarify the physiological significance of K ATP channels in the PNS, we evaluated distributions of proteins that compose K ATP channels in mouse dorsal root ganglion (DRG) neurons. Then, we determined the effects of K ATP channel blockers/activator on the neurite outgrowth of cultured DRG neurons. Thereafter, we evaluated the functional and morphological changes of the PNS in Kir6.2 knockout mice, in which the aberrant K ATP channel causes loss of responsiveness to glucose and sulfonylurea.

Expressions of components of K ATP channels in DRG
The reverse transcriptase-PCR (qPCR) experiments revealed that mRNA of K ATP channel subunits was expressed in murine DRG ( Figure 1). Both pore domains Kir6.1 and Kir6.2 were detected in DRG. The expression levels of Kir6.2 in streptozotocin (STZ)-induced diabetic mice were significantly higher than that of non-diabetic C57BL/6 (BL6) (p < 0.01). The expression level of Kir6.1 in C57BLKS/J-+ Lepr db /+ Lepr db (db/db) diabetic mice was also significantly higher than that of C57BLKS/J (BLKS) mice in their genetic background (p < 0.001) and that of their C57BLKS/J-m+/+Lepr db (db/+) control littermates (p < 0.005) ( Figure S1). Three SURs are known in mice: SUR1, SUR2A, and SUR2B. Among these three SURs, the expression of Sur2A was not verified in DRG (data not shown). The expression of Sur1 and Sur2b increased in STZ-induced diabetic mice (p < 0.01 and p < 0.05, respectively) compared with BL6 ( Figure 1).
Immunostaining revealed the protein expression of SUR1, SUR2B, KIR6.1, and KIR6.2 in neurons and a part of non-neuronal cells (Figures 2A-2D). The protein expression of SUR2A was not able to be detected (data not shown).
Neurite lengths of DRG neurons were shortened with K ATP channel closers and extended with its opener Since axonal degradation precedes depletion of DRG neurons in DPN, neurite outgrowths in a primary culture of DRG neurons are utilized to assess the favorable or unfavorable effects in DPN under various circumstances including the supplementation of chemical compounds or cell stress. Therefore, the neurite outgrowths in DRG cultures were examined to reveal the influences of K ATP channel activity on axonal regeneration ( Figure 3A). K ATP channel inhibition with its closers, glimepiride (non-selective blocker of SUR1 and SUR2) and tolbutamide (selective blocker of SUR1), shortened neurite lengths of DRG neurons ( Figure 3B). In contrast, K ATP channel activation with its opener diazoxide elongated neurite lengths ( Figure 3C). iScience Article On the other hand, tolbutamide and diazoxide had no effects on the neurite outgrowths in DRG neurons obtained from Kir6.2-deficient (Kir6.2 À/À ) mice ( Figures 3D and 3E). However, glibenclamide decreased neurite outgrowths even in DRG neurons from Kir6.2 À/À mice ( Figure 3D).

Unmyelinated fibers in the sural nerves were reduced in Kir6.2-deficient mice
The sural nerve morphometry obtained from 6-month-old mice showed no significant difference between Kir6.2 À/À and BL6 mice. Therefore, we have extended the period of investigation to 18 months old. In the sural nerve morphometry using electron microscopy, the number of unmyelinated nerve fibers decreased in Kir6.2 À/À mice compared with BL6 mice (Table 1). On the other hand, the number of myelinated nerve fibers showed no reduction in Kir6.2 À/À mice. The number and size of axons decreased in unmyelinated fibers (Figures 5A and 5B) but not in myelinated fibers (Table 1). The number and occupancy of Remak bundles, ensheathing multiple unmyelinated axons, in the sural nerve also decreased in Kir6.2 À/À mice (Table 1). In addition, the number and occupancy of C-fibers in Remak bundles decreased in Kir6.2 À/À mice ( Table 1). Morphometry of myelinated fibers in the sural and sciatic nerves (Table S1) showed that iScience Article g-ratio and the total number of Schwann cell nuclei indicated no significant difference between these two types of mice. These results indicate that Schwann cell dysfunction may not be responsible for the morphometric changes in Kir6.2 À/À mice.
Unmyelinated nerve fibers degenerated in the sural nerve of Kir6.2-deficient mice The histogram analysis using percentages of unmyelinated fiber size distribution showed a shift toward smaller sizes in Kir6.2 À/À mice ( Figure 5C). As the number of unmyelinated fibers decreased, we also analyzed absolute fiber numbers in the entire sural nerve for histogram analysis. The histogram using absolute unmyelinated fiber numbers in each size range revealed that this shift was caused by the reduction of numbers in medium and large-sized unmyelinated fibers ( Figure 5D). A characteristic ultrastructural change was that many Remak bundles include empty gaps among axons ( Figure 6A). In some of these empty gaps, possibly caused by inflated Schwann cell tongues, there is an astral scar, implying a degenerated axon ( Figure 6B). Bundles were involved heterogeneously in the unmyelinated axonal loss; an affected bundle was either located next to ( Figures 6A and S2), or connected with, an intact bundle ( Figure S3). Furthermore, Kir6.2 À/À mice exhibited frequent denervated Schwann cell profiles ( Figures 6C and S2).

Myelinated nerve fibers exhibited degeneration and regeneration in the peripheral nerves of Kir6.2-deficient mice
The size-frequency histogram analysis of myelinated fibers in the sural nerve using percentages of axon size distribution showed no significant differences between Kir6.2 À/À and C57BL6/J mice ( Figure S4A). However, the histogram analysis for absolute axon numbers in the entire sural nerve revealed a significant decrease in small to medium-sized axons ( Figure S4B). The histogram using myelin size showed no significant difference between these two mice ( Figure S4C).
Electron microscopy showed that most myelinated fibers had a normal appearance in the sciatic and sural nerves of Kir6.2 À/À mice ( Figure 7A). However, the sciatic nerve frequently contained clusters of regenerating myelinated fibers ( Figures 7B and S5). Occasionally, onion bulb-like lamellated structures were observed ( Figures 7C and S5), which are concentric layers of Schwann cell processes caused by repetitive segmental demyelination and regeneration of myelin.

DISCUSSION
The major findings of the current study are (1) DRG neurons express components of K ATP channels at both transcript and protein levels and the levels of certain transcripts increased in mice with hyperglycemia; (2) K ATP channel inhibition with its closers shortens, but K ATP channel activation with its opener elongates, neurite outgrowths in the primary culture of DRG neurons; (3) Kir6.2 À/À mice present the dysfunction of electrophysiology and sensory perception in the peripheral nerves of the lower extremities; and (4) Kir6.2 À/À mice present ultrastructural abnormalities in peripheral somatosensory nerves.
It has been proven that some components of K ATP channels were expressed (Zoga et al., 2010) and activation of K ATP channels attenuated hyperexcitability in DRG neurons of rats (Du et al., 2011;Luu et al., 2019). In this context, we revealed the transcript levels of K ATP components in murine DRG. We also showed that the constitutive proteins of K ATP channels existed in the DRG using immunostaining. Furthermore, these  iScience Article expression levels in STZ-induced hyperglycemic or db/db mice were increased compared with those in non-diabetic mice. It has been reported in a previous paper that the increased K ATP channel density regulated the protective property against hypoxia in cardiomyocytes (Ranki et al., 2002). Therefore, although the significance of altered expression levels in K ATP channel components in diabetic mice has not yet been clarified, the increase might strengthen the neuroprotective properties of the K ATP channel in DPN. However, as the functional change of K ATP channels cannot be assessed precisely by their expression levels, we will plan to investigate the electrical properties of DRG neurons in diabetic mice in the future.
Although glibenclamide irreversibly binds to SUR1 and SUR2A and has strong effects to accelerate insulin secretion, tolbutamide reversibly binds to SUR1 and inhibits SUR1 but not SUR2A (Gribble and Reimann, 2003). We should consider the possibility that the selective affinity of tolbutamide may induce different behaviors on DRG neurons. Given tolbutamide produced no alteration of neurite outgrowths in DRG neurons obtained from Kir6.2 À/À mice, the selectivity of affinity to SURs might be an important factor in whether each K ATP channel closer has neuroprotective or neurotoxic effects on DRG neurons.
K ATP channel openers, diazoxide and nicorandil, have been proven to prevent neurodegeneration (Kong et al., 2013;Virgili et al., 2013;Watanabe et al., 2008) and reduce neuronal death in cerebral ischemia of rats (Farkas et al., 2006). It has also been published that nicorandil activated endothelial nitric oxide synthase and provided neuroprotective effects as a consequence of reducing oxidative stress on hypoperfusion-induced dementia model mice (Gupta et al., 2016). Although another K ATP channel opener, diazoxide, acts on both SUR1 and SUR2 and inhibits insulin secretion, nicorandil activates only SUR2 and has no influence on insulin secretion. Therefore, nicorandil is widely used as an anti-anginal drug (Moreau et al., 2000) because nicorandil has been proven to have a myocardial protective effect at the onset of ischemia and reduce cardiovascular deaths (Horinaka et al., 2010) via its preconditioning effect (IONA Study Group, 2002;Ishii et al., 2007). Currently, however, no research has suggested long-term usefulness of nicorandil in human diseases of the PNS. We would like to investigate the neuroprotective effects of nicorandil against DPN in clinical settings in the future.
Finally, we used Kir6.2 À/À mice to assess the chronic effects of the K ATP channel on the PNS. In Parkinson's disease, which is one of the neurodegenerative diseases of the central nervous system, activation or closure of K ATP channels is known to promote the death of dopamine neurons (Liss et al., 2005). It has also been reported that the expression of SUR1 is increased in dopamine neurons of the substantia nigra in the human brain with Parkinson's disease (Schiemann et al., 2012). Given these facts, it may be suggested that the increased expression of K ATP channel components may conversely imply a decrease in channel function. In addition, we previously reported that the Kir6.2À/À mice showed depolarized membrane potential and were susceptible to hypoxia-induced seizure (Yamada et al., 2001), suggesting that excitotoxicity may be involved in neurons of these mice. Given the excitotoxicity induced by K ATP channel dysfunction and considering the physiological response of K ATP channels to glucose and sulfonylureas, we evaluated the knockout mice lacking responsiveness to glucose and sulfonylurea as a model that might mimic sustained hyperglycemia or chronic administration of sulfonylureas. Nerve conduction velocities (NCVs) in juvenile Kir6.2 À/À equaled that of wildtype BL6 mice. However, SNCV significantly decreased at 12-24 weeks of age in Kir6.2 À/À compared with BL6 mice. In addition, the knockout mice that were 12 weeks of age or older showed dysfunction of sensory perception. These facts indicate that the K ATP channels assembled with KIR6.2 are uncritical for the development of the PNS but imperative to maintain the integrity of the PNS. Electron microscopy revealed a change of ultrastructure in the sural and sciatic nerves. The neural degeneration primarily occurred in axons, not in Schwann cells. In the axonal degeneration, small fibers, especially unmyelinated fibers, decreased. However, there was evidence of axonal de/re-generation of myelinated fibers and de/re-myelination. As the immunostaining images of DRGs showed the expression of K ATP channel in non-neuronal cells and the findings of demyelination were observed by electron microscopy, we should consider that the degeneration observed in the nerves of Kir6.2 À/À mice may be partly due to the degeneration of Schwann cells. Although these heterogeneous structural changes in the PNS of Kir6.2 À/À mice are consistent with the mixed pathological features of DPN, which consists of prominent axonopathy and less prominent demyelination (Dyck et al., 1986), we should consider the influence of impaired insulin secretion to neurodevelopment of the knockout mice. Although the levels of blood insulin and glucose were not significantly different in the knockout mice from wild-type mice, glucose-stimulated insulin secretion was impaired in the knockout mice . As insulin is an important growth factor for neuronal cells, the impairment of insulin secretion might cause the neuronal impairment in the knockout mice.
Here we verified the significant role of K ATP channels in the homeostasis of the PNS and suggested that the channel could be a new etiology of peripheral neuropathies through demonstrating the neuroprotective ll OPEN ACCESS iScience 25, 103609, January 21, 2022 iScience Article effect of the K ATP channel opener ex vivo. In our further research, we will study the detailed modulatory effects of K ATP channels in the PNS under the diabetic state.

Limitations of the study
First, we should consider the fact that Kir6.2 À/À mice have a complete deficiency of electrophysiological activity of the K ATP channel in pancreatic beta cells and present neonatal transient hypoglycemia followed by normoglycemia with a partial reduction of glucose-stimulated insulin secretion . Therefore, the degeneration of peripheral nerves might be caused by transient hypoglycemia and subsequent impairment of insulin secretion. We will address this question using mice with a PNS-specific Kir6.2 deficiency in the future.
Second, we have not explained the discrepancy between the cellular hyperexcitability and the dysfunction of sensory perception in 12 weeks of age or older Kir6.2 À/À mice. Although the absence of K ATP channel should decrease the excitability thresholds of sensory neurons, neurological functions of Kir6.2 À/À mice showed a decrease in sensitivity compared with wild-type mice. Regarding b-cell function, although these knockout mice transiently exhibited neonatal hypoglycemia, levels of blood insulin and glucose were not significantly different in adult knockout mice from wild-type mice . Given the fact, it is possible that the knockout mice did not exhibit hyperalgesia in this study because the experiments were conducted after infancy and that they might exhibit transient hyperalgesia during the neonatal period.
Third, although we used Kir6.2 À/À mice to elucidate the pathophysiology of DPN, there is insufficient evidence of the involvement of DPN and K ATP channels. The molecular mechanisms of insulin secretion involving K ATP channels and the increase of K ATP channel components in DRG of diabetic mice might indicate the relevance of K ATP and DPN. However, no direct evidence has been accumulated. Therefore, we should consider that the Kir6.2 À/À mice may be applicable to explore other peripheral neuropathies.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Tatsuhito Himeno (thimeno@aichi-med-u.ac.jp).

Materials availability
This study did not generate new unique materials.

Data and code availability
All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. To determine a nociceptive threshold, CPT was measured in 4-, 12-, and 24-week Kir6.2 -/and age-matched BL6 mice (n = 5-15 in each group) using a CPT/LAB Neurometer TM (Neurotron, Denver, CO). Each mouse was settled in a Ballman cage (Natsume Seisakusho, Tokyo, Japan) suitable for light restraint to keep awake. Two electrodes (SRE-0405-8; Neurotron) were attached to the plantar of one foot. Transcutaneous-sine wave electrical stimuli with three different frequencies (2000, 250, and 5 Hz) were applied to each plantar. The intensity of each stimulus gradually increased automatically (increments of 0.01 mA for 5 and 250 Hz, increments of 0.02 mA for 2000 Hz). The minimum intensity at which a mouse withdrew its paw was defined as the CPT. Six consecutive measurements were conducted at each frequency. Six consecutive measurements in each mouse were conducted at each frequency. Mice who were not settled in the cage were excluded from the experiment.

ll
OPEN ACCESS