We examined whether SpTx1 had an influence on the fasting blood glucose levels of wild-type mice. The blood glucose level of the overnight-fasted wild-type mice was 173 (±3.18) mg/dl in the following morning (Figure 1A). Given that the blood glucose level of wild-type mice can be lowered by glibenclamide (Remedi and Nichols, 2008), we used it as a positive control. As expected, a single intraperitoneal application of glibenclamide (40 mg/kg body weight) lowered the blood glucose level of wild-type mice by half at the end of a 2-hr observation period whereas that of the vehicle dimethyl sulfoxide (DMSO) did not have any meaningful effect. To assess the effects of SpTx1 on the blood glucose levels, intravenous (IV) injection was used to administer SpTx1 to avoid the interpretation of the result being confounded by the toxin’s bioavailability. SpTx1 at a dose of 1 mg/kg neither had discernible effect on the blood glucose levels of wild-type mice (Figure 1B) nor caused other noticeable differences during the observation period, when compared to wild-type mice administered with the vehicle saline. Using total blood volume, this dose of SpTx1 is calculated to be 3 µM, ~200 times the K_d value of SpTx1 against hKir6.2. The finding that SpTx1, estimated to be at 3 µM in the blood, had no effect on the blood glucose level, suggests that the toxin does not have any consequential actions (in the present context) by either inhibiting mouse Kir6.2 (mKir6.2) channels or acting on unintended targets in wild-type mice.

Figure 1

Download asset Open asset

Figure 1—source data 1

Related to Figure 1A–C.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig1-data1-v1.xlsx

Download elife-77026-fig1-data1-v1.xlsx

We tested whether SpTx1 affected the amount of glucose-stimulated insulin secretion (GSIS) from β cells in isolated pancreatic islets of wild-type mice using two different stimulating concentrations of glucose. To perform this test, we adopted a static assay of insulin release from isolated islets incubated in microwell plates (Remedi et al., 2009). We used high concentrations of SpTx1 between 0.2 and 1 μM in the bath solutions because there was likely a SpTx1 concentration gradient from the extracellular solution to the interior of an islet. In this concentration range, SpTx1 failed to alter insulin secretion in the presence of either 7 or 15 mM glucose (Figure 1C). Thus, SpTx1 had no significant effects on GSIS either through acting on mKir6.2 or any unintended targets within the wild-type mouse islets.

The prevailing excitation–secretion coupling paradigm regarding β cells postulates that a blockade of K_ATP channels in pancreatic β cells markedly depolarizes V_m to trigger insulin release. Previous studies reported that dissociated β cells exhibited altered electrophysiological properties, for example, a much greater cell-to-cell variation in glucose sensitivity, compared with the cells in isolated islets (Salomon and Meda, 1986; Scarl et al., 2019). Thus, V_m responses of individual cells in isolated but intact islets were recorded in the perforated whole-cell mode (Figure 2). In a typical mouse islet, up to 80% of the cells are β cells (Steiner et al., 2010). In our measurements, β cells were identified by their characteristic glucose-stimulated V_m depolarization and subsequent bursts of (non-overshooting) action potentials. Illustrative V_m changes in responses to an increase of the extracellular glucose concentration to 15 mM from the starting concentration of 0, 5, or 8 mM are shown in Figure 2A–C. For example, increasing the glucose concentration from 0 to 15 mM led to V_m depolarization and a burst of action potentials in a reversible manner (Figure 2A, blue and black segments), thus functionally confirming that the recorded cell was a β cell. Subsequent application of SpTx1 at 0.2 µM had little effect on V_m (Figure 2A, orange segment) but that of the sulfonylurea glibenclamide at 0.4 µM dramatically depolarized V_m and triggered a train of action potentials (Figure 2A, purple segment). Similar results documenting the ineffectiveness of SpTx1 and the contrasting effectiveness of glibenclamide in inducing V_m depolarization were also observed using the starting extracellular glucose concentrations of 5 and 8 mM (Figure 2B, C). Low-pass filtered (Figure 2A–C, light gray traces) time-averaged V_m values from multiple islets under the three different glucose conditions are summarized in Figure 2D–F. Thus, in wild-type mouse islets, glucose and glibenclamide dramatically depolarize V_m but SpTx1, despite its high concentration (>10 times the K_d value for hKir6.2), has no such effect. These results also serve as an additional control study to determine if SpTx1 has impactful off-targets in the islets.

Figure 2

Download asset Open asset

Figure 2—source data 1

Related to Figure 2A–F.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig2-data1-v1.xlsx

Download elife-77026-fig2-data1-v1.xlsx

The most likely cause for the failure of SpTx1 to depolarize V_m and to potentiate GSIS in wild-type mouse β cells is that SpTx1 does not potently inhibit mKir6.2. Indeed, the amino acid sequences of hKir6.2 and mKir6.2 are extremely similar but not identical (96% identity). To determine whether SpTx1 targets the two K_ATP channel orthologs with different affinities, we examined mKir6.2 and hKir6.2 coexpressed with their respective SUR1 in Xenopus oocytes so that we could directly compare the effect of SpTx1 on mK_ATP and hK_ATP channels under the same conditions (Figure 3).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Related to Figure 3A–F.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig3-data1-v1.xlsx

Download elife-77026-fig3-data1-v1.xlsx

As expected, 10 nM SpTx1 suppressed currents through hK_ATP channels comprised of hKir6.2 and hSUR1 (hKir6.2 + hSUR1) by about half (Figure 3A; Ramu et al., 2018). In contrast, a comparable suppression of currents through mK_ATP channels, each of which consists of mKir6.2 and mSUR1 (mKir6.2 + mSUR1), required 1 µM SpTx1, 100 times higher concentration (Figure 3B). The concentration dependence of current inhibition by SpTx1 was fitted using a model for the toxin-to-channel interaction with one-to-one stoichiometry, yielding an apparent K_d of 15 and 673 nM for hK_ATP and mK_ATP, respectively (Figure 3E). For ease of comparison, all apparent K_d values in the entire study are summarized in Figure 3—figure supplement 1. Clearly, SpTx1 inhibits mK_ATP with a much lower potency than hK_ATP.

It is imperative to exclude the possibility that some differences between the auxiliary hSUR1 and mSUR1 subunits are the primary causes of the aforementioned large difference in apparent SpTx1 affinities of the two orthologous channel complexes. First, we compared the effects of SpTx1 on currents through channels that consisted of hKir6.2 + hSUR1, hKir6.2 + mSUR1, mKir6.2 + mSUR1, and mKir6.2 + hSUR1 (Figure 3A–D). SpTx1 inhibited currents through hKir6.2 + hSUR1 and hKir6.2 + mSUR1 with high affinities but, in contrast, the toxin inhibited currents through mKir6.2 + mSUR1 and mKir6.2 + hSUR1 with low affinities (Figure 3E, F). Second, we examined the Kir6.2 mutant that lacks 26 C-terminal residues, dubbed Kir6.2-ΔC26; unlike wild-type Kir6.2, this mutant forms functional channels on the cell surface without needing coexpression with SUR1 (Tucker et al., 1997). For a technical advantage, we used the version of Kir6.2-ΔC26 with an additional mutation at the N-terminal region, V59G, which boosted the apparent expression of the K_ATP current (Proks et al., 2004). The resulting constitutively active channels with this double (background) mutation are hereafter referred to as Kir6.2^bgd, which carry robust currents in whole oocytes that contain millimolar concentrations of ATP. SpTx1 inhibited currents through hKir6.2^bgd much more potently than those through mKir6.2^bgd (Figure 4A, blue squares versus orange circles). The above two lines of results indicate that differences in the two pore-forming orthologous Kir6.2 themselves, but not in their SUR1, primarily underlie the observed differential SpTx1 affinities, and are consistent with the idea that SpTx1 inhibits currents through K_ATP channels by interacting with Kir6.2 (Ramu et al., 2018; Ramu and Lu, 2019). Furthermore, the low affinity of SpTx1 for mKir6.2 explains this toxin’s observed ineffectiveness in influencing GSIS from isolated islets (Figure 1C) and V_m in individual β cells (Figure 2).

Figure 4

Download asset Open asset

Figure 4—source data 1

Related to Figure 4A–C.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig4-data1-v1.xlsx

Download elife-77026-fig4-data1-v1.xlsx

Next, we proceeded to identify the residue variation causing the different SpTx1 sensitivities in hKir6.2 and mKir6.2. Many venom proteins target the external vestibule of K⁺ channels including Kir channels (MacKinnon et al., 1990; MacKinnon and Miller, 1989; Ramu et al., 2004). Comparison of the partial sequences of hKir6.2 and mKir6.2 in the region lining the external vestibule of the pore reveals differences at three positions shown in bold (Figure 4A, top). To determine which of the three differences underlies the observed differing SpTx1 affinities, we performed a mutagenesis study with hKir6.2^bgd first. The triple mutations (S103G, A107N, and E108V), making this segment identical to that in mKir6.2, lowered the affinity of hKir6.2^bgd for SpTx1 to the level of mKir6.2^bgd (Figure 4A, black squares).

We then mutated two of the three residues at a time. The SpTx1 affinity of the mutant S103G + A107N, which did not carry E108V, remained as high as that of hKir6.2^bgd itself (Figure 4B, squares versus blue curve). In contrast, the remaining two mutants carrying the mutation E108V exhibited markedly reduced SpTx1 affinities similar to that of mKir6.2^bgd (Figure 4B, triangles and inverse triangles versus orange curve). These mutagenesis results point to E108V as the mutation responsible for lowering the affinity of hKir6.2^bgd for SpTx1. To further confirm this inference, we examined the effects of individual point mutations. Indeed, the E108V mutation alone lowered the affinity of hKir6.2^bgd to about that of mKir6.2^bgd (Figure 4C, triangles versus orange curve) whereas the other two point mutations had little effects on the affinity of hKir6.2^bgd (Figure 4C, squares and inverse triangles versus blue curve). We also performed the mutagenesis on mKir6.2^bgd, mKir6.2 + mSUR1 and hKir6.2 + hSUR1 (Figure 5). Either the reversed triple or single-V108E mutation conferred such a high SpTx1 affinity on mKir6.2^bgd that was comparable to the SpTx1 affinity of hKir6.2^bgd (Figure 5A–C, F, triangles and circles versus blue curve).

Figure 5

Download asset Open asset

Figure 5—source data 1

Related to Figure 5A–G.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig5-data1-v1.xlsx

Download elife-77026-fig5-data1-v1.xlsx

To demonstrate that the above results also occur in the octameric K_ATP channel–protein complex, we examined hKir6.2 with the E108V mutation and mKir6.2 with the V108E mutation coexpressed with their respective SUR1. Indeed, the mKir6.2^V108E (mKir6.2-V108E) mutation conferred the high SpTx1 affinity of hK_ATP channels on mK_ATP channels coexpressed with mSUR1 (Figure 5D, G, inverse triangles versus cyan curve). Conversely, the hKir6.2-E108V mutation conferred the low SpTx1 affinity of mK_ATP channels on hK_ATP channels coexpressed with hSUR1 (Figure 5E, G, squares versus vermilion curve).

The above results demonstrate that the low affinity of mK_ATP channels for SpTx1 reflects a residue variant, valine versus glutamate, at amino acid position 108 located in the extracellular vestibule of the pore. This finding in turn strengthens the notion that SpTx1 inhibits the K_ATP channel by blocking Kir6.2’s ion-conduction pore (Ramu et al., 2018; Ramu and Lu, 2019).

On the basis of the mechanistic information revealed by the above heterologous mutagenesis studies, we generated a SpTx1-sensitive-mutant mouse line, in which the valine 108 residue of the endogenous (Endo) mKir6.2 was replaced by a glutamate residue, dubbed ^EndomKir6.2^V108E. According to the prevailing paradigm, if the high SpTx1 sensitivity is conferred on the K_ATP channels of ^EndomKir6.2^V108E mice, then SpTx1 should depolarize V_m of their β cells and possibly elicit action potentials.

As performed with wild-type islet cells (Figure 2), we recorded V_m from individual β cells in isolated but intact ^EndomKir6.2^V108E islets (Figure 6, Figure 6—figure supplement 1). Starting from 0, 5, or 8 mM glucose in the extracellular medium, the islets were subsequently challenged with high glucose, SpTx1, or glibenclamide. In a nominal glucose-free condition, SpTx1 at 0.2 µM, 10 times higher than the K_d value for V108E-carrying mKir6.2 (+mSUR1), caused a small depolarization (Figure 6A, orange segment); this SpTx1-induced V_m depolarization was, however, insufficient to elicit action potentials in comparison to that induced by high glucose challenge or glibenclamide that elicited action potentials (Figure 6A, blue and pink segments). With 5 mM glucose, SpTx1 at 0.2 µM caused (variable) V_m depolarization without eliciting any action potentials, which was again less than that induced by high glucose challenge or glibenclamide (Figure 6B). However, with 8 mM glucose, SpTx1 at 0.2 µM strongly depolarized V_m and elicited action potentials (Figure 6C). The results pooled from multiple measurements under all above three glucose conditions are summarized in Figure 6D–F. Thus, the impact of SpTx1 on V_m depends on the glucose concentration.

Figure 6 with 1 supplement see all

Download asset Open asset

Figure 6—source data 1

Related to Figure 6A–F.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig6-data1-v1.xlsx

Download elife-77026-fig6-data1-v1.xlsx

We examined GSIS of islets isolated from the pancreas of ^EndomKir6.2^V108E mice by a perifusion method. Given that SpTx1 in 0 and 5 mM glucose conditions did not trigger action potentials but it did so in the presence of 8 mM glucose, we set the basal glucose level to 3 mM, a known nonstimulating concentration for isolated intact islets, and then raised the glucose concentration to 10 mM to stimulate insulin secretion in the initial set of studies. As a positive control, upon elevating the glucose concentration in the perifusion solution from 3 to 10 mM, insulin secretion rapidly increased in the first phase and then declined in the second phase to a relatively steady level above the baseline (Figure 7A, black trace). After the glucose concentration was lowered back to 3 mM, insulin secretion returned to the baseline level. As an additional control, elevation of glucose and inclusion of 1 μM glibenclamide together yielded insulin secretion much greater than that from elevated glucose alone. As expected, even after removing glibenclamide from the perifusion solution and lowering glucose back to 3 mM, the elevated insulin secretion largely persisted (Figure 7A, cyan trace).

Figure 7

Download asset Open asset

Figure 7—source data 1

Related to Figure 7A–E.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig7-data1-v1.xlsx

Download elife-77026-fig7-data1-v1.xlsx

Next, we tested whether SpTx1 could act as an effective secondary secretagogue to potentiate GSIS in 10 mM glucose. Because 0.2–1 μM SpTx1 did not potentiate insulin secretion from isolated wild-type mouse islets in 7 or 15 mM glucose (Figure 1C), we first tested the effect of a SpTx1 concentration within this range on GSIS from isolated islets of ^EndomKir6.2^V108E mice. Inclusion of 0.5 μM SpTx1 in the perifusion solution potentiated insulin secretion in 10 mM glucose to a level between that in 10 mM glucose alone and in 10 mM glucose plus glibenclamide (Figure 7A, orange trace). Following removal of high glucose by subsequent perifusion with a solution containing 3 mM glucose, which also washed out extracellularly bound SpTx1, insulin secretion returned to the baseline level. This potentiating effect of SpTx1 on GSIS from islets of ^EndomKir6.2^V108E mice, not from those of wild-type mice, strongly implies that SpTx1 generated this effect by blocking Kir6.2 channels harboring the introduced V108E mutation.

We wondered whether SpTx1 diffused slowly into islets to reach β cells, and a preincubation of islets with SpTx1 would thus produce a greater potentiation of GSIS. To examine this possibility, we added SpTx1 10 min before raising glucose concentration to allow more time for equilibration. We found that, whether added earlier or at the same time as raising the glucose concentration, SpTx1 had comparable potentiating effects on GSIS in 10 mM glucose (Figure 7B). The potentiating effects of glibenclamide were also comparable regardless of whether it was applied concurrently with or before glucose elevation (Figure 7C). Adding these two inhibitors 10 min before the high glucose challenge also provided us an opportunity to evaluate SpTx1 and glibenclamide as primary secretagogues, triggering insulin secretion in the presence of a nonstimulating concentration of glucose. In the presence of nonstimulating 3 mM glucose, glibenclamide triggered sizable secretion of insulin (Figure 7C, purple trace) whereas SpTx1 caused no detectable insulin secretion above the baseline level (Figure 7B, blue trace).

Furthermore, we also examined the effect of SpTx1 on insulin secretion in the presence of 16.7 mM glucose, which is an empirically determined optimal concentration commonly used to stimulate a very high level of insulin secretion from mouse islets (Alcazar and Buchwald, 2019; Cerasi, 1975; Malaisse et al., 1967). Either glibenclamide or SpTx1 was added into the perifusion solution containing 3 mM glucose, 10 min before raising glucose to 16.7 mM. In the presence of 3 mM glucose, glibenclamide again stimulated sizable secretion of insulin, which decreased somewhat with time (Figure 7D, purple trace). Raising glucose to 16.7 mM in the presence of glibenclamide caused much greater secretion of insulin than it did in the absence of glibenclamide (Figure 7D, purple trace versus black trace). The same procedure was repeated with SpTx1. In the presence of 3 mM glucose, 0.5 μM SpTx1 did not stimulate insulin secretion above the baseline level but did markedly potentiate insulin secretion in the presence of 16.7 mM glucose (Figure 7D, blue trace versus black trace), albeit less than that potentiated by 1 μM glibenclamide (Figure 7D, blue trace versus purple trace). As we increased the SpTx1 concentration to 2 μM, SpTx1 potentiated insulin secretion in the presence of 16.7 mM glucose to a level comparable to that caused by 1 μM glibenclamide (Figure 7D, vermilion trace versus purple trace in the ~30 to ~60 min region). These observations establish a pair of practically equivalent concentrations for the two inhibitors to act as secondary insulin secretagogues. This is perhaps not surprising because 1 μM glibenclamide in the presence of physiological concentrations of intracellular ATP and 2 μM SpTx1 (K_d = 20 nM; Figure 5G) are both expected to inhibit 99% of the K⁺ current carried by their target channels on the cell surface (Proks et al., 2013). However, at these equivalent concentrations, glibenclamide and SpTx1 did not have comparable abilities as primary insulin secretagogues; SpTx1 stimulated much less insulin secretion than glibenclamide did as evidenced by their differing effects in nonstimulating 3 mM glucose (Figure 7D, vermilion trace versus purple trace in the ~20 to ~30 min region). These comparable secondary secretagogue effects but differing primary secretagogue effects between glibenclamide and SpTx1 are more clearly shown by the ratio of the mean insulin secretion in the presence of 1 μM glibenclamide to that in the presence of 2 μM SpTx1 (Figure 7D, shaded area under the orange curve in the lower panel). Such a difference in the ratio was even more pronounced when 1 μM glibenclamide and 0.5 μM SpTx1 were compared (Figure 7D, shaded area under the cyan curve in the lower panel).

If SpTx1 acts as an effective secondary insulin secretagogue by blocking the V108E-mutant mKir6.2, then SpTx1 should counteract the effect of diazoxide, an opener of K_ATP channels. In a static assay, 16.7 mM glucose stimulated insulin secretion from isolated ^EndomKir6.2^V108E mouse islets to a level above the baseline level in 3 mM glucose (Figure 7E). Furthermore, 100 µM diazoxide lowered the insulin secretion in 16.7 mM glucose to the baseline level. SpTx1 at 1 µM, but not 0.02 µM, could effectively boost GSIS in the presence of diazoxide. As expected, in the presence of nonstimulating 3 mM glucose, the levels of insulin secretion were comparably low under the four compared conditions.

Our ability to confer high SpTx1 sensitivity on mKir6.2 in mice enabled us to examine whether SpTx1 can actually trigger impactful insulin secretion and thus lower the elevated blood glucose level in diabetic mice that overexpress constitutively active mutant mKir6.2 (Figure 8). We started with an NDM-model diabetic-mouse line previously constructed by the Nichols’ group, which contained an inducible mutant mKir6.2 transgene in the Rosa26 locus (Pdx^CreERT2; Rosa26^lsl-Kcnj11), hereafter referred to as ^Rosa26mKir6.2^NDM (Remedi et al., 2009). After a 9-day induction to overexpress the constitutively active mutant mKir6.2 in the pancreas of ^Rosa26mKir6.2^NDM mice, the overnight-fasted blood glucose level of mice remained highly elevated during the 2-hr observation period (Figure 8A). We found that an IV injection of SpTx1 (1 mg/kg) or its vehicle saline had no meaningful effect on the elevated blood glucose levels of diabetic ^Rosa26mKir6.2^NDM mice and did not trigger a rise of insulin in the plasma of their circulating blood collected during the observation period (Figure 8A, C, filled and open orange triangles). These findings are expected because mKir6.2 exhibits very low affinity for SpTx1.

Figure 8 with 2 supplements see all

Download asset Open asset

Figure 8—source data 1

Related to Figure 8A–D.

https://cdn.elifesciences.org/articles/77026/elife-77026-fig8-data1-v1.xlsx

Download elife-77026-fig8-data1-v1.xlsx

To increase the affinity of mKir6.2 in ^Rosa26mKir6.2^NDM mice for SpTx1, we introduced the V108E mutation into all endogenous and transgenic copies of mKir6.2 within their genome (Figure 8—figure supplement 1, Figure 8—figure supplement 2). The resulting mice are denoted here as ^Rosa26mKir6.2^NDM-V108E. Like the original ^Rosa26mKir6.2^NDM mice, after a 9-day induction to overexpress the mutant mKir6.2 in the pancreas of ^Rosa26mKir6.2^NDM-V108E mice, their overnight-fasted blood glucose level also became highly elevated in a sustained manner. A single bolus application of glibenclamide (40 mg/kg) or its vehicle DMSO neither markedly lowered the elevated blood glucose level nor caused a rise of insulin in the plasma (Figure 8A, C, filled and open blue circles) of diabetic ^Rosa26mKir6.2^NDM-V108E mice during the observation period. This finding is consistent with the ineffectiveness of acute glibenclamide treatments reported in the original study of ^Rosa26mKir6.2^NDM mice (Remedi et al., 2009). Thus, in the present context, ^Rosa26mKir6.2^NDM-V108E mice retained the basic characteristics of inducible diabetic ^Rosa26mKir6.2^NDM mice.

Next, we tested whether SpTx1 could mitigate hyperglycemia of diabetic ^Rosa26mKir6.2^NDM-V108E mice. Following an IV injection of SpTx1 (1 mg/kg) but not of saline, the elevated blood glucose level of diabetic ^Rosa26mKir6.2^NDM-V108E mice dropped markedly over 30 min and this lower level of blood glucose persisted over the remaining observation period (Figure 8B, filled and open magenta circles). In contrast, administered SpTx1 or saline had no discernible effects on the fasting blood glucose levels of wild-type and ^EndomKir6.2^V108E mice (Figure 8B, filled black squares and green diamonds or open black squares and green diamonds). Moreover, SpTx1, but not saline, triggered a transient but sizable rise of insulin in the plasma of diabetic ^Rosa26mKir6.2^NDM-V108E mice, but not in that of wild-type or ^EndomKir6.2^V108E mice (Figure 8D). These results show that SpTx1 promotes transient insulin secretion and markedly lowers the highly elevated blood glucose level of diabetic ^Rosa26mKir6.2^NDM-V108E mice.

All mutant Kir6.2 cDNAs were produced through PCR-based mutagenesis and confirmed by DNA sequencing. The Kir6.2 and SUR1 cRNAs were synthesized with T7 polymerase using the corresponding linearized cDNAs as templates. Channel currents were recorded from Xenopus oocytes, which were injected with cRNA encoding specific wild-type or mutant Kir6.2 channels with or without coinjection with cRNA encoding SUR1, through a two-electrode voltage-clamp amplifier (Oocyte Clamp OC-725C, Warner Instruments Corp.), filtered at 1 kHz, and sampled at 10 kHz using an analog-to-digital converter (Digidata 1322A; MDS Analytical Technologies) interfaced with a personal computer. pClamp8 software (MDS Analytical Technologies) was used for amplifier control and data acquisition. All recordings were performed at room temperature. To activate K_ATP (Kir6.2 + SUR1) channels, 3 mM azide was added to the bath solution. The resistance of electrodes filled with 3 M KCl was 0.2–0.4 MΩ. To elicit currents through the channels, the membrane potential of oocytes was stepped from the holding potential of 0 to −80 mV then to +80 mV before returning back to 0 mV. The bath solution contained (in mM): 100 KCl, 0.3 CaCl₂, 1.0 MgCl₂, and 10 N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid (HEPES), and bovine serum albumin (BSA, 50 μg/ml), where pH was titrated to 7.6 with KOH. All salts were from Millipore-Sigma. Recombinant SpTx1, produced as previously described (Ramu et al., 2018), was added to the bath solution in the concentrations as specified in the relevant figures. The number of measurements was determined on basis of previous studies (Ramu et al., 2018; Ramu and Lu, 2019). The variations in data values reflected both biological variability and technical errors. No experiments were excluded.

Electrophysiological membrane potential (V_m) measurements from individual cells in isolated but intact islets cultured on glass coverslips were performed using the perforated whole-cell patch-clamp method as described (Yang et al., 2021). Wax-coated electrodes (G85150T, Warner) were back filled with the intracellular solution containing (in mM): 76 K₂SO4, 10 KCl, 10 NaCl, 6 MgCl₂, 30 mannitol, 30 HEPES, where pH was titrated to 7.2 at 35°C with N-methyl-d-glucamine (NMG). The tip of the pipette was filled with this solution and the remainder was back filled with the same solution plus β-escin (6 or 8 µM). The free Mg²⁺ concentration of this solution, taking the divalent cation-chelating action of SO₄²⁻ into consideration, is estimated to be about 2 mM. The extracellular solution contained (in mM): 135 NaCl, 4 KCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES, where pH was titrated to 7.4 at 35°C with NMG. Glucose was added to this solution as required and the osmolarity was adjusted to ~300 mOsm with mannitol. All salts were from Millipore-Sigma. Using the solutions described above, the initial input resistance of the electrode was typically 4–8 MΩ. The electrophysiological recordings were performed under a continuous perfusion condition, 0.3–0.4 ml/min (Instech), at ~35°C (TC-124A and TC-344B Warner). Islets were equilibrated in the recording chamber for 10–15 min before measurements were made. Adequate whole-cell access was achieved typically within 5–10 min. SpTx1 was applied with 0.3% (wt/vol) BSA.

The output of the patch-clamp amplifier (Axopatch 200B, Molecular Devices) was digitized at 2 kHz and stored for later analysis using Igor Pro (v8 or v9, Wavemetrics). The liquid junction potential has been subtracted from the results. To estimate the average V_m values under different conditions, the V_m traces were digitally Gaussian filtered at 0.1 Hz and the filtered data points in the last 50 s of a segment of interest were averaged. The filtered traces are superimposed on the illustrative original traces in the figures where appropriate. The variations in data values reflected both biological variability and technical errors (Figures 2D–F–6D–F). No data, which were successfully recorded over sufficient durations that allowed for adequate examination the membrane potential under the various specified conditions, were excluded.

To confer the hKir6.2-like, high SpTx1 sensitivity on endogenous mKir6.2 channels in wild-type mice, we employed the CRISPR-Cas9 genome editing technique (Ran et al., 2013). The sequences of in vitro transcription template of a guide RNA (gRNA) and the single-stranded oligodeoxynucleotide (ssODN) donor templates are provided below.

Synthesized DNA templates (IDT Technologies) for gRNA targeting the mKir6.2 DNA sequence (Genbank, Accession Number NC_000073.6) also contained a T7 promoter sequence. The sequence of the DNA template for the gRNA from the 5′ end to the 3′ was:

GAAATTAATACGACTCACTATAGGGAGAGCTTGTGACGCAGGGCACATGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGCTTTTTT where the T7 promoter sequences are underlined, target DNA sequences bolded, and tracrRNA italicized; whereas that for the ssODN donor template was:

GGACCTCGATGGAGAAAAGGAAGGCAGATGAAAAGGAGTGGATGCTTGTGACGCAGGGCTCGTTAGTGCCCTCTCCGGGGGCCAGGTCACCGTGGGCGAAGGCGATGAGCCACCAGACCATGGCAAAGA where the sequence of the restriction enzyme (BanII) cleavage site is underlined, the intended nonsynonymous substitution is bolded and synonymous ones italicized (Figure 6—figure supplement 1 and Figure 8—figure supplement 1).

In vitro transcription from the gRNA template or a Cas9 plasmid (T7-Cas9-HA-2NLS) was performed to produce gRNA or Cas9 mRNA using mMESSAGE mMACHINE T7 ULTRA kit (Invitrogen, AM1344). The ssODN stock solution was prepared according to the manufacturer’s recommendation (IDT Technologies). The freshly prepared solution, which contained 100 ng/µl Cas9 mRNA, 100 ng/µl gRNA, 200 ng/µl ssODN, 0.1 mM ethylenediaminetetraacetic acid (EDTA), and 10 mM Tris titrated to pH 7.5 with HCl, was injected into cytoplasm of FVB mouse embryos at the Transgenic and Chimeric Mouse Facility (TCMF) of the University of Pennsylvania Perelman School of Medicine (UPENN-PSOM). The injected embryos were implanted into the uterus of pre-prepared surrogate female mice for obtaining germline-transmitting founder ^EndomKir6.2^V108E-mutant mice. The intended T323A mutation and the five adjacent nucleotides in the DNA created a cleave site for the restriction enzyme BanII, which was used to assist the identification of genotypes.

Double transgenic ^Rosa26mKir6.2^NDM mice on the C57BL/6J (B6J) background were obtained from the laboratory of Dr. Colin Nichols (Washington University, St. Louis, MO; Remedi et al., 2009). This mutant mouse line contained two unlinked transgenes. One encodes for a tamoxifen-dependent, pancreas-specific Cre recombinase (Pdx^CreERT2) and the other is a Cre-inducible copy of NDM-causing mutant Kir6.2 gene inserted at the Rosa26 locus. The translation of mutant Kir6.2 transgene would harbor the K185Q mutation and lack N-terminal 30 residues (ΔN30). By crossing their ^Rosa26mKir6.2^NDM mice with our mutant ^EndomKir6.2^V108E mice, we created a hybrid mutant mouse line on the B6.FVB background whose genome contained the two unlinked transgenes and the endogenous Kir6.2 gene with homozygous V108E mutation. Furthermore, using the hybrid mutant line and the CRISPR-Cas9 targeting strategy described above, we introduced the V108E mutation into the NDM-causing Kir6.2 transgene at the Rosa26 locus, and the resulting germline-transmitting mutant mice are called ^Rosa26mKir6.2^NDM-V108E, in which only the mutant Kir6.2 transgene contain K185Q and ΔN30 mutations but both transgenic and endogenous Kir6.2 genes contain V108E mutation. All mice used in the study were generated and maintained in a UPENN-PSOM animal facility.

All mice were housed in ventilated, sterilized polysulfone mouse cages (AN75, Ancare Corp., Bellmore, NY, USA) containing Bed-o’Cobs absorbent bedding (Andersons Lab Bedding, Maumee, OH, USA). Irradiated LabDiet 5053 feed (PicoLab Rodent Diet 20, Fort Worth, TX, USA) and autoclaved acidified water (pH 2.5–2.8 titrated with HCl) were provided ad libitum. Shepherd’s shack and nesting material were also provided in the cages to minimize distress of mice. Cages were changed weekly in a Class A2 biological safety cabinet (Baker SG403A-ats, Sanford, ME, USA). The air in the mouse-holding room was exchanged with filtered air at a rate of 15 times of room volume per hour. Room temperature was maintained between 20 and 25°C, and relative humidity between 30% and 70%. Lighting of the room was on and off alternately every 12 hr where the light phase was from 7:00 AM to 7:00 PM. A vivarium-wide rodent health monitoring program showed the absence of the following pathogens: MHV, Sendai virus, MVM, MPV1/2, TMEV, PVM, Reo-3, EDIM, Ectromelia virus, LCMV, MAdV, K-Virus, Polyomavirus, MCMV, Mouse Thymic Virus, Hantavirus, Mycoplasma pulmonis, Citrobacter rodentium, Clostridium piliforme, Corynebacterioum kutscheri, CAR bacillus., Salmonella sp., Klebsiella pneumoniae, Streptococcus pneumoniae, Streptobacillus moniliformis, Encephalitozoon cuniculi, Myobia musculi, Myocoptes musculinus, Radfordia affinis, Aspiculuris tetraptera, Syphacia obvelata, and Giardia muris.

The animal cohort in this study included wild-type and mutant mice at 8–12 weeks of age. The designed endpoint of the study was defined as individual mice reaching that age range and euthanized for various types of experiments. We monitored the general health condition and behavior of each mouse at least three times per week during maintenance and continuously during the 2-hr observation experiment. If a mouse was suspected to have clinical signs of pain and distress, this mouse was then monitored and checked one or several times daily. We used the following criteria to determine the humane endpoints at which mice of any age were euthanized within 1 hr upon our determination: clinical signs of pain and distress, such as hunched posture, inactivity, dehydration, abdominal enlargement caused by intestine obstruction, increased respiratory effort manifested as increased intercostal or subdiaphragmatic retraction and gasping or breathing with an open mouth, raffled fur coat, emaciated body condition (e.g., weight loss of >20%), severe lethargy manifested as unwillingness to ambulate more than a few steps when gently stimulated with a gloved finger, or cold to the touch.

For genomic DNA isolation, a tail sample (~2 mm) or a clipped toe from each mouse was digested at 55 ℃ overnight with gentle mixing in 0.5 ml solution containing 50 mM Tris titrated to pH 8.0 with HCl, 100 mM EDTA, 0.5% sodium dodecyl sulfate (SDS), and 0.5 mg/ml Proteinase K. The high molecular weight genomic DNA in the digestion sample was extracted with a mixture of phenol and chloroform and precipitated using cold ethanol with 0.3 M sodium acetate titrated to pH 6.0. The precipitated genomic DNA was resuspended in 0.1 ml of a solution containing 1 mM EDTA and 10 mM Tris titrated to pH 8.0 with HCl; 1 μl of which was used as the DNA template and added into a 25 μl PCR mixture containing 50 mM KCl, 1.5 mM MgCl₂, 20 mM Tris triturated to pH 8.4, 0.2 mM dNTPs, 0.2 μM of each forward and reverse oligodeoxynucleotide primers described below, and 1 U of Platinum Taq DNA polymerase (Invitrogen, 10966034). PCR reactions took place in a thermal cycler (Mastercycler 5333; Eppendorf) with an initial denaturation step at 95℃ for 2 min and 30 amplification cycles (denaturation at 95℃ for 30 s, annealing at 60℃ for 30 s, and extension at 72℃ for 45 s for <0.7 kb or 150 s for ~2 kb), and followed by a final 5 min extension step at 72℃. 5 μl of individual PCR products was subjected to electrophoresis on 1% agarose gel, stained with 0.5 μg/ml ethidium bromide, and evaluated against a DNA ladder (Thermo Scientific, SM1331). The initial screen was performed on all resulting PCR products by the expected size (~2 kb) and by their susceptibility to BanII restriction enzyme digestion.

By subsequent DNA sequencing of the whole ~2 kb PCR product, we verified that the BanII-cut products contained the mutated codon for a glutamate residue at position 108 (Figure 6—figure supplement 1 and Figure 8—figure supplement 1), whereas the uncut products contained the wild-type codon for valine. Moreover, sequencing of the entire coding region of the Kir6.2 gene in ^EndomKir6.2^V108E mice confirmed the two intended synonymous substitutions in the DNA triplets that encode amino acids threonine 106 and asparagine 107 without any other nucleotide changes. These intended substitutions were designed in the ssODN donor template to prevent re-editing by Cas9 while preserving their original amino acid identities. However, sequencing of the entire coding region of the mutant Kir6.2 transgene in ^Rosa26mKir6.2^NDM-V108E mice revealed a single unintended, nonsynonymous substitution in the DNA codon that would lead to an additional N107Y mutation in the translated Kir6.2 protein (Figure 8—figure supplement 1). Fortunately, this N107Y mutation had little effect on the affinity of channels for SpTx1 (Figure 8—figure supplement 2).

Specific primer pairs were designed using published genomic sequences for the mouse Kcnj11 gene (Genbank accession: NC_000073.6) or the Rosa26 locus (Genbank accession: NC_000072.6). Using Primer-BLAST (NIH) on default settings, each candidate pair was checked against the Refseq representative genomes of Mus musculus (taxid: 10090) to ensure a single expected PCR product. The oligodeoxynucleotide sequences of the synthesized primers (Sigma-Genosys) and the predicted sizes of the PCR products are given below:

1. For amplifying the entire coding region of the Kir6.2 gene at the endogenous locus, a 1954-bp PCR product was expected with primers 5′-GGTAGACTTATCCCGCCGTG-3′ and 5′-TGGGGGCTCAGTAAGCAATG-3′.

2. For amplifying the entire coding region of the Kir6.2 transgene at the Rosa26 locus, a 1981-bp PCR product was expected with primers 5′-GAGGCTACTGCTGACTCTCAA-3′ and 5′-GCTCGTCAAGAAGACAGGGC-3′.

3. For testing the BanII susceptibility, a 760-bp product was expected using PCR product from one or two as the template with primers 5′-CGCCCACAAGAACATTCGAG-3′ and 5′- GGTGATGCCCGTGGTTTCTA-3′. Upon exposure to BanII restriction enzyme, the 760-bp PCR product containing the designed V108E mutation would be cut into 565 and 195 bp fragments whereas the wild-type PCR product would remain uncut.

4. For verifying the presence of the Cre gene, a 429-bp PCR product was expected with primers 5′- GCAAGAACCTGATGGACATGTTCAG-3′ and 5′-GCAATCCCCAGAAATGCCAGATTAC-3′.

Pancreatic islets were isolated as previously described (Doliba et al., 2017; Li et al., 2009). For each experiment, one to four adult mice (8–12 weeks of age) were euthanized by exposing them to an overdose of isoflurane through inhalation and a subsequent cervical dislocation. Following a perfusion through the common bile duct with a Hank’s balanced salt solution (HBSS; GIBCO, 14175) containing collagenase XI (1–2 mg/ml, Millipore-Sigma, C7657) and 1 mM CaCl₂, the pancreas was dissected and digested at 37°C for 5 or 15 min. After the digestion, the collagenase in the pancreas homogenate was removed through several washes with HBSS containing 0.3% (wt/vol) bovine serum albumin, and dissociated islets were then hand-picked into a recovery media RPMI1640 (GIBCO, 21870) with the following supplements: 10% fetal bovine serum, 2 mM GlutaMAX, 1 mM sodium pyruvate, penicillin–streptomycin, and 10 mM HEPES titrated to pH 7.4 with NaOH. For the perifusion assay, the digested pancreases samples were subject to a Ficoll gradient purification, and the islets in the enriched fraction were washed with the recovery solution before handpicking islets. During this process, the samples from four mice were pooled to ensure a sufficient number of islets were used to perform assays under compared conditions with the same population in each independent round of experiment. The handpicked islets were placed in 5% CO₂ incubator at 37°C to rest overnight prior to the static incubation assay or 2 days prior to the perifusion assay.

For insulin secretion in each independent static incubation assays, pancreatic islets from one mouse (8–12 weeks of age) were used as follows. Isolated islets were washed by six sequential transfers to a modified Krebs–Ringer buffer containing 114 mM NaCl, 24 mM NaHCO₃, 5 mM KCl, 2.2 mM CaCl₂, 1 mM MgCl₂, 1 mM NaH₂PO₄, 10 mM HEPES titrated to pH 7.4 with NaOH, 0.3% (wt/vol) BSA, and 3 mM glucose. Washed islets were equilibrated in the final wash step for 1 hr at 5% CO₂ and 37°C. Using a previous study as reference (Remedi et al., 2009) and on the basis of the amount of insulin secretion and the sensitivity of insulin assay, we chose to use 5–10 islets in each assay. After 1 hr of equilibration, each group of 5–10 randomly selected islets was placed into a well of a microwell plate containing the modified Krebs–Ringer buffer, added with additional test reagents: glucose and SpTx1 as indicated in Figures 1C and 7E. Diazoxide (Millipore-Sigma, D9035) was added to the modified Krebs–Ringer buffer 20 min prior to the addition of other test reagents as indicated in Figure 7E. Following 1-hr incubation under the conditions of 5% CO₂ and 37°C, the supernatant containing released insulin was recovered and the islets were placed in acidified ethanol to extract total insulin content as previously described (Leiter, 2009). Insulin in the samples was assayed using a colorimetric ELISA kit (ALPCO, 80-INSRTU-E10). The variations in data values reflected both biological variability and technical errors. No experiments were excluded.

For each perifusion experiment (Doliba et al., 2017), pancreatic islets from four mice (8–12 weeks of age) were used in the following manner. About 180 handpicked islets were placed onto a polycarbonate membrane (Nuclepore, Whatman) inside a polypropylene perifusion chamber (Swinnex, Millipore) and the oxygenated perifusion solution (modified Krebs–Ringer buffer as above) added with additional test reagents: glucose, glibenclamide (Millipore-Sigma, G0639), and/or SpTx1 as indicated in Figure 7A–D. All experiments were performed at 37°C with an HPLC-controlled flow rate of 1 ml/min; consecutive fractions of 1 ml were collected. At the end of each experiment, the islets were recovered and placed in acidified ethanol to extract total insulin content as described above. Insulin in the samples was assayed using an insulin-specific radioimmunoassay kit (RI-13K, Millipore). For each condition, the study was performed three to six times. The variations in data values reflected both biological variability and technical errors. No experiments were excluded.

In ^Rosa26mKir6.2^NDM and ^Rosa26mKir6.2^NDM-V108E mice, the expression of mutant Kir6.2 activity occurred in the pancreas upon the excision of the Neo/WSS cassette by the tamoxifen-dependent Cre recombinase (Remedi et al., 2009). Exposure to tamoxifen permitted the translocation of Cre recombinase from the cytoplasm to the nucleus in order to perform the excision between loxP sites within the targeted Rosa26 locus. To induce the expression of mutant Kir6.2 activity and consequently the diabetic phenotype in these mutant mice (8–12 weeks of age), corn oil containing tamoxifen (75 mg/kg body weight; prepared as described below) was injected intraperitoneally for five consecutive days. Injection sites were sealed with a tissue adhesive (3M Vetbond) to prevent leakage of the injected liquid.

To make a liquid stock of tamoxifen (20 mg/ml) for injection, tamoxifen solids (Millipore-Sigma, T5648) were added into corn oil (Millipore-Sigma, C8267) in a polypropylene tube wrapped with aluminum foil to protect tamoxifen from light. The stock was placed on a nutator to dissolve overnight at room temperature. The dissolved tamoxifen stock was filtered through a sterilized Steriflip unit (Millipore) then aseptically aliquoted, stored at 4°C, and used within 5 days.

We evaluated the blood glucose levels of tested mice (8–12 weeks of age) using a digital glucometer (Clarity Diagnostics, BG1000). Induced diabetic mice were tested daily starting from the day after the final administration of tamoxifen until their blood glucose levels reached above 600 mg/dl. Following overnight (16 hr) fasting, blood glucose monitoring and plasma collection were then performed using specified mouse lines at designated time points over a 2-hr observation period in the next morning. Glibenclamide (40 mg/kg body weight) or its vehicle pure DMSO was administered through a single intraperitoneal injection whereas SpTx1 (1 mg/kg body weight) or its vehicle normal saline solution was intravenously administered by a lateral tail vein or retro-orbital sinus injection. At each given time point, the glucose level of a blood drop obtained from the cut tail tip was measured twice using a glucometer. For any case where the level was above 600 mg/dl, the blood sample was diluted by an equal volume of the phosphate-buffer solution titrated to pH 7.3 before the glucose level was remeasured for two times. Additionally, four to five drops (~12 μl) of tail vein blood were collected at individual specified time points into a tube containing K₂EDTA (1.5 mg/ml blood) and centrifuged at 2000 × g for 5 min at 4°C. 5 μl of the resulting plasma was frozen in liquid nitrogen, stored at −20°C, and assayed using a colorimetric insulin-detecting ELISA kit (CrystalChem, 90080) within a week. Mice were grouped according to genotypes. Using a previous study as reference (Remedi et al., 2009), we determined the number of mice for individual groups. The variations in data values reflected both biological variability and technical errors. No experiments were excluded.

Unless otherwise specified, all data are reported as mean (± SEM). Statistical analyses were performed using software Origin 8.0 (OriginLab) and Igor Pro 8 and 9 (Wavemetrics: https://www.wavemetrics.com). p values were calculated using two-tailed Student’s t-test with unequal variance (Origin 8.0) and presented in the relevant figure legends purely as descriptive parameters of data.

The data graphs and statistical analyses in Figures 1, 3—5, Figures 7 and 8 and Figure 8—figure supplement 2 as well as data values in Figure 3—figure supplement 1 were created using Origin 8.0 (OriginLab: https://www.originlab.com). Islet cell electrophysiological results in Figures 2 and 6 were analyzed using Igor Pro 8 and 9. Snippets of the DNA sequencing chromatograms in Figure 6—figure supplement 1A and Figure 8—figure supplement 1A, B were generated by screen capture, the textual contents of the figure legends were drafted with Word (Microsoft Office Suite version 365; https://www.microsoft.com/en-us/store/apps/windows). All figure panels and legends were made by importing and resizing these vector graphics, image files and texts using Adobe Illustrator (version CS4; https://www.adobe.com).

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

We thank J Bryan (Baylor College of Medicine) and L Jan (University of California, San Francisco, CA) for sharing Kir6.2 and SUR1 cDNAs; C Nichols (Washington University) for sharing the diabetic model mice; and the University of Pennsylvania Transgenic and Chimeric Mouse Facility for the CRISPR mouse line creation service and the Diabetes Research Center Core for performing the perifusion and RIA assays of mouse pancreatic islet β cells (P30-DK19525). The present study in the early phase was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (DK109979) to ZL, and in the later phase when research mice were not directly involved by the National Institute of General Medical Sciences (GM055560) to ZL. TH was in part supported by the National Institute of Diabetes and Digestive and Kidney Diseases (DK098517 and P30-DK19525).

The Institutional Animal Care and Use Committee at the University of Pennsylvania reviewed and approved this study (Protocol Numbers: 804489 and 804811). All animal welfare considerations were taken and the study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal care was provided by experienced staff and researchers, and tissue dissections were performed after euthanasia by researchers who were trained at the University Laboratory Animal Resources of the University of Pennsylvania. Every effort was made to minimize suffering and distressed of mice. The studies were performed in compliance with the ARRIVE guidelines (https://arriveguidelines.org/).

1. Received: January 13, 2022
2. Preprint posted: January 17, 2022 (view preprint)
3. Accepted: January 26, 2022
4. Version of Record published: February 25, 2022 (version 1)

© 2022, Ramu et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.