Kcne2 deletion uncovers its crucial role in thyroid hormone biosynthesis.

Thyroid dysfunction is a global health concern, causing defects including neurodevelopmental disorders, dwarfism and cardiac arrhythmia. Here, we show that the potassium channel subunits KCNQ1 and KCNE2 form a thyroid-stimulating hormone-stimulated, constitutively active, thyrocyte K+ channel required for normal thyroid hormone biosynthesis. Targeted disruption of Kcne2 in mice impaired thyroid iodide accumulation up to eightfold, impaired maternal milk ejection, halved milk tetraiodothyronine (T4) content and halved litter size. Kcne2-deficient mice had hypothyroidism, dwarfism, alopecia, goiter and cardiac abnormalities including hypertrophy, fibrosis, and reduced fractional shortening. The alopecia, dwarfism and cardiac abnormalities were alleviated by triiodothyronine (T3) and T4 administration to pups, by supplementing dams with T(4) before and after they gave birth or by feeding the pups exclusively from Kcne2+/+ dams; conversely, these symptoms were elicited in Kcne2+/+ pups by feeding exclusively from Kcne2-/- dams. These data provide a new potential therapeutic target for thyroid disorders and raise the possibility of an endocrine component to previously identified KCNE2- and KCNQ1-linked human cardiac arrhythmias.

INTRODUCTION

KCNQ1 is a voltage-gated K+ channel α subunit noted for its key role in human ventricular repolarization because it generates the IKs ventricular repolarization current, by co-assembling with the KCNE1 (MinK) single transmembrane domain β subunit1–4. KCNE1 belongs to a family of five genes including KCNE2 (MiRP1), which like KCNE1 can regulate KCNQ1 and other α subunits such as hERG, often endowing unique functional properties5–7. Inherited human gene variants in KCNQ1, hERG, KCNE1 and KCNE2 are all associated with life-threatening cardiac arrhythmias including long QT syndrome3,4,6,8, and may play a role in atrial fibrillation (AF)9–12. These subunits are also expressed in a variety of other tissues, but the possible cardiac effects secondary to their dysfunction in these other tissues are little-studied.

KCNQ1 is unique among the voltage-gated K+ channel α subunits in that it can form constitutively active, K+ ‘leak’ channels; this is achieved by co-assembly with KCNE2 or KCNE3. This permits KCNQ1 to facilitate background K+ flux in some non-excitable, polarized epithelial cells. Thus, KCNQ1 and KCNE3 are thought to form a channel in the basolateral membrane of colonic crypt cells13. KCNQ1-KCNE2 channels support function of the H+/K+-ATPase in the apical membrane of parietal cells; disruption of Kcnq1 or Kcne2 in mice causes achlorhydria and gastric hyperplasia14,15.

Analogous to parietal cells and colonic crypt cells in the gastrointestinal tract, thyrocytes are non-excitable, polarized epithelial cells expressing ion transporters essential for the function of the thyroid gland. The thyroid hormones (TH) triiodothyronine (T3) and tetraiodothyronine (thyroxine, or T4) are critical for normal growth and development of the fetus and newborn as well as for regulation of metabolism in virtually all tissues at all ages. Because of the scarcity of iodine, an essential constituent of T3 and T4, iodide (I−) deficiency disorders are still prevalent in many areas of the world and are thus at the forefront of global health initiatives. I− enters thyrocytes via the basolaterally located Na+/I− symporter (NIS)16,17 and exits apically into the colloid, where it is covalently incorporated into thyroglobulin, the precursor of T3 and T4. NIS-mediated I− transport uses the downhill Na+ gradient generated by the Na+/K+ ATPase at the basolateral membrane of the thyrocyte. The role of K+ channels in the thyroid is unknown.

Here, we show that potassium channel subunits KCNQ1 and KCNE2 - originally recognized for their functional roles in repolarizing cardiac myocytes - form a constitutively-active K+ channel in thyrocytes, and that Kcne2 is required for normal TH biosynthesis. Aging Kcne2−/− mice, and pups from Kcne2−/− dams, exhibit a panoply of symptoms including goiter, cardiomegaly, dwarfism and alopecia. Remarkably, these symptoms are alleviated highly efficiently by administration of T3/T4 to pups, by supplementing dams with T4 pre- and postpartum, or by surrogacy with wild-type dams; conversely, symptoms are triggered in wild-type pups by surrogacy with Kcne2−/− dams. Put in the context of existing studies, our findings raise the possibility of an endocrine component to some cardiac arrhythmias and early-onset myocardial infarction previously associated with human KCNQ1 and KCNE2 genetic variants.

RESULTS

Kcne2−/− pups from Kcne2−/− dams exhibit cardiac hypertrophy, fibrosis and reduced cardiac output

Mutations and polymorphisms in human KCNQ1 and KCNE2 are associated with ventricular and atrial cardiac arrhythmias, presumed to arise from dysfunction of the K+ channels they form in cardiac myocytes4,6,18,19. We previously found that at 3 months of age, Kcne2−/− mice from heterozygous crosses have normal echocardiographic parameters and ventricular myocyte size20. In contrast, in the current study, when we bred Kcne2−/− pups from homozygous knockout crosses, we found that they exhibited striking cardiomegaly: >2-fold increased heart mass, and >3-fold increased heart weight:bodyweight ratio, at 3 weeks (Fig. 1a). The 3-week-old Kcne2−/− pups from Kcne2−/− dams also exhibited >50% increases in end-diastolic left ventricular (LV) internal diameter, LV anterior and posterior wall thickness, and a 45% decrease in fractional shortening (Fig. 1b,c). The anterior and posterior wall thickening demonstrate cardiac hypertrophy, and this is probably the primary effect, with the LV dilation and reduced fractional shortening likely arising from a compensatory response (Starling mechanism) to the impaired contractility resulting from sustained hypertrophy.

Figure 1

Kcne2 disruption causes cardiac hypertrophy, fibrosis and reduced fractional shortening

(a) Left, external view of exemplar hearts from 3-week-old Kcne2+/+ (left) and Kcne2−/− (right) mice from homozygous crosses. Right, mean heart mass, body mass, and heart weight/bodyweight (HW/BW; mg g−1) for mice as in panel a (n = 9–14).* P < 0.000001.

(b)Exemplar transthoracic echocardiograms for a 3-week-old Kcne2+/+ pup from a Kcne2+/+ dam and a Kcne2−/− pup from a Kcne2−/− dam.

(c) Mean echocardiographic parameters from recordings as in panel b. d/s, diastolic/systolic; LVAW, left ventricular anterior wall thickness; LVID, left ventricular internal diameter; LVPW, left ventricular posterior wall thickness, n = 4 per group. * P < 0.05; ** P < 0.005; *** P < 0.0005, by one-way ANOVA.

(d) Left, mean cell capacitance for ventricular myocytes isolated from 3-week-old pups from homozygous crosses, n = 6–8; *P < 0.001. Right, exemplar whole-cell recordings from ventricular myocytes as in left. Insets: voltage protocol and scale bars.

(e) Mean peak current densities for Kcne2+/+ (blue) and Kcne2−/− (red) myocytes as in panel d, n = 6–8; *P < 0.05.

(f) Left, current inactivation τ values for Kcne2+/+ (blue) versus Kcne2−/− (red) myocytes as in panel e. Right, current amplitudes from double exponential fits for Kcne2+/+ (solid) versus Kcne2−/− (open) myocytes as in panel g. *P < 0.001. Current decay was best fit with two exponentials with parameters resembling Ito,f and IK,slow, and a steady-state current, Iss.

(g–i) Cardiac tissue from 3-week-old Kcne2−/− mice bred from Kcne2−/− dams.

(g) Exemplar necropsy.

(h) Left, exemplar Masson’s trichrome-stained left ventricle showing collagen (blue) indicative of fibrosis. Scale bar, 150 µm. Right, exemplar H&E-stained papillary muscle. Scale bar, 30 µm.

(i) Exemplar Masson’s trichrome stained liver showing marked perisinusoidal fibrosis (blue). Scale bar, 200 µm.

(j) Cardiac tissue from 15-month-old Kcne2−/− mice bred from Kcne2+/− dams; similar results were observed in 4/4 mice evaluated. Left, exemplar necropsy showing cardiomegaly; center and right, exemplar Masson’s trichrome-stained LV showing fibrosis (blue). Scale bars: center, 100 µm; right, 30 µm.

Ventricular myocytes of 3-week-old Kcne2−/− pups from homozygous crosses had a 2-fold larger membrane capacitance than those from age-matched Kcne2+/+ pups, also indicative of hypertrophy – defined as increased organ or tissue size due to increase in size of the constituent cells (Fig. 1d). Raw amplitudes of two of the predominant mouse ventricular repolarization K+ currents, IK,slow and Ito,f, were unaltered by Kcne2 disruption, but because of the doubling in capacitance this constituted a 2-fold reduction in current density for each. In contrast, the raw amplitude of the steady-state K+ current (Iss) doubled, indicating that its density did not change with Kcne2 disruption (Fig. 1d–f).

Hearts from 3-week-old Kcne2−/− mice from Kcne2−/− dams (Fig. 1g) also exhibited marked LV fibrosis and papillary muscle degeneration, necrosis and mineralization - a feature of sustained hypertrophy (Fig. 1h). Hepatic fibrosis was also observed, suggesting right heart failure (Fig. 1i) and the liver had an unusually pale appearance (Fig. 1g) possibly indicative of fatty liver. Furthermore, 1-year-old Kcne2−/− mice bred from Kcne2+/− dams also exhibited cardiomegaly with LV fibrosis (Fig. 1j). In sum, it is apparent that there exists a strong influence of parental genotype on cardiac phenotype of the Kcne2−/− offspring, and also that aging Kcne2−/− mice from heterozygous parents show severe cardiovascular hypertrophy and fibrosis not yet manifested in their 3-month-old counterparts20.

Kcne2−/− mice exhibit dwarfism and alopecia

Aside from their cardiac pathology, Kcne2−/− mice exhibited other gross abnormalities that were exacerbated by the maternal Kcne2−/− genotype. Pups from Kcne2−/− dams exhibited 50% embryonic lethality whether the sire was Kcne2−/− or Kcne2+/− (Fig. 2a). Maternal genotype was the determining factor in litter size: litters from Kcne2+/− dams with Kcne2−/− sires were of normal size, and surviving pups in litters from Kcne2−/− × Kcne2+/− crosses showed an approximately Mendelian distribution (Fig. 2a). Kcne2−/− pups from Kcne2−/− dams also exhibited severe dwarfism (Fig. 2b,c). While birth weights were similar in all cases, mean body weight of pups from Kcne2−/− × Kcne2−/− crosses was 40% lower than that of pups from Kcne2+/+ × Kcne2−/− crosses at 5 weeks. Reduced body mass correlated with the maternal Kcne2−/− genotype, but also to pup genotype since pups from (+/−) sires were significantly larger than those from Kcne2−/− sires (both with Kcne2−/− dams) (Fig. 2d). By 15 weeks of age mean bodyweights were similar regardless of pup genotype (Supplementary Fig. 1a).

Figure 2

Kcne2 disruption causes embryonic lethality, dwarfism and alopecia

(a) Left, mean live births per litter for wild-type and Kcne2-disrupted crosses as indicated; n = 19–50 litters per group. *P<0.01. Right, genotype of surviving pups (%). n shown in parentheses.

(b) Exemplar 3-week-old pups from Kcne2−/− × Kcne2−/− and Kcne2+/+ × Kcne2+/+ crosses.

(c) Exemplar X-ray images of 5-week-old pups from Kcne2−/− × Kcne2−/− and Kcne2+/+ × Kcne2+/+ crosses.

(d) Mean body mass at 3–6 weeks of age for pups from wild-type and Kcne2-disrupted crosses as indicated; n = 15–50 pups per group. *P<0.05.

(e) Exemplar 5-week-old Kcne2+/+ and Kcne2−/− mice from homozygous crosses.

(f) Exemplar close-up views of skin from mice as in panel e.

(g) Exemplar micrographs of H&E-stained dermis sections from Kcne2−/− mouse as in panel e. Scale bar = 300 µm.

(h) Exemplar 1-year-old Kcne2+/+ and Kcne2−/− mice from Kcne2+/− × Kcne2+/− crosses.

(i) Exemplar micrograph of H&E-stained dermis sections from Kcne2−/− mouse in panel i. inset, detail from boxed region, showing border zone between normal hair and alopecia, with abrupt cessation of mature hair follicles. Scale bar = 600 µm.

(j) H&E-stained sections showing hair follicles in dermis from Kcne2+/+ and Kcne2−/− mice as in panel i. Scale bars = 20 µm.

Radiological examination revealed retarded skeletal development in the Kcne2−/− pups, producing dwarfism due to slow growth of both long bones and vertebrae. This was also apparent in the form of larger epiphyseal gaps and less ossification of the epiphyses in the large joints, which were irregularly shaped, fragmented, and heterogeneously sclerotic, characteristic of slow multifocal ossification (Fig. 2c).

Surprisingly, Kcne2−/− pups from homozygous crosses also exhibited striking alopecia of the trunk which began at 1–2 weeks of age and peaked at 4–5 weeks (Fig. 2b,e–g). Alopecia was also observed in aging Kcne2−/− mice from Kcne2+/− × Kcne2+/− crosses, initiating between the ears then spreading postero-dorsally with an abrupt loss of mature hair follicles at the transition zones (Fig. 2h–j).

Cardiomegaly, dwarfism and alopecia in Kcne2−/− mice are attenuated by wild-type surrogacy, T4 supplementation of dams, or T3/T4 administration to pups

Considering the combination of cardiac hypertrophy, cardiac and hepatic fibrosis, dwarfism, alopecia, and skeletal abnormalities indicative of retarded development, we investigated whether or not hypothyroidism might be the underlying cause. To test this hypothesis, we first determined serum TH concentrations. Indeed, serum T4 was 2-fold decreased, and TSH 2-fold increased, in 3-week-old Kcne2−/− pups from homozygous crosses compared to Kcne2+/+ pups from homozygous crosses (Fig. 3a), confirming the former were hypothyroid. Serum T4 and TSH were, however, normal in nubile 3–6 month old Kcne2−/− mice from heterozygous crosses (Supplementary Fig. 2), consistent with growth, litter size, and cardiac morphology trends (see Fig. 2 and 20). By age 12–15 months T4 and TSH were trending down and up respectively in mice from heterozygous crosses (Supplementary Fig. 2), consistent with a latent hypothyroidism, and the late onset of alopecia (Fig. 2h), cardiac hypertrophy and fibrosis (Fig. 1j). Thyroid glands from these aged mice bred from Kcne2+/− crosses showed a 40% greater mean mass post-mortem than Kcne2+/+ thyroids, with Kcne2+/− thyroids having intermediate mean mass, indicative of goiter formation due to Kcne2 disruption (Fig. 3 b). In contrast to nubile adult mice, pregnant Kcne2−/− dams exhibited almost 3-fold reduced serum T4 concentration compared to pregnant Kcne2+/+ dams (Figure 3 c), suggesting an explanation for the exaggerated phenotype and reduced litter size of Kcne2−/− pups bred from Kcne2−/− dams (Fig. 2). Consistent with previous reports for mice and rats (but in contrast to humans)21,22, pregnant mice in our study exhibited lower serum T4 than age- and genotyped-matched nubile adults (Fig. 3c; Supplementary Fig. 2).

Figure 3

Kcne2−/− mice are hypothyroid and treatable with T3/T4 or wild-type surrogacy

(a) Left, Serum T4 in Kcne2+/+ & Kcne2−/− mice at 3 wks. Right, serum TSH in Kcne2+/+ & Kcne2−/− mice at 3 wks. *significantly different from Kcne2+/+ by ANOVA, P < 0.001. Numbers in parentheses indicate n.

(b) Mean mass of thyroid glands from 12 month old Kcne2+/+, +/− and −/− mice from Kcne2+/− × Kcne2+/− crosses, weighed most-mortem, numbers in parentheses indicate n. Significant differences: * P < 0.005; ** P < 1 × 10−4; ***P < 1 × 10−9.

(c) Serum T4 in pregnant Kcne2+/+ & Kcne2−/− mice. *significantly different from Kcne2+/+, P < 0.001.

(d) Exemplar 3-week-old pups from homozygous crosses surrogated (Sgt) with dams of opposite genotype

(e) Mean body mass at 3–6 weeks of age for pups from wild-type and Kcne2-disrupted crosses surrogated (Sgt), or treated (Tx) with T3/T4 injection (P) or by T4 supplementation of their mothers (D); n = 9–23 pups per group. Untreated groups (/) from Figure 2 d shown for comparison.

(f) Exemplar 12-week-old Kcne2−/− mouse before (left) and after (right) 10 days QOD T3/T4 administration. Results were consistent in 5/5 mice evaluated.

(g) Left, histology of mouse in panel f after T3/T4 treatment showing recovery of normal hair follicles. Scale bar = 200 µm. Right, percentage of mice with normal hair growth grouped according to parental genotype. Groups were untreated (/), surrogated (Sgt) with dams (genotype as indicated) or treated (Tx) directly by QOD T3/T4injection (P) or by T4 supplementation of their mothers (D); n = 16–23 mice per group.

(h) Exemplar necropsies of 3-week-old Kcne2+/+ and Kcne2−/− mice bred from homozygous crosses, with/without surrogacy with mothers of opposite genotype.

(i) Mean heart weight/bodyweight (HW/BW; mg/g) for mice as in panel h, n = 7–9. *P<0.05 compared to values for non-surrogated pups (taken from Fig. 1c, key in Fig. 3j).

(j) Mean echocardiographic parameters for 3-week-old Kcne2+/+ and Kcne2−/− mice bred from homozygous crosses, with/without surrogacy with mothers of opposite genotype. d/s, diastolic/systolic; LVAW, left ventricular anterior wall thickness; LVID, left ventricular internal diameter; LVPW, left ventricular posterior wall thickness, n = 5 mice. *P<0.05 compared to values for non-surrogated pups (taken from Fig. 1c).

We next tested whether or not surrogacy of pre-weaning pups would alleviate any of the observed abnormalities. Significantly, normal body weight was fully restored in Kcne2−/− pups from Kcne2−/− × Kcne2−/− crosses by pre-weaning surrogacy with Kcne2+/+ dams (Fig. 3d,e). Conversely, pre-weaning surrogacy of Kcne2+/+ pups with Kcne2−/− dams resulted in mean pre-weaning body weight similar to Kcne2−/− pups, and intermediate body weight in post-weaning pups (Fig. 3d,e). These data suggested the possibility that maternal TH passed through milk – perhaps at higher concentrations, or in higher volumes of milk, from wild-type dams – were compensating for the defect in pups. Confirming the role of TH in body weight differences, Kcne2−/− pups born and raised by Kcne2−/− dams and from Kcne2−/− sires, showed significantly improved body weight by 3 weeks of age after T3/T4 administration every 48 hours (QOD) from birth (Fig. 3e, Supplementary Fig. 1b). Furthermore, T4 supplementation of Kcne2−/− dams, from 2 weeks pre-birth to weaning, also resulted in normal pup body weight (Fig. 3e).

Dermatologic disorders occur frequently in hypothyroidism23. Remarkably, here we found that alopecia was completely reversed in adult Kcne2−/− mice with 2 weeks QOD administration of T3/T4 (exemplars in Fig. 3f,g). Alopecia was also completely reversed in 19 of 21 Kcne2−/− pups by 10 days QOD T3/T4 administration, in 19/20 Kcne2−/− pups by surrogacy with Kcne2+/+ dams, and in 16/16 Kcne2−/− pups by T4 supplementation of their mothers from 2 weeks pre-birth to weaning. Conversely, alopecia was observed in 13/23 Kcne2+/+ pups surrogated with Kcne2−/− dams (Fig. 3g).

Hypothyroidism is associated with dilated and hypertrophic cardiomyopathies, reduced fractional shortening and heart failure24. Supporting a link between these cardiac defects and the hypothyroidism observed in Kcne2−/− mice, surrogacy with Kcne2+/+ dams resulted in a dramatic reduction in the relative mass of the heart compared to body weight; conversely, surrogacy of Kcne2+/+ pups with Kcne2−/− dams had the opposite effect (Fig. 3 h,i). As observed with non-surrogated Kcne2−/− pups (Fig. 1g), the liver of Kcne2+/+ pups surrogated with Kcne2−/− dams had an unusually pale appearance possibly indicative of fatty liver (Fig. 3h). Furthermore, echocardiographic determination of the potential beneficial effects of surrogacy of Kcne2−/− pups with Kcne2+/+ dams showed significant reduction in ventricular wall thickness and chamber diameter, and increased fractional shortening compared to non-surrogated Kcne2−/− pups (Fig. 3j).

KCNE2 and KCNQ1 form a TSH-stimulated thyrocyte K+ channel

Our data (Fig. 1–Fig. 3) suggested a potential role for KCNE2 in TH biosynthesis. Previous studies indicate that KCNE2 probably forms channel complexes with the KCNQ1 K+ channel α subunit in gastric epithelium14,15,25,26. Here, we found that both KCNE2 and KCNQ1 are expressed in human (Fig. 4a) and mouse (Fig. 4b–d) thyroid glands; in both species they partially co-localized with NIS, the basolateral membrane glycoprotein that mediates active I− transport, the first step in TH biosynthesis. Furthermore, thyroid follicular epithelia in Kcne2−/− mice exhibited abnormal architecture: compared to those in Kcne2+/+ mice, Kcne2−/− thyrocytes were often flattened and less abundant (Fig. 4e,f).

Figure 4

KCNE2 and KCNQ1 form a TSH-stimulated thyrocyte K+ channel

(a) Immunofluorescence using antibodies raised against KCNE2, KCNQ1 and NIS in human thyrocytes in sections from individuals with Grave’s Disease. DAPI visualization of nuclei shown in blue. *Colloid. Scale bars: 4 µm.

(b) Western immunoblots (IB) using antibodies raised against KCNE2 (arrows indicate the three glycosylation states of KCNE2) and KCNQ1 (lower arrow, monomer; upper arrow, multimer) in thyroid tissue from Kcne2+/+ and Kcne2−/− mice (from Kcne2+/− × Kcne2+/− crosses).

(c) Immunofluorescence using antibodies raised against KCNE2, KCNQ1 and NIS, in thyrocytes of 3-month-old Kcne2+/+ mice (from Kcne2+/− × Kcne2+/− crosses). *Colloid. Scale bars: 4 −µm.

(d) Immunofluorescence using antibodies raised against KCNE2 and KCNQ1, in thyrocytes of 3-monthold Kcne2−/− mice (from Kcne2+/− × Kcne2+/− crosses). *Colloid. Scale bars: 4 µm.

(e) H & E staining of adult Kcne2+/+ and Kcne2−/− thyroid glands (from Kcne2+/− × Kcne2+/− crosses); scale bars, 20 µm.

(f) Electron micrographs of thyroid epithelium from adult Kcne2+/+ and Kcne2−/− mice (from Kcne2+/− × Kcne2+/− crosses). Scale bars, 2 µm.

(g) Western blots of membrane fractions from FRTL5 cells with and without 10 hours incubation with cAMP, using antibodies raised against KCNE2 (arrows putatively indicate the three predicted glycosylation states); and FRTL5 cells with and without 6 days TSH incubation, using antibodies raised against KCNQ1 (lower arrow, monomer; upper arrow, tetramer).

(h) Exemplar whole-cell patch-clamp recordings of XE991-sensitive currents in FRTL5 cells with or without TSH incubation. Upper left inset: voltage protocol. Zero current level indicated by dashed line.

(i) Mean whole-cell total (n = 9–16) and XE991-sensitive (n = 8–9) current relationships for FRTL5 cells with or without TSH incubation. Significant difference: *P < 0.05.

We next sought to determine whether KCNQ1-KCNE2 K+ currents were expressed in thyrocytes, employing the highly functional rat thyroid-derived FRTL5 cell line. We detected endogenously-expressed KCNQ1 and KCNE2 proteins, which appeared to be upregulated by TSH or its major downstream effector, cAMP, in FRTL5 cell membrane fractions (Fig. 4g). We then measured endogenous currents from FRTL5 cells using patch-clamp recording in the whole-cell configuration. A TSH-stimulated K+ current in FRTL5 cells bore the signature linear current-voltage relationship of KCNQ1-KCNE2 channels and was inhibited by the KCNQ-specific antagonist XE991 (Fig. 4h,i). In sum, KCNQ1-KCNE2 channels are expressed in human and rodent thyrocytes, where they generate a TSH-stimulated, constitutively-active K+ current.

KCNE2 is required for normal thyroid I− accumulation

TH requirements are especially high in early development: developing fetuses and neonates rely upon not only their own TH biosynthesis, but also maternal T4 in utero and perhaps via milk. Therefore, we examined thyroid I− accumulation, a critical step in TH biosynthesis, in lactating dams and their pups. We injected 124I only into the tail vein of lactating dams, which were then placed back together with their pups to feed them. Both dams and pups were imaged by positron emission tomography (PET). Kcne2−/− dams showed a striking defect in 124I accumulation in the thyroid, with 4-fold less accumulation over the first hour post-injection and continuing deficiency in the following three days (Fig. 5a–c). In pups, whose sole source of 124I was dams’ milk, Kcne2 deletion caused a 5-fold reduction of pup thyroid 124I accumulation 24-hours post-injection of dams, and continuing deficiency for the following 2 days (Fig. 5d,e). When normalized to stomach 124I count the thyroid 124I count was reduced 8-fold in Kcne2−/− pups compared to wild-type pups at 72 hours post-injection of the dam (Fig. 5f).

Figure 5

Kcne2 is required for normal thyroid I− accumulation

(a) Exemplar microPET images of lactating Kcne2+/+ and Kcne2−/− dams recorded during the first hour after tail vein injection of 124I. Labels: t, thyroid; m, mammary glands. Color intensity scale on right, showing red indicates highest intensity. Scale bar (white) = 5 mm.

(b) Mean 124I accumulation in thyroid relative to mammary gland from imaging as in panel a, n = 4 mice per group, error bars indicate SEM. Measured as ratio of maximum radioactivity in each tissue minus mean background count in each mouse. *P<1×10−8.

(c) Mean 124I accumulation in thyroid relative to mammary gland as in panel a but from 1–72 hours after tail vein injection of 124I, n = 4–5 mice per group, error bars indicate SEM. *P<0.05.

(d) Exemplar microPET images of pre-weaning Kcne2+/+ and Kcne2−/− pups recorded 24–72 hours after their mothers (same dams as in panels a–c) received tail vein injections of 124I. Labels: t, thyroid; s, stomach. Intensity scale as in panel a. Scale bar (white) = 5 mm.

(e) Mean 124I accumulation (in µCi) in thyroid and stomach from imaging as in panel d, n = 7–12 pups per time-point per group, error bars indicate SEM. Measured as maximum radioactivity in each tissue minus mean background count in each mouse. *P<0.05.

(f) Mean 124I accumulation in thyroid relative to stomach, imaging method and pups as in panel e. Measured as ratio of maximum radioactivity in each tissue minus mean background count in each mouse. *P<0.0005.

Thus, Kcne2 deletion causes a thyroid I− accumulation defect, which in turn causes a TH biosynthesis defect, the gross phenotypic effects of which are particularly striking in pre-weaning pups feeding from Kcne2−/− dams. To examine the mechanistic basis for this, and for the beneficial effects of surrogacy by Kcne2+/+ dams, we first performed PET on pups surrogated with dams of opposite genotype, after tail vein injection of lactating dams with 124I followed by imaging of pups feeding from them. Strikingly, we found that Kcne2−/− pups feeding from Kcne2+/+ dams had higher stomach and thyroid 124I counts (measured as peak counts/cc), and higher thyroid:stomach count ratio, than Kcne2+/+ pups feeding from Kcne2−/− dams (Fig. 6a–c). This suggested that the surrogating dams’ genotype was critical in determining thyroid 124I uptake of pups, although pup genotype was also important because when pups of either genotype fed from Kcne2+/+ dams, Kcne2+/+ pups still had an almost 2-fold higher thyroid:stomach count ratio at 48–72 hrs compared to Kcne2−/− pups (Fig. 5f, Fig. 6c).

Figure 6

Kcne2−/− dams have a milk ejection defect and produce low-T4 milk

(a) Exemplar microPET images of pre-weaning Kcne2+/+ and Kcne2−/− pups recorded 24–96 hours after their surrogate mothers of opposite genotype received tail vein injections of 124I. Labels: t, thyroid; s, stomach. Intensity scale shown on right. Scale bar (white) = 10 mm.

(b) Mean peak 124I accumulation (measured as peak µCi cc−1) in thyroid and stomach from imaging as in panel a, n = 7–12 pups per time-point per group, error bars indicate SEM. Measured as maximum radioactivity in each tissue minus mean background count in each mouse. *P<0.05; **P<0.01.

(c) Mean peak 124I accumulation in thyroid relative to stomach, imaging method and pups as in panel a. Measured as ratio of maximum radioactivity in each tissue minus mean background count in each mouse. **P<0.001.

(d) Effects of pup and dam genotype on pup total thyroid and body 124I accumulation (left), and thyroid RAIU (radioactive iodide uptake as a percentage of total body radioiodine) (right), determined using 3D regions of interest from PET analyses of pup 124I accumulation (surrogated and non-surrogated) after tail vein injection of dams; n = 7–12. *P<0.05; **P<0.005; ***P<0.0005.

(e) Milk ejection assay. Mean mass of pups before (0 min) and after 30 min feeding period; n = 8–20; *P<0.01. ‘oxy’ indicates dams injected with oxytocin 10 minutes prior to feeding period.

(f) Milk T4 concentration; n = 5–7; *P<0.05.

(g) Plasma I− concentration in 3-week-old pups; n = 6 per genotype.

Total thyroid and total body activity was also quantified for all surrogated and non-surrogated pups, thereby permitting comparison of thyroid radioactive iodide uptake (RAIU) as a percentage of whole body activity (Fig. 6d). Total thyroid 124I was higher in all pups when feeding from Kcne2+/+ dams than when feeding from Kcne2−/− dams, whereas total body 124I was only significantly higher than other groups for Kcne2−/− pups feeding from Kcne2+/+ dams (Fig. 6d left). In contrast, thyroid RAIU – a measure of the efficiency of the thyroid at accumulating 124I from the available total body 124I – was only significantly higher than other groups in Kcne2+/+ pups feeding from Kcne2+/+ dams (Fig. 6d right). These data again demonstrated that Kcne2−/− pups’ thyroids are less efficient than those of Kcne2+/+ pups at accumulating I−, but also indicated that Kcne2−/− dams supply less I− to their pups than do Kcne2+/+ dams. These results also revealed that Kcne2−/− pups are relatively better at accumulating total body I− than Kcne2+/+ pups.

Examining first the poor delivery of I− from Kcne2−/− dams, we compared milk ejection from Kcne2+/+ and Kcne2−/− dams by weighing pups before and after feeding, as previously described27. This led to the discovery that Kcne2−/− dams have a highly significant milk ejection defect, manifested as pups (of either genotype) failing to gain weight (from milk ingestion) during the first 30 min of feeding from Kcne2−/− dams, in sharp contrast to pups feeding from Kcne2+/+ dams (Fig. 6e). Similar results were seen over the first 60 min of feeding (Supplementary Fig. 3). Importantly, Kcne2+/+ and Kcne2−/− pups showed no significant difference in their feeding rate measured by weight gain (Fig. 6e, Supplementary Fig. 3). Pups were latched on to dams of either genotype for the entire period under study (30 or 60 min). Thus, the milk ejection defects were not related to behavioral differences in either pups or dams, and instead reflected a pathophysiological defect in milk ejection by Kcne2−/− dams. Hypothyroid rats were previously shown to have impaired milk ejection due to reduced serum oxytocin compared to euthyroid rats27. Accordingly, here we found that injection of Kcne2−/− dams with oxytocin returned their milk ejection to the same level as Kcne2+/+ dams (Fig. 6e). The milk ejection defect was probably a dominant factor in the beneficial effects of Kcne2+/+ surrogacy, and the negative effects of Kcne2−/− surrogacy. Additionally, however, we discovered that milk from Kcne2−/− dams contained only half as much T4 as that from Kcne2+/+ dams (Fig. 6f), potentially also contributing to the observed effects of surrogacy.

Finally, addressing the superior total body accumulation of I− by Kcne2−/− pups, we found equal serum I− concentrations in non-surrogated 3-week-old Kcne2+/+ and Kcne2−/− pups (Fig. 6g), suggesting that despite inferior milk ejection by Kcne2−/− dams, Kcne2−/− pups were able to maintain normal plasma I− concentrations. This was not surprising given that hypothyroidism is known to result in decreased I− excretion28.

DISCUSSION

More than a decade ago, KCNQ1 mRNA was found to be expressed at a higher level in human thyroid than in the heart or stomach29, but its role in the thyroid has not previously been reported. Furthermore, kcnq1 gene-disrupted mice were previously found, like our Kcne2−/− mice, to have enlarged hearts and thickened ventricular walls, but the mechanistic basis for this was not described30,31.

T3 and T4 biosynthesis requires active I− transport in the thyroid, where I− concentrations reach 20–40 times that of the plasma. NIS, located on the basolateral side of the thyrocytes – thyroid epithelial cells which encircle the colloid – transports I− into the thyrocyte; at the cell/colloid interface I− ion is oxidized and covalently incorporated into thyroglobulin, for TH production17. NIS function requires a basolateral Na+/K+-ATPase for Na+ efflux but the necessity for other channels or transporters in this process is not known. Here, KCNQ1-KCNE2 is identified as a TSH-stimulated thyrocyte K+ channel critical for normal thyroid I− accumulation, and probably expressed predominantly at the basolateral membrane.

The dramatic effects of surrogacy in the current study add to the debate over whether or not maternal T4 is at high enough concentrations in milk to deliver therapeutic effects in hypothyroxinaemic newborns32. Our findings suggest mouse milk TH could be beneficial in this context, as there are significant levels of T4 in milk, reduced by Kcne2 disruption (Fig. 6f), albeit the poor development of pups feeding from Kcne2−/− dams probably arises from a combination of this and the impaired milk ejection of Kcne2−/− dams (Fig. 6e). The mechanisms underlying the whole-animal and molecular effects of surrogacy appear complex, as one would expect. We speculate that the thyroid I− accumulation of Kcne2−/− pups is diminished by Kcne2 deletion but that this is partially balanced by e.g., adaptation to developing in a low maternal T4 environment in the womb (Fig. 3c) and being initially fed with poorly-ejected, low-T4 milk (Fig. 6e,f). The end result is that Kcne2−/− pups are less efficient at accumulating thyroid I− compared to Kcne2+/+ pups when either are fed by Kcne2+/+ dams, but when fed by Kcne2−/− dams their thyroid RAIUs are similar (Fig. 6d). Part of this adaptation may involve reduced I− excretion by Kcne2−/− pups, as previously reported28 and supported by our current data (Fig. 6d,e,g). Interestingly, as observed for NIS16,33, KCNQ1 is also expressed in mammary gland epithelium, where it may co-assemble with KCNE3 to play a role in K+ homeostasis34. While a role for KCNE2 in mammary epithelial function should not be ruled out, our PET data demonstrate that mammary gland I− uptake is not impaired in Kcne2−/− dams (Fig. 5). Nevertheless, the phenotypes we describe herein for Kcne2−/− pups bred from homozygous Kcne2−/− crosses include features such as alopecia and cardiac hypertrophy, not always observed in hypothyroid mouse models35. While this may at least partly be explained by the combination of both Kcne2−/− dam and pup (heterozygous crosses are typically employed), it could possibly indicate additional pathogenesis beyond thyroid impairment but successfully treatable by TH supplementation.

Human thyroid dysfunction negatively impacts the brain, heart and GI tract; fatalities may occur from thyroid storm in hyperthyroidism, and myxedema coma in hypothyroidism36. In addition, thyroid dysfunction during pregnancy increases the risk of adverse maternal and fetal outcomes37–39. Subclinical human maternal hypothyroxinemia causes severe neurodevelopmental disorders40, may include changes in blood lipid profile, myocardial function, and neuropsychiatric function41–43, and is an independent risk factor in heart failure due to structural and electrical remodeling in the heart24. Importantly, a SNP near KCNE2 was recently shown to associate with early-onset myocardial infarction44 – suggesting the possibility of a genetic link to previously-reported subclinical hypothyroidism-associated accelerated coronary artery disease and myocardial infarction45.

Subclinical hypothyroidism is also associated with prolonged QTc46, a hallmark of loss-off-function mutations in KCNE2 and KCNQ13,6, and with AF, an increasingly prevalent disease in the aging population47,48 that is also associated with some KCNQ1 and KCNE2 gene variants 9,12. As many as 13% of patients with idiopathic AF exhibit biochemical evidence of hyperthyroidism49; in one study, 62% of 163 patients reverted to sinus rhythm within 8–10 weeks after treatment for hyperthyroidism returned them to a euthyroid state50. Therefore, the finding here that KCNE2-KCNQ1 channels contribute to thyroid function raises the tantalizing hypothesis that there is a thyroid component to some KCNE2- or KCNQ1-associated cardiac arrhythmias. In previous studies of sudden cardiac or unexplained death, it was often assumed that ion channel gene mutations were not causative in those cases exhibiting overt structural heart disease upon autopsy51; historically, ‘electrical’ heart diseases arising from ion channel defects have mostly been considered genetically distinct from ‘structural’ heart disease, although variants in the human SCN5A Na+ channel gene have been associated with dilated cardiomyopathy52. Our current findings suggest reconsideration of patients with structural heart disease exhibiting ventricular or atrial arrhythmias, given the possibility that mutations in KCNQ1 and KCNE2 could be arrhythmogenic due both to primary electrical defects in myocyte K+ channels containing these subunits, and to cardiac structural abnormalities arising secondarily from thyroid dysfunction due to defective thyroid KCNQ1-KCNE2 channels.

Identification of KCNE2-KCNQ1 as a thyrocyte channel important for I− accumulation may also have therapeutic implications. Agonists and antagonists of KCNQ1-KCNE2 channels have already been developed. Because the pharmacology of KCNQ1-KCNE2 complexes is markedly different from that of homomeric KCNQ1, KCNQ1-KCNE1, or KCNQ1-KCNE3 channels53, identification of the requirement of KCNQ1-KCNE2 complexes for normal thyroid function may permit semi-specific, reversible pharmacological targeting of the KCNQ1-KCNE2 complex to treat thyroid disease.