Nodule-Specific NRF2-Targeted Upregulation in Patients With KEAP1 Mutations and Familial Nontoxic Multinodular Goiter

Abstract

Context

Kelch-like ECH-associated protein 1 (KEAP1) is associated with nuclear factor erythroid-2–related factor 2 (NRF2) and promotes NRF2 degradation in normal conditions. Genetic abnormality in KEAP1 is a rare disease and presents with familial multinodular goiter.

Objective

This study assessed the clinical and molecular findings concerning nodular formation in the thyroid gland of patients harboring KEAP1 germline mutations.

Methods

Next-generation sequencing analysis targeting goiter-associated genes was performed on 39 patients with familial multinodular goiter. The expression of NRF2-targeted genes from surgical thyroid specimens of patients with KEAP1 mutations were analyzed using a whole-transcript expression array and immunohistochemistry.

Results

We found 5 probands with pathogenic heterozygous mutations in KEAP1 (p.Q86*, p.L136P, p.V411fs, p.R415C, and p.R483H) that had no meaningful concomitance with mutations of other goiter-associated genes at germline and somatic levels. Their common histopathological features showed multinodular goiters in the entire thyroid gland with few degenerative lesions or complications of malignancy and slow proliferation indicating less than 1% at the Ki-67 labeling index. Among 42 NRF2-targeted genes, antioxidant genes were most frequently upregulated (11/12) in the nodule, followed by detoxification genes (6/11). Immunohistochemical analysis showed relatively high expression of glutathione peroxidase 2 and NAD(P)H quinone oxidoreductase 1 (representative NRF2-targeted genes) in the nodules of various patients harboring KEAP1 mutations.

Conclusion

KEAP1 germline heterozygous mutations exert excessive NRF2 activity in the thyroid gland and may confer cytoprotective effects even under abundant reactive oxygen species associated with thyroid hormone production, resulting in thyroid hyperplasia with scarce degradation.

Multinodular goiter presents with multifocal hyperplastic/neoplastic lesions, and these nodules are frequently clonal. In the 2022 World Health Organization classification of thyroid neoplasms, multinodular goiter has been introduced in classifying benign tumors; the term thyroid follicular nodular disease is based on clonal proliferation (1). Most familial multinodular goiter cases present with an autosomal dominant pattern of inheritance. Indeed, several genetic abnormalities, including DICER1, KEAP1, and DGCR8, have been identified in familial multinodular goiter (2-4).

Kelch-like ECH-associated protein 1 (KEAP1) was originally found to be associated with nuclear factor erythroid-2–related factor 2 (NRF2) (5) and functions as a substrate adaptor protein for a Cul3-dependent E3 ubiquitin ligase complex, with subsequent degradation of NRF2 by the proteasome (6). NRF2 activates the transcription of various cytoprotective genes that enhance cell proliferation (7). Somatic mutations in KEAP1 in cancer tissues and cancer-derived cell lines are advantageous for cell growth (7). However, germline mutations in KEAP1 have not been reported in cases of familial cancer; it has been detected in only 2 families with multinodular goiter (3, 8). The thyroid tissue–specific phenotype also applies to Keap1 knockdown mice that develop diffuse goiter without thyroid nodules or hyperplasia (9). Why KEAP1 germline mutations are susceptible to a thyroid-predominant phenotype remains unknown.

Oxidative stress or electrophiles inactivate KEAP1 by modifying cysteine residues, which results in decreased ubiquitin activity and nuclear accumulation of NRF2 from the cytoplasm. In thyroid tissues with a KEAP1 germline heterozygous mutation (R483H), NRF2-positive staining of the nuclei is detected in more than 30% of cells in thyroid nodules and less than 10% of cells in nonnodular parenchyma (8). External stimuli and genetic changes lead to alterations in the cellular localization of NRF2.

In this study, we investigated the clinical and molecular findings associated with 5 different KEAP1 germline heterozygous mutations, including 4 newly identified mutations.

Materials and Methods

Screening of Goiter-associated Gene Variants Using Next-Generation Sequencing

Genetic testing was retrospectively performed on 39 patients with familial multinodular goiter (multinodular goiter detected in a patient and relatives within the second degree of consanguinity) who visited Kuma Hospital between 2014 and 2021 and agreed to genetic testing (Fig. 1). Of the 39 patients, 11 underwent thyroidectomy during their clinical courses. Analysis was performed by next-generation sequencing (NGS) targeting most of the coding exons of 24 goiter-associated genes (Goiter Ampliseq panels A and B, Table 1) using custom primers designed using Ion Ampliseq Designer (Life Technologies), according to the manufacturer's instructions. Briefly, genomic DNA was isolated from whole blood using a QIAamp DNA Blood Mini Kit (Qiagen), and a multiplex polymerase chain reaction of A, 429 and B, 263 amplicons was performed, followed by the addition of IonCode Adaptors using an Ion Chef system (Life Technologies).

Figure 1.

Flowchart of the study population.

Abbreviations: FFPE, formalin-fixed paraffin-embedded; MNG, multinodular goiter.

Table 1.

Genes included in goiter Ampliseq

Gene	Chromosome
Panel A
NRAS	Chr1
TPO	Chr2
THRB	Chr3
IYD	Chr6
SLC26A4	Chr7
TG	Chr8
RET	Chr10
HRAS	Chr11
KRAS	Chr12
TSHR	Chr14
DICER1	Chr14
DUOX2	Chr15
DUOXA2	Chr15
SLC5A5	Chr19
KEAP1	Chr19
Panel B
RGS12	Chr4
GRPEL1	Chr4
WFS1	Chr4
APC	Chr5
SLC26A7	Chr8
WRN	Chr8
PTEN	Chr10
PRKAR1A	Chr17
GNAS	Chr20
Panel C
NRAS	Chr1
TPO	Chr2
ZNF148	Chr3
THRB	Chr3
WFS1	Chr4
GRPEL1	Chr4
RGS12	Chr4
IYD	Chr6
KMT2C	Chr7
SLC26A4	Chr7
BRAF	Chr7
TG	Chr8
SLC26A7	Chr8
ARFGEF1	Chr8
WRN	Chr8
RET	Chr10
PTEN	Chr10
HRAS	Chr11
KRAS	Chr12
TUBGCP3	Chr13
TSHR	Chr14
DICER1	Chr14
DUOX2	Chr15
DUOX1	Chr15
DUOXA2	Chr15
SPOP	Chr17
EZH1	Chr17
PRKAR1A	Chr17
KEAP1	Chr19
SLC5A5	Chr19
GNAS	Chr20
EIF1AX	ChrX
KDM5C	ChrX

To verify somatic mutations in thyroid tumors, DNA was isolated from 4 samples of available formalin-fixed paraffin-embedded (FFPE) thyroid tissues using a QIAamp DNA Mini Kit (Qiagen), and the same procedures as germline mutations for 1414 amplicons were performed by NGS analysis using Goiter Ampliseq panel C.

After enrichment by clonal emulsion polymerase chain reaction on Ion Sphere particles, the barcoded libraries were loaded on an Ion 318 chip, Ion PI chip, or Ion 540 chip, and massively parallel sequencing was conducted on an Ion Torrent PGM sequencer, Ion Proton sequencer, or Ion GeneStudio S5 system with an Ion PGM Hi-Q View Chef kit, Ion PI Hi-Q Chef kit, or Ion 540 Kit-Chef, respectively. We analyzed the raw signal data of NGS using Torrent Suite v.5.2.2 or v.5.4, including adaptor trimming, read alignment to the human genome 19 reference, coverage analysis, and variant calling. We performed variant filtration and annotation using Ion Reporter v.5.0 (Life Technologies) to determine whether the detected sequence variants were known pathogenic mutations or novel variants.

The pathogenicity of 3 missense variants was estimated using computational algorithms, PolyPhen2 (http://genetics.bwh.harvard.edu/pph2/index.shtml), PROVEAN (http://provean.jcvi.org/seq_submit.php), and PANTHER (http://www.pantherdb.org/tools/csnpScoreForm.jsp), and evaluated as “disease-causing or deleterious.” The frequency of each variant in the general population was evaluated using gnomAD (https://gnomad.broadinstitute.org/)

Affymetrix Whole-Transcript Expression Array

Thyroid specimens resected from the nodule and nonnodular parenchyma (∼100 mg each) available from a patient with a KEAP1 mutation (L136P, IV-1) were quickly immersed in RNAlater Stabilization Solution (Thermo Fisher Scientific). Total RNA was extracted using the RNeasy Plus Universal kit (Qiagen) according to the manufacturer's instructions. RNA purity and integrity were evaluated using an ND-2000 Spectrophotometer (NanoDrop, Thermo Fisher Scientific) and an Agilent 2100 Bioanalyzer (Agilent Technologies). The Affymetrix Whole-transcript Expression array process was executed according to the manufacturer's protocol (GeneChip Whole Transcript PLUS reagent Kit). Complementary DNA was synthesized using a GeneChip WT amplification kit, according to the manufacturer's instructions. The sense complementary DNA was then fragmented and biotin-labeled with terminal deoxynucleotidyl transferase using a GeneChip WT Terminal labeling kit. Approximately 5.5 μg of labeled DNA was hybridized to the Affymetrix GeneChip Human Clariom S Array at 45 °C for 16 hours. Hybridized arrays were washed and stained on GeneChip Fluidics Station 450 and scanned on a GCS3000 Scanner (Thermo Fisher Scientific). Signal values were calculated using Affymetrix GeneChip Command Console software. Differentially expressed genes were selected at a P value of less than .05 and a threshold of fold change (FC) value ≥2.0 or ≤−2.0.

Histopathological Evaluation and Immunohistochemistry

After surgical resection, thyroid tissues were fixed in 10% neutral buffered formalin, and the specimens were embedded in paraffin. Serial sections (3-µm thick) were cut from each paraffin block. The sections were stained with hematoxylin-eosin (HE) for light-microscopic examination. Immunostaining for human Ki-67 (MIB1, 1:200; Dako), human thyroglobulin (TG; 1D4, 1:400; Leica Microsystems), human NAD(P)H quinone oxidoreductase 1 (NQO1; A180, 1:400; Santa Cruz), and human glutathione peroxidase 2 (GPX2; EPR8175(2), 1:800; Abcam) was performed using a Leica Bondmax system (Leica Microsystems) and a Bond refine kit (Leica Microsystems), according to the manufacturer's instructions. The Ki-67 labeling index was calculated as the percentage of positively stained nuclei at 400× magnification. Control specimens were obtained from 4 unrelated patients with large multinodular goiters (>100 g), who were negative for antithyroid peroxidase (TPO) and anti-TG antibodies (Abs).

This study was approved by the ethics committee of Kuma Hospital, and informed consent was obtained from the patients and their family members for using their samples in this study.

Results

Identification and Clinical Characteristics of 5 Families With a KEAP1 Germline Mutation

Genetic analysis of 39 patients with familial multinodular goiter (see Fig. 1) identified 4 newly heterozygous variants in germline KEAP1 (c.256C > T, p.Q86*; c.407T > C, p.L136P; c.1231_1232insG, p.V411fs; and c.1243C > T, p.R415C), in addition to the previously reported variant (c.1448G > A, p.R483H) (8). They were located at the cullin 3- and NRF2-binding regions; 2 were stop codons or frameshifts, and the other 3 were missense mutations (Fig. 2A and 2B). The 3 missense variants were evaluated as “probably damaging or deleterious” (Table 2). Although genetic analysis of family members was limited to almost half (9/20) of the patients with multinodular goiter in 5 families (see Fig. 2B), we judged these variants to be very strong to moderate pathogenic variants, based on the American College of Medical Genetics and Genomics guidelines (10).

Figure 2.

Identifying families with Kelch-like ECH-associated protein 1 (KEAP1) mutations. (A) Localization of germline KEAP1 mutations. The structure is represented by functional regions and corresponding amino acid numbers. Five mutations were identified in our hospital and 1 mutation was reported in another institute. (B) Pedigree of the investigated family. Individuals affected by goiter are indicated by filled symbols. The arrow shows the proband. The bars above each symbol indicate individuals who underwent sequencing analysis for KEAP1.

Table 2.

Genetic and clinical characteristics of 5 probands with KEAP1 variants in familial multinodular goiter

KEAP1 variants (heterozygous)	p.Q86*	p.L136P	p.V411fs	p.R415C	p.R483H
Frequency (gnomAD)	<0.001%	<0.001%	<0.001%	<0.001%	<0.001%
PolyPhen2 (cutoff: 0.5)	Not subject	Probably damaging (1)	Not subject	Probably damaging (0.997)	Probably damaging (0.994)
PROVEAN (cutoff: –2.5)	Not subject	Deleterious (−6.609)	Not subject	Deleterious (−7.751)	Deleterious (−4.42)
PANTHER (cutoff: –3)	Not subject	Deleterious (−4.17968)	Not subject	Deleterious (−5.05895)	Deleterious (−4.09048)
Sex	Female	Female	Male	Male	Female
Onset of goiter, age in years	13	22	17	37	35
Complication disease	None	None	None	None	Graves disease
Thyroid function	Euthyroid	Euthyroid	Euthyroid	Euthyroid	Euthyroid∼ hyperthyroid
Tg level, ng/mL	1125	4769	736	91	742
TgAb/TPOAb	Negative	Negative	Negative	Negative	Negative (TRAb positive)
Maximum tumor size, mm	39	46	57	47	41
Tumor localization	Bilateral	Bilateral	Bilateral	Bilateral	Bilateral
Thyroidectomy	Done (another hospital)	Done	Done	Done	Done
Thyroid weight, g	No data	No data	690	No data	256

Abbreviations: KEAP1, Kelch-like ECH-associated protein 1; Tg, thyroglobulin; TgAb, thyroglobulin antibodies; TPOAb, thyroid peroxidase antibodies.

Clinical findings on the probands of 5 families with abnormalities of KEAP1 showed that all variants were heterozygous, and the age of goiter onset ranged between 13 and 37 years. TG levels were elevated, and TGAb and TPOAb levels were negative. All multinodular lesions filled both lobes of the thyroid gland, and the entire thyroid gland was markedly enlarged, which was pathologically confirmed by mixing adenomatous goiter and small follicular adenoma with few degenerative lesions (see Table 2;Fig. 3). The Ki-67 labeling index of all nodules was <1% (see Fig. 3). Next, we evaluated the clinical features of the total number of family members (n = 51) of 5 probands (see Fig. 1). While these 5 families consisted of 20 members with multinodular goiter, 12 patients underwent thyroidectomy based on the surgical indication criteria at our institution (11), 3 of whom underwent multiple surgeries (Figs. 1 and 2B). Thyroid nodules of family members other than the 5 probands had similar pathological findings to those of the probands. A total of 12 family members underwent genetic testing of the KEAP1 gene, including 9 patients with multinodular goiter and 3 family members without multinodular goiter (see Fig. 2B). All patients with multinodular goiter harbored KEAP1 mutations. Of the 3 members without multinodular goiter, 2 (R483H I-6 and III-2; see Fig. 2B) harbored no mutation in KEAP1, and 1 (R483H III-1; see Fig. 2B) had the same mutation as the proband but had no goiter or nodular lesions at age 22 at follow-up. While no thyroid cancer or neoplastic lesions in other organs were observed in the probands (see Table 2), 2 family members (Q86* I-4 and L136P III-1; see Figs. 1 and 2B) had a history of anaplastic thyroid and esophageal cancers, respectively. The patient case with anaplastic thyroid cancer was not tested for the KEAP1 mutation, while the patient case with esophageal cancer was confirmed to harbor the KEAP1 mutation.

Figure 3.

(A-D) Histopathological features and the (E) and (F) Ki-67 labeling index in the thyroid nodules of 4 probands with Kelch-like ECH-associated protein 1 (KEAP1) mutations.

Somatic gene testing of 4 FFPE tissues showed that these 4 samples harbored KEAP1 mutations identical to germline mutation. Moreover, the allele frequency of somatic mutation was greater than 80% in L136P and R483H (Table 3).

Table 3.

Comparison of variant allele frequency between germline and somatic mutations

Germline					Somatic
Gene	Amino acid change	Coding	VAF, %	Coverage reference, variant	Gene	Amino acid change	Coding	VAF, %	Coverage reference, variant
KEAP1	p.L136P	c.407T > C	48.9	T = 2041, C = 1952	KEAP1	p.L136P	c.407T > C	86.8	T = 263, C = 1723
					RET	p.Q796*	c.2386C > T	24.6	C = 211, T = 69
KEAP1	p.V411fs	c.1231_1232insG	43.8	T = 293, TG = 228	KEAP1	p.V411fs	c.1231_1232insG	45.5	T = 1078, TG = 901
KEAP1	p.R415C	c.1243C > T	54.0	C = 182, T = 214	KEAP1	p.R415C	c.1243C > T	53.2	C = 936, T = 1064
					DICER1	p.R1003*	c.3007C > T	7.11	C = 235, T = 18
					PTEN	p.G338R	c.493G > A	9.46	G = 1851, A = 145
KEAP1	p.R483H	c.1448G > A	48.0	G = 302, A = 279	KEAP1	p.R483H	c.1448G > A	82.8	G = 689, A = 3309

Abbreviations: KEAP1, Kelch-like ECH-associated protein 1; VAF, variant allele frequency.

NFR2-targeted Gene Expression in the Thyroid Nodule and Nonnodular Parenchyma in a Patient With Mutated KEAP1

A comprehensive gene expression profile in the thyroid nodule and nonnodular thyroid parenchyma of a patient harboring a KEAP1 mutation (p. L136P) was performed using an RNA microarray. We focused on NRF2-targeted genes that had the binding site antioxidant response element (ARE) at their promoter and enhancer regions and also had experimentally identified functional ARE in humans (12-14). We found 42 NRF2-targeted genes among the approximately 20 000 analyzed genes. The FC values of these NRF2-targeted genes showed that messenger RNA expression of 24 genes was significantly upregulated (2.03 to 496.64) in the thyroid nodule; that of 1 gene was significantly downregulated (−6.75), and the expression of the remaining 17 genes was not significantly different (−1.93 to 1.83; Table 4). Antioxidant genes were most frequently upregulated (11/12) followed by detoxification genes (6/11) in the 4 functional categories.

Table 4.

RNA microarray analysis of experimentally identified functional NRF2-targeted genes in a case of KEAP1 mutation (p. L136P)

Gene	FC valuea	Expression		Function
		Nodule	Nonnodule
GCLC	36.41	15.68	10.49	Antioxidant
GCLM	3.80	10.79	8.86	Antioxidant
GLRX	3.17	10.84	9.17	Antioxidant
GPX2	496.64	14.78	5.82	Antioxidant
GSR	10.38	14.56	11.18	Antioxidant
PRDX1	2.40	16.74	15.48	Antioxidant
PRDX6	1.08	14.83	14.72	Antioxidant
SLC7A11	82.33	14.23	7.87	Antioxidant
SOD1	2.13	15.04	13.95	Antioxidant
SRXN1	4.29	10.44	8.34	Antioxidant
TXN	7.55	16.29	13.37	Antioxidant
TXNRD1	18.54	16.43	12.22	Antioxidant
ABCB11	−1.88	3.20	4.11	Detoxification
ABCB6	9.36	9.23	6.01	Detoxification
AKR1C1	45.84	17.06	11.55	Detoxification
AKR1C2	95.69	18.00	11.42	Detoxification
APOA1	1.15	5.76	5.56	Detoxification
GSTP1	2.23	16.05	14.90	Detoxification
MGST2	−1.13	10.70	10.88	Detoxification
NQO1	55.47	16.93	11.14	Detoxification
NQO2	−1.20	8.94	9.21	Detoxification
UGT1A1	14.70	7.12	3.24	Detoxification
UGT2B7	1.16	3.71	3.50	Detoxification
BLVRB	2.62	13.29	11.90	Heme and iron metabolism
ETS1	−1.93	10.04	10.98	Heme and iron metabolism
FECH	2.60	9.81	8.43	Heme and iron metabolism
FTH1	1.93	17.73	16.78	Heme and iron metabolism
FTL	2.03	16.39	15.37	Heme and iron metabolism
HMOX1	−1.18	6.27	6.50	Heme and iron metabolism
MT1B	−1.93	15.15	16.10	Heme and iron metabolism
ATF3	−1.37	12.69	13.14	Others
BACH1	1.83	12.78	11.91	Others
GNAI2	1.46	12.07	11.52	Others
KRT16	−1.34	8.24	8.66	Others
MAFG	2.32	7.02	5.81	Others
MAPT	−1.11	6.40	6.55	Others
ME1	11.75	13.07	9.51	Others
PSMA3	2.88	12.86	11.34	Others
PTGS2	−6.75	5.15	7.91	Others
S100A6	1.64	12.86	12.14	Others
TBXAS1	5.31	10.24	7.84	Others
TG	1.04	18.28	18.22	Others

Abbreviations: FC, fold change; KEAP1, Kelch-like ECH-associated protein 1; NRF2, nuclear factor erythroid-2–related factor 2.

^a The significant FC values are expressed in bold.

Immunohistochemical Analysis of Thyroglobulin, NAD(P)H Quinone Oxidoreductase 1, and Glutathione Peroxidase 2 in the Thyroid Gland of Patients With KEAP1 Mutations

Immunohistochemical analysis of TG, NQO1, and GPX2 was performed using thyroid tissues of the same patient with the L136P mutation, and FC values in the thyroid nodule were 1.04, 55.47, and 496.64, respectively (see Table 4 and Fig. 4). While TG expression in the nodule was equivalent to that in the nonnodular parenchyma, NQO1 and GPX2 expression was higher in the nodule than in the nonnodular parenchyma (Fig. 5B-5D). In the patient with multinodular goiter who did not harbor a KEAP1 mutation, TG expression in the nodule was equivalent to that in the nonnodular parenchyma, while NQO1 and GPX2 expression was very low in both regions (Fig. 5F-5H). In the 4 probands with different KEAP1 mutations, TG expression in the nodule was equivalent to that in the nonnodular parenchyma (Fig. 6A-6D), although varying degrees of nodule-dominant expression of NQO1 (Fig. 6E-6H) and GPX2 (Fig. 6I-6L) were detected.

Figure 4.

Scatter plots of RNA expression levels. Each gene expression is distributed at the intersection of the thyroid nodular lesion (vertical axis) and normal parenchyma (horizontal axis) of the patient with a Kelch-like ECH-associated protein 1 (KEAP1) mutation (L136P). The expression of GPX2, NQO1, and TG is indicated in the blue circle.

Figure 5.

Immunohistochemical detection of NRF2 target genes in the thyroid. (A-D) The upper panels represent a patient with a Kelch-like ECH-associated protein 1 (KEAP1) mutation (L136P), and (E-H) the lower panels represent a control patient harboring no KEAP1 mutation. Histopathological feature of nonnodular parenchyma (*) and thyroid nodules (#) in A and E, hematoxylin-eosin staining; B and F, TG; C and G, NQO1; and D and H, GPX2 expression. The data are representative of 2 experiments with similar results.

Figure 6.

Kelch-like ECH-associated protein 1 (KEAP1) mutations and nuclear factor erythroid-2 related–factor 2 (NRF2)-targeted gene expression. (A-D) TG; (E-H) NQO1; and I to L, GPX2 expression in the thyroid of 4 probands with different KEAP1 mutations. *Nonnodular parenchyma; ^#Thyroid nodules.

Discussion

To date, only 2 families with KEAP1 germline mutations have been reported (3, 8); therefore, linking the phenotype of affected patients to genetic abnormalities is difficult. In this study, we present 5 families with KEAP1 mutations, including 4 newly identified mutations. Their common histopathological features showed multinodular goiters located in the entire thyroid gland with few degenerative lesions, slow proliferation, and elevated expression of NRF2-targeted genes, specifically in the thyroid nodules.

Familial multinodular goiter caused by germline mutations often presents with coexisting tumors in multiple organs. For example, DICER1 syndrome is accompanied by pleuropulmonary blastoma, ovarian Sertoli-Leydig cell tumor, or cystic nephroma (15), whereas DGCR8 syndrome is complicated by schwannoma (4). While 2 family members with different KEAP1 mutations had anaplastic thyroid and esophageal cancers (Q86* I-4 and L136P III-1), respectively, no frequent complications were detected in this study, suggesting a predisposition to thyroid-specific phenotypes by KEAP1 germline heterozygous mutations. No other significant abnormalities were present in thyroid tissues, including driver gene mutations, although half the cases (2/4) had greater than 80% variant allele frequency of KEAP1, suggesting that the loss of heterozygosity of KEAP1 may be partially involved in forming the thyroid nodules.

Significantly increased messenger RNA expression of many NRF2-targeted genes concerning antioxidant substances and detoxification in nodular lesions was noticed in a patient harboring a KEAP1 mutation (p.L136P), compared to that in normal parenchyma. Immunohistochemistry confirmed the high protein expression of GPX2 and NQO1, representative NRF2-targeted genes, specifically in the thyroid nodules of various patients harboring KEAP1 mutations. In contrast, TG (the precursor of the thyroid hormone) was too highly expressed in normal parenchyma to induce more in the nodule. The Keap1-null cell line derived from thyroid tissue showed a significantly higher expression of all GPX2, NQO1, and TG genes than the wild type (14). We identified nodule-specific upregulation of GPX2 and NQO1, but not TG, in the thyroid gland of patients with heterozygous germline mutations of KEAP1. We consider that NRF2 plays a central role in nodule formation in the thyroid gland with KEAP1 mutations based on the nuclear accumulation of NRF2 by dissociation from mutant KEAP1 (p.R483H) in the cytoplasm (5, 8) and upregulation of its targeted genes in this study. Among the NRF2-targeted genes investigated using the RNA microarray, only prostaglandin-endoperoxide synthase 2 (PTGS2) expression was significantly reduced in the nodule (see Table 4). PTGS2, also called Cox-2, is a representative inflammatory cytokine and has a single consensus ARE sequence between nucleotides −562 and −572 of its promoter (16). Interestingly, heat stress upregulates antioxidant enzyme genes, such as SOD1, but downregulates PTGS2 (17), indicating that although both genes have functional AREs, PTGS2 may develop responses different from antioxidant enzyme genes against various stress factors.

The thyroid gland generates high levels of reactive oxygen species (ROS) that are associated with thyroid hormone production. Minor ROS stress may be necessary for cellular homeostasis, while severe ROS induction leads to the apoptosis of thyroid cells by inducing mitochondrial and endoplasmic reticulum stress (18). In response to ROS stress, the expression of antioxidant substances induced by NRF2 protects against oxidative stress and confers multiple advantages to cell proliferation (7). In patients with KEAP1 germline mutations, excessive NRF2 activity in the thyroid gland may exert a cytoprotective effect even against cells subjected to apoptosis by severe ROS generation, resulting in thyroid hyperplasia with scarce degradation. Of the 5 probands with KEAP1 germline mutations, 1 had Graves disease with excessive hormone production, while the remaining 4 had normal thyroid function, indicating that normal hormone production is sufficient for developing thyroid hyperplasia.

This study had several limitations. First, there was a relatively small number of members with familial multinodular goiter who underwent genetic analysis. Second, thyroid specimens from patients with anaplastic thyroid carcinoma and esophageal cancer for further evaluation were absent. Third, comprehensive gene enrichment and functional annotation analyses using RNA microarray results were not possible because only one sample was available.

Collectively, our findings indicate that a germline heterozygous mutation of KEAP1 is associated with the development of multinodular goiter as one pathogenesis. Genetic analysis of KEAP1 should be considered in cases of familial multinodular thyroid nodules with an autosomal dominant mode of inheritance.

Abbreviations

Abs

antibodies

ARE

antioxidant response element

FC

fold change

FFPE

formalin-fixed paraffin-embedded

GPX2

human glutathione peroxidase 2

HE

hematoxylin-eosin

KEAP1

Kelch-like ECH-associated protein 1

NGS

next-generation sequencing

NQO1

human NAD(P)H quinone oxidoreductase 1

NRF2

nuclear factor erythroid-2–related factor 2

PTGS2

prostaglandin-endoperoxide synthase 2

ROS

reactive oxygen species

TG

human thyroglobulin

TPO

thyroid peroxidase

Funding

This work was supported by the Cosmic Thyroid Research Award 1st prize from the Japan Thyroid Association and Japan Society for the Promotion of Science KAKENHI (grant Nos. 26460653 and 21K07325).

Disclosures

The authors have no conflicts of interest to disclose.

Data Availability

Original data generated and analyzed during this study are included in this published article or in the data repositories listed in “References.”