Type 2 deiodinase polymorphism causes ER stress and hypothyroidism in the brain

Abstract

Levothyroxine (LT4) is a form of thyroid hormone used to treat hypothyroidism. In the brain, T4 is converted to the active form T3 by type 2 deiodinase (D2). Thus, it is intriguing that carriers of the Thr92Ala polymorphism in the D2 gene (DIO2) exhibit clinical improvement when liothyronine (LT3) is added to LT4 therapy. Here, we report that D2 is a cargo protein in ER Golgi intermediary compartment (ERGIC) vesicles, recycling between ER and Golgi. The Thr92-to-Ala substitution (Ala92-D2) caused ER stress and activated the unfolded protein response (UPR). Ala92-D2 accumulated in the trans-Golgi and generated less T3, which was restored by eliminating ER stress with the chemical chaperone 4-phenyl butyric acid (4-PBA). An Ala92-Dio2 polymorphism-carrying mouse exhibited UPR and hypothyroidism in distinct brain areas. The mouse refrained from physical activity, slept more, and required additional time to memorize objects. Enhancing T3 signaling in the brain with LT3 improved cognition, whereas restoring proteostasis with 4-PBA eliminated the Ala92-Dio2 phenotype. In contrast, primary hypothyroidism intensified the Ala92-Dio2 phenotype, with only partial response to LT4 therapy. Disruption of cellular proteostasis and reduced Ala92-D2 activity may explain the failure of LT4 therapy in carriers of Thr92Ala-DIO2.

Keywords: Behavior; Cell stress; Endocrinology; Metabolism; Thyroid disease.

Conflict of interest statement

Conflict of interest: ACB is a consultant for Sentier LLC and Synthonics Inc.

Figures

Figure 1. D2 recycles between Golgi apparatus and ER.

(A–F) Immunofluorescence of Thr92-D2HY stably expressing cells using the indicated antibodies. On the far right is the Pearson’s plot for each immunofluorescence image. The right top number is the Pearson’s coefficient for that specific cell. Pearson’s coefficient was calculated as follows: Thr92-DIO2 × α-GM130 (0.32 ± 0.06); Ala92-DIO2 × α-GM130 (0.36 ± 0.07). (G–L) Same as A–F, except that cells stably express Ala92-D2HY. Arrows point to Golgi AlaD2 staining. Pearson’s coefficient: Thr92-DIO2 × α-p230 (0.33 ± 0.05); Ala92-DIO2 × α-p230 (0.65 ± 0.12; P < 0.01 versus Thr92-DIO2 × α-p230). (M and N) Thr92-D2HY (Thr) or Ala92-D2HY (Ala) pulldown, followed by Western blot analysis with the indicated antibodies. (O) Same as M and N, except that cells transiently express Δ18-D2HY. (P–R) Same as A–C except that cells transiently express Δ18-D2HY. (S) Pearson’s coefficient between the indicated D2 proteins and cis-Golgi marker GM130; Δ18-D2 is Δ18-D2HY (P–R). D2T is Thr92-D2HY (A–C). ΔC-D2T is Δ10C-Thr92-D2HY (T–V). ΔC-D2A is Δ10C-Ala92-D2HY (W–Y). (T–Y) Same as A–C, except that cells transiently express Δ10C-Thr92-D2HY or Δ10C-Ala92-D2HY. Original magnification, APO ×60/1.40 oil objective. Values are shown in box-and-whiskers plot indicating median and quartiles. n = 21/group. Statistical analysis used was Mann-Whitney U test in comparison with D2T. ***P ≤ 0.0001.

Figure 2. Expression of Ala92-D2 causes ER stress, triggers UPR response.

UPR markers in Thr92-D2HY–expressing (Thr) or Ala92-D2HY–expressing (Ala) cells. (A) Western blot of pIRE1α. (B) Quantification of pIRE1α shown in A. (C) sXBP1 mRNA levels. RQ, relative quantification. (D) Western blot of uncleaved (uc) and cleaved (c) ATF6. (E) Quantification of cATF6 shown in D. (F) CHOP mRNA levels. (G) Western blot of BIP. (H) Quantification of BIP shown in G. (I) BIP mRNA levels. (J) Western blot of pPERK. (K) Quantification of pPERK shown in J. (L) Western blot of pEIF2α. (M) Quantification of pEIF2α shown in L. (N) ATF4 mRNA levels. (O) Western blot of ERGIC53. (P) Quantification of ERGIC53 shown in O. (Q) ERGIC53 mRNA levels. (R–W) Immunofluorescence of Thr92-D2HY or Ala92-D2HY stably expressing cells using the indicated antibodies. Original magnification, APO ×60/1.40 oil objective. Values are shown in a box-and-whiskers plot indicating median and quartiles. n = 4–5/group. Statistical analysis used was the Mann-Whitney U test or Kruskall-Wallis test followed by the Dunn’s multiple comparison test. *P ≤ 0.05; **P ≤ 0.01.

Figure 3. ERGIC53 and SCAP play a role in Ala92-D2 trafficking.

(A) Western blot of control HEK-293 and HEK-293-ERGIC–/– cells using the indicated antibodies. (B–D) Immunofluorescence of control HEK-293 cells transiently expressing Ala92-D2HY using the indicated antibodies; on the far right is the Pearson’s plot for each immunofluorescence image. The right top number is the Pearson’s coefficient for that specific cell. (E–G) Immunofluorescence of HEK-293-ERGIC–/– cells transiently expressing Ala92-D2HY; arrows point to Golgi AlaD2 staining. (H) Pearson’s coefficient between the indicated D2 proteins and cis-Golgi marker GM130. (I) FRET in HEK293 cells transiently expressing either D2T-EYFP (Thr) or D2A-EYFP (Ala) and SCAP constructs containing the YFP and CFP fluorophores at the indicated positions; cells were treated with HPβCD to cause cholesterol deprivation, as indicated. (J) Same as I, except that SCAP-ΔWD40 was used. (K–P) Immunofluorescence of Thr92-D2HY- or Ala92-D2HY-expressing cells, which were incubated overnight with 500 μM 4-PBA. (Q) Pearson’s coefficient between the indicated D2 proteins and trans-Golgi marker p230 as shown in K–P. (R–V, X) same as K–P, except the cells were incubated overnight with 10 μg/ml cholesterol. (W) Same as in Q, except that data are from R–V, X. Original magnification, APO ×60/1.40 oil objective. Values are shown in a box-and-whiskers plot indicating median and quartiles. n = 9–69/group. Statistical analysis used was the Mann-Whitney U test or the Kruskall-Wallis test, followed by Dunn’s multiple comparison test. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Figure 4. ER stress in Ala92-Dio2 cerebral cortex and slower rate of T4 to T3 conversion in Ala92-D2 expressing cells.

(A) Western blot of cerebral cortex sonicates of Thr92-Dio2 (Thr) and Ala92-Dio2 (Ala) mice utilizing the indicated antibodies; each lane represents an independent mouse sample. (B) Quantification of uncleaved and cleaved ATF6. (C) In vivo deiodination in intact Thr92-D2HY–expressing (Thr) or Ala92-D2HY–expressing (Ala) cells. (D) Immunoprecipitation followed by Western blot of T and A cells utilizing α-YFP and α-cyclophilin B. (E) In vitro deiodination in T and A cell sonicates. (F) Same as C, except that cells were treated for 24 hours with 500 μM 4-PBA or 10 μg/ml cholesterol. Values are shown in a box-and-whiskers plot indicating median and quartiles or mean ± SEM. n = 3–10/group. Statistical analysis used was the Mann-Whitney U test or the Kruskall-Wallis test, followed by the Dunn’s multiple comparison test. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Figure 5. Impaired new object recognition in Ala92-Dio2 mouse.

NOR memory test displayed as preference index (%) of Thr92-Dio2 (Thr) and Ala92-Dio2 (Ala) mice. (A and B) Intact animals. N, new object; O, old object. (C and D) Intact+LT3 animals. (E and F) Intact+4-PBA animals. (G and H) Hypothyroid animals. (I and J) Hypothyroid+LT4 animals. (K–L) Hypothyroid+LT4+LT3 animals. (A, C, E, G, I, and K) 3-hour recall. (B, D, F, H, J, and L) 24-hour recall. Values are shown in a box-and-whiskers plot indicating median and quartiles. Statistical analysis used was Mann-Whitney U test. n = 5–11/group. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Figure 6. Impaired new social recognition in Ala92-Dio2 mouse.

SI memory test displayed as preference index (%) of Thr92-Dio2 (Thr) and Ala92-Dio2 (Ala) mice. (A and B) Intact animals. (C and D) Intact+LT3 animals. (E and F) Intact+4-PBA animals. (G and H) Hypothyroid animals. (I and J) Hypothyroid+LT4 animals. (K–L) Hypothyroid+LT4+LT3 animals. (A, C, E, G, I, and K) 3-hour recall. (B, D, F, H, J, and L) 24-hour recall. Values are shown in a box-and-whiskers plot indicating median and quartiles. Statistical analysis used was Mann-Whitney U test. n = 5–11/group. **P ≤ 0.01; ***P ≤ 0.001.