Differentiation versus dysfunction: thyroid hormone, deiodinases and retinal photoreceptors

Abstract

A growing body of evidence has established that thyroid hormone (triiodothyronine, T3) is a key factor in the differentiation and survival of the light-sensing photoreceptors in the retina. These functions include a critical role in generating the cone photoreceptor diversity that is required for color vision. Here, we review some of these functions of T3 and the critical mechanisms that regulate the T3 signal in the mammalian retina. The provision of T3, the active form of thyroid hormone, is determined by developmentally rising levels of T3 and its precursor T4 (thyroxine) in the circulation and by intrinsic control within the retina itself by deiodinase enzymes that deplete or amplify the available level of T3. Dynamic profiles of inactivating (DIO3) and activating (DIO2) deiodinases suggest that the T3 signal is progressively calibrated throughout early development, maturation and later functional maintenance of the retina. However, the benefits of T3 come at a cost: photoreceptors are susceptible to impairment and cell death when T3 signaling becomes imbalanced. These findings have implications regarding the influence of T3 in retinal diseases.

Keywords: color vision; deiodinase enzyme; photoreceptor; retina; thyroid hormone.

Figure 1

Systemic and tissue-intrinsic regulation of T3 in the retina. (A) T4 and T3 in the circulation rise during development, and in mice, peak around the second postnatal week then stabilize in adulthood (128, 129). An early peak of DIO3 activity protects the immature retina from excessive exposure to T3. DIO2 rises during later retinal differentiation as DIO3 declines and could amplify T3 levels. A predicted net rise in T3 signaling encompasses terminal differentiation and maturation of visual function. (B) The developmental switch of Dio3 and Dio2 expression is detected in RNA-seq datasets of mouse (23) (pools of retinas) and human retina (58, 130) (each point from a single individual). A similar trend is detected in retinal organoids derived from human H9 embryonic stem cells (131). DIO1 RNA is undetectable at any stage. Analysis of datasets with GEO accession numbers: mouse GSE 224863; human fetal GSE 104827; human adult peripheral retinal punches GSE 115828; human retinal organoids GSE 129104.

Figure 2

Scheme indicating sources of T4 and T3 for the retina. (A) In the circulation, T3, the primary active form of thyroid hormone, and its precursor T4 are present with T3 at relatively low concentrations. Within the retina, DIO2 generates T3 by conversion from T4. DIO3 converts both T4 and T3 into the largely inactive metabolites reverse T3 (rT3) and 3,3′-diiodothyronine (T2), respectively. Plasma membrane transporters mediate uptake of T3 and T4 into the retina, requiring transfer across the blood-retina barrier. An outer barrier is formed by the RPE adjacent to the fenestrated vessels in the choroid. An inner barrier is formed by the tight junctional barrier of inner retinal vessels. (B) Histological section of the adult mouse retina. The photoreceptor layer contains cones (large nuclei with dispersed heterochromatin), which align near the outer zone of the layer, and more numerous rods (smaller nuclei with dense heterochromatin). Rods are >30-fold more abundant than cones in mice (132). The photoreceptor layer is non-vascularized. The photoreceptor segments extend to the microvilli of the RPE (i.e. the outer blood-retina barrier). Scale bar 25 μm.

Figure 3

Functions of T3 in retinal development and disease. (A) Cone development progresses earlier in humans than in mice relative to the time of birth. The time scales differ, with mice born at ∼20 days and humans at ∼40 weeks post-conception. Cones are generated during early gestation in humans (52, 57, 58) but later gestation in mice (45). Opsin patterning initiates early in utero in humans (23, 52) but is established postnatally in mice (30, 47). Retinal function matures over several weeks in mice and over a period of months or years in humans (57, 133, 134). (B) T3 regulates the development and maintenance of the retina and could potentially modify the pathogenesis of different retinal diseases. Retinal disorders due to genetic defects in T3 signaling are very rare (see text for discussion). However, evidence suggests that T3 could modify the outcomes of more common retinal disorders arising from other causes.

Figure 4

Deiodinases and cell-to-cell control of T3 availability for cones. (A) Image of Dio3 expression in undifferentiated progenitor cells and diagram (on right) suggesting that DIO3 protects cones from excessive T3 in the immature mouse retina (embryonic day 17 shown). Some T3 is needed for initiation of M opsin expression and opsin patterning, but excessive T3 triggers cone cell death. Retinal progenitor cells generate a variety of retinal cell types and are intermingled with newly formed cones. In a proposed model, Dio3-expressing cells form a protective ‘sink’ that degrades T3, constraining exposure to T3. DIO3 also depletes T4, although this is not shown for simplicity. Scale bar 20 μm. (B) At later stages, DIO2 in Müller glial cells can amplify T3 levels in the juvenile and adult mouse retina. Müller glial extensions can transport solutes, potentially including T3, to cones and other cell types. Other possible routes of transport of T3 (such as transfer of blood-borne T3 through the RPE) are not shown in this simplified scheme. Confocal microscopy images: Lily Ng and Ye Liu, using cre drivers for Dio2 (79) and Dio3 (Y Liu, personal communicaion) with an Ai6 fluorescent reporter (pale blue). Scale bar 25 μm.