Expression of TR Isoforms in Failing Human Heart

We read with great interest the article by d’Amati et al. (1) published in the recent issue of JCEM. They found increases in the expression of TRα 1, α2, and β1 in failing human heart. In contrast, we have observed that the expression of TRα1 mRNA is decreased in the failing human heart (2). Moreover, our RNase protection assay revealed that TRα2 was increased in these samples whereas total TRα products (i.e. α1 + α2) were not altered, suggesting a change in alternative splicing of the TRα gene. Sylvén et al. (3) also reported that the TRα1 isoform was decreased in failing human heart. The discrepancy with the present report is surprising, since all of these studies were carried out in human ventricular samples from end-stage heart failure patients. There are some possible reasons for these differences that we will try to characterize in this letter. In the report from d’Amati et al. (1), multiplex competitive RT-PCR was the method used to determine expression levels of TRs. We used RNase protection assay. Although the authors state in the article that Sylvén et al. (3) used RNase protection, they actually performed solution hybridization assays in which the protected fragments were not separated by gel electrophoresis. The authors suggest that altered thyroid hormone status might change the expression of the loading control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (although the reference is in brown adipose tissue; Ref. 4). We routinely use GAPDH as an internal control for RNase protection assay because we have not observed significant changes in its expression in cultured cardiac myocytes (5). However, even if thyroid hormone positively regulates GAPDH expression in human heart as seen in the brown adipocyte, actual expression levels of TRα1 per myocyte should be much less in our failing heart after normalization to GAPDH (considering the authors’ hypothesis that failing hearts might have lower thyroid hormone content), and our conclusion on TRα1 expression would not be changed. With regard to the multiplex RT-PCR, quantification of PCR amplified fragments with ethidium bromide staining cannot be very precise. We have also performed quantitative RT-PCR (2 , 6), but have counted radioactivity that is incorporated into amplified fragments as a method of quantification. The method for measurement of band intensity in the present report is unclear. In addition, the control samples obtained from one-vessel diseased hearts may not represent true “normals.” As mentioned throughout the article, many of cardiac-specific genes (e.g. myosin heavy chain) are responsive to thyroid hormone. Referencing our previous study (6), they said that failing human hearts exhibits a “hypothyroid” phenotype with respect to these genes, which we evaluated in the same samples that TR mRNAs were examined. Since the present report has not approached this in the same manner, it is not clear if their failing hearts really had a “hypothyroid” phenotype. This would be unusual considering that normal thyroid hormone levels were found in the serum and increased expression of TRs identified in the heart. Furthermore, the Western signal for TRα1 protein is surprising to us. Although the authors did not provide the catalogue number, we assume it as MA1-215. Is this correct? Using this antibody, we have been consistently unable to detect any specific signal for TRα1 protein synthesized with rabbit reticulocyte lysate, although a clear signal for synthesized TRβ1 protein can be identified. The company says that this antibody reacts with both human TRα1 and β1, but their experience on Western analyses for this antibody has been limited with bacterially expressed TR proteins. Was the specificity of the antibody used in the present report confirmed on in vitro synthesized TR proteins (especially in mammalian system), and did this observation extend to more than the two hearts shown? The controls for Western were again obtained from patients with coronary artery disease, not from true “normal” subjects. Moreover, those controls were pooled samples from 11 hearts, and it is problematic to compare them with each individual failing heart. The authors hypothesize that increased TR expression might be a compensation for “local” hypothyroidism. Although we are not clear what is meant by this term, it seems to suggest a mechanism such as a defect in myocyte uptake of thyroid hormone despite normal levels of circulating thyroid hormone. If the increased expression of TR is, in fact, a compensation for cardiac hypothyroidism, it would be critical to know if thyroid responsive gene expression is reversed in hearts with higher TRα1 expression. In this regard, we have seen that there is a positive correlation between TRα1 and α-myosin heavy chain expression (2). Finally, the authors also discuss the possibility that marked increases in TRα1 might result in more unliganded TRs which could then act as dominant negative receptors. Although it is not clear if such a dominant negative effect by TRs occurs in the presence of normal circulating thyroid hormone levels, the measurement of thyroid hormone content as well as T3-binding assay in heart tissue would be an important complement to these studies and a subject worthy of pursuing in future studies.