Thyroid Hormone Binding and Variants of Transport Proteins
More than 60 years ago, it was shown that circulating thyroid hormones are noncovalently bound to plasma proteins.1 Well over 99% of circulating thyroxine (T4) and triiodothyronine (T3) is protein bound in the circulation, with bound and free moieties in constant rapid equilibrium. An understanding of thyroid hormone binding allows the clinician to better appreciate tissue delivery and interconversion of thyroid hormones, especially when these phenomena change as a result of illness or drug therapy, or when treatment is given to alter thyroid status. This knowledge aids the interpretation of (1) atypical thyroid function tests, in particular unusual relationships between serum thyroid-stimulating hormone (TSH) and circulating thyroid hormones that may indicate hereditary or acquired abnormalities of the three major thyroid hormone–binding proteins; (2) aberrant estimates of serum free T4 that may be due to analytic artifact that results from abnormal tracer binding by a variant protein, and (3) the effects of medications and critical illness on protein binding of the thyroid hormones.
The evolution of thyroid hormone binding to plasma proteins can be traced from fish, which show only albumin binding, through birds, in which T4 binds to both albumin and transthyretin (TTR, previously known as prealbumin). Most larger mammals, with the exception of felines, have in addition a low capacity, high-affinity protein, termed thyroxine-binding globulin (TBG) that carries well over half the total circulating T3 and T4. In humans, only 0.02% to 0.03% (about 1 part in 4000) of T4 and 0.2% to 0.3% (1 part in 400) of T3 circulate in the free or unbound state in undiluted normal human serum or plasma at equilibrium at 37°C. Other iodothyronines, whether synthetic analogues or metabolites, also are generally highly protein bound.2 Numerous other hydrophobic ligands that are unrelated to thyroid hormones can also compete for the various plasma protein–binding sites (see below).
In normal human serum, about 75% of the total circulating T4 concentration of 60 to 140 nmol/L or 4 to 11 µg/dL is carried on TBG, with about 10% to 15% attached to TTR and 10% to 15% bound to albumin. The electrophoretic techniques used to make these estimates allow some dissociation of labeled hormone during separation and thus tend to underestimate proportional carriage on lower-affinity sites. A minor fraction (<5%) of circulating T4 and T3 is associated with lipoprotein.3
Binding to plasma proteins is noncovalent and rapidly reversible; it is important to emphasize that the much larger bound moiety of hormone acts, in effect, as a reservoir. Dissociation of bound hormone almost instantaneously replenishes the free hormone concentration, as this fraction is taken up by tissues or diminished by sample dilution in vitro. A rigid distinction between bound and free hormone moieties may be artificial in light of studies that suggest dissociation and reassociation so rapid that the free and bound moieties interchange several million times per day.4
Marked hereditary differences in the protein binding of iodothyronines lead to wide variations in the total T4 concentration in humans (Fig. 22-1), while the free hormone concentration remains within much narrower limits. This finding supports the free hormone hypothesis (see below), which proposes that the minute free fraction of the total circulating T4 and T3 pool is the major determinant of hormone action, clearance, and negative feedback. Even when different species are compared, the concentration of free hormone varies much less than total concentration.5
FIGURE 22-1 Estimated proportional carriage of thyroxine (T4) on the three major plasma binding proteins is shown for euthyroid subjects with normal or variant T4 binding proteins. In the face of normal free T4 and thyroid-stimulating hormone (TSH) concentrations, the concentration of total T4 can vary from about 25 nmol/L in total TBG deficiency to about 1800 nmol/L in the Pro218 variant of familial dysalbuminemic hyperthyroxinemia (FDH) described in several Japanese kindreds.
Concentrations of the binding proteins can vary independent of thyroid status. When the concentration of TBG changes, the total serum T4 and T3 concentrations tend to alter to restore the preexisting concentration of free hormone, as determined by the set point of the feedback relationship with TSH. Hence, in theory, free hormone estimates will give a more accurate reflection of thyroid status than will measurement of total hormone. However, sample storage and dilution and the presence of competitors can alter the relationship between free and total hormone concentration, so that an analytically correct free hormone estimate may not reflect the in vivo situation (see below).
The definition of a number of terms assists the understanding of thyroid hormone binding. Capacity expresses the molar concentration of a specific class of ligand-binding site; if one binding site is present per protein molecule, capacity and protein concentration will be identical in molar terms. When about half the binding sites are empty, the free and hormone concentrations will change to about the same extent (i.e., the free fraction will show little change). However, as the total concentration of ligand approaches the binding protein capacity, the free hormone concentration will rise disproportionately, as occurs in thyrotoxicosis as the total T4 serum concentration approaches the capacity of TBG.6
Proportional carriage, the distribution of total hormone between a number of heterogeneous binding proteins, is influenced by the concentration of binding protein, the affinity of binding, and the free hormone concentration, but it is not directly influenced by the total hormone concentration.7 In a heterogeneous mixture of binding proteins, as in serum, proportional carriage can change as the free hormone concentration alters in response to dilution or hormone loading (see below).
Free fraction describes the percentage of the total hormone that is unbound. When serum that contains a highly bound ligand is progressively diluted, bound hormone will dissociate, so that the free hormone concentration at first is well maintained, with an increase in free fraction.
Occupancy, the proportion of a particular class of binding site that is filled with hormone, is a direct function of the free hormone concentration and is fundamental to the definition of affinity or kD. TBG is normally about one third occupied by T4; TTR is <1% occupied by T4, and albumin shows negligible occupancy (Table 22-1).
Affinity or dissociation constant (kD, mol/L) describes the free hormone concentration at which a particular binding site is half occupied. The association constant (kA, L/mol) is the inverse of kD. (kA is a theoretical concept of the number of liters at which 1 mole of ligand would occupy half the sites on 1 mole of binding protein.) By definition, a binding protein is 50% occupied when the free hormone concentration is the inverse of kA.
Dissociation rate (TΩ, sec) or rate constant (sec-1) defines the rate of unidirectional dissociation or delivery of hormone from a binding site.4 The unidirectional maximum rate of hormone delivery is relevant under non–steady state conditions, as, for example, when free hormone is rapidly removed from the circulation during tissue transit. The dissociation rate, as well as kA and kD, are highly temperature dependent. The free T4 fraction is higher at 37°C than at room temperature by a factor of up to two8 and dissociation of ligand is much faster.4
When represented in terms of the association constant, kTBG, this relationship becomes
It follows that
At half occupancy of the binding site, [TBG · T4] will equal [uTBG] and
For the total hormone concentration to have a regulatory influence on tissue function, it is appropriate for the kD of the dominant binding protein and the physiologic concentration of free hormone to be of the same order, as is the case for the relationship between normal serum free T4 (10 to 25 pM) and the kD of TBG (100 pM). A wide disparity between the kD of the dominant binding protein and the free hormone concentration would require relatively large changes in total hormone concentration to achieve regulatory variation. In effect, TBG stabilizes the tissue distribution of T4, while allowing normal regulatory variation over an approximately twofold range of total hormone concentration. In the absence of TBG, a greater proportional change in total hormone concentration would be necessary to achieve regulatory variation.
According to this hypothesis, it is the free or unbound equilibrium concentration of a hormone that determines biological activity. The validity of this hypothesis, which generally is well sustained for the thyroid hormones, has been analyzed in detail.9,10 Earlier liver perfusion experiments that showed the loosely albumin-bound moiety of the total circulating hormone pool to be virtually as readily available as the free hormone11 have now been refuted.9 No conclusive evidence suggests that any particular class of binding protein facilitates tissue uptake of thyroid hormones. When isolated rat liver was perfused with T4 bound to various normal and variant binding proteins,12 tissue uptake of T4 was proportional to the spontaneous dissociation of T4 from each protein.12
Under some circumstances, especially when capillary transit is slow, or when mixing of layers across the diameter of a vessel is incomplete, tissue uptake of hormone may be limited by dissociation of bound hormone.9 Under these circumstances, the local concentration of free hormone at a particular site may be lower than the equilibrium concentration. The albumin-bound moiety with the fastest rate of unidirectional dissociation (see Table 22-1) will then make a large contribution in replenishing the free concentration.9
As formulated by Mendel,9,10 the unmodified free hormone hypothesis will be valid when tissue uptake of hormone is limited by influx or elimination. When flow or dissociation is the limiting condition, for example, when flow is slow and clearance is rapid in a tissue such as the liver, the free hormone hypothesis still holds, with hormone dissociation as an additional critical variable.
The characteristics of the three normal major iodothyronine-binding proteins, thyroxine-binding globulin, transthyretin and albumin, are summarized in Table 22-1. The numerous structural variants of the three major thyroid hormone–binding proteins that have been described (Table 22-2) were initially recognized from the investigation of euthyroid subjects who showed markedly abnormal levels of total serum T4 or T3, but recently, additional variants have been identified from population screening and genetic studies. The total concentrations of serum T4 and T3 range widely in association with variant binding proteins, but none of the multiple hereditary binding alterations has been shown to confer any advantage or disadvantage, or to disturb thyroid hormone action, unless thyroid disease is associated. Plasma TSH concentrations remain normal, as do free T4 and T3 concentrations, provided the method of estimation is free of artifact. In general, TBG abnormalities tend to affect T4 and T3 similarly; by contrast, some albumin variants lead to selective abnormalities of T4 or T3 binding.
• Effect on diagnostic tests. In euthyroid subjects, abnormal total T4 or T3 concentrations occur in association with normal free hormone concentrations. Particularly with the albumin variants, method-dependent artifacts may compromise measurements of free, occasionally total, T4 or T3.
• Structure-function relationships. Structure-function relationships of specific hormone-binding sites can be studied by using the large amounts of material available in serum. Albumin variants have shown how structural changes can increase the binding affinity for a particular ligand. In contrast, TBG mutants show either normal or diminished binding affinity, often associated with abnormal heat lability.
• Effects of variant binding proteins. The effects of variant binding proteins on T4 and T3 distribution, clearance, and delivery to tissues can test the hypothesis that it is the free hormone concentration, as determined at equilibrium, that determines hormone clearance and action.
• Changes in binding protein configuration. The recent demonstration that TBG, a member of the serine protease inhibitor (SERPIN) class of proteins, can undergo cleavage or change in configuration at local tissue sites has led to important speculation that changes in binding protein configuration may facilitate tissue-specific hormone delivery (see below).
In contrast to the multitude of clearly defined plasma protein–binding variants, the possibility of genetically determined abnormalities of cell membrane transport, cytoplasmic binding, or deiodination of iodothyronines is less clearly defined (see below).
TBG is a single polypeptide chain α globulin, with molecular weight of about 54 kD synthesized as a 415 amino acid protein.13–15 The first 20 amino acid residues of the TBG peptide are hydrophobic in nature and probably represent the signal peptide, which is removed in the endoplasmic reticulum, leaving a mature protein of 395 amino acids in a single chain with a molecular weight of about 44 kD. Multiple glycosylation sites allow an average of 10 terminal sialic acid moieties. The carbohydrate portions of TBG influence protein half-life in blood, stability in vitro, and microheterogeneity on electrophoresis, with only minor effects noted on immunoreactivity or T4 binding. Although TBG is stable in stored serum at 4°C, it gradually loses its binding affinity for T4 at 37°C or above. Differences in the rate of loss of binding affinity at raised temperature have been important in identifying TBG variants. Of particular interest are a variant with markedly increased heat stability (TBG-Chicago)16 and a variant with an extremely heat-labile protein with an abnormally high concentration of denatured TBG and subnormal total T4 (TBG-Gary).17
The amino acid sequence of human TBG shows homology with rat TBG (70%), human cortisol-binding globulin (55%) and members of the serum protease inhibitor family (SERPINS), which includes α-antitrypsin (53% homology) and a 1-antichymotrypsin (58% homology).18 The significance of the structural similarity between human TBG, CBG, and the SERPINS remains unclear, because the hormone-binding proteins do not exhibit antiprotease activity.
Susceptibility to cleavage and change in configuration may modulate hormone delivery from the SERPIN family of binding proteins. The study of Zhou et al.19 confirms that the binding of thyroxine to TBG can alter in response to changes in configuration of the binding protein. Using nonglycosylated recombinant human TBG, investigators reported that thyroxine is carried in a surface pocket and not within the beta barrel of the TBG molecule. With this structural model, conformational changes that result from relocation of a mobile peptide loop within the TBG molecule can favor binding or release of thyroxine. The demonstration of labile interaction between TBG and thyroxine raises the important possibility of modulated or tissue-specific delivery of thyroxine that could vary in response to changes in local pH, temperature, or redox status. The details of how local tissue factors may enhance or limit local hormone release from SERPIN binding proteins remain to be studied (see later section, “Function of Iodothyronine Binding Proteins”).
The normal concentration of human plasma TBG measured by radioimmunoassay is between 10 and 30 mg/L (0.2 to 0.6 µmol/L). TBG is normally 20% to 40% occupied by T4 and <1% occupied by T3. Occupancy may increase markedly in hyperthyroidism owing to increased total T4 and decreased TBG concentrations, leading to a disproportionate rise in free T4 relative to total T4.6 The T4-binding affinity (kD), is about 50 pmol/L at 37°C, consistent with the estimate that TBG is approximately 30% occupied by T4 at the normal free T4 concentration of about 20 pmol/L.20
The single 8 kilobase human TBG gene has been localized to the long arm of the X chromosome at site Xq21-q22.14 Male hemizygotes who express a single mutant allele can show one of three variant phenotypes for T4 binding to TBG: increase, decrease, or absolute deficiency. Despite the presence of two X chromosomes, normal females have TBG levels similar to those of males. Females heterozygous for complete TBG deficiency usually show less than the anticipated 50% reduction in serum TBG, a phenomenon attributed to selective inactivation of the mutant allele.14,15 However, selective inactivation of the normal allele occasionally may result in females showing complete deficiency of TBG.21
Multiple inherited TBG variants, often designated geographically, can result in partial or complete deficiency of immunoreactive TBG in serum. Of at least 24 known X-linked TBG mutants, 15 may cause complete TBG deficiency, and eight other variants are associated with subnormal concentrations of immunoreactive serum TBG, often with reduced affinity for T4.13–15,22 The structural basis of numerous variants is summarized in Fig. 22-2. Several variants of TBG show decreased heat lability in vitro, which generally correlates with accelerated clearance in vivo. TBG deficiency (complete or partial) results from missense or nonsense mutations in the coding exons or in donor or acceptor splice sites.14,15 Thus, single nucleotide deletions or substitutions can lead to a frame shift with premature termination of translation, resulting in a truncated protein that is retained and degraded intracellularly.13–15 In some reports no mutations could be demonstrated in the TBG gene.23
FIGURE 22-2 Summary of reported thyroxine-binding globulin (TBG) deficiency mutations. The numbered panel designates the five exons of TBG. Mutations above that panel refer to partial deficiency of TBG, and those below refer to total TBG deficiency. Polymorphisms are designated by an asterisk. Below the figure, details of each mutation are given, together with geographic designation and literature reference, as in Mannavola et al.14 (From Mannavola et al., 2006.14)
In general, the various methods for serum free T4 estimation, as well as binding corrections based on T3–uptake measurements, give a useful semi-quantitative correction for TBG abnormalities, whether hereditary or acquired.
In total TBG deficiency, total T4 is about 25 nmol/L (see Fig. 22-1), associated with normal free T4 and TSH. By contrast, in hemizygous TBG excess, the total T4 concentration is typically over 200 nmol/L, of which over 80% is carried on TBG (see Fig. 22-1).
From newborn screening studies, the prevalence of complete TBG deficiency in males is about 1 : 5000, with 1 : 15,000 showing complete deficiency,15 but marked ethnic differences are noted in the frequency of hereditary TBG deficiency, with complete deficiency being highest in the Japanese.14 Diminished TBG binding of T4 is especially prevalent in Australian aborigines, up to 30% of whom have subnormal serum concentrations of total T4, associated with subnormal serum concentrations of an abnormally heat-labile TBG that shows subnormal affinity for T4.24 Owing to a very high gene frequency in this population, the pattern of inheritance was initially thought to be autosomal dominant.25 Abnormal heat lability at 37°C was found in both male and female subjects from affected families, but the pattern of intermediate heat lability was found exclusively in female subjects,26 demonstrating that inheritance must be X-linked (Fig. 22-3). Hereditary TBG excess, probably due to gene duplication,27 appears to have a prevalence of about 1 : 25,000 in newborn males.15 The binding of T4 to TBG in inherited X-linked TBG excess is indistinguishable from the common type of TBG. In contrast to the albumin variants, no known TBG mutant shows increased T4-binding affinity.
FIGURE 22-3 Heat stability of thyroxine-binding globulin (TBG) at 56°C in sera from Australian aborigines. Both male and female subjects showed either normal (upper line) or markedly reduced stability (lower line). No male subject showed the intermediate affinity (middle line) that demonstrates the heterozygous state, thereby confirming X-linked inheritance. (From Refetoff S, Murata Y: X-chromosome-linked inheritance of the variant thyroxine-binding globulin in Australian Aborigines, J Clin Endocrinol Metab 1985;60:356-360.)
Human serum albumin, a highly conserved 66 kD nonglycoprotein,28 has a molar plasma concentration of approximately 600 µmol/L, corresponding to about 40 g/L. As well as being the principal carrier of numerous hydrophobic compounds in serum, albumin binds T4 in its region 2, with an affinity about four orders of magnitude less than that of normal TBG. Albumin normally carries 10% to 15% of circulating T4, but the proportion of albumin occupied by T4 is less than 0.002%.
Hyperthyroxinemia can result from variant albumins with increased affinity for T4 or T3, the total albumin concentration being normal (Table 22-3). In familial dysalbuminemic hyperthyroxinemia (FDH), the total T4 concentration in affected individuals is about 200 nM,29,30 with over 50% of T4 carried on the variant albumin (see Fig. 22-1). FDH appears to be the most common hereditary T4-binding abnormality, with a prevalence as high as 1 : 1000 in some Latin American populations.31 As with other variants that show enhanced binding affinity or capacity, the increased concentration of total circulating T4 appears to be an appropriate response to maintain a normal free T4 concentration in feedback relationships with TSH.30