How changes in affinity of corticosteroid-binding globulin modulate free cortisol concentration.

CONTEXT: Recent studies of corticosteroid-binding globulin (CBG) indicate that it does not merely transport cortisol passively but also actively regulates its release in the circulation. We show how CBG binding affinity can vary to give changes in free cortisol concentration in a physiologically relevant range.
OBJECTIVE: The objective was to determine how the binding affinity of plasma CBG is affected by glycosylation, changes in body temperature, and the conformational change induced by proteases at sites of inflammation.
DESIGN: Binding assays were performed over a range of temperatures with plasma and recombinant CBG to determine the contribution of glycosylation. The role of conformational change was assessed by measuring binding affinities of plasma CBG before and after reactive loop cleavage by neutrophil elastase. MAIN OUTCOME MEASURES: Determination of binding constants allows calculation of clinically relevant changes in CBG saturation and free cortisol concentrations.
RESULTS: On reactive loop cleavage at inflammation sites, CBG can continue to act as a buffered source of cortisol, although with a much reduced affinity, to give a potential quadrupling of free cortisol. Predicted increases in systemic free cortisol resulting from elevated body temperatures, previously reported based on affinity measurements using nonglycosylated recombinant CBG, were shown here to be considerably increased using glycosylated plasma CBG, with a doubling for every 2°C rise in body temperature.
CONCLUSIONS: The ability of CBG to modulate free cortisol levels in blood must be considered in the understanding and management of disease processes, as illustrated here with predictable changes in inflammation and fever.

Corticosteroid-binding globulin (CBG) is a blood plasma protein 50 to 60 kDa in size. In many vertebrate species, it acts as the major transport protein for glucocorticoid hormones (1), which are poorly soluble in aqueous environments and rely on binding proteins to remain in solution in the blood. CBG plays a vital role in homeostasis and metabolism by functioning as a circulating reservoir, binding up to 90% of all the cortisol in the circulation, leaving only about 4% (11–43 nM) free in solution to diffuse into cells to act on intracellular glucocorticoid receptors (2–5). This concentration of the active free cortisol is, however, dependent primarily on the binding affinity of CBG, which, as we document here, varies considerably with glycosylation state, conformational changes, and physiological changes in body temperature.

CBG has evolved as a member of the serpin family of serine protease inhibitors that has lost its inhibitory activity but still retains the typical serpin structural framework, as well as the dramatic conformational change that characterizes this family (6–8). CBG binds corticosteroids in a surface pocket underlying the main β-sheet that is structurally equivalent to the thyroxine site in thyroxine-binding globulin (9). In each hormone carrier, the β-sheet expands as the molecule undergoes the stressed-to-relaxed (S-to-R) serpin conformational change induced by insertion of the reactive loop into the body of the molecule (Figure 1, A and B). This is accompanied by a change in binding affinity, with a high affinity in the loop-exposed S-form and a low affinity in the loop-inserted R-form (8, 10, 11). The extreme conformational change occurs physiologically, with full insertion of the reactive center loop, when it is cleaved by proteases at loci of inflammation, resulting in what was proposed to be a complete discharge of the cortisol into the inflamed tissues (10, 12).

Figure 1.

Conformational change brought about by cleavage of the reactive center loop causes the release of cortisol at sites of inflammation. Structures of CBG with cortisol (space-filling spheres) bound to a surface pocket. A. S-state CBG (Rat, PDB 2V95) binds cortisol with high affinity. It has a 5-stranded β-sheet A (red) and an intact reactive center loop (yellow). Part of the loop is not resolved due to its flexibility (8). B, R-state CBG (Human, chimera, PDB 2VDY) is cleaved in the reactive center loop (yellow), which then inserts into β-sheet A to form a novel fourth strand. Helix D is partially unwound by 2 turns (green). This conformer of CBG has a reduced cortisol-binding affinity (9). C, Glycosylation sites on CBG. Human CBG has 6 glycosylation sites: Asn 9, Asn 74, Asn 154, Asn 238, Asn 308, and Asn 347, 4 of which are shown here in blue. Asn 238 is located closest to the steroid-binding pocket (dark gray) and lies only about 8 Å away. A typical biantennary N-linked sugar comprising 12 monosaccharides measures about 30 Å × 10 Å × 10 Å (36). The presence of a large N-linked glycan at this position would readily influence binding affinity either by contributing direct additional binding contacts or by favoring the conformations of binding residues in their bound state.

This belief in a defined on-off mechanism, with a full release of hormone on cleavage of the reactive loop, has been challenged by recent crystallographic studies of both thyroxine-binding globulin and CBG (9, 11, 13). These x-ray structures showed how the movement of the reactive loop allows a modulated and reversible release of hormone from the intact circulating carrier, supporting the hypothesis that rather than there being a switch between 2 defined on-off conformations, the variation in affinity results from equilibrated changes in the plasticity of the binding-pocket (9, 11). This concept of a modulated release mechanism was further supported by the crystal structures that confirmed the ability of R-form CBG to bind cortisol. Even though the reactive loop was fully and irreversibly inserted, the binding pocket containing the cortisol retained a configuration very similar to that of the S-form. The significance of this finding was challenged by the suggestion that the observed binding of cortisol to CBG in the R-form is an artifact of crystallization (14), but is supported by a recent structure showing binding of R-form CBG to progesterone (13).

There is also conflicting evidence about the effect of glycosylation on CBG function (15–17). CBG is known to be heavily glycosylated (16–20). The human CBG sequence contains 6 consensus glycosylation sites (Asn-Xaa-Ser/Thr), which are seen in the x-ray structure (Figure 1C) to be distributed around the protein, with one (Asn 238) lying within 8 Å of the steroid-binding site (11). The glycosylation profile of CBG may depend on various factors, including stage of fetal development (21), pregnancy (22), and health status (23, 24), but there is still little consensus as to the possible role of oligosaccharide groups in the function and activity of CBG.

In this study, we have sought to resolve whether CBG can indeed physiologically bind corticosteroids after cleavage of the reactive loop by determining the binding affinity of cortisol to plasma-derived CBG in the R-form. We have also demonstrated that, contrary to a previous report (15), the binding affinity of CBG is affected by its glycosylation, and we determine here the remarkable changes in affinity that occur in plasma (glycosylated) CBG with variations in physiological temperatures.

Materials and Methods

Recombinant human CBG

Recombinant human CBG with the wild-type sequence was expressed in the BL21star (DE3) strain of Escherichia coli using the pSUMO3 expression system with an N terminus of S11NHHRGLA and purified from the bacterial cell lysate using fast protein liquid chromatography, adapted from a published method (11). The purified protein was stored at a concentration of 1 mg/mL in 10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA at −20°C until used.

Plasma-derived human CBG

Human CBG isolated from a large pool of healthy adult plasma by affinity chromatography was obtained from Affiland (Liège, Belgium). The purity of the sample was confirmed using SDS-PAGE and matrix-assisted laser desorption/ionization mass spectroscopy. Endogenous cortisol was stripped off using Amberlite XAD-2 resin (Sigma Aldrich, Poole, Dorset, United Kingdom). After this treatment, the protein was stored at a concentration of 1 mg/mL in 10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA at −20°C until used.

Preparing CBG conformer with cleaved reactive center loop

CBG is normally expressed either naturally or recombinantly in the S-state, with an intact reactive center loop. To generate the loop-cleaved R-state conformer, we incubated S-state CBG with human neutrophil elastase at room temperature for 12 hours. The elastase was removed by anion exchange chromatography, and the sample was subjected to SDS-PAGE to check for complete cleavage. Reactive loop cleavage is marked by a characteristic change in molecular weight following the release on unfolding of a peptide 4 kDa in size (8, 25). R-state CBG was stored at a concentration of 1 mg/mL in 10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA at −20°C until used.

Fluorimetric determination of dissociation constants

The x-ray structures of the rat and human forms of CBG show that it binds cortisol with a one-to-one stoichiometry. Therefore, the chemical equation for the binding reaction at equilibrium is given by: and the dissociation constant, which is a measure of binding affinity, is calculated by: where [E]F is the concentration of free protein, [L]F is the concentration of free ligand, and [E · L] is the concentration of the protein-ligand complex.

Binding studies on CBG-cortisol interactions were carried out using fluorescence quenching. The experiments were performed on a LS55 120V fluorescence spectrometer (PerkinElmer, Waltham, Massachusetts), and the data were read and recorded using the company's FL WinLab software. The excitation wavelength was set at 280 nm and emission was read at 350 nm. The temperature of the cell was tightly controlled, and both the buffer and the cuvette were prewarmed to the desired temperature before the experiments.

Lyophilized cortisol (Sigma-Aldrich) was dissolved in 80% ethanol to make a 500 μM stock solution, which was diluted with water to make solutions of 20, 40, and 80 μM. Water was used as a negative control. CBG was added to 800 μL 20 mM Tris/150 mM NaCl/1 mM EDTA/0.1% v/v polyethylene glycol 8000 buffer to give a final protein concentration of 200 nM. Aliquots of the cortisol were then titrated into the protein solution, and changes in the fluorescence were measured.

The data were fitted using Prism 5 (GraphPad Software, San Diego, California), according to the following relationship: where ΔF is the fluorescence change, [L]T is the total concentration of ligand added, [E]T is the total protein concentration, and ΔFM is the maximum change in fluorescence signal (9, 25, 26).

The dissociation constants of various forms of CBG were determined in this fashion over a range of temperatures from 25°C to 42°C, the latter part of which, from 37°C to 42°C, is the physiologically relevant range for human beings under healthy and febrile conditions.

Determining free cortisol concentration and saturation of CBG

Percent/fractional saturation of a carrier protein is calculated using the following equation: The free cortisol concentration at equilibrium, [L]F, depends on the concentrations and binding affinities of cortisol-binding proteins in plasma. Apart from CBG, the main known cortisol-binding proteins are albumin and, of somewhat less importance, orosomucoid (α1-acid glycoprotein). It has been reported that, in the presence of functioning CBG in the body, the effect of albumin on CBG-cortisol interaction is minimal (4, 27). Nonetheless, the effect of other cortisol-binding proteins will become more significant if the affinity of CBG is reduced.

The effect of CBG affinity changes can be assessed by using representative values for the concentrations of albumin at 40 g/L or 600 μM (4), with a dissociation constant of 300 μM (27), and of orosomucoid at 1 g/L or 24 μM (28), with a dissociation constant of 62 μM (29). Under conditions where [L]F is much lower than the dissociation constants of the lower affinity binding proteins, [L]F can be determined by solving an equation for total cortisol, [L]T:

The concentration of free cortisol in human blood plasma has been reported to range between 10 and 15 nM in healthy individuals (4, 30). For the purpose of our calculations, we have used the free and total cortisol concentration of normal healthy adults that was reported by Ho and coworkers to be 13 nM and 352 nM, respectively (30). The concentration of CBG has been reported to be in the range of 312-1324 nM (4); for consistency with other values, we have used a value of 1066 nM so that the calculated free cortisol concentration at 37°C would be 13 nM.

Circular dichroism spectroscopy

The protein samples were first dialyzed against 20 mM sodium phosphate, pH 7.5, and diluted to a final concentration of 0.4 mg/mL. Measurements were made on a J810 spectropolarimeter (Jasco UK, Great Dunmow, Essex, United Kingdom) using a quartz cuvette with a 0.1-cm path length. The temperature was controlled using a PTC-348 WI Peltier temperature controller (Jasco UK). Ten single scans were accumulated for each temperature with a scan rate of 20 nm/min. Thermal melt experiments were performed by heating samples of each protein gradually from 20°C to 95°C while using ellipticity at 222 nm to monitor the integrity of the α-helical elements as a measure of the protein's secondary and tertiary structure.

Determination of kinetic constants by stopped flow spectroscopy

The kinetics of association of cortisol with CBG was measured using an SX-20 stopped-flow spectrometer (Applied Photophysics, Surrey, United Kingdom). The protein was excited at a wavelength of 280 nm and the fluorescence signal was measured at 350 nm with a 320-nm filter. The decay of CBG's intrinsic fluorescent signal when it binds to cortisol was tracked over time to give a reaction trace. Up to 10 traces may be averaged to improve signal-to-noise ratio.

Recombinant and plasma-derived CBG were used at a concentration of 250 nM. Both proteins and cortisol were made up in a buffer containing 10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA. In all cases, we used an excess of cortisol, ranging from 5 to 60 times the protein concentration, which enabled us to study the CBG-cortisol interaction as a pseudo-first-order reaction in which the concentration of free cortisol was essentially constant throughout the course of the experiment. We then plotted kobs at various concentrations of cortisol and fit the data to the linear regression: where kobs is the apparent rate constant obtained experimentally, the slope, kon, is the actual association rate constant, [L] is the starting concentration of cortisol, and koff is an imprecise estimate of the dissociation rate constant (31).

Results

The findings are summarized in Table 1, which shows the binding affinities of cortisol to the intact S-forms and the cleaved R-forms of both unglycosylated recombinant and glycosylated plasma CBG over a range of temperatures.

Table 1.

Cortisol Binding Affinities of Various Forms of CBG

Temperature, °C	Glycosylated CBG (KD, nM)		Unglycosylated CBG (KD, nM)
	S State	R State	S State	R State
25	2.3 ± 0.5	51.3 ± 15.0	42.8 ± 11.7	165 ± 14.9
29	11.5 ± 3.3	110 ± 4.8	69.9 ± 7.7	433 ± 43.7
33	23.4 ± 8.0	200 ± 27.6	162 ± 7.4	618 ± 252
37	32.0 ± 1.8	292 ± 27.8	279 ± 18.0	734 ± 48.0
39	57.2 ± 23.7	481 ± 64.9	388 ± 18.8	912 ± 33.5
42	134 ± 15.0	611 ± 45.2	584 ± 44.3	1020 ± 216

Dissociation constants (KD, nM) of glycosylated and unglycosylated CBG in the S and the R states, measured in triplicate by fluorescence quenching titrations over a range of temperatures.

R-form CBG retains binding affinity

At all temperatures, the cleaved R-form of CBG was found to retain a measurable binding affinity for cortisol with R-form plasma CBG at 37°C having a KD of 292 nM. Although the affinity is drastically decreased from what it was before reactive loop cleavage (KD of 32 nM), the KD of plasma-derived R-CBG was still equivalent to that of the intact unglycosylated recombinant form, which demonstrably binds cortisol as seen in the crystal structures of the complex (10, 11).

Glycosylation increases CBG's cortisol-binding affinity

Table 1 also shows that, contrary to an earlier report (15), at all temperatures tested, both the S-states and R-states of glycosylated CBG have higher binding affinities for cortisol than the unglycosylated forms of the protein. The extent to which glycosylation improved cortisol-binding affinity depended on the temperature and conformation of the protein. At 37°C, the difference between the binding affinities of glycosylated and unglycosylated forms of the protein was nearly 9-fold for S-state CBG. Although the difference was not nearly as dramatic for R-state CBG, the glycosylated protein still bound cortisol with 2.5 times higher affinity than the unglycosylated protein.

Glycosylation increases the temperature sensitivity of CBG

Perhaps more remarkable than just increasing its binding affinity for cortisol is the fact that N-linked glycosylation of CBG makes it much more temperature sensitive than was previously known (27). To demonstrate this, we compared the cortisol-binding affinity of CBG at various temperatures to that at 37°C for each form of the protein. We found that for all species of CBG tested, the binding affinity decreases with an increase in temperature. Moreover, the gradients of the temperature response graphs were steeper within the physiological temperature range from 37°C to 42°C than they were from 25°C to 37°C. We also observed that the binding affinity of glycosylated CBG to cortisol was more sensitive to changes in temperature than that of unglycosyated CBG. In particular, the KD of glycosylated S-state CBG increased by over 4 times when the temperature increased from 37°C to 42°C, while the corresponding change for the unglycosylated protein was a mere 2-fold. This was represented graphically by plotting KD/KD,37 against temperature, as shown in Figure 2.

Figure 2.

Cortisol-binding affinity of S-state CBG at different temperatures. Ratios of the dissociation constants, KD, of S-state CBG for cortisol at various temperatures with respect to that at 37°C. Glycosylated CBG (filled circles) is more temperature sensitive than unglycosylated CBG (open circles), especially over the physiological range of temperatures. Data for glycosylated CBG were in agreement with earlier studies by Mickelson et al. (37), represented by the broken black line.

Could the increased temperature sensitivity be the inadvertent consequence of reduced thermal stability? It is easy to envisage how a less thermally stable protein might unfold more readily when the temperature is raised, resulting in a lower apparent affinity for the ligand. To check this, we performed thermal melt experiments on both glycosylated and unglycosylated forms of S-state CBG. Figure 3, A and B, shows that the glycosylated form of the protein was, in fact, more thermally stable than the unglycosylated form, and the melting temperature, Tm, was 2°C higher for the glycosylated protein. Therefore, the steeper temperature response curve of glycosylated CBG was not due to thermal destabilization of the protein.

Figure 3.

Comparing the thermal stabilities of glycosylated and unglycosylated CBG. Loss of secondary structure as a result of heat denaturation was monitored spectroscopically. A, Circular dichroism data of the thermal melts, monitored at 222 nm. Glycosylated CBG (black) was more thermally stable than unglycosylated CBG (gray). B, The first derivative of the melting curves. Peaks give the melting temperature, Tm.

Glycosylation increases the difference in binding affinity following S-to-R transition

Our study showed that plasma-derived and, therefore, glycosylated CBG exhibited a larger change in binding affinity for cortisol than the recombinantly expressed, unglycosylated form of the protein after its reactive center loop had been cleaved by human neutrophil elastase. This effect is especially pronounced at physiological temperature (37°C), where glycosylated CBG exhibited a 9.1-fold change in binding affinity for cortisol on enzymatic cleavage of the reactive center loop, whereas unglycosylated CBG underwent a mere 2.6-fold change in affinity.

Glycosylation reduces the rates of cortisol association and dissociation

Having determined the effects of glycosylation on the cortisol-binding affinity of CBG at equilibrium, we went on to examine how the kinetics of cortisol association and dissociation might have been affected by the presence of surface glycans.

We obtained the apparent binding rate constant, kobs, by fitting the decay of the intrinsic fluorescence to an exponential function using Pro-Data (Applied Photophysics), as demonstrated in Figure 4A. We calculated the association rate constant, kon, from the gradient of the linear regression represented by equation (6), as shown in Figure 4B, and found that the association rate constant of cortisol for unglycosylated S-state CBG was 3.1 × 107 M−1 s−1 at 25°C, almost twice that of glycosylated CBG, which was 1.7 × 107 M−1 s−1. Using the equation KD = koff/kon, it is possible to work out the koff of both species using the equilibrium constants measured at 25°C. It follows that at room temperature, the glycosylated S-state CBG, with a koff of 0.039 s−1, dissociates from cortisol about 30 times more slowly than unglycosylated S-state CBG, which has a koff of 1.3 s−1. A significantly reduced rate of release of bound cortisol by glycosylated CBG more than compensates for the slight decrease in the rate of binding, giving rise to an overall higher affinity.

Figure 4.

Stopped flow kinetics of cortisol binding by CBG. Association rate of cortisol to CBG was measured. A, The quenching curve was fitted to the exponential equation, A = A0e−k, where kobs is the apparent rate constant. Averaged raw data (light gray). Fitted curve (black). Smoothed data (broken line). B, Values of kobs plotted against cortisol concentration. Association rate constant, kon, was calculated from the gradient of the graph. Unglycosylated CBG (open circles). Glycosylated CBG (filled circles).

Discussion

The fraction of total cortisol in the blood that exists in the free bioavailable form is a crucial parameter in maintaining homeostasis (32–35). In healthy individuals, that figure is about 4%, but as we show here, this depends on CBG's cortisol binding affinity, which will vary with changes in conformation, glycosylation, and, notably, temperature.

First, we have established that although the basic paradigm for our understanding of CBG's mechanism of action (8) still holds true, the simple on-off CBG mechanism does not; cleavage and increasing temperature both modulate the buffering of free cortisol. Following cleavage of its reactive loop by elastase, CBG's cortisol affinity is greatly reduced, with the measured KD increasing from 32 nM to 292 nM, but binding is not completely lost. In a healthy adult, at a free cortisol concentration of 13 nM (30), the CBG will be 29% saturated with cortisol under physiological conditions. However, when CBG is cleaved by elastase released by neutrophils, the resulting R-state CBG will release most of its bound cortisol but will still retain a minimal binding saturation of 4.3% within the active circulation. The accompanying release of cortisol would however have profound effects where the circulation is sluggish or encapsulated, as in local areas of inflammation. In an enclosed compartment, in the presence of typical levels of other cortisol-binding proteins, a total cleavage of CBG would result in a 4-fold spike in the local concentration of free cortisol, rising from the normal systemic 13 nM to 54 nM locally. This would be accompanied by a re-equilibration of the percentage saturation of the cleaved GBG to 16%, a level at which it could effectively buffer the raised free cortisol level within the inflammatory compartment. If cleaved CBG did not bind cortisol at all, free cortisol would rise to 103 nM under the same conditions but would be unbuffered because albumin would be only 0.03% saturated and orosomucoid would be only 0.2% saturated. Thus, in effect, there are 2 buffered settings of the CBG modulating mechanism designed to meet the alternate requirements: as intact S-CBG in the general circulation in health and as cleaved R-CBG in inflammation.

Second, we found that glycosylation plays a key role in maintaining CBG function. The recent solution of the recombinant rat and human CBG structures (10, 11, 13) showed that, contrary to a previous report (16), glycosylation is not required for folding. However, it is evident from our data that, despite assuming the correct conformation, unglycosylated CBG does not possess the full activity of the physiologically relevant glycosylated protein. Indeed, at all temperatures tested, glycosylated CBG binds cortisol with much greater affinity than the unglycosylated form, and at body temperature, this translates to a nearly 7-fold difference in saturation in the case of S-state CBG, the prevalent form found in healthy individuals. In fact, across all temperatures measured, intact S-state but unglycosylated CBG has binding affinities akin to those of reactive loop-cleaved R-state glycosylated CBG.

Importantly, glycosylation does not increase cortisol-binding affinity equally for both the S-forms and the R-forms of CBG. This results in an amplification of the change in binding affinity of CBG when it undergoes the S-to-R transition on cleavage of its reactive center loop. At body temperature, the binding affinity of glycosylated CBG decreases by 9-fold, whereas that for unglycosylated CBG decreases by only 2.6 times. This pattern is replicated across the physiological range of temperatures. If the CBG in the body were unglycosylated, the change in free cortisol level in inflamed tissues would be much less dramatic.

Furthermore, the temperature sensitivity of CBG's cortisol binding activity, previously reported with recombinant CBG by Cameron et al in this journal (27), is hugely augmented by the presence of glycosylation. We compared the temperature response of CBG that had been purified from human plasma to that of CBG expressed recombinantly and identical to that used by Cameron and coworkers and found that within the physiological range of temperatures (37°C to 42°C) glycosylated CBG was twice as sensitive to changes in temperature. This means that, even without reactive loop cleavage, the plasma concentration of free cortisol will nearly double from 13 nM to 21 nM when the body temperature rises from 37°C to 39°C. In the case of a high fever of 42°C, the free cortisol concentration in the blood will nearly triple to 37 nM. This observation is consistent with cortisol's function in the human stress response.

In conclusion, the results here, taken together with recent crystallographic studies, emphasize that CBG acts not just as a carrier of corticosteroids but also as a modulator that will allow a varied release to the tissues. The demonstrated changes in binding affinity show how cortisol levels will rise in inflammatory loci. Our findings show a marked difference in binding affinity between glycosylated CBG from healthy adults and unglycosylated CBG. It is therefore conceivable that CBG from individuals in a different physical condition, and hence glycosylation state, would behave differently. More strikingly, it shows how increases in body temperature will greatly increase the release of cortisol to the tissues. Most importantly, this in vitro demonstration of the modulating function of CBG opens the prospect of as yet undiscovered in vivo variations in cortisol release when CBG interacts with other plasma and cellular factors.