|Home | About | Journals | Submit | Contact Us | Français|
The Cu-PTSM (pyruvaldehyde bis(N4-methylthiosemicarbazonato)copper(II)) and Cu-ATSM (diacetyl bis(N4-methylthiosemicarbazonato)copper(II)) radiopharmaceuticals exhibit strong, species-dependent binding to human serum albumin (HSA), while Cu-ETS (ethylglyoxal bis(thiosemicarbazonato)copper(II)) appears to only exhibit non-specific binding to human and animal serum albumins. This study examines the structural basis for HSA binding of Cu-PTSM and Cu-ATSM via competition with drugs having known albumin binding sites. Warfarin, furosemide, ibuprofen, phenylbutazone, benzylpenicillin, and cephmandole were added to HSA solutions at drug:HSA mole ratios from 0 to 8:1, followed by quantification of radiopharmaceutical binding to HSA by ultrafiltration. Warfarin, a site IIA drug, progressively displaced both [64Cu]Cu-PTSM and [64Cu]Cu-ATSM from HSA. At 8:1 warfarin:HSA mole ratios, free [64Cu]Cu-PTSM and [64Cu]Cu-ATSM levels increased 300–500%. This was in contrast to solutions containing ibuprofen, a site IIIA drug; no increase in free [64Cu]Cu-PTSM or [64Cu]Cu-ATSM was observed except at high ibuprofen:HSA ratios, where secondary ibuprofen binding to the IIA site may cause modest radiopharmaceutical displacement. By contrast, and consistent with earlier findings suggesting Cu-ETS exhibits only non-specific associations, [64Cu]Cu-ETS binding to HSA was unaffected by the addition of drugs that bind in either site. We conclude that the species-dependence of Cu-PTSM and Cu-ATSM albumin binding arises from interaction(s) with the IIA site of HSA.
Plasma protein binding can have a profound effect on drug distribution and pharmacokinetics by slowing or preventing passive extravasation into tissues.1,2 Protein binding must be evaluated as part of the drug development process, and the FDA requires that plasma protein binding be reported.3 While simple documentation of the extent of plasma protein binding has utility in assessment of clinical drug delivery phenomenon, knowledge of the structural details of small molecule-albumin interactions are somewhat limited for therapeutic drugs, and very scarce for radiopharmaceuticals.
Human serum albumin (HSA) is highly concentrated in plasma (46 mg/mL),4 and responsible for the majority of small molecule binding in plasma. HSA (Mol. Wt. 66,438) has a flexible three domain structure with two high-affinity, reversible, drug binding sites on subdomains IIA and IIIA.5,6 Hundreds of drugs and endogenous compounds have been formally classified according to binding in either high affinity site.7–10 High-affinity, limited-capacity, drug binding can occur in one or both of these locations, in addition to more widespread nonspecific, high-capacity, small molecule-albumin interactions.
Cu-PTSM and Cu-ETS (Fig. 1) have shown promise as perfusion tracers for positron emission tomography (PET),11–21 and are suitable for labeling with any of the positron-emitting copper isotopes (60Cu, 61Cu, 62Cu, 64Cu). The structurally related Cu-ATSM radiopharmaceutical (Fig. 1) has shown promise for use in PET imaging to assess tissue hypoxia.22–26 Cu-PTSM and Cu-ATSM extensively bind HSA (approx. 95% bound in 40 mg/mL HSA solution), while Cu-ETS binding to HSA appears limited to non-specific interactions (approx. 60% bound in 40 mg/mL HSA solution).27,28 The extensive, and species-specific, binding of Cu-PTSM to HSA is problematic in translation of animal results to humans, as it seemingly limits radiotracer diffusion into tissues exhibiting high rates of perfusion13,16–18.
To probe the structural basis for strong HSA binding of Cu-PTSM and Cu-ATSM, the present study examined competitive displacement of [64Cu]Cu-PTSM, [64Cu]Cu-ATSM , and [64Cu]Cu-ETS from HSA by known site-selective albumin-binding drugs. Furosemide, warfarin, phenylbutazone, and benzylpenicillin (penicillin G) have diverse chemical structures (Fig. 2), but commonly bind in the IIA site of HSA with affinity constants ranging from 1.2 × 103 M−1 (benzylpenicillin) to 7.0 × 105 M−1 (phenylbutazone) (Fig. 2).10 Ibuprofen has a similar affinity for the IIIA site, (Ka = 2.7 × 106 M−1) (Fig. 2).10 The other competing agents examined include L-tryptophan, which has low affinity (Ka = 1 × 104 M−1) for the IIIA site but can competitively displace fluorescent probes from that location,29–31 and the beta lactam antibiotic, cephmandole, which has been shown to disrupt bilirubin binding to HSA.32,33 Finally, since one low affinity fatty acid binding site directly overlaps the HSA IIA drug binding site,34,35 competitive binding assays were performed to probe [64Cu]Cu-PTSM binding to HSA in the presence of four saturated and unsaturated fatty acids with chain lengths varying from 10 to 18 carbons.
Copper-64 was obtained as no-carrier-added 64Cu2+ in dilute HCl from the Radionuclide Resource for Cancer Applications at Washington University via Isotrace, Inc. (St. Louis, MO). The [64Cu]Cu-PTSM, [64Cu]Cu-ATSM, and [64Cu]Cu-ETS radiopharmaceuticals were prepared as described previously, and purified to remove traces of ionic 64Cu using C18 SepPak Light (Waters, Milford, MA) solid-phase extraction cartridges.27 Following recovery in EtOH, the radiochemical purity of each chelate exceeded 99% (silica gel TLC developed with ethanol). Lyophilized human and canine serum albumins were purchased from Sigma Chemical Company (St. Louis, MO), and reconstituted in 0.9% normal saline at 0.66 mM (44 mg/mL). The canine serum albumin was a fraction V powder, and the HSA was essentially globulin-free, and at least 99.5% fatty acid free. Rat serum was purchased from Sigma and stored frozen until use. Ibuprofen, cephmandole, and benzylpenicillin were also purchased from Sigma. Ibuprofen and cephmandole were sodium salts, with cephmandole having a vendor reported potency of 921 µg per mg. The lot of benzylpenicillin (potassium salt) purchased had a potency of 1596 units per mg reported by the vendor. Phenylbutazone was obtained as Phenylbute®, an injectible 20% solution for equine analgesia from Phoenix Pharmaceuticals (St. Joseph, MO), containing 10.45 mg/mL benzyl alcohol as a preservative. Preservative-free furosemide was obtained as a 5% solution for general veterinary administration. Phenylbutazone and furosemide solutions were stored at 2–8°C until use. Racemic warfarin sodium was purchased from Spectrum Chemical Corporation (Gardena, CA) as a crystalline solid.
Stock solutions of each drug were first prepared by dissolving the drug or drug salts in 0.9% normal saline solution at 53 mM, with the exception of the cephmandole stock solution which was prepared at 26.5 mM due to limited aqueous solubility. Furosemide and phenylbutazone solutions were similarly diluted with saline to 53 mM. The benzylpenicillin and cephmandole stock solutions were adjusted for potency, so the final drug concentrations as tested were 53 mM. Aliquots of the 53 mM stock solutions were serially diluted to produce additional solutions at 26.5, 13.3, 6.7, and 3.3 mM solutions of each drug at pH 7. Then, a 100 µL aliquot of each drug solution was added to 1.0 mL of each serum albumin solution, diluting the albumin solutions as tested to 40 mg/mL (at corresponding drug:HSA mole ratios of 8:1, 4:1, 2:1, 1:1, and 0.5:1). 100 µL of saline was also added to a 1.0 mL aliquot of each albumin solution as a drug-free control. Then, 2–8 µCi (0.074 to 0.30 MBq) of [64Cu]Cu-PTSM, [64Cu]Cu-ATSM, or [64Cu]Cu-ETS was added in a volume of 2–10 µL to each drug-serum albumin solution and vortex mixed. In a typical preparation, 300 µL aliquots of each saline or protein solution were added to the ultrafiltration devices in triplicate. The Amicon Centrifree (Millipore, Bedford, MA) ultrafiltration devices with 30,000-DaNMWL methylcellulose micropartition membranes were centrifuged immediately (within 5-minutes of mixing), as reported previously,27 and the quantity of 64Cu in measured volumes of the albumin and ultrafiltrate solutions determined using an automatic gamma counter. Unbound radiotracer levels ("% free" values) were then calculated as described, and corrected for non-specific binding to the ultrafiltration device using the data obtained by analysis of the ultrafiltrate from protein-free saline solutions.27 To determine if the addition of drug altered non-specific radiopharmaceutical binding to the ultrafiltration device, additional controls were prepared by adding a 100 µL of each 53 mM drug solution to 1.0 mL saline with each of the three radiotracers; for each drug-tracer combination; no such effect was detected (i.e., measured %Free values for the [64Cu]Cu-L in protein-free drug solution were identical to the %Free values measured in saline). In these studies, the presence of [64Cu]Cu-L does not appreciably alter the number of available HSA binding sites. Calculated ([64Cu]Cu-L):HSA mole ratios based on the maximum specific activity (171 mCi/µg) on date of receipt ranged from (6 × 10−8):1 on the date of receipt to (1.8 × 10−6):1 in the most delayed experiments.
Additional assays probed [64Cu]Cu-PTSM binding to HSA in the presence of saturated and unsaturated fatty acids with chain lengths varying from 10 to 18 carbons. Capric acid, myristic acid, palmitic acid, and oleic acid (Sigma) were dissolved in 100% ethanol at 49°C as 145 mM and 24 mM solutions. A 0.60 mM (40 mg/mL) HSA solution was prepared using saline, and maintained at 49°C. This temperature was required to keep the fatty acids in solution, but was well below the denaturation temperature of HSA (60°C).36 A 25 µL aliquot of each fatty acid solution in ethanol, corresponding to 1:1 and 6:1 (fatty acid):HSA mole ratios, and 2 µL of [64Cu]Cu-PTSM, were added to 1.0 mL of HSA and vortex mixed prior to being subjected to the ultrafiltration process.
Additional controls were also required to examine the effects of temperature and the addition of ethanol on HSA binding. Solutions of HSA + [64Cu]Cu-PTSM were incubated at 49°C for 30 min. prior to centrifugation, then assayed by ultrafiltration to examine any effects of temperature on binding. The increased incubation temperature did not affect the binding of [64Cu]Cu-PTSM. Unbound [64Cu]Cu-PTSM values in room temperature control solutions were 4.2 ± 0.6% (n = 5) vs. 3.8 ± 0.2% (n = 5) for the test solutions incubated at 49°C. To determine if a 2.5% v/v ethanol solution would degrade the ultrafiltration membrane, 25 µL ethanol was added to 1.0 mL (saline + [64Cu]Cu-PTSM) and subjected to the ultrafiltration process. The addition of ethanol did not appear to compromise the integrity of the ultrafiltration device’s methyl cellulose membrane. Free tracer in the control (saline) group was 67.2 ± 0.7% free (n = 5), vs. the ethanol test group which was 68.0 ± 2.3% free (n = 5). Finally, to examine any effects of ethanol on HSA conformation, 25 µL ethanol was added to 1.0 mL (HSA + [64Cu]Cu-PTSM). The corrected [64Cu]Cu-PTSM %free value was 4.5 ± 0.2% free (n = 5) for the HSA + ethanol solution, compared to 4.0 ± 0.1% free (n = 8) for [64Cu]Cu-PTSM binding to ethanol-free HSA. HSA solutions containing 2.5% v/v ethanol do not appear to affect the integrity of the ultrafiltration membrane or the binding of [64Cu]Cu-PTSM to HSA. HSA solutions in this experiment were maintained at 49°C, and [64Cu]Cu-PTSM binding to HSA at this temperature is comparable to binding to HSA at room temperature.
To determine whether [64Cu]Cu-PTSM could be displaced from HSA by l-tryptophan or other amino acids, competitive binding assays were also performed using tryptophan and a mixed amino acid solution. l-tryptophan and RPMI-1640 50X amino acid medium solution for cell culture (an amino acid cocktail containing Arg (10 mg/mL), Asn (2.5 mg/mL), Asp (1 mg/mL), Cys (3.3 mg/mL), Glu (1 mg/mL), Gly (0.5 mg/mL), His (0.75 mg/mL), Ile (2.5 mg/mL), Leu (2.5 mg/mL), Lys (2 mg/mL), Met (0.75 mg/mL), Phe (0.75 mg/mL), Pro (1 mg/mL), Ser (1.5 mg/mL), Thr (1 mg/mL), Trp (0.25 mg/mL), Tyr (1.44 mg/mL), Val (1 mg/mL)) were purchased from Sigma Chemical Co. Multiple tryptophan solutions (63 mM, 16 mM, 8 mM, and 4 mM) were prepared in saline, corresponding to 8:1, 2:1, 1:1, and 0.5:1 tryptophan:HSA stoichiometry as tested. Then, 0.075 mL aliquots of each tryptophan solution and the amino acid solution were added to 0.9 mL of 0.65 mM (43.3 mg/mL) HSA solution containing [64Cu]Cu-PTSM. These solutions were briefly vortexed and subjected to the ultrafiltration process.
The binding of [64Cu]Cu-PTSM and [64Cu]Cu-ATSM to HSA appear to be selectively disrupted by the addition of warfarin, phenylbutazone, and furosemide, drugs known to selectively bind with high affinity in the IIA site (Table 1 and Table 2). [64Cu]Cu-PTSM and [64Cu]Cu-ATSM are both highly bound in drug-free HSA solutions, (4.0 ± 0.1 %free, and 5.9 ± 0.4 %free, respectively), and the progressive displacement due to increasing warfarin concentrations is nearly identical for these tracers. At 8:1 warfarin:HSA mole ratios, %free values for [64Cu]Cu-PTSM and [64Cu]Cu-ATSM were 20.1 ± 1.5%, and 19.0 ± 1.7%, respectively. The observed displacement of [64Cu]Cu-PTSM by furosemide and phenylbutazone competition at high (8:1) mole ratios is even more dramatic, 32.9 ± 1.1% free, and 28.7 ± 0.8% free, respectively. Furosemide displacement at an 8:1 mole ratio translates into more than a 700% increase in free [64Cu]Cu-PTSM in solution. Further, this level (approximately 33%) of free tracer approaches the value of free [64Cu]Cu-PTSM in canine serum albumin (CSA) solutions (approximately 42%), believed to indicate only non-specific interactions.27
Phenylbutazone has a two-fold greater affinity for HSA than warfarin, (Ka = 7 × 105 M−1 and 3.4 × 105 M−1, respectively), and more effectively displaced [64Cu]Cu-PTSM. The fractions of unbound tracer in 1:1 and 8:1 phenylbutazone solutions were 7.1 ± 0.1% and 28.7 ± 0.8%, respectively, vs. 5.8 ± 0.7% and 20.1 ± 1.5%, in 1:1 and 8:1 warfarin solutions, respectively. Phenylbutazone (log P = 3.2) is more lipophilic than warfarin (log P = 2.7), and both drugs are considerably more lipophilic than Cu-PTSM (log P = 1.9).37,38
[64Cu]Cu-PTSM displacement from HSA binding by site IIA ligands does not appear to be a simple function of the individual drug affinities for HSA. Warfarin has more than a 10-fold greater affinity for site IIA (Ka = 3.5 × 105 M−1) than furosemide (Ka = 2.6 ×104 M−1), yet even at identical 1:1 mole ratios, the increase in free tracer due to furosemide binding (from 4.0% to 8.4% free) is more dramatic than the increase due to warfarin binding (from 4.0% to 5.8% free). The most significant drug displacement at high drug concentrations was observed in solutions containing furosemide (approx. 33% free). Scatchard plot analysis of furosemide binding to HSA indicates there may be at least one additional, lower affinity (Ka = 1 × 10−4 M−1) HSA binding site.39 At high drug concentrations, furosemide binding to HSA at a secondary site may further displace [64Cu]Cu-PTSM, if Cu-PTSM shares this common secondary site. The extremely limited solubility of Cu-PTSM did not permit direct investigation of a secondary binding site in this study by Scatchard plot analysis, which typically requires a 100-fold range of ligand concentrations.
Benzylpenicillin was included in this study as an additional site IIA ligand to probe the affinity of [64Cu]Cu-PTSM for HSA relative to a competing drug with a much lower affinity (Ka = 1.2 × 103 M−1) for the IIA site.40–42 While [64Cu]Cu-PTSM was displaced by increasing amounts of IIA drugs with high (105 or 107 M−1) affinities for the IIA site, no displacement of [64Cu]Cu-PTSM was observed even at an 8:1 benzylpenicillin:HSA mole ratio (3.3 ± 0.1% free Cu-PTSM at the 8:1 benzylpenicillin:HSA mole ratio vs. 4.0 ± 0.1% free Cu-PTSM in the drug-free HSA control solution). The relative affinity of Cu-PTSM for the IIA site of HSA may be significantly stronger than the binding of benzylpenicillin, or benzylpenicillin binding to HSA does not disrupt the concurrent binding of Cu-PTSM.
Ligand binding in the IIIA site does not appear to directly affect the interaction of [64Cu]Cu-PTSM or [64Cu]Cu-ATSM with HSA. The number of site IIIA drugs with solubilities acceptable for use in ultrafiltration assays is extremely limited. Ibuprofen binding in the IIIA site has been confirmed by a crystal structure.43 Ibuprofen has a high affinity for the IIIA site (Ka = 2.7 × 106 M−1), but there was no displacement of [64Cu]Cu-PTSM or [64Cu]Cu-ATSM from HSA observed for ibuprofen:HSA mole ratios up to 4:1 (Table 1 and Table 2). At an 8:1 ibuprofen:HSA mole ratio, however, there were increases in unbound tracer: 10.1 ± 0.6% free and 10.5 ± 1.1% free for [64Cu]Cu-PTSM and [64Cu]Cu-ATSM, respectively. The reported ibuprofen:HSA adduct crystal structure revealed a secondary ibuprofen binding site, overlapping a fatty acid binding site, which bridges the IIA and IIB subdomains.43 Curry further noted anomalous electron density within the IIA site, suggesting ibuprofen may actually bind with very low affinity in the IIA site, but it was not sufficient to allow inclusion in the final refined model. In our test solutions with the highest high ibuprofen concentrations, we believe this secondary ibuprofen binding may be affecting [64Cu]Cu-radiopharmaceutical binding in the IIA site, rather than reflecting competitive binding at the IIIA site.
Cephmandole binding to HSA (Ka = 1.2 × 103 M−1) is disrupted by ibuprofen,32 suggesting this drug also binds in the IIIA site. However, while bilirubin association with HSA is strong (Ka = 9.5 × 107 M−1) and believed to be distinct from the IIA and IIIA drug sites, cephmandole is, surprisingly, also known to disrupt the bilirubin-HSA interaction. The data in Table 1 indicates [64Cu]Cu-PTSM is not displaced by cephmandole at drug:HSA mole ratios up to 4:1. The apparent lack of [64Cu]Cu-PTSM displacement by cephmandole competition seems consistent with the conclusion that Cu-PTSM does not bind the IIIA site.
Tryptophan is the only amino acid known to strongly bind to HSA, interacting with the IIIA site.29 There was no evidence of [64Cu]Cu-PTSM displacement due to tryptophan; even at the highest concentration of tryptophan, 4.8 mM (1.0 mg/mL), [64Cu]Cu-PTSM remained predominantly HSA bound (3.2 ± 0.1% free). Similar results were obtained when trying to disrupt Cu-PTSM binding to HSA using a mixture of 18 amino acids at (amino acid):HSA mole ratios ranging from 0.2:1 (Trp) to 7.4:1 (Arg). The unbound fraction of [64Cu]Cu-PTSM in test solution containing the mixed amino acid solution was 2.6 ± 0.2% free.
The interaction of [64Cu]Cu-ETS with HSA was unaffected by the addition of representative drugs with known affinities for either the IIA or IIIA sites (Table 2). [64Cu]Cu-ETS binding in the drug-free HSA solution was 40.6 ± 1.1% free. At 8:1 drug:HSA mole ratios with warfarin and ibuprofen, Cu-ETS was 44.7 ± 2.7% free, and 49.6 ± 1.2% free, respectively. Thus, binding of Cu-ETS to HSA appears to be limited to non-specific interactions. Unlike Cu-PTSM and Cu-ATSM, there is no evidence that Cu-ETS binds with high affinity to the IIA or IIIA drug sites.
While we have previously seen no influence of fatty acid depletion on the binding of [67Cu]Cu-PTSM in commercially available HSA preparations,28 ESR studies show Cu-PTSM can be displaced from HSA by the addition of supraphysiologic levels of stearic acid and the spin-labeled analogue, 5-doxyl stearic acid.44 The results in Table 3 indicate varying levels of [64Cu]Cu-PTSM displacement by capric acid (CH3(CH2)8COOH), myristic acid (CH3(CH2)12COOH), palmitic acid (CH3(CH2)14COOH), and oleic acid (CH3(CH2)7CH=CH(CH2)7COOH). At 1:1 fatty acid:HSA ratios, there is little effect of fatty acid addition on the HSA binding of [64Cu]Cu-PTSM, while at 6:1 fatty acid:HSA mole ratios the level of free [64Cu]Cu-PTSM was approximately doubled.
[64Cu]Cu-PTSM binding to rat serum in the presence of representative site IIA drugs warfarin, phenylbutazone, and furosemide was also examined (Table 4). Protein binding of [64Cu]Cu-PTSM is comparable between solutions of rat serum albumin (RSA) and rat serum itself.27 The affinity of [64Cu]Cu-PTSM for RSA is relatively high, but significantly less than the affinity for HSA, 14.1 ± 0.3% free vs. 4.3 ± 0.2% free for RSA and HSA, respectively.27 The affinities of warfarin and phenylbutazone for RSA (Ka = 8.3 × 105 M−1, and 4.6 × 105 M−1), respectively, and HSA (Ka = 3.4 × 105 M−1, and 7.0 × 105 M−1), respectively, are nearly identical.45 An affinity constant for furosemide binding to RSA has not been reported, however, one study reported furosemide binding to RSA as high as 99.1%.46 As observed with HSA, warfarin, phenylbutazone, and furosemide were able to competitively displace [64Cu]Cu-PTSM from protein binding sites in rat serum (Table 4). The observed competitive displacement of [64Cu]Cu-PTSM from RSA suggests that RSA contains a site analogous to the IIA site of HSA capable of binding [64Cu]Cu-PTSM.
Further competitive binding assays evaluated [64Cu]Cu-PTSM, [64Cu]Cu-ATSM, and [64Cu]Cu-ETS binding to canine serum albumin (CSA) in the presence of warfarin and ibuprofen. There is no evidence of high affinity CSA binding for any of these bis(thiosemicarbazone)copper(II) complexes, with approximately 40% free reported for each tracer in CSA.27 Thus, these studies were undertaken with the expectation that the added drugs should not alter the levels of free [64Cu]Cu-PTSM, [64Cu]Cu-ATSM, or [64Cu]Cu-ETS in canine serum albumin solutions. Indeed, the free fraction of [64Cu]Cu-PTSM in 40 mg/mL CSA solutions containing 8:1 mole ratios of warfarin were 36.1 ± 2.8% (n = 4) and 32.8 ± 0.7% (n = 4), respectively. Unbound levels of [64Cu]Cu-ATSM and [64Cu]Cu-ETS assayed in CSA solutions containing warfarin were 37.0% and 40.4%, respectively. Free [64Cu]Cu-ATSM and [64Cu]Cu-ETS in CSA solutions with added ibuprofen were 43.8% and 39.5%, respectively. These results further suggest that [64Cu]Cu-PTSM has a unique affinity for HSA, and that any binding of these three radiopharmaceuticals to CSA is limited to non-specific interactions. The affinity of racemic warfarin for CSA has been reported as 7.4 × 104 M−1; while this is approximately 5 times less than the warfarin affinity for HSA (Ka = 3.4 × 105 M−1), warfarin clearly has a significant affinity for CSA.47
These competitive binding assays provide some structural details of copper chelate binding to HSA. The site IIA drug ligands warfarin, phenylbutazone, furosemide, and to a lesser extent, endogenous fatty acids, were shown to displace [64Cu]Cu-PTSM from binding to HSA. Similar displacement due to warfarin binding was observed in HSA solutions containing [64Cu]Cu-ATSM. The addition of warfarin, furosemide, and phenylbutazone also elevated unbound [64Cu]Cu-PTSM in rat serum. Radiotracer liberation was only detected in HSA solutions containing ligands with high affinities for the IIA site, suggesting the high affinity HSA binding of [64Cu]Cu-PTSM and [64Cu]Cu-ATSM arises from interactions in the IIA binding site of HSA. The crystal structures of warfarin-HSA and phenylbutazone-HSA adducts show drug binding causes only subtle, localized changes in protein configuration, which further suggests the observed [64Cu]Cu-PTSM and [64Cu]Cu-ATSM displacement may be due to direct competition for the IIA site, and not allosteric release from a distant binding site.
[64Cu]Cu-PTSM or [64Cu]Cu-ATSM displacement was not observed in HSA solutions containing even high concentrations of the site IIIA ligand l-tryptophan. The IIIA ligand ibuprofen also did not disrupt HSA binding of [64Cu]Cu-PTSM or [64Cu]Cu-ATSM, except at an 8:1 ibuprofen:HSA mole ratio, where we believe ibuprofen's secondary binding to the IIA site likely explains the observed modest radiopharmaceutical displacement. Cephmandole is also likely a site IIIA ligand (based on its displacement due to ibuprofen binding), that has also been shown to displace bilirubin, which may bind HSA at a site distinct from the IIA and IIIA sites. Like ibuprofen and L-tryptophan, cephmandole exhibited no tendency to displace Cu-PTSM from HSA. Benzylpenicillin, with a relatively low affinity (Ka = 1.2 × 103 M−1) for site IIA, similarly failed to displace radiotracer at any concentration. The extent of benzylpenicillin protein binding in human plasma is only 50%, so a relatively low affinity for the IIA site may be insufficient to disrupt [64Cu]Cu-PTSM binding in that location.
[64Cu]Cu-ETS has a drastically different protein binding profile than its structurally-similar counterparts, Cu-PTSM and Cu-ATSM. The addition of representative drugs known to bind in the IIA and IIIA sites did not affect the level of unbound [64Cu]Cu-ETS in either HSA or CSA solutions, suggesting that any measured binding was due to non-specific interactions. Similarly, the non-specific binding of [64Cu]Cu-PTSM or [64Cu]Cu-ATSM to CSA was not affected by the introduction of any drug ligand. The structural differences between these three bis(thiosemicarbazone) chelates of copper(II) are subtle (Fig. 1). However, it appears that additional steric bulk introduced by the ethyl vs. methyl substituent on the diimine chelate backbone is sufficient to preclude strong binding of Cu-ETS in the HSA site that binds Cu-PTSM and Cu-ATSM.
The poor aqueous solubility of Cu-PTSM prevents the direct experimental calculation of an affinity constant for binding to HSA. Attempts to competitively displace [64Cu]Cu-PTSM from the HSA binding site using excess Cu-PTSM were plagued by precipitation of the poorly soluble Cu-PTSM chelate.
The competitive displacement of Cu-PTSM and Cu-ATSM from HSA by common therapeutic drugs is unlikely to have clinical significance with regard to radiopharmaceutical distribution and kinetics, since these competition studies employed the competing drugs at levels that could not be sustained in vivo. Warfarin is a powerful anticoagulant with a narrow therapeutic range (2 – 5 µg/mL), and a relatively long pharmacological half life, averaging 40 hrs.48 Though warfarin binding was shown to displace [64Cu]Cu-PTSM, it could not be administered at the doses needed to preclude HSA interaction with the [60,61,62,64Cu]Cu-PTSM or [60,61,62,64Cu]Cu-ATSM radiopharmaceuticals in vivo. Furosemide is >98% protein-bound in humans and can be acutely administered as an intravenous bolus, but similarly would not saturate even 50% of available HSA molecules at a typical adult single dose (40 mg).48
Fatty acid binding to HSA is physiologically important, as HSA is the primary transport vehicle of insoluble fatty acids in plasma. Typical fatty acid loading of HSA in circulation is 0.1 to 2 moles of fatty acids of varying chain lengths heterogeneously distributed among as many as seven binding sites.34 Protein X-ray crystallography has revealed the relationship of fatty acid binding sites and chain length, and shown that a single low affinity fatty acid binding site overlaps the IIA drug binding site.35
The combined masses of palmitic and oleic acids represent ~50% of the total fatty acid present in adult human serum,49 while myristic acid represents ~2% of the mass fraction, and capric acid is present only in trace amounts. Thus, while oleic and palmitic acids might be the most likely to displace Cu-PTSM in vivo, we believe this is unlikely under natural circumstances; palmitic acid (the most abundant fatty acid in human serum at 0.074 g/L) would typically be present at a 0.4:1 palmitate:HSA stoichiometry, well below the 6:1 stoichiometry required for even moderate Cu-PTSM displacement from HSA in our study (Table 3).
The binding of Cu-PTSM and Cu-ATSM to HSA is selectively disrupted by drugs known to possess high affinities for the IIA binding site of albumin. Drugs that bind the IIIA site did not generally appear to disrupt Cu-PTSM or Cu-ATSM binding to HSA; however, the IIIA drug ibuprofen, which appears to have some secondary affinity for the IIA site, 43 was found to modestly elevate free Cu-PTSM and Cu-ATSM levels at a high ibuprofen:HSA mole ratio. The interaction of the structurally related Cu-ETS radiopharmaceutical with HSA is unaffected by drugs that bind at either the IIA or IIIA site, consistent with previous conclusions that any observed serum protein binding of this tracer is limited to non-specific interactions.27,28 Displacement of Cu-PTSM due to site IIA drugs was also detected in rat serum solutions, while there was no evidence of increased Cu-PTSM, Cu-ATSM, or Cu-ETS in canine serum albumin solutions using drugs which bind in either site.
This work was supported by a research grant from the Purdue Research Foundation, and R01-CA092403. The development of Cu-64 production at Washington University School of Medicine was supported by NCI grant R24 CA86307.