- Email: [email protected]
The peripheral benzodiazepine receptor ligand PK11195 binds with high affinity to the acute phase reactant α1-acid glycoprotein: implications for the use of the ligand as a CNS inflammatory marker1
Nuclear Medicine and Biology 30 (2003) 199 –206
www.elsevier.com/locate/nucmedbio
The peripheral benzodiazepine receptor ligand PK11195 binds with high affinity to the acute phase reactant ␣1-acid glycoprotein: implications for the use of the ligand as a CNS inflammatory marker1 Andrew Lockharta,*, Bill Davisa, Julian C. Matthewsa, Hassan Rahmounea, Guizhu Honga, Antony Geea, David Earnshawb, John Browna a
GlaxoSmithKline, Translational Medicine and Technology, Addenbrooke’s Centre for Clinical Investigation, Addenbrooke’s Hospital, Cambridge, CB2 2GG, UK b GlaxoSmithKline, New Frontiers Science Park, Harlow, CM19 5AW, UK Received 11 May 2002; received in revised form 27 September 2002; accepted 28 September 2002
Abstract The peripheral benzodiazepine receptor ligand PK11195 has been used as an in vivo marker of neuroinflammation in positron emission tomography studies in man. One of the methodological issues surrounding the use of the ligand in these studies is the highly variable kinetic behavior of [11C]PK11195 in plasma. We therefore undertook a study to measure the binding of [3H]PK11195 to whole human blood and found a low level of binding to blood cells but extensive binding to plasma proteins. Binding assays using [3H]PK11195 and purified human plasma proteins demonstrated a strong binding to ␣1-acid glycoprotein (AGP) and a much weaker interaction with albumin. Immunodepletion of AGP from plasma resulted in the loss of plasma [3H]PK11195 binding demonstrating: (i) the specificity of the interaction and (ii) that AGP is the major plasma protein to which PK11195 binds with high affinity. PK11195 was able to displace fluorescein-dexamethasone from AGP with IC50 of ⬍1.2 M, consistent with a high affinity interaction. These findings are important for understanding the behavior of the ligand in positron emission tomography studies for three reasons. Firstly, AGP is an acute phase protein and its levels will vary during infection and pathological inflammatory diseases such as multiple sclerosis. This could significantly alter the free plasma concentrations of the ligand and contribute to its variable kinetic behavior. Secondly, AGP and AGP-bound ligand may contribute to the access of [11C]PK11195 to the brain parenchyma in diseases with blood brain barrier breakdown. Finally, local synthesis of AGP at the site of brain injury may contribute the pattern of [11C]PK11195 binding observed in neuroinflammatory diseases. © 2003 Elsevier Science Inc. All rights reserved. Keywords: PK11195; PBR; ␣1-acid glycoprotein; PET; Orosomucoid
1. Introduction PK11195 (1-[2-chlorophenyl]-N-methyl-N-[1-methylpropyl]-3-isoquinoline carboxamide) is a high affinity ligand of the peripheral benzodiazepine receptor (PBR). Labeled with carbon-11, PK11195 has been used in positron 1 List of abbreviations: AD, Alzheimer’s disease; AGP, ␣1-acid glycoprotein; APO, apolipoprotein A1; BBB, blood brain barrier; HSA, human serum albumin; MS, multiple sclerosis; PBR, peripheral benzodiazepine receptor; PBS, phosphate buffered saline; PET, positron emission tomography; PK11195, 1-[2-chlorophenyl]-N-methyl-N-[1-methyl-propyl]-3-isoquinoline carboxamide * Corresponding author. Tel.: ⫹44 (0)1223 296077; fax: ⫹44 (0)1223 296063. E-mail address: [email protected] (A. Lockhart).
emission tomography studies as an in vivo marker of neuroinflammation and injury in a range of diseases including multiple sclerosis (MS) [1], Alzheimer’s disease (AD) [2], Rasmussen’s encephalitis [3], ischemic stroke [4] and herpes simplex encephalitis [5]. It is believed that the increased binding of [11C]PK11195 seen in these studies is primarily due to the presence of activated microglia, the brain’s resident phagocytic cell, and that as such it may provide an in vivo marker of disease activity in the brain [1]. PBRs are present at low levels in normal central nervous tissue, located primarily on astrocytes and microglia [6]. A number of studies have found that in animal models of brain injury there is time-dependent increase in the binding of PK11195 at the site of damage [7,8,9,10,11,12]. The increase in ligand binding is believed to be due to an upregu-
0969-8051/03/$ – see front matter © 2003 Elsevier Science Inc. All rights reserved. PII: S 0 9 6 9 - 8 0 5 1 ( 0 2 ) 0 0 4 1 0 - 9
200
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
lation in the expression of the PBR, although in a number of studies this has not been substantiated by immunohistochemical data [7,10]. The question of which glial cell type in CNS disease is primarily responsible for the increased ligand binding is not fully resolved and the answer is most likely model dependent. In a rat experimental allergic encephalitis model of multiple sclerosis it has been demonstrated that activated microglia and macrophages are responsible for a significant part of the observed increase in PK11195 binding [1,11]. However Kuhlmann and Guilarte [9] demonstrated that both activated astrocytes and microglia were capable of expressing high levels of PBR following neurotoxic injury induced by trimethyltin. Despite the increasing use of [11C]PK11195 as a PET ligand in studies of human brain disease, there remain a number of methodological issues surrounding the interpretation of results from these studies [1,2]. Firstly, the very low signal from [11C]PK11195 in the normal brain results in a poorly defined pattern of binding. This in turn makes it difficult to define normal reference tissue in diseases that are disseminated throughout the brain, such as AD and MS. Secondly, standard methods of extracting reference data from the plasma compartment are not possible due to the highly variable kinetic behavior of [11C]PK11195 in plasma. This unstable kinetic behavior has been attributed to variable plasma protein binding [1]. We therefore undertook a study to investigate the distribution of the ligand in blood and to identify its binding to plasma proteins. Our results demonstrate that PK11195 binds with high affinity to the acute phase reactant ␣1-acid glycoprotein and the relevance of this interaction with regards to the use of the ligand in PET studies is discussed.
2. Materials and methods 2.1. Collection of blood and the preparation of plasma Whole blood was collected from healthy volunteers into EDTA coated tubes and either used immediately or centrifuged at 10000g for 2 min at 4°C to prepare EDTA-plasma. EDTA-plasma was aliquoted on ice and frozen at – 80°C. Aliquots of plasma were thawed only once and unused material discarded. 2.2. Measurement of the binding of [3H]PK11195 in whole blood 100 l aliquots of freshly collected EDTA treated human blood were incubated with 10 nM [3H]PK11195 (specific activity 85.5 Ci/mol, PerkinElmer Life Sciences) and incubated at 20°C for the times indicated. The samples were then spun at 6000g for 5 mins at room temperature. The separated plasma was removed to a fresh tube and the cell pellet resuspended in an equivalent volume of PBS (typically 50 l). 10 l aliquots of plasma and resuspended cells
were mixed with 200 l of Microscint 40 (Packard Bioscience) and counted for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter (Packard Bioscience). The [3H]PK11195 counts in whole blood were obtained by removing 10 l aliquots from 100 l of whole blood labeled with 10 nM [3H]PK11195. Experiments were performed in triplicate using blood from three healthy volunteers. 2.3. Separation of plasma by gel filtration chromatography A K9/30 column (Amersham Biosciences) was packed with Sephacryl S-200HR resin equilibrated in phosphate buffered saline (PBS). The column was run at a flow rate of 0.4 ml/min and had a void volume of ⬃7.5 mls and an included volume of ⬃18mls. 50 l EDTA-plasma samples were treated with either 25 nM [3H]PK11195 or 25 nM [3H]PK11195 plus 66 M unlabelled PK11195 at 20°C for 10 min. A 30 l aliquot of the treated plasma was applied onto the top of the column which was then developed with PBS and the fractions were collected into 96 well plates. Total protein was determined by mixing a 5 l aliquot from each fraction with 250 l of 1x Advanced Protein Reagent (Cytoskeleton) and the absorbance at 600 nm recorded using a Fusion Universal Microplate Analyzer (Packard Bioscience). The radioactivity in each fraction was determined by mixing 10 l of sample with 100 l of Microscint 40 followed by counting for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter. 2.4. [3H]PK11195 binding assay using spin desalting columns Micro Bio-spin chromatography columns packed with Bio-Gel P6 (Bio Rad) were equilibrated in PBS according to manufacturer’s instructions. Human serum albumin (HSA) (Sigma), ␣1-acid glycoprotein (AGP) (Sigma) and apolipoprotein A1 (APO) (Calbiochem) solutions were prepared in PBS at concentrations of 50 mg/ml, 1 mg/ml and 1 mg/ml, respectively. 30 l aliquots of PBS, EDTA-plasma, HSA, AGP and APO were incubated with incubated with either 10 nM [3H]PK11195 or 10 nM [3H]PK11195 plus 66 M unlabelled PK11195 at 20°C for 10 min. 25 l aliquots of each incubation were loaded onto the top of the spin columns which were then centrifuged at 1000g for 4 min, 20°C. Total protein in the column loads and eluates were determined by removing a 1 l aliquots from each and mixing with 250 l of 1x Advanced Protein Reagent (Cytoskeleton) and recording the absorbance at 600 nm using a Fusion Universal Microplate Analyzer (Packard Bioscience). Recovery of the protein from the columns in all cases was around 95%. The radioactivity in the column loads and eluates was determined by mixing with 100 l of Microscint 40 followed by counting for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter.
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
201
2.5. Immunodepletion of AGP from plasma A goat anti-human AGP antibody (OG20, Europa Bioproducts Ltd) was immobilized and cross-linked to Protein G using the Seize X immunoprecipitation kit (Perbio) according to manufacturer’s instructions. Approximately 1 mg of the OG20 antibody in PBS was cross-linked to 400 l of the Protein G resin slurry. 200 l of EDTA-plasma was incubated with for 2 hr at 4°C with the OG20 resin to immunodeplete the AGP. An untreated plasma sample was also prepared in parallel and kept at 4°C for the same length of time as the OG20 treated plasma. The OG20 treated and untreated plasma samples were immediately assayed for [3H]PK11195 binding using the spin desalt column method followed by gel filtration chromatography as described above. 5 l of plasma was removed from the OG20 treated and the untreated plasma samples, mixed with 35 l of SDS-loading and boiled for 3 min. 2.5 l samples were loaded onto Novex 4-12% NuPAGE Bis-Tris gels (Invitrogen) and run in MES buffer (Invitrogen) according to the manufacturer’s instructions along with Mark12 unstained protein markers (Invitrogen). For total protein staining gels were stained overnight with 30 mls of SYPRO Ruby stain (Molecular Probes), briefly destained in 10% methanol/7% acetic acid before image capture on a Genegenius imaging system (Syngene). For Western analysis the SDS-gels were blotted using the XCell II blot module (Invitrogen) according to the manufacturer’s instructions onto nitrocellulose. Membranes were stained in Ponceau S, briefly washed in PBS plus 0.05% Tween 20 (PBST), blocked in PBST/5% low fat milk for 30 min, probed overnight with an antihuman AGP monoclonal (AG47, Sigma) used at a dilution of 1:10000 in PBST/5% low fat milk, washed extensively in PBST and probed for 1 hr with an anti-mouse HRP conjugate (Sigma) used at a dilution of 1:10000 in PBST/5% low fat milk. After extensive washing in PBST the blot was developed with Supersignal (Perbio) and exposed to Hyperfilm ECL (Amersham Biosciences). Film images were captured using a Genegenius imaging system (Syngene) and the AGP bands quantitated using GeneTools (Syngene).
2.6. Displacement of fluorescein-dexamethasone from AGP by PK11195 5 M AGP was incubated with 10 nM fluoresceindexamethasone (Molecular Probes) and mixed with PK11195 ranging in final concentration from 5 nM to 10 M. The reactions were performed at 20°C in PBST. The fluorescence polarization signal of the fluorescein-dexamethasone was measured in a Screenstation (Molecular Devices) using excitation and emission filters of 490 nm and 535 nm, respectively. The data were analyzed in GraFit (Erithacus Software Ltd) to determine the IC50 value.
Fig. 1. Time course following the distribution of [3H]PK11195 into the cellular elements of blood.
2.7. Quantitation of AGP in the plasma of healthy volunteers The plasma concentration of AGP in was determined by Western blot analysis using the antibody AG47 as described above. Plasma samples were loaded onto Novex 4-12% NuPAGE Bis-Tris gels (Invitrogen) and run in MES buffer (Invitrogen) according to the manufacturer’s instructions along with AGP stocks of known concentration. Blots were developed with Supersignal (Perbio) and images were captured using a Chemigenius imaging system (Syngene). Plasma AGP levels were in the range 1.1-1.4 mg/ml for volunteers used in the study. 2.8. Statistical analysis The student’s t-test was used to determine the significance of differences in the binding of [3H]PK11195 to normal plasma and treated plasma (Figures 3a and 5).
3. Results Experiments studying [3H]PK11195 binding to whole blood indicated that (i) equilibrium was reached within five minutes and (ii) ⬃15% of the ligand was distributed in pelleted material following low speed centrifugation (Figure 1). The pellet contained the cellular elements of blood and a small amount of trapped plasma. This indicated that the binding of PK11195 to blood cells from healthy volunteers is low and that the majority of the compound was distributed in the plasma. To determine whether PK11195 was protein bound in the plasma the labeled ligand was mixed with EDTA-plasma and separated using gel filtration chromatography which resolved the plasma proteins into two distinct peaks (Figure 2). The [3H]PK11195 signal eluted as a single peak on the shoulder of the larger protein peak suggesting an interaction with one of the lower molecular weight proteins in the plasma. The binding of [3H]PK11195 to plasma was displaced using an excess of the unlabelled ligand indicating that a specific binding interaction was occurring. Interest-
202
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
Fig. 2. Separation of [3H]PK11195 labeled plasma by gel filtration chromatography (x total protein measurement, ■ CPM (x102) [3H]PK11195, Œ CPM (x102) [3H]PK11195 plus excess unlabelled PK11195).
ingly, in this latter experiment the unbound [3H]PK11195 did not elute with the included volume of the column. Preliminary data (not shown) suggest that [3H]PK11195 has a propensity to stick to plasticware and thus may have been retained on the column surfaces. Consistent with this observation the recovery of [3H]PK11195 from the first gel filtration experiment was only around 40%. The most common drug-binding plasma proteins are albumin and ␣1-acid glycoprotein (AGP). Analysis of the column fractions by SDS-PAGE (data not shown) indicated that the main protein in the peak, around fraction 37 (Figure 2), was human serum albumin (HSA). This suggested that [3H]PK11195 did not bind to HSA and that it might instead interact with AGP. The binding of [3H]PK11195 to EDTA-plasma, HSA and AGP was measured using a spin column method (Figure 3). The concentrations of the purified plasma proteins used in the assay reflected their approximate concentrations in healthy human plasma (HSA 50 mg/ml, AGP 1 mg/ml). Two sets of controls using buffer alone and apolipoprotein A-1 (APO, 1 mg/ml) indicated that (i) the ligand by itself did not elute from the column and (ii) the assay yielded specific results as no radioactivity eluted with APO (Figure 3a). The assay confirmed that [3H]PK11195 bound to plasma proteins and that there was an interaction with AGP and, to a lesser extent, with HSA. Although the binding of PK11195 to AGP was almost completely reversible, its binding in plasma was only partially displaced by unlabelled ligand (Figure 3a). This may be due to a weak interaction of the ligand with HSA which at the physiological concentrations used in the assay binds [3H]PK11195. When the amount of radioactivity associated with each protein was expressed on a per milligram basis, the binding of [3H]PK11195 to HSA was reduced to the background levels observed with APO (Figure 3b). This demonstrated that there was a much higher affinity interaction between [3H]PK11195 and AGP compared with HSA. The percentages of the total radioactivity added to the assay which bound to APO, plasma, HSA and AGP were 1%, 82%,
Fig. 3. a. Elution of [3H]PK11195 from spin desalting columns with buffer alone, APO (apolipoprotein A1, 1 mg/ml), EDTA-plasma, HSA (human serum albumin (50 mg/ml) and AGP (␣1-acid glycoprotein, 1 mg/ml). b. Elution of [3H]PK11195 from spin desalting columns expressed per mg of protein. Light gray bars [3H]PK11195 alone, dark gray bars [3H]PK11195 plus excess unlabelled PK11195. Values in (a) marked with *** are significantly lower (p⬍0.001) than plasma incubated with [3H]PK11195 alone.
47.5% and 100%, respectively. These data indicate that PK11195 is extensively protein bound in plasma. Gel filtration chromatography of the purified AGP incubated with [3H]PK11195 resulted in the elution of a single protein peak and a single peak of radioactivity (data not shown). The [3H]PK11195 was slightly displaced from the AGP suggesting a slightly later elution from the column. Western blot analysis for AGP in the fractions from the plasma-labeled [3H]PK11195 (Figure 2) also demonstrated a similar displacement of the AGP and [3H]PK11195 peaks (data not shown). This may reflect an interaction of the ligand with the column matrix as noted above. In order to confirm that AGP was the sole high affinity binder of PK11195 in plasma, AGP was immunodepleted from plasma using a goat anti-human AGP antibody (OG20). Western blot analysis of plasma, following treatment with OG20, demonstrated that AGP levels were reduced to ⬃5% of their original level (Figure 4a). The OG20-treated plasma sample demonstrated no significant changes in the overall pattern of plasma proteins with the exception of a band with a molecular weight of ⬃50 kDa that was significantly reduced in the OG20 treated plasma (Figure 4b). The molecular weight of AGP from human plasma has been reported to be between 41-43 kDa [13] and
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
203
Fig. 5. Elution of [3H]PK11195 from spin desalting columns with control plasma and with AGP-plasma. Light gray bars [3H]PK11195 alone, dark gray bars [3H]PK11195 plus excess unlabelled PK11195. Value marked * is significantly lower (p ⬍0.05) than the control plasma incubated with [3H]PK11195 alone.
1.2 M. The theoretical maximal IC50 value that can be measured in the assay is around half the concentration of AGP used in the assay, which is 2.5 M. The finding that the measured IC50 was around half this value is indicative of a IC50 value of less than 1.2 M. Fig. 4. Immunodepletion of ␣1-acid glycoprotein from EDTA-plasma. a. Western blot analysis of untreated plasma (-) and plasma treated with the anti human AGP polyclonal antibody OG20 (⫹). b. SDS-PAGE of samples stained for total protein with SYPRO ruby stain.
therefore this band is unlikely to represent AGP. A more likely candidate is the heavy chain of IgG which has a molecular weight of ⬃50 kDa. The decrease in the intensity of this band in the OG20 treated plasma may indicate that a small proportion of the plasma IgG may have bound to unoccupied antibody binding sites on the Protein G resin used in the immunodepletion. The data nevertheless indicate that OG20 selectivity depletes human plasma of AGP. The binding of [3H]PK11195 to plasma depleted of AGP by OG20 (AGP-plasma) was initially analyzed using the spin column method. This indicated that the level of [3H]PK11195 binding to the AGP-plasma was significantly reduced (p⬍0.05) compared with control plasma (Figure 5). Addition of a large excess of the unlabelled ligand did not significantly reduce the binding of the [3H]PK11195 further, consistent with its weak interaction with HSA. Gel filtration chromatography of the control plasma and AGP-plasma samples indicated broadly similar protein elution patterns consistent with the results of the total protein staining by SDS-PAGE (Figure 4b). While a single peak of radioactivity eluted with the control plasma sample, no corresponding radioactivity eluted with AGP-plasma. These data confirm that AGP is the major protein in human plasma to which PK11195 binds with high affinity. The interaction of PK11195 with AGP was also confirmed by its ability to displace AGP bound fluoresceindexamethasone. Under the conditions used, PK11195 was able to displace fluorescein-dexamethasone with a IC50 of
4. Discussion A number of studies examining the binding of [3H]PK11195 have reported the presence of the PBR in a wide range of blood cells [14,15,16]. Canat et al. [17] reported the highest levels of [3H]PK11195 binding to monocytes, polymorphonuclear cells and lymphocytes, and much lower levels of binding to platelets and erythrocytes (see Table 1). They demonstrated that granulocytes account for ⬃70% of the total number of cellular PK11195 binding sites in whole blood, with the remaining sites being shared more or less equally between the other cell types. Assuming a single PK11195 binding site on each PBR [17], the concentration of binding sites in blood is ⬃10 nM. The reported binding constant for PK11195 on polymorphonuclear cells is around 3 nM [17] and, at the concentrations of [3H]PK11195 used in the blood binding assay, it was expected that a high proportion of ligand would be bound to the cells. Our initial investigations found that only a relatively small amount (⬃15%) of the added [3H]PK11195 was bound to blood cells and that the remainder was distributed in the plasma. This suggested that PK11195 was extensively bound in plasma. Three lines of evidence are consistent with the conclusion that PK11195 binds with high affinity to AGP. Firstly, incubation of purified AGP with [3H]PK11195 followed by a rapid separation from unbound ligand using microspin columns indicated that all of the added ligand eluted with the protein. Secondly, immunodepletion of AGP from plasma abolished the binding of [3H]PK11195. Thirdly, PK11195 was able to displace fluorescein-dexamethasone
204
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
Table 1 Calculation of the total number of PK11195 cellular binding sites in a microlitre of blood. Cell population
Number of PK11195 binding sites per cell*
Average values for cells in normal human blood (cell/ l)#
Number of PK11195 binding sites per l of normal human blood
Lymphocytes Polymorphonuclear cells Monocytes Platelets Erythrocytes
1 ⫻ 105 7.2 ⫻ 105 7.5 ⫻ 105 1.2 ⫻ 103 1.1 ⫻ 102
2.75 ⫻ 103 6 ⫻ 103 5.4 ⫻ 102 3 ⫻ 105 5.4 ⫻ 106 Total number of PK11195 binding sites per l of normal blood
2.75 ⫻ 108 4.32 ⫻ 109 4.05 ⫻ 108 3.60 ⫻ 108 5.94 ⫻ 108 5.95 ⴛ 109
* Values from Canat et al [17]. # Values from Ganong [30].
from AGP with a IC50 of ⬍1.2 M. There are two consequences resulting from this interaction. Firstly, the high plasma concentration of AGP (⬃20 M) results in a smaller than expected proportion of the ligand binding to the blood cells. Secondly, PK11195 is highly (⬎80%) protein bound in plasma. Our results also indicated that PK11195 interacted very weakly with HSA and at the tracer amounts typically used in PET studies this binding is likely to be insignificant due to its higher affinity interaction with AGP. Consistent with our findings Dougherty et al [18] have recently reported the binding of PK11195 to site II on HSA. AGP, also known as orosomucoid, is a glycoprotein of 183 residues with an extremely high carbohydrate content accounting for ⬃45% of its mass [19]. The plasma level of AGP is relatively stable in healthy subjects (at around 1 mg/ml) but can rapidly increase up to three-fold in response to acute inflammation. It is therefore considered to be an important member of the positive acute phase protein family. Increased plasma levels of AGP are associated with inflammatory diseases such as Crohn’s disease [20], rheumatoid arthritis [21] and MS [22]. Although the precise biological function of AGP is unclear, it appears to have both anti-inflammatory effects and a role in immunomodulation [19]. The displacement assay used to estimate the binding affinity of PK11195 for AGP utilizes fluoresein-dexamethasone as a fluorescent probe. The assay is sensitive to a number drugs, such as propanalol [23] and prazosin [24], which are reported to bind to the common binding site for basic drugs on AGP [19]. In the assay propanalol and prazosin are able to displace fluoresein-dexamethasone from AGP with IC50 values of 4.9 and 2.3 M, respectively (data not shown). The data is therefore consistent with PK11195 binding to this common ligand pocket on AGP although the compound may interact with additional sites on the protein. The potential consequences resulting from the interaction of PK11195 with AGP are three-fold: (i) variations in
levels of AGP could significantly alter the free plasma concentrations of the ligand, (ii) AGP and AGP-bound ligand may contribute to the delivery of ligand to the brain parenchyma (i.e., across a break in the blood brain barrier), (iii) local synthesis of AGP at sites of inflammation in the brain tissue could provide additional PK11195 binding sites. A number of studies have found a strong inverse correlation between the plasma AGP concentration and the amount of free drug (for example propanolol) in the plasma of healthy individuals [25,26]. Normal variation in plasma AGP levels between individuals and changes in the levels of plasma AGP in individuals with underlying diseases, or as a consequence of a bacterial or viral infection, may alter the levels of free PK11195 in the plasma. Both these factors may contribute to the reported unstable kinetic behavior of [11C]PK11195 in the plasma compartment [1]. Longitudinal monitoring of plasma AGP levels during PET studies may provide useful information in modeling the distribution of the ligand. In diseases with significant blood brain barrier breakdown, such as MS [27] and to a lesser extent AD [28], the passage of AGP into the brain parenchyma may contribute to the increased binding of [11C]PK11195 in the damaged tissue in a number of ways: firstly, via the direct passage of AGP-bound ligand into the brain tissue and secondly, an increase in the local concentration of AGP may draw the free ligand from the plasma into the brain. Plasma AGP is mainly derived from hepatic synthesis and its expression is induced by glucocorticoids and proinflammatory cytokines such as interleukin 1 and interleukin 6 [19]. Extrahepatic expression of AGP has, however, been detected in a wide range of tissues and cells, including those of the immune system such as lymphocytes, monocytes, alveolar macrophages and granulocytes [reviewed in 19]. Given the potential immunomodulatory role of AGP, its local synthesis by these cells may help limit tissue damage during inflammation and there is some evidence that these extra-hepatic cell types may be regulated by cytokines in the
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
same manner as hepatic cells [29]. It is also possible that cells, such as glia, which are normally resident in the CNS, synthesis AGP. Thus there is the potential in neurological diseases with a significant cellular infiltrate, such as MS, for the local synthesis of AGP which may contribute to the increased focal binding of PK11195. In summary, we have identified that the PBR ligand PK11195 binds with high affinity to the plasma protein and acute phase reactant AGP. This may explain the observed variability in kinetic behavior of [11C]PK11195 in PET studies and that AGP-bound ligand may contribute to the pattern of PK11195 binding observed in neuroinflammatory diseases. These findings provide the basis for further studies investigating the distribution and potential role of AGP in neuroinflammation.
Acknowledgments We are grateful to Dr Murray Brown and Dr Andrew Pope for assistance with the fluorescent displacement assay and to Dr. Ron Leslie for helpful discussions.
References [1] R.B. Banati, J. Newcombe, R.N. Gunn, A. Cagnin, F. Turkheimer, F. Heppner, G. Price, F. Wegner, G. Giovannoni, D.H. Miller, G.D. Perkin, T. Smith, A.K. Hewson, G. Bydder, G.W. Kreutzberg, T. Jones, M.L. Cuzner, R. Myers, The peripheral benzodiazepine binding site in the brain in multiple sclerosis: quantitative in vivo imaging of microglia as a measure of disease activity, Brain 123 (2000) 2321–2337. [2] A. Cagnin, D.J. Brooks, A.M. Kennedy, R.N. Gunn, R. Myers, F.E. Turkheimer, T. Jones, R.B. Banati, In-vivo measurement of activated microglia in dementia, Lancet 358 (2001) 461– 467. [3] R.B. Banati, G.W. Goerres, R. Myers, R.N. Gunn, F.E. Turkheimer, G.W. Kreutzberg, D.J. Brooks, T. Jones, J.S. Duncan, [11C](R)PK11195 positron emission tomography imaging of activated microglia in vivo in Rasmussen’s encephalitis, Neurology 53 (1999) 2199 – 2203. [4] S. Pappata, M. Levasseur, R.N. Gunn, R. Myers, C. Crouzel, A. Syrota, T. Jones, G.W. Kreutzberg, R.B. Banati, Thalamic microglial activation in ischemic stroke detected in vivo by PET and [11C]PK11195, Neurology 55 (2000) 1052–1054. [5] A. Cagnin, R. Myers, R.N. Gunn, A.D. Lawrence, T. Stevens, G.W. Kreutzberg, T. Jones, R.B. Banati, In vivo visualization of activated glia by [11C] (R)-PK11195-PET following herpes encephalitis reveals projected neuronal damage beyond the primary focal lesion, Brain 124 (2001) 2014 –2027. [6] M. Gavish, I. Bachman, R. Shoukrun, Y. Katz, L. Veenman, G. Weisinger, A. Weizman, Enigma of the peripheral benzodiazepine receptor, Pharmacol Rev 51 (1999) 629 – 650. [7] Agnello, L. Carvelli, V. Muzio, P. Villa, B. Bottazzi, N. Polentarutti, T. Mennini, A. Mantovani, P. Ghezzi, Increased peripheral benzodiazepine binding sites and pentraxin 3 expression in the spinal cord during EAE: relation to inflammatory cytokines and modulation by dexamethasone and rolipram, J Neuroimmunol 109 (2000) 105–111. [8] A.C. Kuhlmann, T.R. Guilarte, Regional and temporal expression of the peripheral benzodiazepine receptor in MPTP neurotoxicity, Toxicol Sci 48 (1999) 107–116.
205
[9] A. C. Kuhlmann, T.R. Guilarte, Cellular and subcellular localization of peripheral benzodiazepine receptors after trimethyltin neurotoxicity, J Neurochem 74 (2000) 1694 –1704. [10] E.L. Conway, A.L. Gundlach, J.A. Craven, Temporal changes in glial fibrillary acidic protein messenger RNA and [3H]PK11195 binding in relation to imidazoline-I2-receptor and alpha 2-adrenoceptor binding in the hippocampus following transient global forebrain ischaemia in the rat, Neuroscience 82 (1998) 805– 817. [11] E. Vowinckel, D. Reutens, B. Becher, G. Verge, A. Evans, T. Owens, J.P. Antel, PK11195 binding to the peripheral benzodiazepine receptor as a marker of microglia activation in multiple sclerosis and experimental autoimmune encephalomyelitis, J Neurosci Res 50 (1997) 345–353. [12] R.B. Banati, R. Myers, G.W. Kreutzberg, PK (’peripheral benzodiazepine’)-binding sites in the CNS indicate early and discrete brain lesions: microautoradiographic detection of [3H]PK11195 binding to activated microglia, J Neurocytol 26 (1997) 77– 82. [13] J.M. Kremer, J. Wilting, L.H. Janssen, Drug binding to human alpha1-acid glycoprotein in health and disease, Pharmacol Rev 40 (1988) 1– 47. [14] S. Maeda, T. Miyawaki, T. Nakanishi, M. Takigawa, M. Shimada, Peripheral type benzodiazepine receptor in T lymphocyte rich preparation, Life Sci 631 (1998) 423– 430. [15] P.N. Moynagh, L.A. O’Neill, D.C. Williams, Interleukin-1 and phorbol myristate acetate modulate the peripheral-type benzodiazepine receptor in lymphocytes and glial cells, Biochem Pharmacol 46 (1993) 821– 827. [16] A. Berkovich, C. Ferrarese, G. Cavaletti, H. Alho, C. Marzorati, G. Bianchi, A. Guidotti, E. Costa, Topology of two DBI receptors in human lymphocytes, Life Sci 52 (1993) 1265–1277. [17] X. Canat, P. Carayon, M. Bouaboula, D. Cahard, D. Shire, C. Roque, G. Le Fur, P. Casellas, Distribution profile and properties of peripheral-type benzodiazepine receptors on human hemopoietic cells, Life Sci 52 (1993) 107–118. [18] T.J. Dougherty, A.B. Sumlin, W.R. Greco, K.R. Weishaupt, L.A. Vaughan, R.K. Pandey, The role of the peripheral benzodiazepine receptor in photodynamic activity of certain pyropheophorbide ether photosensitizers: albumin site II as a surrogate marker for activity, Photochem Photobiol 76 (2002) 91–97. [19] T. Fournier, N. Medjoubi, D. Porquet, Alpha-1-acid glycoprotein, Biochim Biophys Acta 1482 (2000) 157–171. [20] O.H. Nielsen, B. Vainer, S.M. Madsen, J.B. Seidelin, N.H. Heegaard, Established and emerging biological activity markers of inflammatory bowel disease, Am J Gastroenterol 95 (2000) 359 –367. [21] T. Nakamura, P.G. Board, K. Matsushita, H. Tanaka, T. Matsuyama, T. Matsuda, Alpha 1-acid glycoprotein expression in human leukocytes: possible correlation between alpha 1-acid glycoprotein and inflammatory cytokines in rheumatoid arthritis, Inflammation 17 (1993) 33– 45. [22] P.C Dowling, S.D. Cook, Disease markers in acute multiple sclerosis, Archives of Neurology 33 (1976) 668 – 677. [23] F. Albani, R. Riva, M. Contin, A. Baruzzi, Stereoselective binding of propranolol enantiomers to human alpha 1-acid glycoprotein and human plasma, Br J Clin Pharmacol 18 (1984) 244 –246. [24] F. Brunner, W.E. Muller, Prazosin binding to human alpha 1-acid glycoprotein (orosomucoid), human serum albumin, and human serum. Further characterization of the ’single drug binding site’ of orosomucoid, J Pharm Pharmacol 37 (1985) 305–309. [25] K.M. Piafsky, O. Borga, I. Odar-Cederlof, C. Johansson, F. Sjoqvist, Increased plasma protein binding of propranolol and chlorpromazine mediated by disease-induced elevations of plasma alpha1 acid glycoprotein, N Engl J Med 299 (1978) 1435–1439. [26] G. Alvan, K. Bergstrom, O. Borga, L. Iselius, N. Pedersen, Family study of genetic and environmental factors determining the protein binding of propranolol, Eur J Clin Pharmacol 5 (1983) 437– 441. [27] N.C. Silver, P.S. Tofts, M.R. Symms, G.J. Barker, A.J. Thompson, D.H. Miller, Quantitative contrast-enhanced magnetic resonance im-
206
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
aging to evaluate blood-brain barrier integrity in multiple sclerosis: a preliminary study, Mult Scler 7 (2001) 75– 82. [28] T.M. Berzin, B.D. Zipser, M.S. Rafii, V. Kuo-Leblanc, G.D. Yancopoulos, D.J. Glass, J.R. Fallon, E.G. Stopa, Agrin and microvascular damage in Alzheimer’s disease, Neurobiol Aging 21 (2000) 349 – 355.
[29] C.G. Gahmberg, L.C. Andersson, Leukocyte surface origin of human alpha 1-acid glycoprotein (orosomucoid), J Exp Med 148 (1978) 507–521. [30] W.F. Ganong, Circulating Body Fluids. in: W.F. Ganong (Ed), Review of Medical Physiology (1999) pp. 493–521) Appleton and Lange: Stamford.
www.elsevier.com/locate/nucmedbio
The peripheral benzodiazepine receptor ligand PK11195 binds with high affinity to the acute phase reactant ␣1-acid glycoprotein: implications for the use of the ligand as a CNS inflammatory marker1 Andrew Lockharta,*, Bill Davisa, Julian C. Matthewsa, Hassan Rahmounea, Guizhu Honga, Antony Geea, David Earnshawb, John Browna a
GlaxoSmithKline, Translational Medicine and Technology, Addenbrooke’s Centre for Clinical Investigation, Addenbrooke’s Hospital, Cambridge, CB2 2GG, UK b GlaxoSmithKline, New Frontiers Science Park, Harlow, CM19 5AW, UK Received 11 May 2002; received in revised form 27 September 2002; accepted 28 September 2002
Abstract The peripheral benzodiazepine receptor ligand PK11195 has been used as an in vivo marker of neuroinflammation in positron emission tomography studies in man. One of the methodological issues surrounding the use of the ligand in these studies is the highly variable kinetic behavior of [11C]PK11195 in plasma. We therefore undertook a study to measure the binding of [3H]PK11195 to whole human blood and found a low level of binding to blood cells but extensive binding to plasma proteins. Binding assays using [3H]PK11195 and purified human plasma proteins demonstrated a strong binding to ␣1-acid glycoprotein (AGP) and a much weaker interaction with albumin. Immunodepletion of AGP from plasma resulted in the loss of plasma [3H]PK11195 binding demonstrating: (i) the specificity of the interaction and (ii) that AGP is the major plasma protein to which PK11195 binds with high affinity. PK11195 was able to displace fluorescein-dexamethasone from AGP with IC50 of ⬍1.2 M, consistent with a high affinity interaction. These findings are important for understanding the behavior of the ligand in positron emission tomography studies for three reasons. Firstly, AGP is an acute phase protein and its levels will vary during infection and pathological inflammatory diseases such as multiple sclerosis. This could significantly alter the free plasma concentrations of the ligand and contribute to its variable kinetic behavior. Secondly, AGP and AGP-bound ligand may contribute to the access of [11C]PK11195 to the brain parenchyma in diseases with blood brain barrier breakdown. Finally, local synthesis of AGP at the site of brain injury may contribute the pattern of [11C]PK11195 binding observed in neuroinflammatory diseases. © 2003 Elsevier Science Inc. All rights reserved. Keywords: PK11195; PBR; ␣1-acid glycoprotein; PET; Orosomucoid
1. Introduction PK11195 (1-[2-chlorophenyl]-N-methyl-N-[1-methylpropyl]-3-isoquinoline carboxamide) is a high affinity ligand of the peripheral benzodiazepine receptor (PBR). Labeled with carbon-11, PK11195 has been used in positron 1 List of abbreviations: AD, Alzheimer’s disease; AGP, ␣1-acid glycoprotein; APO, apolipoprotein A1; BBB, blood brain barrier; HSA, human serum albumin; MS, multiple sclerosis; PBR, peripheral benzodiazepine receptor; PBS, phosphate buffered saline; PET, positron emission tomography; PK11195, 1-[2-chlorophenyl]-N-methyl-N-[1-methyl-propyl]-3-isoquinoline carboxamide * Corresponding author. Tel.: ⫹44 (0)1223 296077; fax: ⫹44 (0)1223 296063. E-mail address: [email protected] (A. Lockhart).
emission tomography studies as an in vivo marker of neuroinflammation and injury in a range of diseases including multiple sclerosis (MS) [1], Alzheimer’s disease (AD) [2], Rasmussen’s encephalitis [3], ischemic stroke [4] and herpes simplex encephalitis [5]. It is believed that the increased binding of [11C]PK11195 seen in these studies is primarily due to the presence of activated microglia, the brain’s resident phagocytic cell, and that as such it may provide an in vivo marker of disease activity in the brain [1]. PBRs are present at low levels in normal central nervous tissue, located primarily on astrocytes and microglia [6]. A number of studies have found that in animal models of brain injury there is time-dependent increase in the binding of PK11195 at the site of damage [7,8,9,10,11,12]. The increase in ligand binding is believed to be due to an upregu-
0969-8051/03/$ – see front matter © 2003 Elsevier Science Inc. All rights reserved. PII: S 0 9 6 9 - 8 0 5 1 ( 0 2 ) 0 0 4 1 0 - 9
200
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
lation in the expression of the PBR, although in a number of studies this has not been substantiated by immunohistochemical data [7,10]. The question of which glial cell type in CNS disease is primarily responsible for the increased ligand binding is not fully resolved and the answer is most likely model dependent. In a rat experimental allergic encephalitis model of multiple sclerosis it has been demonstrated that activated microglia and macrophages are responsible for a significant part of the observed increase in PK11195 binding [1,11]. However Kuhlmann and Guilarte [9] demonstrated that both activated astrocytes and microglia were capable of expressing high levels of PBR following neurotoxic injury induced by trimethyltin. Despite the increasing use of [11C]PK11195 as a PET ligand in studies of human brain disease, there remain a number of methodological issues surrounding the interpretation of results from these studies [1,2]. Firstly, the very low signal from [11C]PK11195 in the normal brain results in a poorly defined pattern of binding. This in turn makes it difficult to define normal reference tissue in diseases that are disseminated throughout the brain, such as AD and MS. Secondly, standard methods of extracting reference data from the plasma compartment are not possible due to the highly variable kinetic behavior of [11C]PK11195 in plasma. This unstable kinetic behavior has been attributed to variable plasma protein binding [1]. We therefore undertook a study to investigate the distribution of the ligand in blood and to identify its binding to plasma proteins. Our results demonstrate that PK11195 binds with high affinity to the acute phase reactant ␣1-acid glycoprotein and the relevance of this interaction with regards to the use of the ligand in PET studies is discussed.
2. Materials and methods 2.1. Collection of blood and the preparation of plasma Whole blood was collected from healthy volunteers into EDTA coated tubes and either used immediately or centrifuged at 10000g for 2 min at 4°C to prepare EDTA-plasma. EDTA-plasma was aliquoted on ice and frozen at – 80°C. Aliquots of plasma were thawed only once and unused material discarded. 2.2. Measurement of the binding of [3H]PK11195 in whole blood 100 l aliquots of freshly collected EDTA treated human blood were incubated with 10 nM [3H]PK11195 (specific activity 85.5 Ci/mol, PerkinElmer Life Sciences) and incubated at 20°C for the times indicated. The samples were then spun at 6000g for 5 mins at room temperature. The separated plasma was removed to a fresh tube and the cell pellet resuspended in an equivalent volume of PBS (typically 50 l). 10 l aliquots of plasma and resuspended cells
were mixed with 200 l of Microscint 40 (Packard Bioscience) and counted for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter (Packard Bioscience). The [3H]PK11195 counts in whole blood were obtained by removing 10 l aliquots from 100 l of whole blood labeled with 10 nM [3H]PK11195. Experiments were performed in triplicate using blood from three healthy volunteers. 2.3. Separation of plasma by gel filtration chromatography A K9/30 column (Amersham Biosciences) was packed with Sephacryl S-200HR resin equilibrated in phosphate buffered saline (PBS). The column was run at a flow rate of 0.4 ml/min and had a void volume of ⬃7.5 mls and an included volume of ⬃18mls. 50 l EDTA-plasma samples were treated with either 25 nM [3H]PK11195 or 25 nM [3H]PK11195 plus 66 M unlabelled PK11195 at 20°C for 10 min. A 30 l aliquot of the treated plasma was applied onto the top of the column which was then developed with PBS and the fractions were collected into 96 well plates. Total protein was determined by mixing a 5 l aliquot from each fraction with 250 l of 1x Advanced Protein Reagent (Cytoskeleton) and the absorbance at 600 nm recorded using a Fusion Universal Microplate Analyzer (Packard Bioscience). The radioactivity in each fraction was determined by mixing 10 l of sample with 100 l of Microscint 40 followed by counting for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter. 2.4. [3H]PK11195 binding assay using spin desalting columns Micro Bio-spin chromatography columns packed with Bio-Gel P6 (Bio Rad) were equilibrated in PBS according to manufacturer’s instructions. Human serum albumin (HSA) (Sigma), ␣1-acid glycoprotein (AGP) (Sigma) and apolipoprotein A1 (APO) (Calbiochem) solutions were prepared in PBS at concentrations of 50 mg/ml, 1 mg/ml and 1 mg/ml, respectively. 30 l aliquots of PBS, EDTA-plasma, HSA, AGP and APO were incubated with incubated with either 10 nM [3H]PK11195 or 10 nM [3H]PK11195 plus 66 M unlabelled PK11195 at 20°C for 10 min. 25 l aliquots of each incubation were loaded onto the top of the spin columns which were then centrifuged at 1000g for 4 min, 20°C. Total protein in the column loads and eluates were determined by removing a 1 l aliquots from each and mixing with 250 l of 1x Advanced Protein Reagent (Cytoskeleton) and recording the absorbance at 600 nm using a Fusion Universal Microplate Analyzer (Packard Bioscience). Recovery of the protein from the columns in all cases was around 95%. The radioactivity in the column loads and eluates was determined by mixing with 100 l of Microscint 40 followed by counting for 1 min in a TopCount NXT Microplate Scintillation and Luminescence Counter.
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
201
2.5. Immunodepletion of AGP from plasma A goat anti-human AGP antibody (OG20, Europa Bioproducts Ltd) was immobilized and cross-linked to Protein G using the Seize X immunoprecipitation kit (Perbio) according to manufacturer’s instructions. Approximately 1 mg of the OG20 antibody in PBS was cross-linked to 400 l of the Protein G resin slurry. 200 l of EDTA-plasma was incubated with for 2 hr at 4°C with the OG20 resin to immunodeplete the AGP. An untreated plasma sample was also prepared in parallel and kept at 4°C for the same length of time as the OG20 treated plasma. The OG20 treated and untreated plasma samples were immediately assayed for [3H]PK11195 binding using the spin desalt column method followed by gel filtration chromatography as described above. 5 l of plasma was removed from the OG20 treated and the untreated plasma samples, mixed with 35 l of SDS-loading and boiled for 3 min. 2.5 l samples were loaded onto Novex 4-12% NuPAGE Bis-Tris gels (Invitrogen) and run in MES buffer (Invitrogen) according to the manufacturer’s instructions along with Mark12 unstained protein markers (Invitrogen). For total protein staining gels were stained overnight with 30 mls of SYPRO Ruby stain (Molecular Probes), briefly destained in 10% methanol/7% acetic acid before image capture on a Genegenius imaging system (Syngene). For Western analysis the SDS-gels were blotted using the XCell II blot module (Invitrogen) according to the manufacturer’s instructions onto nitrocellulose. Membranes were stained in Ponceau S, briefly washed in PBS plus 0.05% Tween 20 (PBST), blocked in PBST/5% low fat milk for 30 min, probed overnight with an antihuman AGP monoclonal (AG47, Sigma) used at a dilution of 1:10000 in PBST/5% low fat milk, washed extensively in PBST and probed for 1 hr with an anti-mouse HRP conjugate (Sigma) used at a dilution of 1:10000 in PBST/5% low fat milk. After extensive washing in PBST the blot was developed with Supersignal (Perbio) and exposed to Hyperfilm ECL (Amersham Biosciences). Film images were captured using a Genegenius imaging system (Syngene) and the AGP bands quantitated using GeneTools (Syngene).
2.6. Displacement of fluorescein-dexamethasone from AGP by PK11195 5 M AGP was incubated with 10 nM fluoresceindexamethasone (Molecular Probes) and mixed with PK11195 ranging in final concentration from 5 nM to 10 M. The reactions were performed at 20°C in PBST. The fluorescence polarization signal of the fluorescein-dexamethasone was measured in a Screenstation (Molecular Devices) using excitation and emission filters of 490 nm and 535 nm, respectively. The data were analyzed in GraFit (Erithacus Software Ltd) to determine the IC50 value.
Fig. 1. Time course following the distribution of [3H]PK11195 into the cellular elements of blood.
2.7. Quantitation of AGP in the plasma of healthy volunteers The plasma concentration of AGP in was determined by Western blot analysis using the antibody AG47 as described above. Plasma samples were loaded onto Novex 4-12% NuPAGE Bis-Tris gels (Invitrogen) and run in MES buffer (Invitrogen) according to the manufacturer’s instructions along with AGP stocks of known concentration. Blots were developed with Supersignal (Perbio) and images were captured using a Chemigenius imaging system (Syngene). Plasma AGP levels were in the range 1.1-1.4 mg/ml for volunteers used in the study. 2.8. Statistical analysis The student’s t-test was used to determine the significance of differences in the binding of [3H]PK11195 to normal plasma and treated plasma (Figures 3a and 5).
3. Results Experiments studying [3H]PK11195 binding to whole blood indicated that (i) equilibrium was reached within five minutes and (ii) ⬃15% of the ligand was distributed in pelleted material following low speed centrifugation (Figure 1). The pellet contained the cellular elements of blood and a small amount of trapped plasma. This indicated that the binding of PK11195 to blood cells from healthy volunteers is low and that the majority of the compound was distributed in the plasma. To determine whether PK11195 was protein bound in the plasma the labeled ligand was mixed with EDTA-plasma and separated using gel filtration chromatography which resolved the plasma proteins into two distinct peaks (Figure 2). The [3H]PK11195 signal eluted as a single peak on the shoulder of the larger protein peak suggesting an interaction with one of the lower molecular weight proteins in the plasma. The binding of [3H]PK11195 to plasma was displaced using an excess of the unlabelled ligand indicating that a specific binding interaction was occurring. Interest-
202
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
Fig. 2. Separation of [3H]PK11195 labeled plasma by gel filtration chromatography (x total protein measurement, ■ CPM (x102) [3H]PK11195, Œ CPM (x102) [3H]PK11195 plus excess unlabelled PK11195).
ingly, in this latter experiment the unbound [3H]PK11195 did not elute with the included volume of the column. Preliminary data (not shown) suggest that [3H]PK11195 has a propensity to stick to plasticware and thus may have been retained on the column surfaces. Consistent with this observation the recovery of [3H]PK11195 from the first gel filtration experiment was only around 40%. The most common drug-binding plasma proteins are albumin and ␣1-acid glycoprotein (AGP). Analysis of the column fractions by SDS-PAGE (data not shown) indicated that the main protein in the peak, around fraction 37 (Figure 2), was human serum albumin (HSA). This suggested that [3H]PK11195 did not bind to HSA and that it might instead interact with AGP. The binding of [3H]PK11195 to EDTA-plasma, HSA and AGP was measured using a spin column method (Figure 3). The concentrations of the purified plasma proteins used in the assay reflected their approximate concentrations in healthy human plasma (HSA 50 mg/ml, AGP 1 mg/ml). Two sets of controls using buffer alone and apolipoprotein A-1 (APO, 1 mg/ml) indicated that (i) the ligand by itself did not elute from the column and (ii) the assay yielded specific results as no radioactivity eluted with APO (Figure 3a). The assay confirmed that [3H]PK11195 bound to plasma proteins and that there was an interaction with AGP and, to a lesser extent, with HSA. Although the binding of PK11195 to AGP was almost completely reversible, its binding in plasma was only partially displaced by unlabelled ligand (Figure 3a). This may be due to a weak interaction of the ligand with HSA which at the physiological concentrations used in the assay binds [3H]PK11195. When the amount of radioactivity associated with each protein was expressed on a per milligram basis, the binding of [3H]PK11195 to HSA was reduced to the background levels observed with APO (Figure 3b). This demonstrated that there was a much higher affinity interaction between [3H]PK11195 and AGP compared with HSA. The percentages of the total radioactivity added to the assay which bound to APO, plasma, HSA and AGP were 1%, 82%,
Fig. 3. a. Elution of [3H]PK11195 from spin desalting columns with buffer alone, APO (apolipoprotein A1, 1 mg/ml), EDTA-plasma, HSA (human serum albumin (50 mg/ml) and AGP (␣1-acid glycoprotein, 1 mg/ml). b. Elution of [3H]PK11195 from spin desalting columns expressed per mg of protein. Light gray bars [3H]PK11195 alone, dark gray bars [3H]PK11195 plus excess unlabelled PK11195. Values in (a) marked with *** are significantly lower (p⬍0.001) than plasma incubated with [3H]PK11195 alone.
47.5% and 100%, respectively. These data indicate that PK11195 is extensively protein bound in plasma. Gel filtration chromatography of the purified AGP incubated with [3H]PK11195 resulted in the elution of a single protein peak and a single peak of radioactivity (data not shown). The [3H]PK11195 was slightly displaced from the AGP suggesting a slightly later elution from the column. Western blot analysis for AGP in the fractions from the plasma-labeled [3H]PK11195 (Figure 2) also demonstrated a similar displacement of the AGP and [3H]PK11195 peaks (data not shown). This may reflect an interaction of the ligand with the column matrix as noted above. In order to confirm that AGP was the sole high affinity binder of PK11195 in plasma, AGP was immunodepleted from plasma using a goat anti-human AGP antibody (OG20). Western blot analysis of plasma, following treatment with OG20, demonstrated that AGP levels were reduced to ⬃5% of their original level (Figure 4a). The OG20-treated plasma sample demonstrated no significant changes in the overall pattern of plasma proteins with the exception of a band with a molecular weight of ⬃50 kDa that was significantly reduced in the OG20 treated plasma (Figure 4b). The molecular weight of AGP from human plasma has been reported to be between 41-43 kDa [13] and
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
203
Fig. 5. Elution of [3H]PK11195 from spin desalting columns with control plasma and with AGP-plasma. Light gray bars [3H]PK11195 alone, dark gray bars [3H]PK11195 plus excess unlabelled PK11195. Value marked * is significantly lower (p ⬍0.05) than the control plasma incubated with [3H]PK11195 alone.
1.2 M. The theoretical maximal IC50 value that can be measured in the assay is around half the concentration of AGP used in the assay, which is 2.5 M. The finding that the measured IC50 was around half this value is indicative of a IC50 value of less than 1.2 M. Fig. 4. Immunodepletion of ␣1-acid glycoprotein from EDTA-plasma. a. Western blot analysis of untreated plasma (-) and plasma treated with the anti human AGP polyclonal antibody OG20 (⫹). b. SDS-PAGE of samples stained for total protein with SYPRO ruby stain.
therefore this band is unlikely to represent AGP. A more likely candidate is the heavy chain of IgG which has a molecular weight of ⬃50 kDa. The decrease in the intensity of this band in the OG20 treated plasma may indicate that a small proportion of the plasma IgG may have bound to unoccupied antibody binding sites on the Protein G resin used in the immunodepletion. The data nevertheless indicate that OG20 selectivity depletes human plasma of AGP. The binding of [3H]PK11195 to plasma depleted of AGP by OG20 (AGP-plasma) was initially analyzed using the spin column method. This indicated that the level of [3H]PK11195 binding to the AGP-plasma was significantly reduced (p⬍0.05) compared with control plasma (Figure 5). Addition of a large excess of the unlabelled ligand did not significantly reduce the binding of the [3H]PK11195 further, consistent with its weak interaction with HSA. Gel filtration chromatography of the control plasma and AGP-plasma samples indicated broadly similar protein elution patterns consistent with the results of the total protein staining by SDS-PAGE (Figure 4b). While a single peak of radioactivity eluted with the control plasma sample, no corresponding radioactivity eluted with AGP-plasma. These data confirm that AGP is the major protein in human plasma to which PK11195 binds with high affinity. The interaction of PK11195 with AGP was also confirmed by its ability to displace AGP bound fluoresceindexamethasone. Under the conditions used, PK11195 was able to displace fluorescein-dexamethasone with a IC50 of
4. Discussion A number of studies examining the binding of [3H]PK11195 have reported the presence of the PBR in a wide range of blood cells [14,15,16]. Canat et al. [17] reported the highest levels of [3H]PK11195 binding to monocytes, polymorphonuclear cells and lymphocytes, and much lower levels of binding to platelets and erythrocytes (see Table 1). They demonstrated that granulocytes account for ⬃70% of the total number of cellular PK11195 binding sites in whole blood, with the remaining sites being shared more or less equally between the other cell types. Assuming a single PK11195 binding site on each PBR [17], the concentration of binding sites in blood is ⬃10 nM. The reported binding constant for PK11195 on polymorphonuclear cells is around 3 nM [17] and, at the concentrations of [3H]PK11195 used in the blood binding assay, it was expected that a high proportion of ligand would be bound to the cells. Our initial investigations found that only a relatively small amount (⬃15%) of the added [3H]PK11195 was bound to blood cells and that the remainder was distributed in the plasma. This suggested that PK11195 was extensively bound in plasma. Three lines of evidence are consistent with the conclusion that PK11195 binds with high affinity to AGP. Firstly, incubation of purified AGP with [3H]PK11195 followed by a rapid separation from unbound ligand using microspin columns indicated that all of the added ligand eluted with the protein. Secondly, immunodepletion of AGP from plasma abolished the binding of [3H]PK11195. Thirdly, PK11195 was able to displace fluorescein-dexamethasone
204
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
Table 1 Calculation of the total number of PK11195 cellular binding sites in a microlitre of blood. Cell population
Number of PK11195 binding sites per cell*
Average values for cells in normal human blood (cell/ l)#
Number of PK11195 binding sites per l of normal human blood
Lymphocytes Polymorphonuclear cells Monocytes Platelets Erythrocytes
1 ⫻ 105 7.2 ⫻ 105 7.5 ⫻ 105 1.2 ⫻ 103 1.1 ⫻ 102
2.75 ⫻ 103 6 ⫻ 103 5.4 ⫻ 102 3 ⫻ 105 5.4 ⫻ 106 Total number of PK11195 binding sites per l of normal blood
2.75 ⫻ 108 4.32 ⫻ 109 4.05 ⫻ 108 3.60 ⫻ 108 5.94 ⫻ 108 5.95 ⴛ 109
* Values from Canat et al [17]. # Values from Ganong [30].
from AGP with a IC50 of ⬍1.2 M. There are two consequences resulting from this interaction. Firstly, the high plasma concentration of AGP (⬃20 M) results in a smaller than expected proportion of the ligand binding to the blood cells. Secondly, PK11195 is highly (⬎80%) protein bound in plasma. Our results also indicated that PK11195 interacted very weakly with HSA and at the tracer amounts typically used in PET studies this binding is likely to be insignificant due to its higher affinity interaction with AGP. Consistent with our findings Dougherty et al [18] have recently reported the binding of PK11195 to site II on HSA. AGP, also known as orosomucoid, is a glycoprotein of 183 residues with an extremely high carbohydrate content accounting for ⬃45% of its mass [19]. The plasma level of AGP is relatively stable in healthy subjects (at around 1 mg/ml) but can rapidly increase up to three-fold in response to acute inflammation. It is therefore considered to be an important member of the positive acute phase protein family. Increased plasma levels of AGP are associated with inflammatory diseases such as Crohn’s disease [20], rheumatoid arthritis [21] and MS [22]. Although the precise biological function of AGP is unclear, it appears to have both anti-inflammatory effects and a role in immunomodulation [19]. The displacement assay used to estimate the binding affinity of PK11195 for AGP utilizes fluoresein-dexamethasone as a fluorescent probe. The assay is sensitive to a number drugs, such as propanalol [23] and prazosin [24], which are reported to bind to the common binding site for basic drugs on AGP [19]. In the assay propanalol and prazosin are able to displace fluoresein-dexamethasone from AGP with IC50 values of 4.9 and 2.3 M, respectively (data not shown). The data is therefore consistent with PK11195 binding to this common ligand pocket on AGP although the compound may interact with additional sites on the protein. The potential consequences resulting from the interaction of PK11195 with AGP are three-fold: (i) variations in
levels of AGP could significantly alter the free plasma concentrations of the ligand, (ii) AGP and AGP-bound ligand may contribute to the delivery of ligand to the brain parenchyma (i.e., across a break in the blood brain barrier), (iii) local synthesis of AGP at sites of inflammation in the brain tissue could provide additional PK11195 binding sites. A number of studies have found a strong inverse correlation between the plasma AGP concentration and the amount of free drug (for example propanolol) in the plasma of healthy individuals [25,26]. Normal variation in plasma AGP levels between individuals and changes in the levels of plasma AGP in individuals with underlying diseases, or as a consequence of a bacterial or viral infection, may alter the levels of free PK11195 in the plasma. Both these factors may contribute to the reported unstable kinetic behavior of [11C]PK11195 in the plasma compartment [1]. Longitudinal monitoring of plasma AGP levels during PET studies may provide useful information in modeling the distribution of the ligand. In diseases with significant blood brain barrier breakdown, such as MS [27] and to a lesser extent AD [28], the passage of AGP into the brain parenchyma may contribute to the increased binding of [11C]PK11195 in the damaged tissue in a number of ways: firstly, via the direct passage of AGP-bound ligand into the brain tissue and secondly, an increase in the local concentration of AGP may draw the free ligand from the plasma into the brain. Plasma AGP is mainly derived from hepatic synthesis and its expression is induced by glucocorticoids and proinflammatory cytokines such as interleukin 1 and interleukin 6 [19]. Extrahepatic expression of AGP has, however, been detected in a wide range of tissues and cells, including those of the immune system such as lymphocytes, monocytes, alveolar macrophages and granulocytes [reviewed in 19]. Given the potential immunomodulatory role of AGP, its local synthesis by these cells may help limit tissue damage during inflammation and there is some evidence that these extra-hepatic cell types may be regulated by cytokines in the
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
same manner as hepatic cells [29]. It is also possible that cells, such as glia, which are normally resident in the CNS, synthesis AGP. Thus there is the potential in neurological diseases with a significant cellular infiltrate, such as MS, for the local synthesis of AGP which may contribute to the increased focal binding of PK11195. In summary, we have identified that the PBR ligand PK11195 binds with high affinity to the plasma protein and acute phase reactant AGP. This may explain the observed variability in kinetic behavior of [11C]PK11195 in PET studies and that AGP-bound ligand may contribute to the pattern of PK11195 binding observed in neuroinflammatory diseases. These findings provide the basis for further studies investigating the distribution and potential role of AGP in neuroinflammation.
Acknowledgments We are grateful to Dr Murray Brown and Dr Andrew Pope for assistance with the fluorescent displacement assay and to Dr. Ron Leslie for helpful discussions.
References [1] R.B. Banati, J. Newcombe, R.N. Gunn, A. Cagnin, F. Turkheimer, F. Heppner, G. Price, F. Wegner, G. Giovannoni, D.H. Miller, G.D. Perkin, T. Smith, A.K. Hewson, G. Bydder, G.W. Kreutzberg, T. Jones, M.L. Cuzner, R. Myers, The peripheral benzodiazepine binding site in the brain in multiple sclerosis: quantitative in vivo imaging of microglia as a measure of disease activity, Brain 123 (2000) 2321–2337. [2] A. Cagnin, D.J. Brooks, A.M. Kennedy, R.N. Gunn, R. Myers, F.E. Turkheimer, T. Jones, R.B. Banati, In-vivo measurement of activated microglia in dementia, Lancet 358 (2001) 461– 467. [3] R.B. Banati, G.W. Goerres, R. Myers, R.N. Gunn, F.E. Turkheimer, G.W. Kreutzberg, D.J. Brooks, T. Jones, J.S. Duncan, [11C](R)PK11195 positron emission tomography imaging of activated microglia in vivo in Rasmussen’s encephalitis, Neurology 53 (1999) 2199 – 2203. [4] S. Pappata, M. Levasseur, R.N. Gunn, R. Myers, C. Crouzel, A. Syrota, T. Jones, G.W. Kreutzberg, R.B. Banati, Thalamic microglial activation in ischemic stroke detected in vivo by PET and [11C]PK11195, Neurology 55 (2000) 1052–1054. [5] A. Cagnin, R. Myers, R.N. Gunn, A.D. Lawrence, T. Stevens, G.W. Kreutzberg, T. Jones, R.B. Banati, In vivo visualization of activated glia by [11C] (R)-PK11195-PET following herpes encephalitis reveals projected neuronal damage beyond the primary focal lesion, Brain 124 (2001) 2014 –2027. [6] M. Gavish, I. Bachman, R. Shoukrun, Y. Katz, L. Veenman, G. Weisinger, A. Weizman, Enigma of the peripheral benzodiazepine receptor, Pharmacol Rev 51 (1999) 629 – 650. [7] Agnello, L. Carvelli, V. Muzio, P. Villa, B. Bottazzi, N. Polentarutti, T. Mennini, A. Mantovani, P. Ghezzi, Increased peripheral benzodiazepine binding sites and pentraxin 3 expression in the spinal cord during EAE: relation to inflammatory cytokines and modulation by dexamethasone and rolipram, J Neuroimmunol 109 (2000) 105–111. [8] A.C. Kuhlmann, T.R. Guilarte, Regional and temporal expression of the peripheral benzodiazepine receptor in MPTP neurotoxicity, Toxicol Sci 48 (1999) 107–116.
205
[9] A. C. Kuhlmann, T.R. Guilarte, Cellular and subcellular localization of peripheral benzodiazepine receptors after trimethyltin neurotoxicity, J Neurochem 74 (2000) 1694 –1704. [10] E.L. Conway, A.L. Gundlach, J.A. Craven, Temporal changes in glial fibrillary acidic protein messenger RNA and [3H]PK11195 binding in relation to imidazoline-I2-receptor and alpha 2-adrenoceptor binding in the hippocampus following transient global forebrain ischaemia in the rat, Neuroscience 82 (1998) 805– 817. [11] E. Vowinckel, D. Reutens, B. Becher, G. Verge, A. Evans, T. Owens, J.P. Antel, PK11195 binding to the peripheral benzodiazepine receptor as a marker of microglia activation in multiple sclerosis and experimental autoimmune encephalomyelitis, J Neurosci Res 50 (1997) 345–353. [12] R.B. Banati, R. Myers, G.W. Kreutzberg, PK (’peripheral benzodiazepine’)-binding sites in the CNS indicate early and discrete brain lesions: microautoradiographic detection of [3H]PK11195 binding to activated microglia, J Neurocytol 26 (1997) 77– 82. [13] J.M. Kremer, J. Wilting, L.H. Janssen, Drug binding to human alpha1-acid glycoprotein in health and disease, Pharmacol Rev 40 (1988) 1– 47. [14] S. Maeda, T. Miyawaki, T. Nakanishi, M. Takigawa, M. Shimada, Peripheral type benzodiazepine receptor in T lymphocyte rich preparation, Life Sci 631 (1998) 423– 430. [15] P.N. Moynagh, L.A. O’Neill, D.C. Williams, Interleukin-1 and phorbol myristate acetate modulate the peripheral-type benzodiazepine receptor in lymphocytes and glial cells, Biochem Pharmacol 46 (1993) 821– 827. [16] A. Berkovich, C. Ferrarese, G. Cavaletti, H. Alho, C. Marzorati, G. Bianchi, A. Guidotti, E. Costa, Topology of two DBI receptors in human lymphocytes, Life Sci 52 (1993) 1265–1277. [17] X. Canat, P. Carayon, M. Bouaboula, D. Cahard, D. Shire, C. Roque, G. Le Fur, P. Casellas, Distribution profile and properties of peripheral-type benzodiazepine receptors on human hemopoietic cells, Life Sci 52 (1993) 107–118. [18] T.J. Dougherty, A.B. Sumlin, W.R. Greco, K.R. Weishaupt, L.A. Vaughan, R.K. Pandey, The role of the peripheral benzodiazepine receptor in photodynamic activity of certain pyropheophorbide ether photosensitizers: albumin site II as a surrogate marker for activity, Photochem Photobiol 76 (2002) 91–97. [19] T. Fournier, N. Medjoubi, D. Porquet, Alpha-1-acid glycoprotein, Biochim Biophys Acta 1482 (2000) 157–171. [20] O.H. Nielsen, B. Vainer, S.M. Madsen, J.B. Seidelin, N.H. Heegaard, Established and emerging biological activity markers of inflammatory bowel disease, Am J Gastroenterol 95 (2000) 359 –367. [21] T. Nakamura, P.G. Board, K. Matsushita, H. Tanaka, T. Matsuyama, T. Matsuda, Alpha 1-acid glycoprotein expression in human leukocytes: possible correlation between alpha 1-acid glycoprotein and inflammatory cytokines in rheumatoid arthritis, Inflammation 17 (1993) 33– 45. [22] P.C Dowling, S.D. Cook, Disease markers in acute multiple sclerosis, Archives of Neurology 33 (1976) 668 – 677. [23] F. Albani, R. Riva, M. Contin, A. Baruzzi, Stereoselective binding of propranolol enantiomers to human alpha 1-acid glycoprotein and human plasma, Br J Clin Pharmacol 18 (1984) 244 –246. [24] F. Brunner, W.E. Muller, Prazosin binding to human alpha 1-acid glycoprotein (orosomucoid), human serum albumin, and human serum. Further characterization of the ’single drug binding site’ of orosomucoid, J Pharm Pharmacol 37 (1985) 305–309. [25] K.M. Piafsky, O. Borga, I. Odar-Cederlof, C. Johansson, F. Sjoqvist, Increased plasma protein binding of propranolol and chlorpromazine mediated by disease-induced elevations of plasma alpha1 acid glycoprotein, N Engl J Med 299 (1978) 1435–1439. [26] G. Alvan, K. Bergstrom, O. Borga, L. Iselius, N. Pedersen, Family study of genetic and environmental factors determining the protein binding of propranolol, Eur J Clin Pharmacol 5 (1983) 437– 441. [27] N.C. Silver, P.S. Tofts, M.R. Symms, G.J. Barker, A.J. Thompson, D.H. Miller, Quantitative contrast-enhanced magnetic resonance im-
206
A. Lockhart et al. / Nuclear Medicine and Biology 30 (2003) 199 –206
aging to evaluate blood-brain barrier integrity in multiple sclerosis: a preliminary study, Mult Scler 7 (2001) 75– 82. [28] T.M. Berzin, B.D. Zipser, M.S. Rafii, V. Kuo-Leblanc, G.D. Yancopoulos, D.J. Glass, J.R. Fallon, E.G. Stopa, Agrin and microvascular damage in Alzheimer’s disease, Neurobiol Aging 21 (2000) 349 – 355.
[29] C.G. Gahmberg, L.C. Andersson, Leukocyte surface origin of human alpha 1-acid glycoprotein (orosomucoid), J Exp Med 148 (1978) 507–521. [30] W.F. Ganong, Circulating Body Fluids. in: W.F. Ganong (Ed), Review of Medical Physiology (1999) pp. 493–521) Appleton and Lange: Stamford.