SEARCH

SEARCH BY CITATION

Keywords:

  • Allergens;
  • Plasma;
  • Protein adducts;
  • Target for drugs

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

Flucloxacillin is a synthetic penicillin used in the treatment of Staphylococcal infections. Adverse reactions to the drug are believed to arise through covalent modification of proteins, with tissue damage occurring secondary to an immune reaction. Serum proteins have been shown by adduct-specific antibodies to be modified by flucloxacillin, but the nature and sites of modification have not been characterised. Here, in vitro studies on HSA have shown by MS that the modification of protein lysine residues occurs in a dose-, time- and site-dependent manner. Affinity, cation exchange and reversed phase chromatography prior to MS revealed in vivo modification of HSA with flucloxacillin in tolerant patients, with up to nine modified lysine residues being detected in each patient, and with modification of Lys190 and Lys212 being detected in 8/8 patients. It was also revealed for the first time that plasma proteins could be modified with the 5-hydroxymethyl metabolite of flucloxacillin, and that essentially the same Lys residues were targeted by both the parent drug and its metabolite. This study provides a detailed characterisation of sites of chemical modification of an endogenous target and reveals candidate peptides for T-cell and antibody assays of flucloxacillin hypersensitivity.


1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

Flucloxacillin is a synthetic, penicillinase-resistant isoxazolyl penicillin 1 that is used for the treatment of Staphylococcal infections, particularly those of the skin, urinary tract and respiratory tract. The antibiotic is well tolerated in most individuals at oral doses of up to 4 g per day 2, but there are some individuals who develop adverse reactions such as hepatic cholestasis (bile duct obstruction), nephritis, rash, fever and eosinophilia 3, 4. The adverse reactions are characterised by delayed onset of symptoms, sometimes several weeks after treatment has been discontinued 5, and prolonged recovery periods 2. In some, mainly elderly patients, flucloxacillin induces fulminant hepatic failure, which can be fatal 2, 4, 6. Indeed, flucloxacillin has been reported to be the second most common cause of drug-induced liver failure after paracetamol 7.

The mechanism of flucloxacillin toxicity has not been fully determined, but may vary according to drug disposition, and be dependent on both metabolism- and immune-mediated events 4, 5, 8. Cytochrome P450 3A4 metabolises the drug to 5′-hydroxymethyl flucloxacillin, which has been shown to damage bile duct epithelial cells, but not hepatocytes, in culture 8. This is consistent with patient studies where bile duct damage has been observed with little or no hepatic necrosis 8. However, an immune-mediated pathogenesis has also been postulated for flucloxacillin cholestasis 4. By contrast, in patients who develop skin rash and fever without any hepatic involvement, immune-mediation has been presumed on the basis of symptoms 5. Drug-specific T cells have also been detected in the blood of patients experiencing flucloxacillin-induced interstitial nephritis 3, and flucloxacillin-specific IgE has been detected in the serum of allergic patients 9. Taking all the evidence together, it seems likely that the different manifestations of flucloxacillin idiosyncratic reactions are immune-mediated.

According to the hapten hypothesis, drug modification of host proteins is an important event in the pathogenesis of an immune reaction 10. In relation to flucloxacillin, several studies have detected drug bound to both tissue and serum proteins in vivo using adduct-specific antibodies 5, 11. However, the nature and sites of the modification have not been characterised. In this study, the modification of isolated amino acids and the model protein HSA by flucloxacillin in vitro have been determined by MS. The incidence and sites of modification of HSA were also determined in serum from individual non-hypersensitive patients undergoing flucloxacillin antimicrobial therapy. We also report for the first time the detection of 5′-hydroxymethyl flucloxacillin modification of protein in vivo.

2 Materials and methods

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

2.1 Chemicals and supplies

HSA (approx. 99% pure, essentially globin free and fatty acid free), DTT and other standard chemicals were obtained from Sigma-Aldrich (Poole, Dorset, UK), and modified trypsin from Promega (Southampton, Hampshire, UK). HPLC grade solvents were purchased from BDH.

2.2 Clinical details

Patients (n=8) receiving either i.v. and/or oral (p.o.) therapeutic flucloxacillin were recruited. Venepuncture samples were extracted into heparinised tubes prior to the first daily administration of the drug, with a minimum of 8 h since the previous dose. Samples were immediately placed on ice and were centrifuged at 2000×g at 4°C for 15 min. Small aliquots were prepared and stored at −80°C. Clinical details of the patients are presented in Table 1. Ethical approval was obtained from Liverpool local research ethics committee and each patient gave their informed consent.

Table 1. Details of patients used in the study
PatientAge (years)GenderFlucloxacillin dosing scheduleTotal dose (g)Average dose/day (g)Ion counts for modified K190Ion counts for modified K212No. of residues modified by FLUNo. of residues modified by 5OH-FLUConcentration free FLU in plasma (mM)Primary medical conditions
187F
  • 4 g i.v., 3 days

  • 1 g p.o., 3 days

152.57762879940.7Cellulitis
250M4 g i.v., 9 days364.024 1321499710Cellulitis
370M
  • 4 g i.v., 11 days

  • 8 g i.v., 10 days

1245.917 94033679412.6Septic arthritis
423M
  • 4 g p.o., 1 day

  • 2 g p.o., 2 days

82.71477553041.4Cellulitis
561M8 g i.v., 4 days328.012 33233939815.6Cellulitis
629M
  • 8 g i.v., 1 day

  • 1.5 g p.o., 9 days

21.52.22918621840.6Cellulitis
766M2 g p.o., 5 days102.025 58937828626.4Cellulitis
840M
  • 4 g i.v., 8 days

  • 4 g p.o., 1 day

364.0389516779724.3Cellulitis

2.3 In vitro reaction of flucloxacillin with N-acetyl-lysine and N-acetyl-cysteine

Flucloxacillin (10 mM) was incubated for 16 h at 37°C with the methyl esters of either N-acetyl lysine or N-acetyl-cysteine (10 mM) in phosphate buffer (13.08 mM KH2PO4, 62.27 mM K2HPO4, pH 7.4). The sample was extracted with ethyl acetate (3×20 μL). The pH of the aqueous phase was adjusted to 3 using concentrated phosphoric acid, and it was then extracted again with ethyl acetate (3×30 μL). The combined extracts were evaporated to dryness and were reconstituted in 25% methanol in 0.1% acetic acid prior to LC-MS analysis.

2.4 Concentration of free flucloxacillin in plasma

In order to generate a standard curve, control plasma (10 μL) was spiked with flucloxacillin at concentrations of 5, 10, 50, 100 or 1000 μM. The protein was removed by the addition of four volumes of ice-cold acetone, incubation at −20°C for 2 h and centrifugation at 14 000×g at 4°C for 10 min. The supernatant was collected, and the pellet was washed twice with acetone. The combined supernatants were evaporated to dryness and were reconstituted in 100 μL distilled water prior to LC-MS analysis. Patient samples were processed for MS analysis in the same manner as described above.

2.5 Concentration-dependent modification of HSA by flucloxacillin

HSA (130 μg, 40 μM) was incubated with flucloxacillin (4, 40 or 400 μM) in 50 μL phosphate buffer, pH 7.4, at 37°C for 16 h. The drug was removed by precipitation of the protein with ten volumes of ice-cold methanol followed by centrifugation at 14 000×g at 4°C for 10 min. This was repeated three times, and the protein pellet was reconstituted in 25 μL phosphate buffer. The protein was reduced by incubation with 10 mM DTT w/v at 55°C for 15 min, and alkylated by incubation with 166 mM iodoacetamide w/v for a further 15 min at room temperature. The samples were again subjected to methanol precipitation and were reconstituted in 30 μL 50 mM ammonium bicarbonate buffer. Trypsin (1 μg) was added, and the samples were incubated overnight at 37°C. The digestions were desalted using C18 Zip-Tips (Millipore) and dried prior to LC-MS/MS analysis.

2.6 Time-dependent modification of HSA by flucloxacillin

HSA (1.3 mg, 40 μM) was incubated with 400 μM flucloxacillin in 500 μL phosphate buffer, pH 7.4, at 37°C for 16 h. Aliquots of 50 μL were removed after 30 min, 1, 2, 3, 16 and 24 h and processed for MS analysis as described in Section 2.5.

2.7 Isolation of HSA from plasma by affinity chromatography

Serum albumin was isolated from whole plasma by means of affinity chromatography, as described in Section 2.5 12. A POROS anti-HSA affinity cartridge (Applied Biosystems, Foster City, CA, USA) attached to a Vision Workstation (Applied Biosystems) was used to capture HSA, which was then eluted with 12 mM hydrochloric acid. Fractions containing protein were identified by means of UV detection at 280 nm, and were methanol precipitated and processed for MS analysis, as described previously.

2.8 Peptide fractionation by cation exchange chromatography

For higher sensitivity detection of adducts in human plasma, samples from eight patients exposed to flucloxacillin during the course of therapy were processed individually for 3-D LC-MS/MS analysis. Aliquots of 400 μg of affinity-isolated HSA were precipitated and digested as described in Section 2.5, and the digests were diluted to 2 mL with 10 mM potassium dihydrogen phosphate/25% w/v ACN. The pH of the samples was adjusted to <3 using phosphoric acid prior to fractionation on a Polysulfoethyl A strong cation-exchange column (200×4.6 mm, 5 μm, 300 Å; Poly LC, Columbia, MD, USA). Fractions of 2 mL were collected and were dried by centrifugation under vacuum (SpeedVac, Eppendorf).

2.9 MS

Free flucloxacillin present in patient plasma and flucloxacillin-modified amino acids were analysed by LC-MS. Samples were separated on a Prodigy 5 μm C8 column (150×4.6 mm, Phenomenex) using a gradient of methanol (0–50% over 10 min) in 0.1% v/v formic acid with a flow rate of 1 mL/min. The samples were delivered into an API2000 triple quadrupole mass spectrometer (Applied Biosystems) by a Series 200 pump and autosampler (Perkin Elmer). The column eluate was split to provide a flow rate of 200 μL/min to the TurboIonSpray source. The ionspray potential was set to 5500 V, the nebuliser gas to 80 and the interface heater to 400°C. Transitions for multiple-reaction monitoring MS were selected based on experimental data. One MRM transition specific for the free drug was used, combining the m/z of the drug (454 amu) with that of the dominant fragment ion (160 amu, [M+H]+ of thiazolidine ring). The MRM transitions for the flucloxacillin-modified amino acids were selected as follows: the m/z values of flucloxacillin-modified methyl esters of N-acetyl-lysine and N-acetyl-cysteine are 656 and 631, respectively, and both of these are associated with dominant fragment ions of m/z 160 and m/z 454 ([M+H]+ of cleaved drug adduct). MRM transitions were acquired at low resolution in both the Q1 and Q3 quadrupoles to maximize sensitivity, and in positive ion mode. Collision energies were optimised for each MRM, collision cell exit potentials were optimized for the 160 m/z fragment and dwell times were 100 ms.

For LC-MS/MS analysis, digested and ZipTipped samples were reconstituted in 10 μL 5% ACN/0.05% v/v TFA, and aliquots of 0.5–2 μL were analysed, whereas cation exchange fractions were reconstituted in 120 μL of 5% ACN/0.05% TFA, and aliquots of 60 μL were analysed. Samples were delivered into a QSTAR® Pulsar i hybrid MS by automated in-line LC (integrated LCPackings System, 5 mm C18 nano-precolumn and 75 μm×15 cm C18 PepMap column (Dionex, California, USA)) via a 10 μm inner diameter PicoTip (New Objective, MA, USA). For prefractionated samples, the nano-precolumn was washed for 30 min with 5% ACN/0.05% TFA prior to initiation of the solvent gradient in order to reduce the salt content of the sample. A gradient from 5% ACN/0.05% v/v TFA to 48% ACN/0.05% v/v TFA in 60 min (unfractionated samples) or 70 min (fractionated samples) was applied at a flow rate of 300 nL/min. MS and MS/MS spectra were acquired automatically in positive ion mode using information-dependent acquisition (Analyst, Applied Biosystems). Survey scans of 1 s were acquired for m/z 400–2000, and the three most intense ions were selected for MS/MS, with accumulation times of 1 s and with a dynamic exclusion of 40 s. Database searching was performed using ProteinPilot version 2 (Applied Biosystems) against the latest version of the SwissProt database, with biological modifications allowed and with the confidence level set to 90%. Flucloxacillin (mass addition 453 amu) or 5′-hydroxymethyl flucloxacillin (mass addition 469 amu) were included as high probability user-defined modifications of Lys and carboxamidomethyl as a fixed modification of Cys. The data were also assessed manually for the presence of a dominant fragment ion of 160 amu, indicative of cleavage of the thiazolidine ring from the drug adduct, and the presence of the remaining adduct fragment of 294 amu (flucloxacillin) or 310 amu (5′-hydroxymethyl flucloxacillin) present on the lysine residue.

2.10 Computer modelling of the reactive Lys residues in HSA

One measure of the reactivity of ionisable amino acid residues is the pKa value, which provides information on the pH dependence of protein characteristics such as folding, protein–protein interactions, ligand binding and so on 13, 14. The local environment of Lys, Glu, Asp, His, Arg and Cys residues within a protein will influence their pKa values in a reasonably predictable manner, so that computer algorithms may be used to determine positional pKa values empirically from structural information listed in databases such as the Protein Data Bank (PDB) 13–16. The crystal structure of HSA determined at 2.5 Å resolution (1 bm0) 17 was downloaded from the PDB website and was analysed using two free web-based packages, PropKa (http://propka.ki.ku.dk/∼drogers/) 13, 14 and H++ (http://biophysics.cs.vt.edu/H++/index.php) 15. For the latter, the following settings were used: salinity was set to 0.15, internal dielectric to 6, external dielectric to 80, a pH of 7.5 was assumed and the method of calculation applied was that of Poisson–Boltzmann.

The methods used to calculate the pKa of an amino acid within a protein are thought to be less precise for residues buried within hydrophobic pockets 18. Thus, the following values were calculated for HSA: the distance of each Lys from each Ser or His residue (PyMol v1.0, Delano Scientific, USA), as these are thought to be involved in deprotonation of Lys 19, 20; the fractional solvent-accessible surface area (FSASA) as a measure of the accessibility of the Lys residues 21 and the position of the Lys residues with respect to the hydrophobic binding pockets (Sudlow sites I and II) 22. The same PDB file as above was used. The mean of each of these values was calculated for the unmodified and modified Lys residues, and the statistical significance of any differences was assessed by the appropriate 2-tailed t-test.

3 Results

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

3.1 Characterisation of flucloxacillin adducts on isolated amino acids

Flucloxacillin was incubated with the methyl esters of N-acetyl lysine or N-acetyl cysteine in order to determine the nature of the chemical modification and to assess the MS fragmentation pattern of the adduct. Figure 1A shows the chemical structure of flucloxacillin, and Fig. 1B shows the drug conjugated to amino acid or protein (R) and the most dominant ions formed from fragmentation of the adduct. Figures 1C and D show the MRM traces for the methyl esters of N-acetyl lysine and N-acetyl cysteine, respectively, exposed to flucloxacillin at pH7.4, and reveal that both of these amino acids may be modified when exposed to molar ratios of amino acid to drug of 1:5.

thumbnail image

Figure 1. Characterisation of flucloxacillin-modified amino acids by LC-MS. (A) Chemical structure of free flucloxacillin. (B) Chemical structure of flucloxacillin bound to an amino acid or peptide (R) and showing the possible MS-induced fragmentation sites. (C) Q1 scan of N-acetyl-lysine (NAL) modified with flucloxacillin (FLU) showing the mass and retention time of the conjugate. (D) Detection of the flucloxacillin-modified N-acetyl-lysine by MRM using the transition 650/160. (E) Q1 scan of N-acetyl-cysteine (NAC) modified with flucloxacillin showing the mass and retention time of the conjugate. (F) Detection of the flucloxacillin-modified N-acetyl-cysteine by MRM using the transition 631/160.

Download figure to PowerPoint

3.2 Modification of HSA with flucloxacillin in vitro

Time and dose modification analyses of HSA exposed to flucloxacillin in vitro revealed ten sites of modification within the protein, all at Lys residues (Fig. 2A) and none at the Cys residues. Modification induced a missed cleavage at the modified residue, presumably due to impedance of the trypsin by the bulky adduct on Lys. Modification of Lys190 and Lys212 was observed at the earliest time points and lowest concentrations of drug used, indicating that they may be the most reactive with flucloxacillin. Figures. 2B and C show the MS/MS spectra of the corresponding peptides, highlighting some of the diagnostic fragment ions observed for flucloxacillin-modified peptides. A dominant ion at m/z 160 indicates fragmentation of the adduct at the thiazolidine ring (circled, solid line) and a less dominant ion at m/z 454 reveals cleavage of the entire adduct from the peptide (circled, dotted line). In the survey scan, the m/z of the modified peptide is equivalent to the mass of the parent peptide plus 453, but the full-length peptide is observed with no mass addition in the MS/MS spectrum, suggesting that the bond with the flucloxacillin adduct is more labile than the peptide bond. The m/z of the full-length peptide plus a mass addition of 294 amu is also visible, indicating the cleavage of the adduct at the thaizolidine ring. These multiple fragmentation sites within the modified peptide result in weak, noisy and difficult to interpret MS/MS spectra as multiple patterns of fragmentation are overlaid. However, sufficient information may be gleaned to confidently identify Lys as the target residue in most cases.

thumbnail image

Figure 2. Time- and dose-dependency of flucloxacillin modification of HSA in vitro. (A) Spreadsheet listing the Lys residues of HSA modified with flucloxacillin at different time points and with different molar ratios of drug to protein. The modified Lys residues are indicated by an asterisk. (B and C) MS/MS spectra of tryptic peptides 210–218 and 182–195 revealing flucloxacillin modification of Lys212 and Lys190, respectively. Cleavage of the thiazolidine ring of the adduct yields a characteristic fragment ion of m/z 160 (circled, solid line) and cleavage of the entire adduct from the peptide yields an ion of m/z 454 (circled, dotted line) in each spectrum. (D) and (E) Semi-quantitative analysis of the time- and dose-dependent increase in the peptides containing modified Lys212 and Lys190: these are based on the area under the curve for the relevant extracted masses followed by normalisation using the total ion count for the sample.

Download figure to PowerPoint

A crude quantitative analysis of modification at Lys212 and Lys190 was performed by determining the area under the curve for the relevant extracted masses for the modified peptides, followed by normalisation of the ion intensity using the total ion count for the sample. This revealed a time- and dose-dependent increase in the relative levels of the modified peptides (Figs. 2D and E).

3.3 Modification of HSA with flucloxacillin in vivo

Patients undergoing flucloxacillin treatment for Staphylococcal infections at the Royal Liverpool University Hospital were recruited (Table 1). None of these patients had experienced an adverse reaction to flucloxacillin. Seven of the eight patients were male, and the mean age was 53 years (range 23–87). Both the dose and route of administration varied, while three of the patients (6, 7 and 8) had a history of kidney problems that may have affected renal clearance of the flucloxacillin.

In order to achieve the highest sequence coverage of HSA from patients, and thereby maximise the number of sites of modification detected, the protein was first affinity isolated from plasma using an anti-HSA column. HSA was eluted from the column in 12 mM HCl, providing an essentially pure solution of the protein 12. This enabled a sequence coverage of approximately 80% to be achieved routinely on a QStar MS instrument for the digested protein (data not shown), and between 0 and 3 sites of modification with flucloxacillin to be unequivocally identified in the patient samples (Fig. 3A). The same samples were subjected to cation exchange chromatography prior to LC-MS/MS analysis, and this enabled a sequence coverage of 99% to be achieved (data not shown) and between two and nine sites of modification to be detected (Fig. 3B). In addition, the modification of an extra site (Lys137) that had not been detected in the in vitro samples was observed in patients 1, 3, 5 and 8 following the extended sample processing.

thumbnail image

Figure 3. Characterisation of flucloxacillin-modified HSA isolated from patients undergoing antibiotic treatment. (A and B) Spreadsheets listing the flucloxacillin-modified Lys residues detected in HSA isolated from patient plasma without (A) and with (B) fractionation of the tryptic peptides by cation exchange chromatography. (C) MS/MS spectrum of tryptic peptide 210–218 revealing 5-hydroxymethyl flucloxacillin (5OHFLU) modification of Lys212. Cleavage of the thiazolidine ring of the adduct yields a characteristic fragment ion of m/z 160 (circled, solid line) and cleavage of the entire adduct from the peptide yields an ion of m/z 470 (circled, dotted line).

Download figure to PowerPoint

Metabolism of flucloxacillin in the liver results in the formation of 5-hydroxymethyl flucloxacillin 8 but little is known about the potential protein targets of the metabolite. This study reveals for the first time that plasma proteins may be modified by 5-hydroxymethyl flucloxacillin in vivo and that the metabolite targets the same residues as the parent drug in HSA (Fig. 3B, detection of both flucloxacillin adduct and metabolite adduct represented by double tick). Thus, for example, the peptide containing Lys212 (210AFKAWAVAR218) may be detected as a doubly charged ion with m/z 744.9, a mass increase compared with the unmodified peptide of 469 amu (Fig. 3C). A dominant ion at m/z 160 indicates fragmentation of the adduct to release the thiazolidine ring (circled, solid line) and a less dominant ion at m/z 470 reveals cleavage of the entire adduct from the peptide (circled, dotted line). In addition, ions present in the MS/MS spectrum for the full-length peptide plus 310 (m/z 1329.6) and the full-length peptide with no mass addition (m/z 1019.6) indicate that the adduct is cleaved at the thiazolidine ring and at the side chain amino group of the Lys residue, respectively. The MS/MS spectra for all other flucloxacillin and 5-hydroxymethyl flucloxacillin-modified peptides are shown in Figs. 1 and 2 of the Supporting Information, respectively.

The study of HSA incubated with flucloxacillin in vitro revealed that there was a dose- and time-dependent modification of the protein. For the patient samples, the area under the curve for the extracted masses for modified Lys190 and Lys212-containing peptides normalised using the total ion count for the sample was determined. Table 1 presents the lack of a clear correlation between these factors and the dose or period of treatment with flucloxacillin in this small cohort of patients.

3.4 Computer-based structural analysis of HSA

In an attempt to understand the selectivity of flucloxacillin binding to HSA, a series of software programs was exploited to create a theoretical map of the most reactive Lys residues. Table S1 (Supporting Information) shows that there was little correlation between the pKa values of the Lys residues and the formation of adducts with flucloxacillin, with the exception of Lys190 and 199, which displayed reduced pKa values in calculations by both web-based tools used (PropKa 7.42 and 7.46, respectively; H++ 6.03 and 0.69, respectively). However, overall there was no significant difference between the mean pKa of the unmodified and modified Lys using PropKa, and the difference was just significant (p=0.04, t-test) when using H++. The results for the FSASA analysis revealed that buried Lys residues were more prone to modification by flucloxacillin than surface ones, which was reflected in the fact that the mean accessible surface area was lower for the modified than the unmodified residues (p=0.03, t-test). It has been reported that the sites of covalent modification of proteins by benzylpenicillin may be influenced by the proximity of certain other residues, particularly serine or histidine: both of these residues have been suggested as the original target for penicilloylation, followed by transesterification to a nearby lysine 19. In the case of flucloxacillin, there was no significant difference between modified and unmodified lysine residues with respect to the calculated distance between the ε-amino nitrogen of the lysine and the imidazole ring of the nearest histidine residue, or the hydroxyl oxygen of the nearest serine residue (Table S1, calculated with PyMol v1.0, Delano Scientific). The most interesting observation was the apparent greater reactivity of Lys residues situated in Sudlow sites I and II 22. The 3-D structure of HSA with the flucloxacillin-modified Lys residues highlighted is shown in Fig. 3 of the Supporting Information.

4 Discussion

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

The penicillins are good examples of very effective drugs that are tolerated at high doses by the majority of patients, but which nonetheless induce severe and sometimes fatal reactions in some recipients 1–4, 6. The mechanism of toxicity has not been fully elucidated, but it is believed that covalent modification of proteins plays a role 5, 11. This could result in the generation of neoantigens, which is believed to be a key step in the induction of immune reactions towards penicillins. HSA non-covalently binds up to 95% of the circulating flucloxacillin in humans 23 and it also accounts for ∼90% of covalently bound penicilloyl groups in serum from patients treated with benzylpenicillin 24. In addition, penicilloylated HSA has been shown to activate penicillin-specific T cells from hypersensitive patients 25. These factors prompted this thorough investigation of the sites of modification of HSA with flucloxacillin in vitro and in vivo: this is important as better tools are essential for diagnosing hypersensitivity induced by the different penicillins, and for understanding the mechanisms of cross-reactivity between the β-lactam antibiotics.

Flucloxacillin was shown to covalently bind to the methyl esters of N-acetyl-cysteine and N-acetyl-lysine at pH 7.4, and to generate characteristic fragment ions on MS/MS analysis (Fig. 1). The ions were derived from several consistent fragmentations of the adduct: the thiazolidine ring was cleaved from the flucloxacillin to yield an ion of m/z 160 (Figs. 1D and F), with the remaining 294 amu associated with N-acetyl-lysine and N-acetyl-cysteine methyl esters to give ions of m/z 497 and 472, respectively. The adduct was also cleaved at the central amide bond to release the halogenated benzene ring linked to the isoxazole ring, yielding an ion of m/z 217; finally, the entire adduct was cleaved from the amino acids to yield an ion of m/z 454. However, the most dominant fragment ion observed was the one at m/z 160, and this provided a signature ion for flucloxacillin modification in all subsequent analyses.

A series of in vitro experiments was performed in order to develop the methods required to investigate low-level modification of proteins, and to gain an insight into the pattern of Lys reactivity that might be expected in vivo. Since HSA non-covalently binds a large proportion of the circulating flucloxacillin in vivo23, one concern was that covalent modification could occur post sampling and even during tryptic digestion of the protein. In order to eliminate this risk, all samples were stored in the short term at 4°C and in the long term at −80°C, and it was demonstrated that HSA incubated with high levels of flucloxacillin at 4oC displayed little or no modification. Modified HSA was also subjected to repeated methanol precipitation prior to overnight digestion in order to remove non-covalently bound drug. In samples incubated with high levels of drug where this was not done, tryptic peptides were detected with the flucloxacillin adduct on the N-terminal amine of peptides as well as on the Lys residues. In addition, modification of Lys results in a characteristic missed cleavage with trypsin, indicating that where this happens, modification occurred before digestion.

HSA was incubated with flucloxacillin at molar ratios of drug to protein of 1:100 to 10:1 for 30 min to 16 h at 37°C, with adducts being visible from 30 min at the highest level of drug (Fig. 2A). These conditions were selected because in patients, more than 80% of an oral dose is available, leading to a plasma Cmax of 15 mg/L (33 μM) after 0.5–1 h following administration of 500 mg per os23, 26, 27. The approximate concentration of HSA in plasma is 35 g/L (0.53 mM) 28 so that at the peak plasma concentration, the molar ratio of drug to protein is 1:16. Higher levels of drug were used here to assess the time dependency of the covalent modification in vitro, but Fig. 2A shows that adducts could be detected following incubation for 16 h with a molar ratio of drug to protein of 1:10, and therefore within the expected therapeutic concentration range. HSA has a half-life of approximately 19 days in vivo29 and although this is liable to be reduced by the presence of chemical adducts 30, the modified protein would be expected to accumulate over the course of antibiotic treatment. Thus, we were confident that the conditions used for the detection of protein adducts in in vitro studies were consistent with those that would be encountered in vivo. These studies revealed an approximately dose- and time-dependent increase in both the level of individual modified peptides (Figs. 2D and E) and in the range of modified Lys residues, with Lys190 and Lys212 being amongst the most reactive in vitro.

The in vivo half-life of flucloxacillin has been reported to be between 0.5 and 1.5 h 23 following oral administration, although this is increased in elderly patients to between 1.7 and 2.7 h following i.v. administration due to a lower rate of clearance 26. The half-life of the metabolite in elderly patients is between 2.2 and 3.75 h 26. It has also been reported that the circulating level of flucloxacillin is zero 8 h after ingestion of 750 mg of the drug 23. The blood samples analysed in this study were taken at least 8 h after the previous dose of flucloxacillin so that the level of free drug was generally below that required for the detection of protein modification in vitro (Table 1): the highest molar ratio of drug to protein (1:13) was detected in patient 4 who exhibited low levels of protein modification. This provides further assurance that there was no modification of proteins ex vivo. Modification of HSA was observed when whole plasma was subjected to methanol precipitation and processing for tryptic digestion (data not shown), but in order to increase the sensitivity of detection of modified residues in HSA, a multidimensional HPLC method was utilised. This increased the number of modified peptides detected in each patient compared with the unfractionated samples (Figs. 3A and B), and made possible the detection of modified Lys137, which had not been observed in the in vitro samples. Up to nine Lys residues modified with flucloxacillin and up to eight Lys residues modified with 5-hydroxymethyl flucloxacillin were detected. This is important as it reveals that albumin may be used as a biomarker for liver metabolism of flucloxacillin to 5-hydroxymethylflucloxacillin, which may also be related to the severity of adverse events, both chemical-induced and immune-mediated.

As determined for the in vitro-modified HSA, Lys 190 and Lys212 were also the most reactive residues in vivo. The pKa values reflect this for Lys190, which has one of the lowest values calculated by both PropKa (7.42) and H++ (6.03) 13–15. However, Lys212 has high-predicted pKa values (10.36 by PropKa and 11.09 by H++) that are not consistent with the deprotonated and therefore reactive state of the side chain amino group 13–15, 18, 31. Indeed, Table S1 (Supporting Information) shows that there appears to be little or no relationship between the calculated pKa, FSASA or distance from His or Ser and the apparent reactivity of the Lys residues of HSA with respect to flucloxacillin binding. One factor that did seem to impact on Lys modification was the presence of the residue within one of the hydrophobic pockets of HSA (Supporting Information Table S1). Non-covalent binding of ligands tends to occur preferentially in Sudlow site I or Sudlow site II, but can occur in both 17, 32, 33. Ligands held in these sites are brought into close proximity with the amino acids lining the pockets, thereby increasing the chances that some of them may be covalently modified 32, 33. Flucloxacillin modifies residues in sites I and II, suggesting that non-covalent binding of the drug occurs in both hydrophobic pockets (See Fig. 3 in Supporting Information). An interesting observation here is that, despite the detection of modification at each of Lys190, Lys195 and Lys199 within an individual patient, no tryptic peptides were detected with more than one of these sites modified. This is probably due to either steric hindrance by the adduct or to an alteration in the local conformation and therefore reactivity and/or availability of each Lys in the intact protein. This is an important factor when considering the hapten hypothesis of drug hypersensitivity as it is probable that antigenic peptides will comprise only a single flucloxacillin adduct.

The mechanisms underlying hypersensitivity reactions to flucloxacillin and other β-lactam antibiotics have not been elucidated, but the formation of neoantigens is believed to be a critical step in the process. Drug-modified albumin has been implicated as a possible source of neoantigens, but modification of protein can be detected in all patients studied to date: what determines why a minority will go on to develop hypersensitivity reactions is not known. With the bioanalytical tools described herein, we are now in a position to relate the site, extent and time-course of drug/metabolite-antigen formation in vivo to immunogenicity and clinical hypersensitivity in patient cohorts. This study has provided the set of candidate peptides necessary for qualitative and quantitative studies in man, and revealed flucloxacillin-modified albumin as a marker for bioactivation of the drug.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

The authors would like to thank the Medical Research Council, the National Institute of Health Research (NIHR – Department of Health) and the Northwest Development Agency (NWDA) for infrastructural and project support. We would also like to thank Anita Hansen and Karen Hawkins for sample collection. V.L.E. is sponsored by Pfizer. X.M. is supported by the NIHR Biomedical Research Centre in Microbial Diseases.

The authors have declared no conflict of interest.

5 References

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and methods
  5. 3 Results
  6. 4 Discussion
  7. Acknowledgements
  8. 5 References
  9. Supporting Information

Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.

FilenameFormatSizeDescription
prca_200800222_sm_SupplInfo.pdf56KSupplInfo

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.