«THE BINDING OF ANTIBIOTICS TO SERUM PROTEINS BY G. N. ROLINSON AND R. SUTHERLAND From the Beecham Research Laboratories, Research Division, Brockham ...»
Brit. J. Pharmacol. (1965), 25, 638-650.
THE BINDING OF ANTIBIOTICS TO SERUM PROTEINS
G. N. ROLINSON AND R. SUTHERLAND
From the Beecham Research Laboratories, Research Division, Brockham Park, Betchworth, Surrey
(Received February 11, 1965)
The published work on the binding of antibiotics to serum proteins is extensive.
Nevertheless, data are still incomplete in certain respects. For example, there is little information on the rate at which antibiotics are bound to the proteins and the rate at which this complex breaks down. Similarly, although the binding of antibiotics to serum proteins is known to be essentially reversible it is not certain whether there is any proportion of the drug which is bound irreversibly. On certain aspects, published data are available but are conflicting and some lack of agreement exists on the precise extent of binding of certain antibiotics, particularly of tetracyclines. There is also a general lack of comparative results on the extent of binding for related antibiotics obtained from experiments carried out by the same method.
In the present paper, results are given for the effect of certain factors on the binding of penicillins and other antibiotics in serum and comparative data are also given for the extent of binding in human serum for all the penicillins at present in clinical use.
METHODSMeasurement of binding of antibiotics to serum proteins The extent of binding of penicillins and other antibiotics to the proteins of serum was measured by ultrafiltration through Visking viscose-cellulose dialysis tubing. Preliminary experiments indicated that there was no significant difference between the extent of binding in plasma obtained from heparinized blood and that in serum. The antibiotic present in the protein-free ultrafiltrate was measured by microbiological assay and this quantity represented the free, unbound, fraction of antibiotic in serum. The amount of antibiotic bound to protein was derived by subtracting the level of free antibiotic from the known total concentration in serum. Before comparative experiments were made to measure the extent to which different antibiotics were bound to the proteins of human serum, preliminary investigations were carried out into the effects of various experimental factors thought likely to influence the extent of binding. These included the effects of antibiotic concentration, nature of protein, temperature, individual variation, rate of binding and nature of binding.
Ultrafiltraton techniques Suitable lengths of Visking tubing (0.25 in. internal diameter) were knotted at one end and attached at the other to a manifold connected to compressed air. Serum samples were introduced into the tubing before connexion to the manifold and ultrafiltration was then carried out at room temperature at a positive pressure of 15 lb/in2. The ultrafiltrate was collected in a glass tube which closely surrounded the Visking sac. In most experiments the volume of serum used was 5 ml., and the volume of ultrafiltrate collected for assay was about 0.5 ml. In some experiments a volume of 10 to 12 ml. of serum was used which permitted an adequate sample of ultrafiltrate to be collected in about 7 min.
ANTIBIOTICS AND SERUM PROTEINSHorse, sheep, rabbit, and calf sera were obtained as commercial samples from Burroughs Wellcome. Human serum was obtained from healthy volunteers and pooled before use. The pH of the fresh human serum was approximately 7.4, but on standing in the refrigerator for a few days the pH rose to as high as 8.0 or above. Because of this change in pH, in all the experiments the pH of the serum was adjusted to 7.4 with carbon dioxide before use, although in a limited number of experiments with benzylpenicillin and oloxacillin the extent of binding did not appear to be greatly influenced by changes in pH over the range 7.4 to 8.2.
With each antibiotic it was confirmed that there was free passage of the compound through the membrane without selective filtration when aqueous solutions were placed inside the Visking tube and air pressure was applied. It was also shown that the ultrafiltrate from normal serum was devoid of protein. There was no evidence that blocking of the membrane took place during filtration, and in repeat experiments with different penicillins consecutive samples of ultrafiltrate showed no significant difference in assay results. In contrast, with the tetracyclines there was some evidence that the concentration in the first sample of ultrafiltrate was erroneously low and a subsequent sample was routinely used for assay. This phenomenon with tetracycline has already been reported by Remington & Finland (1962).
Measurement of free and total antibiotic The antibiotic content of samples of serum or serum ultrafiltrate was determined by microbiological assay. Rectangular glass plates (12 x 15 in.) were poured with a layer of nutrient agar (Oxoid No. 2) inoculated with a suspension of a suitable assay organism. For assay of penicillins and cephalothin, Bacillus subtilis (N.C.T.C. 8236) or Sarcina lutes (N.C.T.C. 8340) was employed. The tetracyclines were assayed with Bacillus cereus (N.C.TC 8035), and novobiocin with the Oxford strain of Staphylococcus aureus (N.C.T.C. 6571). Holes (7 mm) were punched in the seeded agar with a No. 4 corkborer, and these were filled with the test samples or appropriate standard solutions.
For estimation of free concentrations of penicillins in ultrafiltrate samples, standard solutions were prepared in 0.05 M-phosphate buffer (pH 7.4), preliminary experiments having shown that standard penicillin solutions in phosphate buffer yielded results identical to equivalent standard solutions prepared in ultrafiltrate from normal human serum. With the tetracycline antibiotics, however, preliminary experiments showed that standard solutions in buffer produced inhibition zones which were significantly larger than those caused by the equivalent solutions prepared in serum ultrafiltrate.
Accordingly, with this group of antibiotics, and with novobiocin and fusidic acid, standard solutions were prepared in normal serum ultrafiltrate. Total antibiotic concentrations in serum were measured against standard solutions prepared in normal human serum, pH 7.4. Inhibition zone diameters were measured after incubation overnight at 30' C and the concentrations of antibiotic were estimated by reading from a graph derived from similar assays of standard solutions.
Determination of minimum inhibitory concentrations Minimum inhibitory concentrations required to prevent growth of bacteria for 24 hr at 370 C were determined by serial dilution of the antibiotic in 1 ml. volumes of nutrient broth (Oxoid No. 2).
In addition, minimum inhibitory concentrations were also determined in serum in the same fashion by serial dilution in 1 ml. volumes of normal human serum, pH 7.4. For the determinations of minimum inhibitory concentrations in serum, aqueous solutions of the antibiotic were added to give the desired range of antibiotic concentrations with a final serum concentration of 95%. One drop of a 1/1,000 dilution of an overnight culture was added to each tube to give an inoculum of about 1O4 organisms. Visual determination of minimum inhibitory concentrations in serum was not always definite, and with the penicillins the end points were checked by subculturing from the serial dilution onto antibiotic-free agar, and observing growth after a further 24 hr incubation.
2.5 1.0 a Unbound antibiotic (%) Fig. 2. Effect of variation in the concentration of serum protein on the extent of binding of benzylpenicillin and cloxacillin. Human serum was diluted with 0.05 M-phosphate buffer (pH 7.4) to give solutions of various protein concentrations. These solutions containing benzylpenicillin (1,ug/ml.) or cloxacillin (50,ug/ml.) were filtered through Visking membranes and the proteinfree ultrafiltrates, containing the unbound penicillin of each solution, were assayed for antibiotic content.
of extent of binding were made by ultrafiltration of samples of aqueous sera containing benzylpenicillin at 2,ug/ml. and cloxacillin at 50,ug/ml. of serum. As would be expected, the extent of binding of both penicillins decreased as the concentration of protein was reduced, and the relative increase in unbound antibiotic with decreasing serum concentration was very similar for benzylpenicillin and cloxacillin. Nevertheless, with cloxacillin only about 50% of the penicillin was available as unbound antibiotic even after a ten-fold dilution of the serum.
Temperature. The percentage binding of benzylpenicillin and cloxacillin in human serum was determined by ultrafiltration at 37, 22 and 4' C. Solutions of benzylpenicillin (1 pug/ml. of serum) and cloxacillin (50 pg/ml. of serum) were adjusted to the required temperature before filtration and the experiments were carried out in temperature controlled rooms. The temperatures of the samples of ultrafiltrate were checked as they G. N. ROLINSON and R. SUTHERLAND were collected and did not differ significantly from the initial temperatures. In several experiments with benzylpenicillin and cloxacillin, variation in temperature had little effect on the extent of binding of the penicillins, and with both compounds results obtained at 370 C differed little from those obtained at 4° C.
Binding in sera of individual human subjects. Serum was obtained from each of six healthy subjects. Benzylpenicillin was added to give a concentration of 2 /Ag/ml. and the extent of binding was determined by ultrafiltration. Ultrafiltration was repeated the following day using a further aliquot of each of the samples of sera. One week later serum was again obtained from the same six subjects and the determination of binding with benzylpenicillin was carried out on two successive days as before. The results are shown in Table 2. Differences in binding capacity between one subject and another were slight and might well be less than the experimental errors inherent in the methods. Further
experiments with another group of six subjects using both benzylpenicillin and cloxacillin confirmed these results and failed to indicate marked differences in binding capacity between one person and another.
Rate of binding and dissociation of penicillin in human serum. Experiments were carried out with benzylpenicillin, phenoxymethylpenicillin, phenethicillin, methicillin, cloxacillin and ampicillin to determine the rate at which binding takes place in human serum. Ultrafiltration was carried out immediately after the addition of the penicillin to the serum and further samples were also taken at intervals of time up to 1 hr. The time taken to obtain the first sample of ultrafiltrate was about 8 min after adding the penicillin to the serum. With all the penicillins tested the assay of the initial sample did not differ significantly from any of the subsequent samples indicating that equilibrium had been reached by the time the first sample had been collected.
The following experiments were also carried out to determine the rate of breakdown of the penicillin-protein complex. Cloxacillin was dissolved in serum to give a penicillin concentration of 400 ug/ml. and a sample of ultrafiltrate was obtained. At this concentration 13% of the penicillin was present in the serum as free, unbound, cloxacillin. The
ANTIBIOTICS AND SERUM PROTEINS 643serum was then diluted fortyfold with 0.05 M-phosphate buffer (pH 7.4) to give a serum concentration of 2.5%, and samples of ultrafiltrate were taken immediately after dilution and also at intervals of time up to 1 hr. Assay of the sample taken immediately after dilution showed 49% unbound antibiotic, and assays of the subsequent samples showed that the new equilibrium had been established by the time the first sample was taken.
Extent of irreversible binding in serum. Cloxacillin was dissolved in human serum at a concentration of 400 pg/ml. and a sample of ultrafiltrate was obtained in the usual way. Under these conditions 87% of the antibiotic was bound to protein. An aliquot of the original serum solution was then diluted fortyfold with 0.05 M-phosphate buffer, pH 7.4, which resulted in dissociation of the antibiotic-protein complex and yielded a solution consisting of 2.5% serum in buffer and cloxacillin at a concentration of 10 jug/ml.
A sample of ultrafiltrate was then obtained and compared with ultrafiltrate from a solution of 10 Jug/ml. cloxacillin in 2.5% serum prepared directly by adding cloxacillin to 2.5% serum. Results, typical of those obtained when the experiment was repeated, are shown in Table 3. After dilution of the serum with buffer the concentration of unbound antibiotic was slightly lower than in the solution prepared directly at the same TABLE 3
THE EXTENT OF IRREVERSIBLE BINDING OF CLOXACILLIN IN HUMAN SERUMHuman serum containing cloxacillin (400 pg/ml.) was diluted fortyfold in 0 05 M-phosphate buffer (pH 7 4), and the unbound penicillin present was measured by ultrafiltration. The extent of binding of a solution of cloxacillin (10 ph/ml.) prepared directly in 2.5% human serum was also determined. The amount of irreversible binding of cloxacillin to serum is indicated from a comparison of the amounts of free cloxacillin in the two solutions Antibiotic Cloxacillin Serum Unbound Bound concentration concentration (NO (Y.) (pgfml.) Solution (%) Original 100 Diluted fortyfold 25 Prepared directly 10 25 40 serum concentration. Further experiments were also carried out with oxacillin at two concentrations, 400 and 20,ug/ml., before dissociation by dilution with buffer. As in the experiments with cloxacillin the concentration of unbound drug was slightly higher in the solution prepared directly in 2.5% serum than in the corresponding solution obtained by dilution of normal serum with buffer. These results suggest that some degree of irreversible binding may take place, although it appears that this is only a small proportion of the total antibiotic in normal serum.
Competitive binding between cloxacillin and other substances. The results of experiments to determine the extent of binding of cloxacillin in human serum in the presence of substances which are themselves highly bound to serum proteins are shown in Table 4.