THE interaction of an antibiotic with serum is generally A believed to result in a decrease in antibacterial activity. When the natural penicillins were originally examined, it was found that in the absence of serum, penicillin K was more active than penicillin X, and penicillin X was more active than penicillin G, but in the presence of serum, penicillin G was the most active. The reversal in activity was attributed to the fact that penicillin G was less bound to protein than penicillin K or X., and hence a greater concentration of free penicillin was available for antibacterial action1. The methods used to examine protein binding of penicillins have reHed largely on in vitro static dialysis, but such methods, however, bear little relationship to the dynamic conditions in the body. We have, therefore, studied the phenomenon of protein binding with the view of determining more precisely its significance to antibacterial activity in vivo. Gel-filtration and both static and continuous dialysis methods have been used in this investigation.The penicillins which were selected were penicillin G, 'Ampicillin' and 'BEL 1071' (triphenylacetylpenicillin). Static dialysis. To 5 ml. of human serum, 5 mg of the pencillin under examination were added. This was placed in a dialysis bag of 'Visking' tubing 0.75 in. diameter and suspended in 20 ml. normal saline and dialysed at 8 C for 24 h. Several dilutions of retentate and dialysate were prepared in M/20 phosphate buffer pH 7.0, and the amount of antibiotic was assayed by the cup-plate technique using Sarcina lutea as the test organism. The samples of each antibiotic were assayed against dilutions of known concentration of the same antibiotic.Continuous dialysis. 5 mg of each antibiotic in 5 ml. serum were placed in 'Visking' tubing as described. The bags were suspended in a beaker through which water flowed continuously. The final concentration of the antibiotic was assayed in the serum at 24 h and the percentage obtained was calculated in terms of the concentration in serum at zero and 24 h. Gel-filtration. 5 mg of each antibiotic were dissolved in 5 ml. serum and allowed to stand for 18 h at 10 C. 3 ml. of the serum were applied to a column jacketed at 10 C and containing 100 ml. 'Sephadex (225' gel equilibrated with water, the flow rate being 2025 ml./h. Fractions (5 ml.) were collected and protein elution was followed by optical density measurements at 280 mon a 'Unicam' S.P. 500. The fractions were assayed for antibiotic content. Aqueous solutions containing 1 mg/ ml. of the antibiotics were similarly treated. Protein fractionation. Further examination of the bound fraction from the 'Sephadex G25' column was carried out using DEAE-'Sephadex .450'. After concentration by ultra-filtration the bound fraction was passed down a jacketed column kept at 10 C and containing 30 ml. of fully swollen DEAE-'Sephadex 450' equilibrated with 0.02 M phosphate buffer at pH 8.6. Step wise elution was then carried out with 0.02 M phosphate buffer followed by saline of molarities 0.07, 0.12, 0.18 and 0.3 M, the change-over point being based on the optical densities of the fractions at 280 mResults(a) Dialysis. The results obtained for the degree of protein binding are shown in Table 1. By static methods 'BRL 1071' is bound 80.90 per cent to the serum proteins; penicillin G is much less bound, while 'Ampicillin' is only slightly bound. When examined by continuous dialysis, the final percentages bound were reduced but the reduction in the binding of 1BEL 1071' was less than with the other penicillins (Table 1). Table 1Bound Penicillin Continuous StaticPenicillin G 14-2 (5) 46-4 (21) 'BRL 1071' 41-4 (6) 91-6 (3)'Ampicillin' 0-9 (3) 10-2 (6) (b) 'Sephadex.' Typical elution patterns from three penicillins of different binding properties obtained on 'Sephadex' columns are shown in Fig. 1. With the penicillin alone the antibiotic activity is confined to fractions 8.10. With serum alone the protein as determined by optical density at 280 mm emerged in fractions 4.6. When the incubated penicillin serum mixture was passed down the 'Sephadex' column the protein again emerged in fractions 4.6, whereas the antibiotic activity showed two peaks, if there was significant binding. One peak was associated with the protein and the other again emerged in the protein-free fractions 8, 9 and 10. The degree of protein binding was then calculated from the antibiotic content of the appropriate fractions. The 'Sephadex' method2 offers considerable advantages over dialysis methods since the fractionation occurs in less than 2 h, and after a short wash with water the column is ready for further use. The dilution factor of both the bound and unbound material is small, and this aids the sensitivity of the method. The reduction in time needed for the separation of bound and unbound compound is also an advantage when dealing with labile materials.Fig. 1. Elution patterns from 'Sephadex' columns 'Sephadex G255 grade has a lower limit for complete exclusion of molecular weights of 3,5004,500 and is the preferred grade for protein binding studies. With the 'G50 and 'G75 grades of different degrees of cross-linkage, the protein peak is flattened and tends to tail into the unbound antibiotic peak.(c) Protein fractionation. The bound fraction from the 'Sephadex G25' column, on further examination on DEAE-'Sephadex A50', gave the bulk of the antibiotic activity in the 0.18 M saline fractions, whereas aqueous penicillin solutions gave a peak of antibiotic activity in the 0.07 M saline fractions with some tailing into the 0.12 M saline fractions. Starch-gel and cellulose acetate electro-phoresis was then carried out on the protpin fractions. The material eluted. with 0.02 M phosphate 'Vasiiounxl to consist of pure g-giobulin. The other fractions were all multi-component, but the 0.025 M, 0.07 ivlrand, 0.12 M fractions contained no albumin and were made up of mixtures of a- arid b-globulins. The material Muted'with 0.18 M saline consisted almost entirely of albumin whereas the material eluted with 0.03 M saline contained, some material of the same mobility as albumin-together with high electrophoretic 'pre'-albumin content. It is evidence from this method that the bound penicillin is associated with the albumin fraction of the serum protein. DiscussionFrom the results of the three methods described, it is evident that penicillins can differ in the degree and nature of their binding to serum proteins. Static dialysis shows that 'Ampicillin' is slightly bound, penicillin O is moderately bound while BRL 1071' is considerably bound. The nature of the binding of BRL 1071 and penicillin 0 is evident from the results under the different conditions of continuous dialysis and gel-filtration. The binding of penicillin O is shown to be of a very loose nature and the binding of 'BEL 1071' to be very firm. Such differences in binding can affect activity in vivo. If a compound is firmly bound to the serum proteins the concentration of free penicillin is reduced and hence its intrinsic activity is reduced. The in vivo activity, however, may riot be significantly influenced if the binding is of such a loose character that the penicillin is readily available for antibacterial activity. This is the case with penicillin G which is bound to the extent of 46 per cent to serum protein yet its activity is apparently uninfluenced by the presence of serum in the same way as a compound, for example, 'Ampicillin' which is only slightly bound. On the other hand, 'BRL 1071' is very firmly bound and its activity is much reduced in the presence of serum in vitro and in vivo. Therefore, before conclusions on in vivo activity can be drawn from in vitro results, information'should be available on both the actual degree and nature of the binding to serum proteins.