CHEMICAL STUDIES ON BACTERIAL AGGLUTINATION III. A REACTION ME-W AND A QUANTXTATMC THEmu* BY MICHAEL HEIDELBERGER, PH.D., AND ELVIN A. KABAT (Fmm tke Dc#xz~hnent of Medicine, Colkge of Physicians and Surgeons, Columbia Uniuwsity, ad tko Ptwbyti~ Eo#al, N+w York) (Received for publication, March 2, 1931) Recent and even current articles on specific bacterial agglutination reveal wide differences of opinion regarding the mechanism of the re- action and the factors influencing its course. Since much of this divergence appeared traceable to the essentially qualitative and rela- tive nature of the methods used for the estimation of agglutinin titer, it was decided to attempt the assembly of more precise data on agglu- tination than had hitherto been available. As a first step an absolute, quantitative method was developed for the micro estimation of certain agglutinins (l), in which these are measured as the amount of antibody nitrogen actually combined with a given volume of bacterial suspen- sion and in which an accuracy of 0.01 to 0.02 mg. of nitrogen is easily attainable. With this method and with the absolute, quantitative methods previously developed for the estimation of precipitins (24, it was possible to demonstrate the identity of agglutinin and precipitin in absorbed antisera containing only antibody to the type specific polysaecharide of Type I pneumococcus (5). It has now been possible to extend the analogy and to account quantitatively for this instance of specific bacterial agglutination by a purely chemical theory, much as had been accomplished in the case of the precipitin reaction (6). EWERIAXENTAL H&t-Lilled suspensions of Type I S (Dawson M) pneumococci we= prepared as described in (1). It was found that on standing for some time appr&able amounts of polysaccharide appeared in the supernatants of even carefully washed * The work reported in this communication was carried out under the Harkuess Research Fund of the Presbyterian Hospital. A preWnary note has been pub- lished (Pm. Sot. Exp. Biol. and Med., 1936, I, 301). 885 886 BACTEBJAL AGGLEMNATTON. III suspensions, just as Craigie and Wishsrt (7) found antigen to leach out slowly from washed suspensions of the elementary bodies of vaccinia virus. In order to avoid confusion of the agghrtinii reaction under invest&ion with a simul- taneously occurring precipitin reaction, the suspensions were freshly centrifuged and washed once with saline before use. After thii treatment only insigniit' traces of specific polysaccharide remained in the supernatant. Type I antipneumococcus horse sera H 701, H 702, and a Felton antibody solu- tion B 78 were completely freed from group specific antibody by repeated absorp- tion with I R (Dawson S) pneumococci as described in (5). Horse serum H 79 was absorbed with C substance (8) and protein prepared from a Type I R strain. Felton antibody solution B 75 was described in (1). The pH of antibody solution B 78 was 6.96, that of serum 701 was 7.34. Estimation of the amount of S 1' in the pneumococci was carried out as in (5) by dissolving a measured volume of the suspension in alkali, neutralizmg after 72 hours at 37"C., and making up to a volume calculated to yieiclO.9 per cent salt TABLE I Eject of Volume and Final Cdrdion of Antibody OB the Atmu& of Antibody N in Aggluthakd Pncnmowcci d. au. grad. 4.0 0% 0.15 8.0 0.34 0.074 12.0 0.33 0.050 concentration. Aliquot portions of this solution were analyzed according to (4) by means of an antibody solution calibrated with alkali-treated S I. Agglutinln determinations were carried out by addition of accurately measured volumes of washed bacterial suspension to measured vohunes of serum or antibody solution and estimation of the increase in nitrogen content. Heat-hilled sus- pensions of Pn I were used. All determinations were run in duplicate unless otherwise stated and tubes were covered with tight fitting rubber caps. Data on the effect of volume on the agglutination reaction in the region of excess antigen were included in (1). Additional data in the region of excess antibody are given in Table I. It will be seen that the amount of antibody nitrogen taken out by the bacteria after 48 hours at O", with occasionel stirring, is independent of the volume just as in the precipitin reaction (6). Thus the agglutinii nitrogen 1 This designation is used for the specific polysaccharide of Type I pneumo- coccus. Pneumococcus I suspensions are referred to as Pn I. MIC!HAEL EIEIDELBEBGEB AND ELVIN A. KABAT 887 removed is independent of the concentration of antibody nitrogen in the super- natant. In Table II are given data in the region of excess antibody on the effect of vary- ing lengths of exposure of the Pn I suspensions to antibody. Suspension and serum were mixed in tubes and allowed to stand with occasional stirring. In- creasing amounts of antibody nitrogen were taken up with increasing lengths of time. The extremely alow rate under these conditions at 0" suggested the possibiity of measuring the velocity of agglutiuation. 15 ml. of Pn I and 3 ml. of antibody 37' IS" 0' 3s. b?r. mc. m. SC. 1 oY6+ 2 0.37 6 0.53 19 0.27 0.36 0.44 24 0.58 43 0.39 48 0.44 0.58 72 &YS 5 0.43 10 0.52 12 0.34 * Control tubes containing serum alone tended to gel at this temperature unless diluted with saline. t One determination. TABLE II f Standing OS Antibody N Taken up by Pn I at Various Tcmpmcrtawes = I serum H 19 solution B 78 were added to 32 ml. of saline and the mixture was stirred mechan- ically in an ice bath. The temperature mmained constant at +O.S'C. 5 ml. samples were withdrawn at varying intervals and pipetted into 5 ml. of cold saline and centrifuged at about 3OMl P.P.X. in a refrigerated antrifuge.a The precipi- tates were washed and analyzed in the usual manner. It will be seen from Table III that with eficient stirring the reaction is au extremely rapid one, beiig more than 80 per cent complete in 5 minutes. hkimum combination of antibody requires a considerably longer period. Even when this is attained the aggregates * Manufactured by the International Equipment Company, Boston. 888 BACTERIAL AGGLDTINATION. III formed are much smaller than in the incomplete reaction under the usual con- ditions. In accordance with the above erperiments it was found possible to use the usual arrangement of tubes and take out the maximum antibody nitrogen for any given set of conditions if the mixtures were shaken on a shaking machine oscillat- ing with sufficient rapidity to keep the Pn I in suspension. A ma&ii of the type' adopted for streptolysin determinations proved adequate. Shaking at 370 was carried out in an incubator, and at 0" the tube racks were placed in a box, packed with crushed ice, and shaken, the ice being replenished as often as neces- sary. Table IV shows the result of shaking for varying lengths of time. Values in parentheses were obtained by use of a mechanical stirrer instead of shaking. It will be San that within the limit of error of the determinations stirring and shaking yielded identical results. Since the dilution in the stirred mixtures was more than two and one-half times as great, this cenSrms the &ding (Table I) that antibody N removed is independent of the concentration. Two diierent lots TABLE III V&city of Gwnbindh of Antibody and Pn I Thne min. 5 15 2s 40 Antibody N remod oY.2 0.14 0.14 0.14 Time min. 60 90 120 150 Antibody N removed 0:s 0.16 0.16 0.16 Samples taken contained 0.25 mg. bacterial N. of Pn I were used for the 370 and 0" experiments with B 78, but for sera H 701 and H 702 the same suspension was used for all determinations and results at diikent temperatures and salt concentrations are strictly comparable. It will be observed that in the region of excess antibody much more agglutiuii N was taken out by the same amount of suspension at 37' than at 0". This will be discussed below. After determination of the time necessary to reach equilibrium, a series of tubes containing 1.0 ml. of serum or antibody solution and increasing amounts of freshly washed Pn I was set up in duplicate and shaken at the desired temperature until equilibrimn was reached. The tubes were then centrifuged at the same temperature and the precipitates washed twice in the cold (1) and analyzed for nitrogen. Agglutinin N was found by subtracting the bacterial N. At the same time an accurately measured portion of Pn I was dissolved in alkali and the S content determined as described above. In the experiments in which the tinal salt concentration was to be 2 Y, a volume of 3.8.5 M salt equal to the volume of serum and suspension used was first added. Table V shows the results obtained Tim0 hn. 0.5 1 1.5 2 3 4 5 t.5 BliCt.&ll NWUl TABLE IV Antibody N Renwwd from Anti~~cus I Solution and Sara by Pa I on Shakhg for Varying Per&is = I BT8 I H 101 D.9 pa cent N&l 1 0.9 pa olt (0.15 u) N&l T 37' 0' I -I- rl. w. 0.22 0.22 0.23* 0.23 0.23 0.26+ 0.25 0.23 31. ml. 0.31 0.35 0.35 (0.32) 0.35 1 0.34 1 0.12 0' w. 0.13 0.14 0.14 (0,13) 0.09 2YNdl 37' w. 0.24 0.24. 0.23 0.23 (0.22) w. ry. 0.15 0.15 0.16 0.16*(0.15) 0.27 0.28 0.29t (0.25) 0.11 0.09i 0.16 0. H 702 I5 0.9 pa cCnt (0.15 10 Ndl 1 2xNaCl 379 0' I 37s -r- w. 0.16 0.25 0.17 0.25 0.17 (0.21) 0.25 (0.23) 0.16 0.16 -_ I 0. w. P Ii !z 0.17 ii 0.19 5 0.20 (0.19) p ii! 0.16 ? Values in purnthcscs obtained with mechanical stirring instead of shaking. Aliquot portions analyzed in triplicate. * One determination only. E t A duplicate set of tubes, after 6 hours at 3P, was shaken 8 hours at o", yielding 0.32 mg. of antibody N. 5 # Bacterial N of the suspension in 0.9 per cent NaCl wss used, owing to the di&ulty of centrifuging non-agglutinated PnIin2uNaCL Q TABLE V Addition of Inceashg Amom& of Pnemwcoccus I S (M) SuspcrrtMn to 1 Ml. of Serum or Antibody Solution d Various TemperaWes and Salt C~n~rafh 00 25 , I BacteridN d%lent "n s SoDtMt w. 0.122 0.244 0.366 0.488 0.732 0.976 1,220 1.464 sexllm# aalt 0.064 0.096 0.127 0.191 0.255 0.382 0.509 0.637 suum,sall ry. 0.017 0.034 0.051 0.068 0.102 0.136 0.170 0.204 0.0165 0.025 0.033 0.0495 0.066 0.099 0.132 0.165 SC. i i i mK. w- Antibody !hlution B 78,O.U Y Salt, at 37" 0.255 0.17 7.1* 0.494 0.24 7.1 0.24 0.718 0.34 6.7 0.34 0.916 0.41 6.0 0.42 1.302 0.56 5.5 0.54 1.576t 0.59 4.3 0.59 1.840 0.61 3.6 2.076 0.60 2.9' 0.014 mg. antibody N pptd. - 8.0 S - 26.9 S' SlIliU. - 0.149 N mar. - 0.594 c&d. 0.61 found !3enuu H:tOl, 0.15 Y Salt, at 37' 0.254 0.18 10.9' 0.369 0.27 10.8 0.28 0.476 0.34 10.3 0.36 0.686 0.49 9.9 0.49 0.868 0.61 9.2 0.59 1.060 0.67 6.8 0.72 1.202 0.69 5.2 1.340 0.70 4.2+ 0.008 mg. antiiy N pptd. - 12.5 S - 53.2 Y S max. = 0.1175 N max. p 0.734 cakd. 0.70 found rr. B 78,0.15 Y Salt, at 0" 0.21q 0.07 4.1. 0.454 0.17 5.0. 0.642 0.24 4.7' $ 0.852 0.33 4.9 0.33 3 1.216 0.45 4.4 0.46 1.574 0.56 4.1 0.56 E 1.880 0.62 3.6 0.63 2.188 0.69 3.4 0.68 0.036 mg. a&body N pptd. = 5.7 S - 11.6 9 Fj Smax. -0.2% NIMI. - 0.7Ocalcd. 4 L. 0.187 0.290 0.376 0.534 0.722 1.030 1.244 1.414 0.000 tug. anti 0.69 found H 701,0.15 Y !hlt., at 0" 0.12 7.3' 0.19 7.6. 0.25 7.6* 0.34 6.9+ 0.47 7.1 0.47 0.65 6.6 0.63 0.74 5.6 0.74 0.78 4.7 0.79 xly N pptd. = 8.8 S - 24.5 S' s max. - 0.180 N max. = 0.79 c&d. 0.78 found 0.064 0.0165 0.096 0.025 0.127 0.033 0.191 0.0495 0.255 0.066 0.382 0.099 0.509 0.132 0.637 0.165 serum, salt 0.000 H 701,2 Y salt, at 37' 0.222 0.16 9.7 0.290 0.19 7.6 0.354 0.23 7.0 0.434 0.26 5.3 0.540 0.29 4.4 0.688 0.31 3.1. 0.828 0.32 2.4+ 0.980 0.34 2.1* 0.15 0.20 0.24 0.28 mg. antiiy N pptd. = 10.6 S - 100 S' S max. - 0.053 N max. = 0.28 calcd. 0.34 found H 701,2 Y Salt, at 0' 0.208 0.14 8.5* 0.304 0.21 8.4 0.21 E 0.394 0.27 8.2 0.26 0.534 0.34 6.9 0.34 0.638 0.38 5.8 0.39 E 0.7% 0.41 4.1 0.952 0.44 3.32 1.120 0.48 2.9. i O.ooO P mg. antibody N pptd. = 9.9 S - 59.8 S' 8 Smax. -0.083 N&u. - 0.41calcd. Ef 0.48 found E For serum H 702 the equations were: 0.15 Y saltz At 3f0, mg. antibody N pptd. - 6.9 S - 16.5 S' Smax. = 0.209 Nmax. - 0.72 c&d. 0.74 found 2 x salt: At 37", mg. antbody N pptd. - 7.7 S - 44.8 S' SU. = 0.086 N max. - 0.33 c&d. 0.40 found * Points not considered in calculating equation. t One detemhation discarded. $ Only two point3 not at maxinwm ratio available for calculation. At O', mg. nntihdy N pptd. = 5.8 S - 11 fi S max. - 0.263 N max. = 0.76 calcd. zi 0.83 found : At 0', mg. antibody N pptd. = 6.3 S - 20.7 S' S max. = 0.152 N mar. = 0.48 calcd. 24 0.53f0ulld ?- E 5 892 BACTERIAL AGGLUmATION. III in these experiments. The ratios given were calculated from the determinations of antibody N in the agglutinated Pn I and the S I content of the Pn I suspension used, although it is by no means certain that all of the S I in the suspension is capable of reacting. If it is assumed that the same proportion of the S I reacts at a given temperature throughout the entire reaction range, the observed N:S ratios would then be divided by the same factor and the character of the results would be unchanged. It will be seen from Table V that the ratios reach an upper limit with decreasing amounts of suspension. This was also determined by the 8 0.80 $ iif o.soq 0.40 z 0.30 i .rJD' s 0.20 0.10 0.03 QiO a15 0.20 Hg. S added FIG. 1 addition of a constant amount of Pn I to increasing quantities of serum H 79 at 3F. VolunleofseNm,mL. .,.. 0.5 1.0 1.5 2.0 2.5 3.0 AntibodyNremoved,mg.. 0.47 0.53 0.60 0.63 0.62 0.63 The amount of antibody N removed by 0.26 mg. of bacterial N reached a definite limit as the volume of serum added wss increased. In the region in which a change of ratio is observed the equation N- 2RS-;Ss' MICHAEL HEIDELBEBGEB AND ELVIN A. KABAT a93 derived for tbe precipitin reaction (6) was found to hold. In Table V are given the equations caicuiated from tbe data at each temperature and salt concentra- tion. The antibody N values found ars also compared with those calculated from the above equation. Fig. 1 shows tbe curves obtained in the csse of antibody solution B 78 at 0" (curve A) and at 37' (curve B) in 0.9 per cent saline by plotting antibody N in the agglutinated Pn I against the polysaccbaride content of the Pn I. Limes satisfying the equation N - 2R - It* s AS were obtained by plotting against S I the ratios of agglutinin N taken out to S I in the Pn I used (Fig. 1, lines A and B). Tbe method of least squares was used in TABLE VI COnrparis~ of Matinrum CM and Found Antibody N :S I Ratios 701, 0.15 x salt, 37" `I " `I `I 00 " 2 " `I 37" `I 1` ,` 6` 00 B78,0.15 I` " 37" " " `I " 0' 702, 0.15 `I " 37" L` `I " `, 0" " 2 " " 37" " " " `I 0" "U&-Q&S I . 12.5 10.0 8.8 7.4* 10.6 9.7 9.9 8.5 8.0 7.1 5.7 4.6. 6.9 5.4* 5.8 4.2* 7.7 6.7 6.3 5.6* r2dcubtc%lntio Foundntia 1.25 1.19 1.09 1.17 1.13 1.24 1.28 1.38 1.15 1.12 *Meanof maximum ratios determined experimentaily. every case to calculate the line best fitting the data. Points in the region of excess antigen, 85 well as those beyond the first point with maximum ratio, were not used in the calculations. These poi~& are marked with asterisks in Table V and Fig. 1. The data obtained in 2 Y salt solution are aiso in agreement with corre spending results for the precipitin reaction (9). Similar data were obtained with serum H 702 in 0.15 and 2 Y salt solution at both temperatures. A comparison of the &ximum calculated and found ratios of antibody N to S is, made in Table VI. It will be seen that the theoretical maximum ratio, 2 R, (LG., the intercept of tbe line on tbe y axis) is approximately equal to 1.2 times tbe experimentally determined maximum ratio. It will be noted from Table V and Fig. 1 that throughout the region of antibody excess a given amount of Pn I removes more antibody N at 37" than it does when set up at 0" with serum in the same proportions. Tbe following eapekmmt was 894 BACXEBIAL AGGLVMNATION. ID ~pa'h'med in an efiort to explaii this effect. Four tubes containing 1.0 ml. of F'n I suspension and 1.0 ml. of mum were shaken for 6 hours at 37". One pair was then shaken at 0" for 8 hours. Agglutinin N was determined in both sets of tubes. It was found that more aggbitinin N wss removed by shaking at 37" and then at 0" than was taken out at 37" alone (see footnote, Table IV). The diier- ence in N removed at 37' and at 0" is very markedly reduced in 2 Y salt. As in the precipitin reaction it wss not found possible to take out at 0" antibody remaining after the maximum antibody had been taken out at 37O with excess Pn I. In the region of excess antibody, the agglutinated bacteria are very easily and uniformly resuspended in saline. In tbe region of excess polysaccharide, as in the precipitin reaction, large clumps are formed which cannot be easily dispersed. To determine whether or not Pn I, aggIutinated in the region of excess antibody, is capable of combining with more Pn I, the following experiment was set up. To each of a series of tubes were added 0.5 ml. of Pn I S suspension (0.19 mg. N) and 1.5 ml. of serum, a large excess, The agglutinated bacteria were allowed to stand in tbe ice box for 48 hours and were centrifuged and washed twice with dine. The second washing was set up with 0.25 ml. of fresh Pn I sus- pension and failed to show agglutination, indicating that all uncombined antibody bad been washed out. Tbe original agglutinated (sensitized) Pn I was then resuspended uniformly in saline, and 0.25 ml. portions of various pneumococcus suspensions or measured amounts of specific polysaccharide (S I) were added. Tbe results are shown in Table VII. Eagle, Smith, and Vickers (10) have found that the partial coupling of pneumococcus Type I antibody with diazotized sulfanilic acid prevented the antibody from precipitating S I although it still gave definite agglutination with Pn I. It was concluded that S I failed to combine with the azo antibody since addition of untreated Type I antipneumococcus serum gave a precipitate. However, agglutination would also fail to occur if the polysaccharide no longer combined with antibody, since it has been shown that Type I a&carbohydrate precipitin and agglutinin are identical (5). If, on the other hand, soluble compounds of S I and azo antibody were formed, such com- pounds should be as readily capable of reacting with unaltered anti- body to form precipitates as is uncombined S I. It has long been known that polysaccharide and antibody can combine in proportions varying over a wide range (11). Eagle, Smith, and Vickers' experi- ment was accordingly repeated with additional controls. MICHAFiL HEIDELBEBGEB AND ELVIN A. KABAT 895 6 ml. of Type I antipneumococcus horse serum were coupled with 6 ml. of diazotized sulfanilic acid ae described in (10). 2 ml. of the resulting azo antibody, freed from excess diazosulfanilic acid, failed to precipitate 0.05 mg. of S I or its deacetylated degradation product, but agglutinated 2 ml. of Pn I suspension (0.2 mg. N per ml.). 1 ml. of Type I antiserum was added to the tubes containing S I, resulting in immediate precipitation. As an additional control, an egg albumin-anti-egg albumin precipitate was formed in 2 ml. of the azo antibody solution by addition of 0.34 mg. of egg albumin aud 0.5 ml. of a potent anti-egg albumin serum. All tubes were centrifuged and washed in the cold with aaline until the washings were colorless. The egg albumin-anti-egg albumin precipitate was white, whiie the precipitates in the other tubes were definitely yellow. S I had therefore combined specitkally with the azo antibody, not only on the surface of Pn I, where agglutination occurred, but in solution aa well DISCUSSION The mechanism of specific bacterial agglutination (reviewed in (12,13,14)) has been considered either as an adsorption process or as a precipitin reaction taking place at the cell surface (15). Although adsorption is now generally considered in its chemical aspects, Iva- no&s (16) as late as 1935 chose to interpret bacterial agglutination as an example of physical adsorption. This conclusion is open to criticism on account of the large error of the experimental method used, the failure to eliminate simultaneously occurring precipitin reactions, and the complexity of the antigen-antibody system studied. From the data given in Table I and reference 1 it is evident that a simple treatment according to classical chemical laws is as inadequate in accounting quantitatively for specific bacteriaI agglutination as is the Freundlich adsorption isotherm, since both depend on the concen- trations of the reactants at equilibrium. As in the precipitin reaction (6) the composition of the agglutinated mass depends rather on the proportions in which the components are mixed (Table V). In order that figures such as those given in Table V should be of significance it was necessary first to find the conditions under which the maximum amount of antibody nitrogen would be taken out by Pn I throughout the reaction range. It was observed that removal of antibody was too slow when tubes were mixed only at intervals (Table II), although this procedure was satisfactory in the precipitin reaction. This was due not to a slower rate of combination of polysaccharide and antibody in agglutination, but more probably to the fewer collisions between Pn I and antibody molecules. Mechanical stirring (Table 896 BACTERIAL AGGLUTINATION. III III) resulted in rapid attainment of maximum combination, as did also the use of a shaking machine (Table IV). A technique was thus available for a study of the agglutination reaction similar to that made of the precipitin reaction. The data so obtained (Table V) indicate that equations of the same form as were valid for the precipitin reaction (6) serve also to describe quantitatively the agglutination reaction between Pn I and homol- ogous type-specific anticarbohydrate. In the derivation of these equations it was assumed that S (polysaccharide) and A (antibody) Crst combine in a bimolecular reaction to yield AS, followed by the competing bimolecular reactions (in the region of excess antibody), AS+ A-+AS.A AS+AS+AS*AS and that reactions of this type continue with increasing complexity until aggregates are formed which are large enough to separate from solution. In the derivation, also, it was found that volume factors cancelled, so that the end-result was independent of the final concen- trations of the components. Since this condition is also shown experimentally to obtain in the agglutination reaction studied (Table I), and since a similar type of equation is found to apply to this reac- tion (Table V) as well as to specific precipitation, it would appear justitiable to consider this instance of specific bacterial agglutination as a precipitin reaction at the bacterial surface,8 the more so as the same substances enter into both reactions (5). According to this mechanism the mass of agglutinated Pn I would consist of pneumococci held together by combination of S I on the pneumococcus surface with antibody molecules. This three-dimen- sional process might be represented two-dimensionally as follows: . * -A+n$A- * .-(Pi, ' A-(PnS)-* . . *aA-(P&)-A-- . . Tti discussion "bacterial surface" is de&d as the portion of the Pn I cell to which reactive S I molecules are attached. Thii would not necessarily~ coincide with the external surface. MKHAEL HEIDELBEBGER AND ELVIN A. KABAT 897 in which (PnS) represents pneumococcus with an unknown number of molecules of S at the reactive surface or surfaces. A somewhat similar qualitative representation of agglutination has been given by Marrack (14). Agglutination of Pn I difIers in at least three respects from the precipitin reaction. It will be noted from Table V that the antibody N : S I ratios reach a m&mum in the region of excess antibody, beyond which more antibody cannot be forced into reaction. The value of 2 R given by the equation is usually about 1.2 times the experimentally determined maximum, as shown in Table VI. A possible explanation would be that antibody molecules, when present in large excess, are prevented by spatial exigencies from combining with all of the im- munologically reactive S groupings available. It appears reasonable to assume that this crowding would occur at lower N: S I ratios at the Pn I surface than in the precipitin reaction, in which S I is in solution. Another dXerence from the precipitin reaction is in the apparently lower combining ratios of the components. Since it is not known whether or not all of the S found in Pn I is available for reaction, as is assumed in calculating the ratios given, these values have only a relative significance. If, under a given set of conditions, the same fraction of S I reacts throughout the whole range, the entire set of ratios could be multiplied by a factor inversely proportional to the fraction reacting. A third difference is that throughout the antibody excess range more antibody is taken out at 37" than at 0" (Table V and Fig. 1). An even larger amount of antibody N is removed by the same quan- tity of Pn I when the reaction is carried out first at 37' and then at 0' (footnote, Table IV). It would seem that a larger proportion of the S I present is capable of reacting at 37" than at O', but when this has once entered into reaction, continuation of the process at 0' takes place much as in the precipitin reaction, with more antibody removed at the lower temperature. Possibly the effect noted is due to swelling of the Pn I at 37O, resulting in the exposure of more of the S I content. The differences between the precipitin and agglutinin reactions are thus seen to be mainly mechanical, rather than differences in principle, and to be conditioned by the spatial limitations of a chemical reaction taking place, in agglutination, at the bacterial surface. Since con- 898 BACTERIAL AGGLUTINATION. III siderable confusion would be caused by the adoption of conclusions of Eagle, Smith, and Vickers (10) indicating a fundamental difference between agglutination and precipitation, it is necessary to mention here that these concHsions are easily shown to be based on insufficient evidence. By the repetition of their experiments with the ne,cessary controls (page 895) it has been shown that Type I pneumococcus anticarbohydrate which has been coupled with diazotized sulfanilic acid actually combines with S I, although it does not yield a precipi- tate. It is therefore not surprising, in view of the existing knowledge of the precipitin reaction and the demonstrated identity of anti- carbohydrate precipitin and agglutinin (5) that agglutination of Pn I is effected by the same axo antibody which merely combines with S I in solution to give soluble compounds. According to the new quantitative theory the entire process of specific bacterial agglutination may be most simply treated as a con- tinuous progression of competing bimolecular chemical reactions. The theory makes no distinction between the initial combination of antigen with antibody and the subsequent aggregation, the latter being considered as a continuation of the process of combination of multivalent antigen with multivalent antibody until aggregates are formed which are large enough to flocculate. While the presence of salts is necessary for flocculation, and their function is discussed in greater detail below, it is believed that the experiments recorded in Table VII indicate the essentially chemical nature of even the flocculation stage of speciCc bacterial agglutination. By interruption of the process of specific bacterial agglutination at a definite stage it was found possible to study the specificity of further particulation under constant conditions of salt concentration and to test predictions based on the theory. Pn I cells, agglutinated with a considerable excess of antiserum, were washed with saline until the supernatant contained no agglutinin. The agglutinated (sensitized) Pn I cells were then evenly resuspended in saline and divided into several portions. According to the new quantitative theory, Pn I, agglutinated in the region of excess anti- body, would still have available on the surface of the particles some of the specifically reactive groupings of the originally multivalent anti- body. These particles, then, should be able to combine with S I on MK!EAEL EEIDELBEBGER AND ELVIN A. KABAT 899 the surface of freshly added unsensitized Pn I, and reagglutination should take place to form larger aggregates. It will readily be seen from Table VII that this prediction was verified, and that the effect is specific, since it is not given by Pn II or III, or by Pn I R (Dawson S) cells under identical conditions of salt concentration and, in the TABLE VII Addition of Pnawwsoccus or S I lo Washed, Rem+ndcd, Agglutinokd Pn I S (M) Un~~sitized Pn I S @I) suspension Rapid reagglutination into large clumps, clear supematant `I Pn I R suspeusion No visible reaction Pn I S &xmatant* " `I I` u*tised Pn III s (M) suspension `I " " Pn II S (M) suspension LI " " t 1 rug. SI(Pn I specific polysaccbcuide) " `I " 0.10 mg. s I Reagglutination$ 0.01 mg. s I `1 # 0.001 .plg. s I Partial reagglutination o.ooo1 mg. s I " `8 0.10 In& s II No visible reaction 0.9 per cent saline " `I `8 Reagglutination failed to take place if sensitized Pn I and fresh Pn I were washed with 5 per cent glucose until supematants no longer reacted for Cl-. Addition df N&l caused immediate flocking. The function of electrolyte is discusd in the text. * Freshly washed, unsensitized Pn I S suspension was again centrifugedand 0.25 ml. supermatant was added to determine whether effect was due to any S I which might he dissolved from bacteria. t Pn I S suspension was next added. Immediate flocculation took place. The large clumps formed settled rapidly. Type II serum added to the turbid su- pematant caused immediate agglutination. $ The contents of these tubes formed large clumps, resembling those with Pn I suspension. case of Pn II and III, presumably similar conditions of potential and cohesive force. Reagglutination is, moreover, produced almost as completely by suitable amounts of S ,I in solution,' so that the conclu- sion seems inescapable that particulation, as well as the original anti- 4 And not by S II, the specific polysaccharide of Pn II. 900 BACTERIAL AGGLUTINATION. III gen-antibody combination, is a chemical process. The quantitative data given indicate that the particulation is a result of the original bimolecular combination as continued in a series of competing bi- molecular reactions, much as in the precipitin reaction (6). It is also shown in Table VJI that most of the Pn II added fails to participate in the reagglutination when Pn I is subsequently added, but may be agglutinated by pouring off the turbid supernatant and adding Type II antiserum. A similar separate agglutination of mixed microorgan- isms was observed by Topley, Wilson, and Duncan (17), who also concluded that the particulation phase of agglutination was strictly chemical, in confirmation of Marrack's views (14). The results of all of these experiments, in accord with the new quan- titative theory, indicate the primary importance of the chemical inter- action of multivalent antigen with multivalent antibody in completing, as well as initiating, the process which the bacteriologist calls specific bacterial agglutination. If this be true, specific bacterial agglutina- tion differs fundamentally from other instances of agglomeration and agglutination of suspended particles, and the analogy with these, so often cited, does not apply. The function of the electrolyte in specific bacterial agglutination would seem to be the secondary one of pro- viding ions for the ionized salt complexes in which form antibody probably reacts (18), and in addition, of miniiing electrostatic effects due to the presence of many ionized groupings on the particles, effects which might interfere with the completion of particulation by chemical interaction. Whether or not the initial bimolecular antigen-antibody reaction on the bacterial surface can take place in the absence of electrolyte, the reactants carry ionized groups and it is evident that the succeeding competing bimolecular interactions between polysaccharide molecules on partly sensitized cells and additional antibody in solution or on other cells would soon result in the formation, in the absence of electro- lyte, of particles carrying large numbers of ionized groups. Coulomb forces on such particles are known to cause abnormally great viscosities and Donnan effects, so that it would not be surprising if these forces would prevent the continuation of the chemical reactions resulting in the completion of what is commonly recognized as specific bacterial agglutination. The effect of removal of salts is shown at the bottom MICEAEL EEJDELBEBGEX AND ELVIN A. ICABAT 901 of Table VII. Only when the effect of these forces is eliminated by a sufficient ionic atmosphere, on addition of electrolyte, is it possible to obtain significant figures for viscosity, osmotic pressure, sedimenta- tion constants, and the like. To ascribe a similar r81e to electrolytes in specific bacterial agglutination would seem reasonable and consist- ent, for after reduction of the Coulomb forces the growing particles could again interact chemically, and the process of agglutination be completed. This conception of the effect of electrolytes is in part based on the charge-reducing properties of salts, like the older views of Bordet (19), Northrop and De Kruif (20), Sibley (21), and others, but involves a shift of emphasis in that, regardless of any such reduction, specific agglutination takes place only when the primary chemical interac- tion between multivalent antigen and, antibody can proceed toward completion. Many irregularities and inconsistencies in the relation of physical forces to agghrtination are thus eliminated and explained. SUMMARY 1. By the application of an absolute, quantitative microchemical method for the estimation of agglutinins, precise .data have been obtained on the course of the agglutination of Type I pneumococcus by homologous anticarbohydrate. 2. Within the limitations imposed by the necessity for the aggluti- nation reaction to take place at the bacterial surface, the reaction is shown to be analogous to the precipitin reaction and subject to the same laws. 3. The entire process of a typical instance of spec%c bacterial agglutination has been quantitatively accounted for on a purely chemical basis and expressed in the form of equations derived from the law of mass action. 4. Experimental verification of predictions based on the theory has shown a fundamental difference between this instance of specific bacterial agglutination and the commonly adduced analogies, and necessitated a revision of current conceptions regarding the rdle of electrolytes and of physical forces in the reaction. BIBLIOGRAPHY 1. Heidelbqer, M., and Kabat, E. A., Pm. Sot. Exp. Bid. and Med., 1934, 31,595; f. Exp. Med., 1934,60,643. 902 BACTERIAL AGGLUTINATION. III 2. Heidelberger, M., and Kendall, F. E., J. Ezp. Med., 1935,61,559. 3. Heidelherger, M., Kendall, F. E., and Soo Hoo, C. M., J. &p. Med., 1933, 68, 137. 4. Heidelberger, M., and Kendall, F. E., J. Exp. dlcd., 1932,66,555. 5. Heidelberger, M., and K&at, E. A., J. Exp. Med., 1936,6!$737. 6. Heidelberger, M., and Kendall, F. E., J. Ezp. Med., 1935, 61, 563; 62, 467, 697. 7. Craigie, J., and Wii, F. O., J. J&p. Xti., 1936,64,803. 8. Tiiett, W. S., .Goebel, W. F., and Avery, 0. T., J. Bzp. Med., 1930,611, 896. Heidelberger, M., and Kendall, F. E., J. Exp. dded., 1931,68,625. 9. Heidelberger, M., Kendall, F. E., and Teorell, T., J. hp. Med., 1936,63,819. 10. Eagle, H., Smith, D. E., and Vickers, P., 1. Ezp. d8cd., 1936,0,617. 11. Heidelberger, M., and Kendall, F. E., J. Ex). Med., 1929,60,809. 12. Wells, H. G., The chemical aspects of immunity, New York, Chemical Catalog Co., 2nd edition, 1929. 13. Topley, W. W. C., Outline of immunity, Baltimore, Wiim Wood h Co., 1933, chapter V. Culbertson, J. T., in Gay, F. P., Agents of disease and host resistance, Springfield, Ill., C. C. Thomas Co., 1935, chapter 20. 14. Marrack, J. R., Chemistry of antigens and antibodies, Great Britain Med. RGscarch Council, S&&l Rep. Series, No. 194, 1934. 15. Francis, T., Jr., J. Exp. Med., 1932,66,55. 16. Ivanovics, G., 2. Inrmzmit&farsch., 1933,80,209; 1934,81,518; 1935,86,165. 17. Topley, W. W. C., Wilson, J,, and Duncan, J. T., BY& J. EL+. Pdh., 1935, 16,116. 18, Pauli, W., and Valk6, E., Kolloidchemie der Eiweisskiirper, Dresden, Stein- kopff, 2nd edition, 1933. 19. Bordet, J., Trait6 de l'immunit8, Paris, Masson et Cie, 1920, pt. 3, chapter 2. 20. Northrop, J. H., and De Kruif, P., J. Cm. Pkysiol., 1922,4,629,639,6!%. 21. Shibley, G. S., J. Exp. Med., 1924,40,853.