QUANTITATIVE STUDIES ON ANTIBODY PURIFICATION

I. Tae DissoctaTION OF PRECIPITATES FoRMED BY PNEUMOCOCCUS
SrEcIFIC POLYSACCHARIDES AND HomoLocous ANTIBODIES*

By MICHAEL HEIDELBERGER, Pu.D., ano FORREST E. KENDALL, Pu.D.

(From the Department of Medicine, College of Physicians and Surgeons, Columbia
University, and the Presbyterian Hospital, New York)

(Received for publication, April 23, 1936)

Although the partial dissociation of immune precipitates was first
reported many years ago and has since been accomplished in a number
of instances (1) the methods used have been empirical and no rigid
test could be made of the yield or purity of the recovered antibody
until the recent development of absolute, quantitative methods for the
determination of precipitins (2-4) and agglutinins (5). Felton has
since stated that Types I and II pneumococcus specific precipitates,
dissolved with calcium or strontium hydroxide and treated with phos-
phate to precipitate the polysaccharide, yield antibody which is pre-
cipitable by the homologous specific polysaccharide to the extent of
80 to 85 per cent (6).

In a paper from this laboratory it was shown that increasing con-
centrations of sodium chloride resulted in a progressive decrease in the
amount of antibody nitrogen precipitated by the specific polysac-
charide of Type III pneumococcus (S IIT) from homologous antisera
(7), and that this decrease was reflected by a diminution in both con-
stants of the equation (8) describing the behavior of the serum:

R2
mg. antibody N precipitated = 2RS — A 3

It was also shown that the decrease was not due to increased solubility
of the precipitate but to a shift in the reaction equilibrium by which a
given amount of S III combines with less antibody in the presence of

* The work reported in this communication was carried out under the Harkness
Research Fund of the Presbyterian Hospital.

161
162 ANTIBODY PURIFICATION. I

a high concentration of salt than under ordinary conditions. Since
the precipitin reaction between S III and antibody has been found to
be reversible (9) it seemed that the salt effect might also be reversible.
In this event a theoretical basis would be available for the hitherto
empirical method of dissociating specific precipitates by means of
strong salt solutions. On this basis it should be possible, starting
with a washed S-antibody precipitate obtained at a proper point in
the reaction range, to dissociate a portion of the antibody with strong
salt solution and use the residual precipitate repeatedly for the absorp-
tion of additional quantities of serum and dissociation of the antibody
taken up.

It will be shown below that these expectations were realized in some
measure. It was found possible, with Types I and IT antipneumo-
coccus horse sera and Type III antipneumococcus rabbit sera to pass
from unconcentrated sera in a single step to antibody solutions of
which 85 to 93 per cent of the total nitrogen was immune nitrogen.
From the extracted precipitate by a modification of the Felton dis-
sociation procedure (6) additional antibody of a similar degree of
purity could be obtained.

Experimental

Unless otherwise stated all sera used were absorbed with C substance (10) and
with pneumococcus protein (cf. 5) in order to remove most of the antibodies other
than anticarbohydrate.

1, Preliminary Experiments on Variation of Hydrogen Ion and Salt Concentration
of Extracting Solution Experiments on the effect upon the specific precipitate of
buffer solutions of varying pH and of sodium chloride solutions of various strengths
are summarized in TableI. 1.5 to 3.0 ml. portions of the rather dilute Felton (11)
Type III pneumococcus antibody solution (horse) and 1.0 ml. portions of rabbit
3.50, Type III antipneumococcus serum (1:2) and rabbit 3.88, anti-egg albumin
(Ea) serum (1:1) were used. In the case of Pneumococcus III antibodies, S$ III
prepared according to Reference 12 was added in amounts just sufficient to reach
the beginning of the equivalence zone (8), while in the anti-egg albumin serum, Ea
was added in amounts sufficient to. reach the same point. In the experiments
designated series A, however, 1.0 ml. portions were precipitated with a smaller
amount of Ea containing 0.049 mg. N. After 1 hour in the cold the tubes were
whirled in a refrigerating centrifuge,' and the supernatants (series B) were pre-
cipitated at room temperature with one-half the amount of Ea. Both series of

1 Manufactured by the International Equipment Co., Boston.
MICHAEL HEIDELBERGER AND FORREST E. KENDALL 163

tubes were washed once with 5 ml. of chilled saline, and were used after centrifuga-
tion in the cold. In these and in the S ITI series 5 ml. of the appropriate buffer
or salt solution were added. After incubation at the temperature indicated, with
frequent stirring, the tubes were centrifuged and the nitrogen in the supernatants
was determined by the micro Kjeldahl method, deducting the small amount in a
control sample extracted with 0.9 per cent sodium chloride solution.

The pH measurements were made with the glass electrode.* The pH 3.95
solution was a phthalate buffer prepared according to Clark,’ while the pH,
8.96, 9.84, and 11 buffers were borate solutions, prepared according to Clark.‘
The concentration of the pH 3.73 buffer, which contained 24.5 gm. crystalline
sodium acetate, 48.5 gm. glacial acetic acid and 48.0 gm. sodium chloride per
liter, was molar in respect to both sodium and acetate.

Equilibrium was probably reached rapidly in the extractions, as a second -
treatment with the same solvent rarely yielded much more nitrogen than did the
control tube with 0.9 per cent saline. Precipitates from which considerable
amounts of nitrogen had been extracted were, however, more gelatinous than at the
start, and this possibly contributed to the smaller effect of a second extraction.
In many instances it was necessary to filter the supernatant through a small
Munktell No. 1F paper in order to remove suspended particles.

2. Dissociation of Specific Precipitates with Strong Sodium Chloride Solution. —
10 to 50 ml. of type specific antipneumococcus serum or Felton antibody solution
(11) were chilled, diluted with 2 to 3 volumes of chilled 0.9 per cent saline, and
precipitated with an amount of homologous specific polysaccharide (12) calculated
to bring the system to the beginning of the equivalence zone (8) or to leave a small
amount of antibody. After the precipitate had flocked, the mixture was cen-
trifuged in the cold and the precipitate evenly suspended and washed with 10 to
50 ml. portions of chilled saline until the amount of heat coagulable protein ex-
tracted was ata minimum. The residue was then evenly suspended in a volume
of 10, 15, or 20 per cent (gm. per 100 ml. solution) sodium chloride solution equal
to that of the original serum and immersed in a water bath at about 30°, with
frequent stirring. After 1 hour the mixture was centrifuged and the supernatant
filtered through a small Munktell 1F filter paper if necessary to remove suspended
particles. The solution, which was generally water clear, was dialyzed in the cold
in cellophane tubing against repeated changes of 0.9 per cent salt solution until
these gave the same interferometer® reading as did a control sample of the salt
solution used. In general the dialyzed solutions from the 10 per cent salt extrac-
tions remained clear, while small amounts of precipitate settled from the 15 per cent

2 By Mr. F. Rosebury of the Department of Biological Chemistry.

* Clark (13), page 200, pH 4.

4 Clark (13), pages 208 and 209.

5 Purchased through a grant from the Bache Fund of the National Academy
of Sciences.
164 ANTIBODY PURIFICATION. I

salt extracts and larger precipitates from the 20 per cent salt extracts, owing,
doubtless, to the actual solvent action of salt solutions of this strength super-
imposed on their effect of shifting the S-antibody equilibrium. The 15 per cent
salt solution was used in most instances. The rabbit antipneumococcus exttacts
were usually dialyzed under negative pressure in order to reduce the final volume.
This was desirable in connection with the analytical control of the solutions,
owing to the appreciable solubility of rabbit S-anti-S precipitates. At the end of
the dialysis the solutions were centrifuged if necessary, preserved by the addition
of 1 per cent by volume of 1 per cent merthiolate solution, and used for analysis
after 24 hours, after centrifugation to remove any additional precipitate caused
by the merthiolate.

Analyses for precipitin nitrogen were made according to References 2, 3, and 4
by addition of a slight excess of homologous S to duplicate 2 to 5 ml. portions of
the chilled solution and determination of the amount of nitrogen in the washed
precipitate after 48 hours in the cold. Since the total nitrogen in the supernatants
was extremely low, one washing with 3 ml. of chilled saline was considered sufi-
cient. Blank tubes were run under the same conditions with the same amount of
antibody solution and micro Kjeldahl determinations were run separately on the
supernatants (plus washings) of the blank tubes and the generally negligible
residues,® the sum of the two giving the total nitrogen of the antibody solution.
In the case of S I the small amount of S I nitrogen precipitated was deducted from
the total nitrogen precipitated in order to give antibody nitrogen. As large aliquot
portions as possible of the supernatants from the precipitin determinations were
analyzed for agglutinin nitrogen according to Reference 5, by addition to a meas-
ured volume of a suspension of homologous type specific pneumococci, and deter-
mination of the increase in nitrogen over that in the pneumococcus suspension
alone after centrifugation and washing once. It was also possible to analyze the
supernatants from the centrifuged pneumocecci for nitrogen and determine
agglutinin N by difference. Both methods of analysis gave the same result in the
case in which a comparison was made. Control determinations on Type I and
Type III pneumococci with as much as 1 ml. of undiluted normal horse serum
showed no absorption of nonspecific nitrogen, while the small amount taken up by
the Type II suspension was deducted from the agglutinin nitrogen actually deter-
mined in the Type II antibody supernatant.

The analytical data are summarized in Table II. Values are also given in the
table for the amount of antibody nitrogen taken, the amount and percentage
recovered, and the percentage of the total nitrogen in the recovered solutions
accounted for as antibody nitrogen. In the case of antisera the precipitin content
was taken as “antibody,” since it had been shown for Type I antipneumococcus

® These residues, when present, appeared to be due to the film of antibody
protein remaining in the pipette and generally appeared only after pipettes were
refilled. The small amount of nitrogen thus precipitated was deducted from the
precipitable nitrogen found.
MICHAEL HEIDELBERGER AND FORREST E. KENDALL 165

sera that the anticarbohydrate precipitin and agglutinin content are identical
(14). In both Felton antibody solutions (cf. also 14) and in those now reported,
a portion of the antibody could only be recovered as agglutinin.

In the first series of experiments summarized in Table I], 64 ml. of a TypeI
antipneumococcus Felton solution, B 79, containing 2.22 mg. of total N and 1.09
mg. of N precipitable by S I, per ml., were diluted and precipitated at 0° with 8 mg.
of SI. The washed precipitate was first treated with 15 per cent sodium chloride
solution and the recovered antibody designated 79 A. After analysis the solution
was run through a Chamberland Lz filter and again analyzed (79 Ay). The specific
precipitate remaining after the salt extraction was washed in the cold with 25
ml. of water, centrifuged in the cold, and then evenly suspended at 0° in 30 ml.
of unabsorbed New York State Type I antipneumococcus horse serum 444’ con-
taining 1.92 mg. of anti-'SINperml. After letting stand in the refrigerator over-
night, the mixture was centrifuged in the cold and the first 50 ml. of 0.9 per cent
saline washings were added to the supernatant serum. After an additional
washing the precipitate was extracted for 1 hour at 30° with 30 ml. of 15 per cent
saline. The extract (79 B) was dialyzed as described above and the precipitate
washed with water and treated again in the same way with the partially exhausted
serum to yield antibody solutions 79 C and D. The remaining precipitate was
then dissociated further by means of the method described in the following section.
The various solutions obtained are numbered 1 to 5 in Table II to indicate suc-
cessive treatments of the same precipitate.

The Type ITI specific precipitate used was derived from the pooled serum of
horses which had been immunized simultaneously to Types I and Ii pneumo-
cocci by addition of somewhat less S II to the unabsorbed serum than was cal-
culated as necessary to remove all of the antibody. After thorough washing
equal portions of this precipitate were dissociated with 15 per cent (portion A)
and 20 per cent (portion B) salt solution. Data on the recovered antibody solu-
tions are given in Table II. The surprising observation was made that these
solutions contained much anti-S I as well as anti-S II, a point which will be taken
up in the discussion.

In the case of Type III antipneumococcus specific precipitates from antisera
obtained from the horse, the recovered antibody solutions were of a considerably
lower degree of purity than in the case of Types I and II, and also contained
relatively less precipitin and more agglutinin. Efforts will be made toimprove
on these results. Preparation 792 B was derived from absorbed whole serum,
while 792 C was prepared from a Felton solution made from the same serum.
Type III antipneumococcus rabbit sera, however, readily yielded antibody solu-
tions of which over 90 per cent of the nitrogen was immune nitrogen. It will be
noted throughout that regardless of the type of antiserum, there is no advantage
in starting with partially purified antibody, such as Felton solutions.

3. Dissociation of Type I and Type III S-Anti-S Precipiiates by Means of Barium

7 Kindly. supplied by Dr. Augustus B. Wadsworth.
166 ANTIBODY PURIFICATION. I

Hydroxide and Barium Chloride —This modification of Felton’s dissociation pro-
cedure (6) was suggested by the insolubility of the barium salts of S I (15) and
S IQI (12). The Type I antipneumococcus precipitate which had been used four
times for dissociation was washed with cold water and suspended in 25 ml. of
water at 0°. 0.1 normal barium hydroxide was then added drop by drop until no
more of the precipitate appeared to dissolve, about 1 ml. being necessary. To the
strongly alkaline solution were added 4 ml. of 10 per cent barium chloride solution
containing 1 per cent by volume of the same barium hydroxide solution. This
increased the amount of precipitate, and as it settled readily, leaving a clear
solution, the mixture was centrifuged in the cold after 5 minutes. After an addi-
tional 10 minutes the clear supernatant was made very faintly acid to litmus with
dilute acetic acid and dialyzed against 0.9 per cent sodium chloride solution until
only traces of barium ion remained. These were removed after addition of 1 per
cent merthiolate solution by means of 0.6 ml. of 2per cent sodium sulfate solution.
After 24 hours in the cold the centrifuged solution was subjected to analysis and
proved to be highly pure antibody (Table IT, 79 E).

Type III S-anti-S precipitates from horse serum were worked up similarly,
but did not yield antibody of the same degree of purity. Preparation 792 D,
obtained from an antibody solution, contained a somewhat higher proportion of
immune nitrogen than did a corresponding preparation, E, from absorbed whole
serum,

DISCUSSION

In continuation of our work on a quantitative theory of the pre-
cipitin reaction, the observation was made (7) that a given amount of
pheumococcus specific polysaccharide combined with less antibody in
the presence of high sodium chloride concentrations than under normal
physiological conditions. Further study showed that this shift in
the reaction equilibrium was reversible, and that a theoretical basis
was thus provided for the dissociation of certain specific precipitates
by means of strong salt solutions. From the data given it is apparent
that a rapid, simple, and in several instances, readily reproducible
method is available for the preparation, in a single step from uncon-
centrated antiserum, of pneumococcus anticarbohydrate in a high
state of purity.

In the preliminary experiments summarized in Table I it was found
that pneumococcus S-anti-S precipitates formed at 0.9 per cent salt
concentration gave up far more nitrogen to strong salt solutions than
did egg albumin-antibody precipitates, in agreement with observations
on the reverse reaction (7) showing that the amount of nitrogen pre-
MICHAEL HEIDELBERGER AND FORREST E, KENDALL 167

cipitated in the egg albumin-antibody reaction was almost the same
in strong or weak sodium chloride solutions, while in the S-anti-S
system far less antibody was precipitated in the presence of strong
salt. On the other hand, the egg albumin precipitates proved to be
more sensitive than S-anti-S to the action of acid (pH 4) or relatively

TABLE I

Effect of Hydrogen Ion and Salt Concentration on Amount of Antibody Nitrogen
Extracted from Specific Precipitates

 

 

 

1 Vv
3 3 3 3 giz Seid |S /3
= : Biss |S |S |2)88l2 (2 12
> $4 ry Bae 2/3) 8] Rs ag i ai e¢
38 ba | 2 fee] g |g | 19d] ge | ee! ge
as Fe | 2 jsee) ¢ | | € jee | 2s] 2a] Be
< a a | a c ma |» & 2 g
me. degrees me. me. mg. | mg. | mg. | mee. | me. | mg me.
Pneumococcus Type III Antibody Solution, B 66, Horse
0.55 0, 3 hrs. | 0.00 0.01] 0.03] 0.06
t 28 .5-30 0.35} 0.43] 0.47
Pneumococcus Type III Antiserum, 3.50;, 1:2, Rabbit
1.2 21 0.26 0.11] 0.19)
1.1 28-30 0.40] 0.44| 0.43
Anti-Egg Albumin Serum, 3.88, 1:1, Rabbit
1.8 0, 2 hrs. | 0.35t 0.03} 0.07! 0.34
0.62 (series A) 21-22 | 0.13§ 0.08) 0.35
0.16 (series B) 21-22 | 0.08 0.06] 0.16
0.88 22-23 0.46] 0.03) 0.04
0.81 28-30 0.06; 0.05) 0.04

 

 

 

 

 

 

 

 

 

 

*In Ea-anti-Ea system only.

+ Calculated from quantities of components used.

¢ Second and third extractions removed 0.13 and 0.03 mg. N.

§ Extraction at pH 10 after removal of N extractable at pH 4 failed to increase
the amount extracted, nor did changing to pH 4 after 2 extractions at pH 11.

| A second extraction removed 0.04 mg. N.

strong alkaline (pH 11) buffers. In the pH 4 buffer, at least, the
effect was mainly one of solvent action, rather than a shift in the
proportions of the components, since neutralization of the solution
resulted in precipitation. When the anti-egg albumin was fractionally
precipitated and the two fractions were tested separately with the
buffers, the second fraction (Table I, series B) was found far more
168 ANTIBODY PURIFICATION. I

soluble than the first (series A), an additional confirmation of the
presence of more than one antibody in antisera to this crystalline,
homogeneous antigen (16).

The data on the actual isolation of purified antibody by dissociation
of washed pneumococcus S-anti-S precipitates are given in Table II.

 

 

 

 

 

 

 

 

TABLE II
Analyses of Antibody Solutions Obtained by Dissoctation of Specific Precipitates

Precipitin + . |Agglu-| Anti- .

Serum or antibody Antibody jitcovered eatery ve Teal my t we oy Antibody

antibody | antibody N | per mi. of recovered antibody | Total N

solution solution
me. meg. fer cent mg. | mg. mE. mg. per cent
Type I Antipneumococcus, Horse
1, 79A 70 (P)* | 21.6 (P) | 31 (P) 0.323/0.261 |0.02910.290]' 90
Al 0.300/0.236 |0.024/0.260] 87
2. B 57.3 75 13.1 0.177/0.138 |0.01810.156| 88
3. 4.7 8.2 0.11610.089 |0.00910.098) 85
4 D 3.0 5.2 0.076/0 .061 |0.003/0.064) 84
5. E 9.3 (P) | 13.3 (P) | 0.227/0.217 |0.00510.222} 98
Type II Antipneumococcus, Horse (Pn I, II)

NYCI, A | 108 21.1 (P) | 19.5 (P) | 0.36710 .321f¢ 87 (P)

NYCI, 0B | 108 24.2 22.5 0.358/0.321$/0.008)0.329} 92
Type III Antipneumococcus, Horse
792 B 32.5 3.1 9.5 0.142/0.074 |0.025/0.099| 70
792 C 43.8 3.5 8 0.092/0.048 10.02210.070| 76
792 D 14.4 2.8 19.5 0.159/0.127 (0.01410.141) 89
792 E 37 4.4 11.9 0.159|0.110 |0.016)0.126) 79
Type III Antipneumococcus, Rabbit

3.50, 18.0 4.6(P) | 25.5 0.11310. 102 90 (P)

3.511 72 16.5 23 0.433)0.383 {0.020/0.403/ 93

 

 

 

 

 

 

 

* Values followed by (P) refer to precipitin N only,

{15.2 mg. of antibody N recovered from serum 444 up to this point. Analysis
of the serum supernatant showed that 23.7 mg. had been removed.

t Separate determinations with S II, followed by S I, gave anti-S II, 0.237,
0.239, and anti-S I, 0.081 and 0.076 per ml. for the two solutions.

It is seen that the dissociation by strong salt solutions of precipitates
obtained directly from whole sera, or sera absorbed with C substance
and pneumococcus protein, yields antibody solutions of as high a
degree of purity as does dissociation of precipitates originally derived
from partially purified antibody solutions.
MICHAEL HEIDELBERGER AND FORREST E. KENDALL 169

The five Type I pneumococcus antibody solutions given in the
table were obtained by successive dissociations of a single S-anti-S
precipitate. Between all dissociations except the last two this pre-
cipitate was used again for the removal of additional quantities of
antibody from an unabsorbed whole serum. After the fourth dis-
sociation, which yielded relatively little antibody, an additional
amount, of exceptionally high purity, was recovered by the barium
method described in the experimental part. If this portion is con-
sidered as derived from the original precipitate 44 per cent of the
precipitin in the Felton antibody solution used for its formation was
recovered, as well as 26 per cent of the antibody in the serum used for
the second, third, and fourth absorptions. After these absorptions
59 per cent of the precipitin content was still present in the serum.
Attempts will be made to improve on the yields of recovered antibody.

Although one of the recovered Type I solutions may be said to be
95 to 100 per cent pure antibody within the limits of error of the
analytical methods used, it is apparent that even the simple procedure
used in this and the other cases has resulted in a slight change in the
properties of the antibody. Thus it has been shown that anticarbo-
hydrate in Type I antipneumococcus sera is entirely precipitable by
the homologous specific polysaccharide (14), while the anticarbo-
drate in Felton solutions (14, 17), as well as in the more highly purified
solutions now reported, is in part recoverable only as agglutinin.

In the experiments on the recovery of antibody from an S II-anti-S
precipitate obtained from a combined Type I, IT antipneumococcus
horse serum, the surprising discovery was made that the recovered
solution, although prepared from a precipitate formed by addition of
S II (12) to the serum, contained significant amounts of antibody to
the specific polysaccharide of Type I pneumococcus. Since Types I
and II anticarbohydrate do not cross react with the heterologous
polysaccharides when tested separately, it is believed that the cross
reaction in the combined serum is another instance (cf., for example,
16) in which the antibody in question contains too few reactive group-
ings in its molecule to take part by itself in the building up of aggre-
gates large enough to precipitate, but may add to an aggregate in
process of formation between antigen (or hapten) and antibody which
are multivalent with respect to each other. The alternative explana-
170 ANTIBODY PURIFICATION. I

tion, that the anti-S I and anti-S II groupings are on the same mole-
cule is excluded by the experiment recorded in the footnote to Table
II, in which it is shown that the anti-S I is left in the supernatant
when the purified antibody is precipitated with S 1]. That this is
possible in the recovered antibody solution although the anti-S I was
firmly bound in the original precipitate is probably due to an equilib-
rium between free and bound anti-S I.

While the results with Type III antipneumococcus horse serum are
the poorest of the series as to yield and quality of the recovered anti-
body, it is believed that they are better in the latter respect than those
obtained by any other single step procedure now available. Attempts
will be made to improve on the S II-anti-S results. The dissociation
method proved applicable, however, to the corresponding rabbit
antisera, and it is probable that both of the solutions described in the
table represent 95 to 100 per cent pure antibody, since rabbit S-anti-S
precipitates are more soluble than those formed with antibodies pro-
duced in the horse and no solubility correction has been applied to the
figures in the table.

Felton (6) has reported the preparation, from Types I and IL
pneumococcus antibody solutions, of zinc and aluminium containing
solutions of which 100 per cent of the protein was precipitable by the
homologous specific polysaccharide. After removal of the metal
salts, the remaining protein was still precipitable to the extent of
80 to 90 per cent. Felton has also reported 80 to 85 per cent pre-
cipitable antibody by an alkaline earth, hydroxide-phosphate dis-
sociation method. Chow and Goebel (18), also starting with partially
purified Type I pneumococcus antibody, have stated that their end-
products were, under optimal conditions, precipitable to the extent of
85 to 90 per cent.

Thus, although antibody of the degree of purity attained by the
present method has already been reported, the methods used are not
simple of execution and cannot be carried out on the original serum,
two disadvantages which are eliminated by the salt dissociation
method herein described. Moreover, analytical data are now pre-
sented for the first time on antibody of over 80 per cent purity.

A study of the physical, chemical, and immunological properties
of the new antibody material is under way.
MICHAEL HEIDELBERGER AND FORREST E. KENDALL 171

The authors wish again to express their thanks to Dr. Torsten
Teorell, with whose assistance many of the observations (7) were
made which led to the present work.

SUMMARY

1. Quantitative data are given on the effect of changes in hydrogen
ion concentration and of salt solutions of high concentration on certain
immune precipitates obtained at lower salt concentration.

2. Advantage is taken of the shift in reaction equilibrium brought
about by the salt in the case of pneumococcus carbohydrate-anti-
carbohydrate precipitates to enable the preparation, in a single step
from unconcentrated serum, of antibody solutions in which up to 93
per cent of the total nitrogen is immune nitrogen. The method per-
mits successive absorptions of a serum to be made with the same
specific precipitate.

3. A modification of Felton’s alkaline earth hydroxide dissociation
procedure is proposed which yields highly purified antibody with
precipitates which have been subjected to several successive salt
dissociations.

BIBLIOGRAPHY

1. Marrack, J. R., The chemistry of antigens and antibodies, London, His
Majesty’s Stationery Office, 1934, 96. Landsteiner, K., The specificity of
serological reactions, Springfield, Dlinois, Charles C. Thomas, 1936, 95.

2. Heidelberger, M., Sia, R. H. P., and Kendall, F. E., J. Exp. Med., 1930, 52,
477.

3. Heidelberger, M., Kendall, F. E., and Soo Hoo, C. M., J. Exp. Med., 1933,
§8, 137.

4. Heidelberger, M., and Kendall, F. E., J. Exp. Med., 1935, 61, 559.

5. Heidelberger, M., and Kabat, E. A., Proc. Soc. Exp. Biol. and Med., 1934,
31, 595; J. Exp. Med., 1934, 60, 643.

6. Felton, L. D., J. Zmmunol., 1932, 22, 453; also contains references to earlier
work on dissociation of specific precipitates.

7. Heidelberger, M., Kendall, F. E., and Teorell, T., J. Exp. Med., 1936, 63,
819,

8. Heidelberger, M., and Kendall, F. E., J. Exp. Med., 1935, 61, 563.

9. Heidelberger, M., and Kendall, F. E., J. Exp. Med., 1929, 60, 809.

10. Tillett, W. S., Goebel, W. F., and Avery, O. T., J. Exp. Med., 1930, 52,
896. Heidelberger, M., and Kendall, F. E., J. Exp. Med., 1931, 63, 625.
11. Felton, L. D., J. Infect. Dis., 1928, 42, 248; and earlier papers.
172 ANTIBODY PURIFICATION. I

12. Heidelberger, M., Kendall, F. E., and Scherp, H. W., Proc. Soc. Exp. Biol.
and Med., 1935, 33, 188; J. Exp. Med., 1936, 64, in press. Heidelberger,
M., and Kendall, F. E., Proc. Soc. Exp. Biol. and Med., 1935, 83, 445.

13. Clark, W. M., The determination of hydrogen ions, Baltimore, Williams &
Wilkins Co., 3rd edition, 1928.

14, Heidelberger, M., and Kabat, E. A., J. Exp. Med., 1936, 63, 737.

15. Heidelberger, M., Goebel, W. F., and Avery, O. T., J. Exp. Med., 1925, 42,
727,

16. Heidelberger, M., and Kendall, F. E., J. Exp. Med., 1935, 62, 697.

17. Felton, L..D., J. Immunol., 1931, 21, 341.

18. Chow, B. F., and Goebel, W. F., J. Exp. Med., 1935, 62, 179.