SOURCES OF THE ANTIBODIES DEVELOPING AFTER
BY OSWALD H. ROBERTSON, M.D., AND PEYTON ROUS, M.D.
(From the Laboratories of The Rockefeller Institute for Medical Research.)
(Received for publication, October 1, 1921.)
The recent wide utilization of transfusion as a therapeutic measure
has brought to light many facts of theoretical as well as practical
significance. Perhaps most interesting from both points of view is
the gradual decrease in beneficial effect, and the appearance sometimes
of positive injury from the frequently repeated injection of alien
blood. Were it not for this complication one might reasonably ex-
pect to maintain cases of pernicious anemia in good blood condition
for an indefinite period of time. Needless to say, the sources
of the failure have been the subject of much discussion and of
some research. One fact of great weight has been clearly shown.
Blood derived from a donor originally compatible, as proven both by
in vitro tests and by the clinical result, may not only cease to be use-
ful when too frequently injected into the same individual but may
give rise to serious reactions. A change has occurred, not in the donor
but in the recipient, such that the alien blood is no longer tolerated in
Boycott and Douglas 1 have noted that blood is destroyed more
rapidly after repeated transfusions in normal animals than it is at
first, as attested by an increase in the rate at which plethora disappears.
One of us, with Oliver, 2 has utilized the phenomenon to induce in
rabbits a hemosiderosis closely resembling that of hemochromatosis
in man. No doubt a part of the pigmentation observed at autopsy
in pernicious anemia patients who have been repeatedly transfused
is due to a like destruction of alien blood. But how is this blood
destroyed? By circulating antibodies or within special organs?
t Boycott, A. E., and Douglas, C. G., J. Path. and Bact., 1909, xlii, 414.
Rous, P., and Oliver, 7., J. Exp. ]fled., 1918, xxviii, 629.
ANTIBODIES DEVELOPING AFTER TRANSFUSION
Recent observations indicate that circulating antibodies are to a
considerable degree involved, s It has been found that repeated
transfusions of compatible blood in rabbits are followed often by the
appearance in the recipients' plasma of hemagglutinins so strong
that the red cells come together into a firm mass practically as soon
as the blood has been shed, while, furthermore, a fulminant destruc-
tion of corpuscles may take place in vivo with result in anemia. Rob-
ertson 4 has presented evidence that the elements destroyed are the
alien cells which, little by little, under the circumstances of plethora
and diminished bone marrow activity consequent thereon have taken
the place of cells proper to the host.
The hemagglutinins just mentioned are in the immediate sense
autoantibodies. They clump practically all of the circulating eryth-
rocytes, are especially effective at low temperatures, and persist
in high titer for months after the transfusions have been discontinued
and after recovery from the severe intercurrent anemia which may
develop soon after their appearance.
engaged in eliminating unusually large amounts of blood will elaborate
antibodies directed against its own cells? This is a point of major
interest in any study of the source of the hemagglutinins, and one
not without a practical bearing. In human beings true autohemag-
glutinins have repeatedly been observed 5 in association with anemia
of obscure origin; while autohemolysins are known to bear an impor-
tant relation to paroxysmal hemoglobinuria.
Is it possible that an organism
Three to six compatible donors were selected for each recipient by the examina-
tion of mixtures of the citrated bloods, e The recipients received, 6 days in every
7, 10 cc. of blood taken by cardiac aspiration into 1 to 2 cc. of 0.9 per cent salt
solution containing 1 per cent of sodium citrate. The donors were employed in
rotation. The small amount of citrate mentioned was sufficient to prevent
clotting during the short period required to introduce the blood into the recipient's
ear vein. About half of the transfused animals failed to develop autoagglutinins
at any time, even when the injections were continued for many weeks. In the
8 Rous, P., and Robertson, O.H., J. Exp. Med., 1918, xxvii, 509,
4 Robertson, O. H., J. Exp. Med., 1917, xxvi, 221.
5 Clough, M. C., and Richter, I. M., Bull. Johns Hopkins Hosp., 1918, xxix, 86.
6 Rous, P., and Turner, J. R., J. Am. Meal. Assn., 1915, lxiv, 1980.
OSWALD H. ROBERTSONAND PEYTON ROD'S
other individuals they appeared early, as a rule after only five to ten transfusions;
and when three or four more had been given, that is to say, after some 10 days
to 3 weeks in all, they were well marked. The shed blood of the rabbits, when
examined at room temperature, now showed prior to clotting a massive clumping
of the red cells due, as our previous work has shown, to the presence in the
plasma of true hemagglutinins, a
The Mixed Content of the Blood.
The transfused rabbits became, as it were, mixing vessels for
several alien corpuscles and plasmas, perhaps indeed concentrators of
immune principles through the elimination of the fluid wherewith
the latter were introduced. The clumping of the cells of the shed
blood might conceivably have been the result of:
1. The injected sodium citrate---a possibility ruled out by nega-
tive findings in control animals given citrate alone.
2. Antibodies introduced with the donor's plasma and (a)active
against the recipient's cells. These might either have developed
during the period of experiment, or have been so weak as to escape
detection in vitro. No tests had been made to rule out (b) interag-
glutination among the donor bloods.
3. Antibodies elicited in the recipients and directed against (a) the
strange corpuscles and (b) against the animal's own corpuscles.
4. Changes in the circulating cells resulting in a greater
5. Antibodies within the injected corpuscles, liberated by their
More remote possibilities could be invoked, as for example, a non-
specific agglutination brought about through physical changes in
the plasmas pooled in vivo, but, as will be shown, those above listed
suffice to account for the findings.
Tests with Preserved Cells.
In an initial series of experiments for the analysis of the conditions,
blood specimens were taken aseptically from the individuals of sev-
eral donor-recipient groups prior to any transfusions, and set aside
in a citrate-glucose mixture in which rabbit corpuscles remain viable
ANTIBODIES DEVELOPING APTER TRANSFUSION
for several weeksJ Later when the transfusions had elicited the
clumping phenomenon, these preserved corpuscles were several times
washed, made to a 5 per cent suspension with salt solution, and used
in agglutination tests paralleling others carried out at the same time
with freshly taken cells of the same animals.
For the purpose, equal parts of cell suspension and serum were mixed, allowed
to stand for 15 to 30 minutes at room temperature, and examined microscopically.
Room temperature is far more favorable to hemagglutination than blood heat.
Clumping of the cells is ordinarily completed in about 10 minutes.
The tests yielded clear-cut results. Prior to transfusion no agglu-
tination reaction was ever observed between the bloods of the pro-
spective donors and recipients, when examined either according to
the method just described or by that with whole citrated bloods
previously referred to. Weak interagglutinations were sometimes
noted between the bloods of certain donor groups but they were
entirely absent from others. The fresh and preserved cells of the
same individual were regularly found to behave almost identically
on test in vitro. No evidence was obtained for possibility No. 4
of those above listed, namely an increase in agglutinability of cells
circulating in an alien organism; but several other causes for the
clumping were readily demonstrated. Most important, and most
frequent, were isoagglutinins, often of high titer, developing in
the recipient and effective against the cells of one or more donors
(possibility 3, a). Such antibodies commonly appeared, and were
readily demonstrable in serum separated from the clot at room
temperature. Their association with the most marked clumping
noted makes it highly probable that the sudden anemia supervening
on plethora which occurs only in association with such marked clump-
ing is due to isoantibodies.
Agglutinins interactive between the donor bloods (possibility 2,
b) sometimes developed, but they were never strong. Occasionally
agglut~n~ns for the recipients' cells appeared ill the donors' plasma dur-
ing the period of transfusion (possibility 2, a). These were always
so weak as to be negligible in vivo.
There remained a series of instances in which, with all such anti-
bodies ruled out, there were yet hemagglutinins in the recipient that
7 Rous, P., and Turner, J. R., J. Exp. Med., 1915, xxiii, 219.
OSWALD H. ROBERTSON AND PEYTON ROUS
led to a marked clumping not only of the mixture of cells present after
transfusion but of the recipient's cells taken prior to transfusion and
kept in vitro.
Such antibodies, like true autoagglutinins, s were read-
fly bound to, and completely freed from, the cells by cooling and wann-
ing respectively. Tests on their origin were now begun; and to rule
out wholly the possibility that undetected antibodies had been in-
jected with the donors' bloods, the transfusions were carried out with
Transfusions with Washed Cells.
Five donor-recipient groups were chosen in the usual way and the donors bled
in rotation. 10 cc. of whole blood was taken each time into about 20 cc. of a
sterile isotonic solution of sodium citrate (3.8 per cent) in water and allowed to
stand in the ice box for 2 days, during which period practically complete sedimen-
tation occurred. The supernatant fluid was now pipetted off and the cells were
suspended in about 20 cc. of Ringer's solution, to which had been added 0.125
per cent of gelatin and 0.2 per cent of sodium citrate. The gelatin prevented
mechanical injury of the cells during the manipulations, and the citrate did away
with the slight tendency to clotting. After a further 2 days of sedimentation in
the cold the corpuscles were again suspended, but this time in gelatin-Ringer's
solution lacking citrate, and immediately thrown down with the centrifuge; made
to a total bulk of 10 cc. with an ordinary Ringer's mixture containing only 0.76
per cent NaC1; and injected. During all the handling no hemolysis occurred
except occasionally to a negligible degree in the case of a donor with unusually
Prior to the first transfusion, blood specimens from the prospective recipients
were taken for preservation as already described.
In three of five rabbits repeatedly injected with the washed cells
a well defined clumping in the shed blood was soon noted.
two better marked cases the serum of the recipient, separated from
the cells at 37°C., caused at room temperature outspoken agglutina-
tion of the preserved ceils of the same individual.
which caused the red ceils to come together remained fixed upon them
in the cold, and the cell masses could be repeatedly washed in salt
solution without its liberation. But when the fluid was wanned to
body heat the agglutinins passed into it and the cell mass separated
into its components. By such means the antibody proved readily
obtainable in salt solution, as is the case with autoagglutinins. 8
s Landsteiner, K., M iind,, reed. Woch., 1903, 1, 1812.
ANTIBODIES DEVELOPING AFTER TRANSFUSION
In considering these results one asks first whether they might not
have been duplicated with normal blood. For it will be recalled
that Landsteiner s was able to demonstrate autoagglutinins in the
normal blood of several animal species, among them the rabbit. But,
the blood of our recipients showed no clumping prior to the transfu-
sions. Furthermore, Landsteiner's report shows that his agglutinins
were vastly weaker than ours, being demonstrable at room tempera-
ture only when a large amount of serum was allowed to act on a few
corpuscles. Nevertheless, we made attempts to demonstrate auto-
agglutinins in five normal rabbits by the same measures that had been
employed with the transfused ones, but obtained consistently nega-
A utofixation of Isoantibodies.
A remote possibility which suggested itself to account for the
autoagglutination just described was the fixation on the donor cells
and persistence throughout the washing process of isoantibodies
present in the donor plasma and effective aga~st the blood of the
recipient. So unlikely did this seem that we had not considered it
at the time of the above mentioned transfusions, but, secure in the
realization that no foreign serum would reach the recipients,
had omitted compatibility tests of them and the donors. Now an
experiment was undertaken to determine whether normal isoagglu-
tinins become fixed on the possessor's own cells when these latter
are allowed to sediment out of citrated plasma in the cold. Bloods
suitable for the work were obtained with some difficulty owing to the
fact that isoantibodies in normal rabbits are weak and rather
Experiment L--By the examination of mixtures of whole citrated bloods five
rabbits possessing isoagglutinins were selected. In mixtures of three parts of
serum separated from the clot at 37°C. and one part of 5 per cent suspension in
salt solution of the susceptible cells, definite dumping occurred in all cases at
room temperature. Now portions of blood from the five animals were treated to
the same slow, repeated washing by sedimentation in the ice box as in the case
of the donors in the transfusion experiments with washed cells already described;
90ttenberg, R., Kaliski, D. J., and Friedman, S. S., J. Med. Research, 1913-14,
OSWALD H. ROBERTSO~AN~ PEYTO~ ROUS 147
and after the final washings, carried out with the help of the centrifuge, each cell
sediment was tested as follows for serum agglutinins which might have become
fixed on the corpuscles:
All possible supernatant fluid was drawn off; Locke's solution to about one-
third the quantity of the original plasma was poured on; the cells were suspended
for 10 minutes at 38-40°C.; and then they were thrown down by rapid, brief
centrifugation. The salt solution was then immediately pipetted away and mixed
in various proportions with 5 per cent suspensions of the cells that had originally
been agglutinated by the plasma. As a control similar mixtures were made of
cell suspensions and serum from the five animals, all kept from the time the
original citrated specimens were drawn. The control serum had been removed
from the clot after a few hours at incubator temperature and preserved in the ice
box until needed.
In every case the salt solution clumped the cells acted upon by the control
serum, while comparative tests in graded dilution showed that practically the
entire original isoagglutinln content of the serum had passed into it.
The point disclosed by this experiment was most unexpected and
of considerable theoretical interest. It is evident that normal iso-
agglutinins of the rabbit may become fixed upon the animal's own
ceils in the cold, to be given up again into salt solution at 37oc. Pos-
sibly in our transfusions with cells washed by sedimentation in the
cold some isoagglutinins were thus carried over into the recipient.
But this 'does not warrant the conclusion that an introduction of
such isoagglutinins is the explanation of the clumping previously
interpreted as autoagglutination. In the transfusions with whole
citrated blood from compatible donors which resulted in what seemed
to be autoagglutination the total amount of plasma injected into the
recipient prior to the development of the clumping phenomenon
was often considerably less and never very much more than the ani-
mal itself had to start with. Now granting that all of the alien plasma
save the antibodies was removed from circulation, this would
mean that the latter were left in about their original concentration
in the donors, and at this concentration they had not been demon-
strable i~ vitro when the donors were chosen. They cannot, then,
have been responsible for the autoagglutination now seen in the re-
cipient's shed blood and ih suspensions of his preserved cells mixed
with fresh serum. Nevertheless, transfusion experiments with ceils
washed in the warm were planned, but before they had been under-
taken a new fact came to light. The corpuscles of normal rabbits
ANTIBODIES DEVELOPING AFTER TRANSFUSION
having no demonstrable isoagglutinins in the plasma were themselves
found to contain such antibodies in considerable amount.
Intracellular A gglutinins.
Klein I° in 1902 claimed that extracts of red cells in salt solution or
water showed agglutinative properties.
general recognition, perhaps in part because of their conflict with the
accepted nile that circulating antibodies are confined to the plasma,
but also doubtless because of the irregularity of his results and the
fact that the method used to obtain them is open to criticism.
tain the agglutinins he ground the corpuscles with quartz sand; and
salt solution ground with sand becomes endowed with the ability to
cause a non-specific agglutination as we have repeatedly found.
the following experiment the cells were broken up by another means.
His findings have met no
Experiment 2.--Three rabbits with plasmas that entirely failed to clump the
cells of certain other rabbits in 9 to 1 and 1 to 9 mixtures of the whole bloods were
bled 5 cc. each into 10 cc. of isotonic sodium citrate solution at 37°C. The cor-
puscles were at once packed with the centrifuge in a room maintained at body
heat, and the supernatant fluid pipetted away as completely as possible, while
still warm. The temperature of the room tended to go slightly above 38°C.
because of the heating up of the centrifuge.
Now portions of each specimen of sedimented red cells were spread in thin
layer on watch-glasses and dried in a current of air, a process requiring about 20
minutes. To the dried material of 0.7 to 1.0 cc. of cells, 1.0 to 1.7 cc. of doubly
distilled water was added; and after the resulting thin paste had stood for a few
minutes, it was centrifuged; the clear, deep red, supernatant fluid pipetted off;
and to ten parts of this last one part of 9.0 per cent sodium chloride solution was
added to restore approximate isotonicity. The fluid remained clear. It was now
used in agglutination tests with corpuscles of the sort that had remained unagglutin-
ated by the plasmas of the animals from which the cells for drying had come.
Nine parts of the extract were used to one of a 50 per cent suspension of corpuscles
in salt solution. In every case there was an almost immediate clumping of the cells
into coarse flocculi. Under the microscope the clumping was found to be a
characteristic agglutination, with a candy-like stringing out of the adherent cor-
puscles when pressure was made on the cover-glass, and a resumption of the
original form when pressure was relieved. There was no crenation of the cells.
The flocculi failed to break up when salt solution was added in quantity at room
temperature, but fell apart immediately into the component cells whenwarmed to
37°C. in salt solution or the original cell extract; and in the latter at once reformed
on cooling to or below 34.5°C.
to Klein, A., Wien. klin. Woch., 1902, xv, 413.
OSWALD H. ROBERTSO~ ~D PEYTON ROUS
A closer study of the intracellular agglutinins of rabbits such as
were here found should have considerable interest. We have examined
the corpuscles of eight human bloods for similar principles, according
to the method described, but the findings were negative.
With the demonstration that rabbit red cells contain a principle
capable of agglutinating characteristically the corpuscles of other
rabbits, our study of the autoagglutination induced by transfusion
came to an end. In brief, we had found that the remarkable clump-
ing of the cells in the shed blood of repeatedly transfused rabbits is
due in most instances to the action of isoagglutinins developing in the
recipient and effective upon the alien elements circulating amongst
its own cells. Interagglutination of the donor cells by the plasmas
introduced with them may sometimes play a subsidiary part in the
phenomenon, as may possibly agglutination of the recipient's cor-
puscles by weak antibodies appearing in the donor's plasma during
the course of the experiment. There remain instances in which,
with all such causes of clumping ruled out, the recipient's corpuscles,
taken prior to transfusion and preserved in vitro, are agglutinated by
the animal's own serum obtained after several injections of alien
blood. That these are genuine instances of induced autoagglutina-
tion remains uncertain because the possibility was not excluded that
intracellular agglutinins derived from the donors underwent liberation
in the recipient. The best evidence for a true autoagglutination
lies in the not infrequent long persistence of the antibodies in the
recipient's blood. In one such instance the clumping phenomenon
remained well marked for at least 133 days after the last transfusion.
During this period the plethora consequent on the introduction of
the blood had been quickly succeeded by a profound anemia during
which the hemoglobin fell to 27 per cent (Palmer), and this in turn
was followed by a rapid recovery to normal and many weeks of good
health. 3 Here the destruction and replacement of the circulating
elements gave assurance that whatever the original source of the
agglutinin, it was directed against the rabbit's own cells., An anti-
body of foreign origin would scarcely have remained so long in the
blood stream. Tests with serum taken from other recipients when
ANTIBODIES D~.VELOPING AFTER TRANSFUSION
the clumping phenomenon Was at its height, and preserved in the
ice box, reinforce such findings. Thus, for example, it was observed
that cells obtained from a recipient 86 days after the last transfusion,
with an intervening plethora-anemia-recovery sequence, were ag-
glutinated by serum from the animal preserved all this while. There
are several facts which speak for the view that true autoagglutinins
were induced in the transfused rabbits as a by-product to isoagglu-
tinins. Our experiment in which normal isoagglutinins became fixed
in the cold on their possessor's own cells shows how close is the anti-
genic relationship between the erythrocytes of different rabbits.
And Ottenberg and Thalheimer n have made the significant observation
with regard to the normal autoagglutinins of cats that: "Those
strongly agglutinative sera which affect the red cells of a large num-
ber of other animals are usually those which are auto-agglutinative."
The fact that the normal isoantibodies of cats are weak at most,
while isohemolysins are entirely absent from the plasma would seem
to indicate that, as in rabbits, the antigenic differentiation of the red
cells is but slight. Numerous recent authors have noted that immuni-
zation with a single antigen may cause the development of antibodies
of rather wide application. One need only suppose for the present
case that this widening of application is such that the antibodies
find a mark in the animal's own cells, as was Ehrlich's original
The clumping observed in the shed blood of transfused rabbits
never fails to bring together practically all of the cells. Yet our
analysis of the content of the sera has shown in a number of instances
that agglutinins were present only for certain ceils in the circulating
mixture. The apparent contradiction here seen is explained by the
distribution of the susceptible cells within rouleaux. These latter,
which fail to break up save when agglutination is strong, are brought
together with the cells. If they have been broken up by washing
with warm salt solution, and the cells dispersed, the agglutination
becomes selective. Whenever the clumping of the shed blood is
strong enough to disorganize the rouleaux and mass all the cells con-
fusedly, agglutinins for all are regularly found to be present on serum
u Ottenberg, R., and Thalheimer, W., J• Me~. Re~e~ch, 1915-16, xxxiii, 213.
OSWALD H. ROBERTSONAND PEYTON ROUS
The fact that repeated subcutaneous or intraperltoneal injections
with a compatible blood will cause the development of antibodies
effective against such blood has been generally accepted since Ehrlich's
study of induced isolysins in goats; but it has failed of recognition
in its important practical bearing on the outcome of repeated trans-
fusions in man. Our experience with rabbits repeatedly injected
from compatible donors, especially such animals as showed a sudden
and great blood destruction, serves to illustrate in little what must
not infrequently occur in human beings. It follows that the most
careful blood tests are called for with patients repeatedly retrans-
fused at short intervals. In special, a thick spread of the patient's
blood should be examined for autoagglutinatlon which is often, as
we have shown, an evidence that newly developed isoagglut~n~n.~
are in circulation. Throughout the tests the temperature factor
should be carefully controlled. Serum separated from the clot at
37°C. will often yield antibodies not demonstrable in that taken in
the cold, and conversely agglutination mixtures examined at room
temperature are far more likely to yield positive findings than those
kept at blood heat.
Whether massive transfusions are preferable in pernicious anemia
to repeated small ones remains uncertain. Sudden large increments
of blood tend to lessen the reparative activity of the bone marrow. 4
But, on the other hand, they may also act to prevent fulminant de-
struction of the introduced blood by the recipient's serum antibodies.
In the work already mentioned on experimental hemosiderosis in
rabbits ~ some very large transfusions were given to animals that had
developed strong serum antibodies, in the hope that abundant and
continued blood destruction would ensue.
• The amount of antigen introduced so greatly exceeded that of the
circulating antibodies that these latter failed to be destructive and
plethora was maintained. It is a curious fact that in animals trans-
fused 6 days out of every 7 during a period of 6 months the agglu-
tinins gradually and completely disappeared, although following the
first few injections they were often strong. A similar disappearance
of precipitins upon long immunization has been recorded by Tschisto-
witsch and Nuttall. n Yet the transfused blood is somehow destroyed
with great rapidity in animals frequently injected.
This was never the case.
12 Nuttall, G.H.F., Bloodimmunityand blood relationship, Cambridge, 1904,127.
ANTIBODIES DEVELOPING A~TER TRANSFUSION
The massive agglutination observable in the shed blood of trans-
fused rabbits, and associated not infrequently with sudden marked
blood destruction, has a practical significance in connection with the
untoward results of repeated transfusion from donors originally
compatible; and it has special theoretical interest because the clumping
of the cells is apparently all autoagglutination. To determine the
actual source of the antibodies has been the object of the present work.
The agglutination ill its most marked form has been traced to
isoantibodies elicited by the presencein the body of corpuscles originally
found compatible; and the frequently associated, rapid blood destruc-
tion is doubtless of similar origin. Occasionally antibodies develop
in the donor bloods during the period of transfusion, but they are so
weak as to be negligible. There remain instances of what would
seem to be true autoagglutination due to serum bodies induced by the
transfusions as a by-product, so to speak, in the manufacture of iso-
agglutinlns. The antigenic relationship between the red cells of
different rabbits is so close that normal isoagglutinins became fixed ill
the cold upon their elaborator's own corpuscles.
Agglutinins exist within the red cells of rabbits--as has been claimed
by Klein. They are readily demonstrable ill watery extracts of the
dried corpuscles. Whether similar agglutinins ever exist within
human cells remains to be determined. We have not found them
in the normal corpuscles.