A BACTERIAL DISEASE OF LIMULUS POLYPHEMUS
The Marine Biological Laboratories, Woods Hole, Massachusetts, and the Department of Pathobiology,
School of Hygiene and Public Health, The Johns Hopkins University
Received for publication October 12, 1955
Supported in part by a Grant-in-Aid from the National Mogical Institute (E-135) of the U. S. Public Health Service.
|Materials and Methods
In Vitro Effects on Serum
Bacteria obtained at random from fresh sea water were injected into a series of horseshoe crabs (limulus polyphemus) of varying sizes. One limulus became sluggish and apparently ill. Blood from its heart did not clot when drawn and placed on glass, and yet instant clotting is a characteristic of normal limulus blood (2). Cultures taken from the heart blood at intervals for several days yielded quantities of new bacteria. These were injected into limulus and a few other marine invertebrates. and the reactions of the hosts to the bacteria were studied.
The bacteria caused an active progressive disease marked by extensive intravascular clotting and death. Injection of a heat stable derivative of the bacterium also caused intravascular clotting and death. The intravascular clots had the same appearance as clots that formed when normal limulus blood was placed in clean glassware.
Other gram negative bacteria or toxins also provoked intravascular clotting in normal limuli. When these same bacteria or toxins were added to sera from normal limuli, a stable gel was formed.
Materials and Methods
All of the experimental animals were obtained from the Marine Biological Laboratory Supply Department and were kept in running sea water. Their sizes varied from 1 to 12 inches across the shell. No attempt was made to isolate them from each other, and since they were wild caught, nothing is known about previous exposure to pathogenic agents. In general the limuli appeared in good health, and under the conditions in the laboratory only one death among the several hundred which were not inoculated was noted. The work was done during July and August 1953, and July, August, and September 1954.
Bacterial cultures were made on sea water agar enriched with peptone and ferric salts (1). During the first season aged sea water was used (1) but since it was found not to be needed for our cultures, subsequent agar plates were prepared with fresh sea water obtained from the tap.
Glassware was cleaned with the commercial detergent "Alconox" rinsed with running and then with distilled water, and air dried.
Blood cultures of large limuli were obtained by rubbing the surface of the joint of a leg with 70 per cent ethyl alcohol, inserting a sterile hypodermic syringe into the joint, and withdrawing the blood. After discarding the first few drops, one or two drops of this fresh blood were immediately placed on a sea water agar plate and spread over the surface with a sterile platinum loop. Cultures were incubated at room temperature and examined for i three consecutive days. With care, sterile cultures were routinely obtained from normal animals. The procedure for smaller limuli was similar except that the blood was obtained by cardiac puncture by inserting the needle at the joint between the thoracic and the abdominal segment. Cultures were always taken from some area other than the one in which the bacteria were injected.
Normal sera were obtained by withdrawing blood sterilely from the hearts of large limuli, placing it in sterile petri dishes, and removing the supernatant 24 hours afterwards.
Description of the Organism. Bacteria obtained from the blood of sick limuli were gram negative rods which formed discrete shiny translucent colonies on sea water agar within 24 hours. They had a slight greenish tint by indirect light and were light brown with transmitted light. The colonies seemed to effect confluent growth under conditions which are described under "pathogenicity". The percentage of motile forms was much greater in the confluent colonies. The organisms grew on Salmonella-Shigella agar, desoxycholate and chocolate agar, but did not grow on Christiansens' urea agar. No gas was produced. Some of the reactions produced in fermentation media are indicated in Table I.
|Lactose||Dextrose||Sucrose||Xylose||Mannite||Urease||Indole||Plain Agar||Human Blood Agar||Rabbit Blood Agar||Selenite F Enrichment|
|No NaCl added ...||Neg.||Acid||Neg.||Neg.||Acid||Neg.||Neg.||Good growth||Good growth||Good growth||Growth|
|2% NaCl added ...||Neg.||Acid||Neg.||Neg.||Acid||Neg.||Neg./pos.||Good growth||Good growth||Good growth||Growth|
|2.5% NaCl added ...||Neg.||Acid||Neg.||Neg.||Acid||Neg.||Neg.||Good growth||Good growth||Good growth||Growth|
|3% NaCl added ...||Neg.||Acid||Neg.||Neg.||Acid||Neg.||Neg.||Good growth||Good growth||Good growth||Growth|
Courtesy Miss Regain Schneider, Biological Division, Johns Hopkins Hospital
There are few references to organisms pathogenic for marine invertebrates. A few bacterial diseases have been described (3, 4, 5). One paper reports that injections of a marine bacterium, Gaffkia, killed lobsters, and another records luminescent bacterial infections in a sand flea (5). However, it is not made clear whether the deaths were the result of overwhelming initial effects, or the consequence of progressive disease. We had previously found that several marine crustacea were often killed by the injection of small amounts of fresh heavy suspensions of various marine bacteria, including a bright yellow chromogen. The colored colonies produced by the chromogen when it was cultured from the blood made it possible to show that it was disseminated throughout the vascular system. Progressive disease, however, was not produced at any time. Death of a host shortly after an injection of many bacteria therefore does not in itself prove pathogenicity. It may be the result of toxic effects.
|Limulus #||Inoculation of Bacterial Suspension||Number of Bacteria/Drop on Sea Water Agar Plate|
|6 hrs.||1 day||2||3||5||7|
|26||Undiluted||500+, no clot||Dead||--|
|27||500+, no clot||Dead||--|
One problem, then, was to determine whether an active and progressive infection could be produced. Table II shows that this was produced with both large and small limuli when smaller doses of bacteria were injected. At no time did we produce active infection with relatively few organismssay 10 or 100. However, we did no careful dilution experiments to determine the minimum number of organisms which would initiate infection.
Within a few minutes after bacteria are injected into several marine crustacean including limulus, they are disseminated throughout the vascular system and tissue spaces (Table II). This is probably a mechanical dissemination, since other bacteria, including pneumococci, beta- hemolytic streptococci, and staphylococci, showed the same spread.
In each experiment several limuli at first had negative blood cultures, then as they became ill (poor blood clotting, sluggish movement) blood cultures became positive. The limulus usually died soon after. In one case an infection apparently localized in one leg, which moved poorly, and had a distended joint. Fluid taken from the joint was watery, did not clot and had few intact white cells. A culture yielded a confluent spreading type of growth. A blood culture taken from a leg on the other side was negative, and this latter blood clotted readily. The animal was found moribund 13 days later, or 24 days after the original injection, and at this time had a positive blood culture.
We feel that the data establishes this organism as a pathogen for limulus, but it is hard to evaluate its degree of pathogenicity, because so little is known about the effect of other bacteria on limulus. In two experiments in which large quantities of chromogens were injected into a total of six limuli, no progressive infection was established.
Between the summers of 1953 and 1954 the cultures of bacteria were transferred about ten times and kept at 4°C. on agar slants. Because of apparent poor infections at the beginning of the 1954 season the bacterium was "passed" through five series of limuli. During this series of passages a change from discrete round and regular colonies to confluent spreading growth with irregular edges was noted. In most cases, limuli which at first produced discrete types of colonies later furnished the spreader type, and this change preceded the death of the host by perhaps a day. A direct comparison was made between the original discrete organism and its presumed derivative. Both were prepared from peptone sea water fluid cultures and inoculated into limuli.
One of eight small (approximately 2 inches) limuli inoculated with the discrete type organism died after 6 days, whereas four of eight small limuli inoculated with the spreader died within two days. A more complete comparison is needed.
Blood obtained from normal limuli clots rapidly in glass vessels, despite the use of citrate or heparin. This clotting was extensively investigated by Leo Loeb (2, 6) who showed that the process, like that in many other marine invertebrates, has two stages. Primarily there is a clumping of the white cells which lose their granular content and form long connecting extensions (Figs. 1, 2, 3). Associated with this is the formation of a clear gel. Much of this gel, however, disappears within a half hour, even when the blood is kept under sterile conditions, and the original white cell clot remains. The gel consists of extremely fine (50 to 110 mµ in diameter) filaments. Studied in the electron microscope (7), a periodicity similar to vertebrate fibrin was not observed in the filament structure, but periodicity is difficult to determine in clots of vertebrate blood made directly with unpurified components.
|Unheated||D. 12 hr.
D. 7 days
|D. 18 hr.
D. 18 hr.
|50-55 degrees C||D. 12 hr.
D. 12 hr.
|D. 18 hr.
D. 24 hr.
|75 degrees C||Survived 14 d.
Survived 14 d.
|D. 18 hr.
D. 8 hr.
|100 degrees C||D. 12 hr.
D. 12 hr.
|D. 18 hr.
D. 2 d.
Small inoculae of concentrated bacterial suspensions frequently killed adult limuli within twelve hours. To determine whether death was caused by toxins we heated the bacteria to the temperature et which activity might be destroyed. To our surprise we found that bacterial suspensions which had been boiled for five or ten minutes still killed the limuli (Table III). Furthermore, a clear supernatant of the suspensions, after centrifugation at 15,000 rpm.,* still killed limuli and caused intravascular clotting. Large adult limuli were more readily killed by this toxin than were small, young animals even when identical amounts of toxin were given to animals which differed in mass by more than a hundredfold. One-tenth to two-tenths ml. of this clear supernatant had an effect within ten minutes after injection. The animal became stiff and tetanic and at first it was difficult to obtain blood. If some was obtained, the white cells were already clumped into stringy masses. Both old and young limuli failed to bleed when blocks of body tissue were cut out for pathological study.
|* Such crude preparations will be referred to as "crude toxin" or "toxin".|
Since granules appeared in the serum and the white cells disappeared, and since blood taken after the injection did not clot, it was assumed that an internal clot had been induced. This process was readily followed in the living animal by examining the individual leaves in the gill books. As Lankester (8) and Loeb (9) pointed out the white cell within the normal animal is a nucleated disc (Figs. 5 and 6). The cell moves freely in the serum, the cytoplasm appears packed with granules and cytoplasmic processes are rare. When traumatized or removed from the circulation the white cell loses the granules and sends out many processes (Fig. 1). Pieces of gill leaflets were cut off from the living animal after injection of the toxin and were examined by phase microscopy. Within a few minutes the white cells no longer moved freely within the vascular spaces. They stuck to the surfaces of the dividing partitions and clumped together in small groups (Figs. 7 and 8). The discharge of granules and the formation of filamentous processes occurred continuously, with frequent sudden snapping of cells as a filament broke.
The same cellular changes were seen in animals with active bacterial infections (Fig. 8) but were even more striking in the acute experiment. Dilutions of toxin which did not kill limulus nonetheless produced the changes in white cells. Furthermore, amounts of toxin which killed adult limuli caused immediate intravascular clotting in the young limuli, but the young animal in contrast to the adult recovers within a few days after the injection of the toxin.
The intravascular clotting was also produced by a partially purified preparation of Shigella toxin, but not by the injection of 1 gram of urea (10), or by plague toxin (11), fresh suspensions of pneumococci, staphylococci, or streptococci.* The urea was used because Loeb and Bodansky in some studies of urease in limulus blood found that this amount of urea killed limuli (10). We confirmed the lethal effect, but found no change in white cells. Thus there is a degree of specificity, although our early experience with the effects of large doses of non-pathogenic chromogens, etc., obtained from sea water would suggest that a variety of marine bacteria may produce the effect. Most marine bacteria are gram negative motile rods.
The pathological changes following the injection of the toxin or following active infection with bacteria was also studied in paraffin sections. Figures 9, 11, and 13 show the normal vascular system of the gills and body of young limuli. Figures 10, 12, and 14 show the effect of the injection of toxin.
This type of change in white cells following the injection of "toxin" is not confined to limulus. A variety of marine arthropods was injected and their blood cells were studied within the first few hours after injection (Table IV). In those in which it was possible to obtain cells in free fluid, we noted that the number of cells decreased markedly following the injection of toxin. Small animals were killed by the toxin. The Shigella toxin was also found lethal for lobsters and apparently produced the same intravascular dotting. Blood removed from lobsters after injection of Shigella toxin failed to dot and had few white cells.
IN VITRO EFFECTS ON SERUM
In a preliminary test for the antibacterial effects of fresh serum it was noted that the serum which had been obtained by allowing a clot to separate was in turn made viscid by the addition of bacteria. Table V shows that "elation, as measured by the lack of movement of the serum when the tube containing the serum was tilted, could readily be brought about by the inoculation of living bacteria or by the addition of dilute toxin. The ability of the serum to form such a gel was destroyed by heating the serum to 56°C. for 10 minutes (Table V).
Limulus serum contains hemocyanin as a large molecule free in the serum (12). Its presence is readily apparent by the blue color of the serum on exposure to oxygen. To determine whether it had a significant role in the formation of the clot when toxin was added, it was separated from the serum by ultracentrifugation. Since it was not possible to dilute the serum to any significant degree and still get a visible gel, most of the tests were performed with undiluted serum to which a small volume of toxin in sea water was added. Table VI with accompanying protocol shows that the molecule which is responsible for the formation of the gel upon addition of toxin is not hemocyanin and presumably is much smaller.
Protocol: Blood obtained sterilely from healthy limulus by cardiac puncture. Placed in sterile petri dish and kept at room temperature overnight. Serum withdrawn from clot and used in this and subsequent experiments after refrigeration at 4°C. One-half ml. of serum added to 1/4 ml. of dilution of toxin in artificial sea water. Test read at half hour.
As the gelation of 0.5 ml. of serum was regularly produced by 0.1 ml. of a 1/500 dilution of most preparations of toxin, it was apparent that the effect was powerful enough to explain the extensive "elation of the entire blood of large limuli, of weights in the order of 1 kg., when 0.2 ml. of undiluted toxin was injected.
Since large limuli were much more susceptible to the lethal effect of the toxin, and since effects on the circulating blood could readily be produced by amounts of toxin short of that which killed the animal, an accurate end point by which the toxin could be measured had not been available. With the in vitro test, and with the formation of a gel as the end point, it was possible to show that the toxin was not dialyzable through a collodion sac and that it maintained activity within the sac when dialyzed against forty times its volume of distilled water.
Protocol: 24 hr. Limulus sera placed in celloidin cups and centrifuged in Refrigerated Spinco Ultracentrifuge at 35,000 rpm. for 6-1/2 hours. Clear colorless supernatant separated. Sticky blue pellet resuspended in artificial sea water to reach original volume. Test for "elation by adding 0.1 ml. of fresh undiluted toxin to 0.6 ml. of serum or hemocyanin. Gels formed within half hour and remained stable for more than six hours.
The suffix L or S refers to the size of the Limulus. Most large (L) ones were 10-12 inches across, the small (S) were less than 6 inches.
A preparation of plague toxin* which killed mice failed to induce formation of a gel when 0.1 ml. of solution containing 100 gamma was added to limulus serum. As previously noted this toxin had no effect on adult living limuli. Gels were formed when living cultures of hemolytic streptococci were added to the serum, but not with pneumococci or staphylococci aureus.
The capacity of a given serum to form a gel when toxin was added to it was a constant finding. However, not all adult limulus sera formed gels when toxin was added to them. In general an individual limulus repeatedly furnished sera which formed gels, but in several cases (Table VII) this was not so. To test the idea that a serum which did not form a gel was merely quantitatively deficient in this capacity, we turned to viscosity measurements. It was possible to detect a change in viscosity by an Ostwalt viscometer ten minutes after the toxin was added to a 1/10 dilution of a positive serum. Several so-called negative sera were shown to increase in viscosity when toxin was added.
|* Courtesy of Dr. Joel Warren of the Army Medical School (14)|
The direct addition of limulus bacteria to sera of varying dilutions and observations of these mixtures under phase microscopy showed that the bacteria were immobilized in 10 minutes by the gel, and indeed even Brownian movement disappeared. Controls which consisted of the addition of bacteria to sera which had been heated to 56°C. for minutes showed no immobilization of the bacteria.
Since Metchnikov's famous observations on fungi growing inside the water flea Daphnia (4) little has been reported on invertebrate defenses against natural infections. Indeed, except in the case of the insects (15) there are few well established infectious diseases of invertebrates.
It is not safe to assume that a given bacterium is pathogenic for an invertebrate because the animal dies after it is infected with large amounts of the bacterial culture. We have observed that active and progressive infection occurs and can be demonstrated by repeated positive blood cultures and by the death of the host some days after the original infection is established. Furthermore certain pathological changes found late in the infection have been' reproduced by injection with a heat stable "toxin". Most of the work subsequent to Metchnikov's excellent studies (16) on invertebrates has followed his lead in emphasizing the role of phagocytosis. There has been no reason to challenge this emphasis since there seemed to be no "antibody" response among the invertebrates.
Limulus was chosen for these studies largely as a matter of convenience and availability but it is interesting to consider that limulus has an ancient history. Similar forms classed within the same genus have been found as far back as Paleozoic times (17). Since amoebocytes morphologically closely related to those of limulus are found in its relatives the scorpions, (8) the mechanisms of reacting to bacterial infection which are described here are presumably ancient in origin, and may represent a type of protective mechanism as basic as phagocytosis. The similarity to the so-called Schwartzman phenomenon of rabbits will be discussed later (18).
Since our evidence indicated that a massive intravascular clot follows the intravenous injection of the bacteria or of the toxin, it was rewarding to review current knowledge of clotting in invertebrates, and particularly in limulus. Although W. H. Howell was one of the first to describe the interesting jelly-like clot of limulus blood (19), most of our knowledge derives from the extensive studies by Leo Loeb (2, 6) during his repeated summer visits to Woods Hole. His work may be divided into several parts. First, he demonstrated that amoebocytes changed in form when removed from their natural vascular beds either when placed on glass (20) or when tissue was injured (21). Many granules were lost from the cytoplasm during this change in form. One feature of the change was the extrusion of many fine filaments which tied the cells together in a loose network. On top of this a gel was formed. The process could be slowed greatly by lowered temperatures, out standard anticoagulants had no effect.
Loeb followed the behavior of these cells in primary explant cultures, made simply by placing a few drops of blood, removed under sterile conditions, in hanging drop cultures (22). It was then possible to follow their movements under different conditions, (23) and in one study he showed that they collected around certain foreign bodies (9). Although in an entirely different study (24) he also showed that bacteria (staphylococci) may cause goose plasma to gel, we were unable to find any record of studies on the reaction of limulus or its cell cultures to bacteria. Finally he showed that certain enzymes were liberated from the amoebocyte of clotted blood into the serum (25).
Our study has centered around the toxicity of the bacteria and the destructive effects produced by them. Yet the evidence that the reaction has a protective effect is good. A gel formed at a traumatized area would prevent the ingress of the bacteria surrounding the animal and would also prevent further leakage of blood. We have found that the highly motile bacteria obtained from the confluent colonies are completely immobilized within a few minutes after they have been placed in a serum which can be gelled. Secondly the gels which had been formed in vitro either by addition of the toxin or by the introduction of the bacteria themselves, remained stable at room temperature for several weeks.
Certain similarities to the current concept of the Schwartzman phenomenon in rabbits are striking. Like this reaction, a clumping of white cells within the vascular channels follows the injection of the toxin (26, 27). The formation of a gel in the limulus serum both in vitro and in vivo after the addition of the toxin is similar to the recent description of the changes in the generalized Schwartzman phenomenon in which there is intravascular deposition of fibrinoid following the second injection of toxin (28, 29). The substances used to provoke the present reaction are similar to those used in the Schwartzman. It is further amusing that in both cases young animals (30) are less susceptible to the effect of the toxin.
However, the preliminary injection has not been found important in the limulus reaction, whereas some preparatory treatment is all important in the Schwartzman. In the limulus reaction it is impossible to know the degree of sensitization already achieved by the wild-caught animals when they are tested, since they are continually exposed to a gram negative flora of motile bacteria.
In one recent case report (31), an acutely ill patient was found to have a failure of blood clotting due to lack of fibrinogenassociated with bacteremia with Escherichia cold and acute yellow atrophy. Although Loeb believed fibrinogen to be absent from Limulus the question as to whether the secondary clot is due to a fibrinogen-fibrin type of conversion needs further study In the horseshoe crab, Limulus polyphemus an active bacterial infection which eventually killed the host was experimentally produced by gram negative marine bacteria. Its course was followed by blood cultures. The main pathological effect is referable to generalized intravascular clotting.
A heat stable derivative was obtained by boiling a suspension of the bacteria and separating the bacteria by centrifugation. The "toxin" killed limuli a few hours after injection. The toxin caused extensive intravascular clotting a few minutes after injection. This effect was observed by direct inspection of the living amoebocytes in the vascular compartment of the gill leaflets. These cells lost their granules, stuck to the wall of the compartment, and sent out long thin processes similar to those seen on extravascular clotting. Shigella toxin produced this same effect, but several gram positive bacteria failed to produce it. Although limuli were readily killed by urea, no intravascular clot was produced. Large (old) limuli were much more susceptible than young.
Most of the sera from adult limuli formed a gel when combined in vitro with dilutions of the bacterial toxin or of the living bacteria. This gel immobilized the bacteria but did not kill them. The gel or clot was stable and tough and remained so for several weeks at room temperature. It is thought that this may represent an important protective mechanism against bacterial infections, in that immobilization of entering bacteria would delimit the infectious process.
Similar changes in white cell behavior were observed after the injection of toxin in several other arthropods.
The similarity of this reaction to the Schwartzman phenomenon is noted.
1. ZoBell C. E.: Marine Microbiology. A Monograph on Hydrobiology. Chronica Botanica Co., Waltham, Mass., 1946.
2. LOEB, L.: Über die Blutgerinnung bei Wirbellosen. Biochem. Ztschr., 1910, 24: 478.
3. SAWYER, W. H. AND TAYLOR, C. C.: The effect of shell disease on the gills and chitin of the lobster (Homarus americanus Res. Bull. No. 1, Dept. Sea and Shore Fisheries, 1949.
4. GETCHELL, J. S. Personal communication.
5. Inman O. L.: A pathogenic luminescent bacterium. Biol. Bull. Bost., 1927, 53: 197.
6. LOEB, L.: Über die Zweite Gerinnung des Blutes von Limulus. Biochem. Ztschr., 1909, 16:157.
7. BANG, F. B. AND FROST, J. L.: The toxin effect of a marine bacterium on limulus and the formation of blood clots. (Abstract) Biol. Bull. Bost., 1953, 105: 361.
8. LANKESTER, E. RAY: On the skeleto-trophic tissues and coxal glands of Limulus, Scorpio and Mygale. Quart. J. Biol., 1884, 24: 129.
9. LOEB, L.: On the blood Iymph cells and inflammatory processes of limulus. J. Med. Res., 1902, 7: 145.
10. LOEB, L. AND BODANSKY, O.: The specific effect of salts in the extraction of urease from the amoebocytes of limulus. J. Biol. Chem., 1927, 77: 415.
11. WARREN, J. Personal communication.
12. PROSSER, C. L., ET AL.: Comparative Physiology. W. B. Saunders Co., Philadelphia, Pa. 1950.
13. CLUFF, L.: Immunochemical study of a bacterial endotoxin: Shigella Flexneri Type Z. J. Exper. Med., 1954, 100: 391.
14. Metchnikov E.: Über eine Sprosspelzkrankeit der Daphnien. Virchows Arch., 1884, 96: 177.
15. Steinhaus, E. A.: Principles of Insect Pathology, McGraw Hill Book Co., New York, N.Y., lsted.
16. METCHNIKOV, E.: Lectures on Comparative Pathology of Inflammation K. Paul, Trench, Trübner Co., Ltd. London England, 1893.
17. MOORE, R. C., Lalicher, C. C., AND FISHER, A. G.: Invertebrate Fossils, McGraw Hill Book Co., New York, N. Y. 1952.
18. Schwartzman G.: The Phenomenon of Local Tissue Reactivity. Paul R. Hoeber, Inc. New York, N. Y., 1937.
19. Howell W. H.: Observations upon the chemical composition and coagulation of the
blood of Limulus polyphemus, Callinectes hastatus, and Cucumaria sp. Johns Hopkins
University Circulars, 1885, No. 43: 4.
20. LOEB, L.: Amoebocyte tissue and amoeboid movement. Protoplasma, 1928, 4: 596.
21. LOEB, L.: The movements of the amoebocytes and the experimental production of amoebocytes (cell-fibrin) tissue. Washington University Studies, 1920, 8: 3.
22. LOEB, L.: Amoeboid movement and agglutination in amoebocytes of limulus and the relation of these processes to tissue formation and thrombosis. Protoplasma, 1927, 2:512.
23. LOEB, L. AND Blanchard, K. C.: The effect of various salts on the outgrowth from experimental amoebocyte tissue near the isoelectric point and with the addition of acid or alkali. Am. J. Phys., 1922, 60: 277.
24. LOEB, L.: The influence of certain bacteria on the coagulation of the blood. J. Med. Res., 1903, 10: 407.
25. LOEB, L. AND BODANSKY, O.: The occurrence of urease in the blood cells, blood plasma, and tissues of limulus. J. Biol. Chem., 1926, 67: 79.
26. STETSON, C. A.: Studies on the mechanism of the Schwartzman phenomenon: Certain factors involved in the production of the local hemorrhagic necrosis. J. Exper. Med., 1951, 93: 489.
27. Cluff, L. AND BERTHRONG, M.: The inhibition of the local Schwartzman reaction by heparin. Bull. Johns Hopkins Hosp., 1953, 92: 353.
28. THOMAS, L., BRUNSON, J. AND Smith R. T.: Studies on the generalized Schwartzman reaction. VI. Production of the reaction by the synergistic action of endotoxin with three synthetic acidic polymers (sodium polyenethol sulfonate, dextran sulfate, and sodium polyvinyl alcohol sulfonate). J. Exper. Med., 1955, 102: 249.
29. Thomas L. AND GOOD, R.: Studies on the generalized Schwartzman reaction. J. Exper. Med., 1952, 96:605.
30. FREUND, J. AND HOSMER, E. P.: Hemorrhagic reaction at the site of injection of toxins after intravenous injection of starch in young and adult rabbits. J. Immunol., 1935, 29:279.
31. Conley, C. L., RATNOFF, O. D. AND HARTMANN, R. C.: Studies of afibrinogenemia I. Afibrmogenemia in a patient with septic abortion, acute yellow atrophy of the liver and bacteremia due to E. coli. Bull. Johns Hopkins Hosp., 1951, 88: 402.
The electron microscope preparations and photographs were made by Mr. James Frost to whom we are indebted for their inclusion in the paper.