Effects of Escherichia coli on germ-free and gnotobiotic pigs

J. COMP. PATH.

1970, VOL. 80.

EFFECTS

OF

ESCHERICHZA AND

II.

233

SERUM

COLZ

GNOTOBIOTIC PROTEINS

AND

ON PIGS

GERM-FREE

ANTIBODIES

BY

P. PORTER

and R. KENWORTHY

Unileoer Research Laboratory,

Colworth

Howe,

Shambrook,

Bedford

INTRODUCTION

Considerable progress in studies of the immunopathogenesis of enteric disease could be made by understanding the mechanism of the immune response specifically directed against antigens-bacterial or dietary-as they occur in the alimentary tract. The germ-free colostrum-deprived pig in many ways provides a good model for such studies since the animal may be reared without the complications of a wide range of antibody activity passively acquired by the colostrum or transplacentally from the sow. In previously reported studies (Kenworthy, 1970) detailed histological investigation by light and electron microscopy have demonstrated in gnotobiotic pigs fed a diet based on cows’ milk that the intestinal response to infection with E. co& serotype 0141: K85 a,c:H4 was a classical inflammatory reaction at the mucosal level. The lymphocytes and plasma cells demonstrated in the lamina propria of these.animals are potential sourcesof antibody. Thus any changes in the agammaglobulinaemic state of the gnotobiotic pig may be a direct response generated in the intestinal tissue against antigens derived from the alimentary tract, which includes contributions from bacterial and dietary sources. An investigation of immunoglobulins and antibodies synthesisedin the intestinal immune response to these antigens is described in this paper; the main groups investigated are antibodies to E. coli polysaccharides, bovine milk whey proteins and casein. METHODS

Microelectrophoresis Sera and fractions were examined by cellulose acetate electrophoresisand immunoelectrophoresis (Porter, 1964). Disc electrophoresiswas carried out in polyacrylamide gels (Orstein and Davis, 1964). Protein Determination Protein determinations were made from celluloseacetate electrophoretic separations of serum components (Briere and Mull, 1964) and total protein was determined by micro Kjeldahl. Gel Filtration Chromatografihy The techniques of Porter, Porter and Shanberge (1967) for thin layer and column chromatography in Sephadex gels were used.

234 Ion Exchange

IMMUNOGLOBULINS

IN

GNOTOBIOTIC

PIGS

Chromatography

Anion exchange chromatography was carried out on diethyl-aminoethyl-cellulose (Whatman DE-11) using the stepwise elution technique of Augustin and Hayward (1960). Five buffers were used (1) 0.01 M NaH,PO, pH 7.6; (2) 0.02 M NaH,PO, pH 6.3; (3) 0.05 M NaH,P04pH 4.5; (4) 0.3 M NaH,PO,; (5) 0.4 M NaH,PO, 2M NaCl. Determination of Immunoglobulin ,4gammaglobulinaemic Serum

Concentrations

in

The levels of immunoglobulins in gnotobiotic pigs are too low to be measured by conventional electrophoretic techniques and therefore immunological methods were devised. The first requirement was to obtain the characterised immunoglobulins in pure form. These were prepared as under. IgG isolation and purification. A crude immunoglobulin preparation was obtained by precipitation from pig serum using 10 per cent. w/v Na,SO,. This material was washed and resuspendedin 0.01 M phosphate buffer pH 7.5 and fractionated on DEAE anion exchange cellulose. A pure preparation of 7s IgG was present in the “fall through fractions”. IgM isolation and purification. This immunoglobulin is present in serum in very low concentrations; it was prepared in more concentrated form by an anion exchange batch processof 500 ml. of serum using DEAE cellulose. The serum was buffered by dialysis using 0.05 M phosphate buffer pH 4.5; DEAE cellulose equilibrated with this buffer was stirred in and the mixture was filtered. The cellulose was washed free of excessprotein with 0.05 M phosphate buffer and the macroglobulin fraction was eluted with 0.3 M NaH,PO,. The 19s IgM was isolated from this fraction by chromatography on Sephadex G200. IgA isolation and purification. A crude immunoglobulin preparation was obtained from sow colostrum by taking the eluates at the exclusion limits after fractionation on Sephadex G150. IgA was separated from the macroglobulins by chromatography on DEAE cellulose, being eluted with phosphate buffer pH 7.6 and 0.125 M NaCl. Preparation of specific antisera. The immunoglobulin preparations were tested for purity by immunoelectrophoresis against a rabbit antiserum to total pig serum proteins. Antisera were prepared in rabbits using the technique and immunisation schedulesdescribed by Porter (1964). There is a degree of cross-reactivity between the isolated immunoglobulins and their antisera owing to common antigenic determinants on the molecules.This was removed by absorption with the pure cross-reacting immunoglobulin in order to prepare a specific antiserum. Non-specific activity to other pig serum proteins was removed from the antisera by absorption with precolostral piglet serum. Immunochemical determination of immunoglobulins. The diffusion gradient precipitin technique of Hayward and Augustin (1957) for the quantitative assay of antigens was used. The test was calibrated with known concentrations of the specific immunoglobulin determined by Kjeldahl. A diffusion gradient of the specific antiserum was establishedin agar gel. The end point for the assay of antigen (in this caseimmunoglobulin IgG, IgA or IgM) was determined as the highest dilution which produced an observable precipitin line against the antibody gradient. The technique was calibrated for each assay with a standard solution of the purified protein. A typical assaygiving detail of the method is shown in Fig. 2. Antibody Assays Bacterial antibodfes.

Haemagglutination and antiglobulin haemagglutination tests were. used as described by Buxton and Thomlinson (1961). Sheep red cells were modified wrth ,?$ colt polysaccharides prepared by the phenol method of Westphal, Luderitz and Bister (1952) or by ultrasonic disintegration of a heavy suspensionof

P.

PORTER

AND

R.

KENWORTHY

235

bacteria in a 1.25 per cent. formalin solution in saline (10 min. using a 5 mm. probe at 1.2 amp.). rlntibodies to bovine casein and milk whey proteins. Casein and milk whey soluble proteins were prepared from untreated cows’ milk by ultracentrifugation (Bohren and Wenner, 1961). The casein gel was transformed to colloidal alkali caseinate by ion exchange with Na-H-casein Amberlite IR 120. A 2 per cent. suspension of casein was stirred with Na-H-Amberlite for 40 to 50 min. at room temperature, after which time the Na caseinate was decanted from the Amberlite cooled in ice-water and centrifuged at 15,000 r.p.m. for 20 min. The colloidal solution was treated with N/100 NaOH till clear and buffered to pH 6.4 with @15 M KH,PO*. The whey protein was adjusted to a protein concentration of approximately 1 g./lOO ml. by dilution with 0.15 M phosphate buffer pH 6.4. Antibodies to casein and milk whey antigens were assayed by the sensitised tanned sheep erythrocyte agglutination technique. Washed sheep erythrocytes were treated with tannic acid at a concentration of 1:40,000 in buffered saline pH 7.2 at 37OC. for 10 min. The tanned cells were washed and suspended in 200 vol. of saline pH 6.4 and sensitised with the required protein fraction by adding 1 vol. of protein solution to 4 vol. of tanned cell suspension and maintaining the suspension at 37OC. for 30 min. The sensitised cells were centrifuged at 1,500 r.p.m., washed with a solution or rabbit serum (1: 100) in saline and resuspended in 10 vol. of the rabbit serum solution, which acts as a stabiliser to the sensitised cells. The cells were added to equal volumes of the test. Animals Nineteen out of 23 piglets from 4 litters whose derivation has been previously described (Kenworthy, 1970) were included in the present studies and the relevant data for their identification is included in Table 1. To avoid confusion in this paper, the term germ-free has been used to denote an animal free from all organisms which would be revealed by the applied screening techniques. The term gnotobiote is used to denote an animal harbouring one or more known organisms, introduced either intentionally or accidentally (Kenworthy, 1970). RESULTS

Serum Proteins

in Germ-Free

and Gnotobiotic

Pigs

Preliminary studies of the serum proteins using thin layer gel filtration (Fig. 3) and polyacrylamide disc electrophoresis (Fig. 4) indicated the extent of hypogammaglobulinaemia in these animals. Germ-free animals, even at 30 days, had no obvious development of immunoglobulin demonstrable by these techniques indicating that dietary antigens had brought about no apparent change in the gamma globulin content of the serum. Bacterial contamination stimulated immunoglobulin synthesis. This is readily evident in the comparative studies by thin layer gel filtration (Fig. 2) which shows an obvious increase in protein in the 7S immunoglobulin region. With the exception of the gamma globulins, these animals showed some maturation of the serum profile. For example, the albumin level increased and the alpha “fetuin,” which is a characteristic foetal protein, disappeared. These changes are demonstrated in the cellulose acetate electrophoresis studies (Fig. 1) and the total proteins measured by this technique are given in Table 1. The increase in albumin accounts for most of the change in the serum protein profile of these animals from birth. In the 30-day-old germ-free pig, albumin may account for almost three-quarters of the total serum protein giving an A/G ratio of E

236

IMMUNOGLOBULINS

IN

GNOTOBIOTIC

TABLE DETAILS

Infectid with

Pig no.

E. coli

OF GERM-FREE

Age (days) when infected with E. coli

6 7

08 -

s s 10

16 17

0141 08 0141

::

OR GNOTOBIOTIC

Additional

1 PIGS WITH

micmflora *

0141 0141

23

08 0141

:t

OF SERUM

Age (days) at autopsr

Serum Alb.

PROTEINS

proteins a

g./ 100 ml. p+y Total

3”: 29 ::

2.63 2.80 2.50 2.90 2.80 3.70 2.50 1.67

0.58 0.48 0.51 0.70 0.96 0.90 2.53 0.24

0.45 0.35 0.41 I.55 1.34 0.70 1.58 1.0

3.66 3.63 3.42 5.15 5.10 5.30 4.32 5.20

29

3.60

0.70

0.50

4.80

.:;4 12

3.10 1.0 2.01

0.70 1.37 0.70

0.86 0.70 1.19

4.50 3.23 3.90

31

3.80

0.70

0.70

5.20

21

2.08

0.40

I-53

4.01

B. cereus, B. subtilis B. cereus, B. sphericus, B. subtilis

!i 24 31

2.19 3.60 2.71 3.80

0.70 0.90 0.72 0.80

0.80 1~84 5.30 4.73 2.55 5.98 0.70 5.30

L. Plan&rum

37

3.20

0.80

0.80

G+

3”:

coccus

G+rod S Yeast G-rod, G+rod S, Gfcoccus G+COCCUS,G+rod S G+rod non S, Yeast G+COCCUS,G+rod

22 2;

ASSAYS

30

NIL NIL NIL NIL NIL

S

:fi

:i zi

PIGS

4.80

* Additional microflora! probably due to infection at birth (with exception of L. plantancm which was deliberate mfection at 11 days of age). G + gram positive G - gram negative S sporing

approximately 3 compared with 1.5 in the conventional animal. The gamma globulin development even in gnotobiotic pigs harbouring organisms for more than 3 weeks was too low to measure satisfactorily by cellulose acetate electrophoretic techniques using calorimetry. Immunoelectrophoresis confirmed the findings of the thin layer studies indicating 7S IgG development (Fig. 5) in the gnotobiote and virtual absence in the germ-free animal. Immunoglobulins and Antibodies in Germ-Free and Gnotobiotic Pigs Quantitative estimations of serum immunoglobulin levels were carried out by the gradient diffusion precipitin technique using specific antisera. These assays were obtained as an end point at a given serial dilution and this was calculated as a protein concentration for immunoglobulin by reference to the end point for a standard solution of the pure immunoglobulin calibrated by biuret. No attempts were made to prepare intermediate dilutions to narrow the end point because of the need to conserve the reagent antisera. Although an accurate end point was not obtained the assays do give a good index of the development of a given immunoglobulin at the very low levels demonstrated in gnotobiotic pigs. The results of the assaysare presented in Table 2. The assaysconfirmed the previous impressionsgained by thin layer and electrophoretic studies that there was little or no immunoglobulin development in the germ-free pig and that the predominant immunoglobuhn development in the

P. PORTER

AND Fig.

1.

New born

Zday old suckled

Germfree@

Electrophoretic

profiles

days)

after cellulose

acetate

electrophoresis

TABLE

ASSAYS OF IMMUN~GL~BULINS

Gradient I& .%b. of pig

mg./IOO

16 8 16 4 2

f 1 2 3 Germ-free

j

AND

D&rim Id

1”;

’ i Gnotobiotic i 9 no. E. coii I. 10

i 16 4 256 4 128 8 128 64 2%

r;; 16

128 64

uls

Endo

-

237

R. KENWORTHY

E. cou

ml.

-

5.5 2.8 5.5 -

-

-

IN GERM-FREE

AND

-

10

--

10

10

--

20 10 5

S.

10 10

-

PIGS

Tanned cell haemagglutination

Milk

u&y ii

Casein ::

24

-

5

-

3.3

11

320

6% 6.6

Ii22

6iO

4.8 ;:“5

221

6.6 6.6

it

polysaccharides polysaccharide

Ponceau

GNOTOBIOTIC

Antiglobulin haenqglutination tl/sOHl Endo u/s 08 Endo -

5.5

-

with

2

ANTIBODIES

Precipitin assay rgnr

of sera, staining

640 1280 160

20 20 -

itI -

24 -.-

2;

5

12 12 12

12 24 24 -

ii

::: -

640 320 320

derived by ultrasonic dkentegration derived by phenol method, Westphal

of cells et al. (1952)

-

238

IMMUNOGLOBULINS

IN

GNOTOBIOTIC

PIGS

gnotobiotes was associated with 7s IgG. The level of IgG in the 30-day-old germ-free pig remained essentially the same as the new born (2 to 16 mg. / 100 ml.) and probably relates to that gained in utero. This is approximately two hundred times less than the level in the conventional pig of the same age. The gnotobiotes showed an IgG development which in general increased with the time of exposure to bacteria. Immunoglobulins IgA and IgM could not be detected in the serum of the new born pig. Immunoglobulin IgA could not be measured in the Serum of germ-free animals (1 to 5) or in gnotobiotes (7 to 10) which had not been infected with E. c&i. In addition, IgA could not be measured in the serum of the two gnotobiotes infected with E. coli (15 and 17). These were the youngest animals in the E. cob group and had short periods of infection (2 and 5 days respectively). In the other gnotobiotes which were aged 24 to 37 days IgA was assayed in the range 2.2 to 6.6 mg. / 100 ml. For comparison the serum level of IgA in conventional animals of the same age was 38 rt4.6 mg./lOO ml. (6 samples). Thus the development of IgA in the gnotobiote was never greater than one-sixth of the normal level. The findings for IgM were similar to those for IgA, but IgM was measurable in the three germ-free animals 1, 2 and 3 ; it was assayed in the range 3 to 22 mg./lOO ml. For comparison the level of IgM recorded in conventional animals was 96+34 mg./lOO ml. (6 samples). Antibodies to the dietary antigens in bovine milk whey and casein were assayed by the tanned cell haemagglutination technique. Antibodies to bacterial antigens were assayed by the antiglobulin haemagglutination technique using E. coli endotoxin and polysaccharides extracted after ultrasonic disintegration. The results are presented in Table 2. There was little or no measurable antibody response to the dietary antigens. This was probably due to the very low levels of available soluble antigen left in the milk preparation after autoclaving and is demonstrated in Fig. 6 where comparative studies of whey fractions of T.T. milk, spray dried milks and autoclaved milk are made by electrophoresis in polyacrylamide gels. Antibody to dietary antigen was only detected in animals which had been receiving the diet for more than approximately 4 weeks. There was a poor immune response measurable to endotoxin; it was detectable only as incomplete antibody in the antiglobulin haemagglutination test. Much higher antibody titres were obtained in the antiglobulin haemagglutination test using E. coli polysaccharides prepared by ultrasonic disintegration, thus reflecting the immune response to other undefined bacterial antigens. DISCUSSION

It should be stressed that the present investigations were carried out as a supporting study to experiments designed for other purposes (Kenworthy, 1970). The conditions relating to the samples obtained are too varied to take full advantage of the possible control afforded for immunological experiments using the germ-free animal. Nevertheless, a number of interesting points arise out of these studies which warrant further investigation of the immune response in the gnotobiotic pig.

P. PORTER

AND

R. KENWORTHY

239

The gnotobiotic pigs do not appear to have exhibited an immunoglobulin development characteristic of that described for inoculated antigens in conventional animals and in man. The reported sequential synthesis of antibody commences predominantly with 19s immunoglobulin followed by increasing synthesis of 7s antibody (Benedict, Brown and Ayenger, 1962 ; LoSpalloto, Miller, Dorward and Fink, 1962). Initial synthesis of immunoglobulin in gnotobiotic pigs used in the present studies appears predominantly as 7s IgG, and 18s IgM is either not detectable, or present at extremely low levels. This latter finding correlates with assays of antibody to E. coli endotoxin which was low in most animals infected with either or both strains of E. coli under investigation. This antibody has been studied in conventional animals and demonstrated to be mainly associated with 18s IgM (Porter and Kenworthy, 1969). Kohler and Bohl (1966) in similar studies with gnotobiotic pigs were unable to detect any 0 somatic antibody by the bacterial agglutination test and only occasional low antibody titres by the direct haemagglutination test. Two possible explanations can be offered for these observations. Firstly, there is some evidence that in the pig pre-experience of an antibody is required for its synthesis (Segre and Kaeberle, 1962). Precolostral piglet serum is almost entirely deficient of immunoglobins, but a component having shared antigenic determinants with IgG is present in concentrations less than 50 pg./ml. (Stertzl, Kostka, Riha and Mandel, 1960). Thus without pre-experience of IgM there may be a poor capacity for synthesis. It is interesting in this respect that Kim, Bradley and Watson (1966) working with colostrum-deprived pigs claim that antibody synthesis commences with a heavy 19s antibody which is antigenically distinct from IgM but identical with 7s IgG. The second possibility is that the route of administration of antigen may affect the nature of the immune response, In recent studies in the chicken (Duffus and Allan, 1968) oral infection with S. gallinarum produced a simultaneous early response of IgG and IgM antibodies to somatic antigen, and IgG predominated as the course of the immune response was followed. After an intramuscular inoculation most of the activity resided in the IgM fraction. Thus the development of immunoglobulin in the serum of the gnotobiotic pigs compares favourably with the findings of Duffus and Allan for orally infected chickens. The histological studies of intestinal tissue of gnotobiotic pigs demonstrated a marked increase in plasma cells in the deeper layers of the mucous membrane in response to oral inoculation with E. coli 0141 (Kenworthy, 1970). Immunofluorescent studies in man have demonstrated that the plasma cells in the lamina propria of the gastro-intestinal tract contain predominantly IgA, and presumably synthesise and store the immunoglobulin (Crabbe, Carbonara and Heremans, 1965; Galzayd, Kraft and Khmer, 1968). This has been suggested as the first line of defence against intestinal flora (Tomasi, 1967). Similar observations of the IgA system have been recorded for the conventional pig (Porter and Allen, 1969). There is evidence that the plasma cells contribute some of this immuneglobulin to the circulation as well as secrete it into the lumen. In view of the plasma cell response in the intestine of the gnotobiote it is surprising that the serum levels of IgA should be so low compared with normal animals and com-

240

IMMUNOGLOBULINS

IN

GNOTOBIOTIC

PIGS

pared with the serological development of IgG. Possibly IgA synthesis in intestinal plasma cells contributes less to the circulation than is generally appreciated. The acute inflammatory response previously described appears to be due to the bacterial component of the intestine and not to the diet. No inflammatory reaction occurred in germ-free animals on the same diet and the immune response to the diet was very low in all animals, presumably due to the effect of autoclaving on antigenicity. However, this does not exclude the diet as a source of bacterial substrate and hence its participation in the synthesis of bacterial antigens. There is always the possibility that the dietary substrate may have a bearing on virulence. It will be interesting to ascertain the nature of the immunoglobulin and antigens which participate in the acute reaction. From the present studies the immunoglobulin which suggests itself is IgG. Because of its smaller molecular size and greater abundance, both in the gnotobiote and conventional animals, this immunoglobulin appears at extravascular sites with every possibility of contributing to immune reactions of a deleterious nature. SUMMARY

Serological

studies of colostrum-deprived

gnotobiotic

pigs orally

infected

with

E. cc& (0141: K85 a,c: H4) and reared on a diet based on cows’ milk, indicate that the immune response is dominated by the production of 7s IgG. Very low levels of IgA and IgM were detectable and assayed with specific rabbit antisera. There was a poor antibody response to bacterial endotoxin attributable to the poor response in the IgM fraction and little detectable antibody to the dietary antigens in bovine milk whey and casein attributable to poor availability of antigen after autoclave treatment of the diet. There is no measurable evidence of transplacental transfer of IgA or IgM, but IgG is detectable in the serum of the newborn pig. The level of IgG in the serum of the germ-free pig remained essentially the same as the new born, IgM was measured in the serum of 3 out of 5 germ-free pigs, but IgA could not be assayed. The observations are discussed in relation to the cellular response observed in the lamina propria of the intestine. ACKNOWLEDGMENT

We are indebted to Mrs. L. Pugh and Mr. M. E. Prior for valuable assistance in conducting these studies.

technical

REFERENCES

Augustin, R., and Hayward, M. J. (1960). Immunology, 3, 45. Benedict, A. A., Brown, R., and Ayenger, R. I. (1962). J. ext. Med., 115, 196. Bohren, H. W., and Wenner, J. R. (1961). J. dairy Sci., 44, 1213. Briere, R. O., and Mull., J. D. (1964). Amer. J. clin. Path., 42, 547. Buxton, A., and Thomlmson, J. R. (1961). Res. vet. Sk., 2, 73. Crabbe, P. A., Carbonora, A. O., and Heremans, J. F. (1965). Lab. Invest., 14, 235. Duffus, P. H., and Allan, D. (1968). Immunol., 15, 653. Galzayd, E. A., Kraft, S. C., and Kirsner, J. B. (1968). Gastroenterol., 54, 334. Hayward, M. J., and Augustin, R. (1957). Int. Arch. Allerg., 11, 192. Kenworthy, R. (1970). J. camp. Path., 80, 53.

P.

PORTER

AND

R.

KEPI:M’ORTHT

Fig. Fig.

2 (top,. Quantitative assay of immunoglobulins bv gradient diffusion precipitin test in tubes. 3 ilefl cen&e). Thin layer gel filtration studies of sera from germ-free, gnotobiote and conventional pigs, demonstrating the effect of bacterial contamination of the germ-free animal on the clev~~lopment of 7S immunoglobulins. ii. New-born pig; B. Germ-free pig; C. Sormal pig; D. GnotobirAc piq: E. Gnotobiotic pig. 1. r*,M & IgM; 2. 7S IgG; 3. Albumin. Fig. 4 (ri,.$ cen~e). Polyacrylamide disc electophoretic studies of srra from germ-free, gnotobiote and conventional pigs. A. Duocontammated gnotobiote E. coli 0141 and R. coli 08 (30-day-old); B. Sow rolostral whey; C. 2-day-old piglet (colostrum fedj ; D. New-horn piglet (colostrum deprived, : and E. Germ-free piglet (30-day-old). l*‘ig. .i lbottorn left). Immunoelectrophoretograms ofsera from germ-fire and gnotobiotic pigs. Gl, monocontaminated gnotobiote E. coli 0141 ; GO. germ-free: G2. duocontaminated gnotobiott E. co/i 0141 and E. coli 08. Fig. 6 (hotton ri&). Polyacrylamidc disc electrophowtic studies of milk proteins after various treatments demonstrating the effect of autoclaving on the level of soluble protein in v.hcy. 1’ pastcurized, G Golden Glo\v spray dried, U untreated. G .\ Autoclavcd Goldrn Glo\r. 131 ISah? fbod 1 *. B2 Babyfood 21, S sow serum. * Cow

and

Gate,

Guildford.

Surrey.

t S.bf..A.

\Vveth

I.td..

Havant,

Hants.

P. PORTER

AND

R. KENWORTHY

Kim, Y. B., Bradley, S. G., and Watson, D. W. (1966). J. Zmmunol., 97, 189. Kohler, E. K., and Bohl, E. H. (1966). Can. 1. camp. Med. vet. Sci., 30, 169. LoSplal:u5to,J., Miller, W., Dot-ward, B., and Fink, C. W. (1962). J. clin. Invest., Orstein, L., and Davis, B. J. (1964). Ann. N.Y. Acad. Sci. N.Y., 121, 321. Porter, P. (1964). 1. camp. Path., 74, 108. Porter, P., Porter, M. C., and Shanberge, J. N. (1967). Biochemistry, 6, 1854. Porter, P., and Kenworthy, (1969). J. camp. Path., 79, 553. Porter, P., and Allen, W. D. (1969). Experentia, in press. Segre, D., and Kaeberle, M. L. (1962). 1. ZmmunoZ., 89, 790. Stertzl, J., Kostka, J., Riha, I., and Mandel, I,. (1960). Folk. Microbial., 5, 29. Tomasi, T. B. (1969). Hosp. Prac., 2, 26. Westphal, O., Luderitz, D., and Bister, F. (1952). Z. Naturforsch., 7, 148. [Received

for publication,

July 25th, 19691

241

41,