ty of Bristol, E. Mary Cooke, Department of Microbi- ology, University of Leeds, and Bernard Rowe, Cen- ... Wehr/Baden, Germany. Specially pure sodium lauryl.
Vol. 39, No. 1
INFECTION AND IMMUNITY, Jan. 1983, p. 315-335
0019-9567/83/010315-21$02.00/0 Copyright C 1983, American Society for Microbiology
Six Widespread Bacterial Clones Among Escherichia coli Kl Isolates M. ACHTMAN,l* A.
MERCER,1 B. KUSECEK,' A. POHL,' M. HEUZENROEDER,' W. AARONSON,2 A. SUTTON,2 AND R. P. SILVER2
Max-Planck-Institut fur Molekulare Genetik, D-1000 Berlin 33, Germany, ' and Division of Bacterial Products, Bureau of Biologics, Food and Drug Administration, Bethesda, Maryland 202052 Received 24 May 1982/Accepted 7 September 1982
Variable properties among Escherichia coli isolates include serotype, electro-
phoretic migration of major outer membrane proteins, metabolic properties, production of hemolysin or colicin or both, and plasmid content. These characteristics were compared in E. coli strains of capsular types Kl, K5, K92, and K100 and in non-encapsulated isolates. The 234 bacterial strains from the United States and Europe which we studied had been isolated from healthy or diseased individuals recently or as long ago as 1941. Regardless of source, most 07:K1, 016:K1, and 075:K100 isolates could be assigned to three unique, serotypespecific groups, which were interpreted as representing three bacterial clones. Two bacterial (sub)clones each were discerned among the 018:K1 and 018:K5 isolates, and two further, distinct clones were discerned among the 01:K1 isolates. The implications of these results for epidemiological analyses and for virulence are discussed. Escherichia coli isolates have been classified according to their 0 (lipopolysaccharide [LPS]), K (capsular), and H (flagellar) antigens (9, 18, 25). The uniformity of the bacteria within any one serotype remains unclear. In some cases, epidemiological analysis has been aided by assaying biochemical reactions (biotyping) as additional or alternative criteria for subdividing bacterial isolates (8, 9). Recently, the characteristic migration patterns of outer membrane proteins during sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) has been used for subdividing Neisseria meningitidis (10) and Haemophilus influenzae (7, 20) isolates. Because different migration patterns of outer membrane proteins have been observed with isolates of E. coli (28), such an analysis might have been expected to yield information on the relatedness of E. coli isolates. However, no unique relationships between these patterns and 0 serotype or pathogenicity have been found (15, 27). Given the plethora of E. coli serotypes, it is interesting to note that most extraintestinal E. coli infections of humans are associated with only few of the 0 and K groups (25); 80% of the E. coli strains isolated from newborn patients with meningitis were found to be of the Kl capsular type (29), and more than one-half of these isolates were either 01:K1, 07:K1, 016:K1, or 018:K1 (32). In addition, Kl bacteria are frequent among E. coli strains isolated from patients with urinary tract infections and
sepsis (16, 21), thus making this one of the major groups of E. coli which can cause extraintestinal disease in humans. Kl bacteria with the same 0 serotypes can also be isolated from stools of healthy individuals; it remains unclear whether these isolates are as virulent as the isolates from diseased individuals. One analysis of the plasmid contents of Kl bacteria indicated that all 018:K1 bacteria tested contained a common plasmid and might represent one bacterial clone (34). The same analysis suggested that 07:K1 bacteria might also be closely related to each other. In the work reported here, Kl bacteria were analyzed by using SDS-PAGE as an alternative method to detect clonal relationships. As controls, three other groups of encapsulated E. coli, which possessed K5, K92, and K100 capsules were also tested. The conclusions drawn from SDS-PAGE of membrane proteins and those drawn from the biotypes of the bacterial isolates studied were compared. (A preliminary report of the results was presented at the International Symposium on Bacterial Vaccines, Bethesda, Md. 15-18 September 1980 [3].) MATERIALS AND METHODS Bacterial strains. Bacterial strains isolated recently in several countries were obtained from the following individuals: P. Helena Makela, Central Public Health Laboratory, Helsinki, Finland; Ben Lugtenberg, Vak315
316
ACHTMAN ET AL.
INFECT. IMMUN.
groep moleculaire Celbiologie, University of Utrecht, in ornithine decarboxylase test broth (E. Merck AG, Utrecht, The Netherlands; Bertil Kaijser, Institute of Darmstadt, Germany) containing 0.5% ornithine. InMedical Microbiology, University of Goteborg, Gote- dole production was tested as described by Edwards borg, Sweden; Klaus Jann, Max-Planck-Institut fur and Ewing (9), except that growth was for 20 h. Antiserum agar plates (33) contained (per liter of Immunbiologie, Freiburg, West Germany; A Thunert, Zentralinstitut fur Versuchstiere, Hannover, West water) 11.6 g of minimal broth Davis without dextrose Germany; and Stanley Falkow, Department of Micro- (Difco), 2 g of Casamino Acids (Difco), 12.0 g of biology and Immunology, University of Washington, agarose (type I; low EEO; Sigma Chemical Co., St. Seattle, Wash. English strains were obtained from Louis, Mo.) and 0.2% (wt/vol) glucose. After the Peter Bennett, Department of Bacteriology, Universi- temperature dropped to 50°C, antiserum was added to ty of Bristol, E. Mary Cooke, Department of Microbi- a final concentration of 7.5% for horse 46 anti-N. ology, University of Leeds, and Bernard Rowe, Cen- meningitidis group B serum or burro 211 anti-N. tral Public Health Laboratory, Colindale. Older strains meningitidis group C serum or 5% for burro 132 antifrom the collection of F. Kauffmann were obtained H. influenza type b serum. These antisera were kind from Ida and Frits 0rskov, World Health Organization gifts from J. B. Robbins. Reactions were read after International Escherichia and Klebsiella Center, Sta- streaking bacteria to single cells and 16 to 24 h of tens Seruminstitut, Copenhagen, Denmark. The re- incubation at 37°C. Sarkosyl NL-97 was obtained from Ciba-Geigy, maining strains, mostly of United States origin, were obtained from the Cooperative Neonatal Meningitis Wehr/Baden, Germany. Specially pure sodium lauryl Study via John B. Robbins, Division of Bacterial sulfate (SDS; catalog no. 30176) was obtained from Products, Bureau of Biologics, Food and Drug Admin- BDH Chemicals, Poole, England. Urea (ultrapure) istration, Bethesda, Md. was obtained from Becton, Dickinson GmbH, HeidelThe individual bacterial strains used are listed in berg, West Germany. N,N'-methylene bisacrylamide Table 1. The sequential numbers in Table 1 (as op- (catalog no. 161-0200) was obtained from Bio-Rad posed to the strain designations) are used throughout Laboratories, Richmond, Calif., and acrylamide (catathis report for convenience. All strains were single- log no. 5521) was obtained from Eastman Kodak Co., colony isolates and were stored at -30°C in 30% Rochester, N.Y. Other chemicals were of research (vol/vol) glycerol-1% (wt/vol) peptone (Difco Labora- grade and were obtained from E. Merck, Darmstadt, tories, Detroit, Mich.), as well as in the lyophilized Germany, or Serva Feinbiochemica, Heidelberg, Gerstate. Strains listed as Kl, K92, and K100 yielded many. precipitation halos on antiserum agar plates containing Plasmid and colicin analysis. Cells were lysed and antisera against Neisseria meningitidis group B (17), plasmid DNA was isolated and analyzed by agarose Neisseria meningitidis group C (30), and Haemophilus gel electrophoresis as described by Helmuth et al. (11). influenzae type B (33), respectively. Strains listed as Colicin production was detected after chloroform 01, 07, and 018 were agglutinated by 1:128 dilutions treatment of bacterial growth on L agar by using soft of rabbit immune sera raised against reference 01, 07, agar overlays containing E. coli K-12 strain P400 or and 018 E. coli strains, respectively. The serotypes of JC3272. Colicin producers were then tested against a strains listed as 016, 075, and K5 were determined by variety of reference test strains, each immune to a tests conducted by Ida and Frits 0rskov. The H different colicin (Table 4). We tentatively concluded serotypes listed are those included in the original that certain strains, which were active against all test strain descriptions and were not retested. All of the strains, produced more than one colicin. These strains strains are being maintained in Berlin, Germany. Ref- were cultivated in L-broth to the stationary phase of erence strains for individual membrane patterns and growth, and after centrifugation the growth medium for the clonal groups are listed in Tables 2 and 3. was concentrated by lyophilization and analyzed by Media and chemicals. L-broth contained (per liter of agarose gel electrophoresis. In some cases, colicins water) 10 g of tryptone (Difco), 5 g of yeast extract were identified after electrophoresis on the basis of (Difco), 10 g of NaCl, and one pellet of NaOH. Tryptic migration which was the same as that of a colicin from soy broth was obtained from Difco. Fermentation tests a reference producer strain (Table 4) and because the were conducted in purple broth base (Difco) supple- corresponding reference tester strain was immune to mented with appropriate carbohydrates at concentra- that colicin band. tions of 0.5%. SDS gel electrophoresis. (i) Gel method 1. DiscontinAll reactions except the rhamnose, dulcitol, and uous, linear gradient gels containing 11 to 20% acrylsalicin reactions were scored as positive or negative amide and 0.2 to 0.38% methylene bisacrylamide were after 2 days at 37°C; rhamnose reactions were scored made as described by Achtman et al. (4), except that after 1 day, and dulcitol and salicin reactions were only specially pure SDS was used. (ii) Gel method 2. Discontinuous SDS slab gel elecscored after 3 days. These times were chosen after the reactions of a representative set of strains was record- trophoresis was performed by the method of Laemmli ed at intervals up to 5 days. In general, the reactions (19), using 10o acrylamide and 0.26% methylene recorded at the times described above were not bisacrylamide, except that urea was present at a changed by prolonged incubation. Nutritional require- concentration of 4 M in the running gel and SDS was ments were tested on minimal agar plates (56/2 medi- present at a concentration of 0.15% (wt/vol) in the um [1]) containing 0.2% glucose and the test supple- electrode buffer. Electrophoresis was conducted at 20 ments. When required, nicotinamide was added to a mA. Sample buffer and staining and destaining soluconcentration of S ,ug/ml. Hemolysin production was tions were as described previously (4). tested with tryptose blood agar base (Difco) containing Isolation of outer membranes by detergent solubiliza5% (vol/vol) fresh, defribrinated sheep blood. Orni- tion. Cells were grown in 100-ml volumes of L-broth to thine decarboxylase was scored after 2 days of growth mid-exponential phase (4 x 108 cells per ml) and
VOL. 39, 1983
v~ .s~z_4w ~t _4-es _z - e4_u
BACTERIAL CLONES AMONG E. COLI
317
CL z~~~~~~~~~~~~~~~~~~~
..L .2 .23 .; . .
; E
U
enX6X en en6en e- r2->m
mIt
en o
t It en en en en en en en en n
Z
0 0 m m m
0E NNN >NN
NNN
NN>
m m m
NO£
Cd
NxN
3 en
Z
t 4NN 4_ _ 4N
^^t^£W9£
m~~~~~~~ 0 0 0 0 0 0 0 0 0 0 0 0 0 0n ^e~ O ~ ~ ~ ~
INFECT. IMMUN.
ACHTMAN ET AL.
E zc o
+ + + + + + + + + + + + + + + + + +
oc~~~~~~~~0
00 tn
-4 JY
-
0.
m 00 00 0
H
4 0.
0-
0 ~o 0 W) (7
0'Io ~o ~o r-
0- 4
_t-4 0-4 *ooo
_
0-.
0 00
0 00
N
ON cr ON tc 0011014
4
W ot4
.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ E
a X e
0O
C
o
-
0f Q
NNNN _e
o o \0 0t \0 t s0
0000 00 x £ \0 £ b 00 s000 m o 00 00 °O 00 ffi ^ 00 u F 00 C 00 Y W
o
Cd
X oO OO t _t _s _e e _s_c>OX4 to00
e . . r .~~~~~~enM rrsssMsMMr rsbM E
== X Y4" 0oo xYe _ = [L4 X [;4 X3 Y~~~~~~~~~~~~~~~~~~~~~~~~~~~~~o 3 ^ 2 X X; X;;;;; ;xz 3 >;
= t| r O O O O O O O O O 0 | 0> X X X X = X X X X ^ ^ ^ ^ e |m W)
St
r-;
|~~~~~~~~~~~~~~~~~~~~~~~~t
Ne
c) in m
3
t-
x-
_-
o-
&c " l>v
~~~~~~~~~~~~~~~~~~~~~~oo
c 0 0 00 00 0 00
cn
n0
0 00 0 s0 0 00 0 00 0 oo00 0 00 _0 0 00 0fi °0 00 00 00 00 0> 00 00 00 0000000_00000U00001000 0 l_D1100 tN_ e_00- ntmNo
lr C4 rT" 00 ON m m ) == = =~~ = :==V: ~ ~cq m en 44 ................... -4z.. . .
-I4_
_4 _4
4 _4 _l4 _4 _I4 4a 4 _4 _4
--- -
000000000000)C0000000000000000000000000000000000
WI) W) r1
_e
eel 0_ e _-0 o% _0 0 0 r- rr- r-0o r r4 rI r_e "_
_14
_t4
_I4 V-
_1
4
_s4
_e4
_I4
_4
_4
_-
_t4
_4
_-4
m 0 w 00 0 0 "-I "O_~T.
_ACN(71
-I V-
V-
CI V-.
ON V-
aD -4T.
0 -
-
0_e N " "
v
~0
ACHTMAN ET AL.
322 E
C.,
|+ + + + + + +
C
INFECT. IMMUN. + + IIIIIIIIIIII
II
9
E¢ E3
oM
v) ) tn in tn n t VD co00 00 Y 00 qT £v I' 00 Y s: ~~~~~~~~~00 F o~ .° ;8 i8 ; ; r,ii 6- ISroo; ON ONo 6f
Aa,
-
-
3~ ~ ~ ~: Z
w I I ' EH D D => D =) D F 3 D
Z
Z 3Z
0
=iae g*
|bXF
a w~~~~~~~~~~~~~~~~~~~~~~~~~~~C %n,s.=Qx la
ku~~~~~~~~~~~~~~~dC
0
UCO)¢1) |bme-
c
0 V
iiiiiiXw ^~~~~~~~~~~~~~c u
5
D
.
2
2"
C
0.
vC
oX* (A
|- M o Gg ~~~~~~~~~~~~~
3
V)~~~~~~~~~~~~~~~~~~~~~~~~~4 cis0
S
VOL. 39, 1983
BACTERIAL CLONES AMONG E. COLI
323
TABLE 2. Serotypes of strains assigned to the individual membrane patterns Membrane pattern
Reference strain
No. of strains with membrane pattern
1 2 3
226 59 189
4
228
5 6
229 102
7
230
8 9 10 11
231 27 117 215
12 13 14 15 16 17 18 19 20 21 22 23
201 200 232 192 190 118 123 140 141 148 149 180
1 (O?:Kl)a 1 (O1:K?) 39 (07:Kl), 1 (07:K-), 1 (O1:Kl), 1 (SpAg:Kl)b 1 (018:K92), 1 (013:K92), 1 (073:K92) 25 (01:Kl), 1 (SpAg:Kl) 25 (018:K1), 3 (O1:K1), 1 (018:K5) 1 (012:K1), 1 (07:K?), 1 (07:K100), 1 (SpAg:K100) 1 (SpAg:K1) 29 (01:K1), 29 (018:K1) 1 (018:K1) 11 (018:K5), 7 (075:K100), 1 (075:K5), 1 (075:Kl), 1 (075:K1) 13 (016:K1) 2 (07:K-) 1 (O?:K1) 1 (07:K1) 1 (07:K1) 1 (018:K1) 7 (018:K5) 1 (018:K?) 8 (018:K?) 2 (018:K-) 1 (018:K-) 2 (01:K?), 2 (07:K?), 1
(07:K1) 191 1 (07:K1) 214 1 (016:K?) 62 1 (O1:K?) 1 (07:K?) 197 199 1 (07:K?) a Serotypes are given in parentheses. b SpAg, Spontaneously agglutinating. 24 25 26 27 28
suspended after centrifugation in 10 ml of 10 mM Tris (pH 8.0). The cells were broken by sonication (80 s, 50% cycle, Branson model B-30 cell disrupter), and unbroken cells were removed by centrifugation (Sorvall SS34 rotor, 3,000 rpm, 20 min, 4°C). The resulting supernatant was centrifuged (Sorvall SM24 rotor, 20,000 rpm, 60 min, 4°C), and the pellet was suspended in 150 ,ul of distilled water and stored at -30°C. After thawing, 50-,ul samples were extracted with 8 volumes of a detergent solution containing 1.67% (wt/vol, final concentration) Sarkosyl and 11.1 mM (final concentration) Tris (pH 7.6) for 20 min at room temperature. The insoluble outer membranes were pelleted by centrifugation (Eppendorf microvials in adapters in a Sorvall SS34 rotor, 20,000 rpm, 90 min, 20°C) and suspended in 50 ,ul of electrophoresis sample buffer. Separation of inner and outer membrane vesicles. The method of Witholt et al. (37) was adapted as follows. Cells were grown with agitation in 500 ml of L-broth containing 50 ,uCi of [2-3H]glycerol (500 mCi/mmol; Amsersham Corp., Arlington Heights, Ill.)
324
ACHTMAN ET AL.
Reference
straina 9 41
INFECT. IMMUN.
TABLE 3. Reference strains for individual clonal groups Hemolysin Membrane Current St eroype
designation
A109 O1:K1:HO1:K1 A110 A76 07:K1 165 016:K1:HA284 205 018:K1:HRS228 102 A53 018:K1:H7 64 A62 018:K5:H7 123 A158 018:K5:H5 133 075:K100 RS501 215 a These strains are the strains which have been investigated represent the clonal groups in future experiments. to mid-exponential phase. All further operations were performed at 0 to 4°C. The cells were concentrated by
centrifugation, and the pellet was suspended in 5 ml of buffer (200 mM Tris, 1 mM MgSO4, pH 8.0). The cells were centrifuged again and then suspended in 5 ml of 200 mM Tris (pH 8.0). The following solutions were then added sequentially with thorough mixing with a Vortex blender after each addition: 5 ml of 1.0 M sucrose-200 mM Tris (pH 8.0); 20 p.l of 0.25 M EDTA; and 75 ,u1 of lysozyme (5 mg/ml; Serva). Then 10 ml of distilled water was added forcefully by pipetting, and the mixture was rocked back and forth five times. After 30 min, 200 ,ul of 20 mM dithiothreitol was added and the spheroplasts were disrupted by sonication as described above. Unbroken cells were removed by centrifugation (Sorvall SS34 rotor, 3,000 rpm, 20 min), and the supernatant was diluted with an equal volume of distilled water. Membrane vesicles were concentrated by centrifugation (Beckman 6OTi rotor, 58,000 rpm, 2 h).
pattern 5
9 3 12 6 9 18 11 11
production
+ +
+ +
-
413 317 666 716 313 317 737 776 777
most intensively and which will continue to
The sucrose solutions described below all contained 1 mM EDTA and 0.2 mM dithiothreitol. The membrane vesicle pellet was suspended in 1.0 ml of a 30% (wt/vol) sucrose solution and applied to a step gradient containing (from the bottom up) 2.2 ml of a 55% (wt/vol) sucrose solution, 2.3 ml of a 50% (wt/vol) sucrose solution, 2.3 ml of a 45% (wt/vol) sucrose solution, 2.2 ml of a 40% (wt/vol) sucrose solution, and 2.2 ml of a 35% (wt/vol) sucrose solution. The tube containing this gradient was centrifuged (Beckman SW41 rotor, 36,000 rpm, 36 h), and samples were collected from the bottom. Subsequent procedures were performed as described by Osborn et al. (26), except that samples for gel electrophoresis were concentrated by trichloroacetic acid precipitation (4 volumes, 5% [wt/vol], overnight, 0WC), followed by suspension in 0.2 volume of electrophoresis sample buffer. The results of a typical experiment are shown in Fig. 1 and 2. Light microscopy showed that all of the cells
TABLE 4. List of reference strains for colicin testing Colicin(s)
producer strain'
Reference
Source"
CA31 T20 BZB2103 (CA23) K53 BZB2135 (P9) BZB2106 (CA38) BZB2107 (CT9) BZB2108 (099) BZB2109 (CT14) BZB2110 (K317) BZB2279 (CA53) BZB2202 (P9) K235 M32T-19 BZB2123 (284)
1 1 2 1 2 2 2 2 2 2 2 2 1 1 2
Reference tester
strain
BZB2101 BZB2134 BZB2103 BZB2104 BZB2125 BZB2106 BZB2107 BZB2108 BZB2109
Sourceb
Additional resistance to colicin(s):
2 B 2 2 2 N El 2 N E2 2 E3 2 E4 2 E5 2 E6 2 BZB2110 E7 BZB2279 2 H,L Ia 2 BZB2202 lb 2 BZB2116 K 1 M32T-19 M 2 BZB2123 N 1 P525 E1-E7 + A P460 1 K+ L a Strain CA31 is a Citrobacterfreundii strain. The other strains are E. coli derivatives. Strains designated BZB are E. coli K-12 strains that carry a single plasmid and produce a single colicin. The original strain from which this plasmid was isolated is listed in parentheses. b 1, Peter Reeves, Department of Microbiology and Immunology, University of Adelaide, Adelaide, South Australia; 2, Anthony Pugsley, Biozentrum, University of Basel, Basel, Switzerland.
A B D
BACTERIAL CLONES AMONG E. COLI
VOL. 39, 1983
STRAIN 73
750k
1J825 p~~~ ~ a
(M~~~~I
im
a U.,
500F
2501
NADH OXIDASE ACTIVITY RATIO OM:IM= 1:63 L
5
10
15 FRACTION
20
25 TOP
325
analyses have shown that encapsulated E. coli isolates possess major outer membrane proteins which have characteristics similar to the characteristics of these proteins but which are variable in migration during SDS-PAGE (12, 13, 15, 27, 28). In addition, encapsulated E. coli strains, regardless of K serotype, also possess a major outer membrane protein called protein K, which is relatively constant in migration during SDS-PAGE (28). Because protein K was found only rarely among non-encapsulated bacteria, it seemed possible that protein K may be needed for or associated with synthesis of acidic polysaccharide capsules (28). However, in other analyses of major outer membrane proteins, the existence of protein K has not been explicitly noted. Outer membranes were isolated by sucrose gradient centrifugation from 14 encapsulated strains whose major outer membrane proteins showed differences in migration during SDS-PAGE. Protein K was identified, and its existence was confirmed on the basis of the following criteria. (i) With all 14 encapsulated strains we detected a single major protein which did not vary markedly in migration during SDS-PAGE. Both gel methods gave the same result. (ii) This protein comigrated with purified protein K (the generous gift of E. C. Gotschlich) with both gel methods. (iii) This protein was absent in several natural isolates
FIG. 1. Sucrose gradient centrifugation of membrane vesicles from strain 73. The fractions labeled OM (outer membrane) and IM (inner membrane) were pooled, assayed for NADH oxidase, and concentrated after precipitation with trichloroacetic acid. p, Density (expressed in grams per cubic centimeter).
1
2
3
were converted to spheroplasts before sonication and that no cell fragments were detectable. After sucrose
gradient centrifugation, two major radioactive peaks were detected (Fig. 1). We concluded that these peaks represented inner membrane and outer membrane vesicles because an inner membrane enzyme, NADH oxidase, was enriched 63-fold in the inner membrane fraction compared with the outer membrane fraction and because proteins similar to those found in outer and inner membrane vesicles of laboratory strains were selectively enrichled in the outer membrane and inner membrane fractions (Fig. 2). The densities of the inner and outer membrane fractions were slightly higher than the densities reported previously for laboratory strains (1.22 and 1.14 to 1.16 g/cm3) (26). Identification of the major outer membrane proteins. Similar to E. coli K-12 proteins, proteins with apparent molecular weights between 30,000 and 40,000 predominated in the outer membrane fractions of the encapsulated bacteria. We refer to these as the major outer membrane proteins below. The major outer membrane proteins were resolved better by gel method 2 than by gel method 1, although the latter method yielded sharper bands over a broad molecular weight range. Figure 2 shows the results when gel method 1 was used; the remaining figures show only the gel region spanning the major outer membrane proteins after separation by gel method 2. Studies with E. coli K-12 have defined porins la and lb and the ompA protein (14). Recently, more limited
ou ter
membraine proteins
FIG. 2. SDS-PAGE of pooled fractions from sucrose
gradient centrifugation. WM, Unseparated
membrane vesicles applied to the sucrose gradient. OM and IM correspond to the fractions indicated in Fig. 1. Gel method 1 was used.
._ O
INFECT. IMMUN.
ACHTMAN ET AL.
326 Irl
f) IQ;-
K . . .
