Evaluation of new enzyme-linked immunosorbent assay based on a ...

Vol. 31, No. 11

JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1993, p. 3036-3039

0095-1137/93/113036-04$02.00/0

Copyright ©D 1993, American Society for- Microbiology

Evaluation of New Enzyme-Linked Immunosorbent Assay Based on a Supernatant Containing Staphylococcus aureus Alpha-Toxin Produced by Bacillus subtilis PER EGNELL,1* BERTIL CHRISTENSSON,2 ROLAND MOLLBY,3 AND JAN-INGMAR FLOCK1 Center for Biotechnology, Karolinska Institute Novum, S-141 57 Huddinge, 1 Department of Infectious Diseases, Lunds University Hospital, S-221 85 Lund,2 and Department of Bacteriology, Karolinska Institute, S-104 01 Stockholm, 3 Sweden Received 18 June 1993/Returned for modification 15 July 1993/Accepted 25 August 1993

The

gene

encoding alpha-toxin from Staphylococcus

aureus was

cloned into

a

BaciUus subtilis expression

vector (pEF 231/a-Tox). The protease-deficient B. subtilis strain DB 104 transformed with pEF 231/ae-Tox

expressed and secreted 5 mg of alpha-toxin per liter into the growth medium. The alpha-toxin-containing supernatant was diluted 200-fold and used as coating antigen in an enzyme-linked immunosorbent assay (ELISA) for serodiagnosis of septicemia and endocarditis caused by S. aureus. Paired sera from patients in acute and convalescent stages of S. aureus and non-S. aureus infections were used to evaluate this ELISA. To evaluate the effectiveness of the crude preparation, the results were compared with those of an ELISA based on a commercially available alpha-toxin. Similar rises in serum titers were obtained with either type of alpha-toxin preparation. This is the first time a crude supernatant without any further purification has been used as an ELISA coating antigen. We therefore conclude that B. subtilis is a suitable host organism for cheap and simple production of prokaryotic recombinant antigens to be used in serodiagnosis.

Measurement of antibody titers against certain microbial antigens in an enzyme-linked immunosorbent assay (ELISA) has been shown to be a useful tool in the diagnosis of microbial infections (1, 3). Such serodiagnosis is particularly useful in cases where cultivation of the infecting microorganism taken from the infectious foci is unreliable. The choice of antigens employed for serodiagnosis of infections is important in order to obtain a high specificity, such as no cross-reactivity with other infectious agents, and high sensitivity. It is also important to have an antigenic preparation of high purity to improve the quality of the test by reducing the nonspecific background. However, purification of a specific antigen from its native host is often complicated, and often the natural host is difficult to propagate in vitro. Therefore, many genes encoding different proteins suitable as antigens have been cloned and expressed in an alternative host, such as Escherichia coli. Contaminating antigens in the preparation will, however, give unacceptable background in the ELISA, since E. coli naturally colonizes humans, giving rise to antibodies (7). So even if it is possible to produce high levels of different antigens in the E. coli production systems available, purification is often necessary for the production of a pure protein preparation suitable as a coating antigen in the ELISA. An alternative to this approach would be to produce the antigen in an organism normally not seen by the human immune system. It would also be of obvious advantage if the antigen is secreted into the growth media, necessitating only a centrifugation step. We have cloned the gene from Staphylococcus aureus coding for alpha-toxin in Bacillus subtilis and obtained secretion of this antigen into the culture supernatant. Such crude supernatants were used to coat microtiter plates which *

were used in the ELISA to determine anti-alpha-toxin antibody levels in patient sera. The alpha-toxin gene from S. aureus Wood 46 was cloned into plasmid pEF 231 (2) as a DNA fragment encompassing the promoter, ribosomal binding site, and structural gene. Plasmid pEF 231 contains a T5 promoter active in B. subtilis (6), thus resulting in two promoters constitutively driving the alpha-toxin gene. The protease-deficient B. subtilis strain DB 104 (aprA nprE) (5) was transformed with this plasmid construct, pEF 231/a-Tox, and when the strain was plated on blood agar plates, hemolytic zones were displayed around the colonies. Plasmid segregational stability was secured by cultivation in the presence of 20 ,ug of kanamycin per ml,

since kanamycin resistance is encoded by the pEF 231/aTox plasmid. The quantity and quality of the alpha-toxin produced from a Luria broth culture of B. subtilis DB 104(pEF 231/a-Tox) was assessed by making a coating curve as shown in Fig. 1. One hundred microliters of various dilutions of a culture supernatant was used to coat microtiter wells overnight at room temperature. The wells were then coated with 5% skim milk. Either a positive serum sample from a patient with complicated septicemia or serum from a rabbit immunized with purified alpha-toxin (National Bacteriology Laboratory [SBL], Stockholm, Sweden) was diluted 800-fold and used in the ELISA. Binding of antibodies was monitored by the addition of alkaline phosphatase-conjugated antibodies against either human or rabbit immunoglobulin G (Sigma Chemical Co., St. Louis, Mo.), diluted 1,000-fold, followed by alkaline phosphatase substrate, p-nitro phenyl phosphate (Sigma). Each incubation lasted for 1 h at 37°C, and the microtiter plate was washed three times between each step with phosphate-buffered saline containing 0.05% Tween. As controls, a crude supernatant from a B. subtilis culture

Corresponding author. 3036

3037

NOTES

VOL. 31, 1993

3.0

2 C

0

2.5

1.5 a) 0

cb

CZ

-0L- 1

OosdX~~M

o

.0 0.5

0.5~~ o I-

II

0

1

00

I

3

9

27

101

o

81 243 729 2187

.5

0

100

15

20

.

Dilution factor FIG. 1. Coating curves with different types of alpha-toxin. The initial concentration of alpha-toxin from SBL was 30 1Lg/ml. Symbols: O, B. subtilis(pEF 231/a-Tox)-produced alpha-toxin batch 1; El, B. subtilis(pEF 231/ot-Tox)-produced alpha-toxin batch 2;*, alpha-toxin from SBL; M, B. subtilis(pEF 231) negative control.

without alpha-toxin production and purified alpha-toxin from the SBL were included in the coating curve experiment in Fig. 1. From these curves, the amount of alpha-toxin in the crude B. subtilis supernatant was estimated to be approximately 5 mg/liter. The coating concentration for subsequent experiments was chosen to be 5 ng of alpha-toxin per well. The supernatant from B. subtilis DB 104(pEF 231/a-Tox) was run on a sodium dodecyl sulfate-polyacrylamide gel, and the proteins were transferred to nitrocellulose filter (Western blot [immunoblot]) as described in reference 4 and probed with rabbit anti-alpha-toxin or a serum sample from a patient with S. aureus endocarditis. With these sera, single bands corresponding to a molecular size of 30 kDa were seen, indicating that the alpha-toxin was intact (Fig. 2). Furthermore, no bands were detected with a supernatant from a B. subtilis strain without the alpha-toxin-encoding plasmid by using these sera or sera from 15 healthy blood donors (data not shown). Sera taken from 49 patients both at the acute phase (within

A

B

1 234

1234

94 67 ---O 43 30 _ 20 -o -4 14

94 67 r

_W

43

;1

30

204 14 -0--o

FIG. 2. Western blot of alpha-toxin preparations. Lanes 1, crude supernatant from B. subtilis DB104(pEF 231); lanes 2, crude

super-

subtilis(pEF 231/a-Tox); lanes 3, crude supernatant from S. aureus V8; lanes 4, commercial alpha-toxin from SBL. Rabbit anti-alpha-toxin serum (A) and serum from a patient with S. aureus endocarditis (B) were used. Anti-rabbit (A) and anti-human (B) alkaline phosphatase-conjugated antibodies raised in a goat were subsequently used. natant from B.

pEF 231/ a-toxin FIG. 3. The absorbance values at the 200-fold dilution for each serum sample by using either commercial alpha-toxin (SBL) or DB

104(PEF 231/a-Tox)-produced alpha-toxin as a coating antigen are plotted against each other. The background values were first subtracted from each absorbance value.

10 days after onset of infection) and at the convalescent phase (10 to 30 days after onset of infection) were measured in the ELISA described above, using either crude supernatant from B. subtifis(pEF 231/aL-Tox) or purified alpha-toxin (SBL). The categories of patients included were as follows: S. aureus uncomplicated septicemia (n = 5), S. aureus complicated septicemia (n = 20), S. aureus endocarditis (n = 7), non-S. aureus septicemia (n = 13), and non-S. aureus endocarditis (n = 4). Titration curves were made for each serum sample at dilutions of from 50- to 6,400-fold. The absorbance values at the 200-fold dilutions for each sample were plotted with either type of antigen (Fig. 3). The regression coefficient was found to be 0.49, implying a correlation between the antigenic performance of the two alpha-toxin preparations. It is important to point out that it is too unspecific to measure only the individual antibody titers and to compare

them with a standard titer value as an indication of S. aureus infection; the individual variations normally occurring in a healthy population are considerable. It is thus the increase in titer, i.e., the difference between the acute-phase serum titers and the convalescent-phase serum titers, which is a relevant indication of an S. aureus infection. Titer rises in the paired serum samples were calculated as the percent increase of absorbance obtained with sera taken in the acute and convalescent stages. An increase exceeding 50% was considered a positive sign of S. aureus infection; this threshold is indicated in Fig. 3. In Fig. 3, the increases in antibody titers with either type of antigen are compared. For clarity the data are divided according to patients with different types of infections (Fig. 4a, infections caused by S. aureus; Fig. 4b, non-S. aureus infections). Different symbols denote different categories of patients. If the two alpha-toxin preparations had given the same result, i.e., same sensitivity, the points would be expected to be placed on a line with the slope of 1 (in Fig. 4 such a line is inserted). However, many points, especially at high titer increases, are below the line, indicating that the increase is more pronounced, i.e.,

3038

NOTES

J. CLIN. MICROBIOL.

300

A -e

200 CD

' -

x

oe

1

0

u8

-01,

*

100 +

0-.

-4e

.-e

.e

.041

0

1#0

' -+

___

0 +0

0

0

.4-

-100'

0

-1CC 0

100

200

'O

B. subtilisa-toxin; percent titer rse

B 400

200 I

-

x

a) *-

0C 6 ;p

1 00

0)a

.r

Q

CD

-1 00 IC-100

450

B. subtilis a-toxin; percent titer rise FIG. 4. Percent increase in antibody titers. Alpha-toxin from SBL (y axis) or a B. subtilis supernatant containing alpha-toxin (x axis) was used, and the percent increases in absorbance value in ELISA at a 200-fold dilution of the sera are plotted against each other. (A) Symbols: +, S. aureus uncomplicated septicemia; 0, S. aureus complicated septicemia;*, S. aureus endocarditis. (B) Symbols: 0, non-S. aureus non-S. aureus endocarditis; septicemia; -, line with the slope of 1; , regression line. Horizontal and vertical lines indicate a 50% increase in antibody titer. E,

--

VOL. 31, 1993

higher sensitivity is obtained, with B. subtilis supernatant containing alpha-toxin. A possible explanation for this is that the B. subtilis expression system is less damaging to the alpha-toxin molecules and therefore the antigenicity would be better preserved in the B. subtilis preparation. We conclude that B. subtilis is a promising host organism for the production of recombinant proteins of a prokaryotic extracellular origin to be used in serodiagnosis. REFERENCES 1. Christensson, B. 1986. Serological diagnosis of staphylococcal bacteraemia. Lancet ii:1165-1166. 2. Egnell, P., and J. I. Flock 1991. The subtilisin Carlsberg proregion is a membrane anchorage for two fusion proteins produced

NOTES

3039

in Bacillus subtilis. Gene 97:49-54. 3. Ericsson, A., M. Granstrom, R. Mollby, and B. Strandvik. 1986. Antibodies to staphylococcal teichoic acid and alpha toxin in patients with cystic fibrosis. Acta Paediatr. Scand. 75:139-144. 4. Harlow, E., and D. Lane. 1988. Antibodies: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 5. Kawamura, F., and R. Doi. 1984. Construction of a Bacillus subtilis double mutant deficient in extracellular alkaline and neutral proteases. J. Bacteriol. 160:442-444. 6. Melin, L., and E. Dehlin. 1987. Functional comparison of an early and a late promoter active DNA segment from coliphage T5 in B. subtilis and in E. coli. FEMS Microbiol. Lett. 41:141-146. 7. Orskov, F., and I. Orskov. 1992. Escherichia coli serotyping and disease in man and animals. Can. J. Microbiol. 38:699-704.