ANALYTICAL

BIOCHEMISTRY

207,

311-316

(1992)

Polymyxin B-Horseradish Peroxidase as Tools in Endotoxin Research Ben J. Appelmelk,* Donghui Su,* A. Marian Bert G. Thijs,? and David M. MacLaren* *Department of Medical van der Boechorststraat

Received

April

J. J. Verweij-van

Microbiology and tMedica1 Zntensive Care Unit, 7, 1081 BT Amsterdam, The Netherlands

Vught,* Vrije Universiteit,

30, 1992

The peptide antibiotic Polymyxin B (PMB) binds to bacterial endotoxin (lipopolysaccharide, LPS). We prepared covalent conjugates of PMB and horseradish peroxidase (HRP) by periodation of HRP-linked oligosaccharides followed by direct condensation with PMB. In addition we prepared monoclonal antibodies (Mabs) to PMB. The PMB-HRP conjugates and anti-PMB Mabs were used to study in ELISA the binding of PMB to LPS from Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa, In addition, PMB-HRP was used to quantify lipid A in ELISA, and to stain gramnegative bacteria histochemically. For the study of PMB-LPS interaction, PMB-HRP proved to be superior to the anti-PMB Mabs. PMB-HRP conjugates are useful general probes to detect or measure lipid A and LPS of various species using very simple methods and to stain bacteria, and they may obviate the need for many specific antisera. Thus, PMB-HRP conjugates 0 ivw Academic are useful probes for endotoxin research. Press,

Conjugates

Inc.

Polymyxins’ (PMB) are cyclic amphipathic peptide antibiotics that contain free amino groups derived from a,T-diaminobutyric acid and a fatty acid tail (Fig. 1). PMB bind to endotoxins (lipopolysaccharide, LPS) of gram-negative bacteria. The interaction of PMB with LPS involves ionic forces between amino groups in PMB and negatively charged phosphate and carboxyl

’ Abbreviations use& PMB, polymyxin B; LPS, lipopolysaccharide; Kdo, 3-deoxy-D-munno-octulosonate; HRP, horseradish peroxidase; RT, room temperature; O/N, overnighti Mabs, monoclonal antibodies; KLH, keyhole limpet hemocyanin; PBS, phosphate-buffered saline; PBST, PBS containing 0.02% Tween; GDA, glutaraldahyde; BSA, bovine serum albumin; OD, optical density; OPD, o&o-phenylenediamine. 0003-2697/92 $5.00 Copyright @ 1992 by Academic Press, All rights of reproduction in any form

groups in the lipid A-Kdo region (Kdo = 3-deoxy-Dmanno-octulosonate); in addition, hydrophobic interactions take place between the respective acyl groups. PMB has relevance for endotoxin research in at least two ways; first, PMB (as opposed to specific anti-LPS antisera) reacts with LPS of many species regardless of varied serospecificity, and thus it can be used as a general probe for measuring or detecting LPS or lipid A. Second, binding of PMB to LPS may result in neutralization of the detrimental effects of LPS either in u&o (1) or in ~~IJO(2). Clearly it is of prime importance to be able to study quantitatively the interaction of PMB and LPS. Until now binding has been measured by use of nonlabeled PMB in (rather cumbersome) precipitation assays (3), or by use of PMB labeled either radioactively or with a fluorescent dansyl group (4). In this paper, novel PMB reagents are described, i.e., covalent conjugates of PMB with horseradish peroxidase (PMBHRP). We describe three applications of PMB-HRP in endotoxin research: (a) for studying quantitatively the interaction between PMB and LPS, (b) for quantitative measurement of lipid A or LPS, and (c) for histochemical staining of gram-negative bacteria. MATERIALS

Synthesis of

AND PMB-HIW

METHODS

Conjugates

Polymyxin B sulfate was obtained from Pfizer (Brussels, Belgium); a single lot (G OYl82) was used throughout. Horseradish peroxidase type VI was purchased from Sigma. Aldehyde groups were introduced into HRP by incubating 5 mg HRP in 0.7 ml water with 50 ~1 0.2 M periodate for 20 min at room temperature (RT) while rotating end-over-end. Three drops ethylene glyco1 were added and likewise incubated for 10 min. Activated HRP was purified by gel filtration on Sepharose 311

Inc. reserved.

312

APPELMELK

NH,’ I

/ DAE\

l-Leu

DAB-NH,’

I o-Phe

! t -1hr

I

I DAB-NH,

‘NH,-DAB

’

‘DAB/ t -1ht

I + NH,

0

-DAB \

FIG. 1. Covalent diaminobutyric acid.

structure

NH/‘of

Polymyxin

B.

DABNH:

= a,~-

ET

AL.

dialysis bag was transferred to PBS-0.2% glutaraldehyde (GDA, 4“C, O/N). This was followed by dialysis (4’C, O/N, 0.05 M carbonate buffer, pH 9). Finally, 5 mg PMB and 1 mg GDA-activated KLH were mixed and incubated (O/N, RT). For blocking, 0.1 ml 3 M ethanolamine (pH 9) was added (2 h, RT); finally, the conjugate was dialyzed (PBS, O/N, 4’C) and stored at -8O’C until use. Mice received four intraperitoneal vaccinations at 2-week-intervals of 100 pg PMB-KLH in 0.5 ml, mixed with complete Freund adjuvant or IFA for first and later injections, respectively. Finally, an intravenous booster was given and spleen cells were fused 3 days afterward. Anti-PMB Mabs were produced using conventional hybridoma technology. Screening for anti-PMB Mabs took place in ELISA with plates coated with 10 hg/ml PMB or PMB-BSA, prepared as described above for PMB-KLH. Three hybridomas were subcloned thrice and this resulted in clones 44 (IgG2a), 45 (IgM), and 46 (IgG3). Culture supernatant of clone 45 contained 50 pg/ml Mab.

G25 (Pharmacia). One milligram purified activated Other Mubs HRP in 1 ml water was mixed with 0.1, 1, or 10 mg PMB Clone 20, a murine IgM Mab, binds to Kdo in the (in 1 ml water) and 0.2 ml 0.5 M carbonate buffer (pH 9). inner core of LPS as well as to the hydrophobic part of This mixture was incubated for 2 h at RT, then chilled lipid A (5). Clone 20 was purified by modified gel filtrato 4OC and thereafter the Schiff bases formed were station (6). Murine Mabs 48 and 49 (both IgGl) are dibilized by reduction with 25 ~1 of a 10 mg/ml sodium borohydride solution; incubation proceeded 1 h at 4’=‘C. rected against the idiotype of clone 20; clone 49 binds very close to the lipid A binding site of clone 20, while After overnight (O/N) dialysis, done in order to remove clone 48 binds further away (manuscript in preparaunbound PMB that would act as a competitor to the tion). The following isotype-matched negative control conjugate, the final product was preserved with 0.01% Mabs for clones 44, 45, 46, 48 and 49 were included merthiolate and stored at 4 or -8O’C. clone 3 (IgG2a), clone 38 (IgM), clone 4 (IgG3), and clone 2 (IgGl), respectively. Clones 2,3, and 4 react with Bacteria and Bacterial Antigens LPS of E. coli J5 (7). Clone 38 is directed against a nonLPS surface antigen of S. minnesota (unpublished obSynthetic diphosphoryl lipid A (compound 506), was servations). Unless stated otherwise, all Mabs were purchased from Daichi Co. (Osaka, Japan). Escherichiu coZi J5 (Rc chemotype), natural diphosphoryl-lipid A used in the form of hybridoma culture supernatant. prepared by mild acetic acid hydrolysis of E. coli F515 LPS (HaC Lipid A), and LPS from SuZmoneZZu minneApplication of anti-PMB Mabs and PMB-HRP Conjugates to sotu R595 were obtained from H. Brade (Borstel, GerEndotoxin Research many). LPS from E. coli Olll:B4 was from Sigma; LPS Studying the Interaction between PMB und LPS from clinical isolates of E. coli, Pseudomonus aeruginosa, and Klebsiella pneumoniue were from N. Mehta (LonUse of PMB-HRP to measure binding of PMB to soliddon). Lipid A and LPS were dissolved in 0.01% triethylFirst, the performance of the three phase bound LPS. amine in water by sonication for 1 min and stored different PMB-HRP conjugates was compared. One at 4’C. hundred microliters HaC lipid A or E. coZi Olll:B4 LPS (1 pg/ml in PBS) was left overnight at RT in flatbottom polystyrene microtiter plates. Negative control wells Monoclonal Antibodies (Mabs) were filled with PBS only. Plates were washed thrice in Monoclonal Antibodies to PMB PBS containing 0.02% Tween 80 (PBST). Doubling dilutions of PMB-HRP in PBST starting at 1:lOOO were Mabs to PMB were prepared by vaccinating Balb/c added and incubated for 4 h at 37’C. After washing, mice with PMB coupled to a KLH carrier (KLH = keyhole limpet hemocyanin). KLH (Calbiochem, San substrate solution (100 ~1 1 mg/ml ortho-phenylenediamine (OPD) and 1 pi/ml 30% HzOz in 0.1 M citrateDiego, CA) at 2 mg/ml was dialyzed (4’C, O/N, PBS) to phosphate buffer, pH 5.5) was incubated for 15 min at remove low-molecular-weight contaminants; then, the

IMMUNOSORBENT

ASSAY

OF

POLYMYXIN 2.0 2.0 -b 1.8 .

2.0 1.E

a ~

313

B CONJUGATES

-

1.6.

1.6 -

s

1.4 -

: ~ .z !g g 5

1.2 1.0 0.8 -

1.4 1.2s l.Ob 0.6 .

0.6 . : 0.4 0.2 -

o.oL

0.2 . 1 0.0 ’

I l/l000

1/4000 dilution

l/l6000 of PMB-HFIP

0.0

FIG. 2. Binding of various PMB-HRP Results of PMB-HRP conjugates prepared

l/4000 dilution

l/l000

l/4000 dilution

l/l6000 of PM-HRP

l/64000 of PUE-HRP

conjugates to natural HaC lipid from a PMB:HRP ratio of l:lO,

RT and thereafter 50 ~1 2 N HzSOd was added. Optical density (OD) was read at 492 nm. The reactivity of PMB-HRP (1 mg/l mg) was similarly tested on synthetic lipid A (compound 506) and LPS from clinical isolates of E. coli, P. aeruginosa, and K. pneumoniae. Use of anti-PMB Mabs to study the interaction between PMB and solid-phase bound LPS. First, the reactivity of anti-PMB Mabs with PMB was determined. PMB at 10 pg/ml in PBS was coated at RT, O/N. After washing, doubling dilutions of Mabs 44,45,46, and 47 in PBST were added and incubated for 4 h at 37OC. Then HRP-conjugated goat anti-mouse antibodies (American Qualex) were added and incubated for 2 h at 37OC. ELISA was finished as described above. The ability of Mabs 44 to 46 to bind to LPS-complexed PMB was also evaluated S. minnesota R595 LPS (10 pg/ml) was coated onto ELISA plates; after washing 10 pg/ml PMB was added and incubated and ELISA proceeded as described above. Negative controls were (a) LPS-coated wells without PMB, (b) PMB-coated wells, (c) “PBS coated’ wells, and (d) isotype-matched controls clones 3, 4, and 38. Finally we determined the reactivity of

OA 0.4 -

A (Cl), Escherichiz coZi Olll:B4 (b) for 1:1, and (c) for 1O:l.

LPS

(A),

and

polystyrene

(0).

clone 45 for PMB complexed to LPS of clinical isolates of E. coli, P. aeruginosa, and K. pneumoniae.

Measurement

of Antibody-Bound

Lipid A

In this application, the ability of the anti-idiotype Mabs 48 and 49 (directed to the idiotype of clone 20, a clone reactive with lipid A) to compete with lipid A for binding to the antigen-binding site of clone 20 was studied. PMB-HRP was used as a probe to quantify lipid A that bound to clone 20 in the absence or presence of inhibiting anti-ids. Purified clone 20 (3 pg/ml in PBS) was coated to ELISA plates (O/N, RT), after washing 0.25 pg synthetic 506 lipid A was added and incubated (2 h, 37OC) and finally, bound lipid A was detected with HRP-PMB (1 mg/l mg, diluted 1:5000). Under these noncompetitive conditions, an OD of approximately 1.5 was obtained. To study the ability of clones 48 and 50 to compete with lipid A for binding to clone 20, doubling dilutions of hybridoma culture supernatant (100 ~1 in PBST) of clones 48 and 49 were added to solid-phase

314

APPELMELK

l/4000 dilution

1~16000

Reactivity of PMB-HRP with synthetic 506 lipid LPS from clinical isolates of Escherichiu coli (A), Pseudomonus ginosa (O), and Klebsiella pneumonias (+).

A (II), aeru-

bound clone 20 and incubated for 1 h, then lipid A was added and ELISA was performed as described above.

for Staining

AL.

of PME-HRP

FIG. 3.

Use of PMB-HRP Bacteria

ET

FIG. 5. ready clone

Reactivity of monoclonal bound to LPS of Salmonella 45 (A), clone 46 (0).

This was investigated in a model system. A single colony of E. coli J5, strain 2877 was suspended in water and diluted 10 times. Thirty-microliter aliquots were applied to multiwell immunofluorescence glass slides, air dried, and heat fixed in a flame. PMB-HRP was added in serial dilutions (30 ~1 in PBST) ranging from 1:lO to 1:5120. After 1 h at 37’C the slides were washed under running tap water and repeated immersion in PBST. For color development, 0.5 mg/ml diaminobenzidin in PBS and 1 PI/ml 30% HzOz was used; the stained bacteria were visualized by bright field microscopy, and mi-

sorbed

Reactivity to polystyrene.

of monoclonal antibodies with Polymyxin B adClone 44 (O), clone 45 (A), clone 46 (0).

B al44 (0),

AND

DISCUSSION

In this paper we describe the preparation of conjugates consisting of the LPS-binding antibiotic polymyxin B, covalently linked to horseradish peroxidase (PMB-HRP) and we give three applications of PMBHRP for endotoxin research. Synthesis of PMB-HRP Conjugates PMB-HRP conjugates at ratios of PMB/HRP of l:lO, l:l, and 1O:l were synthesized by the use of a procedure very much like the one used for making HRPantibody conjugates. The reactivity of the three conjugates with HaC lipid A and LPS is shown in Figs. 2a-2c.

FIG. 6. FIG. 4.

with Polymyxin R595. Clone

crographs were taken on Kodak 100 ASA film using a Leitz MPS 51s camera attached to Zeiss Axioscope. RESULTS

of Gram-Negative

antibodies minnesotu

Reactivity of clone 45 with PMB already clinical isolates of Escherichia coZi (A), Pseudomonas and Klebsiella pneumoniae (+).

bound to LPS of aerughosa (0),

IMMUNOSORBENT

ASSAY

2.0 r

OF

I

0.4 0.2

I

FIG. 7. Detection by HRP-PMB of lipid A bound to the antigenbinding site of clone 2.0 in the presence of increasing amounts of antiids to clone 20, clone 48 (+), clone 49 (U).

A low PMB-HRP (1:lO) ratio (Fig. 2a) resulted in conjugates that were not so reactive. On the other extreme, at a high PMB-HRP (1O:l) ratio (Fig. 2c) the conjugate bound nonspecifically to the ELISA plate. This finding is not unexpected given the hydrophobic nature of PMB. The other conjugates did not stick to the plates, nor did HRP alone; HRP alone also did not bind to lipid A (data not shown). Thus, a 1:l ratio of PMB-HRP (Fig. 2b) resulted in the best conjugates and these were used for further study. The reactivity of this conjugate for synthetic 506 lipid A and LPS of clinical isolates is

FIG.

8.

Immunohistochemical

detection

by PMB-HRP

POLYMYXIN

315

B CONJUGATES

shown in Fig. 3. Apparently, the affinity of PMB-HRP for lipid A and intact LPS is similar, but when lipid A and LPS were coated at lower concentrations, PMBHRP had a (much) higher reactivity for lipid A than for smooth LPS (data not shown); probably this is due to steric hindrance caused by the O-side chain of smooth LPS. Our results show that PMB reacts with LPS of various species, and that the lipid A is the binding site. Thus, with PMB-HRP conjugates the interaction between PMB and LPS can be studied quantitatively and in ELISA format without radioactivity. We also tested the performance of anti-PMB Mabs. First, clones 44 to 46 were tested for reactivity to solid-phase bound PMB (Fig. 4). The highest reactivity was found for clone 45. Clones 44 to 46 did not react with “PBS coated” wells (not shown). Clones 44 to 46 were also tested for reactivity with PMB, already bound to S. minnesota R595 LPS. Again, clone 45 was superior (Fig. 5); the negative controls, Mabs 3, 4, and 38, did not react. Finally clone 45 was tested for reactivity with PMB already bound to LPS of clinical isolates (Fig. 6). From a comparison of Figs. 3 and 6 it can be calculated that PMB-HRP was superior to clone 45: while use of 12.5 pg/ml of clone 45 resulted in ODs of 1.2-1.6, use of 0.5-l pg/ml PMBHRP yielded ODs >2. Thus, we decided to use PMBHRP for further studies. A second application of PMB-HRP as a general probe for LPS is shown in Fig. 7. In this experiment, both lipid A and anti-idiotypes (Mabs 48 and 49) are allowed to react with clone 20, already bound to polysty-

of gram-negative

bacteria.

Final

magnification

was

1200~.

316

APPELMELK

rene of ELISA plates. PMB-HRP served to measure lipid A that finally bound to clone 20 in the presence of increasing amounts of inhibiting anti-ids. Competition experiments demonstrated that clone 49 competes more strongly than clone 48 with lipid A for binding to clone 20. These data are not inconsequent with what we know on clones 48 and 49, i.e., that clone 49 binds close to the lipid A binding site while clone 48 binds further away (manuscript in preparation). Clearly without the use of PMB-HRP measurement of bound lipid A would have been much more complicated. Thirdly, we used PMBHRP to stain gram-negative bacteria (Fig. 8). Conjugate concentrations were used in a range from 1:lO to 15120. At low dilutions a high background was obtained (not shown); at dilutions greater than 1:640 the signal gradually faded out. At 15120 bacteria were barely visible. A dilution of 1:320 was judged to be optimal and was used in experiments of Fig. 8. We think it should be possible to use PMB-HRP for detecting bacteria in tissue sections, e.g., obtained from organs of experimentally infected animals. In summary we have prepared conjugates consisting of the LPS binding antibiotic Polymyxin B, covalently linked to a “reporter enzyme,” i.e., horseradish peroxidase. Preparing these conjugates followed a very simple recipe. We demonstrated the use of these conjugates as general probes for detection and measurement of various LPS, and finally as a staining reagent to detect bacteria immunohistochemically. Thus a single reagent (i.e., PMB-HRP) may obviate the need for a large col-

ET

AL.

lection of specific anti-LPS/bacterial antibodies, and it allows the study of PMB-LPS interactions with much more simple methodology than hitherto described. In short, PMB-HRP are useful probes for endotoxin research. ACKNOWLEDGMENTS tut

We thank Borstel,

Dr. H. Brade and Dr. E. Th. Rietschel (ForschungsinstiBorstel, Germany) for material and immaterial support.

REFERENCES 1. Cavaillon, J. M., munol. 23,965-969.

and

Haeffner-Cavaillon,

N.

(1986)

Mol.

Zm-

2. Baldwin, G., Alpert, G., Caputo, G. L., Baskin, M., Parsonnet, J., Gillis, Z. A., Thompson, C., Siber, G. R., and Fleisher, G. R. (1991) J. Znfect. Dis. 164, 542-549. 3. Vaara, M., Vaara, T., Helander, I., Nurminen, M., Rietschel, E. Th., and Make& P. H. (1981) FEMS Microbd Lett. 129, 145-149. 4. Bader, J., and Teuber, M. (1973) 2. Naturforsch. 28c, 422-430. 5. Appelmelk, B. J., Su, D., Cohen, J., Verweij-van Vught, A. M. J. J., Hekker, T. A. M., Thijs, L. G., Buurman, W. A., Komuro, T., Playfair, J. H. L., and MacLaren, D. M. (1990) in Cellular and Molecular Aspects of Endotoxin Reactions (Nowotny, A., Spitzer, J. J., and Ziegler, E. J., Eds.), pp 347-353, Excerpta Medica, Amsterdam. 6. Bouvet, J. P., Pires, R., and Pillot, J. 66,299-305. 7. Appelmelk, B. J., Verweij-van Vught, Schouten, W. F., de Jonge, A. J. R., D. M. (1988) J. Med. Microbial. 26,

(1984)

J. Zmmunol.

Metho&

A. M. J. J., Maaskant, J. J., Thijs, L. G., and MacLaren, 107-114.