Binding of a membrane proteoglycan from Klebsiella pneumoniae and its derivatives to human leukocytes

lmmunobiol., vol. 186, pp. 183-198 (1992)

Original Papers

1 Laboratoire d'Immunologie, INSERM USO CNRS URA 1177 UCBL, H6pital E. Herriot, Pavillon P, 69437 Lyon Cedex 3, and 2 Centre d'Immunologie et de Biotechnologie Pierre Fabre, St Julien en Genevois, France

Binding of a Membrane Proteoglycan from Klebsiella pneumoniae and Its Derivatives to Human Leukocytes 2 3 1 ZAKARIAHMAMA , GERARD NORMIER , EDOUARDKoUAssr , 1 2 1 MONIQUE FLACHER , HANS BINZ , and JEAN-PIERRE REVILLARD

Received September 24, 1991 . Accepted in Revised From March 16, 1992

Abstract The binding of a membrane proteoglycan from a non-encapsulated strain of Klebsiella pneumoniae (Kp-MPG) and four derivatives thereof, to human leukocytes, was investigated by indirect immunofluorescence using biotinylated F(ab')2 fragments of anti-Kp-MPG antibodies and the streptavidin-phycoerythrin amplification system in flow cytometry. Four KpMPG derivatives were studied: 11 an acylpoly(I,3)galactoside (APG), 2/ an APG preparation submitted to acid hydrolysis which removed all fatty acids, but left intact the galactose chain of APG (GC-APG), 3/ a preparation obtained by mild alkaline hydrolysis, containing additional ester-linked C t4 and C 16 fatty acids bound to the APG molecule (EFA-APG) and 4/ a polymer of the latter compound (APG pol). Kp-MPG, APG and EFA-APG were shown to bind exclusively to monocytes at the lowest concentrations (from 0.15 to 3 IlM APG). At higher concentrations, these compounds interacted with polymorphonuclear leukocytes, and with lymphocyte subsets in the following decreasing order: B cells, NK cells, CD8+ and CD4+ lymphocytes. Neither APG pol or GC-APG nor K. pneumoniae smooth LPS showed significant binding to leukocytes. However Kp-LPS treated by drastic alkaline hydrolysis displayed binding properties similar to those of APG. Removal of the ester-linked C 14 and C t6 fatty acids from EFA_APG did not affect the binding of the molecule. The capacity of cells from the myelomonocytic lineage to bind Kp-MPG and APG was very low in phenotypically immature cell lines (HL60 and U937) as compared with monocytes or polymorphonuclear cells. Treatment of U937 cells with interferon-y up-regulated their APG binding capacity along with the expression of the integrin CD 11 b and the CD 14 molecule, whereas monocytes J Current address: Research Center Maisonneuve-Rosemont Hospital, University of Montreal, Montreal Quebec, Canada

Abbreviacions: APG = Acylpoly (1,3) galactoside; EFA-APG = esterified fatty acids-APG; GC-APG = galactose chain of APG; APG pol = polymer of EFA-APG; Bior-F(ab')2 = biotinylated F(ab'h fragments of anti-Kp-MPG antibodies; 1'1 MFI = specific mean fluorescence intensity; IFN-y = Interferon gamma; Kp-MPG = membrane proteoglycan of K. pneumoniae; Kp-LPS = lipopolysaccharide from K. pneumoniae; Kp-LPS (NaOH) = KpLPS treated by alkaline hydrolysis; MNC = mononuclear cells; PE = phycoerythrin; PMN = polymorphonuclear cells; SAPE = Streptavidin-PE

184 . Z.

HMAMA

et al.

exposed to interferon-y showed an increased binding of APG associated w ith an elevated expression of the galactose specific lectin Mac-2. The data demonstrate a preferential binding of Kp-MPG and APG to cells of the monocyte/macrophage lineage. APG binding does not involve the poly (1,3) galactose chain and the ester-linked C l4 and C l6 fatty acids but requires the presence of the hydrophobic part of the molecule.

Introduction

Defense against gram-negative bacteria is achieved by phagocytosis and intracellular killing followed by the synthesis of various cytokines, including Tumor Necrosis Factor (TNF)-a, Interleukin (IL)-1 and IL-6, which generate a local inflammatory reaction, while IL-6 triggers the synthesis of acute phase proteins by hepatocytes. Exaggerated responses to gramnegative bacteria may provoke a syndrome of shock with fever, neutropenia, dessiminated intravascular coagulation and multiple organ failure. Lipopolysaccharide (LPS) can elicit most of the inflammatory and immune reactions induced by gram-negative bacteria, including the endotoxic shock (1). Antibodies to LPS are presently assessed with regard to their capacity to prevent or reverse shock in gram-negative septicemia (2, 3). Recent studies outlined the complexity of the structure-activity relationships within LPS molecules. Hence most biological activities of LPS could be reproduced by using synthetic analogues of lipid A which mimicks the lipid moeity of LPS (4, 5). Furthermore the lipid A region of LPS molecule was shown to play a major role in the attachment to leukocyte surface molecules (6, 7). Among these receptors three classes of leukocyte surface molecules, the CD ll/CD 18 integrins, the scavenger (acetyl-low-density lipoprotein) receptor and the CD 14 molecule which recognizes complexes of LPS with the lipopolysaccharide-binding-protein (8), have recently been identified. In contrast with the data accumulated on LPS structure and activity, little is yet known as regards other components of bacterial membrane and cell wall which may display some of the biological properties of LPS (9). For instance membrane proteoglycans (MPG) have been isolated from various strains of K. pneumoniae and were shown to stimulate NK cell activity, cytokine synthesis by monocyte/macrophages and to activate B cells from LPS non-responder mice (10-12). From a MPG isolated from a nonencapsulated strain of K. pneumoniae (Kp-MPG) several derivatives were prepared and characterized. One of them, an acylpoly (1, 3) galactoside (APG) is presently evaluated in the clinic for its capacity to label inflammatory foci when administrated as a Technetium99 conjugate (13). We report here the binding of Kp-MPG and its derivatives to leukocytes as assessed by indirect immunofluorescence with labeled antibodies to Kp-MPG. The availability of several derivatives enabled us to exclude the contribution of certain defined parts of the APG molecule to the binding to leukocytes. Furthermore, we report the differential fixation of APG to lymphocyte subsets, as assessed by double-labeling experiments.

Binding of Kp-MPG and Its Derivatives to Leukocytes· 185

Material and Methods Preparation of Kp-MPG and its derivatives The MPG from a non encapsulated strain of K. pneumoniae (Institut Pasteur 1-145) was obtained as already described (10). Kp-MPG activity was 100 times lower than that of Escherichia coli LPS in the Limulus amoebocyte lysate assay. LPS from the same strain of K. pneumoniae (Kp-LPS) was prepared according to WESTPHAL and JANN (14). Kp-LPS submitted to alkaline hydrolysis (NaOH 0.5 N, 1 h, 56 QC), designated as Kp-LPS (NaOH) was studied along with Kp-LPS. Analytical data on Kp-MPG, Kp-LPS and Kp-LPS (NaOH) are presented in Table 1. LPS from E. coli (strain 0111 B5) and from S. minnesota were purchased from Sigma (St. Louis, MO, USA). Four different derivatives were prepared from Kp-MPG. Their analytical composition is presented in Table 1. A first preparation was made by mild alkaline hydrolysis (NaOH 0.1 N, 24 h, 22 QC) followed by preparative gel chromatography (Sephacryl S200 HR, Pharmacia, Uppsala, Sweden, with Tris 0.01 MEthylene diamine tetracetic acid 10 mM buffer, pH 7.4) yielding rwo fractions: an homogeneous 34 kDa fraction containing ester-linked fatty acids (EFA-APG) and> 100 kDa fraction designated as APG pol which was characterized as a polymer of EFA-APG. A second preparation was obtained from Kp-MPG by two cycles of drastic alkaline hydrolysis (NaOH 0.5 N, 1 h, 56 QC) with delipidation (Chloroform/methanol 3:1 v/v) followed by gel chromatography yielding the 34 kDa fraction APG. This fraction was submitted to acid hydrolysis (acetic acid 1 % v/v, 45 min, 100 QC). The precipitate which contained all the fatty acids was discarded and the supernatant which contained mostly poly (1,3) galactose chains was designated as GC-APG. The common structure of all 34 kDa derivatives is a linear polysaccharide chain composed exclusively of D-galactose and containing three anomers of D-galactose: (a Gal p), (~ Gal p) and (~ Gal f). The polysaccharide chain is terminated by a core-like region, the structure of which has not yet been completely characterized in the absence of R-mutant (galactose free) of the strain. The lipid part contains rwo glucosamine residues each bearing one phosphomonoester group (in positions 1 and 4) and one amide linked C t4 ~ hydroxymyristic acid. EFA-APG differs from APG by the presence of additional ester-linked C t4 myristic acids and C t6 palmitic acid. These ester-linked fatty acids were present in the same proportion in APG pol (Table 1).

Preparation of biotinylated F(ab'}2 fragments [biot-F(ab'}21 of anti-Kp-MPG antibodies Anti-Kp-MPG serum was prepared by repeated injections of Kp-MPG in rabbits until obtaining a titer of about 50,000; defined as the reciprocal of the serum dilution permitting the development of an absorbance of 0.2 in ELISA. Kp-MPG was used as solid phase, peroxidaseconjugated goat anti-rabbit IgG Oackson Immunoresearch, Baltimore, MD, USA) as secondary antibody and o-phenylenediamine (Sigma) as substrate. The detailed procedure was adapted from a previously described ELISA (15). Purified F(ab'lz fragments were prepared from IgG by a combination of pepsin digestion and immunoaffinity chromatography as described (16). F(ab'lz fragments were biotinylated by using the succinimidyl-biotin (IBF biotechnics, Villeneuv~ la Garenne, France) according to the manufacturer's instructions. The biot-F(ab')2 fragments were shown to react in ELISA with Kp-MPG and EFA-APG as solid phases, streptavidin-biotinylated horseradish peroxidase complex (Amersham Int pic, Amersham, UK) being used as secondary reagent. In order to minimize the non-specific binding to cells, the biot-F(ab')2 were extensively absorbed on human leukocyte cell suspensions.

Reactivity of the biot-F(ab'h fragments with K. pneumoniae extracts Liquid phase absorption of antibody (17) was used as a quantitative assay to determine the capacity of K. pneumoniae derivatives to bind to biot-F(ab'lz fragments. In brief, biot-F(ab'lz fragments (10 !-Ig/ml, final concentration) were mixed with increasing amounts of either derivative (0.08 to 500 !-Ig/ml in final) in conical Eppendorf tubes and the mixture was

650 30 54 91.0 2 ) 20.9 6.8 8.5

620 30 43 102.02 ) 28.0 7.0 9.0

1291 ) 4.8 26 450 9.8

5.9 18.7

1) Content expressed as I-1g/ mg of total mass (not corrected for water content) 2) Peptidic contamin ants which can be removed by alkaline treatment

Galactose Ketodeoxyoctonic acids Glucosamines Arninoacids C 14 ~O H - my ristic acids (amide-linked) Ester-linked C 14 fatty acids Ester-l inked C 16 fatty acids

APG pol

Kp-MPG

EFA-APG

Table 1. Analytic composition of Kp-MPG, Kp-LPS and their derivates

0.0 0.0

700 30 39 9.7 16.2

APG

0.0 0.0

870 10 20 5.0 0.0

GC-APG

10.0 11.7

542 34.1 47.5 10.0 45.0

Kp-LPS

9.4 7.0

524 31.3 46.6 8.0 38.0

Kp-LPS (NaOH )

00

~

~

~

>-

s::

::r:

~

'"

Binding of Kp-MPG and Its D erivatives to Leukocytes . 187 incubated fo r 30 min at 37 °C. The precipitating complexes were removed by centrifugation (15 min at 4 °C) in an Eppendo rf centrifuge and the amount of antibodies in the supernatant was determined in ELISA with EFA-APG as solid phase. Absorbance at 492 nm (reference 620 nm) was inversely proportional to the capacity of the listed compound to complex biot-F(ab'}z fragments.

Labeling of APC and EFA -A PC with flu orescein isothiocyana te APG or EFA-APG were dissolved (8 mg/ml) in sodium carbonate buffer (0.1 M and pH 9.5) and mixed (v/v) with isothiocyanate (Isomer I , Sigma) at 8 mg/ml in sodium carbonate buffer (0.1 M and pH 9.5). The mixture was incubated for 30 min at 3rc on a rotary shaker, then dialysed 48 h against NaCl (0.15M, 4 °C). Unbound isothiocyanate was removed by filtrati on on an Ultrogel column (AcA202, IBF). Fluorescein isothiocyanate conjugated (FITC)-APG, or -EFA-APG , was collected, yl ophilized and kept at 4°C in the dark until use. FITC content of the ligand was determined by absorbance at 492 nm by reference to a standard curve of fluorescein isothiocyanate. The molar ratios FITC/ APG and FITC/ EFA-APG were 1.8 and 1.4, respectively.

Leukocyte preparation Blood samples of healthy volunteers, collected on heparin (20 IU/ml), were fractionated by standard Ficoll-Hypaque (LymphoprepTN, N ycone A.S., N orway) density gradient centrifugation. Mononuclear cells (MNC) were harvested from the interface Ficoll-plasma, diluted in Hanks' balanced salt solutio n (HBSS, Vietech, St Bonnet de Mure, France) and centrifuged at 4 °C for 10 min (400 x g) followed by two washes in HBSS 10 min at lOa x g to remove the platelets. Polymorphonuclear cells (PMN) were obtained from the pellet after brief hypotonic lysis of erythrocytes and washed as MNC.

Cell lines The human promyelocytic HL60 and monocyte-like U937 cell lines were grown in humidi fied atmosphere of 5 % CO 2 at 37 °C in RPM I 1640 medium (Vietech) supplemented with 10 % heat-inactivated fetal calf serum (Seromed, N oisy-Ie-Grand, France), 2 mM glutamine, penicillin (100 IU/ml) and streptomycin (lOa fJ-g/ml). For activation, cells were adjusted to a concentration of 5 X 10s/ml and treated with recombinant human interferon-y (IFN-y, kindly provided by Dr. WEISSMA NN, Zurich) for 48 h.

Monoclonal antibodies Phycoerythrin (PE) conjugated mAbs anti- Leu-3a (CD 4), anti-Leu-2a (CD 8), and antiLeu-l1 c (CD 16, NK cells) were obtained from Becton Dickinson (Mountain View, CA, USA). P E-co njugated B 1 (anti-CD 20, B cells) was from Coulter (Coultronics, Mergency, France). Unconjugated mAbs (yl isotype) IO Tl 6 (anti-CD Il a), 10M 1 (anti- CD 11 b) and 10M II C (anti-CD li e) were from Immunotech (Marseille, France). BL 5 (anti-CD 18, y l isotype) was previously described (18) and B-A8 (anti-CD 14, y l isotype) was kindly provided by Dr. W'JDENES (Centre de Transfusion, Besan~on, France). A yl isotype mAb (8545 A), anti-secretory component (19), was used as irrelevant control antibody. The unconjugated mAbs w ere used at 10 fJ-g/ml. Their binding was revealed by a seco nd step incubation with FITC goat anti-mouse Ig (Tago Immunologicals, Burlingame, CA, USA). The anti-Mac-2 mAb was from Boehringer M annheim (Mannheim, Germany), it was revealed by a FITC goat anti-rat Ig (Jackson Immunoresearch).

Binding of Kp-MPC and its derivatives 6

to

human leukocytes

Suspensions of 10 MNC or 2.5 X 105 PMN were washed twice with binding buffer (PBS at 1 % BSA and 0.1 % azide: PBS-BSA-Az) and the pellet was stained in binding buffer by either followin g procedures: (I) direct immunofluorescence by using the FITC-labeled ligands (20 fJ-l!pellet) 30 min at 4 °C followed by two washes in ice-cold PBS-BSA-Az. (2) indirect immunofluorescence usin g a method that consists of three s tage incubation at 4 °C of 30 min

188 . Z. HMAMA et al. duration: (a) K. pneumoniae extract (20 ttl), (b) biot-F(ab')2 anti-Kp-MPG (100 ttg/ml, 20 ttl) (c) Streptavidin-PE (SAPE) (Catlag laboratories, San Francisco, CA, USA) (1 :20,20 ttl). After each step, unbound reagents were removed by two washes in PBS-BSA-Az and cells were fixed with 1 % formaldehyde in PBS-BSA-Az. Cell lines HL60 and U937 (2.5 x lOs cells/ pellet) were stained as MNC and PMN. To analyze the binding of the K. pneumoniae extracts to subpopulations of lymphocytes, FITC-ligand labeled cells were further incubated for 30 min at 4 DC with specific surface PEconjugated markers, then the cells were washed twice and fixed.

Flow cytometry Cell suspensions were analyzed by using a FACScan cytofluorometer (Becton Dickinson), which was equipped with an argon-ion laser operating at 488 nm and 250 mV light output. To analyze the total population of lymphocytes, monocytes or polymorphonuclear cells separately, side scatter parameters were used to apply optimal computerized gating (5000 events by subset). Relative fluorescence intensity of each subset was recorded as single parameter histograms (log scale, 256 channels, 48 channels/log decade). Fluorescence measurements were converted from logarithmic to linear using a logarithmic-linear calibration factor, and the rjb cts were expressed as specific mean fluorescence intensity (L'.. MFI) after subtraction of the background MFI (control: cell + buffer + biot-F(ab')2 + SAPE; or cell + buffer).

Results

Specificity of biot-F(ab'JJ fragments for the polysaccharidic chain of Kp-MPG derivatives The relative capacities of K. pneumoniae derivatives to bind biot-F(ab'h were measured by liquid-phase absorption/ELISA (Fig. 1). In spite of structural differences, the antibody binding capacities of the EF A -APG,

~

E

2.0

• EFA-APG 0 APGpol 0 Kp-LPS • APG 11 GC-APG Ell Kp-LPS (NaOH) .A. E. coli- LPS

m

•

Kp-MPG EFA-APG

0 Ell

Kp-LPS Kp-LPS (NaOH)

c

~

t)

c

oa

-eo

0.5

til

.0

<

0.0 +.,.,'""'"". . . . . . ..,..---.-......,......"...!~~~ lO-3 lO-2 10- 1 lOo 10 1 10 2 lO3 lO-3 lO-2 lO-1 lOo

lOl

10 2 lO3

Absorbent (Ill!/ml) Figure 1. Reactivity of biot-F(ab'h anti-Kp-MPG with the 34 kDa derivatives, APG pol, and Kp-LPS, Kp-LPS (NaOH) and E. coli LPS. The biot-F(ab'h fragments were absorbed on the mentioned derivatives as described in Materials and Methods and the remaining antibodies were assessed in ELISA with EFA-APG (left) or Kp-MPG (right). Results are expressed as absorbance at 492 nm (reference 620 nm) as function of concentration of antigen used in absorption.

Binding of Kp-MPG and Its Derivatives to Leukocytes . 189

APG and GC-APG were nearly identical as shown by the superposition of the corresponding absorption curves and the complete absorption of the antibodies. The binding site of biot-F(ab'h fragments was therefore limited to the common hydrophilic polysaccharidic chain of these derivatives. The polymeric derivative (APG pol) displayed a binding capacity greater than that of its monomeric counterpart. In addition both Kp-LPS and Kp-LPS (NaOH) could completely absorb the anti-APG F(ab'h antibody fragments indicating that LPS and APG from the strain shared similar polysaccharidic antigenic determinants (Fig. la). Conversely anti-APG antibodies did not cross-react with LPS from E. coli (Fig. la) nor with LPS from S. minnesota (not shown). U sing the same experimental device with Kp-MPG instead of EF A -APG as solid phase, we observed only a partial absorption of the antibodies by the 34 kDa derivatives and by both forms of Kp-LPS, whatever their concentration, indicating that most of Kp-MPG epitopes were not expressed on its derivatives nor on Kp-LPS (Fig. 1b).

A ~ ~

B

MONOCYTES

MONOCYTES

LYMPHOCYTES

LYMPHOCYTES

~ I()

~

u.l CO

:2 ~

Z

...J ...J u.l

a b

U

u.l

> f::
...J u.l ~

~ I()

FLUORESCENCE INTENSITY Figure 2. Histograms of red fluorescence intensity (log scale) derived from 5000 events after electronical gating of the cytograms. 106 MNC or 5 x 105 PMN were incubated with Kp-MPG (100 [lg/ml) (panel A), or EFAAPG (3 [lM) (panel B) for 30 min at 4°C, then the fixed ligands were revealed by subsequent incubation with biot-F(ab')2 anti-Kp-MPG and SAPE as described. Background fluorescence (cells incubated in absence of ligands) is shown as •• a» histograms; and fluorescence with KpMPG or EFA-APG as .. b» histograms.

190 . Z . HMA MA et al.

Binding of Kp-MPG and EFA-APG

to

leukocytes

Knowing that biot-F(ab')z fragments recognized the Kp-MPG and all derivatives under study, we investigated the binding of these compounds to human leukocytes and revealed it by indirect immunofluorescence, by using biot-F(ab'h/SAPE as amplification system. The fluorescence histograms in Figure 2 demonstrate that Kp-MPG and EFA-APG were fixed on all monocytes of the suspension, while lymphocytes gave a very weak but heterogenous signal, suggesting that only a lymphocyte subset was labeled (Fig. 2). With regard to PMN, they bound EFA-APG but not Kp-MPG.

Differential binding of Kp-MPG derivatives In order to compare the capacity of monocytes, lymphocytes and PMN to interact with different Kp-MPG derivatives, cells were processed as described and the specific mean fluorescence intensity achieved at various concentrations of the tested derivatives was established (Fig. 3). Kp-MPG, EFA-APG, and APG were shown to be taken up in dose-dependent manner by monocytes, and less by lymphocytes and PMN. Of note APG and EFA-APG could label monocytes when introduced at concentrations as low as 0.15 !-1M and were specific for these cells since lymphocytes and PMN required lO-time higher concentrations to be labeled. Conversely, APG pol showed little binding to monocytes, whereas GC-APG showed no binding except at the highest concentrations tested. Native Kp-LPS did not bind to leukocytes under these experimental conditions, but LPS modified by alkaline treatment [Kp-LPS (NaOH)] displayed binding characteristics similar to those of APG and EF A -APG (Fig. 3). In view of the major differences in biological properties between EF AAPG and APG, their binding capacities to monocytes at 0.3 and 3 f.tM were measured on eight different cell suspensions. For EFA-APG, ~ MFI values were 68.8 ± 6.5 (mean ± SD) at 0.3 f.tM and 84.8 ± 3.9 at 3 f.tM. For APG, ~ MFI was 69.7 ± 8.4 and 81.6 ± 6.9, respectively. Comparison of paired values by using the Wilcoxon test did not reveal any significant difference. Lymphocyte subsets recognized by APG and EPA-APG In view of the very low and heterogenous fixation of EF A -APG on lymphocytes (Fig. 2) we performed double labeling experiments in order to determine possible differences of binding among lymphocyte subsets. Results obtained with FITC-APG are shown in Figure 4. A very weak binding of APG could be demonstrated on all lymphocyte subsets but the ~ MFI was slightly greater in NK cells than in CD4+ or CD8+ lymphocytes and further elevated in B cells. Note that the binding was homogeneous within each subset. Therefore the heterogeneity of the fluorescence histogram of the whole population of lymphocytes (Fig. 2) can be attributed to the greater fixation of APG on B cells and intermediate on NK cells. Identical results were obtained with EFA-APG (data not shown).

Binding of Kp-MPG and Its D erivati ves to Leukocytes . 191 o o o o

o

III

o

o o

III

o III

o III

o

....o

III

'"

o

o ..,

'..," "0....

~ E 0.. o :I.

o

'"

o.... o o

o o <-

o

o <-

o

~ :I.

o

.,.o

Vl

o

o <-

o

o o

v

ill'.! V

Figure 3. Binding of Kp-MPG, its four derivatives Kp-LPS and Kp-LPS (NaOH) to human leukocytes. The indicated ligands were added to MNC (106 ) or PMN (5 x 105 ) for 30 min at 4 °C, then the binding was revealed as in Figure 2. Results are expressed as specific MFI (ll. MFI) after subtraction of the MFI values correspondig to the cells incubated without li gands. Native KpLPS is shown as closed symbols and Kp-LPS (NaOH) as open symbols.

192 . Z.

HMAMA

et al.

::r

A

B

C

D

(38.5)

(46.5)

(49.5)

(68.7)

GREEN FLUORESCENCE Figure 4. Binding of FITC-APG to lymphocyte subsets. After incubation of MNC with FITC-APG (25 f-IM, 30 min at 4 QC), cells were further incubated with PE conjugated mAb (A) anti-Leu-3a (CD4+), (B) anti-Leu-2a (CDS+), (C) anti-Leu llc (NK cells) and (D) B 1 (anti-CD 20, B cells). Red and green stained cells were identified by FACS contours; the bar indicates the limit of green fluorescence background (non-stained cells) and the number in brackets indicate II MFI (calculated as in Fig. 3) corresponding to FITC-APG binding to identified cell population.

Binding of Kp-MPG and its derivatives according to myelomonocyte differentiation Several experiments addressed a possible relationship between APG binding and the stage of differentiation of cells of the myelomonocytic lineage. For this purpose we compared the binding of Kp-MPG and APG with the expression of integrins of the CD l1/CD 18 family and the CD 14 antigen on various cells (Table 2). The immature myelomonocytic cell line HL 60 did not bind the Kp-MPG and showed only a borderline binding of APG. The promonocytic cell line U937 bound APG and Kp-MPG in greater amounts than did HL 60, though much less than monocytes or PMN. U 937 cells were induced to partial differentiation by incubation with IFN-y from 10 to 1000 IU/ml during 48 h. The dose-dependent response to IFN-y was characterized by increased uptake of Kp-MPG and APG along with increased expression of CD 11 b and CD 14 antigens (Table 2). When blood monocytes were incubated with IFN-y (100 IU/ml for 48 h) we observed a significant increase of the binding of Kp-MPG and APG, along with an up-regulation of the Mac-2 antigen (20) a slight downregulation of CD 14 without notable change in CD l1/CD 18 (Table 1). As regards alveolar macrophages their binding of APG was comparable to that of monocytes, whereas they expressed higher density of Mac-2 antigen. Discussion In the present study, we have investigated the binding properties of KpMPG, a proteoglycan extracted from K. pneumoniae. Several 34 kDa components were isolated from Kp-MPG. APG, the prototype 34 kDa component, comprises a long polysaccharidic chain, a complex core and a lipid part. Its overall structure therefore bears similarities with that of LPS

APG

7.3 15.5 26.0 28.7 31.2 74.5 104.9 47.3 80.4

Kp-MPG

0.0 3 ) 3.5 9.6 9.3 11.1 55.6 88.9 19.8 ND

K. pneumoniae extracts

36.3 29.9 31.2 28.8 28.4 60.7 54.6 38.6 ND

BL 5/ CD 18 30.4 28.6 29.5 27.2 26.5 63.4 60.3 35.3 ND

lOT 16/ CD 11 a 1.3 10.0 12.7 13.1 14.9 72.9 69.4 63.1 44.4

10M 1/ CD 11 b 0.1 0.6 0.4 0.4 0.5 63.9 56.3 41.3 ND

10M 11 C/ CD 11 c

Antibody

0.5 12.4 17.5 23.2 25.3 66.4 54.3 9.3 44.0

B-A 8/ CD 14

ND4) ND ND ND ND 7.4 21.6 ND 23.2

M 3/38/ Mac-2

I) Cells were incubated with Kp-MPG (10 !-Ig/ml) or APG (0.3 !-1M) or mAbs (30 min at 4 QC). Binding of Kp-MPG and EFA-APG was revealed as in Figure 2 and that of mAbs by a second incubation with FITC goat anti-mouse Ig. 2) Alveolar macrophage were obtained by bronchoalveolar lavage from a patient with lung carcinoma (controlaterallung). 3) ~ MFI for the binding of Kp-MPG and APG were determined as in Figure 2. ~ MFI corresponding to the binding of the mAbs were obtained after subtraction of the MFl from cells stained with mAb 8545A. 4) ND: not determined.

HL 60 U 937 U 937 + IFN-y 10 100 1000 IU/ml Monocytes Monocytes + IFN-y 100 u/ml PIVN Alv. macrophages 2 )

Cells l )

Table 2. Binding of Kp-MPG and APG to myelomonocytic differentiation and surface expression of integrins CD14, and Mac-2 antigens

tp

-'"'"

D:

(")

;,

5"

c

(t)

r<

8

D:

<:

~.

:;r

(t)

..,

tJ

V>

..... ....

0...

~

'"CJ

~

.q

o>+>

(JQ

s·

0...

5·

194 . Z.

HMAMA

et al.

from Gram negative bacteria but differs from them by the composition of the hydrophobic pole. The lipid A is made of a biphosphoglucosamine with 4 chains of C 12 to C 14 fatty acids. Some degree of heterogeneity of the lipid A structure has been reported even in LPS derived from the same strain (e.g. Salmonella abortus eqUl), with regard to relative amounts and length of fatty acid chains (21, 22) but all lipid A molecules appear to have both amide-linked and ester-linked fatty acids. These fatty acids are present in Kp-MPG, EFA-APG and APG pol, but the drastic alkaline hydrolysis used to prepare APG resulted in the complete loss of ester-linked C 16 and C 14 fatty acids (Table 1). While most LPS are strongly hydrophobic, APG though still amphiphatic is highly hydrophilic. Two APG-related compounds provided critical information in the present study: EFA-APG which contains additional ester-linked C 14 and C 16 fatty acids with some contaminating peptides and GC-APG which lacks the biphosphodiglucosamine residues and the fatty acids. The binding of Kp-MPG and its derivatives was assessed by amplified indirect immunofluorescence, using biotinylated F(ab'h fragments of antibodies to Kp-MPG and flow cytometry. This method ensured exquisite sensitivity and non-specific signal could be minimized by the use of F(ab'h fragments instead of whole IgG antibodies, and by extensive absorption of the antibodies on human leukocytes. The three 34 kDa derivatives, which contain the same saccharidic chain but differ by their hydrophobic regions, demonstrated identical antibody binding capacities on a molar basis, suggesting that most, if not all of the antibodies measured in this assay recognized the polysaccharidic chain. This observation is in keeping with the literature on the antigenicity of LPS. Nearly all the antibodies produced during gram negative bacterial infections recognize the polymorphic polysaccharidic chain (<<0» antigen) allowing to define serotypes (23), while production of antibodies against the hydrophobic region of LPS requires the use of rough LPS (24) and has often proved unsuccessful. All anti-APG antibodies cross-reacted with Kp-LPS but not with other LPS such as E. coli and S. minnesota rough LPS, suggesting that identical or cross-reactive epitopes are expressed on the polysaccharidic chains of LPS and MPG of the same strain. Conversely, most Kp-MPG epitopes were not expressed on Kp-LPS or APG. We show here that Kp-MPG, EF A -APG and APG bind to human leukocytes. The binding was studied at 4°C in order to minimize non specific hydrophobic interactions between bacterial lipids and cell membranes, and because bound Kp-MPG or APG were rapidly internalized at 3rc (data not shown). The binding was concentration-dependent and of different magnitude on monocytes, PMN and lymphocytes subsets. KpMPG but even more so APG and EFA-APG bound preferentially to monocytes and much less to PMN and lymphocytes. At very low EFAAPG and APG concentration (from 0.15 to 0.3 [lM), only monocytes but not other cell types were labeled. Increasing concentrations induced a progressive increase of the amount of APG or EF A -APG bound per cell,

Binding of Kp-MPG and Its Derivatives to Leukocytes . 195

and permitted detection of a dose-dependent binding to PMN and to lymphocytes. This profile of response suggests that several cell surface molecules, differentially expressed on monocytes, PMN and lymphocytes, could be involved in Kp-MPG and APG binding, as is the case for LPS (25-27). As regards lymphocytes, each subpopulation appeared to express a homogeneous number of APG binding sites in the following decreasing order: B cells, NK cells, CD8+ and CD4+ lymphocytes. In contrast, LPS from rough S. minnesota was reported to bind to human B lymphocytes but not to NK cells (28). The lack of binding of native Kp-LPS was unexpected. It cannot be attributed to a technical failure, since this LPS was readily recognized by the F(ab')2 fragments. Furthermore, no binding of FITCKp-LPS could be demonstrated (data not shown). In contrast Kp-LPS (NaOH) displayed binding characteristics similar to those of APG, suggesting that the structures involved in binding to cell membranes could be hindered in native LPS but exposed after alkaline treatment. Immature myelomonocytic (HL 60) or monocytic (U 937) cell lines expressed very low Kp-MPG and APG binding capacities as compared with monocytes or PMN. However exposure of either U937 cells or monocytes to IFN-y markedly increased their capacity to bind Kp-MPG or APG, in parallel with the augmented expression of CD 11 blCD 18 integrin and CD 14 molecule on U 937, and Mac-2 antigen, a macrophage lectin specific for galactose (20), on monocytes (Table 2). These observations support a possible role of these molecules as APG receptors, in agreement with blocking experiments using monoclonal antibodies (HMAMA et al., manuscript in preparation). The correlation between CD 14 expression and APG binding observed at different maturation stages of the myelomonocytic lineage (Table 2) can be taken as further indirect evidence for a contribution of CD 14 to APG binding. Some conclusions can be drawn from our study as regards the structures derived from Kp-MPG which may be involved in the binding to leukocyte membranes. By exclusion it appears that the lipid fraction remaining in APG andlor the core part contained in the 34 kDa derivatives are sufficient to ensure the binding properties of these compounds. The polysaccharidic chain made of (1,3) galactose residues cannot contribute to Kp-MPG or APG binding to monocytes, in view of the lack of demonstrable fixation of GC-APG which bears the same saccharidic chain as APG. Accordingly the lack of binding of APG pol was expected since this compound is likely to result from hydrophobic interactions of EFA-APG molecules, resulting in the masking of the hydrophobic pole by the polysaccharidic chain. The unimpaired fixation of polysaccharide specific antibodies to membranebound APG or EFA-APG could be taken as further evidence against the masking of polysaccharide epitopes by ligation to monocyte surface lectins, but the structure of the chain made of 150 residues of galactose would definitely permit the simultaneous binding of lectins and antibodies specific for identical epitopes. However the present data do not exclude the possibility that the polysaccharidic chain could be involved in the binding

196 . Z. HMAMA et a!.

to galactose receptor, such as the Mac-2 expressed by activated macrophages (29), fibroblasts (30) and some tumor cell lines (31), but very weakly or not on peripheral blood leukocytes as shown in the present study. The identical binding capacities of APG and EFA-APG, which differ by the presence of ester-bound C 14 and C 16 fatty acids, suggest that these fatty acids are unlikely to contribute to the binding. This is remarkable because EFA-APG, but not APG, can trigger IL-1~, IL-6, and TNF-a synthesis as well as polyclonal B cell activation, indicating a critical role of the esterified fatty acids in the biological activities of these derivatives (32). Quite different observations were reported with Bordetella pertussis Lipid A whose esterified fatty acids were shown to be critical for both binding and biological activities (33). We may conclude that different molecular substructures of Kp-MPG derivatives are involved in the membrane binding, and the induction of cellular responses, respectively. Acknowledgements We thank Dr. C. VINCENT and M.-J. GARRIAZZO for expert advice and help in immunochemical techniques, Dr. S. DELASSAN and DR. T. BAUSSANT for all data concerning the chemical analysis of the compounds, Dr. A. LE PAPE for preparation of the anti-Kp-MPG serum and Dr. J. F. MORNEX for bronchoalveolar lavage sample. Cytofluorometry analysis was achieved in the Centre Commun de Cytofluorometrie (Univ. Claude Bernard, Lyon). Part of this work was supported by Institut National de la Sante et de la Recherche Medicale and Centre National de la Recherche Scientifique. Dr. Z. HMAMA was supported by a Moroccan Government fellowship.

References 1. MORRISSON, D. C. and J. L. RYAN. 1987. Endotoxin and disease mechanisms. Ann. Rev. Med. 38: 417. 2. LACHMAN, E. and S. B. PITSOE. 1984. Anti-lipopolysaccharide immunotherapy in management of septic shock of obstetric and gynaecological origin. Lancet i: 981. 3. TENG, N. N. H., H. S. KAPLAN,]. M. HEBERT, C. MOORE, H. DOUGLAS, A. WUNDERLICH, and A. 1. BRAUDE. 1985. Protective against Gram-negative bacteremia and endotoxemia with human monoclonal IgM antibodies. Proc. Nat!' Acad. Sci. USA 82: 1790. 4. RAETZ, C. R. H. 1990. Biochemistry of endotoxins. Ann. Rev. Biochem. 59: 129. 5. FREUDENBERG, M. A. and C. GALANOS. 1990. Bacterial lipopolysaccharides: structure, metabolism and mechanisms of action. Int. Rev. Immuno!. 6: 207. 6. TAHRI-JOUTI, M.-A. and R. CHABY. 1990. Specific binding of lipopolysaccharides to mouse macrophages 1. Characteristics of the interaction and inefficiency of the polysaccharide region. Molec. Immuno!. 27: 751. 7. LEI, M. G., T. Y. CHEN, and D. C. MORRISON. 1990. Lipopolysaccharide/lipid A receptors on lymphocytes and macrophages. Int. Rev. Immunol. 6: 223. 8. WRIGHT, S. D. 1991. Multiple receptors for endotoxin. Curr. Op. Immunol. 3: 83. 9. STEWART-TuLL, D. E. S. and M. DAVIES. 1985. Immunology of the bacterial cell envelope. John Wiley and Sons, New York. 10. NORMIER, G., A. M. PINEL, L. DUSSOURD O'HINTERLANO, U. RAMSTEOT, and H. WIGZELL. 1985. NK cell stimulating properties of a membrane proteoglycan from noncapsulated Klebsiella pneumoniae biotype a. Acta Path. Microbio!. Immuno!. Scand. Sect. C 93: 233.

Binding of Kp-MPG and Its Derivatives to Leukocytes . 197 11. MILLET, I., S. LAFFONT, A. DEFRAISSINETTE, M. JEANNIN, J. P. REVILLARD, G. NORMIER, and L. DUSSOURD D'HINTERLAND. 1987. Polyclonal activation of murine B cells by a membrane proteoglycan of Klebsiella pneumoniae. Clin. expo Immunol. 70: 201. 12. VACHERON, F., S. PERIN, E. KODARI, P. SMETS, R. ZALISZ, and M. GUENOUNOU. 1989. Immunological activities of RU-41740, a glycoprotein extract from KlebsielJa pneumoniae. III. Role of LPS-like and LPS-non-related molecules. Res. Immunol. 140: 159. 13. DIOT, P., A. LEPAPE, D. NOLIBE, G. NORMIER, H. BlNz, J. P. REVILLARD, G. LASFARGES, M. LAVANDIER, and E. LEMARIER 1991. Scintigraphy with JOOIX, a KlebsieJla membrane glycolipopeptide, for the early diagnosis of chronic berylliosis. Results in an experimental model of berylliosis in baboons. Brit. J. Indust. Med. (in press). 14. WESTPHAL, O. and K. JANN. 1965. Bacteriallipopolysaccharides: extraction with phenol/ water and further applications of the procedure. Meth. Carbohydr. Chern. 5: 83. 15. VINCENT, C. and J. P. REVILLARD. 1985. Differential measurement by ELISA of free and bound al-microglobulin in human serum without prior fractionation. J. Immunol. Meth. 82: 111. 16. HARLOW, E. and D. LANE. 1988. Antibodies, a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 726 pp. 17. BAUMGARTNER, J. D., T. X. O'BRIEN, T. N. KIRKLAND, M. P. GLAUSER, and E. J. ZIEGLER. 1987. Demonstration of cross-reactive antibodies to smooth Gram-negative bacteria in antiserum to Escherichia coli J5. J. Infect. Dis. 156: 136. 18. KRAFT, N., A. STEIN OKLAY, N. M. THOMSON, and R. C. ATKINS. 1987. Functional effects of non-lineage antibodies. In: McMICHAEL, A. J. (ed.), Leukocyte typing, vol. 4. Oxford University Press, Oxford, p. 864. 19. VINCENT, C. and J. P. REVILLARD. 1988. Sandwich-type ELISA for free and bound secretory component in human biological fluids. J. Immunol. Meth. 106: 153. 20. CHERAYIL, B. J., S. CHAITOVITZ, C. WONG, and S. PILLA!. 1990. Molecular cloning of human macrophage lectin specific for galactose. Proc Nat!. Acad. Sci. USA 87: 7324. 21. GALANOS, c., B. H. JIAO, T. KOMURO, M. A. FREUDENBERG, and o. LODERITZ. 1988. Large-scale fractionation of S-form lipopolysaccharide from SalmoneJla abortus equi. Chemical and serological characterization of the fractions. J. Chromatogr. 440: 397. 22. JIAO, B., M. FREUDENBERG, and C. GALANOS. 1989. Characterization of the lipid A component of genuine smooth-form lipopolysaccharide. Eur. J. Biochem. 180: 515. 23. CROSS, A. S., H. SIDBERRY, and J. C. SADOFF. 1989. The human antibody response during natural bacteremic infection with Gram-negative bacilli against lipopolysaccharide core determinants. J. Infect. Dis. 160: 225. 24. BAUMGARTNER, J. D., J. A. MCCUTCHAN, G. V. MELLE, M. VOGT, R. LUETHY, M. P. GLAUSER, E.J. ZIEGLER, M. R. KLAUBER, E. MUEHLEN, R. CHIOLERO, and S. GEROULANOS. 1985. Prevention of Gram-negative shock and death in surgical patients by antibody to endotoxin core glycolipid. Lancet ii: 59. 25. WRIGHT, S. D., and M. T. C. JONG. 1986. Adhesion-promoting receptors on human macrophages recognize E. coli by binding to lipopolysaccharide. J. Exp. Med. 164: 1876. 26. WRIGHT, S. D., S. M. LEVIN, M. T. C. JONG, V. Z. CHAD, and L. G. KABBASH. 1989. CR-3 (CD lIb/CD 18) express one binding site for Arg-Gly-Asp-containing peptides, and a second site for bacterial lipopolysaccharide. J. Exp. Med. 169: 175. 27. WRIGHT, S. D., R. A. RAMOS, P. S. TOBIAS, R. J. ULEVITCH, and J. C. MATHISON. 1990. CD 14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science 249: 1431. 28. WEERSINCK, A. J. L., K. P. M. VAN KESSEL, R. TORENSMA, J. A. G. VAN STRI]P, and J. VERHOEF. 1990. Binding of rough lipopolysaccharides (LPS) to human leukocytes: inhibition by anti-LPS monoclonal antibody. J. Immunol. 145: 318. 29. Ho, M. K. and T. A. SPRINGER. 1982. Mac-2, a novel 32,000 Mr mouse macrophage subpopulation-specific antigen defined by monoclonal antibodies. J. Immunol. 128: 1221. 30. MOUTSATOS, I. K., M. WADE, M. SCHINDLER, and J. L. WANG. 1987. Endogenous lectins from cultured cells: nuclear localization of carbohydrate binding protein 35 in proliferating 3T3 fibroblasts. Proc. Nat!. Acad. Sci. USA 84: 6452.

198 . Z. HMAMA et al. 31. RAz, A., G. PAZERINI, and P. CARMI. 1989. Identification of the metastasis-associated, galactoside-binding lectin as a chimeric gene product with homology to an IgE-binding protein. Cancer Res. 49: 3489. 32. HMAMA, Z., G. LINA, C. VINCENT, J. WIDJENES, G. NORMIER, H. BINz, and J. P. REVILLARD. 1992. Monocyte cytokine secretion induced by chemically-defined derivates of Klebsiella pneumoniae. Clin. Exp. Immunol. 89: in press. 33. TAHRI-JOUTI, M.-A., M. MONDANGUE, A. LEDUR, F.-I. AUZANNEAU, D. CHARON, R. GIRARD, and R. CHABI. 1990. Specific binding of lipopolysaccharides to mouse macrophages. II. Involvement of distinct lipid A substructures. Molec. Immunol. 139: 1361.

Dr. JEAN-PIERRE REVILLARD, Laboratoire d'Immunologie, INSERM U80 CNRS URA 1177 UCBL, H6pital E. Herriot, Pavilion P, 69437 Lyon Cedex 03, France