THE JOURNAL OF B~OLOCKXL CHEMISTRY Vol. 265, No. 19, Issue of July 5. pp. 11251L11258,199O 0 1990 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

Characterization of the Binding of Actinomyces naeslundii (ATCC 12104) and Actinomyces viscosus (ATCC 19246) to Glycosphingolipids, Using a Solid-phase Overlay Approach*

(Received for publication, December 19,1989)

Nicklas Stromberg and Karl-Anders Karlsson From the Department of Medical Biochemistry, University of Gb;teborg, P. 0. Box 33031, S-400 33 Gdteborg, Sweden

Actinomyces naeslundii (ATCC 12104) and Actino- myces viscosus (ATCC 19246) were radiolabeled ex- ternally (lzsI) or metabolically (36S) and analyzed for their ability to bind glycosphingolipids separated on thin layer chromatograms or coated in microtiter wells. Two binding properties were found and char- acterized in detail. (i) Both bacteria showed binding to lactosylceramide (LacCer) in a fashion similar to bac- teria characterized earlier. The activity of free LacCer was dependent on the ceramide structure; species with 2-hydroxy fatty acid and/or a trihydroxy base were positive, while species with nonhydroxy fatty acid and a dihydroxy base were negative binders. Several gly- colipids with internal lactose were active but only gan- gliotriaosylceramide and gangliotetraosylceramide were as active as free LacCer. The binding to these three species was half-maximal at about 200 ng of glycolipid and was not blocked by preincubation of bacteria with free lactose or lactose-bovine serum al- bumin. (ii) A. naeslundii, unlike A. viscosus, showed a superimposed binding concluded to be to terminal or internal GalNAc@ and equivalent to a lactose-inhibit- able specificity previously analyzed by other workers. Terminal Gala was not recognized in several glyco- lipids, although free Gal and lactose were active as soluble inhibitors. The binding was half-maximal at about 10 ng of glycolipid.

A glycolipid mixture prepared from a scraping of human buccal epithelium contained an active glyco- lipid with sites for both binding specificities.

Specific adherence as a determinant of tissue selectivity of bacterial colonization was early suggested within oral micro- biology including studies with Actinomyces naeslundii and Actinomyces viscosus (for a review, see Ref. l), which have been regarded as potential pathogens in oral infections (2-4). Both A. naeslundii and A. vi.scosus constitute typical members of the tooth plaque but differ from each other in that A. nueslundii is more prominent at oral epithelial surfaces but relatively less common at tooth surfaces (5). The molecular basis for this situation is not fully understood, although it has recently been observed that there are two antigenically dis- tinct types of fimbriae (type 1 and type 2), which differ both in their adhesive properties and their distribution between typical members of the two Actinomyces species (6, 7). Type 2 fimbriae, present on most strains of both species, are asso- ciated with a lactose-sensitive binding and mediate adherence

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

to streptococci of tooth plaque (8-lo), agglutination of human erythrocytes (11, 12), as well as adhesion to oral epithelial cells (13,14). Type 1 fimbriae, on the other hand, are thought to mediate attachment of bacteria to the salivary pellicle coating the tooth surface, and these fimbriae appear on most A. viscosus strains but not on typical A. nueslundii strains (6, 7).

The lactose sensitivity associated with type 2 fimbriae has been examined by inhibition studies with various soluble saccharides (15 and references therein), by a model aggrega- tion system based on beads coated with various glycoconju- gates (16), and more recently by the overlay technique devel- oped by us and used in the present work (17). The general conclusion from these studies has been a recognition of ter- minal Gal@ residues. However, owing to limited access to diverse natural sequences, the specificity has not been worked out in detail.

We have previously used a direct binding of bacteria to glycolipids on thin layer chromatograms (18, 19) and charac- terized carbohydrate receptors for uropathogenic Escherichia coli (20,21) and a group of bacteria classified as lactose binders (22-26),’ including N. gonorrhoeae (25). Using this approach, we now demonstrate that A. naeslundii (ATCC 12104), unlike A. viscosu.s (ATCC 19246) exhibits a high affinity binding to glycolipids with the sequence GalNAcP, and that such com- pounds probably represent receptors on human erythrocytes and oral epithelial cells for A. nueslundii. For both these Actinomyces strains, we also identified an unrelated low affin- ity binding to lactosylceramide (LacCer). At present, we do not know the biological role of this low affinity binding, although receptor glycolipids were detected in oral epithelial cells.

MATERIALS AND METHODS’

Preparation of Glycolipids-Buccal epithelium was collected from one of us fN. S.) by scraping with a spatula every other day for 2 weeks. The collected material was stored frozen in methanol until it was extracted, first twice with 20 ml of methanol at 70 “C for 30 min, and finally three times with 20 ml of chloroform/methanol (2:~ bv volume) for 30 min. From this and other tissue extracts (Table Ii, total neutral and acidic glycolipids were purified by mild alkaline degradation, dialysis, silicic acid, and DEAE-cellulose column chro- matography of acetylated and native substances, as previously de- scribed (27). Individual glycolipid species were purified by repeated

’ N. Stromberg and K.-A. Karlsson, unpublished results. * The glycolipid nomenclature and abbreviations used in this paper

follow the recommendations bv the IUPAC-IUB Commission on Biochemical Nomenclature (CB6J for lipids: Eur. J. Biochem. (1977) 79, 11-21 and J. Biol. Chem. (1982) 257,3347-3351). It is assumed that Gal, Glc, GalNAc, GlcNAc, and NeuAc are of the D configuration, Fuc of the L configuration, and that all sugars present are in the pyranose form. The abbreviations used are: PBS, phosphate-buffered saline; BSA, bovine serum albumin.

11251

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

11252 Binding of Bacteria to Glycolipids

continuous and stepwise gradient column chromatography of acety- lated and native substances, as described in detail elsewhere (26).

To generate compound no. 26, a frog brain di- and trisialoganglio- side fraction containing compound no. 32 (28) was treated with Newcastle disease virus, which specifically hydrolyzes NeuAccu2+3 Gal linkages but does not hydrolyze NeuAcnP-+GGalNAc linkages (29). A virus stock (10 ml) obtained after culturing in chick embryos, and kindly donated by Dr. S. Jeansson of the Department of Virology, University of G6teborg, was pelleted at 40,000 rpm for 3 h and finally dissolved in 0.5 ml of phosphate-buffered saline (PBS), pH 7.3. The ganglioside fraction (2 mg) was dissolved and sonicated in 100 ~1 of PBS followed by incubation with 100 ~1 of concentrated Newcastle disease virus solution at 37 “C for 18 h. A further 100 ~1 of virus was added, and the incubation continued for an additional 18 h. The incubation mixture was then desalted on a reverse-phase Sep-Pak column (Waters, Milford, MA) preconditioned with 20 ml of chloro- form/methanol (2:1, by volume), then 20 ml of methanol, and finally 20 ml of PBS. The column was eluted with 10 ml of PBS, then with 5 ml of methanol, and finally with 10 ml of chloroform/methanol (2:1, by volume). The combined methanol/chloroform fractions were dried and further purified (26).

To generate compounds nos. 23 and 24, 2 mg each of nos. 29 and 30 were incubated in 3 ml of 0.05 M HCI at 80 “C for 2 h. The reaction mixtures were then partitioned by adding 8 ml of chloroform and 4 ml of methanol to each followed by thin layer chromatography analyses using various solvent systems, including chloroform/meth- anol/H?O (60:35:8, by volume), chloroformlmethanol 5 M NH:, in HyO (60:40:9, by volume) and chloroform/methanol, 0.2% CaCI,! in H,O (w/v) 60:40:9, volume). These analyses all revealed the genera- tion of two compounds, constituting about 5-100/o of each fraction, and which both migrated slightly above their precursor glycolipids but not in the position of H-5 type 1 or H-5 type 2 (i.e. nos. 19 and 20 of Table I).

Identity of Isolated Fractions-Individual glycolipids were initially assessed for purity and quantity by thin layer chromatography and various staining reagents (30-32). Chemical structures (Table I) were confirmed by mass spectrometric and proton NMR spectroscopic analyses of permethylated and permethylated-reduced substances (33-39), and through gas chromatographic analyses of the alditol acetates obtained after acid hydrolysis of the native (40) and deriva- tized glycolipids (41).

Bacterial Strains and Labeling-A comparative study was made of A. naeslundii, strain ATCC 12104 isolated from the human sinus, and A. oiscosus, strain ATCC 19246, isolated from a case of human cervico-facial actinomycosis. For all assays, the bacteria were cultured anaerobically in PYG broth media or on Brucella agar plates for 3 days, harvested, and washed three times with PBS, pH 7.3. For binding to glycolipids, bacteria were labeled either externally with lrsI using the Bolton-Hunter reagent (42), as described earlier (18, 26) or metabolically by adding [““Slmethionine (600 pCi/mmol, 150 pCi/plate, Amersham Corp.) diluted with 50 ~1 of broth media onto 50-mm Brucella agar plates inoculated with bacteria (26). Labeled bacteria were washed three times in PBS and used immediately. The two labeling procedures gave the same results and incorporated about the same amount of radioactivity, which was 0.05-0.1 cpm/bacterium.

Binding to Solid-phase Glycolipids-The thin layer chromatogram (18, 19) and microtiter well (19) binding assays were conducted as described in detail elsewhere (26). Glycolipid chromatograms, pre- treated with plastic and bovine serum albumin (BSA), were overlaid with labeled bacteria (2 ml of 10’ cpm/ml, 10’ cells/ml), washed five times with PBS, dried, and exposed to x-ray film (XAR-5, Eastman, Kodak) for 70 h. Glycolipids in methanol (50 ~1) were dried overnight onto polyvinylchloride plates followed by blocking with BSA. The wells were then incubated with radiolabeled bacteria (50 jd/well with 5 X lo” cpm and 5 X 10” cells) for 4 h, washed five times with PBS, dried, cut out, and measured for radioactivity.

Hemagglutination and Sugar Inhibition Experiments-For hemag- glutination, 10 ~1 of fresh bacteria (lOK cells/ml) and 10 pl of human erythrocytes (4%), both suspended in PBS, pH 7.3, were mixed on a glass slide. After gentle mixing for 5 min at room temperature, agglutination was examined by the naked eye and graded. The relative ability of various saccharides to prevent hemagglutination was estab- lished by incubation of the bacteria with varying sugar concentrations for 60 min at room temperature prior to hemagglutination tests with unwashed bacteria. Bacteria treated in this way were also used to evaluate the ability of free sugars to inhibit bacterial binding to the thin layer chromatogram used in Fig. 1.

Chemicals-Lactose, galactose, and galactosamine were obtained

from Sigma, (IactoseflO-CETE)n-BSA (2-(2-carboxy-ethylthio) ethylglycosides) coupled in amide linkage to Lys; 20-30 mol/mol of BSA) and GalNAcpl-3Galcu-O-ethyl were purchased from Socker- bolaget, Arlav, Sweden.

RESULTS

Binding of bacteria to glycolipids separated on thin layer chromatograms is a convenient way of detecting and charac- terizing receptors based on carbohydrate (18-26). The ability of reference glycolipids of different carbohydrate and lipo- philic compositions to attach A. naeslundii (strain ATCC 12104) and A. uiscosu.s (strain ATCC 19246) was investigated in detail (Figs. 1 and 2 and Table I). Both a lactose- and a GalNAc@based binding property were identified.

Detection of Glycolipids Mediating in Vitro Adherence of A. naeslundii and A. uiscosus-Screening various mixtures of glycolipids (Fig. l), A. oiscosus bound selectively to certain molecular species of LacCer and to Gb3b (nos. 4 and 7 of Fig. 1 and Table I). Besides attaching to these compounds, A. naeslundii bound several additional glycolipids ranging in size from about three to eight sugars (e.g. nos. 12 and 17). Several major glycolipids (e.g. nos. 16 and 36), including monohexo- sylceramides (nos. 1 and 2), were negative, even though pres- ent in amounts of 2 pg or more, documenting the selectivity of these interactions. The binding pattern noted for both bacteria is accounted for by a recognition of terminally (no. 4) and internally located lactose (no. 7), and the second and superimposed binding restricted to A. naeslundii by a recog- nition of GalNAcP (see below).

Apparent Specificity of A. viscosus and A. naeslundii for LacCer-The glycolipid binding pattern of A. uiscosus (Table I) is typical of most of the bacteria previously classified as lactose binders (22-26). The features of this binding were recently analyzed in detail for a strain of Propionibacterium granulosum (26).

As seen in Table I and Fig. 3, LacCer (no. 4) is among the most active glycolipids (nos. 4, 9, and 13) and carries the

Anisaldehyde A.viscosus A.naeslundii

ABCDEF ABCDEF ABCDEF FIG. 1. Binding of ““I-labeled A. uiscosus (ATCC 19246)

and A. naeslundii (ATCC 12104) to thin layer chromatograms with non-acid glycolipid fractions of various tissues. Shown are representative autoradiograms of the binding of each strain and the total glycolipid pattern of the test fractions as visualized by the anisaldehyde reagent. Total non-acid glycolipids (20 rg) used were from human erythrocytes of blood group A (lane A), human meconium of blood group B (lane B), monkey intestine of species Macaca cynomolgus (lane C), dog small intestine (lane D), rabbit small intes- tine (lane E), and guinea pig small intestine (lane F). The numbers used to denote glycolipid bands refer to the structures listed in Table I. The oertical numbering to the left indicates the approximate num- ber of sugar residues in glycolipids with a corresponding position. Thin layer plates with resolved glycolipid mixtures were treated for bacterial binding and autoradiography (70 h), as described under “Materials and Methods.”

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

Binding of Bacteria to Glycolipids 11253

ABCDEFG A BCDE FG FIG. 2. Incidental failure of A. naeslundii (ATCC 12104) to

bind to LacCer present in thin layer chromatograms with non-acid glycolipid mixtures of various tissues. II is the auto- radlogram showing the binding of this particular batch of “‘I-labeled 11 nneslundl~. and I the total glycolipid pattern as chemically detected with anisaldehyde. Total non-acid glycolipids (20 ~8) used were from human erythrocytes of blood group A (lane A), human meconium of blood group B (lane B), monkey intestine of species Macaca cyno- molg~ (lane C), dog small intestme (lane I)), rabbit small intestine (lane E). gumea pig small intestine (lane F), and rat small intestine (lane C). The numbermg of glycolipid bands refers to the structures listed m Table I. The numbers to the left of II indicate the number of sugar residues m the glycolipids with corresponding position. The thin layer plate was treated for bacterial binding and autoradi- ographed (70 h) as described m “Materials and Methods.”

smallest sugar head group. It is also clear that the Gal@ terminus of other glycolipids (nos. 18, 21, 25, and 26), includ- ing Galp-Cer (no. l), do not react with A. uiscosus, supporting a requirement for lactose. Nor did glycolipids lacking the lactose sequence bind the bacteria (nos. 2 and 3). Some sugar additions to lactose completely abolished the binding (e.g. Galnl-4 for nos. 8 and 12), whereas others retained activity (e.g. Galal- of no. 7, GlcNAcbl-3 of no. 10, and GalNAcfll- 4 of no. 9). This indicates that A. viscosus also accommodates lactose that is placed internally in the chain. However, only GalNAcpl-4, as present in nos. 9 and 13, allowed binding at a level of free 1acCer. Also, the size of sugar additions at the Gal of lactose critically affected the accessibility of the binding site (Table I). This was evident when comparing compound no. 9 with 19, no. 7 with 11, and no. 10 with 35. Another finding from Fig. 1 is that binding to LacCer is clearly de- pendent on the ceramide structure. The bacteria bound to LacCer with a trihydroxy base and/or hydroxy fatty acid (no. 4, lane D) but did not bind to the faster migrating LacCer species of human erythrocytes with a dihydroxy base and nonhydroxy fatty acid (no. 5, lane A).

Studies with selected glycolipids absorbed in microtiter wells (Fig. 3) confirmed both the preferential binding of A. uiscosus to certain species of LacCer (nos. 4 and 5) and the equal activity of LacCer (no. 4) and its GalNAcp-substituted counterparts (nos. 9 and 13). In this system, the shift from a non-adhesive to an adhesive surface occurred between 20 and 500 rig/well (and with a half-maximal binding at 200 rig/well).

As shown in Fig. 1, A. naeslundii also bound to LacCer (no. 4) and compound no. 7 but not to LacCer of human erythro- cytes (no. 5). Additional reactive bands (e.g. nos. 12 and 17), which did not bind A. uiscosus, were attributable to the second and superimposed recognition of GalNAcP (see below). The results nevertheless indicated that A. naeslundii had a lactose- based specificity very similar, if not identical, to that of A. viscosus.

Apparent Additional Specificity of A. naeslundii for GalNAcP

in Glycolipids-Besides the lactose-based recognition (i.e. nos. 4 and 7) A. naeslundii, unlike A. uiscosus, recognized more slow-migrating bands (~4 sugars) corresponding to GalNAcfl- containing glycolipids (Table I). For instance, the reactive band in the four sugar region of all samples (lanes A-F) in Fig. 1 reflects recognition of the GalNAcb terminus of Gb4a (no. 12). The series of glycolipids recognized in the rat intes- tinal sample (Fig. 2, lane G) are Galnl-3 repetitive structures all terminating in a GalNAcb residue (no. 28). The relatively low amount of glycolipid needed for this binding was first noted when A. naeslundii stained several compounds not visible with the chemical detection reagent (marked with X in Fig. 1). When glycolipids coated in microtiter wells were used (Fig. 3), this binding was detectable at about 1 rig/well and was half-maximal at 10 rig/well. By comparison, these values are lo-loo-fold lower than those observed for the lactose-based bindings (Fig. 3). Additional evidence for two chemically distinct and unrelated binding sites was provided by the sugar inhibition experiments presented below and by the finding that a certain batch of A. naeslundii was found to express the GalNAcP based activity only (Fig. 2). This batch of A. naeslundii failed to interact with LacCer (no. 4) and Gb3b (no. 7), although the binding to the GalNAc@-containing glycolipids remained unaffected (see nos. 12 and 17).

The binding of A. naeslundii to serial dilutions of Gb4a (no. 12) in thin layer chromatograms was detectable at l-10 ng, and at 500 ng it was maximal (data not shown). When this larger amount was used for the qualitative testing of the pure and structurally defined compounds shown in Table I, the most intensively stained glycolipids (marked ++) all con- tained internally or terminally located GalNAcfl. The glyco- lipid binding properties already attributed to the lactose-based specificity (compare with A uiscosus) showed much weaker staining intensities at this concentration (marked with + and (+)). So, except for these less active compounds, glycolipids lacking the GalNAcP structure did not bind the bacteria.

The specificity for GalNAc@ in glycolipids may be summa- rized as follows. The demand for a P-anomerity was first deduced from the inactivity of the Forssman hapten (no. 16) and of blood group A active glycolipids (nos. 29 and 30). These compounds (i.e. nos. 29 and 30) also remained inactive after removal of their fucosyl residues (nos. 23 and 24). Lack of binding to substances with a terminal GalP residue (i.e. nos. 1, 5, 14, 15, and 35) implied a substantial specificity for the acetamido group at C-2. The importance of the stereochem- istry at C-4 was indicated from the results with GlcNAc@ (nos. 10 and 15), which only showed a weak activity (+), probably attributable to the lactose-based specificity (com- pare with A. viscosus). In addition, the hydroxyls at C-3 and C-6 of GalNAcfl may be substituted with Galfi (nos. 13 and 18) and NeuAccu (no. 26), respectively, without abolishing its binding activity.

According to the data in Table I, it appears that the GalNAcP structure contains the whole recognition site. There is no difference in activity between compounds nos. 9 and 13 (Fig. 3), indicating that the terminal GalP1-3 substituent of substances nos. 13 and 18 does not contribute to the interac- tion. Similarly, the equivalent activity of compounds termi- nating in a GalNAc@l-4Gal sequence (no. 9), as compared with those with a GalNAcpl-3Gal terminus (nos. 11, 12, 17, and 28), suggests that the interaction is also independent of the penultimate Gal.

Even though it was substituted with both Galbl-3 and NeuAccu2-6, the epitope on GalNAcP was apparently available to the bacteria (no. 26). However, further extensions of the Galpl-3GalNAcfl terminus with either sialic acid (e.g. nos. 32

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

11254 Binding of Bacteria to Glycolipids

TABLE I

No.

Results from binding of A. naeslundii (ATCC 12104) and A. uiscosus (ATCC 19246) to pure and struct&ly defined glycolipids on a thin-layer chromatogram

Binding’

Structur& A. uis- A. noes- Symbol Source (Ref.)d

EOSlLs lundii 1. Gsl/31-&r(h) 2. Glc#?l-&r(h) 3. Galal-4Gal@l-&r(h) 4. @l@l-QGlc@l-‘&r(h) 5. G&?-4Glc@l-Cer 6. NeuAca2-3Galfll-4Glc@l-Cer 7. Galnl-3Galal-4Glc~l-Cer(h) 8. Galal-4Ga181-4Glc~l-Cer 9. GalNAc~1-4~al~~-~Glc@l-Cer

10. GlcNA~l-3&1l,T1-4Glc~l-Cer 11. GalNAc~l-3Gala1-3Gal~i-~Glc~l-Cer 12. GalNAc~1-3Galal-4Gal~l-4Glc/3l-Cer 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.

- -

+ - - (+)

+ (+)

-

CL (+)

-

- -

- -

-

- -

+

(+) -

++ (+) ++ ++

t+f (+)

++ ++

++ ++

++

++ - -

-

Most tissues (52) Most tissues (52)

LacCer LacCer

Gb3b Gb3a Gg03

Gbla

Gg04

Human mec&i& (53) Dog small intestine (54) Human erythrocyte (52) Human erythrocyte (52) Dog small intestine (54) Human erythrocytes (52) Human brain (56)’ Malignant melanoma’ Rat colon carcinoma (55) Human erythrocyte (52) Mouse smsll intestine (58) Human meconium (53) Dog small intestine (59)g Dog small intestine (54) Human erythrocyte (60) Vera cells (61) Mouse large intestine (58) Dog small intestine (59)# Dog small intestine (59Y Human erythrocyte (52) Dog smell intestine (59P Human meconium (53)” Human brain (56) Frog brain (28)&l Rat small intestine (57)’ Rat small intestine (57)’ Dog small intestine (59)s Human meconium (53) Human brain (56) Frog brain (28)8 Human meconium (62) Human kidnev (63) Various sources (61 54)*,’ Various sources (59: 64)” Human brain (56)

‘The receptor-active GalNAcPand Galpl-4Glc sequences have been underlined with a solid and dotted bar, respectively.

’ Cer indicates ceramide containing nonhydroxy fatty acid and dihydroxy long-chain base; Cerh indicates the presence of hydroxy fatty acids and/or trihydroxy base; Cer(h) indicates that both types of ceramide were present in the fraction tested.

’ ++ marks optimal binding when 500 ng of glycolipid was used, + marks optimal binding only at 2 rg, (+) marks weak binding at 2 fig, - marks no binding at all at 2 fig. The ++ binders repeatedly showed highly intensive autoradiographic staining even at 500 ng, as well as residual staining below 50 ng. The + binders repeatedly showed highly intensive staining at 2 rg and could not be diluted below 50 ng without loss of staining. The (+) binders usually showed, but not always (no 7 in Fig l), a weak staining at 2 pg and could not be diluted below 500 ng with residual staining of the autoradiogram.

’ A main source and a reference is given for each compound. In general, the same compound of several sources was tested.

e Prepared by acidic hydrolysis of the predominant human brain ganglioside (no. 25) as described in Ref. 25. ‘K.-A. Karlsson, unpublished data. B N. Stramberg and K.-A. Karlsson, unpublished data. h Nos. 23 and 24 were obtained by acidic hydrolysis of compounds nos. 29 and 30, respectively, and as described

under “Materials and Methods.” ’ Prepared by neuraminidase treatment of substance no. 31 as described under “Materials and Methods.” ’ n = 1-3. ‘The formulae summarize various combinations (brackets combined with no brackets) of blood group ABH

determinants based on type 2 (no. 35) and type 1 (no. 36). All theoretical combinations have been tested except those with terminal Gal but lacking Fuc bound to the penultimate Gal.

and 37) or blood group ABO determinants (e.g. nos. 19, 33, and 34) blocked the binding. The Forssman hapten (GalNAcal-GalNAcp) was also inactive (no. 16). On the other hand, sialic acid residues linked to the internal Gal of sub- stance no. 13 did not introduce steric hindrances, as judged by the strong activity of certain brain gangliosides (e.g. nos. 25 and 31). Finally, when tested against a mixture of acidic glycolipids prepared from human erythrocytes, A. nueslundii bound several bands within the 5-12 sugar interval (data not shown).

Sugar Inhibition Studies-By preincubation of the two actinomycetes with free soluble sugars, we investigated the ability of saccharides corresponding to the identified lactose and GalNAcP recognition sites to inhibit the glycolipid bind- ings of Fig. 1 (Table II).

As earlier observed for other lactose-binding bacteria (23- 26), both lactose (up to 10 mg/ml) and multivalent lactose- BSA (5 mg/ml, -25 mol/mol) failed to interfere with bacterial binding to LacCer. On the other hand, GalNAc@l-SGalcuO- ethyl at the concentration of 1 mg/ml, as well as free lactose,

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

Binding of Bacteria to Glycolipids

A. viscosus

m s- 4,A

o

2

ff 4- 13.0

10 1000

rig/we 11

A. naeslundii

s 4-

; 0. 0 2-

7 ‘rrml .I 10 1000

rig/we 11

FIG. 3. Binding of 12?-labeled A. viscosus (ATCC 19246) and A. naeslundii (ATCC 12104) to glycolipids coated in microtiter wells. The glycolipids (numbers referring to Table I is given to the right of each curve) were added to microtiter wells in the amount indicated by the bottom scale (ng of glycolipid/50 ~1 metha- nol). The wells were then treated for bacterial binding as described under “Materials and Methods.” Data are expressed as mean values of triplicate determinations.

completely inhibited the binding of A. naeslundii to the GalNAc&based glycolipids of Fig. 1. Lactose was, however, at least 10 times less potent, requiring 10 mg/ml.

As expected from the finding that A. naeslundii, but not A. uiscosus, attached to thin layer chromatograms of human erythrocyte glycolipids (Fig. 1, lane A and above), only A. naeslundii agglutinated these cells (Table II). GalNAc@l- 3Gala-O-ethyl inhibited this hemagglutination down to 50 fig/ ml; free lactose was lo-fold less potent, while both Gal and

GalNAc were even less efficient (two to three times), and lactose-BSA was negative when 2 mg/ml was used.

DISCUSSION

This study identifies and characterizes a low affinity lactose and a high affinity GalNAc/3 binding property of Actinomyces, both of which may contribute to the intra-oral distribution and establishment of this genus. This was accomplished by binding of radiolabeled A. naeslundii (strain ATCC 12104) and A. uiscosus (strain ATCC 19246) to a panel of naturally occurring glycolipids resolved on thin layer chromatograms (Fig. 1) or coated in microtiter wells (Fig. 3). Our study also provides evidence that glycolipids with the GalNAcP sequence serve as human erythrocyte and epithelial cell surface recep- tors for Actinomyces. Interestingly, lactose of LacCer is also recognized by several other commensal and pathogenic bac- teria in man (22-26). Such a broadly distributed binding to a common cell-surface carbohydrate structure has earlier been demonstrated only for the many bacteria classified as man- nose binders (43).

Among the bacteria hitherto classified as lactose binders, we have noted both variations in detailed specificities (24,26) and a typical mode of binding (22, 23) recently characterized in detail for P. granulosum (26). The binding characteristics for A. uiscosus, however, appeared very similar to those for P. granulosum and are therefore only briefly discussed here. A. uiscosus bound to both terminally and internally located lac- tose (underlined in Table I), although most of the lactose- containing glycolipids were negative, as explained by steric hindrances from proximal sugars. There was a clear depend- ence on the ceramide structure with regard to the binding to free LacCer as only those species with P-hydroxy fatty acid and/or a trihydroxy base (the epithelial type of ceramide (44)) was active (compare nos. 4 and 5 in Fig. 1 and Table I). This has previously been explained by influences from the ceramide structure on the conformation of the lactose head group (24, 26). The relatively large amount of LacCer needed for binding (half-maximal binding at -200 rig/well in Fig. 3), together with the inability of free lactose to work inhibitorally (Table II), is explained by a low affinity at individual binding sites (23-26). A similar situation was recently found for the Shiga toxin and its recognition of GalLul-4Gal (45).

A. naeslundii exhibited two superimposed binding proper- ties of which one was very similar, if not identical, to that of A. viscosus. This was initially indicated by the binding of both species to LacCer and Gb3b (nos. 4, 5, and 7 in Fig. 1 and Table I). In support of this, preincubation of A. naeslundii

TABLE II

Effect of free soluble saccharides on the glycolipid binding properties of A. viscosus and A. naeslundii and on A. naeslundii-induced hemagglutination

Bacteria preincubated with listed sugars were tested for binding to the thin layer chromatogram of Fig. 1 or for hemagglutination. For experimental details see “Materials and Methods.”

Preincubation

PBS Lactose GalNAcbl-3Gala-O-ethyl GalNac Gal Lactose-BSA

Binding to’

La&x (no. 4) Gb4a (no. 12)

A. uiscosus A. naeslundii A. oiscosw A. naeslundii

+ + - + + (10) + (10) - - (10) + (1) + (1) - - (1)

+ (5)

A. naeslundii-induced agglutination of

human erythrocytes”

+++ - (0.5) - (0.05) - (1.5) - (1.5) +++ (2.0)

‘Positive binding, +, means binding with the same staining intensity as in Fig. 1 and a minus, -, means no binding. The numbers in parenthesis indicate either the approximate amount (mg/ml) needed for complete inhibition or that used for testing.

*The numbers in parenthesis indicate the approximate minimal inhibiting amount (mg/ml).

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

Binding of Bacteria to Glycolipids

with GalNAcfil-3Galcul-O-ethyl prevented the binding spec- ified by GalNAcP but not the binding to LacCer (Table II). Also, one unique batch of A. naeslundii failed to bind LacCer (nos. 4 and 7), although its binding to the GalNAcp-containing glycolipids was unaffected (Fig. 2). One explanation for this is that the lactose-binding component of the bacteria was selectively turned off.

A. naeslundii also recognized GalNAcP when substituted with GalP1-3 or NeuAcol2-6 (Table I). However, extensions of the Gal@l-3GalNAc terminus with either blood group ABO- determinants or sialic acid apparently introduced steric hin- drances, as judged by their blocking effect on binding. Rec- ognition by microbial ligands of internally located sequences has been demonstrated earlier for uropathogenic E. coli (20), the Shiga toxin (45), and other lactose-binding bacteria (24- 26). The very similar activity of some glycolipids with differ- ent neighboring groups to GalNAcP substantiates our conclu- sion of a small adhesin-combining site that does not extend beyond GalNAcP (compare no. 9 with 13, no. 9 with 12). The fi-anomerity, the acetamido group, and the axial hydroxyl at C-4 were essential, and the hydroxyls at C-3 and C-6 may carry substitutions. The affinity of binding was relatively high, as indicated by the detection level at l-10 ng of glyco- lipid, amounts that are lo-100 times lower than those needed in the case of LacCer (Fig. 3) and that free sugars inhibited the binding to thin layer chromatograms (Table II) as well as hemagglutination (Table II and below).

In view of the sugar inhibition with free lactose (Table II), the proposed binding to GalNAcP in glycolipids is attributable to the lactose-sensitive lectin activity reported earlier for Actinomyces (see Introduction). Although GalNAcPl- 3Galcul-O-ethyl was about 10 times more potent than lactose, both saccharides selectively inhibited the binding of A. naes- lundii to GalNAcp-containing glycolipids on thin layer plates. Consistent with this is that A. naeslundii, unlike A. viscosus, efficiently agglutinated human erythrocytes (Table II) and that the inhibitory effect of GalNAcpl-3Galcu-O-ethyl was again 10 times more potent than that of lactose, which in turn was two to three times better than Gal and GalNAc. These data, together with the noted specificity for glycolipids (Table I), suggest that the lactose sensitivity of Actinomyces is spec- ified by GalNAcP in internal or terminal positions rather than by terminal Gal/3 residues as earlier claimed (see Introduc- tion). McIntire et al. (15) recently proposed a lectin of A. naeslundii (ATCC 12104) with specificity for Gal@l- 3GalNAc, on the basis of the 50-fold lower efficiency of Galpl- 4GalNAc in inhibiting the aggregation of this strain with Streptococcus sanguis S34. Alternatively, the GalP1-4 substit- uent could induce a steric hindrance, since our data suggest a critical role of the axial hydroxyl at C-4 of GalNAc& Also, the extension of terminal GalNAcp with Galfil-3 did not, in our case, improve the binding (compare nos. 9 and 13 in Fig. 3). On the other hand, perhaps the ATCC 12104 strain used by these workers differs in detailed sugar specificity, thereby also explaining why Brennan et al. (17) were unable to show binding of strain ATCC 12104 to Gg03 (no. 9). Whether the GalNAcp sequence is the interaction site on S. sanguis S34 remains to be determined, although it should be noted that the site has been traced to a hexasaccharide with 2 GalNAc residues (46). In addition, the ability of this lectin to accom- modate free lactose and Gal is of interest in principle, since A. naeslundii apparently failed to associate with certain gly- colipids that have these sugars in the terminal position (nos. 1, 5, and 21), as well as with lactose-BSA (Table II). The presentation and activity of terminal sugars in glycolipids and glycoproteins may, therefore, be largely affected by their

internal structures. This observation, together with the earlier hypothesis of a lectin with specificity for terminal Gal/3 resi- dues (15 and references therein), emphasizes that sugar inhi- bition studies may be misleading if used to predict membrane- bound receptor sequences.

As expected from the hemagglutination pattern seen in Table II, A. naeslundii, in contrast to A. viscosus, bound both acid and non-acid human erythrocyte glycolipids (lane A in Fig. 1, and under “Results”). The two receptor-active non- acid glycolipids are known to contain the sequence GalNAcP (nos. 12 and 17 in Table I) and one is the predominant erythrocyte glycolipid Gb4a (no. 12), known to be accessible at the surface of these cells (47). Based on these data and the sugar inhibition experiments (Table II), we propose that glycolipids with the sequence GalNAcp serve as the human erythrocyte receptors for Actinomyces. At present, we cannot exclude that protein-bound GalNAcP sequences may also be receptor sites (48). However, it should be noted that GalNAcP in glycoproteins has so far been found only as part of the Sid blood group antigen (49). Even though the hemagglutination induced by A. naeslundii was a strong and immediate event (Table II), most other strains of A. viscosus and A. naeslundii require prior priming of human erythrocytes with neuramin- idase for lactose-sensitive hemagglutination (11, 12). This may mean that there is another lectin which recognizes a slightly different epitope masked by sialic acids in the native erythrocytes. Alternatively, if the amounts of lectins on these strains are lower than for A. naeslundii (ATCC 12104), the neuraminidase treatment may expose additional GalNAcP sites necessary for a clear hemagglutination.

A scraping of human buccal epithelium, a target tissue for adhesion of A. viscosus and A. naeslundii (5), was preliminary analyzed for the presence of receptor-active glycolipids (Fig. 4). A. viscosus (ATCC 19246) bound one of the major glyco- lipids in this material. In view of the strict specificity of this strain for lactose (Table I), this glycolipid probably contains an active lactose site. Since it is also a lactose binder (Table I), A. naeslundii (ATCC 12104) stained the same glycolipid but apparently much more strongly. This glycolipid therefore

C

1 1 1 FIG. 4. Binding of lz51-labeled A. uiscosus (ATCC 19246)

and A. naeslundii (ATCC 12104) to non-acid glycolipids of a scraping of human buccal epithelium. A shows the total glycolipid pattern as visualized with the anisaldehyde reagent, and B and C show the autoradiograms (70 h) after binding of A. uiscosus and A. meslundii, respectively. Lane 1 contains total non-acid glycolipids (20 ~g) prepared from a scraping of buccal epithelium. The thin layer plate with resolved glycolipids was treated for bacterial binding and autoradiography as described under “Materials and Methods.”

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

Binding of Bacteria to Glycolipids 11257

probably also contains an active GalNAcP site, and, on this basis and in view of its mobility, may possibly be Gg04 (no. 13). These results possibly imply that, in the oral epithelium, there are two populations of receptor-sites for Actinomyces, a low affinity lactose site and a high affinity GalNAcP site. Adhesion of A. naeslundii ATCC 12104 to buccal cells is, at least partially, inhibitable with lactose (13) and appears to be critically dependent on type 2 fimbriae (14), which carry the lactose-sensitive lectin. Even though this strongly implies a GalNAcP sequence of the yet unidentified glycolipid as the main epithelial receptor site for Actinomyces, the lactose site may also be needed for maintenance or stabilization of in oiuo adhesion. As far as A. naeslundii ATCC 12104 is concerned, this strain only expresses type 2 fimbriae (50), indicating a nonfimbrial appearance of the lactose-binding component. A pilus-independent LacCer binding was recently identified for N. gonorrhoeue (25).

The data reported here further emphasize the usefulness of solid-phase glycolipids for identification and characterization of carbohydrate receptor sites for bacteria. We suggest that A. naeslundii (ATCC 12104) binds with high affinity to gly- colipids with the sequence GalNAcP on human erythrocytes and oral epithelial cells. The additional binding of Actino- myces (both ATCC 12104 and ATCC 19246) to lactose in the oral epithelium glycolipid may indicate functional low affinity receptor sites. The proposal (6) that type-2 fimbriae with the GalNAcp specificity are also important with regard to the predominance of Actinomyces in the dental plaque is of par- ticular interest in view of the fact that this bacterial genus may be a key intermediate for the colonization of “parodon- topathogenic” Bacteroides (51). On this basis, one may con- sider the possible use of synthetic receptor analogues to prevent such a series of events.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

REFERENCES

Gibbons, R. J., and van Houte, I. (1975) Annu. Reu. Microbial. 29,19-44

Jordan, H. V., and Hammond, B. F. (1972) Arch. Oral Biol. 17, 1133-1342

Loescbe, W. J., and Syed, S. A. (1978) Infect. Immun. 21, 830- 839

Williams, B. L., Pantelone, R. M., and Sherris, J. C. (1976) J. Periodontal Res. 11, l-18

Ellen, R. P. (1982) in Host-Parasite Interactions in Periodontal Disease (Genco, R. J., and Mergenhagen, S. E., eds) pp. 98- 111, American Society for Microbiology, Washington, D. C.

Cisar, J. 0. (1986) in Microbial Lectins and Agglutinins (Mirel- man, D., ed.) pp. 183-196, John Wiley & Sons, New York

Cisar, J. 0.. Sandberg, A. L.. and Meraenhaaen. S. E. (1984) J. Dent. Res. 63, 3931396

I I,

Gibbons, R. J., and Nygard, W. (1970) Arch. OralBiol. 15,1397- 1400

44.

45. M&tire, F. C., Vatter, A. E., Baros, J., and Arnold, J. (1978)

Infect. Immun. 2 1, 978-988 Cisar, J. 0. (1982) in Host-Parasite Interactions in Periodontal 46.

Disease (Genco, R. J., and Mergenhagen, S. E., eds), pp. 121- 131, American Societv for Microbioloav. Washinaton. D. C.

Costello, A. H., Cisar, J: O., Kolenbran& P. E., and Gabriel, 0. (1979) Infect. Zmmun. 26, 563-572

Ellen, R. P., Fillery, E. D., Chan, K. H., and Grove, D. A. (1980) Infect. Immun. 27, 335-343

Saunders, J. M., and Miller, C. H. (1980) Infect. Zmmun. 29, 981-989

Brennan, M. J., Cisar. J. 0.. Vatter. A. E.. and Sandbera. A. L. (1984) Infect. Immun. 46,‘459-464

II

McIntire. F. C.. Crosbv. L. K.. Barlow. J. J.. and Matta. K. L. ( 1983) Infect. ‘Zmmun. 4 1,848-850 '

Heeb, M. J., Marini, A. M., and Gabriel, 0. (1985) Infect. Zmmun. 47,61-67

Brennan, M. J., Joralmon, R. A., Cisar, J. O., and Sandberg, A. L. (1987) Infect. Zmmun. 55, 487-489

Hansson, G. C., Karlsson, K.-A., Larsson, G., Stromberg, N., and

19.

20.

Thurin, J. (1985) Anal. Biochem. 146, 158-163 Karlsson, K.-A., and Stromberg, N. (1986) Methods Enzymol.

138,220-232 Bock, K., Breimer, M. E., Brignole, A., Hansson, G. C., Karlsson,

K.-A., Larson, G., Leffler, H., Samuelsson, B. E., Stromberg, N., Svanborg Eden, C. S., and Thurin, J. (1985) J. Biol. Chem. 260,8545-8551

21.

22.

Lund, B., Marklund, B.-I., Strijmberg, N., Lindberg, F., Karlsson, K.-A., and Normark, S. (1988) Mol. Microbial. 2, 255-263

Hansson, G. C., Karlsson, K.-A., Larson, G., Lindberg, A., Strom- berg, N., and Thurin, J. (1983) in Gycoconjugates (Chester, A., Heineaard, D.. Lundblad, A., and Svensson. S.. eds) pp. 631- 632, Rahm Publications, Lund

23. Holeersson. J.. Karlsson. K.-A.. Karlsson. P.. Norrbv. E.. Orvell. Cz and Stromberg, N.‘(1985) in World’s Debt to &stew (Ko: prowski, H., and Plotkin, S., eds) pp. 273-301, Alan R. Liss, Inc., New York

24.

25.

Stromberg, N., Ryd, M., Lindberg, A. A., and Karlsson, K.-A. (1988) FEBS Lett. 232, 193-198

Stromberg, N., Deal, C., Nyberg, G., Normark, S., So, M., and Karlsson, K.-A. (1988) Proc. N&l. Acad. Sci. U. S. A. 85,4902- 4906

26. 27. 28.

29.

30.

31. 32.

33. 34. 35. 36.

Stromberg, N., and Karlsson, K.-A. (1990) submitted Karlsson, K.-A. (1986) Methods Enzymol. 138, 212-220 Ohashi, M. (1982) in Advances in Experimental Medicine and

Bio1og.y (Makita, A., Handa, S., Taketomi, T., and Nagai, Y., eds) Vol. 152, pp. 83-91, Plenum Publishing Co., New York

Paulson. J. C.. Weinstein. J., Dorland. L.. van Halbeek. H., and Vliegenthart, J. F. G. (1982) J. Biol’Chem. 257, 12734-12738

Stahl, E. (1962) Dunnschichtxhromatographie Springer-Verlag, Berlin

Svennerholm, L. (1957) Biochim. Biophys. Acta 24,604-611 Fewster, M. E., Burns, B. J., and Mead, J. F. (1969) J. Chroma-

togr. 43, 120-126

37.

38.

39.

40.

41.

42.

43.

Hakomori, S. (1964) J. Biochem. (Tokyo) 55,205-208 Ciucanu, I., and Kerek, F. (1984) Carbohydr. Res. 31,209-217 Karlsson, K.-A. (1974) Biochemistry 13, 3643-3647 Breimer, M. E., Hansson, G. C., Karlsson, K.-A., Larson, G.,

Leffler, H.. Pascher. I.. Pimlott. W.. and Samuelsson, B. E. (1980) Ado: Mass Spectr. 8, 1097-1168

Falk, K.-E., Karlsson, K.-A., and Samuelsson, B. E. (1979) Arch. Biochem. Biophys. 192, 164-176

Falk, K.-E., Karlsson, K.-A., and Samuelsson, B. E. (1979) Arch. Biochem. Biophys. 192, 177-190

Falk, K.-E., Karlsson, K.-A., and Samuelsson, B. E. (1979) Arch. Biochem. Biophys. 192,191-202

Yang, H. J., and Hakomori, S.-i. (1971) J. Biol. Chem. 246,1192- 1200

Stellner, K., Saito, H., and Hakomori, S.-i. (1973) Arch. Biochem. Biophys. 155,464-472

Bolton, A. E., and Hunter, W. M. (1973) Biochem. J. 133, 529- 539

Sharon, N., and Ofek, I. (1986) in Microbial Lectins and Agglutin- ins (Mirelman, D., ed.), pp. 55-81, John Wiley & Sons, New York

Karlsson, K.-A. (1982) in Biological Membranes, (Chapman, D., ed) Vol. 4, pp. l-74, Academic Press, London

Lindberg, A. A., Brown, J. E., Stromberg, N., Westling-Ryd, M., Schultz, J. E., and Karlsson, K.-A. (1987) J. Biol. Chem. 262, 1779-1785

47.

48.

49.

McIntire, F. C. (1987) in Molecular Basis for Oral Microbial Adhesion (Mergenhagen, S. E., and Rosan, B., eds) pp. 153- 158, American Society for Microbiology, Washington, D. C.

Lampio, A., Finne, J., Homer, D., and Gahmberg, G. G. (1984) Eur. J. Biochem. 145, 77-82

Brennan, M. J., Cisar, J. O., and Sandberg, A. L. (1986) Infect. Zmmun. 52,840-845

Donald, A. S. R., Soh, C. P. C., Yates, A. D., Feeney, J., Morgan, W. T. J., and Watkins, W. M. (1987) Biochem. Sot. Trans. 15, 606-608

50.

51. 52.

Cisar, J. 0.. David, V. A., Curl, S., and Vatter. A. E. (1984) Infect.

53.

Immun. 46.453-458 .,

Slots. J.. and Gibbons. R. J. (1978) Infect. Zmmun. 19. 254-264 Hakdmori, S-i. (1983) in Sphingolipid Biochemistry (Kanfer, J.

N., and Hakomori, S.-i., eds) Vol. 3, pp. 1-165, Plenum Pub- lishing Co., New York

Karlsson, K.-A., and Larson, G. (1981) J. Biol. Chem. 256,3512- 3524

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

11258 Binding of Bacteria to Glycolipids

54.

55.

56.

57.

58.

Hansson, G. C., Karlsson, K.-A., Larson, G., McKibbin, J. M., Stromberg, N., and Thurin, J. (1983) Biochim. Biophys. Acta 750,214-216

Falk, P., Holgersson, J., Jovall, P.-A., Karlsson, K.-A., Stromberg, N., Thurin, J., Brodin, T., and Sjogren, H.-O. (1986) Biochim. Biophys. Acta 878, 296-299

Ledeen, R. W., and Yu, R. K. (1978) Res. Methods Neurochem. 4, 371-410

Breimer, M. E., Hansson, G. C., Karlsson, K.-A., and Leffler, H. (1982) J. Biol. Chem. 257, 557-568

Hansson, G. C., Karlsson, K.-A., Leffler, H., and Stromberg, N. (1982)FEBSLett. 139,291-294

59. McKibbin, J. M., Spencer, W. A., Smith, E. L., Mansson, J.-E., Karlsson, K.-A., Samuelsson, B. E., Li, Y.-T., and Li, S.-C. (1982) J. Biol. Chem. 257, 755-760

60. Angstrom, J., Karlsson, H., Karlsson, K.-A., Larson, G., and Nilson, K. (1986) Arch. Biochem. Biophys. 251,440-449

61. Blomberg, J., Breimer, M. E., and Karlsson, K.-A. (1982) Biochim. Biovhvs. Acta 711. 466-477

62. Larson, G. (1986jArch. Biochem. Biophys. 246, 531-545 63. Breimer. M. E.. and Jowall. P.-A. (1985) FEBS Lett. 179. 165-

172’ 64. Bjork, S., Breimer, M. E., Hansson, G. C., Karlsson, K.-A., and

Leffler, H. (1987) J. Biol. Chem. 262, 6758-6765

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from

N Strömberg and K A Karlssonoverlay approach.

Actinomyces viscosus (ATCC 19246) to glycosphingolipids, using a solid-phase Characterization of the binding of Actinomyces naeslundii (ATCC 12104) and

1990, 265:11251-11258.J. Biol. Chem. 

  http://www.jbc.org/content/265/19/11251Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/265/19/11251.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on June 26, 2020http://w

ww

.jbc.org/D

ownloaded from