Expression of a divalent cation-dependent erythroblast adhesion receptor

by stromal macrophages from murine bone marrow

LYNN MORRIS*, PAUL R. CROCKERf, IAIN FRASER, MAXINE HILL and SIAMON GORDON*

Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford 0X1 3RE, UK

* Present address: Walter and Eliza Hall Institute, Melbourne, Australiat Present address: Lnstitut Pasteur, Paris, France% Author for correspondence and reprint requests

Summary

Stromal macrophages in haemopoietic organs ex-press novel surface receptors that are implicated introphic interactions with developing blood cells.Macrophages isolated from foetal liver bind eryth-roblasts (Eb) by a divalent cation-dependent recep-tor (EbR), whereas stromal macrophages in adultbone marrow and lymphoid organs express a lectin-like receptor, sialoadhesin, which interacts withsialylated structures on sheep erythrocytes andmurine haemopoietic cells. In order to learn moreabout the regulation of these haemagglutinins, weexamined binding of Eb by stromal macrophagesthat had been isolated from adult murine tissues orgenerated in Dexter-type cultures of bone marrow.Macrophages were purified from bone marrow bycollagenase digestion, adherence to a substratumand detachment of clustered haemopoietic cells, andtested for their ability to bind Eb from foetal Uver oranaemic adult spleen. Freshly isolated bone marrowmacrophages bound Eb mainly by a divalent cation-dependent activity that was not inhibited by neuram-inidase treatment of Eb or by specific anti-siaload-hesin monoclonal antibodies, although these macro-phages express sialoadhesin and Eb bear a potentialligand for this receptor. Macrophages obtained bydigestion from other adult lymphoid tissues alsobound Eb by a divalent cation-dependent activity,whereas blood monocytes and lavaged peritonealmacrophages failed to do so. Peritoneal macro-phages could be induced to express high levels ofsialoadhesin by cultivation in homologous mouseserum, but such macrophages did not acquire

sialoadhesin-independent EbR activity. By contrast,high levels of EbR (~75 % rosettes) were detected onbone marrow macrophages derived by cultivationfor 3—10 days in medium containing foetal bovine orhorse serum and supplemented with 10~6M dexa-methasone, similar to that required to obtain afunctionally active stroma in long-term bone marrowculture. EbR activity remained low (<20%) onperitoneal macrophages cultivated under the sameconditions. Our studies establish that the EbR is amajor cell adhesion receptor on macrophages inadult as well as foetal tissues, that macrophagesexpress distinct, independently regulated haemopoi-etic cell interaction receptors and that the EbR is aconsistent marker of stromal macrophages in situand in vitro.

Abbreviations: BMDMO, bone-marrow culture derivedmacrophage(s); CSF-1, colony stimulating factor, type 1; Eb,erythroblast; EbR, erythroblast receptor: FBS, foetal bovineserum; FL, foetal liver; GM-CSF, granulocyte-macrophagecolony stimulating factor; HS, horse serum; LCM, L-cellconditioned medium; mAb, monoclonal antibody; MO,macrophage(s); RBMO, resident bone marrow MO; RPMO,resident peritoneal MO; SE, sheep erythrocytes; TPM<t>,thioglycollate elicited peritoneal MO.

Key words: stromal macrophages, haemopoiesis,microenvironment, erythroblast receptor, sialoadhesin (sheeperythrocyte receptor), long-term bone marrow culture,glucocorticosteroids.

Introduction

The stromal microenvironment plays an essential role inhaemopoiesis in vivo and in vitro. Evidence for this comesfrom studies using mice genetically deficient in haemopoi-etic cell growth (McCulloch et al. 1965) and fromexperiments with long-term bone marrow cultures(Dexter, 1982), which show that stromal cells influence thegrowth and differentiation of haemopoietic cells. Macro-phages are a prominent constituent of the stroma in vivo(Hume et al. 1985) and in vitro (Allen and Dexter, 1982),together with several other cell types and extracellular

Journal of Cell Science 99, 141-147 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

matrix. Little is known about the specialised properties ofmacrophages (MO) in lymphohaemopoietic stroma com-pared with those of other MO populations.

Recent studies have shown that tissue MO expressreceptors for ligands present on sheep erythrocytes (SER,now termed sialoadhesin) and on erythroblasts (EbR), thatmediate binding, but not ingestion of attached cells(Crocker et al. 19886), and that may also be involved ininteractions of MO with other haemopoietic cells (unpub-lished). Sialoadhesin, a lectin-like receptor for sialylatedglycoconjugates, was initially described on isolated bonemarrow stromal MO (Crocker and Gordon, 1986). A

141

specific mAb raised against this receptor, SER-.4, labelsmature stellate M<J> within haemopoietic clusters in bonemarrow, and reacts with specialised M<J> subpopulations inlymphoid organs (Crocker and Gordon, 1989). The EbR(Morris et al. 1988) was first identified on foetal liver M<J>,which form clusters with developing erythroid cells. TheEbR requires divalent cations for binding, unlike siaload-hesin, and shows no specificity for sialylated ligands.Sialoadhesin expression depends on the continued pres-ence of a species-restricted plasma/serum factor that isable to induce high levels of the receptor on peritoneal M<t>(Crocker et al. 1988a).

In this study we show that the EbR is also expressed bystromal M<t> from adult animals and that non-stromal M<J>populations lack this receptor. Although the two hae-magglutinins can be co-expressed, they are distinct andregulated independently. EbR activity can be induced inlong-term steroid-supplemented bone marrow cultures inwhich M<t> proliferate and differentiate as part of acompetent stroma, whereas M<I> derived from the perito-neal cavity do not acquire high levels of EbR under similarconditions. The EbR is the major haemopoietic celladhesion receptor expressed by stromal M<J>, and dis-tinguishes these populations from non-stromal M3> popu-lations.

Materials and methods

AnimalsC57B1/6 and Swiss PO (Pathology, Oxford) mice between 8 and 12weeks of age were used. Foetal mice were obtained from a POmating colony.

Media and reagentsRPMI-1640 was purchased from Gibco Biocult Ltd, Paisley,Scotland. The defined serum-free medium HB102 was obtainedfrom New England Nuclear, Boston MA. Media were sup-plemented with 2mM glutamine and 20/igml"1 gentamycin.20 mM Hepes buffer (Gibco Biocult Ltd) was routinely added toRPMI stocks. Fetal bovine serum (FBS) and horse serum (HS)were purchased from Imperial Laboratories, Salisbury, Wiltshire,UK, and heat inactivated at 66°C for 30min. Phosphate-bufferedsahne without Ca2+ and Mg2"1" (PBS) was obtained from Oxoid Ltd(Basingstoke, Hampshire, UK), EDTA, dexamethasone andphenylhydrazine from Sigma Chemical Company Ltd (Poole,Dorset, UK), and neuraminidase (Vibrio cholerae) from Calbio-chem-Behring Corp., La Jolla, CA. Sheep erythrocytes werepurchased from Gibco Biocult Ltd.

AntibodiesThe following rat monoclonal antibodies (mAb) were used atsaturation: F4/80 (Austyn and Gordon, 1981), which is specific formature mouse MO, and SER-4 (Crocker and Gordon, 1989), whichrecognises sialoadhesin on stromal MO in munne lymphohaemo-poietic organs. Fab fragments of FPLC-purified 3D6, a secondmAb to sialoadhesin were prepared by P.R.C. (unpublished).

L cell conditioned medium (LCM)Conditioned medium from the L929 fibroblast cell line grown in10 % FBS was filtered and stored at -20°C before use as a sourceof CSF-1 (colony-stimulating factor, type 1). An L cell fibroblastline transfected to express high levels of GM (granulocyte-macrophage)-CSF was obtained from Dr J. Karri, Laboratory ofMolecular Biology, Cambridge, UK.

Isolation of Af<J> populations and monocytesFoetal liver MO and resident bone marrow MO (RBMMO) wereprepared as described previously (Morris et al. 1988; Crocker and

Gordon, 1985). Briefly, day 14 foetal livers and adult femoral bonemarrow plugs were digested with collagenase and the MO wereselectively adhered to glass coverslips for 3-6 h. Attachedhemopoietic cells were removed by flushing coverslips in PBS.Resident peritoneal MO (RPMO) were collected from normal miceafter lavage of the peritoneal cavity with PBS. Thioglycollate-elicited peritoneal macrophages (TPMO) and Biogel-elicitedperitoneal MO were obtained 3-5 days after injection of 1 mlBrewer's complete thioglycollate broth or Biogel polyacrylamidebeads (Rabinowitz and Gordon, 1989). Peritoneal MO wereadhered to glass coverslips and non-adherent cells removed afterfiflmin at 37°C. Monocytes were isolated as described (Crockerand Gordon, 1985).

Bone marrow culturesBone marrow cells were flushed from femora with PBS using asyringe and a 19G needle. Plugs were loosely dissociated andwashed twice in PBS at 200 # for lOmin. Cells were plated at5xl06/coverslip in HB102 or in RPMI plus serum (FBS or HS),and dexamethasone or growth factor supplement as described.Cultures were refed when the experiment was continued for morethan 7 days. For long-term 'Dexter'-type cultures, cells wereplated in 10% FBS and 10% HS plus 10~ 6 M dexamethasone(10~2 M stock made up in 100 % Analar ethanol) and maintainedat 37 °C in a 5 % CO2 humidified incubator for up to 4 weeks.

Erythroblasts (Eb)Foetal liver erythroblasts (FLEb) were prepared from day 14foetal liver after collagenase digestion or mechanical disruption.Adult Eb were prepared by teasing apart the spleen from miceinjected i.p. with ~ l m g phenylhydrazine 7-10 days previously.Large cellular aggregates were removed by centrifugation at 50 gfor 2min. Supernatants were washed three times at 300 # for10 min in RPMI and 2x 107 cells plated onto 100 mm tissue culturePetri dishes in 10 % FBS. After 2-4 h at 37 °C, dishes were flushedgently to recover erythroid cells.

Neuraminidase treatmentNeuraminidase (0.01 Behringwerke unitml"1 RPMI) was addedto washed packed sheep erythrocytes (~25 /d) or FLEb (~108) andincubated for 60 min at 37 °C. The treatment was ended by addingFBS to a final concentration of 20 % and the cells were incubatedfor 20 min.

Rosetting assaysResetting assays were carried out as described (Crocker andGordon, 1986; Morris et al. 1988), in media with (RPMI) orwithout (PBS+0.5mM EDTA) divalent cations. Coverslips con-taining adherent MO were rinsed three times in the assaymedium before transfer to a 24-well tray. Erythroid ligands werewashed four times in PBS at 300 g for 5 min and resuspended inassay medium. A 100/d sample of diluted cells (lxlO7 toSxlO'ml"1 Eb or 2%, v/v, sheep erythrocytes) was added toadherent MO and incubated for 30 min at 37 °C. Unbound ligandwas removed by dipping the coverslips four times each in fourbeakers of RPMI or PBS and the cells were fixed in 0.25 %, v/v,glutaraldehyde. The percentage of MO that formed rosettes wasdetermined by phase-contrast microscopy or, after immunocyto-chemistry, by counting 200 MO on replicate coverslips. MO wereconsidered positive if they bound ==2Eb or 34 sheep erythrocytes.

ImmunocytochemistryCells on coverslips were processed to reveal MO antigens by use ofan avidin-biotin-peroxidase detection system as described(Crocker et al. 1990).

Results

Adult stromal macrophages bind foetal livererythroblastsWe have previously described a receptor on foetal liver

142 L. Morris et al.

Table 1. Adult stromal M<P-bound Eb

Cell population

Foetal liver M<1>RBMMOT P M *RPM<J>Monocytes

% M * thatbound 2 or more Eb

94±498±216±88±25±4

Cells were tested in at least three independent experiments, in which200 M<t> were counted on duplicate coverslips Values show mean±s.D.More than 90 % of Eb binding to foetal liver M<t> and RBM<t> wasabolished by removal of divalent cations. The difference in expression ofEbR activity among different MO populations could not be ascribed todifferences in the isolation procedure (not shown).

M<t> that is implicated in the divalent cation-dependentbinding of Eb (Morris et al. 1988). In order to establishwhether resident stromal bone marrow macrophages(RBMO) from adult mice express this receptor, haemopoi-etic clusters were isolated from femoral marrow andadherent MO prepared by removal of attached haemopoi-etic cells (Crocker and Gordon, 1985) before incubationwith Eb. Eb obtained from foetal liver or adult spleen ofanaemic mice after phenylhydrazine treatment gaveidentical binding properties on bone marrow MO (notshown) and were used interchangeably. Table 1 showsthat all RBMMO were able to bind Eb by a divalent cation-dependent mechanism similar to that described on foetalliver MO. Bound cells consisted of a heterogeneousmixture of differentiating Eb and were not ingested, asreported for foetal liver (Morris et al. 1988). Bloodmonocytes and non-stromal resident (RPMO) or thiogly-collate-elicited peritoneal MO (TPMO) displayed only lowlevels of EbR activity. The same distribution has beenreported for sialoadhesin expression by these adult MOpopulations (Crocker and Gordon, 1986). Sialoadhesin isalso known to be expressed by other tissue MO in liver,spleen and lymph node, and not by pleura! MO. Prelimi-nary experiments with MO isolated from all these adultmurine tissues showed that EbR had a similar pattern ofexpression to sialoadhesin.

EbR is the dominant adhesion receptor for erythroblastsSince adult stromal MO from bone marrow bound both Eband sheep erythrocytes it was necessary to define thenature of the receptors involved in the binding of Eb. Twopopulations of MO that express only one of the candidatereceptors were used to define the ligands present on Eb.TPMO can be induced to express high levels of sialo-adhesin activity by cultivation in mouse serum (Crocker etal. 1988a). Fig. 1 (mid panel) shows that serum-inducedTPMO bound foetal liver Eb entirely via sialoadhesin,since binding was independent of the presence of divalentcations, sensitive to treatment of Eb by neuraminidaseand fully inhibited by the SER-4 mAb. Binding of Eb toinduced TPMO was therefore indistinguishable fromsheep erythrocyte binding. By contrast, foetal liver MO(Fig. 1, lower panel) failed to bind sheep erythrocytes andbound Eb through EbR alone. Eb binding was totallydependent on divalent cations, resistant to neuraminidasetreatment of Eb, as shown previously (Morris et al. 1988),and not inhibited by 3D6, an anti-sialoadhesin mAb.Therefore Eb express ligands for both EbR (foetal liverMO) and sialoadhesin (serum-induced TPMO).

Fig. 1 (top panel) shows that RBMMO bound Ebpredominantly via EbR, but that sialoadhesin activitymight also participate in this reaction. Binding was fully

resistant to neuraminidase treatment of Eb and to anti-sialoadhesin mAb, which is evidence for EbR involvement,but was only partially prevented by removal of divalentcations. Residual binding of Eb to RBMMO in divalentcation-free medium was sensitive to neuraminidase and3D6 mAb, indicating a role for sialoadhesin under theseconditions. These results established that the EbR was thedominant receptor in adhesion of Eb when both receptorswere present, as in adult bone marrow stromal MO.

Exp. 1 Exp. 2

Untreated Cation-free N dasc Untreated anti-SER mAb

c'•3c'.E

BL.

80

60

40

20

SE

Eb

I

Exp. 1

L1

Exp. 2

1

Untreated Cation-free N'dascExp. 1

Untreated anti-SER mAbExp. 2

Untreated Cation-free N'dasc Untreated anti-SER mAb

Fig. 1. EbR and sialoadhesin expression by adult residentbone marrow M<t> (RBMMO), serum-induced TPMO and day 14foetal liver MO (FLMO). MO were assayed for binding of Eband sheep erythrocytes in RPMI ('untreated') and cation-freemedium (PBS+0.5mM EDTA), after neuraminidase (NMase)treatment of the ligand, or in the presence of sialoadhesin-specific inhibitory mAbs, SER-4 or 3D6. Eb used with serum-induced TPMO were from foetal liver. Results from 2independent experiments in which 200 cells were counted onduplicate coverslips, were within 10% of the mean. SE, sheeperythrocytes.

EbR on stromal macrophages 143

Table 2. Induction ofEbR on stromal bone marrow MOcultured in steroid-containing media

Culture media

HB102 (serum-free)10%FBS+10%LCM10% FBS+10% LCM+dexamethasone10%FBS+10%HS10% FBS10% HS+dexamethasone10% FBS+dexamethasone10% FBS+10% HS+dexamethasone

% M<t> binding

BM culture-derived M<1>

4±123±127±333±1039±2566±1170±1674±19

Eb

TPM4>

412±511±321±910±l

NDND

16±10

BM-derived M<t> and TPM<t> were cultured for 4-8 days in the abovemedia and assayed for Eb binding activity. Dexamethasone was addedto a final concentration of 10~6 M. M<t binding two or more Eb werescored as positive. Results are pooled from 12 independent experimentsin which 200 M<J> were counted on each of duplicate covershps using aphase-contrast microscope. Values show mean±s.D. ND, not done.

Regulation of EbR in vitroThe above experiments showed that stromal MO in adulttissues were able to co-express EbR and sialoadhesin, butthe absence of EbR activity on serum-induced TPMO,which express high levels of sialoadhesin, indicated thatthe two haemagglutinins are regulated by distinctmechanisms. Haemopoiesis can be maintained in long-term bone marrow cultures in specialised, steroid-contain-ing media, and since MO are prominently involved inhaemopoietic clusters in these cultures (Allen and Dexter,1982), we asked whether stromal MO derived in vitroexpressed haemagglutinin activities. Bone marrow cul-tures were grown in media supporting the establishmentof stromal layers to varying extents and the expression ofEbR and sialoadhesin assayed at different stages. Perito-neal MO obtained from the peritoneal cavity of mice afterthioglycollate broth or Biogel bead injection were used forcomparison. In a series of experiments, EbR was inducedconsistently on bone marrow-derived MO (BMDMO),whereas sialoadhesin was present only variably. Table 2and Fig. 2A show that -75 % of BMDMO bound Eb whencultured in medium containing RPMI 1640+10%FBS+10% horse serum+10"6M dexamethasone, corre-sponding to complete medium for long-term stromalcultures. Dexamethasone accounted for ~50% of theactivity when added to media containing either FBS or HSalone (Table 2, Fig. 2B). There was no EbR activity incultures grown in a defined serum-free medium (HB102)and the effects of adding a dexamethasone supplement tothis medium could not be determined, since MO did notthen proliferate in the absence of serum components. Adose-response experiment of dexamethasone supplementversus EbR induction is shown in Fig. 3. Maximal levels ofdivalent cation-dependent binding of Eb to bone marrowculture-derived MO were observed at 10~6 M, and signifi-cant induction of EbR over background was detected at10~ M. Morphologically, the cultures in complete medium(i.e. containing FBS, HS and dexamethasone) varied incomplexity and haemopoietic activity. MO were alwaysstrongly EbR+ whether other stromal elements such asfibroblasts and adipocytes were present or not. Addition ofLCM, a source of CSF-1, as a further proliferative stimulusfor BMDMO in culture, did not induce EbR and counter-acted the effect of dexamethasone supplement (Table 2)and similar results were obtained with GM-CSF (notshown). Only MO, defined by phase-contrast microscopy

(Fig. 2A) and F4/80 labelling (not shown), bound Eb andcontrol experiments showed that Eb binding was notinhibited by anti-sialoadhesin mAb. In marked contrastwith the results obtained with BMDMO, EbR could not beinduced on TPMO (Fig. 2C) or on Biogel-elicited perito-neal MO (Fig. 2D) maintained in fully supplementedmedia containing dexamethasone; low levels of EbRactivity (<20 %) were detected in 12 independent exper-iments with TPMO cultures (Table 2). Peritoneal adher-ent cell populations cultivated in such media were lesscomplex than those derived from bone marrow andcontained mainly non-proliferating MO and few otherstromal elements.

Fig. 4 shows the time course of appearance of EbR,dependence of its activity on divalent cations, and the levelof sheep erythrocyte rosetting activity in an experiment inwhich bone-marrow cultures in complete medium wereassayed daily from day 3, after sufficient numbers of MOhad been generated. High levels of EbR were detected fromday 3 to day 10, during which period there was continuousMO proliferation. Up to day 8, 60 % of Eb binding wasdivalent cation-dependent and this decreased to 50 % afterfurther culture. Sheep erythrocyte binding was uniformlylow throughout this period, hence cation-dependent bind-ing of Eb was unlikely to be mediated by sialoadhesin. Inother experiments Eb binding was resistant to neuramini-dase treatment and to anti-sialoadhesin mAb, confirmingbinding via EbR. Variable sheep erythrocyte binding andsialoadhesin antigen expression were observed in threeout of nine independent experiments, in all of which EbRwas induced on the majority of bone marrow culture-derived MO.

Discussion

Our new findings are: (1) the EbR initially described onfoetal liver MO is present on several fixed tissue MOpopulations in the adult, but not on circulating monocytesor peritoneal, non-stromal MO; (2) EbR is the dominant Ebadhesion receptor on adult resident bone-marrow stromalMO, which also express sialoadhesin, a sialic acid bindinglectin; (3) EbR cannot be induced on peritoneal MO byexogenous serum inducer protein(s); and (4) high levels ofEbR are present after growth and differentiation of MOfrom bone-marrow progenitors in media supplementedwith dexamethasone, whereas peritoneal MO do notacquire EbR activity under similar conditions. Expressionof the EbR therefore correlates with a specialised stromalphenotype in haemopoietic tissues and in cell culture, andis regulated independently of sialoadhesin, another hae-mopoietic cell interaction molecule.

We have added to the defining characteristics of EbR(non-phagocytic binding of erythroblasts, divalent cationrequirement, neuraminidase resistance of the ligand) thefurther observation that binding of Eb to foetal liver MO iscompletely resistant to specific mAb inhibitors of siaload-hesin. Expression of EbR is restricted to stromal MOsubpopulations isolated from lymphohaemopoietic organs,but further studies with specific mAb are needed todetermine its expression on MO in other adult and foetalmurine tissues. The ligand for EbR binding also needs tobe defined; preliminary unpublished observations indicatethat immature myeloid cells in bone marrow bind to EbR+

MO with similar characteristics. However, in the absenceof molecular characterisation of the EbR and specificinhibitors it remains to be shown whether EbR-like

144 L. Morris et al.

Fig. 2. Induction of EbR activity in vitro on bone marrow-derived, but not peritoneal MO. (A,B) Bone marrow MO were cultivatedin RPMI+10% FBS+10% HS in the absence (A) or presence (B) of 1 0 " 6 M dexamethasone for 10 days. Eb rosettes were morefrequent (29 versus 73%) in B. Fibroblast-like stromal cells (open arrowheads) did not bind Eb. Associated myeloid precursors areindicated by filled arrowheads. Addition of dexamethasone (B) reduced the yield of macrophages. (C,D) Few peritoneal MO elicitedby thioglycollate broth (C) or Biogel polyacrylamide bead injection (B) and cultivated in complete steroid-supplemented medium,bound Eb. Adherent cultures consisted of MO alone. Bar, 20 fan.

EbR on stromal macrophages 145

100

XJ

tu 80

1 60

40

5? 20

15 14 13 12 11 10 9 8 7 6 5 4Dexamethasone concentration

(-login)

Fig. 3. Dose-response of dexamethasone supplement and EbRinduction. Bone marrow M<t> were derived in RPMI 1640+10%FBS+10% HS supplemented as shown and assayed for EbRactivity after 15 days. Cultures were analysed for Eb bindingby F4/80+M<t>. At least 200 cells were counted for eachcoverslip and results show means of duplicate assays. The yieldof adherent cells decreased with the concentration ofdexamethasone (not shown).

haemopoietic cell binding is due to a single or multiplereceptors.

Our study shows that some MO populations expressedonly one haemagglutinin and that it was possible todistinguish the EbR and sialoadhesin receptor even whenboth were present on the same cell. Freshly isolated day 14foetal liver MO expressed EbR alone, whereas mouseserum-induced TPMO expressed only sialoadhesin. Ebcarry a binding site for sialoadhesin as shown withinduced TPMO, which bound Eb in the absence of divalentcations, via a reaction completely inhibited by anti-sialoadhesin antibody. When both haemagglutinin recep-

100 r

Ul<£ 80o

XIIX)

X

eD2CQ

60

40

20 f--i AEb withoutcations

5 6 7 8

Days in culture

3SE10 SE

Fig. 4. Time course of EbR expression by BM-denved M<Pafter culture in complete steroid-containing medium. M<t> weregrown from bone marrow progenitors in 10% FBS+10%H S + 1 0 " 6 M dexamethasone on coverslips in 24-well trays.Rosetting assays with spleen Eb were carried out from day 3when sufficient numbers of MO had been generated. There wasconsiderable growth in these cultures and cells were refed onday 7. Results show a single experiment in which 200 MOwere counted on each of duplicate coverslips using a phase-contrast microscope. Values show mean±s.D.

tors were co-expressed, as on freshly isolated resident bonemarrow M<t>, inhibition studies established that the EbRwas the major Eb adhesion receptor, although sialoadhe-sin contributed to divalent cation-independent binding ofEb. Other studies have shown that sialoadhesin antigensconcentrate at points of contact between bone marrowstromal MO and developing myeloid, but not erythroidcells (Crocker et al. 1990). Furthermore, sialoadhesin isnot present or apparently required for foetal erythropoi-esis in liver, and its induction during development iscorrelated with ontogeny of myelo- and lymphopoiesis(Morris et al. unpublished). These observations suggestthat the EbR and sialoadhesin are independently regu-lated and participate in distinct interactions involvingdifferent lymphohaemopoietic populations. Furtherstudies are needed to define the nature, ligands andfunctions of each receptor.

The EbR may provide a marker for a distinctive stromalMO phenotype in bone-marrow cultures. Eb bindingactivity was induced on the majority of MO by supple-menting culture medium with heterologous sera anddexamethasone, a synthetic glucocorticosteroid. FBS orhorse serum alone gave only partial induction of EbR andaddition of 10~ 6 M dexamethasone consistently yieldedM<J> with high levels of EbR. No attempt was made toremove endogenous steroid present in serum, and itcannot be determined from these experiments whethersteroids are essential or the only constituent in serarequired for EbR induction. Other MO markers wereeither inconsistent in expression, as shown here forsialoadhesin, or did not discriminate between stromal andnon-stromal MO, for example, F4/80 antigen. The Forss-man heterophile glycolipid antigen, which is widelydistributed on many cell types (Tanaka and Leduc, 1956),is also expressed by stromal macrophages in vivo (Bethkeet al. 1987; Sadahira et al. 1988) and in bone marrowculture (Mori et al. 1990), but its regulation and possiblerelationship to EbR remain to be studied.

Our experiments did not establish whether dexametha-sone acted directly on MO (Werb et al. 1978) or via othercells, but we observed partial inhibition of M<J> growth, asexpected from previous studies on proliferating MO(Hume and Gordon, 1984), in parallel with induction ofEbR activity. Growth and differentiation of bone marrowMO may play an essential role in EbR induction in vitro,since identical culture conditions failed to induce EbR onperitoneal MO populations elicited by two differentagents, thioglycollate broth or Biogel polyacrylamidebeads. Peritoneal MO did not proliferate in the completeculture medium with dexamethasone and the cultures didnot contain significant numbers of other stromal orhaemopoietic cells (not shown). Attempts to enhance theyield of bone marrow MO by growth factors (CSF-1, GM-CSF or IL-3, not shown) failed to increase EbR expression.Antagonistic effects of glucocorticosteroids and MOgrowth factors have been reported previously (Hume andGordon, 1984). The possibility that other mesenchymalcells or extracellular matrix present in bone marrowstromal cultures influence MO EbR expression should beborne in mind, although we noted that high levels of EbRcould be induced on MO when fibroblasts were virtuallyabsent.

The culture conditions found to have the optimal EbRinducing activity for MO were similar to those used togenerate haemopoietically active long-term bone-marrowcultures. It is known that Dexter-type cultures are rich inF4/80+ MO (Simmons et al. 1983). The EbR may play an

146 L. Morris et al.

important role in haemopoietic cellular interactions inthese cultures; for example, between stromal MO anddeveloping myeloid cells (Allen and Dexter, 1984). Furtherstudies are needed to define the effects of dexamethasoneand other potential regulators of EbR in defined culturemedia, and to account for the striking difference in EbRexpression on M<t> derived from bone marrow and theperitoneal cavity. Our findings on EbR regulation in vivoand in vitro provide a basis for further studies on theunique properties and functions of stromal M<J> inlymphohaemopoietic tissues.

This work was supported by grants from the LeukaemiaResearch Fund and the Medical Research Council, UK. IainFraser is a Rhodes Scholar.

ReferencesALLEN, T. D. AND DEXTER, T. M. (1982). Ultrastructural aspects of

erythropoietic differentiation in long-term bone marrow culture.Differentiation 21, 86-94.

ALLEN, T. D. AND DEXTER, T. M. (1984). The essential cells of thehaematopoietic microenvironment. Expl Haem. 12, 517-621.

AUSTYN, J. M. AND GORDON, S. (1981). F4/80, a monoclonal antibodydirected specifically against the mouse macrophage. Bur. J Immun11, 805-815

BBTHKE, U., KNIEP, B AND MUHLRADT, P. F. (1987). Forsamanglycolipid, an antigenic marker for a major subpopulation ofmacrophages from murine spleen and peripheral lymph nodes. J.Immun. 138, 4329-4335.

CROCKER, P. R. AND GORDON, S. (1986). Isolation and characterization ofresident stromal macrophages and hematopoietic cell clusters frommouse bone marrow. J. exp. Med. 162, 993-1014

CROCKER, P. R. AND GORDON, S. (1986). Properties and distribution of alectin-like hemagglutinin differentially expressed by murine stromaltissue macrophages. J. exp Med. 164, 1862-1865.

CROCKER, P. R. AND GORDON, S. (1989). Mouse macrophagehemagglutinin (sheep erythrocyte receptor) with specificity forsialylated glycoconjugates characterized by a monoclonal antibody. J.exp. Med. 169, 1333-1346.

CROCKER, P. R., HILL, M. AND GORDON, S. (1988a) Regulation of amurine macrophage hemagglutinin (sheep erythrocyte receptor) by aspecies-restricted serum factor. Immunology 65, 516-522.

CROCKER, P. R., MORRIS, L. AND GORDON, S. (19886). Novel cell surface

adhesion receptors involved in interactions between stromalmacrophages and haematopoietic cells. J. Cell Sci. Suppl. 9, 185-206.

CROCKER, P. R., WKRB, Z , GORDON, S. AND BAINTON, D. (1990).Ultrastructural localisation of a macrophage-restricted sialic acidbinding hemagglutinin, SER, in macrophage-hematopoietic cellclusters. Blood 76, 113-131.

DEXTER, T. M. (1982). Stromal cell associated haemopoiesis. J. cell.Physwl. Suppl. 1, 87-94.

HUME, D. A. AND GORDON, S. (1984). The correlation betweenplasminogen activator activity and thymidine incorporation in mousebone marrow-derived macrophages. Opposing actions of colony-stimulating factor, phorbol myristate acetate, dexamethasone andprostaglandin E. Expl Cell Res. 150, 347-355.

HUME, D. A., ROBINSON, A. P., MACPHBRSON, G. G. AND GORDON, S.(1985). The mononuclear phagocyte system of the mouse defined byimmunohistochemical localization of antigen F4/801 relationshipbetween macrophageB, Langerhans cells, reticular cells, and dendriticcells in lymphoid and hemopoietic organs J exp. Med 158,1522-1536.

MCCULLOCH, E. A., SIMINOVITCH, L., TILL, J. E., RUSSELL, E. S. ANDBERNSTEIN, S. E. (1965). The cellular basis of the geneticallydetermined hemopoietic defect in anaemic mice of genotype Sl/Sld.Blood 42, 611-669.

MORI, M., SADAHIRA, Y., KAWASAKI, S., HAYASHI, T. AND AWAI, M(1990). Macrophage heterogeneity in bone marrow culture in vitro. J.Cell Sci. 95, 481-485.

MORRIS, L., CROCKER, P. AND GORDON, S. (1988) Murine foetal livermacrophages bind developing erythroblasts by a divalent cation-dependent hemagglutinin. J. Cell Biol. 106, 649-656.

RABINOWITZ, S. AND GORDON, S. (1989). Differential expression ofmembrane Bialoglycoproteins in exudate and resident mouseperitoneal macrophages. J. Cell Sci. 93, 623-630.

SADAHIRA, Y., MORI, M., AWAI, M., WATAKAI, S. AND YASUDA, T. (1988).Forssman glycosphingolipid as an immunohistochemical marker formouse stromal macrophages in hematopoietic foci. Blood 72, 42-48.

SIMMONS, P. J., ALLEN, T. D., DEXTER, T M , HIESCH, S. AND GORDON, S.(1983). Development of the haematopoietic microenvironment in vitro.Expl Haem (Suppl. 2) 14, 144.

TANAKA, N. AND LEDUC, E. H. (1956). A study of the cellulardistribution of Forssman antigen in various species J. Immun. 77,198-212.

WERB, Z., FOLEY, R. AND MUNCK, A. (1978). Interaction ofglucocorticoids with macrophages. Identification of glucocorticosteroidreceptors in monocytes and macrophages. J. exp. Med. 147,1684-1694.

(Received 14 December 1990 - Accepted 24 January 1991)

EbR on stromal macrophages 147