British ]uffrnuf of Haemutofogy, 1989, 71, 323- 330

Granulocyte macrophage colony stimulating factor induced changes in cellular adhesion molecule expression and adhesion to endothelium: in-vitro and in-vivo studies in man

S. DEVEREUX,* H. A . BULL,' D. C A M P O S - ~ O S T A , ~ R. SAIB A N D D. c. LINCH Departments of Haematology, 'Dermatology and 2Nuclear Medicine, U,aiversity College and Middlesex Hospital School of Medicine, London

Received 6 June 1 9 8 8 ; accepted for pub1,ication 1 8 October 1 9 8 8

Summary. The administration of recombinant human gra- nulocyte-macrophage colony-stimulating factor (rhGM-CSF) causes a transient leucopenia. Radionuclide labelling studies showed this to be due to margination 01' neutrophils and monocytes predominantly in the pulmonary vasculature. No evidence of complement activation was found. A rapid in-viva rise in neutrophil cellular adhesion molecule (CAM) ex- pression was observed paralleling the development of the neutropenia. Neutrophils exposed to rhGM-CSF in-vitro showed similar rapid increases in CAM expression. The adherence of chromium-labelled neutrophils to endothelial cell cultures was modestly but highly significantly increased

by rhGM-CSF, an effect that was reduced by the binding of a monoclonal antibody to the beta chain of neutrophil CAM. The margination of phagocytic cells induced by rhGM-CSF administration is therefore likely to be due at least in part to increased expression of adhesion promoting glycoproteins. The demargination, however, occurred at a time when neutrophil CAM expression was still high, suggesting that dissociation of the neutrophil-endothelial cell interaction depends on factors other than downregulation of CAM expression. In-vivo modulation of phagocyte CAMS and adhesive properties by GM-CSF may be of importance in the normal inflammatory response.

Granulocyte-macrophage colony-stimulating factor is a hae- mopoietic growth factor originally recognized by its ability to stimulate the growth of granulocytic and nionocytic colonies in bone marrow culture systems (Metcalf, 1986). Growth of erythroid and megakaryocyte progenitors is also supported under the appropriate conditions (Sieff et al, 1985). In addition to these effects on primitive cells, GM-CSF also has actions on mature phagocytes. GM-CSF will inhibit neutro- phi1 migration in some in-vitro assays (Gasson et al, 1984). Neutrophil superoxide production in response to the chemo- tactic peptide N-formyl-methionyl-leucyl-phenylalanine (fmlp) is augmented (Weisbart et al, 1985) and phagocytosis increased (Lopez et al, 1986). Antigen presentation by mononuclear cells is increased (Morrissey et al. 1987) and

* Present address: Department of Haematology. Kent and Canter- bury Hospital, Canterbury, Kent.

Correspondence: Dr D. C. Linch. Department of Haernatology, University College and Middlesex Hospital School of Medicine, London WClE 6HX.

cytotoxicity of monocytes to tumour cell lines enhanced (Grabstein et al. 1986).

Recombinant human GM-CSF is now available for clinical use. In leucopenic patients with the acquired immune deficiency syndrome, administration of rhGM-CSF results in increased bone marrow cellularity and a rise in white blood cell count (Groopman et al, 1987). Encouraging results have also been reported in myelodysplasia with correction of cytopenias in some patients (Vadhan-Raj et al, 1987). rhGM- CSF also appears to ameliorate the leucopenia following chemotherapy (Antman et al. 1987) and shorten the period of neutropenia after very high dose chemotherapy and autologous bone marrow transplantation (Brandt et al, 1988: Devereux et al, 1988).

Several side effects of GM-CSF treatment have been noted. A low grade fever occurs in some patients and thrombophle- bitis is common if rhGM-CSF is administered by a peripheral vein (Devereux et al, 1988). Bone pain has been reported in some patients especially during bolus infusions (Groopman et al, 1987). At the highest doses (32-64 pg/kg/d). more severe toxicity has been observed with central venous thrombosis

3 2 3

324 S. Devereux et a1 and a capillary endothelial ‘leak’ syndrome (Brandt et al, 1988). We have recently reported that the administration of yeast derived rhGM-CSF causes a transient leucopenia with sequestration of cells in the lungs (Devereux et al. 1987). No adverse effects attributable to this were observed, but it is possible that pre-existing pulmonary inflammatory processes would be exacerbated.

A family of related adherence promoting glycoproteins, expressed in many cell types, have been characterized (Springer et al. 1987). Both neutrophils and monocytes express a heterodimer adhesion protein whose alpha chain is recognized by monoclonal antibodies of the CD11 series and beta chain by anti C D l 8 antibodies. A receptor called ICAM-I has been identified as the ligand for lymphocyte function antigen1 (LFA- 1 CD 1 l a / 18) dependent leucocyte adherence (Makgoba et al, 1988). Arnaout et a1 (1986) showed that GM- CSF increases the expression of neutrophil adhesion proteins in vitro. In this paper we show that the transient GM-CSF induced leucopenia is due to margination/demargination of neutrophils and monocytes and demonstrate that neutrophil margination is associated with rapid changes in the ex- pression of adhesion molecules. A rapid increase in the number of phagocytes adhering to the vascular endothelium could explain many of the side effects of rhGM-CSF adminis- tration.

PATIENTS

Studies were performed on two groups of patients. One group of four patients with advanced solid tumours not responding to conventional therapy received 10 d treatment with rhGM- CSF as part of a phase 1 study. The second group of six patients with resistant Hodgkin’s disease received rhGM-CSF as an adjunct to high-dose chemotherapy and autologous bone marrow transplantation. The effects of rhGM-CSF in this group were determined prior to chemotherapy in three patients and after chemotherapy but before leucopenia developed in the other three. Local ethical committee appro- val and informed written consent from each patient were obtained.

MATERIALS A N D METHODS

Two types of rhGM-CSF were used in these studies. Yeast derived rhGM-CSF was provided by Immunex Corporation Seattle. Wash., U.S.A., and E. coli derived material by Behringwerke, Marburg, West Germany. Two patients in the solid tumour group received escalating doses from 15 to 400 pg/mz/d by twice daily 1 h intravenous infusion for 10 d. The remaining patients continuous infusions of 100-400 pg/m2/ d for 10-21 d. Endotoxin levels in these materials were less than 0.4 pg/mg as determined by the limulus amoebocyte assay.

Leucocyte kinetic studies. 50 ml of venous blood was taken into acid citrate dextrose solution (ACD), sedimented with 1% Hetastarch and labelled with hydroxymethypropylenearni- neoxime (HMPAO) yymTc as described by Peters et a1 (1986). The labelled cells were reinfused and dynamic quantitative

imaging performed using a gamma camera linked to a computer system. After an equilibration period of 30 min, 15 g/mZ of rhGM-CSF was infused over 60 min. Scanning was performed during this time and for 90-1 50 min afterwards. Values for radioactivity over the heart, lungs and spleen were computed and expressed as a percentage of the initial activity over the heart.

Serial blood counts were performed and aliquots of each separated on Ficol-Hypaque discontinuous density gradients into plasma, mononuclear cells and platelets and a pellet containing red cells and granulocytes. These fractions were counted in a gamma spectrophotometer with correction for radioactive decay. The circulating half-life of the labelled neutrophils was measured by counting labelled cells after 12 h.

Stained peripheral blood smears were obtained from four patients before, during and after 1 h infusions of rhGM-CSF and neutrophil lobularity assessed as previously described (Wintrobe, 198 1).

Adhesion to endothelial cell monolayers. Adhesion of X r - labelled neutrophils to human umbilical vein endothelial cell (HUVEC) monolayers was assayed by a modification of the method of Harlan et a1 (1 98 5). HUVECs were harvested and grown to confluence as described by Jaffe et a1 (1973) in flat- bottomed 5 mm x 96-well plates (Nunc). Venous blood from healthy volunteers was collected into ACD and sedimented with 1% hetastarch for 4 5 min at room temperature. The buffy coat was layered onto a Ficol-Hypaque discontinuous density gradient and centrifuged at 400 g for 2 5 min at 20°C. The cell pellet was subjected to hypotonic lysis at 4°C to remove contaminating red cells and washed once in calcium- free phosphate buffered saline (PBS) containing 5 mmol/l glucose. Neutrophils were then resuspended in PBS/glucose at 20 x 10b/ml and incubated with 51Cr sodium chromate (Amersham) 1 pCi/lOh cells for 1 h at 37°C. After four washes in large volumes of PBS/glucose, cells were resus- pended at 5 x 106/ml in RPMI immediately before incubation with GM-CSF 10 ng/ml or medium control for 30 min at 37°C. These conditions were shown to give optimal adher- ence in preliminary studies. In some experiments monoclonal antibodies were included at this stage.

HUVEC cultures were then washed twice with RPMI, drained of medium and 50 pI of stimulated cell suspension containing 2.5 x lo5 labelled neutrophils added and incu- bated for 45 min at 37°C in 5% CO,. After incubation, each well was aspirated and washed twice with 200 p1 of RPMI. The aspirate and washes were saved for counting of non- adherent cells. 200 pl of 1 N NH,OH was added to each well and left overnight to lyse the adherent cells. The lysate was aspirated and each well washed twice with 200 p1 1 N

NH40H. The supernatants and lysates were counted in a gamma spectrophotometer and the percentage adherent cells calculated by dividing the lysate counts by the total counts. Each experiment was performed in quadruplicate at least and stimulated and unstimulated adherence compared using Student’s t test for paired data.

Cellular adhesion molecule expression. Leucocyte-rich plasma was prepared as outlined above except heparin 20 u/ml was added to prevent coagulation upon recalcification. The buffy

coat was removed and incubated at 3 7°C for varying times with rhGM-CSF or medium control (RPMI 1 640 Gibco). CaClz (5 mmol/l) was added at the start of the 37°C incubation. After stimulation, samples were cooled to 4 "C and kept on ice thereafter. 1 ml aliquots were fractionated on Ficol-Hypaque gradients by centrifugation at 300 g for 25 min and the granulocyte pellet subjected to hypotonic lysis at 4°C to remove residual red cells. After washing in medium once, lo5 cells were incubated for 30 min with saturating amounts of MHM23, an IgGl monoclonal anti CD18 antibody and then washed a further three times and incubated with a fluores- cein-conjugated rabbit anti-mouse immiinoglobulin anti- body (DAKO) for 30 min. After three further washes, mean channel fluorescence of the stained neutrophils was deter- mined on a FACS 4 flow cytometer with linear amplification. For each data point control samples were analysed with medium only as the first layer and UCHT1, an IgGl anti CD3 antibody to control for Fc receptor binding. The addition of GM-CSF caused no change in the fluorescence of either control. The mean of the background fluorescence (medium as first layer) was subtracted from each test sample result to obtain a value for the specific fluorescence. Stimulated and unstimulated CAM expression were compared by Student's t test for paired data.

For measurement of in vivo adhesion molecule expression. blood taken into sodium ethylenediaminetetracetate (EDTA) I mg/ml was immediately cooled and then kept on ice. Leucocyte-rich plasma was obtained by hetastarch sedimen- tation then fractionated and stained as described above.

C5a des arg measurement. Samples of vimous blood were taken into sodium EDTA 1 mg/ml before, during and after

140

120

100 c

0

0 8o

8 40

c

c 60 0

20

GM-CSF Induces Changes in CeZZuZar Adhesion 325 infusions of rhGM-CSF and separated at 4°C within 30 min. The plasma was stored at - 20°C for less than 1 month and C5a des arg estimated by radioimmunoassay (Amersham).

RESULTS Leucocyte kinetic studies In accord with our previous observations, a rapid but transient fall in circulating neutrophil and monocyte numbers occurred at the start of rhGM-CSF treatment in all patients with normal white cell counts. This phenomenon was documented on 16 occasions in nine patients (Fig 1). Recovery occurred after 2-3 h regardless of whether bolus or continuous infusions of GM-CSF were employed and was seen with both the bacterial and yeast-derived material.

Radionuclide leucocyte kinetic studies were performed on two patients, one with a resistant sarcoma and one with resistant Hodgkin's disease prior to autologous bone marrow transplantation. As previously reported (Devereux et al, 1987). there was a dramatic but transient accumulation of labelled cells within the lungs.

The recovery in circulating neutrophil and monocyte numbers coincided with the release of labelled cells from the lungs. Serial measurement of the numbers of circulating labelled neutrophils (Fig 2) showed that the recovery in neutrophil count paralleled the return of labelled cells into the circulation, indicating that the recovery was due to demargination rather than release from bone marrow stores. This was supported by the lack of change in neutrophil lobularity observed during the recovery phase. The half-life within the circulation of the labelled neutrophils was 4.5 and 5.5 h.

(a)

I I

0 6 0 120 180

Tlme (rnlnutes)

140

120 c c

0 2 100

0 80 c - 60

8 40

20

0

" 1 I I I

0 6 0 120 180 Tlme (mlnutes)

Fig 1 . Percentage changes in circulating neutrophil (a) and rnonocyte (b) numbers during GM-CSF infusions. Mean of 16 infusions in nine patients fSEM.

326 S. Devereux et a1

30 - al 0 C

E 2 2 0 - a $

U

10 -

120

100

0

60 - c c 0

40 $

20

I GM-CSF I -

A Neutrophils - Mononuclears

0 : I 1 I

0 60 120 180 Tlme (minutes)

Fig 2. Percentage change in circulating numbers of 99"T-labelled cells during an infusion of GM-CSF. Similar results were obtained during a second identical study performed on another patient. Peripheral blood samples were separated into granulocyte/red cell and mononuclear cell fractions on a Ficol-Hypaque gradient before counting. Results are expressed as a percentage of pre-infusion counts per minute.

Y Expt. 1 - Expt.:! - Expt.3 - Expt.4 - Expt.5

Y- Expt.6

0 ' I I

NO GM-CSF GM-CSF

l 0 1

V- Expt 4 I- Expt 5 Y- Expt 6

o ! I I

NO GM-CSF GM-CSF

Fig 3 . Adherence of 5'Cr-labelled neutrophils to endothelial cell monolayers after incubation with 10 ng/ml GM-CSF and medium control for 30 min without (a) and with (b) anti-CD18 monoclonal antibody.

GM-CSF Zndrices Changes in Cellular Adhesion 32 7 120 -

* p=< 0.05 - 100 - *

- RPMl

20 I I I

0 60 120 180

Time (minutes)

Fig 4. The effect of incubation with GM-CSF 10 ng/ml or medium control, for varying periods of time, on neutrophil cellular adhesion molecule (CAM) expression. The average changes in mean cell fluorescence for four experiments (&standard error of mean) are shown.

- Y GM-CSF RX

300 LL

Y 0

E 200 F

s! c 0

100

1 I I

0 6 0 120

Time (minutes)

Fig 5. Changes in neutrophil CAM expression during in-vivo administration of GM-CSF. Results from studies on three patients are shown expressed as the average percentage change from time 0 mean cell fluorescence. A control experiment shows changes in neutrophil CAM expression over time in three healthy individuals not receiving GM-CSF.

Adhesion of neutrophils to endothelial cells in-vitro Adhesion of 51Cr-labelled neutrophils to human umbilical vein endothelial cell (HUVEC) monolayers was significantly increased by preincubation with GM-CSF in all of six experiments (Fig 3a). The adherence increased from a mean of 13.5&3.1%with mediumalone to20-4+2-4%withGM- CSF (P<O.OI). In three experiments, the effect of anti-CD18 antibody on neutrophil adherence was studied. In each experiment, both GM-CSF induced and unstimulated adher- ence were reduced by an average of 52% ,and 60% respect-

ively (Fig 3b). The increased adherence induced by GM-CSF was, however, modest in comparison to that produced by PMA. In five experiments adherence increased from a mean of 13.5&2.9% with medium alone to 49 .2f2 .2% with PMA. Addition of anti CD18 antibody in two experiments reduced PMA induced adherence by an average of 23%. An IgGI anti CD45 antibody which binds to all leucocytes was used to control for Fc receptor binding in two of the blocking experiments. No reduction in adherence was observed (data not shown).

328 S. Devereux et a1 Neutrophil CAM expression In vitro studies in which buffy coat was stimulated for varying periods with rhGM-CSF at 10 ng/ml showed a rapid and significant increase in neutrophil CAM expression (Fig 4). No fall in neutrophil CAM levels were seen at up to 3 h. A spontaneous increase in CAM expression was seen in neutro- phils held at 37°C for more than 1 h.

In three patients receiving rhGM-CSF (15 pg/m2/h) as a test dose prior to chemotherapy and ABMT. a similar in-vivo rise in neutrophil CAM expression was seen, occurring at the same time as the leucopenia (Fig 5). During recovery from the leucopenia however, neutrophil CAM expression remained high.

C5a des arg levels No detectable levels of the activated complement metabolite C5a des arg were found in plasma from four patients during rhGM-CSF infusions. The assay system used was sensitive to levels above 4 ng/ml.

DISCUSSION

These studies show that the leucopenia associated with rhGM-CSF administration is due to a transient shift of leucocytes into the marginated pool and that the recovery is due to re-entry of these cells into the circulation. A similar leucopenia is observed during haemodialysis or endotoxae- mia (Craddock et al. 1960) and with infusions of chemotactic factors in animals (O’Flaherty et al, 1977). Haemodialysis leucopenia is thought to be due to activation of the alterna- tive complement pathway on the dialysis membrane and generation of chemotactic complement fragments (Craddock et al, 19 77). No evidence ofcomplement activation was found in this study and the leucopenia occurred after doses of rhGM- CSF containing insignificant amounts of endotoxin (< 1 ng). These observations, together with the rapidity of onset of the leucopenia, strongly suggest a direct effect of GM-CSF on circulating phagocytes.

We have confirmed the results of Arnaout et a1 (1 986) who showed that GM-CSF causes a rise in neutrophil CAM expression in vitro and show here that a similar phenomenon also occurs with in-vivo administration, the kinetics parallel- ing the development of leucopenia. These in-vivo studies indicate that GM-CSF induced upregulation of CAMS is not dependent on the manipulation associated with in-vitro studies.

Adhesion-promoting glycoproteins are stored in the secondary granules of neutrophils and the peroxidase- negative vacuoles of monocytes and are rapidly translocated to the cell surface in response to stimuli such as chemotactic factors (Todd et al. 1984: Miller et a!. 1987). It is likely that the rapid CAM upregulation seen in vitro and in vivo is due to mobilization of this intracellular pool and that increased expression of these molecules results in margination.

In our studies, neutrophils stimulated with rhGM-CSF in vitro show increased adherence to endothelial cell cultures. This contrasts with the results of Lopez et a1 (1986) who reported that GM-CSF did not cause increased adherence to endothelial cultures whilst tumour necrosis factor (TNF) and

PMA did. This difference may relate to the way in which the neutrophils were prepared since preparative procedures may themselves upregulate CAMs to such an extent that GM-CSF has little further effect (Fearon & Collins, 1983). The margination of neutrophils that occurs during GM-CSF administration demonstrates that adherence to the vascular endothelium occurs in vivo. in support of our in-vitro findings.

The adherence of neutrophils to endothelial cells was reduced in vitro by antibody to the CAM beta chain suggest- ing that the binding of neutrophil CAM to its endothelial receptor is important for GM-CSF induced adherence. In two experiments in which the effects of anti-CAM antibody on PMA induced adherence were tested only partial inhibition was seen, suggesting that this agent causes adherence by mechanisms other than binding of CAMS to endothelial receptors.

There is some disagreement as to the role of neutrophil CAM in adherence to endothelial cells. Zimmerman & MacIntyre (1 988) reported that the binding of neutrophils stimulated by fmlp. leukotriene B4, platelet activating factor, PMA, calcium ionophore and TNF was blocked by anti-CD18 antibody whilst thrombin and leukotriene C4 induced adher- ence was not. It was suggested that CAM dependent and independent mechanisms of adherence exist. In contrast to these findings, Vedder & Harlan (1988) showed that neutro- phils pretreated with the anion channel blocking agent DIDS (diisothiocyanostilbenedisulphonic acid) failed to upregulate surface membrane CAM but still showed enhanced adher- ence in response to fmlp. PMA and calcium ionophore, suggesting that CAMs are not involved in adherence stimu- lated by these agents. Our studies, whilst not primarily addressing this issue, seem to indicate that GM-CSF induced adherence is largely mediated by neutrophil CAMs but that adherence due to PMA may involve additional mechanisms.

Although GM-CSF induced margination can be explained by CAM mediated adherence to endothelial cells, demargina- tion is not, since no significant fall in CAM expression occurred during prolonged exposure to GM-CSF either in vitro or in vivo. This further suggests that factors other than CAMs are involved in the interaction between neutrophils and the endothelium.

Recent work presented by Attieri and colleagues has shown that monocyte CAMs have high affinity binding sites for coagulation factor X and fibrinogen (Altieri et al, 1987). Endothelial cells also possess binding sites for fibrinogen and it may be that the interaction between phagocytes and endothelial cells is stabilized by fibrinogen/fibrin in the same way as that between platelets and the vessel wall. Re-entry of cells into the circulating pool might therefore be due to fibrin(0gen)olysis rather than down regulation of CAMS alone.

Many of the observed side effects of rhGM-CSF therapy could be explained by abnormal adherence of phagocytic cells to the vascular endothelium. Bone pain might be caused by microvascular occlusion by leucocytes analogous to the painful vaso-occlusive crises of sickle cell disease. Damage to endothelial cells induced by neutrophils and monocytes could produce thrombophlebitis, thrombosis or capillary endothe- lial leak.

Gh4-CSF Induces Changes in Cellular Adhesion 329 We have demonstrated tha t neutrophils and monocytes

induced to adhere to the vessel wall by GM-CSF return into the circulation and do not cross the endothelial barrier and enter the tissues. It is probable tha t other factors are necessary to induce this, explaining the lack of serious pulmonary injury following sequestratioin of cells in the lungs. Finally, the cellular sources of (GM-CSF, namely activated lymphocytes, monocytes, endothelial cells and fibroblasts (Clark & Kamen, 1987), are all present in areas of inflammation. Locally produced GM-CSF may therefore act i n concert with other inflammatory mediators to control the accumulation and activation of phagocytic cells in these lesions.

Note added in proof

Since the submission of this manuscript Socinski and col: leagues have reported in vivo upregulation of the C D l l b antigen in response to rh GM-CSF (Blood, 172, 691-697).

ACKNOWLEDGMENTS

This work was supported by the Kay Kendall Leukaemia Fund and a grant from the North East Thames Regional Research Committee. We wish to thank the Immunex Corporation Seattle, Washington, U.S.A., and Behringwerke Marburg, West Germany. for the provision of recombinant human GM-CSF.

REFERENCES

Altieri. D.. Morrissey. J., Springer, T. & Edgington. T.S. (1987) Binding of coagulation factor X to ADP stimulated monocytes: a novel biological function of the adhesive receptor Macl. ASH Meeting. Blood. 70, Suppl.. 185.

Antman, K.. Griffin, J., Elias. A,, et a1 (1987) Use of rGM-CSF to ameliorate chemotherapy induced myelosuppression in sarcoma patients. Blood. 1987; 70 (suppl) 373.

Arnaout, M.A., Wang, E.A.. Clark, S.C. & Sieff, C.A. (1986) Human recombinant granulocyte-macrophage colony-stimulating factor increases cell to cell adhesion and surface expression of adhesion promoting glycoproteins on mature granulocytes. Journal of Clinical Investigation, 78, 597-601.

Brandt, S.J.. Peters, W.P.. Atwater. S.K.. et ul (1988) meet of recombinant human granulocyte-macrophage colony-stimulat- ing factor on hemopoietic reconstitution fcdlowing high dose chemotherapy and autologous bone marrow transplantation. New England Journal of Medicine. 318, 869-875.

Clark. S.C. & Kamen, R. (1987) The human hematopoietic colony stimulating factors. Science, 236, 1229-1 237.

Craddock. C.C., Perry, S., Ventzke. L.E. & Lawrence. J.S. (1960) Evaluation of marrow granulocytic reserves in normal and disease states. Blood, 1 5 , 840-855.

Craddock. P.R.. Fehr. J., Dalmasso. A.P., Brigham. K.L. &Jacob, H.S. (1 977) Hemodialysis leukopenia. Journal of Clinical Investigation. 59, 879-888.

Devereux. S.. Linch. D.C.. Campos-Costa. D.. Spittle, M.F. & Jelliffe, A.M. ( I 987) Transient leucopenia induced by ;granulocyte macro- phage colony stimulating factor. (Letter). Lancet, ii, 1523.

Devereux. S., Patterson.K.P.,Goldstone. A.H.G. &Linch, D.C. (1988) The use of recombinant human granulocyte macrophage colony stimulating factor to accelerate haemopoietic recovery after

autologous bone marrow transplantation. (Abstract). British Journal of Haematology. 69, 103.

Donahue, R.E., Wang. E.A.. Stone. D.K., et a1 (1986) Stimulation of haematopoiesis in primates by continuous infusion of recombi- nant human GM-CSF. Nature, 321, 872-875.

Fearon. D.T. & Collins, L.A. (1983) Increased expression of C3b receptor on polymorphonuclear leukocytes induced by chemotac- tic factor and by purification procedures. Journal of Immunology.

Gasson. J.C., Weisdorf, R.H.. Kaufman. S.E.. et al (1984) Purified human granulocyte macrophage colony stimulating factor: direct action on neutrophils. Science, 226, 1339-1 342.

Grabstein, K.H., Urdal, D.L., Tushinski, R.J.. eta1 (1986) Induction of tumoricidal activity by granulocyte-macrophage colony-stimulat- ing factor. Science. 232, 506-508.

Groopman, J.E.. Mitsuyasu. R.T.. DeLeo, M.J., Oette, D.H. & Golde, D.W. (1 98 7) Effect of granulocyte-macrophage colony stimulating factor on myelopoiesis in the acquired immune deficiency syn- drome. New England Journal of Medicine, 317, 593-598.

Harlan, J., Killen, P.D., Senecal. F.M.. et al (1985) The role of neutrophil membrane glycoprotein GP-1 SO in neutrophil adher- ence to endothelium in vitro. Blood, 66, 167-1 78.

Jaffe. E.A.. Nachman. R.L.. Becker. C.G. & Minick. C.R. (1973) Culture of human endothelial cells derived from umbilical veins. Journal of Clinical Investigation. 52, 2745-2756.

Lopez, A.F., Williamson, J.. Gamble. J.R.. et al (1986) Recombinant human granulocyte-macrophage colony stimulating factor stimu- lates in vitro mature human neutrophil and eosinophil function, surface receptor expression and survival. journal of Clinical Investi- gation, 78, 1220.

Makgoba. M.W., Sanders, M.E.. GintherLuce, G.E.,etal(1988)ICAM 1 a ligand for LFA 1 dependent adhesion of B. T and myeloid cells. Nature, 331, 86-88.

Metcalf, D. (1986) The molecular biology of the granulocyte- macrophage colony-stimulating factors. Blood, 67, 25 7-267.

Miller, L.J.. Bainton. D.F., Borregaard. N. & Springer, T. (1987) Stimulated mobilisation of monocyte Macl and pl50,95 adhesion proteins from an intracellular vesicular compartment to the cell surface. Journal of Clinical Investigation. 80, 535-544.

Morrissey. P.J., Bressler. L.. Park, L.S., Alpert, A. & Gillis, S. (1987) Granulocyte-macrophage colony-stimulating factor augments the primary antibody response by enhancing the function of antigen presenting cells. Journal of Immunology. 139. 11 13-1 119.

OFlaherty, J.T., Showell, H.J. & Ward, P.A. (1977) Neutropenia induced by systemic infusion of chemotactic factors. Journal of Immunology, 118. 1586-1 589.

Peters, A.M., Danpure, H.J., Osman. S.. et al (1986) Clinical experience with 99mTc Hexamethylpropyleneamineoxime labelled leucocytes and imaging inflammation. Lancet. ii, 946- 949.

Sieff, C.A.. Emerson, S.G., Donahue. R.E.. et al (1988) Human recombinant granulocyte-macrophage colony-stimulating factor: a human multilineage hemopoietin. Science. 230, 1171.

Springer.T.A.,Dustin, M.L..Kishimoto.T.K.&Marlin,S.D.A. (1987) The lymphocyte function associated LFA-1, CD 2 and LFA-3 molecules: Cell adhesion receptors of the immune system. Annual Review of Immunology, 5, 223-252.

Todd, R.F.. Arnaout, M.A., Rosin, R.E., Crowley. C.A., Peters, W.A. & Babior, B.M. (1 984) Subcellular localisation of the large subunit of Mol (Mola formerly GPI 10) a surface glycoprotein associated with neutrophil adherence. Journal of Clinical Investigation, 74, 1280-1290.

Vadas, M.A., Nicola. N.A. & Metcalf. D. (1983) Activation of antibody dependent cell mediated cytotoxicity of human neutro-

130,370-375

3 3 0 S . Devereux et a1 phils and eosinophils by separate colony stimulating factors. Journal of Immunology. 130, 795-799.

Vadhan-Raj, S., Keating, M., LeMaistre, A., et al (1987) Effects of recombinant human granulocyte-macrophage colony-stimulat- ing factor in patients with myelodysplastic syndromes. New England Journal of Medicine, 317, 1545-1551.

Vedder, N.B. & Harlan, J.M. (1988) Increased expression of CD1 lb/ CD18 (Macl) is not required for stimulated neutrophil adherence to cultured endothelium. Iournal of Clinical Investigation, 81, 676- 682.

Weisbart, R.H.. Golde. D.W.. Clark, S.C.. Wong. G.C. & Gasson. J.C. (1 985) Human granulocyte macrophage colony-stimulating fac- tot is a neutrophil activator. Nature, 314, 361-363.

Wintrobe, M.M. ( 7 981) Cfjnical Hernafu~ogy. 8th edn. Lea & Feibiger. Philadelphia.

Zimmerman. G.A. & McIntyre. T.M. (1988) Neutrophil adherence to human endothelium in vitro occurs by CDwl8 (Mol, MAC-1/ LFA/GP 1 50.9 5) glycoprotein-dependent and -independent mechanisms. Journal of Clinical Investigation. 81, 531-537.