Biological characterization of purified macrophage-derived neutrophil chemotactic factor

We have recently described the purification of a 54 kDa acidic protein, identified as macrophage-derived neutrophil chemotactic factor (MNCF). This protein causes in vitro chemotaxis as well as in vivo neutrophil migration even in animals treated with dexamethasone. This in vivo chemotactic activity of MNCF in animals pretreated with dexamethasone is an uncommon characteristic which discriminates MNCF from known chemotactic cytokines. MNCF is released in the supernatant by macrophage monolayers stimulated with lipopolysaccharide (LPS). In the present study, we describe some biological characteristics of homogenous purified MNCF. When assayed in vitro, MNCF gave a bell-shaped dose–response curve. This in vitro activity was shown to be caused by haptotaxis. Unlike N-formyl-methionylleucyl- phenylalanine (FMLP) or interleukin 8 (IL-8), the chemotactic activity of MNCF in vivo and in vitro, was inhibited by preincubation with D-galactose but not with D-mannose. In contrast with IL-8, MNCF did not bind to heparin and antiserum against IL-8 was ineffective in inhibiting its chemotactic activity. These data indicate that MNCF induces neutrophil migration through a carbohydrate recognition property, but by a mechanism different from that of the known chemokines. It is suggested that MNCF may be an important mediator in the recruitment of neutrophils via the formation of a substrate bound chemotactic gradient (haptotaxis) in the inflamed tissues.


Introduction
Emigration of neutrophils from the circulating blood to the site of the injury is a cardinal event during the inflammatory process. This phenomenon is complex and mediated by various molecules acting in different ways. Interleukin 1 and 8 (IL-1, IL-8), tumour necrosis factor (TNF), and the complement fragments and leukotriene B4 (LTB4) are macrophage products thought to be involved in the migration of neutrophils to the inflamed site, a phenomenon characteristic of the acute phase of the inflammatory process. [2][3][4][5] Some products such as C5a and LTB 4 may have a direct chemotactic effect on neutrophils. [6][7][8] Others may act indirectly by mechanisms which depend on the tissue's resident cells and on the endothelium. For example, in addition to their effects on neutrophils and endothelial cells, where they increase the expression of adhesion molecules, IL-1 and TNF also stimulate the release of chemotactic factors by resident macro-phages, il In fact, the neutrophil-induced migration into the peritoneal cavity by cytokines such as IL-1 and TNF, as well as various inflammatory stimuli (LPS, carrageenin, etc), is dependent on the number of resident cells. This idea is supported by the alteration in the magnitude of the neutrophil migration induced by various stimuli when the resident cell population in the peritoneum is increased by previous treatment with thioglycollate or reduced by cell depletion.11, [14][15][16][17] Glucocorticoids greatly reduce the neutrophil migration induced by inflammatory stimuli or by cell-dependent chemotactic cytokines. The mechanism by which glucocorticoids inhibit cell migration is still controversial. There are some indications that these substances block the expression of adhesion molecules in the endothelium, la However, there is also evidence that glucocorticoids may not have a direct effect on the expression of these proteins in neutrophils or endothelial cells but instead may block the release of the mediators involved in the final rats pretreated or not with glucocorticoids, or in stage of neutrophil emigration into tissue. 19 '2 vitro for their ability to induce neutrophil migra-In the companion paper 2 we describe the iso-tion in a microchamber in the presence of either lation of a homogeneous protein which, like anti-IL-8 serum or of control serum. crude MNCF, induced neutrophil migration in vitro as well as in vivo even in animals pre-Determination of biological activity: treated with dexamethasone. The purified factor was acidic and had a molecular mass of 54 kDa. In vivo migration assays. MNCF activity was In the present study, we describe some biologi-assayed by its ability to induce neutrophil migracal characteristics of this purified homogeneous tion into the peritoneal cavity or air pouch of MNCF fraction associated with their neutrophil rats pretreated with a glucocorticoid (0.5 mg of chemoattractant activity in vivo and in vitro, dexamethasone acetate ester/kg, subcutaneously; These activities are dependent upon the carbohy-Merck, Darmstadt, Germany). Each sample was drate recognition property of MNCF. tested in groups of five to six rats. Air pouches were produced on the dorsum of rats as described by Edwards et a/. 23 One h after dex-Materials and Methods amethasone, 1 or 3 ml samples were injected into the air pouch or peritoneal cavity, respectively. Animals: Male, albino, Wistar rats (Rattus norve-The animals were sacrificed by cervical dislocagicus) weighing from 180 to 200g and maintion 4 or 6 h later for the peritoneal or air pouch tained in temperature-controlled rooms at 23tests, respectively. The cells were immediately 25C with free access to food and water, were harvested by washing the respective cavities with used as the source of peritoneal macrophages as 5 or 10ml PBS containing albumin (0.1%w/v) well as for the in vivo tests of cell migration, and heparin (5 IU/ml). Total counts of harvested cells were performed in a Neubauer chamber.
Production and purification of neutrophil chem-Differential counts were made on smears stained otactic factor (MNCF): MNCF was produced and using Rosenfeld's panchromic method. The purified as described by Cunha and Ferreira, and results are reported as the mean number Dias-Baruffi et al. 14,22 Briefly, the supernatant (4-S.E.M.) of neutrophils per ml of cavity wash. from LPS-stimulated macrophage monolayers To test the inhibitory effect of D-galactose or (crude MNCF) were submitted to a simple two-)-mannose on the neutrophil migration induced step purification process involving adsorption to by MNCF or other stimuli, the samples were a )-galactose column followed by gel filtration incubated with these sugars (final concentration on Superdex 75, for the isolation of MNCF. In 0.4M) for 30 min at room temperature. The neuthe present paper MNCF refers to the homotrophil migration activity of the stimuli was then geneous fraction obtained after the last purifica-tested in the air pouch model as described tion step. above.
Heparin binding assay of MNCE. Binding to In vitro neutrophil chemotaxis migration assay.
heparin was assessed with a heparin-agarose Assays of in vitro neutrophil migration were percolumn. The column was washed and equilib-formed as described by Bignold  fraction while that retained and eluted with 0.5 M medium, the neutrophils were resuspended in NaCl in PBS was denoted heparin-bound. Both RPMI 1640 medium containing 0.1% (w/v) BSA fractions were applied to a Fast Desalting HR 10/ (RPMI-BSA), to provide 106 cells/ml. Typical pre-10 column (Pharmacia LKB Biotechnology, parations contained more than 95% viable neu-Uppsala, Sweden). The column was developed trophils. Purified neutrophils were placed in the with sterile deionized water at 20C and a flow upper chamber while the lower chamber conrate of 0.5ml/min. The eluate absorbance was tained the test samples dissolved in RPMI-BSA. monitored at 206 and 280nm, and fractions of Random migration was assessed by using RPMI-0.5 ml were collected. These fractions were then BSA in the lower chamber. The peptide FMLP tested either in vivo for their ability to induce (10 -7 M) was used as the reference chemoatneutrophil migration into the peritoneal cavity of tractant. The number of cells that migrated through the entire thickness of a 5/.tm polycarbonate filter (Millipore Corp., Bedford, MA, USA) during the i h incubation at 37C in a 5% COg atmosphere was counted. Five fields were counted for each assay and each sample was assayed in triplicate. The results are reported as the mean number (_+ S.E.M.) of neutrophils per field. The samples tested were crude MNCF, purified MNCF or human recombinant interleukin-8 (rhIL-8; from National Institute for Biological Standards and Control-NIBSC) as well as the heparin-bound and heparin-not-bound fractions of the LPS-stimulated macrophage supernatant. Crude MNCF, purified MNCF or rhIL-8 were incubated in medium with )-galactose or Dmannose (0.4M in RPMI-BSA) for 30 min at room temperature before testing the effect of )-galactose or D-mannose on the chemotactic activity of each sample. The heparin-bound and heparin-not-bound fractions, as well as rhlL-8, were incubated separately in medium containing anti-IL-8 goat serum or control goat serum (final dilutions 1:100 and 1:10 in RPMI-BSA) for 30 min at room temperature before testing the effect of anti-IL-8 on the chemotactic activity of each sample.
In vitro neutrophil haptotaxis assay. The haptotaxis assay was conducted as described by Rot 25 in a 48-well chemotactic chamber (Neuroprobe) by forming a gradient across a 51.tm polycarbonate filter (Millipore Corp.) Positive haptotactic gradients of MNCF were preformed by filling some of the bottom wells with purified MNCF derived from 3.6 x 10 7 ceils and the corresponding top wells with RPMI-BSA. Negative haptotactic gradients were obtained by reversing the above procedure, i.e., by filling a set of top wells with MNCF derived from 3.6 x 10 7 cells and the corresponding bottom wells with RPMI-BSA. When chemotaxis, chemokinesis or random migration were examined, both the upper and lower wells were filled with RPMI-BSA. After incubation at 37C for 20 min, the chambers were disassembled. The filter was thoroughly washed in an RPMI-BSA bath to remove the nonbound attractant, after which it was air dried and placed in another chemotactic chamber. In this second chamber, RPMI-BSA was used to fill the bottom wells corresponding to those wells with preformed positive or negative gradients in the first chamber. The wells corresponding to those that did not contain attractant in the first chamber were used to assess chemotaxis (by adding the attractants MNCF or FMLP (10-7 M) to the bottom chamber), chemokinesis (by adding equal quantities of MNCF to both the top and bottom chambers), or random migration (by adding RPMI-BSA to both the top and bottom chambers). Fifty l.tl of a suspension of 10 neutrophils/ml were placed in each top well of the second chamber, and their migration was measured by counting the cells that migrated into or towards the second chamber across the filter during a 30 min incubation at 37C. Five fields of filters were counted for each assay and each sample was assayed at least in triplicate. The results are reported as mean 4-S.E.M.

Results
Biological characterization of chemotactic activities of MNCE. Purified MNCF induced neutrophil migration in the same dose-dependent manner in both normal and dexamethasone-pretreated rats (Fig. 1A). The dose-response curve obtained for the in vitro chemotactic activity of purified MNCF was bell-shaped (Fig. 1B). The chemotactic activity was shown to result from haptotaxis rather than chemokinesis since the neutrophil chemokinetic migration induced by purified MNCF was similar to that of a negative control. While in vitro neutrophil migration was observed with a positive haptotactic gradient with MNCF, the negative haptotactic gradient of MNCF was unable to induce neutrophil migration (Fig. 2).
These data indicate that the neutrophil migration induced by filter-bound MNCF was gradientdriven.  Table 1 shows that pretreating the test animals with dexamethasone blocked the neutrophil migration induced by the heparin-bound fraction and by rhlL-8. However, as with the results obtained for purified MNCF (Fig. 1), this treatment did not affect the neutrophil migration induced by the heparin-not-bound fraction. In the in vitro assay, the chemotactic activity of the heparin-bound fraction was inhibited by preincubation with anti-IL-8 serum. A similar result was observed with rhlL-8. On the other hand, the in vitro chemotactic activity of the heparin-notbound fraction was not affected by the same amount of anti-IL-8 serum.
Relationship between migration-inducing and carbohydrate-binding activities: Fig. 3A shows that 0.4M D-galactose did not affect the chemotactic activity of FMLP or IL-8 in vitro, but did inhibit the activity of purified MNCF by more than 90%. As expected, the chemotactic activity of crude MNCF was only partially reduced (28%) by the same concentration of D-galactose. At the same concentration, a non-specific sugar ()mannose) had no effect on the chemotactic activities of FMLP or purified MNCF (Fig. 3B).
The incubation of crude or purified MNCF with D-galactose (0.4 M) also inhibited the ability to induce neutrophil migration into the air pouch  MNCF, a lectin-like ytokine Crude MNCF (from 3 x 10 macrophages), or purified MNCF (from x 10 macrophages) previously non-incubated or incubated with Dgalactose or D-mannose (final concentration 0.4 M), was injected into the air pouches of dexamethasone-pretreated rats (0.5 mg/kg). Neutrophil migration was evaluated 4h later. The results represent the percentage inhibition compared with the migration induced by the same stimulus when non-incubated with D-mannose or D-galactose and incubated with each sugar. of dexamethasone-pretreated rats by 40 and 46%, respectively. At the same concentration, a nonspecific sugar (>mannose) had no effect on the in vivo or in vitro chemotactic activity of crude or purified MNCF (Table 2).

Discussion
In a companion paper, 21 we have described the purification to homogeneity of a .54kDa acidic protein extracted from the supernatant of LPS-stimulated macrophage monolayers. Like the crude macrophage-derived neutrophil chemotactic factor (MNCF), this protein was able to stimulate neutrophil migration in animals pretreated 14 21 with dexamethasone. This in vivo chemotactic activity is an uncommon cytokine characteristic and seems peculiar to MNCF. MNCF behaves like a lectin on the basis of its ability to bind to immobilized )-galactose and the inhibition of its biological activity by the same sugar. 22 In line with our previous observations for crude MNCF, the isolated protein induced dosedependent neutrophil migration in vivo in naive and as well as in dexamethasone-pretreated animals. >galactose, but not )-mannose, significantly inhibited (46%) the capacity of purified MNCF to induce neutrophil migration in viva This limited inhibitory effect of D-galactose in the in vivo assay may reflect the diffusion and/or the reabsorption of the sugar from the air pouch cavity during the 6 h required for the neutrophil migration test.
MNCF also caused chemotaxis in vitro with the dose-response being similarly bell-shaped to that described for other chemotactic agents such as FMLP and LTB4, 26 i>mannose-bindin/_ lectin from Artocarpus integrifolia (KM ) and  The decreased activity observed with high doses of the factor may result from its diffusion into the upper compartment of the chamber, thereby abolishing the chemoattractant gradient. The in vitro chemotactic activity of MNCF was strongly inhibited by pre-incubation with ,)-galactose. This inhibition appears to be specific for MNCF since pre-incubation of FMLP or IL-8 with the same concentration of I)-galactose did not affect their chemotactic activity. Pre-incubation with a non-related sugar, >mannose, did not modify the activity of MNCF or FMLP. The crude MNCF was only partially inhibited by )-galactose since other chemotactic mediators such as IL-1, TNF and IL-8 5'28 were also present. These results support the suggestion that MNCF activity in vitro and in vivo is associated with a sugarbinding domain of the purified protein.
There are at least three distinct stages in the process of neutrophil migration into perivascular tissues: (1)  and (3) neutrophil migration across the basal membrane into perivascular tissues. 9'2 Haptotaxis is thought to be an important mechanism throughout this whole process. Binding of the chemoattractant on the subendothelial matrix is required for the formation of a haptotactic gradient. In the present paper, we have shown that MNCF causes in vitro chemotaxis by both chemotactic and haptotactic mechanisms, properties shared by IL-8. 25 However, in contrast to IL-8, MNCF did not bind to heparin and its chemotactic activity was blocked by -galactose but not by anti-IL-8 serum (Table 1). Thus, the sugar binding moiety which is involved in the stimulation of neutrophil chemotaxis by MNCF is different from that involved in IL-8-induced chemotaxis.
Early in vivo experiments suggested that the inhibitory effect of glucocorticoids on neutrophil migration was due to an action on the neutrophil-endothelium adhesion step. 4-There are, however, conflicting results regarding the inhibitory effect of glucocorticoids on the foregoing interaction in vitro 37 and on the expression of Eselectin and 1CAM-1. 8 Recent in vivo monitoring of the different stages of neutrophil migration has shown that the rolling on or adhesion to venular endothelium induced by LTB 4 or FMLP was not inhibited by pretreating the animals with 0 dexamethasone.
In addition, glucocorticoid treatment did not inhibit neutrophil penetration via the endothelial cell junctions, although it was noted that the neutrophils subsequently remained between the endothelium and the basal membrane without reaching the perivascular tissue. 19 '2 From these experiments it is clear that glucocorticoids affect neutrophil migration mainly by inhibiting an unidentified mechanism responsible for the transmigration through the basal membrane to the perivascular tissue.
The obseeeation that dexamethasone does not affect MNCF-induced chemotaxis in vivo but does block the migration induced by IL-8 allows one to speculate about distinct mechanisms involved in the neutrophil migration induced by both cytokines. MNCF-induced chemotaxis in vivo must result from the protein's ability to promote the three stages of neutrophil emigration to tissues described above. IL-8-induced chemotaxis in vivo, however, may result from mechanisms involving a dexamethasone-sensitive release of chemotactic factors by resident cells such as mast cells 5 and/or the formation of a haptotactic gradient. 25 '38 In relation to the generation of the haptotactic gradient, IL-8 and MNCF may recognize different ligands on the subendothelial matrix, and glucocorticoids may be able to block the expression of the IL-8 ligand while not affecting that of MNCF. The hypothesis of different ligands for IL-8 and MNCF is reinforced by the fact that MNCF, in contrast to IL-8 did not bind to heparin (Table 1), which is structurally related to heparan sulfate. 39 Heparan sulfate has been proposed to be the extracellular matrix ligand for IL-8 responsible for inducing neutrophil migration by the haptotactic mechanism. 9-1 Based on the results of this study, we hypothesize that MNCF forms a dexamethasone-insensitive substrate bound gradient (haptotactic gradient) responsible for the last phase of neutrophil emigration through the basal membrane of acutely inflamed tissues. The previously reported dexamethasone blockade of the release of MNCF by macrophage monolayers stimulated by inflammatory stimuli 4 may at least in part explain why the last phase of neutrophil migration is impaired in dexamethasone pretreated animals.
In conclusion, a 54kDa acidic protein, identified as macrophage-derived neutrophil chemotactic factor (MNCF), has lectin-like properties and causes neutrophil migration in vivo in animals treated with dexamethasone, as well as neutrophil chemotaxis in vitro by a haptotactic mechanism. MNCF is a candidate for a new cytokine and may play an important role in the transmigration of neutrophils to perivascular tissues.