Mobilization of Opioid-containing Polymorphonuclear Cells by Hematopoietic Growth Factors and Influence on Inflammatory Pain

Animal protocols were approved by the animal care committee of the State of Berlin (Landesamt für Arbeits-schutz, Gesundheitsschutz und Technische Sicherheit, Berlin, Germany) and were in accordance with the guidelines of the International Association for the Study of Pain. 21Male Wistar rats weighing 140–200 g were injected with 150 μl Freund complete adjuvant (FCA; Calbiochem, La Jolla, CA) into the right hind paw (intraplantar) and developed an inflammation confined to the inoculated paw. Experiments were conducted at 2 h after injection of FCA unless stated otherwise. For all injections, rats were briefly anesthetized with halothane using an open circuit without nitrous oxide.

Rats were randomly distributed into cages by the animal caretakers, who did not participate in the experiments and were not informed about the study protocol. Before the distribution of rats, cages were marked with letters (A or B) for either the control or the treatment group. Over 5 consecutive days before FCA inoculation, rats were injected subcutaneously once daily with a combination of 50 μg/kg recombinant rat SCF (Amgen Inc., Thousand Oaks, CA) and 100 μg/kg recombinant human G-CSF (filgrastim; Amgen) in 100 μl glucose 5% (Braun-Melsungen, Melsungen, Germany), as described previously. 15Rats in the control group were injected subcutaneously with solvent (100 μl glucose 5%).

Heparinized blood (20 units/ml, Braun-Melsungen) was obtained by cardiac puncture from deeply anesthetized rats, fixed and lysed as described by the manufacturer (FACS lysing solution®; BD Biosciences, Heidelberg, Germany). PMN and mononuclear cells (including monocytes and lymphocytes) were identified by forward–sideward scatter characteristics. For adhesion molecule staining, blood was incubated with fluorescein isothiocyanate–conjugated monoclonal antibodies (mAbs; all at 20 μg/ml and all from BD Biosciences) for 15 min at room temperature in the dark and further processed as described at the beginning of this paragraph. The following antibodies were used: hamster anti-rat CD62L (L-selectin) antibody (isotype: immunoglobulin [Ig] G, clone HRL-1), mouse anti-rat CD49d (VLA-4) mAb (isotype: IgG^2a, clone MRα4-1), and mouse anti-rat CD18 mAb (isotype: IgG¹, clone WT.3). Controls were incubated with the appropriate isotype-matched antibodies: hamster IgG antibody (clone B81-3), mouse IgG^2a, and mouse IgG¹. The intensity of staining was measured by geometric mean fluorescence intensity (MFI). To correct for background nonspecific staining, the MFI of the isotype-matched control mAb was subtracted from the MFI of the adhesion molecule mAb.

Paw tissue was dissected from the inflamed paw, leaving the deep flexor tendon in situ . The overlying skin was removed and the tissue was cut into small fragments, enzymatically digested for 60 min, and pressed through a nylon filter to obtain single-cell suspensions as previously described. 11,22All samples were stained with Cy-Chrome (BD Biosciences Pharmingen, San Diego, CA) -conjugated mouse anti-rat CD45 mAb (against all hematopoietic cells; 4 μg/ml). For intracellular staining, cells were fixed with 1% paraformaldehyde, permeabilized with saponin buffer, and stained with the following mAbs: mouse anti-rat ED1 fluorescein isothiocyanate (monocytes/macrophages, 2 μg/ml; Serotec, Oxford, England), mouse anti-rat RP-1 phycoerythrin (PMN, 12 μg/ml; BD Biosciences), and mouse anti-rat CD3 phycoerythrin (lymphocytes, T cells, 4 μg/ml; BD Biosciences). Intracellular opioid peptides were stained using the pan-opioid mouse 3E7 (subtype IgG^2a; Gramsch Laboratories, Schwabhausen, Germany), an isotype-matched murine mAb as a control, and secondary antibodies. 11,22 

Fluorescence-activated cell staining events from fluorescent TruCOUNT® beads (BD Biosciences) and from stained cells were collected simultaneously as previously described (70,000 events/paw and 20,000 events in the cell migration assay). 11,22The number of cells per tube was calculated in relation to the known number of fluorescent TruCOUNT® beads. Data were analyzed using CellQuest software (BD Biosciences).

After sternotomy, rats were perfused through the left ventricle with 0.9% NaCl containing 2.7 mm EDTA at room temperature (Sigma Chemicals, Taufkirchen, Germany) using an automatic roller pump (Reglo Analog; Ismatec, Glattbrugg, Switzerland). A total of 120 ml perfusate was collected from the right ventricle by passive drainage. After dextran sedimentation (Amersham Biosciences, Uppsala, Sweden) to remove erythrocytes, the cell suspension was layered on a two-phase Percoll (Amersham Biosciences) gradient (58.6% Percoll with 0.15 m NaCl and 78% with 0.18 m NaCl) (Amersham Biosciences, Freiburg, Germany) and centrifuged at 700 g  for 40 min. PMN were further purified by negative selection with mouse anti-rat CD3 fluorescein isothiocyanate and anti-rat CD4 fluorescein isothiocyanate mAbs using magnetic cell separation (MACS; Miltenyi, Bergisch-Gladbach, Germany) as described previously, 11and cell pellets were frozen until further usage. Cell purity was determined by staining with phycoerythrin-conjugated anti-RP-1 (PMN marker) mAb and analyzed by flow cytometry. Cell pellets were thawed, reconstituted with radioimmunoassay buffer, and lysed by repetitive freeze–thaw–sonication cycles. Met-enkephalin and β-endorphin content was measured using commercially available kits (Bachem, Weil am Rhein, Germany) as previously described. 11According to the manufacturer, the cross-reactivity of the anti-Met-enkephalin antibody is 0.1% to human β-endorphin and that of the anti-rat β-endorphin antibody is 0% to Met-enkephalin. The sequence of Met-enkephalin is identical in rats and humans.

To remove erythrocytes, peripheral blood leukocytes were purified by dextran sedimentation and subsequent hypotonic lysis. The chemokine macrophage inflammatory protein-2 (MIP-2; PeproTech, London, England) was dissolved in complete medium (Roswell Park Memorial Institute 1640 medium; Life Technologies, Paisley, Scotland) containing 100 U/ml penicillin, 10 mg/ml streptomycin, and 10% fetal calf serum (all from Invitrogen, Groningen, The Netherlands) at a dose range of 10⁻⁶to 10⁻¹⁰m and was placed into the lower well of a Boyden chamber (Costar, Bodenheim, Germany). After trypan blue staining and counting, 4 × 10⁵cells in 100 μl complete medium were added to the upper well, separated from the lower well by a cell permeable nylon membrane. After incubation for 60 min at 37°C and 5% CO², the migrated cell suspension from the lower well was stained with mouse anti-rat CD45 Cy-Chrome (BD Biosciences Pharmingen) Ab (4 μg/ml) and was quantified by TruCOUNT® beads. A migration index was calculated by dividing the number of migrating cells toward MIP-2 by the number of migrating cells under control conditions (medium alone).

Subcutaneous plantar paw tissue was dissected, and a piece of 25–35 mg from the central inflamed part of the paw was homogenized in Trizol (Invitrogen) as previously described. 23To synthesize complementary DNA (cDNA), RNA was annealed to 0.125 μm oligodesoxythymidine for 3 min at 65°C and transcribed in the presence of reverse-transcriptase buffer, 0.01 m dithiothreitol, 1 mm deoxynucleoside triphosphate, 10 U avian myeloblastosis virus reverse transcriptase, and 4 U ribonuclease inhibitor (all from Roche Diagnostics, Mannheim, Germany, except dithiothreitol, Sigma Chemicals) at 42°C for 2 h.

Polymerase chain reaction (PCR) was performed with DNA Sybr green following the manufacturer’s instructions (Roche Diagnostics) using 10 μm of both sense and antisense primers (listed in the next paragraph). cDNA was amplified in a four-step-per-cycle PCR consisting of 95°C for 30 s, primer-specific annealing, elongation, and a fluorescence detection (temperature step [Tm]) step (table 1). In pilot experiments, the temperature was determined just below the product-specific melting temperature, and this Tm was incorporated into the PCR for detection of fluorescence. Chemokine messenger RNA (mRNA) was quantified using duplicate samples and a standard dilution with a known number of copies for a specific PCR product as described previously. 23Ribosomal protein L19 (RPL-19) was used as a reference gene for quantification.

Primers were intron spanning, and genomic amplification was excluded by agarose visualization. The following primers were used: RPL-19 (sense [SE] 5′ AAT CGC CAA TGC CAA CTC TCG and antisense [as] 5′ TGC TCC ATG AGA ATC CGC TTG); keratinocyte-derived chemokine (SE 5′ AAT GAG CTG CGC TGT CAG TGC and as 5′ TTG GGG ACA CCC TTT AGC ATC); MIP-2 (SE 5′ CAA CCA TCA GGG TAC AGG GGT and as 5′ GTC CTG GAG GGG TCA CCG T); lipopolysaccharide-induced chemokine (SE 5′ CCA AGG TGG AAG TCA TAG C and as 5′ GCT TTG TTT TCT TAT TTT CAC TGC); cytokine-induced chemokine 2 (SE 5′ CAG CCT TCA GGG ACT GTT G and as 5′ AGC TGG ACT TGT CAC TCT TC).

Nociceptive thresholds were assessed using the paw pressure algesiometer as previously described (modified Randall-Selitto test; Ugo Basile, Comerio, Italy). 8,11,22Briefly, pressure was applied to the dorsal surface of the hind paw. The average pressure required to illicit paw withdrawal, the paw pressure threshold (PPT), was determined by three consecutive trials both on the inflamed and on the contralateral noninflamed hind paw. PPT determination was performed before treatment (baseline) and 1 min after CWS 8,11or 5 min after intraplantar injection of CRF. 10Previous dose–response analysis of CRF injection revealed a peak effect at 8 ng that was fully antagonized by intraplantar but not subcutaneous naloxone, demonstrating peripheral opioid-mediated antinociception (data not shown; subcutaneous injections were performed at a remote location on the back to exclude any systemic effect). Data are presented as percent of maximum possible effect (calculated as [PPT^treated− PPT^pretreated]/[250 − PPT^pretreated; a cutoff value of 250 g was used in all experiments to exclude tissue damage). 22All experiments were performed by an examiner blinded to the treatment protocol.

In each experiment, rats were randomly assigned to one of two groups that were injected with G-CSF+SCF (treatment) in 5% glucose or with the solvent only (5% glucose as a control) as described above (Material and Methods, PMN Mobilization by Hematopoietic Growth Factors). On the fifth day and 3 h after the last injection, an examiner blinded to the group assignment injected all rats with FCA intraplantarly and performed the subsequent experiments (removal of paw tissue or behavioral experiments; number of animals per group are given in the figure legends): (1) determination of circulating PMN by flow cytometry and of opioid peptide content by radioimmunoassay; (2) expression of adhesion molecules on circulating PMN by flow cytometry; (3) chemotaxis of circulating PMN in the Boyden chamber; (4) quantification of chemokine mRNA in the inflamed paw by LightCycler (Synthegen, Houston, TX) PCR; (5) quantification of leukocyte subpopulations and of opioid-containing leukocytes in the inflamed paw by flow cytometry; and (6) measurement of analgesia in behavioral experiments before and after CWS or injection of CRF.

Normally distributed data were analyzed by t  test. If the data lacked equal variance or normality, the Mann–Whitney rank sum test was used. Comparisons of multiple time points were analyzed by analysis of variance (ANOVA) or ANOVA on ranks with a post hoc  comparison by the Dunnett and Dunn method, respectively. Migration indexes were compared by two-way ANOVA. A P  value less than 0.05 was considered significant. Data are expressed as mean ± SEM.

Treatment with G-CSF+SCF significantly increased the number of circulating PMN (fig. 2A; PMN: 11-fold increase; P < 0.05, Mann–Whitney rank sum test) but not of mononuclear cells (all leukocytes except PMN, i.e. , mostly lymphocytes; control: 1,487 ± 182 vs.  2,548 ± 510 cells/μl blood; P > 0.05, t  test). For measurement of opioid peptide content, PMN were isolated from circulating blood with a purity of 90 ± 1% (control) and 95 ± 1% (G-CSF+SCF) as determined by flow cytometric staining with the mAb RP-1. Cells were more than 95% viable by trypan blue exclusion. In the PMN fraction, β-endorphin content per cell was significantly decreased and Met-enkephalin content was unchanged in rats treated with G-CSF+SCF (fig. 2B; P < 0.001, P = 0.148, respectively, t  test).

In rats treated with G-CSF+SCF, the MFI of CD62L staining was significantly lower, whereas the MFI of CD49d was significantly higher on circulating PMN (figs. 3A and B; P < 0.05, P < 0.001, t  test). CD18 expression was unchanged (P > 0.05, t  test). The in vitro  cell migration was not influenced by G-CSF+SCF treatment at low MIP-2 concentrations but was significantly lower in rats treated with G-CSF+SCF than in controls at concentrations of 10⁻⁶to 10⁻⁸m (fig. 3C; P < 0.05, two-way ANOVA).

mRNA expression of all four chemokines significantly increased with the duration of inflammation (table 2; all P < 0.05, ANOVA). With the exception of keratinocyte-derived chemokine, this increase did not attain significance during the first 2 h after FCA inoculation.

Flow cytometric analysis of immune cell subpopulations in the inflamed paw showed that total hematopoietic (CD45⁺) cells and PMN were significantly increased in rats treated with G-CSF+SCF (figs. 4A and B; both P < 0.05, t  test; few CD45⁺cells were detected in the noninflamed contralateral paw, n = 3). G-CSF+SCF treatment was significantly less effective to mobilize PMN to the inflamed paw than to expand PMN in blood (1.4-fold vs.  11-fold;figs. 2A and 4A; P < 0.001, Mann–Whitney test). No changes occurred in the number of monocytes/macrophages or lymphocytes (figs. 4C and D; both P > 0.05, t  test). Opioid-containing cells were significantly increased by 1.5-fold in the inflamed paw of rats treated with G-CSF+SCF (fig. 5A; P < 0.05, t  test); the percentage of opioid-containing leukocytes did not differ among groups (control 46.8 ± 2.4%, G-CSF+SCF 50.2 ± 2.2% of CD45⁺cells, P > 0.05, t  test). Baseline PPTs in the inflamed or noninflamed paw were not significantly changed by hematopoietic growth factor treatment (inflamed: control 38.6 ± 3.2 g vs.  G-CSF+SCF 37.9 ± 3.6 g; noninflamed paw: control 64.5 ± 1.6 g vs.  G-CSF+SCF 67.9 ± 1.7 g; both P > 0.05, t  test). Both CRF injection and CWS lead to a significant PPT increase in the inflamed paw of both control rats and rats treated with G-CSF+SCF (figs. 5B and C; P < 0.05, t  test), which was not significantly different between rats treated with G-CSF–SCF and control rats (P > 0.05, t  test).

In this study, we examined whether endogenous peripheral opioid analgesia in early FCA inflammation can be augmented by systemic immunomodulation. We demonstrated that mobilization of PMN by the hematopoietic growth factors G-CSF and SCF induced a major increase in PMN in the circulation. Mobilized PMN had decreased β-endorphin but unchanged Met-enkephalin contents and minor changes in the expression pattern of the adhesion molecules CD62L (L-selectin), CD18 (β²integrin), and CD49d (VLA-4). The in vitro  migration ability of mobilized PMN was intact in response to low doses but impaired in response to high doses of the chemokine MIP-2. Chemokine mRNA transcription in the paw was hardly detectable in early inflammation. The in vivo  migration of opioid-containing leukocytes to the inflamed paw was slightly increased, and endogenous peripheral analgesia was unaltered.

Endogenous opioid-mediated peripheral analgesia has been demonstrated in early stages of an experimental inflammation 9,11as well as immediately after surgery in humans. 4In circulating blood of Wistar rats, lymphocytes constitute the predominant population (60–70% of all leukocytes, 20–30% PMN, 5–10% monocytes 24). In contrast, PMN are the major leukocyte and opioid-containing population in the inflamed paw during early inflammation (2 and 6 h after intraplantar FCA; of all leukocytes: 60–70% PMN, 20–25% monocytes/macrophages, < 5% lymphocytes 11). Importantly, because changes in opioid peptide plasma concentrations do not correlate with analgesia, 25,26an expansion of opioid-containing leukocytes in the blood will not predict the degree of pain control. In contrast, opioid-containing leukocytes in the inflamed paw induce peripheral opioid analgesia correlating with the number of opioid-containing cells. 8,11During early inflammation (2 h after intraplantar FCA), opioid-containing leukocytes are present in lower absolute numbers, and peripheral analgesia is significantly weaker than in late inflammation (96 h after intraplantar FCA), 11suggesting that analgesia could potentially be augmented by increasing migration of opioid-containing leukocytes to the inflamed paw. Previous studies have shown that recruitment of leukocytes by intraplantar injection of a single stimulatory agent (e.g. , leukotriene B4, formyl-methionyl-leucyl-phenylalanine, or nerve growth factor) induced pain that could be abolished by PMN depletion. 27–29On the contrary, in a model of peritonitis (i.e. , acetic acid–induced writhing), pain was enhanced in mice depleted of PMN but not of monocytes/macrophages, indicating that PMN conferred analgesia, whereas activation of resident cells (rather than PMN) enhanced inflammatory pain. 29,30Furthermore, PMN have been shown to contain opioid peptides such as β-endorphin and Met-enkephalin 5,31and to express CRF receptors. 32,33Activation of CRF receptors mediates opioid peptide release from leukocytes in vitro  34,35and peripheral opioid analgesia in vivo . 6,7 

In concurrence with previous studies, 15,16we have now demonstrated—using flow cytometric quantification—that combined G-CSF+SCF treatment was highly effective in increasing the number of circulating PMN (fig. 2A). Because the 3E7 mAb used for flow cytometry does not permit distinction between β-endorphin and Met-enkephalin, we measured opioid peptide content by radioimmunoassay in purified circulating PMN: cellular β-endorphin content was significantly decreased and Met-enkephalin content was unchanged in mobilized PMN (fig. 2B), which may be explained by observations that G-CSF+SCF treatment induces large numbers of immature PMN 15and has differential effects on granular protein content. 36To examine the potential of these cells to migrate to inflamed sites, we first characterized their expression of adhesion molecules. Migration of PMN requires rolling (mediated by selectins) and sticking (mediated by integrins). 20Blockage of CD62 (selectins) by fucoidin has been shown to reduce migration of opioid-containing leukocytes to the inflamed paw and to essentially abolish analgesia. 9Although previous studies have demonstrated that mobilization of circulating PMN by G-CSF is highly effective, whether these cells change the expression of adhesion molecules remains controversial. 37In addition, expression of adhesion molecules has not been studied in combined G-CSF+SCF treatment. This strategy mobilizes an even higher degree of immature PMN than G-CSF treatment alone, 15which might have a more pronounced effect on the adhesion molecule expression pattern. Similar to studies using G-SCF treatment in humans, 37we demonstrate here that CD62L expression was only reduced to a minor extent (by 25%;fig. 3B) in rats treated with G-CSF+SCF. This is unlikely to contribute to a reduced capacity to migrate because even in the complete absence of CD62L in genetically deficient mice, PMN migration was reduced by only 50%. 38The second stage of adhesion involves PMN sticking mediated by CD18 and CD49d expression. In the current study, mobilized PMN had normal CD18 and increased CD49d expression (fig. 3B), as has been shown in human PMN after G-CSF treatment. 39Taken together, adhesion molecule expression is slightly altered in PMN mobilized by G-CSF+SCF but seems adequate for migration of opioid-containing leukocytes to an inflammatory site. Therefore, enhancement of adhesion molecule expression would not be likely to further augment analgesia.

In addition to adhesion, PMN migration requires intact chemotaxis. A decreased chemotactic capacity of G-CSF–mobilized PMN has been demonstrated in the majority of previous studies. 36,37,40In rodents, recruitment is largely mediated by the chemokine receptor CXCR2 on PMN. 18,41,42Because the four known ligands of the CXCR2 receptor (MIP-2, keratinocyte-derived chemokine, lipopolysaccharide-induced chemokine, and cytokine-induced chemokine 2) 43,44all induce PMN recruitment at similar molar concentrations (10⁻⁸m), 45–47we examined in vitro  chemotaxis toward one representative CXCR2 ligand (MIP-2) in the current study. We observed a chemotactic defect of mobilized PMN at higher chemokine concentrations (≥ 10⁻⁸m;fig. 3C). The question arises whether these concentrations are physiologically relevant in vivo . Chemokines are produced within inflamed tissue, but they exert their chemotactic effects while being bound to glycosaminoglycans on the endothelium in the circulation. 17This concentration cannot be determined in vivo . In an inflammatory model in mice, chemokine concentrations were determined in an air pouch 2 h after injection of carrageenan, and a concentration of 10⁻⁸to 10⁻⁹m was measured. 48In accordance with this and other studies, 48,49we detected only very low levels of chemokine transcription at early stages of inflammation (1 and 2 h vs.  12 h after FCA;table 2) by PCR. Taken together, our results suggest that mobilized PMN show a moderately reduced capacity to migrate toward the chemokine MIP-2 at concentrations potentially relevant in vivo , and only one of four chemokines known to recruit PMN is significantly expressed in early inflammation. These data indicate that augmentation of chemokine expression in the inflamed paw may be a promising strategy to enhance recruitment of opioid-containing PMN and analgesia.

In contrast to our findings in peripheral blood (11-fold increase), PMN mobilization by G-CSF+SCF resulted only in a modest increase (1.4-fold increase) in the number of PMN in the inflamed paw, without significant changes in the number of monocytes/macrophages or of lymphocytes (fig. 4). Previous studies have shown that G-CSF treatment increased the number of circulating PMN (by 3- to 10-fold), 50–52but the effect on cell migration to inflamed tissue was variable: G-CSF treatment effectively increased PMN recruitment to the inflamed lung, 13,14,52but PMN recruitment was decreased to human skin or to the cerebrospinal fluid in experimental meningitis. 40,53In line with the modest increase of PMN in the inflamed paw by G-CSF+SCF treatment, the number of opioid-containing leukocytes increased to a minor extent (fig. 5A). Consistently, CWS- and CRF-induced analgesia were unchanged by PMN mobilization (figs. 5B and C). Because baseline hyperalgesia was unaltered, the slightly increased number of PMN at the site of inflammation in rats treated with G-CSF+SCF did not intensify pain.

Taken together, the lack of change in analgesia could be attributed to two factors: First, chemokine transcription during the early stage of inflammation is low (table 2), and mobilized PMN have an impaired ability to migrate (fig. 3C). This would explain the rather modest effect of PMN mobilization by G-CSF+SCF treatment in the paw compared to the circulation (figs. 2A and 4B). Second, although the number of opioid-containing (3E7⁺) cells modestly increased in the paw examined by flow cytometry (fig. 5A), this does not reflect opioid peptide content per cell. In circulating PMN, β-endorphin content per cell was significantly decreased. Because our previous studies in later stages of inflammation indicate that β-endorphin is the major peptide responsible for peripheral opioid analgesia, 8this may contribute to the net lack of effect on analgesia observed here. In addition, opioid-mediated analgesia in early inflammation might be limited by an insufficient permeability of the perineural barrier to opioid peptides. 54Finally, low levels of opioid receptor expression or G-protein coupling on the peripheral sensory nerve may account for the unchanged analgesic efficacy. 55,56 

In summary, our study examined an approach to augment peripheral endogenous analgesia. Treatment with hematopoietic growth factors was effective in expanding the number of circulating PMN, but these cells had a significantly decreased content of β-endorphin. In contrast to circulating blood, G-CSF+SCF increased the number of opioid-containing leukocytes at the site of an inflammation only to a minor extent, and it did not alter peripheral opioid-mediated analgesia. Future strategies to augment the recruitment of opioid-containing PMN and to enhance endogenous analgesia should be directed at increasing chemokine production but not at enhancing the expression of adhesion molecules.