| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Published online before print December 19, 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Infectious Diseases Division and Partners AIDS Research Center, Massachusetts General Hospital, Harvard Medical School, Boston;
Department of Physics, Massachusetts Institutes of Technology, Boston;
Cardiovascular Medicine Unit, National Heart and Lung Institute, Faculty of Medicine, Imperial College London, Hammersmith Hospital, United Kingdom;
Center for Engineering in Medicine and Surgical Services, Massachusetts General Hospital, Shriners Hospitals for Children, and Harvard Medical School, Boston; and
¶ Endocrine Unit, Massachusetts General Hospital, Boston
1 Correspondence: Infectious Diseases Division, Massachusetts General Hospital (East), 149 13th Street, Room 5212, Charlestown Navy Yard, Boston, MA 02129. E-mail: mpoznansky{at}partners.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: chemotaxis • microfluidics • gradient
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
It has recently been demonstrated that certain chemokines, including stromal cell-derived factor-1 (SDF-1; CXCL12), eotaxin-3 (CC chemokine ligand 26), and CXCR3 ligands can act as chemorepellents and chemoattractants [21 22 23 ]. Leukocyte chemorepulsion in this context was shown to mediated by chemokine receptors and was concentration-dependent and pertussis toxin (PTX)-sensitive. We and others have suggested that the active movement of leukocyte subpopulations away from a chemokine, which we termed fugetaxis, may play a significant role in the emigration of these cell types from the thymus or sites of inflammation or immune cell infiltration [21 22 24 25 26 27 ]. In view of this, we postulated that neutrophils may also be capable of undergoing chemorepulsion.
The mechanism by which a single chemokinetic agent can subserve roles as a chemoattractant and chemorepellent has been explored previously for bacteria and axonal growth cones. In bacteria, the directional bias of movement toward or away from an agent has been shown to be dependent on concentration of the chemokinetic agent, receptor occupancy, and the long-range cooperative interactions of receptors for the agent and intracytoplasmic molecules involved in signal integration and amplification [28 29 30 31 ]. Neuronal growth cones have similarly been shown to generate attractant and repellent responses to netrins and semaphorins in a manner that is dependent on gradient steepness and absolute concentration of the ligand within the gradient [32 33 34 ].
Neutrophils play a central role in the initiation and maintenance of the inflammatory response to tissue injury [35 ]. It is well established that neutrophils undergo chemoattraction or persistent, directionally biased movement toward a number of chemokinetic agents elaborated at sites of tissue injury, including the chemokine IL-8 [36 37 38 ]. Production of IL-8 by inflamed tissues and infiltrating inflammatory cells is known to result in the amplification of this signal and the rapid accumulation of neutrophils at a specific site of injury [39 40 41 ]. The question arises as to how an inflammatory response is down-regulated to prevent an overzealous reaction. Models have been proposed to explain this based on in vitro observations, including the repression of de novo IL-8 production, inhibition of chemotaxis, and chemotactic desensitization [42 43 44 45 ]. We propose that neutrophil chemorepulsion may contribute to the down-regulation of an inflammatory response.
In this study, we used microfabricated gradient generators in vitro and diffusive gradient generation in vivo to examine the hypothesis that neutrophils sensed IL-8 gradients, which could induce a chemorepellent response. We observed and quantified persistent, directionally biased migration of neutrophils up and down, defined IL-8 gradients and demonstrated dependencies on gradient steepness, the absolute concentration of chemokine within the gradient, the CXCR2 receptor, and established chemokine signaling and cellular motility pathways. Neutrophils undergoing chemorepulsion polarized down the gradient, and actin and Akt colocalized to the leading edge as in chemoattraction. Chemorepulsion, like chemoattraction was sensitive to inhibition by inhibitors of PI-3K and proteins involved in the maintenance of polarization and persistent movement including the targets of RhoGTPases such as p160-Rho-associated kinase (ROCK) and PKC
. A differential dependency of neutrophil chemorepulsion on PKC isoforms and the cyclic adenosine monophosphate (cAMP)/PKA axis, as compared with chemoattraction, was observed. Finally, neutrophils polarized and underwent fugetaxis from the CXCR2-binding IL-8 ortholog, cytokine-induced neutrophil chemoattractant-1 (CINC-1), in vivo in a model of mesenteric inflammation.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fabrication, preparation, and characterization of microfluidic linear gradient generators
The microfluidic linear gradient generators were fabricated in poly(dimethylsiloxane; Sylgard 184, Dow Corning, NY), and gradient generation was verified and calibrated with fluorescein isothiocyanate (Sigma-Aldrich, St. Louis, MO), as described previously [47
, 48
].
Microfluidic migration assay and time-lapse microscopy
Neutrophils at a concentration of 5 x 106 cells/ml were loaded uniformly across the migration channel (dimensions 4000 µmx450 µmx130 µm) and allowed to adhere within a microfluidic device for up to 30 min. On average, 1200 neutrophils were delivered to and adhered within the migration channel of the device. Under control conditions in which no chemokine was applied, 99% ± 0.5% of the originally applied neutrophils were observed to adhere within the device at the start of each experiment. Adherent neutrophils within the device were then exposed to the absence or presence of a linear gradient of human IL-8 (72 amino acids, PeproTech, Rocky Hill, NJ), as described previously [47
]. Migration was observed in a Nikon Eclipse TE2000-S microscope (Nikon, Japan). Brightfield images (10x) were taken every 30 s for 40 min using a digital camera (Hamamatsu, Japan) controlled by IPLab 3.6.1 (Scanalytics, Fairfax, VA). Cell movement and gradient verification were always observed at a set point along the migration channel. The number of neutrophils adhering within the device was also determined at the end of the experiment, after the neutrophils had migrated for 40 min. Secondary effects of chemokinetic agents secreted by cells during the course of the experiments were effectively ruled out previously, as a result of the flow rate of fluid through the device throughout the assay.
In certain experiments, cells were preincubated with PTX (100 ng/mL, 30 min at 37°C), GF109203X (25 mM for 30 min at 25°C), wortmannin (10 µM, 20 min at 37°C), LY294002 (50 µM, 15 min at 37°C), 8-Br-cAMP, 8-Br-cyclic guanosine monophosphate (cGMP; 100 µM or 1 mM, 15 min at 20°C, Sigma-Aldrich), or the nonpeptide CXCR2 antagonist SB225002 (1 pM, 100 pM, 1 nM, 10 nM, or 1 µM for 15 min at 37°C, Calbiochem, San Diego, CA) and then washed and loaded into the device. In certain experiments, Iscove’s modified Dulbecco’s medium (with 0.5% w/v fetal bovine serum) media were augmented with Y-27632 (100 nM, 10 µM), H-1152P (10 nM), a myristolated-PKC-pseudosubstrate inhibitor (25 µM), Rp-cAMPS (2 µM, 20 µM), or Rp-8-Br-cGMPS (2 µM, 20 µM, Calbiochem).
Immunostaining of neutrophils in microfluidic linear gradient generators
At predetermined time-points, cells migrating in gradients of IL-8 in microfluidic gradient generators were fixed with 3.7% (v/v) paraformaldehyde, permeablized with 0.1% (v/v) Triton X-100 in phosphate-buffered saline (PBS), and then blocked with 1% (w/v) bovine serum albumin in PBS. Samples were washed twice with PBS between each step. To visualize actin localization, samples were stained with phalloidin-AlexaFluor-594 (Molecular Probes, Eugene OR); to visualize Akt/PKB localization, samples were stained with a biotinylated anti-Akt/PKB monoclonal antibody (Cell Signaling Technology, Beverly, MA) and an antibiotin-AlexaFluor488 (Molecular Probes). Cells and stains were observed by epifluorescence and phase contrast microscopy as described above.
Intracellular calcium measurements and CXCR2 staining following IL-8
Calcium flux in primary neutrophils in response to IL-8 was measured as described previously [49
]. Calcium flux, in response to indicated concentrations of IL-8, was measured by fluorescence of cells excited at 340 and 380 nm using a PTI Deltascan dual-wavelength fluorimeter (Photon Technologies Inc., Lawrenceville, NJ), observed through a Nikon Diaphot 200 microscope (Melville, NY) and Sensys charge-coupled device camera (Photometrics, Ltd., Tucson, AZ) and was analyzed using the Poenie-Tsien ratio with Imagemaster 2 software (Photon Technologies Inc.) [50
]. Signaling responses to IL-8 were calibrated against maximal calcium flux following treatment with 5 µM ionomycin (Sigma-Aldrich). CXCR2 expression on neutrophils pre- and post-exposure to 120 nM, 1.2 µM, and 2.4 µM IL-8 was also determined using phycoerythrin (PE)-labeled anti-CXCR2 immunoglobulin G1 (IgG1) antibody (R&D Systems, Minneapolis, MN) or PE-labeled IgG1 isotype control antibody. Mean fluorescent intensity (MFI) was determined using FloJo software (Tree Star, San Carlos, CA). Statistical differences between MFI values were determined by Kolmogrov-Smirnoff test.
Intravital microscopy of rat mesentery
Male Sprague Dawley rats (200–300 g) were prepared for mesenteric intravital microscopy and subsequent leukocyte adhesion and transmigration as described previously [51
]. Videos of migrating cells were constructed for quantitation and mathematical analysis as described above; at the end of certain in vivo experiments, the mesentery was stained with acridine orange (Sigma-Aldrich), a nuclear dye, scanned with a 488-nm laser line, generated from an Argon laser, and observed by confocal microscopy (LSM5 PASCAL, Axioskop II FS, Carl Zeiss, Thornwood, NY) to verify that migrating cells were neutrophils. Neutrophil migration was only recorded in mesentery in which there was no evidence of activation prior to the application of exogenous CINC-1 (PeproTech). Chemokine was added at a steady rate onto the mesenteric tissue throughout each experiment to establish a stable gradient and to dilute out any chemokines released de novo from mesenteric tissue.
Mathematical analysis of cell migration in linear gradient generators
Cell movement was tracked using MetaMorph 4.5 (Universal Imaging, Downington, PA) object-tracking application, which generated tables of Cartesian coordinate data for each migratory step during the migration of a cell. Experiments were run in triplicate, and consequently, three time-lapse videos were analyzed for each control and experimental condition. Each time-lapse video was divided into three 150 µm sections, corresponding to the lower, middle, and upper ranges of each gradient, and equal proportions of randomly chosen cells in each section were tracked. On average, 50 cells were tracked for each experimental and control condition, and more than 4000 migratory steps were analyzed for each set of tracked cells. Tracking data were analyzed in Excel (Microsoft, Redmond, WA) and MATLAB 13 (The Mathworks, Inc., Newton, MA) to determine angular frequencies, mean squared displacements, migration velocities, persistence times, random motility coefficients, and random walk path lengths; cell movement was characterized based on a persistent random walk model as follows [20
, 52
]: For each cell, the squared displacement R2(t) was calculated at time interval t, R2(t) = [x(t0+t)–x(t0)]2 + [y(t0+t)–y(t0)]2, 1, where t0 is the time at the origin. The origin was shifted along the dataset, and the displacements were averaged for overlapping time intervals. A global average was performed over all cells in the set to calculate the mean squared displacement. Mathematically modeling cell movement as a correlated, biased random walk, this can be written as R2(t) = 2S2P[t–P(1–e–t/P)], 2, where S and P are measures of the mean speed of movement and persistence time, respectively. When time interval t is much greater than persistence time P, the mean squared displacement becomes linearly proportional to t, analogous to Brownian diffusion R2(t) = 2S2Pt = 4µ t, 3, where µ is the motility coefficient. The slope and intercept of a least squares regression fitted to the linear section of the mean squared displacement give an estimate of µ and P, respectively.
Further, a "persistence index" (PI) of the motion or mean free path was calculated as the total displacement of the cell divided by the total path length. The PI is an indicator of turning behavior, and 1 indicates motion in a straight line, and 0 indicates no net displacement. To quantify directional bias of neutrophil migration in defined IL-8 gradients in microfabricated devices, we calculated a "chemotropism index" (CI) based on McCutcheon [52] and Nossal and Zigmond [53
] and defined as the net path length traversed by a cell with respect to the direction of the established gradient divided by the total distance traveled and modified to include a measurement of directionality toward or away from the maximal chemokine concentration: CI= 
li cos 
i/li, 4, where li is the length of a cell’s movement vector, and i is the angle the movement vector makes with respect to the established gradient. The CI is therefore an indicator of accuracy of orientation with respect to the gradient and will be 1 if a cell moved directly up the gradient, 0 if there is no gradient bias, and –1 for migration directly down the gradient. The index was calculated for each cell and then averaged over the population of migrating cells to obtain the "mean chemotropism index" (MCI).
Mathematical modeling of continuous gradients in vivo
The steady-state chemokine concentration profile in the mesentery was predicted based on classical diffusion equations applied to a spherical model of the postcapillary venule and the assumption that the receptor-dependent transport of the chemokine by the endothelial cells is the main mechanism for generating the gradient in the vicinity of postcapillary venules [54
55
56
]. This model also assumes that all chemokine binding sites on matrix proteins are saturated. Thus, the steady-state concentration C at a distance r from the postcapillary venule wall was calculated, considering constant concentration of the chemokine far from the venule and constant flux across the endothelium, as: C(r) = C0 – F0a2/D(r+a), where C0 is the chemokine concentration in the perfusion solution, a the vessel radius, F0 the rate of chemokine uptake, and D the diffusion coefficient. The rate of chemokine uptake by the endothelial cells was estimated in the range of 1000–10,000 molecules/cell/min by comparison with endocytosis rates for other proteins [57
]. A value of 0.6 x 10–7 cm2/s for diffusion coefficient of the CINC-1 [molecular weight (MW) 7800] in the mesentery was interpolated from the diffusion coefficient of albumin (MW 66,000), determined experimentally in similar tissues [58
].
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
than those that elicited T cell chemoattraction, we examined IL-8 gradients with peak concentrations of 12 nM, 120 nM, 1.2 µM, and 2.4 µM. Neutrophil migration in the device was tracked, and angular frequencies of all directional movements were quantified (Fig. 1a
1b
1c
1d
1e
1f
1g
1h
; Supplemental Videos 1–4; see below for full quantification of motility). Shear forces were calculated to be less than 0.1 dyne/cm2, which is below shear forces experienced in blood vessels [47
, 59
], and cells exposed to no chemokine or uniform 120 nM IL-8 underwent chemokinesis, demonstrating an even distribution of angular frequency (Fig. 1
b and data not shown). In contrast, cells placed in gradients between 0 and 12 nM or 120 nM underwent robust chemoattraction from all starting positions within the gradient and demonstrated distribution of angular frequencies skewed toward the peak chemokine concentration (data not shown, Fig. 1c
and d, and Video 2, respectively).
|
The majority of cells in all control and experimental conditions described above adhered within the device after their initial introduction and remained adherent within the device throughout the 40-min video-recording period and in particular, in those IL-8 gradients in which robust chemoattraction (0–120 nM) or chemorepulsion (600 nM–1.2 µM) was seen (Table 1 ). There were no significant differences in the proportion of cells adhering in devices in control conditions in the absence of chemokine in comparison with conditions in which cells were exposed to these IL-8 gradients, which induced robust chemoattraction and chemorepulsion (Table 1) . There was a trend toward reduced T cell adhesion in the 0–12-nM IL-8 gradient in which cells predominantly underwent chemoattraction, but this did not reach significance (Table 1) . When neutrophils were exposed to a steeper, 0–2.4-µM IL-8 gradient, cells showed reduced adhesion, and only the minority of cells (10.8±2.9%) remained adherent within the migration channel. The few adherent cells under this condition moved randomly (data not shown).
|
Morphology and localization of cytoskeletal and signaling components during neutrophil chemorepulsion
Neutrophils undergoing chemoattraction in a 0–120-nM gradient of IL-8 or chemorepulsion in a 0.6–1.2-µM gradient of the chemokine demonstrated a persistently polarized phenotype (Fig. 2a
and 2b
, respectively). Immunohistochemical staining of polarized neutrophils with known migratory histories and undergoing chemoattraction or chemorepulsion within an IL-8 gradient demonstrated actin and Akt colocalization at their leading edges; cells exposed to no chemokine were unpolarized and had no significant colocalization of actin or Akt (Fig. 2 c, column 1); cells undergoing chemotaxis to a 0–120-nM IL-8 gradient or undergoing fugetaxis in a 0.6- to 1.2-µM IL-8 gradient were polarized, and actin and Akt colocalized to the leading edge regardless of orientation bias (Fig. 2c
, columns 2 and 3, respectively).
|
The CI for each cell and MCI for all cells migrating in the gradient or section of a gradient were quantified and measured the magnitude and accuracy of the orientation bias of movement toward (+) or away (–) from IL-8 (see Mathematical Methods) [20 , 52 , 53 ]. Cells exposed to no chemokine or a uniform concentration of 120 nM IL-8 demonstrated no directional orientation bias (MCI=–0.02±0.010 and –0.00±0.022, respectively; Table 1 and Fig. 3 a ). Movements of cells exposed to 0–12 nM or 0–120 nM IL-8 gradients were highly biased toward increasing concentration of chemokine along the gradient axis (MCI0.32±0.033 and +0.39±0.025, respectively; Table 1 and Fig. 3a ). In contrast, cells exposed to a steeper gradient of 0–1.2 µM demonstrated a MCI of –0.13 ± 0.024, supporting the observation that the predominant bias of cellular movement was away from the peak concentration of IL-8; cells exposed to a shallower gradient of 600 nM–1.2 µM IL-8 underwent robust chemorepulsion down the gradient (MCI=–0.42±0.042; Table 1 and Fig. 3a ). Division of each gradient into three equal segments and separate analysis of cell populations commencing movement in each segment demonstrated that neutrophils undergo chemotaxis or fugetaxis in a concentration-dependent manner within the same gradient. Cells migrating in all sections of the 0–12-nM and 0- to 120-nM gradient demonstrated positive MCIs between +0.20 and +0.44; cells migrating in the lower segment of the 0–1.2-µM gradient had a MCI of +0.2; and cells in the middle third and upper third of the gradient had MCIs of –0.14 and –0.22, respectively. Cells in all segments of the 600-nM to 1.2-µM gradient demonstrated negative MCIs between –0.29 and –0.57 (Table 1) . These data indicate that neutrophil fugetaxis, like chemoattraction, is persistent, directionally biased movement.
|
Chemorepellent concentrations of IL-8 elicit differential levels of calcium flux
Ionic calcium serves as a secondary messenger in chemokine signaling, and magnitude of intracellular calcium flux serves as an indicator of GPCR-mediated signal transduction [63
]. We examined whether signaling output could be generated by chemoattractant (120 nM) and chemorepellent (1.2 µM) concentrations of IL-8. Peak calcium flux responses were almost twofold greater for 1.2 µM (83%±7.9, percentage of maximal calcium flux) than for 120 nM IL-8 (48%±5.1; P<0.05; data not shown). Previous reports had revealed maximal calcium fluxes up to 100 nM IL-8 without exploring the effect of higher concentrations of the chemokine [62
]. Differential levels of calcium flux to high and low concentrations of a chemokine have also been observed previously for SDF-1 (CXCL12) [64
]. To replicate conditions within a stable chemokine gradient, we examined whether signal output, as measured by calcium flux, could be generated from sequential low or high doses of IL-8. Sequential calcium fluxes were seen in cells exposed to repeated pulses of low-dose IL-8 (120 nM), which had been seen to induce maximal chemoattraction (Fig. 4a
), and high-dose (1.2 µM) IL-8 (Fig. 4b)
and a dose of 600 nM followed by a dose of 1.2 µM (Fig. 4c)
, which were associated with a maximal chemorepellent response. Sequential calcium flux was not seen when cells were exposed to the highest dose of 2.4 µM, and when this was followed by a further high concentration of chemokine, implying that receptor saturation occurred above 1.2 µM (Fig. 4d)
. We also examined whether CXCR2 was still detectable on the surface of the neutrophil after the first pulse of chemokine. Neutrophils were immunostained for CXCR2 following a single pulse of IL-8 at concentrations of 12 nM, 120 nM, 1.2 µM, and 2.4 µM. CXCR2 was down-regulated from baseline expression levels immediately after chemokine exposure, with chemoattractant and chemorepellent concentrations of IL-8 with consistent two- to threefold reductions in MFIs (data not shown), and remained four- to fivefold greater than MFI values for control isotype antibody staining (P<0.01; data not shown). CXCR2 was clearly still detectable on the cell surface following exposure to low or high concentrations of IL-8. These data are consistent with our finding that neutrophils were capable of gradient sensing and directional migration up to a peak concentration of 1.2 µM.
|
i-independent chemotaxis pathways [8
]. Preincubation of cells with 10 µM wortmannin resulted in 97% inhibition of directional movement in a 0– 120-nM IL-8 gradient (P<<0.0001; Fig. 3c
), and persistent movement was reduced (PI=0.30). However, in a 0- to 1.2-µM IL-8 gradient, the directional bias of neutrophil migration, preincubated with 10 µM wortmannin or 50 µM LY294002, was converted to a chemoattractant response in all regions of the gradient (Fig. 3c)
. Reduced neutrophil adhesion was seen within devices in the presence of PTX or PI-3K inhibitors, but these reductions did not reach statistical significance (data not shown). In summary, these data demonstrate that for neutrophils, Gi and PI-3K-dependent and -independent chemokine signal transduction pathways can be activated differentially when cells are exposed to varied relative and absolute concentrations of IL-8 and can convert the directional bias of the migratory response. Given these results in addition to the ubiquitous and necessary role of PI-3K in gradient sensing, polarization, and recruitment of Akt to the leading edge, coupled with recent data suggesting that classical PI-3K inhibitors are relatively nonspecific [65
, 66
], we hypothesized that control of chemorepulsion was further downstream. Inhibition of the Rho-dependent protein kinase p160-ROCK, a downstream component of the pathway governing RhoA action at the trailing edge by Y-27632 [inhibition constant (Ki)=140 nM], has been shown to inhibit maintenance of polarity and directed cell motility in neutrophil chemotaxis and axonal growth cone guidance [16 , 67 ]. Media augmentation with Y-27632 at 10 µM in a 0–120-µM IL-8 gradient demonstrated a 35% inhibition in chemoattraction (P<<0.0001; Fig. 3c ) and reduced persistent motion (PI=0.39). Addition of 100 nM Y-27632 to a 600 nM–1.2-µM IL-8 gradient resulted in a 51% reduction in neutrophil fugetaxis (P=0.0016) but did not affect the cell’s ability to maintain polarized, persistent movement (PI=0.49; Fig. 3c ), and augmentation with 10 µM Y-27632 impaired the cell’s ability to maintain persistent movement (PI=0.25) and inhibited directional movements by 94% (P<<0.0001; Fig. 3c ). Augmentation of the media within the device with H-1152P, a second p160-ROCK inhibitor (Ki=1.6 nM) at 10 nM, inhibited 90% of directional migration (P<<0.0001) with respect to the IL-8 gradient and slightly inhibited persistent movement (PI=0.38; Fig. 3c ). This perturbation of the RhoA pathway, via inhibition of p160-ROCK, demonstrated a dose-dependent inhibition of directed chemorepulsion followed by inhibition of stabilized polarity.
Adhesion of neutrophils within the microfluidic device was reduced significantly by pretreatment of cells with the p-ROCK inhibitor, H-1152P, as compared with untreated cells (P=0.046; Student’s t-test). In this case, the minority of neutrophils (43±8%), which were originally introduced into the device, migrated and remained adherent within the device following 40 min of exposure to an IL-8 gradient. Therefore, data in Figure 3c reflect the migratory behaviors of a minority of neutrophils exposed to this inhibitor.
PKC isoforms and intracytoplasmic cAMP concentrations mediate chemorepulsion
Upstream of or in parallel with the classical chemotaxis effectors, GPCR signaling relies heavily on ubiquitous secondary messengers such as cytosolic calcium and cyclic nucleotides [68
]. It has previously been shown that the magnitude of intracytoplasmic calcium flux is associated with the degree of activity of calcium-dependent signal transduction molecules including certain isoforms of PKC [69
, 70
]. In addition increased intracytoplasmic calcium has been shown to be correlated with PKC activation and triggering of nerve growth cone turning [71
, 72
]. As calcium-dependent PKCs have been shown to be involved in directional cell migration [72
73
74
75
76
], we examined whether the higher level of calcium flux associated with chemorepellent concentrations of IL-8 was associated with increased involvement of signaling via calcium-dependent PKC isoforms. We found that GF109203X, a PKC inhibitor which reduces the filling state of intracellular Ins (1, 4, 5) P3 sensitive Ca2+ stores by inhibiting Ca2+ uptake into these stores), mildly inhibited chemoattraction in a 0–120-nM IL-8 gradient (82% of normal response; P<<0.0001) and converted a chemorepellent response into a chemoattractant response in a 0–1.2-µM IL-8 gradient (Fig. 3d)
. Inhibition of PKC, a putative downstream component of Cdc42 action at the leading edge, has been shown to block integrin-mediated adhesion, directed movement, and F-actin accumulation [77
]. Media augmentation with a cell-permeable, myristolated PKC pseudosubstrate at 10 µM inhibited all motion with respect to a 0–120-nM IL-8 gradient (Fig. 3d) , and augmentation in the same gradient range with the inhibitor at 25 µM induced apoptosis (data not shown). Addition of 25 µM PKC pseudosubstrate to a 600-nM–1.2-µM IL-8 gradient also resulted in 100% inhibition of directed motion (P<<0.0001; Fig. 3d
), although some undirected, persistent motion was observed in the higher gradient range (PI=0.22).
In light of the confluence between signaling for directional leukocyte migration and axon growth cone guidance, we examined the effect of cell membrane-permeable cyclic nucleotide analogs on the bias of neutrophils migrating up or down a 0–1.2-µM IL-8 gradient [78 , 79 ]. Furthermore, we have previously shown that T cell fugetaxis is inhibited selectively by a cAMP analog, and perturbation of intracytoplasmic cAMP levels could convert a chemorepellent to a chemoattractant response in this system [21 , 25 ]. Preincubation of neutrophils with 100 µM 8-Br-cAMP or 100 µM 8-Br-cGMP and subsequent exposure to a 0–1.2-µM IL-8 gradient converted the fugetactic responses to robust, persistent chemotactic migration (MCI=0.40±0.037 and 0.38±0.048; PI=0.56 and 0.62, respectively; Fig. 3d ) by artificially stimulating the actions of and signal flux through the cyclic nucleotide-dependent protein kinases PKA and PKG, processes that have been implicated in regulating signal flow to discrete portions of the signal transduction network [28 , 29 , 68 , 80 ]. Increasing the preincubation dose of 8-Br-cAMP to 1 mM also converted chemorepulsion to chemoattraction (MCI 0.24±0.023; PI=0.44), whereas preincubation with 1 mM 8-Br-cGMP resulted in the inhibition of directional responses (MCI 0.10±0.034; Fig. 3d ) without abrogation of persistent motion (PI=0.41).
Addition of thiolated cyclic nucleotide analogs to the assay media served to inhibit the actions of and signal flux through PKA and PKG. Rp-cAMPS, a cell-permeable, nonhydrolyzable inhibitor of PKA (Ki=11 µM) at 2 µM, abrogated directional movement of neutrophils in response to 0–120 nM and 0.6–1.2 µM IL-8 gradients (80% and 97% inhibition, P<<0.0001, and P<<0.0001, respectively), and addition of 20 µM Rp-cAMPS to a 0.6–1.2-µM IL-8 gradient resulted in 100% inhibition of directional motion (P<<0.0001; Fig. 3d ); inhibition of PKA demonstrated an attenuation of persistent motion (PI=0.22, 0.31, 0.18, respectively). Addition of 2 µM Rp-8-Br-cGMPS, a cell-permeable, nonhydrolyzable inhibitor of PKG (Ki=4 µM) to the assay media, resulted in an inhibition of directional movement of neutrophils in response to a 0–120-nM or 0.6–1.2-µM IL-8 gradient (94% and 96% inhibition, P<<0.0001, and P<<0.0001, respectively). Addition of 20 µM Rp-8-Br-cGMPS to a 0.6–1.2-µM IL-8 gradient resulted in 77% inhibition of chemorepulsion (P<<0.0001; Fig. 3d ), all with similar reduction of persistent motion (PI=0.23, 0.26, and 0.31, respectively). These data support the thesis that perturbation of secondary messenger signaling axes, whether associated with intracytoplasmic calcium, cyclic nucleotides, and some of their respective target proteins, can differentially affect control of the directional bias of neutrophil migration under attractive and repulsive conditions in a manner similar to that observed for axon guidance.
Persistent neutrophil chemorepulsion in diffusive chemokine CINC-1 gradients in vivo
Having demonstrated robust neutrophil chemorepulsion in vitro, we explored whether neutrophil fugetaxis could be observed and quantified in vivo. We examined neutrophil migratory responses to CINC-1, a rat IL-8 ortholog and CXCR2-binding chemokine, in an established rat model of inflammation in which vital mesentery is exteriorized, and neutrophil migration is observed and quantified in the context of chemokine or cytokine gradients [43
, 51
, 81
82
83
84
85
]. Two types of experiments were established in this model. First, mesenteric tissue was superfused with Tyrode’s-balanced salt solution or CINC-1 dissolved in Tyrode’s at concentrations of 1 nM, 10 nM, or 100 nM for 90–120 min at a fixed rate. This concentration range of CINC-1 was chosen, as it was thought to be physiologically relevant in view of published measurements of tissue concentrations of CINC-1 and IL-8 in settings of acute inflammation [86
, 87
]. Neutrophil recruitment and migration were recorded by time-lapse video microscopy and migrating neutrophils positively identified and counted following acridine orange staining of the mesentery. Neutrophil adhesion, transendothelial migration, and chemotaxis into the perivascular tissues, away from the postcapillary venule, occurred toward peak concentrations of 10 nM CINC-1, and concentrations of 100 nM led to similar adhesion, reduced transendothelial migration, and accumulation of neutrophils in the perivascular tissue immediately adjacent to the postcapillary venule (Fig. 5a
and b, and see below). Concentrations of Tyrode’s buffer alone or 1 nM CINC-1 led to minimal neutrophil adhesion to the luminal surface of the venule and transendothelial migration (Fig. 5
, a and b, and data not shown).
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The directional decision of primary neutrophils appeared to be dependent on the steepness of the chemokine gradient and the absolute concentration of the chemokine. These findings are consistent with those made in grasshopper neuronal growth cone attraction or repulsion in which gradient steepness and absolute concentration of the semaphorin, Sema 2a, determine directionality of pathfinding [32 ]. We also observed that IL-8, at concentrations that were significantly higher than the documented dissociation constant (Kd) of CXCR2, could be interpreted as a chemorepellent signal as a result of increased but not saturating levels of CXCR2 receptor occupancy in a process that was exquisitely sensitive to antagonism by a specific CXCR2 antagonist. Furthermore, we showed that this level of receptor occupancy was capable of generating a differential calcium flux in comparison with lower chemoattractant concentrations of IL-8. Although CXCR2 was clearly down-regulated by IL-8 at chemorepellent concentrations, the receptor was still detectable on the cell surface, and cells remained responsive to subsequent pulses of chemorepellent concentrations of the chemokine. Our data support the view that primary human neutrophils are able to sense, process, and migrate in response to chemokine signal input across at least two orders of magnitude of chemokine concentrations and that a chemoattractant at high concentration can function as a chemorepellent in a manner consistent with the way in which certain ligands can induce chemoattraction and chemorepulsion in a concentration- and gradient-dependent manner in bacteria and neuronal growth cones [28 , 29 , 32 33 34 35 36 ].
The demonstration that neutrophils undergo chemoattraction at low chemokine concentrations but are repelled by higher concentrations of the same ligand supports a probabilistic model for GPCR occupancy and signaling, which goes beyond a simple "orthosteric" interaction between ligand and receptor [88 , 89 ]. This model proposes that GPCR can exist in a number of different activation states, dependent or independent of ligand binding [26 , 88 , 89 ]. The model takes into account rapid GPCR recycling, dimerization of chemokines such as IL-8 at concentrations above 100 nM, multiple binding sites of differing affinities for chemokines on their cognate GPCR, of which there are two for IL-8 on CXCR2 [61 ], the previously demonstrated oligo or homodimerization of GPCR such as CXCR2 with receptor and nonreceptor proteins [90 ], and other allosteric mechanisms. This model suggests that recombinant or pharmacological systems for studying GPCR behavior including measurements of Kd may not reflect the action of ligand binding to the wild-type receptor on primary cells under physiological conditions.
We demonstrated that primary neutrophil chemorepulsion, like chemoattraction to IL-8, is dependent on CXCR2 and conserved GPCR signaling pathways as evidenced by its sensitivity to inhibition by PTX, wortmannin, and LY294002. It is not surprising that neutrophil chemorepulsion, like chemoattraction, involves the activity of certain RhoGTPases, which are known to mediate cytoskeletal rearrangement, polarization, maintenance, and polymerization of actin at the leading edge. In addition, we demonstrated that Akt, a part of the PI-3K signal transduction pathway, colocalizes within the leading edge of chemotaxing and fugetaxing cells, although unlike in chemotaxis, the leading edge of the fugetaxing neutrophil is directed away from the peak concentration of chemokine. However, there were signaling elements that differentiated neutrophil chemorepulsion from chemoattraction. We demonstrated that higher concentrations of the chemokine and therefore, higher levels of CXCR2 receptor occupancy by IL-8 are required for chemorepulsion as compared with chemoattraction, and this finding was reflected in the exquisite sensitivity of chemorepulsion to inhibition and conversion to a chemoattractant response by picomolar concentrations of the specific CXCR2 antagonist, SB225002. This resembles the effect that picomolar concentrations of the specific opioid receptor antagonist, naloxone, are known to have on an opioid-binding GPCR. In this case, ultra-low concentrations of the antagonist, which potentially occupy less than 1% of receptors, result in an excitatory response, whereas high concentrations are inhibitory at opioid receptors on neurons [91 , 92 ]. High concentrations of IL-8 evoking a chemorepellent response were associated with an increased amplitude of calcium flux and sensitivity to inhibition by a PKC inhibitor, as compared with lower chemoattractant concentrations of the chemokine. This finding is consistent with recently published work demonstrating the direct involvement of calcium-dependent signaling in neuronal growth cone chemoattraction and chemorepulsion [72 , 93 ] and the previously described association of increased calcium with preferential activation of calcium-dependent signaling pathways including PKC and phospholipase C [69 , 70 , 94 ].
We also demonstrated that a directionally biased motile response can be converted from a chemorepellent to a chemoattractant response as a result of the perturbation of intracytoplasmic cAMP levels, in some cases without inhibiting persistent motion. Taken together, these findings resemble the direct effects that intracytoplasmic cyclic nucleotide and calcium concentrations have been shown to have on converting an axon growth cone chemorepulsive response to chemoattractive responses to netrins and SDF-1 using discrete pathways [72
, 78
, 95
, 96
]. The modulation of directional bias by intracellular calcium and cyclic nucleotides suggests a role in the secondary messenger amplification for gradient sensing in neutrophils, which has been described previously for axon guidance [71
, 72
, 78
, 96
97
98
], and reveals the need to expand current models of gradient perception to include a low gain threshold at which the directional decision is made by the neutrophil to move away from the peak chemokine concentration. Involvement of secondary messengers may also explain, in part, the classical "bell-shaped" chemotactic response, which describes how chemoattraction of leukocytes to increasing concentrations of a chemokine ultimately declines with increasing concentrations of a chemokine [15
]. We propose that beyond this signaling threshold, higher concentrations of chemokine generate a subsequent peak of chemorepellent activity, which itself declines when chemokine concentrations saturate and desensitize the directional responsiveness of the neutrophil. The mechanism by which the calcium/PKC and cAMP/PKA axes may differentially affect neutrophil chemorepulsion and how a chemokine G-protein receptor signaling network controls leukocyte chemoattraction and chemorepulsion clearly warrants further study.
Finally, we demonstrated for the first time robust neutrophil chemorepulsion to the IL-8 ortholog CINC-1 in vivo. Like neutrophil chemoattraction in vivo, neutrophil chemorepulsion involved cell polarization and resulted in persistent, directionally biased movement down a gradient of the chemokine. We speculate that neutrophil chemorepulsion may play a role in the homeostatic control of inflammation and believe that our in vitro reductionist approach and studies of neutrophil migration in vivo support this view. Published studies of the in vivo effects of IL-8 have examined its activity as a potent neutrophil chemoattractant, in which a single application of the chemokine was observed to result in the transient migratory response of neutrophils toward a specific site including the mesentery or joint space [99 100 101 102 ]. However, these studies were performed without modeling the chemokine gradient or examining temporal changes in the applied absolute chemokine concentration. The current paradigm would suggest that a "shallow" gradient of IL-8, established at the onset of inflammation, results in the recruitment of neutrophils to that site through chemoattraction. Our data support a bimodal action of IL-8 and an addition to this model that makes the prediction that as IL-8 production increases at a site of inflammation, the absolute concentration of the chemokine rises until neutrophils undergo fugetaxis from the site of inflammation. The "steeper" gradient may ultimately result in the exclusion of neutrophils from the site, as the inflammatory response transitions from a neutrophil predominant to a mononuclear cell infiltrate. The current paradigm argues that neutrophils only enter tissues as a result of chemoattractants, and our findings support the hypothesis that neutrophil chemorepellents or fugetaxins also exist and under certain conditions, may exclude this cell type from specific anatomic sites in vivo.
|
ACKNOWLEDGEMENTS |
|---|
Received September 13, 2005; revised October 25, 2005; accepted November 3, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and its lipid products Trends Cell Biol 10,466-473[CrossRef][Medline]i is not required for chemotaxis mediated by Gi-coupled receptors J. Biol. Chem. 274,2824-2828
