Non‐Resolving Aspects of Acute Inflammation after Spinal Cord Injury (SCI): Indices and Resolution Plateau

Abstract

Inflammatory resolution is an active, highly regulated process already encoded at the onset of inflammation and required to prevent the transition into chronic inflammation associated with spreading of tissue injury and exacerbated scarring. We introduce objective, quantitative measurements [resolution indices (R_i) and resolution plateau (R_P)] to characterize inflammatory resolution and to determine the persistence (“dwell time”) of differential leukocyte subpopulations at the lesion site after acute experimental spinal cord injury (SCI). The cell type‐specific resolution interval R_i (time between maximum cell numbers and the point when they are reduced to 50%) ranges from 1.2 days for neutrophils, 1.5 days for T lymphocytes, to 55 days for microglia/macrophages. As the resolution interval neglects exiting cell trafficking in the later period of resolution (49%–0% of lesional cells), we introduced the R_P, a marker for the persisting, chronified leukocyte subsets, which are likely to participate in late degeneration and non‐resolving inflammation. Here, we identify the acute inflammatory response in central nervous system (CNS) lesions as partly non self‐limiting. Both extended resolution intervals (reduced leukocyte clearance) and elevated plateaus (permanent lesional cell numbers) provide quantitative measures to characterize residual, sustained inflammation and define cognate timeframes of impaired resolution after acute SCI.

INTRODUCTION

It is well established that inflammation in the central nervous system (CNS) follows a distinct course compared with other organs as being “shielded” by an immune privilege 13, 25. The immune privilege, however, also includes the leukocyte exit and/or their clearance out of the inflammatory CNS, which in contrast to other organs lacks lymphatic drainage (13). Acute inflammation is a highly regulated process characterized by adhesion molecule upregulation, vasoactive mediator expression, leukocyte diapedesis, phagocytosis and release of chemical mediators such as eicosanoids, prostaglandins and complement components. As acute inflammation could perpetuate and result in spreading, chronicity and tissue injury, a complete resolution is required. While there is ample information available about the recruitment of leukocyte entry into the CNS, little is known about their exposure time to the lesioned CNS parenchyma and their exit.

Likewise, resolution is also a finely orchestrated, active process, which is distinct from anti‐inflammation as the inflammatory stimuli not just fizzle out 2, 14, 43, 44. Resolution is already encoded at the onset of inflammation and includes suppression of further leukocyte influx and clearance of inflammatory cells, suppression of pro‐inflammatory factors and release of pro‐resolving factors 14, 44. Complete resolution should eventually lead to restoration of tissue homeostasis 14, 44. For this, the resolution phase at a histological level is defined as the interval from maximum cell infiltration to the point when they are lost from the tissue (44). Macrophages could persist for longer periods after inflammation and might participate in tissue repair; however, they also finally disappear from the lesion in order to bring about complete resolution 21, 44.

Increasing appreciation of the independent mechanisms of resolution has led to interest in developing drugs (14) that support endogenous resolution or directly act as pro‐resolving agents, and first assays have been suggested to test the potency of drugs in order to drive inflammation down a pro‐resolving pathway (29). These efforts resulted in the need to introduce quantitative indices that precisely describe the changes in cell influx into and efflux from the tissue (44). The main events in resolution (Figure 1) can be quantified by resolution indices consisting of Ψ_max (the maximum number of infiltrating cells), T_max (the time point when cell numbers reach maximum) and the resolution interval R_i (the time between the maximum and the point when cell numbers are reduced by 50%) (45). In particular, changes in the resolution interval are generally used to describe the beneficial effects of pro‐resolving drugs as this index reflects the early clearance of apoptotic neutrophils and macrophages 29, 41. Following these lines, “resolution impeding” effects of widely used drugs can also be revealed, for example in case of selective cyclooxygenase‐2 (COX‐2) blockers, which may cause a “resolution deficit”(41).

Figure 1.

Indices that describe the resolution of inflammation. Based on cell trafficking into and away from inflammatory lesions, integrative indices were established to quantitatively determine the main events of resolution. These include Ψ_max (the maximum number of infiltrating cells), T_max (the time point when cell numbers reach maximum) and the resolution interval R_i (the time between the maximum and the point when cells numbers are reduced by 50%).

However, focusing only on the resolution interval neglects cell trafficking subsequent to clearance of the first half of cells from the lesion site. This later period might be independent of the resolution interval. In contrast, it might be rather dependent on whether resolution is complete (returning to sham levels) or incomplete. Moreover, it is likely that cell recruitment and resolution substantially vary between different organs and lesion etiologies (29).

Here we define operative and quantitative resolution indices (R_i) complemented by resolution plateaus (R_p) in order to provide objective parameters to analyze the efficacy of intrinsic inflammatory resolution programs in an immune‐privileged organ after spinal cord injury (SCI).

MATERIALS AND METHODS

Animal preparation

Male Lewis rats (Charles River, Sulzfeld, Germany), 8 to 12 weeks old, weighing between 220 and 280 g, were randomly assigned to groups subjected to SCI and perfused 1, 3, 7, 14 days, 4 or 10 weeks later (five rats/group). Control (sham‐injured) rats underwent bilaminectomy (complete removal of the dorsal arch of the vertebrae: processus spinosus and bilateral lamina arcus vertebrae) without injuring the spinal cord (three animals). All rats were kept under controlled conditions of light and temperature, had free access to food, and surgical and post‐operative care were approved and in accordance with the German guidelines for animal research.

Spinal cord lesions

Rats were anesthetized by intraperitoneal injection of ketamine hydrochloride (Ketanest, Parke Davis, Berlin, Germany; 100 mg/kg) and xylazine hydrochloride (Rompun, Bayer, Leverkusen, Germany; 10 mg/kg). To prevent xerophthalmia during anesthesia, both eyes were covered with retinopalmitate (Oculotect gel, CIBA Vision Vertriebs GmbH, Großostheim, Germany). The skin overlying the vertebral column was incised, and the muscles were detached from the vertebra. A single‐level bilaminectomy was then performed at level T8 to expose the spinal cord. After opening the dura mater, the dorsal spinal cord was symmetrically lesioned with fine iridectomy scissors (FST, Heidelberg, Germany), resulting in a four‐fifth hemisection. This was assured by marks on the microscissors correlating with four‐fifth spinal cord incision depth. To obtain homogeneous tissue samples with comparable transection depths, we included only those tissue sections within two standard deviations according to the method described earlier (26). An incision depth analysis was previously published for model characterization (5). Because this methodology is suitable for determining and comparing the sensitive locomotor outcome after SCI, we adopted it for the immunohistochemical analysis. The targeted neuronal structures were of motor (crossed pyramidal tract, part of the extrapyramidal tract) and sensory (dorsal columns) origins. The wound was rinsed with normal saline and closed in layers. All animals were warmed with infrared light until they recovered from anesthesia.

Postoperative care of rats and tissue preparation

All rats received postoperative analgesic treatment, underwent manual bladder compression (twice daily) and were bathed daily to prevent urine burns until spontaneous bladder function occurred (usually within 10–14 days). The animals were frequently weighed and were killed if their weight loss exceeded 20% of the preoperative weight. For immunohistochemical analysis, rats were killed and perfused intracardially with fixative (4% formalin in 0.1 M phosphate‐buffered saline, pH 7.5). The spinal cords and brains were removed and postfixed overnight at 4°C. The fixed tissue was embedded in paraffin, serially sectioned (3–5 µm) and mounted sequentially on Silan‐covered slides.

Immunohistochemistry

Immunohistochemical analysis was performed on 5‐µm paraffin‐embedded tissue sections that were deparaffinized, rehydrated and boiled for 15 minutes in a 600‐W microwave oven in sodium citrate buffer (2.1 g/L, pH 6). Endogenous peroxidase was inhibited by addition of 1% H₂O₂ in methanol for 15 minutes. To block nonspecific binding of immunoglobulins, sections were pretreated with 10% normal goat serum (Linaris, Wertheim‐Bettingen, Germany). Cell type‐specific antigens were detected by the following primary antibodies: rabbit polyclonal anti‐MPO (Life Span, Seattle, WA, USA; dilution 1:100) for neutrophils (PMN), mouse monoclonal anti‐ED1 (Serotec, Düsseldorf, Germany; 1:100) for microglia/macrophages, mouse monoclonal anti‐W3/13 (Serotec; 1:100) for pan T lymphocytes. Given the co‐labeling of few MPO⁺ monocytes, only MPO⁺ cells revealing a granulocytic lobular nucleus were considered as granulocytes. Similarly, W3/13‐positive cells with a lobulated nucleus within abundant cytoplasm (indicative of granulocytes) were excluded from counting as T lymphocytes. Primary antibodies were applied overnight at 4°C, followed by biotinylated goat anti‐rabbit or anti‐mouse secondary antibody (1:400; Dako, Hamburg, Germany) for 30 minutes. Bound antibody complexes were visualized with a peroxidase‐conjugated streptavidin–biotin complex using diaminobenzidine as a substrate (Sigma, Munich, Germany). Sections were counterstained with Mayer's hemalaun. Negative controls consisted of tissue sections incubated with an unspecific IgG1 isotype antibody (Beckman Coulter, Fullerton, CA, USA; 1:100) or omission of primary antibody.

Double‐labeling experiments

Sections were prepared as above, and the following primary antibodies were applied: rabbit monoclonal anti‐CD86 (Abcam, Cambridge, UK; 1:100) for M1 macrophages; mouse monoclonal anti‐CD206 (Serotec; 1:100) for M2 macrophages, mouse monoclonal anti‐CD8α (Serotec; 1:50) for CD8 co‐expression. After adding biotinylated secondary antibodies (dilution 1:400) for 30 minutes, slices were visualized using alkaline phosphatase‐conjugated avidin–biotin complex (Dako) for 30 minutes and Fast Blue salt. Before the second labeling step, slices were boiled in citrate buffer in a microwave for 10 minutes. ED1 staining was performed as above.

Evaluation and statistical analysis

Mean numbers of labeled cells from injured spinal territories (n = 5 rats) were compared with the control spinal cord tissue from sham‐operated rats (n = 3). Cells were counted in 10 high‐power fields (400 × magnification, together representing 0.625 mm²) restricted to the lesion area (pannecrotic lesion center and lesion margin). Data were calculated as means ± standard deviation. Nonlinear regression analysis assuming a half‐life kinetic was performed for the calculation of the plateau (equation: one phase exponential decay) using Prism 5.0 (GraphPad Software Inc., La Jolla, CA, USA).

RESULTS

Twenty‐eight rat spinal cords following transection SCI and three control spinal cords were evaluated in order to investigate the differential cell influx and disappearance of macrophages, lymphocytes and neutrophils following SCI by immunohistochemistry. In contrast to investigations of the early onset of inflammation, we focused on the dynamics of leukocyte cell populations after reaching maximum levels characterizing the clearance of polymorphonuclear cells (PMN), microglia/macrophages and T lymphocytes. Resolution indices were applied to the acute inflammatory CNS lesion based on the absolute numbers of lesional accumulating cells and time kinetics (Figure 1).

Here, we report about the intrinsic (i) leukocyte subtype‐specific resolution indices (R_i) and (ii) sustained non‐resolving aspects of acute CNS inflammation (resolution plateau, R_p) using a rat spinal cord injury (Table 1). As the time kinetic of resolution (cell clearance) is asymmetric (first 50% of the cells are cleared in a shorter time compared with the remaining cells), we introduce two main parameters to characterize the resolution of CNS inflammation. The clearance efficacy of the first 50% of leukocyte is characterized by the resolution interval (R_i), whereas the subsequent complete or incomplete resolution of the remaining cells (“residuum”) is represented by a sustained resolution plateau (R_p) in relation to sham animal levels.

Table 1.

Cell type‐specific resolution programs (indices) after SCI. The accumulation of neutrophils, microglia/macrophages and T lymphocytes at the lesion site after SCI was characterized using the quantitative resolution terms Ψ_max, T_max and R_i. The proposed resolution plateau R_P represents the percentage of cells that are detectable at the lesion site for extended periods of time (sham levels serve as baseline, 0%).

Cells	Immunocyto‐chemical marker	Ψmax (cells/0.625 mm2)	Tmax (days)	Ri (days)	RP (%)
Neutrophils	MPO	111.6	3	1.2	0
Macrophages/microglia	ED1	365.6	7	55	45
T lymphocytes	W3/13	174.4	3	1.5	10

Defining the resolution interval (R_i) of the inflammatory cellular milieu after SCI

After SCI, lesional neutrophils accumulated early with maximum infiltration at day 3. The PMN were clearly detectable using immunostaining for MPO (Figure 2E,F). MPO⁺ cells with morphological signs of macrophages (monomorphic nucleus, large cell size) were excluded from the analysis. Negative controls using isotype antibodies or omission of the primary antibody confirmed the specificity of the staining (Figure 2A–C). Neutrophil numbers rapidly declined early after SCI with about similar speed as they appeared (Figure 2D). The R_i for PMN was 1.2 days reflecting the “half‐life” in the inflammatory milieu. In contrast, ED1⁺ microglia/macrophages accumulated later with peak numbers at day 7 and revealed a protracted resolution kinetic, illustrated by slower clearance/efflux from the lesion site (Figure 2G–I). In these areas of both primary and secondary injury, microglia/macrophage numbers remained elevated after SCI. The R_i of lesional activated microglia/macrophages accounts 55 days resulting in a 40 times slower clearance rate compared with PMN. W3/13⁺ T lymphocytes started to accumulate at the inflammatory site at day 1, reaching maximum levels at day 3 (Figure 2J–L). Their number rapidly declined thereafter (R_i = 1.5 days).

Figure 2.

Time course of accumulation and clearance of leukocytes at the lesion site after SCI. Negative controls with IgG1 isotype antibodies (A) or omission of the primary antibody using only secondary goat anti‐mouse (B) and anti‐rabbit antibodies (C) confirmed the specificity of the immunostainings. Lesional accumulation of different inflammatory cell populations was quantified after 1, 3, 7, 14, 28 and 70 days post SCI (n = 5 animals per group) and compared with sham‐operated animals (n = 3). MPO⁺ neutrophil numbers reached maximum at day 3 and declined rapidly to the control level (D). Histological staining confirmed the presence of polymorphonuclear MPO⁺ neutrophils at the lesion site 3 days post injury (E), whereas they were completely absent within the scar formation 70 days after SCI (F). The time course of ED1⁺ microglia/macrophages reached maximum numbers at day 7 (G) with huge numbers accumulating at the lesion core (H). Microglia/macrophage numbers declined slowly and remained markedly elevated (G) with many foamy microglia/macrophages intermingled in the scar formation at day 70 (I). Maximum numbers of W3/13⁺ T lymphocytes were observed at day 3 and decreased rapidly but did not disappear completely from the injury site (J). W3/13⁺ T lymphocytes were detected frequently within the lesion area (K). A small but constant proportion of T lymphocytes could be detected until day 70 at the injury site (L). Numbers of labeled cells are plotted as mean ± SD. Green bars indicate the period of prevailing cell clearance after peak accumulation (even though resolution starts already at the onset of inflammation, the green bars indicate the time window where leukocyte clearance >cell entry). Scale bar: 100 µm in panels A–C, 50 µm in panels E, F, H, I, K, L.

Defining the resolution plateau (R_p) of the inflammatory cellular milieu after SCI

Thus, PMNs are the only completely resolving leukocyte subset after SCI. Only very few PMNs were still detectable after 14 days, and they were fully cleared from the lesion site after 10 weeks (Figure 2D). In contrast, the analysis of accumulating microglia/macrophages and lymphocytes at the lesion area after SCI demonstrated that large numbers of inflammatory cells persisted for extended periods of time, representing a sustained “inflammatory residuum” at the lesion site. Even after 10 weeks, almost half of the maximum microglia/macrophage number was still detectable at the lesion area. At that time, ED1⁺ cells revealed morphological hallmarks of chronically activated microglia/macrophages (see below). Based on the general assumption that these cells can participate in secondary damage (at least caused by bystander inflammation) 4, 48, we propose that the sustained plateau of accumulating cells (“inflammatory residuum”) (2, 3) interferes with proper resolution of inflammation, which in turn is essential for functional tissue remodeling and wound healing (21). We here introduce the term “resolution plateau” (R_P) in order to better describe the persisting cellular component of sustained inflammation, which might participate in late degeneration and non‐resolving inflammation (Figure 5). In our study, we identified R_P values of 0% for PMN (“complete resolution”; eg, returning back to sham levels) and of 10% (residual cell numbers) for T cells (“incomplete resolution”) after SCI. Activated microglia/macrophages are only partially cleared with 45% remaining at 10 weeks post SCI.

Figure 3.

Leukocyte subtype‐specific resolution capacity after acute inflammation. Whereas neutrophils completely disappeared between 4 and 10 weeks after SCI (A), macrophages and T lymphocytes constituted a resolution plateau of persisting cells in the lesion area (B–C). Macrophages decreased slowly and approximated to their plateau of about 45% that can be predicted with reasonable certainty from nonlinear regression analysis (B). T cells reached their plateau early with about 10% of cells compared with the maximum numbers (C). The cell‐specific resolution index R_i can be determined from the curve at the time when cell numbers decreased by 50% as indicated by shaded areas between the dotted lines. Broken lines indicate the cell‐specific resolution plateau R_p. Percentages of labeled cells are plotted as mean ± SD.

Figure 5.

Non‐resolving acute CNS inflammation. Model of a resolution deficit with respect to the resolution plateau R_P. Based on (i) the assumption that tissue‐persisting leukocytes might enhance secondary damage after CNS injury, and (ii) given the current evidence of a quantifiable resolution plateau R_P, we propose that the resolution plateau represents a numerical equivalent of a resolution deficit, which determines whether resolution of inflammation is complete (A, R_P = 0%), incomplete (B, 0 < R_P<100%) or absent (C). Thus, the resolution plateau is proposed to complement the R_i in organs of non‐complete resolution. Here the R_P defines sustained chronic CNS inflammation at the cellular level.

Thus, they represent a major cellular fraction of the “inflammatory residuum.” With values between 0% and 45% compared with the respective maximum cell number (Figure 3, Table 1), R_p defines the wide range of leukocyte subset‐specific delayed intrinsic resolution capacity in this SCI model. Consequently, the time kinetics of lesional inflammatory cells accumulating after SCI reflects a differential cell type‐specific persistence (“dwell time”) of leukocytes.

Phenotyping the resolution of immune cells in the lesion area

In order to characterize the differential intrinsic resolution capacities of distinct leukocyte subtypes, we analyzed the time course of resolution from the acute inflammatory site. For quantification of the sustained leukocyte accumulation, cells can be categorized in three groups: (i) complete disappearance from the lesion (“complete resolution”); (ii) reaching a stable plateau already during the observation period of 10 weeks (“incomplete resolution—type A”); and (iii) slow decrease of cell numbers without reaching a plateau during the observation time (“incomplete resolution—type B”). Neutrophils belong to the first group with complete disappearance between 4 and 10 weeks; thus, neutrophils did not constitute a persisting plateau (Figure 3A). In this case, the resolution interval (R_i = time between maximum infiltration and disappearance of 50% of the cells) can be read directly from the curve and was 1.2 days (Figure 3A). Also, the independently calculated plateau using nonlinear regression analysis confirms that the disappearance of neutrophils can be described using a first‐order (half‐life) kinetic (Figure 3A). T cells belong to the second group reaching a plateau during the observed disease course. Analysis of the curve revealed persistence of about 10% of T cells after 4 weeks, and R_i represents 1.5 days (Figure 3C). Interestingly, assuming a half‐life kinetic, the exact calculation again revealed similar values with a plateau of 9.6%. Thus, the plateau can be mathematically defined and was properly validated by the actual cell numbers based on histological analysis. As a result, mathematical calculation could allow prediction of the plateau in cells belonging to the third group where the remaining cell numbers likely further decline beyond the observation period. In the case of macrophages, R_i was 55 days, and the plateau was calculated to be 44.6% (Figure 3B).

Microglia/macrophage subtype characterization during active resolution

The majority of persisting cells after SCI consisted of ED1⁺ microglia/macrophages. In order to estimate their role for secondary damage, we characterized the morphological and immunological phenotype in more detail. During the resolution phase, lesional ED1⁺ cells underwent time‐dependent changes in morphology (Figure 4). At days 7 and 14, a moderate proportion (20%–30%) of ED1⁺ cells displayed shortened and thickened protrusions indicating microglia in a state of early activation (19). These were prevailingly detected in the perilesional rim directly adjacent to the lesion core (Figure 4A). At the same time, densely packed foamy cells with large lysosomal inclusions and eccentric nuclei—indicative of macrophages (19)—were observed mainly in the lesion core but also in the perilesional rim (Figure 4B). Between day 28 and day 70, most ED1⁺ cells in the scar formation were large‐sized foamy macrophages with strictly eccentric nuclei and numerous irregular inclusions surrounded by small edges of cytoplasm (Figure 4C). Ramified ED1⁺ cells with short processes were still present at later time points, but their proportion decreased to <10%. In contrast, the fraction of granular/foamy macrophages at the lesion site represented more than 90% of ED1⁺ cells at day 70 (Figure 4C).

Figure 4.

Changes in the macrophage/microglia phenotype during the resolution period. At days 7 and 14 after SCI, a moderate proportion (20%–30%) of ED1⁺ cells (DAB) displayed thickened branched protrusions indicating activated microglia, which were mainly detected at the lesion border (A). At the same time, densely packed foamy macrophages with many irregular inclusions and eccentric nuclei were observed at the lesion site (B). At day 70, most ED1⁺ cells were large‐sized foamy macrophages (>90%), intermingled in the scar formation and showing small cytoplasmatic edges between the inclusions (C). Double labeling with CD8 (Fast Blue) revealed that about 50% of the ED1⁺ cells were CD8⁺ during the observation period between day 7 (D) and day 70 (E). Scale Bar: 25 µm. DAB = diaminobenzidine.

Given the limitations of a functional characterization of microglia/macrophages based on morphology (31), we performed double‐staining experiments at the time of maximum cell infiltration (day 7) and at the latest study time point (day 70). ED1⁺ cells were divided into CD86⁺“classically activated” (M1) macrophages vs. CD206⁺“alternatively activated” (M2) macrophages 18, 23. CD86⁺ macrophages were mainly detected in the necrotic lesion core where they represented 10% of ED1⁺ cells at day 7 and 30% at day 70. In addition, CD86 was strongly expressed on astrocytes, which is in line with earlier findings (30). CD206 was detectable in 60% of ED1⁺ cells at day 7 but only in 20% at day 70. Moreover, numerous ED1⁺ cells were co‐labeled with CD8, a marker shown to be expressed on cytotoxic macrophages in different lesion models 17, 35, 38, 39. The frequency of ED1/CD8 double‐positive cells in the lesion core was about 50% of ED1⁺ cells and did not change between day 7 (Figure 4D) and day 70 (Figure 4E).

DISCUSSION

We have analyzed the resolution of acute inflammation in an immune‐privileged organ after spinal cord injury. The SCI model was chosen to detect even discrete responses, as trauma to the spinal cord elicits a more profound immune response compared with similar injury of the brain 36, 37.

We characterized the clearance of cells accumulating at the lesion site, which follows principal rules and occurs in a leukocyte cell type‐specific manner. The R_i range from 1.2 days (PMN) being the most short‐lived leukocyte population, followed by T lymphocytes (1.5 days) to microglia/macrophages (55 days). Thus, both the clearance rate of distinct leukocyte subpopulations and also the R_p at the CNS lesion site are regulated differentially, suggesting distinct encoded resolution programs, thereby pointing to diverging resolution receptor repertoire. This is of note to map the residual lesional inflammatory milieu in a spatio‐temporal context. The changing cellular composition at the lesion site, caused by “complete,”“incomplete” and “failing cellular resolution” (Figure 5) defines a fluctuating residual inflammatory milieu—relevant to give rise to an infection‐triggered deterioration, as non‐resolving lesional cells identified as microglia/macrophages express the lipopolysaccharide (LPS) receptor CD14 3, 31. Indeed, the remaining, non‐resolving inflammatory milieu may be considered as an “immunological scar” interfering with the fine‐tuned “immunological homunculus” proposed earlier (8). The here introduced indices that quantify resolution of inflammation might allow to identify different steps of resolution efficacy related to distinct cognate receptors being expressed (20), as well as to characterize organ‐specific differences in the capacity of resolution. Finally, R_i and R_p are helpful to identify “resolution impeding” factors such as COX‐2 blockers (41) or anesthesia, which might be relevant for iatrogenic CNS injury during neurosurgery (7).

The general pattern of lesional inflammatory response after SCI was similar to previous investigations in rodents and human 10, 12, 16, 34, 36, 46. It has been shown that neutrophils in the injured spinal cord were most numerous 1–3 days after SCI and disappeared within 5–10 days 12, 47. Microglia/macrophages achieved their highest number at 5–7 days and persisted for weeks to months 6, 12, 34. Recruitment of T lymphocytes after SCI was variable, also depending upon species and strain of animal 12, 46, with maximal infiltration between 3 and 7 days post injury (34).

It has become a consensus that the clearance of all inflammatory cells including microglia/macrophages from the site of injury is required for complete resolution and tissue homeostasis 21, 44, irrespective of whether invaded cells just “fizzle out” or are supplemented by in situ proliferation or secondary invasion after injury. Macrophages maintain the ability to respond to inflammatory stimuli resulting in tissue damage 11, 32, 33, although they might also mediate beneficial effects for tissue repair in a “non‐phlogistic” fashion 9, 27. Here, characterization of macrophage subtypes revealed the presence of foamy macrophages expressing CD8 during the whole observation period and a relative increase of M1 macrophages with longer intervals post SCI. CD8⁺ macrophages are of hematogenous origin, accumulate in pannecrotic CNS lesions after ischemia, EAE and SCI 35, 38, 39, 40, express the cytotoxic effector molecule perforin and play an important role in demyelination during chronic EAE 17, 40. The M1/M2 macrophage dichotomy distinguishes between the production of pro‐inflammatory cytokines as well as release of cytotoxic agents (such as nitric oxide) in M1 cells and the preferential anti‐inflammatory cytokine profile in rather tissue‐protective macrophages of the M2 phenotype 18, 22, 23. Together, these functional macrophage characteristics support the notion that cells of the resolution plateau participate in pathophysiological processes after SCI.

Our analysis underlines that the temporal pattern of cell clearance can be exactly described, which supports the notion of a well‐regulated resolution process and provides the rationale for further analysis of pro‐resolution strategies. This appears of conceptual relevance as the prolonged “dwell time” of lesional leukocytes might be at least equally important as the maximum amount of cells accumulating at the lesion site, which served as the dominating target of CNS immune‐modulatory therapies in the past.

We were particularly interested in the presence and quantification of a resolution deficit after inflammation. Previous attempts focused on changes in the resolution interval, and treatments with resolution agonists convincingly documented the beneficial effects for accelerated resolution. In murine peritonitis, for example, resolvin E1 (RvE1) and 10,17S‐docosatriene (10,17S‐DT) significantly reduced Ψ_max (maximal neutrophil numbers) and T_max (time until maximum), and 10,17S‐DT further profoundly shortened the resolution interval R_i (time from maximum to 50% reduction) (2). Based on these quantitative indices, it has been postulated that a failure of resolution may result in chronic inflammation (21), which is considered a major driver of disease (28). Consequently, in the CNS, application of resolution agonists resulted in attenuated inflammatory pain (49) or reduced secondary damage after ischemic CNS injury (24).

Given the potential impact of non‐resolving leukocytes in the acute CNS lesion, we further introduced a quantifiable index that represents an important feature of resolution: the resolution plateau, R_P (Figure 5). The resolution plateau represents the late phase of resolution and describes to which extent the inflammatory cells decline (relative to sham control levels). Theoretically, R_P can range between 0% and 100%. An R_P of 0% indicates complete cellular resolution of inflammation with cell numbers equaling those in sham animals. Complete resolution (Figure 5A) is found in superficial skin wounds where all cardinal signs of inflammation are present (rubor, calor, tumor, dolor), but scarless wound healing is paralleled by complete loss of cellular infiltrates (14). The other extreme is an R_P of nearly 100% representing ongoing inflammation without any resolution leading to sepsis, spreading to multiorgan inflammation and eventually death of the animal (Figure 5C). An example for this is the persistent inflammatory response in animals after knockout of SOCS1 (15). Most inflammatory processes, however, are likely to result in partial resolution (Figure 5B). In the present SCI model, R_P was 0% for neutrophils, 10% for T lymphocytes and 45% for macrophages.

The evidence for a resolution plateau has also been demonstrated in human CNS neuropathology, where blood‐borne cells survive in the tissue. For example, lipid‐laden microglia/macrophages can result in a smoldering, non‐resolving inflammatory milieu for up to years in human ischemic CNS lesions (42) or traumatic brain injury (3). Noteworthy, after acute human CNS lesions, perivascular spaces get cleared of microglia/macrophages months after injury, whereas at lesion parenchyma, microglia/macrophages fail to return to control levels (3). This suggests restricted clearance of the lesional milieu or of the accumulated leukocytes—or even implies the option of “late” recruitment of inflammatory cells. The putative relevance of the here defined resolution plateau R_P is further supported by the finding that stimulation of macrophages by innate immunity with LPS is sufficient to induce demyelinating lesions (11) as candidate mechanism involved in delayed ongoing demyelination and subsequent neuronal degeneration after SCI 1, 4, 48.

Given that the resolution plateau R_P is likely related to incomplete resolution of inflammation, additional studies should determine important further questions, such as (i) whether a higher resolution plateau containing more persisting blood‐borne cells leads to worse clinical outcome and histopathological damage and (ii) which effects do resolution agonists have in modifying R_P behind the blood barrier?

In summary, we characterized the non‐resolving aspect of inflammation after CNS injury using objective quantitative measures. These include “resolution indices” (R_i) complemented by “resolution plateaus” (R_p) to analyze the efficacy of intrinsic resolution programs in an immune‐privileged organ after SCI. These specific indices are a prerequisite to (i) understand organ‐specific differences in the efficacy of inflammatory resolution, (ii) decipher distinct phases of resolution, (iii) identify “CNS‐ effective” resolution agonists and (iv) search for “resolution toxic” factors and finally for the development and testing of pro‐resolution strategies. We also introduced the concept of the “resolution plateau” R_P as a marker for the persisting cellular inflammatory component that is likely to participate in neurodegeneration in the later stages after injury.