Hemorrhagic shock shifts the serum cytokine profile from pro- to anti-inflammatory after experimental traumatic brain injury in mice.
Go to:Secondary insults, such as hemorrhagic shock (HS), worsen outcome from traumatic brain injury (TBI). Both TBI and HS modulate levels of inflammatory mediators. We evaluated the addition of HS on the inflammatory response to TBI. Adult male
C57BL6J mice were randomized into five groups (n=4 [na?ve] or
8/group): na?ve; sham; TBI (through mild-to-moderate controlled cortical impact [CCI] at 5?m/sec, 1-mm depth), HS; and CCI+HS. All non-na?ve mice underwent identical monitoring and anesthesia. HS and CCI+HS underwent a 35-min period of pressure-controlled hemorrhage (target mean arterial pressure, 25–27?mm Hg) and a
90-min resuscitation with lactated Ringer's injection and autologous blood transfusion. Mice were sacrificed at 2 or 24?h after inj ury. Levels of 13 cytokines, six chemokines, and three growth factors were measured in serum and in five brain tissue regions. Serum levels of several proinflammatory mediators (eotaxin, interferon-inducible protein 10 [IP-10], keratinocyte chemoattractant [KC], monocyte chemoattractant protein 1 [MCP-1], macrophage inflammatory protein 1alpha [MIP-1α], interleukin [IL]-5, IL-6, tumor necrosis
factor alpha, and granulocyte colony-stimulating factor [G-CSF]) were increased after CCI alone. Serum levels of fewer proinflammatory mediators (IL-5, IL-6, regulated upon activation, normal T-cell expressed, and secreted, and G-CSF) were increased after CCI+HS. Serum level of anti-inflammatory IL-10 was significantly increased after CCI+HS versus CCI alone. Brain tissue levels of eotaxin, IP-10, KC, MCP-1, MIP-1α, IL-6, and G-CSF were increased after both CCI and CCI+HS. There were no significant differences between levels after CCI alone and CCI+HS in any mediator. Addition of HS to experimental TBI led to a shift toward an anti-inflammatory serum profile—specifically, a marked increase in IL-10 levels. The brain cytokine and chemokine profile after TBI was minimally affected by the addition of HS.
Key words: : blast injury, chemokine, head injury, hypotension, interleukin, polytrauma, resuscitation
D EATH AND UNFAVORABL
E NEUROLOGIC OUTCOME after traumatic brain injury (TBI) are strongly associated with secondary insults, such as hypotension.1 This secondary insult has taken on great importance related to blast TBI of U.S. soldiers injured in attacks by improvised explosive devices in both Operation Iraqi Freedom and
Operation Enduring Freedom.2 In blast injury, TBI is often accompanied by polytrauma. Hemorrhagic hypotension thus results from extremity injuries and/or shrapnel, and high mortality rates have been reported in this setting.3Hypotension after TBI is also common in civilian cohorts, affecting 26% of TBI patients in one study.4
To better understand the mechanisms underlying the exacerbation of damage by hypotension after TBI, we have developed mouse models of combined TBI plus hemorrhagic shock (HS). In these models, TBI is produced by a mild-to-moderate controlled cortical impact (CCI), followed by either volume-5–7 or pressure-controlled8 HS. In our model of pressure-controlled HS, we recently showed that the addition of HS exacerbates contusion volume, hemispheric brain tissue loss, hippocampal neuronal death, and functional
deficit.8 However, the cellular and molecular mechanisms contributing to this unfavorable outcome have only begun to be examined, and the effect of HS on the brain and systemic inflammatory response to TBI remain to be defined.
Experimental and clinical studies of TBI have demonstrated robust increases in inflammatory mediators, such as cytokines, chemokines, and growth factors, in serum, cerebrospinal fluid (CSF), brain interstitial fluid, and brain tissue.9–34 The overall effect of
inflammation on the brain after TBI, however, is complex, and earlier studies indicate that inflammation after TBI is a ―dual-edged sword.‖ For example, blockade of proinflammatory interleukin (IL)-1β is associated with improvements in functional outcome and brain tissue loss after CCI,35whereas knockout (KO) of proinflammatory inducible nitric oxide synthase is associated with worsened functional outcome and hippocampal neuronal loss after CCI.36 Similarly, conflicting roles for the associations of pro- and
anti-inflammatory cytokines with favorable and unfavorable outcomes have been reported in clinical studies.10,22,34,37 To our knowledge, no studies have specifically examined the effect of HS on inflammatory response to TBI in either the experimental or clinical setting.
In the current report, Multiplex technology was used to measure brain tissue and serum levels of a large number of inflammatory mediators, including cytokines, chemokines, and growth factors, in mice after sham injury, HS, CCI, and a combined CCI plus HS insult. We hypothesized that the addition of HS to CCI would modulate cerebral and systemic inflammatory responses observed after TBI alone toward a more proinflammatory state.
Go to: Methods
Experimental protocol and study groups
The institutional animal care and use committee of the University of Pittsburgh School of Medicine (Pittsburgh, PA) approved all experiments. The details of our CCI+HS model have been previously described.8 In the current study, male C57BL6J mice (The Jackson Laboratory, Bar Harbor, ME), 12–15 weeks of age, were used. Anesthesia was induced with 4% isoflurane in oxygen and maintained with 1% isoflurane in 2:1?N2O/O2 by nose cone. Inguinal cut down and insertion of femoral venous and arterial catheters were accomplished under sterile conditions using modified polyethylene 50 tubing. After placement of the mouse in a stereotaxic frame, a 5-mm craniotomy was performed over the left parietotemporal cortex with a dental drill, and the bone flap was removed. Immediately after craniotomy, the inhalational anesthesia was changed to 1% isoflurane and room air for a 10-min equilibration period before beginning the injury protocols. Monitoring included a brain parenchyma temperature probe inserted through a separate burr hole. While brain temperature was maintained at 37±0.5°C, CCI was performed with a pneumatic impactor. A 3-mm flat-tip impounder was deployed at a velocity of 5?m/sec and a depth of 1?mm. This injury level for CCI was specifically chosen to produce a contusion
without significant hippocampal neuronal death, unless HS is superimposed upon the insult.5,8,38,39
HS was induced by phlebotomy to a target mean arterial blood pressure (MAP) of 25–27?mm Hg (HS phase). Mice remained hypotensive for a total of 35?min. After completion of the HS phase, a 90-min prehospital ph ase was initiated and 20?mL/kg of lactated Ringer's (LR) was rapidly infused. Additional aliquots of 10?mL/kg were infused as needed to achieve a MAP of 70?mm Hg. To simulate arrival at definitive care (hospital phase), the inhaled gas mixture was switched from 1% isoflurane in room air to 1% isoflurane in oxygen, and all shed blood was reinfused. At completion of the hospital phase, catheters were removed, anesthesia discontinued, and mice recovered in supplemental oxygen for 30?min before being returned to their cages.
Mice were randomized to one of five study groups (n=4 [na?ve] or 8 per group) and underwent procedures or equivalent anesthesia and monitoring as designated: 1) na?ve (no anesthesia, craniotomy, monitoring, CCI, or HS); 2) sham (anesthesia, monitoring, and craniotomy, without CCI or HS); 3) CCI alone; 4) HS alone (including monitoring, without craniotomy); or (5) CCI+HS. Mice in the CCI-only group underwent CCI without HS, but were maintained under identical anesthesia and monitoring as the
combined injury group for a 90-min interval. Heart rate (HR), MAP, arterial blood gas results (pH, PaCO2, PaO2, and base deficit), hematocrit, and lactate were recorded serially.
Mice were euthanized at either 2 or 24?h after finishing the experimental protocol (n=4 per time point per study group). Mice were anesthetized with 4% isoflurane, and a blood sample was obtained by cardiac puncture. Animals were then transcardially perfused with heparinized saline. Brains were dissected into cerebellum, parietal cortex (ipsi- and contralateral), and hippocampus (ipsi- and contralateral) and snap-frozen with liquid nitrogen. Tissue was weighed and homogenized with an 8×volume of 1×phosphate-buffered saline (PBS), sonicated, and spun for 30?min at 14,000g to obtain the supernatant. Protein concentration was measured using the bicinchoninic acid assay (Thermo Fisher Scientific, Inc., Vernon Hills, IL). Supernatant aliquots were assayed for inflammatory mediators by Multiplex (n=3 for 24-h CCI group; Millipore, Billerica, MA). Individual protein levels were indexed to total measured level of protein in the sample (pg/mg of protein). Levels of 10 proinflammatory cytokines (IL-1α, IL-1β, IL-2, IL-5,
IL-6, IL-9, IL-12, IL-17, interferon-gamma [IFN-γ], an d tumor necrosis factor alpha [TNF-α]), three anti-inflammatory cytokines
(IL-4, IL-10, and IL-13), six chemokines (eotaxin [C-C motif chemokine ligand (CCL)11], IFN-inducible protein 10 [IP-10; C-X-C motif chemokine ligand (CXCL)10], keratinocyte chemoattractant [KC; CXCL1; analog of human IL-8], monocyte chemoattractant protein 1 [MCP-1; CCL2], macrophage inflammatory protein 1 alpha [MIP-1α; CCL3], and regulated upon activation, normal T-cell expressed, and secreted [RANTES; CCL5]), and three growth factors (granulocyte colony-stimulating factor [G-CSF], granulocyte macrophage/colony-stimulating factor [GM-CSF], and vascular endothelial growth factor) were measured. Though chemokines are a type of cytokine (chemotactic cytokine), the term ―cytokine‖ is us ed throughout the article to mean cytokines that are not chemokines (i.e., IL-4, IL-6, and so on).
Comparison of sensitivity of Multiplex and enzyme-linked immunosorbent assay
To determine how the sensitivity of Multiplex compared to
enzyme-linked immunosorbent assay (ELISA) specifically in brain tissue samples, we carried out two additional experiments. First, we studied mice in the same three groups (na?ve, CCI, and
CCI+HS; n=3/group) and sacrificed them at 2?h after insult using an identical protocol to that used for Multiplex assessments. Brain samples from the same five regions were processed identically and again homogenized in PBS as described above and assayed for
TNF-α by ELISA (R&D Systems, Minneapolis, MN). Second, we again studied mice in the same threegroups (na?ve, CCI, and
CCI+HS;n=3/group) and sacrificed them at 2?h after insult using an identical protocol to that used for Multiplex and aforementioned ELISA, except that lysis buffer (radioimmunoprecipitation assay [RIPA] buffer; Thermo Fisher Scientific) was used rather than PBS. Brain samples from the same five regions were again assayed for TNF-α by ELISA (R&D Systems). The assay detection limit was
Brain levels of mediators are reported in pg/mg of protein (median [range]) and serum levels in pg/mL of serum (median [range]). Nonzero mediator levels outside of the lab's reference range were transformed into a standardized high or low value (2000 or 0.1?pg/mg of protein for brain tissue samples; 20,000 or 1.0?pg/mL for serum samples). Similar to earlier studies, mediators with levels frequently outside of the reference range (>65%) were excluded from analysis.40,41 Results of mediator levels were grouped by injury mechanism, tissue sample type, and time point. Given that the data were not normally distributed, Wilcoxon's rank-sum test with Bonferroni's correction for multiple comparisons was used to make pair-wise group comparisons with sham as control for brain tissue
levels and na?ve as control for serum levels. Physiology parameters are reported as mean. Physiology parameters were compared with a one-way analysis of variance test and with Holm-Sidak's test. A value ofp<0.05 was deemed significant. Only results with significant differences between groups are shown.
Go to: Results
Similar to earlier studies in this model,8 there were no significant differences in baseline physiologic parameters between groups (Table 1). After injury, HR was increased in all injury groups versus sham and was increased in the CCI+HS group versus CCI and HS-alone groups. The significant decrease in MAP in the HS and CCI+HS groups that was observed after the HS and prehospital phases was not present at the end of the hospital phase, after return of the shed blood. There were no significant changes in pH other than a mild acidosis (7.34±0.02) in CCI+HS versus both sham and CCI groups after the HS phase. However, compared to sham and CCI, the HS and CCI+HS groups had significant increases in base deficit and blood lactate after the HS phase, with significant respiratory compensation as reflected by hypocarbia. Blood lactate was also increased at the end of the prehospital phase in the CCI+HS group
versus sham and CCI groups. Hematocrit was decreased in the HS and CCI+HS versus the sham and CCI groups after injury. PaO2 was increased in the HS and CCI+HS versus the sham and CCI groups at the end of the HS phase. One animal in the CCI+HS group died during the prehospital phase.
PHYSIOLOGY PARAMETERS AT VARIOUS STAGES OF EXPERIMENT
Brain tissue levels
Of the 22 inflammatory mediators tested by Multiplex, seven were increased (p<0.05 vs. sham) in brain tissue both after CCI alone and after CCI+HS, including five chemokines (eotaxin, IP-10, KC,
MCP-1, and MIP-1α), one cytokine (IL-6), and one growth factor (G-CSF; see Fig. 1A–G). The highest median levels of all seven mediators were observed at 24?h after both CCI alone and CCI+HS. Significant increases in eotaxin, IP-10, MCP-1, and MIP-1α observed at 24?h after CCI+HS in ipsilateral brain tissue were not observed at 2?h after CCI+HS in the same brain region. Similarly, unique
increases in brain tissue IP-10 and MIP-1α were observed after CCI only at the later time point. There were no statistically significant differences between levels after CCI alone and after CCI+HS in any mediator, and no mediators were increased after only CCI or only CCI+HS. No mediators were increased after HS alone.
Measured values of inflammatory mediators. Bars represent minimum to maximum values, and the dividing line represents the median. Brain tissue regions are ipsilateral hippocampus (blue), contralateral hippocampus (gray), ipsilateral parietal cortex (yellow), ...
Figure 2A,B shows results of the two ELISA determinations of TNF-α carried out to compare the sensitivity of Multiplex to ELISA in brain tissue. In brain samples processed for ELISA with PBS, TNF-α levels were, in general, low (<0.5?pg/mg protein). Increases versus na?ve were not observed in 13 of 15 brain regions that were assayed, consistent with Multiplex. However, TNF-α levels were
significantly increased in the hippocampus ipsilateral to injury
(p<0.05 vs. na?ve; Fig. 2A).
Tumor necrosis factor alpha (TNF-α) levels assessed by
enzyme-linked immunosorbent assay in cerebellum (CER), left and right parietal cortex (LPC and RPC, respectively), and left and right hippocampus (LH and RH, respectively) in na?ve ...
In samples processed with RIPA lysis buffer and assessed by ELISA, TNF-α levels were somewhat higher than in samples processed with PBS and assessed by either Multiplex or ELISA. Significant increases were observed versus na?ve in both parietal cortex and hippocampus ipsilateral to injury in both CCI and CCI+HS, although the highest level observed was 1.05?pg/mg of protein.
Two hours after HS, serum levels of nine mediators were increased (p<0.05 vs. na?ve), including three chemokines (IP-10, MCP-1, and MIP-1α), five cytokines (IL-1α, IL-5, IL-6, IL-10, and TNF-α), and one growth factor (G-CSF; see Fig. 1C–K). At 24?h after HS, only
four mediators were increased (KC, RANTES, IL-6, and
G-CSF;p<0.05 vs. na?ve; see Fig. 1L). Notably, the median level of
IL-6 was higher at 2?h after HS than at 24?h. Similarly, median levels of KC and RANTES were highe r at 2 than at 24?h, despite only being significantly increased at the later time point.
After CCI alone, four chemokines, four cytokines, and one growth factor were increased at 2?h (KC, MCP-1, MIP-1α, IP-10, IL-5, IL-6, IL-10, TNF-α, and G-CSF, respectively; p<0.05 vs. na?ve) and two chemokines, one cytokine, and one growth factor were increased at 24?h (eotaxin, MIP-1α, IL-6, and G-CSF, respectively;p<0.05 vs.
na?ve). Notably, median levels of MIP-1α and IL-6 were highest at the earlier time point.
Few mediators were increased in serum after CCI+HS. IL-6, IL-5, and IL-10 (Fig. 1F,I,J) were increased at 2?h (p<0.05 vs. na?ve), and IL-5, RANTES, and G-CSF were increased at 24?h (p<0.05 vs. na?ve). Median levels of IL-6 were lower after CCI+HS than at the same time point after CCI. The level of anti-inflammatory IL-10 at 2?h after HS+CCI was statistically significantly increased versus CCI alone, the only instance of a mediator in any tissue type being significantly different in serum after the two injury mechanisms. Specifically, IL-10 levels were >200?pg/mL in CCI+HS, with contrasting levels of ～2?pg/mL in CCI and ～7?pg/mL in HS (Fig. 1J).
Of note, to ensure that IL-10 production in the shed blood that was stored on the benchtop and reinfused after resuscitation was not contributing to the increase in serum levels after CCI+HS, we assessed IL-10 levels by ELISA (Cell Sciences, Inc., Canton, MA) in shed blood maintained at room temperature on the benchtop. In 5 mice, IL-10 levels in serum were assessed from blood samples obtained either immediately or after 90 min of storage. Levels were 7.83±0.85 and 10.25±0.61?pg/mL, respectively. Thus, reinfusion of shed blood cannot explain the marked increase in serum IL-10 level at 2?h after CCI+HS.
Go to: Discussion
There are three key findings from this study. First, the addition of HS to CCI led to a reduction in proinflammatory response and uniquely increased IL-10 in serum. Second, the inflammatory mediator response in the brain after CCI in our mouse model exhibited a robust chemokine response. Third, surprisingly, HS did not lead to an exacerbation of the inflammatory mediator response to CCI in the mouse brain.
Hemorrhagic shock shifts serum cytokine response to an anti-inflammatory profile after traumatic brain injury
The serum inflammatory profile after CCI+HS was more
anti-inflammatory than after CCI. IL-10, a potent anti-inflammatory
cytokine, was significantly increased 2?h after CCI+HS versus CCI alone. Six proinflammatory mediators (IP-10, TNF-α, KC, MCP-1, eotaxin, and MIP-1α) were all increased after CCI, but not CCI+HS, and IL-6 median levels were lower after combined injury.
To our knowledge, this is the first study of the effect of HS on inflammatory response to experimental TBI, though earlier studies have analyzed the effect of systemic trauma. Similar to our results, in a model of TBI with and without tibia fracture,42 serum levels of
IL-10 were highest in combined injury; however, higher levels of
IL-6 were also observed in that combined injury. Data from two clinical studies also support our findings. Shiozaki and
colleagues29 measured levels of proinflammatory (IL-1β and TNF-α) and anti-inflammatory (IL-10, IL-1 receptor antagonist, and soluble TNF receptor I) mediators in CSF and serum in adults with either TBI or TBI plus systemic trauma. Mirroring our findings, systemic trauma did not affect brain inflammatory response, but increased serum levels of anti-inflammatory mediators. Hensler and colleagues19 found increased serum levels of IL-10 and proinflammatory mediators in TBI patients with systemic injury. These similarities between murine and human inflammatory responses in serum to TBI+HS are very important in light of recent evidence that inflammatory responses to various stimuli differ
between humans and mice.43 The findings in our report and these clinical studies are thus similar despite our model superimposing HS on TBI, not systemic trauma. Clinical work has shown that traumatic HS is associated with increased levels of IL-6 and IL-10 versus controls, whereas nontraumatic HS is only associated with increased levels of IL-10.44 Thus, systemic tissue injury may mediate proinflammatory mediator response, whereas HS mediates
anti-inflammatory response. Further studies are needed to fully understand the mechanism by which IL-10 is increased in HS, though Kupffer cell damage45 and catecholamine release46 likely play a role.
The shift to a systemic anti-inflammatory cytokine profile after traumatic brain injury Though our CCI+HS model has an unfavorable outcome versus CCI alone,8 the effect of the anti-inflammatory phenotype observed in combined injury requires further study. Anti-inflammatory mediators may worsen outcome through induction of an immunoparalyzed state, which has been suggested to contribute to the high prevalence of nosocomial infections in TBI
patients.47 However, several studies in experimental TBI suggest that an anti-inflammatory shift could be neuroprotective. Systemic infusion of IL-10 reduced neutrophil influx into the brain after CCI in rats48 and improved functional outcome after fluid percussion
injury.49 Attenuation of inflammation in the brain by IL-10, with resultant neuroprotection, has been suggested.50 Thus, the
anti-inflammatory profile in CCI+HS could attenuate injury induced by HS. However, we have not studied the effect of IL-10 infusion on brain or serum cytokine levels or used the IL-10 KO in our combined injury paradigm.
Marked chemokine response in brain after controlled cortical impact
Previous studies of clinical TBI that measured multiple inflammatory mediators in CSF,12 brain tissue,15 and brain microdialysate18 all detected increased levels of several cytokines, but fewer chemokines, contrasting with our finding of a chemokine predominance. However, our findings are similar to the report of Bye and colleagues, who also used Multiplex to assess brain tissue cytokines and chemokines at 4?h after diffuse TBI produced with the Marmarou model in rats. 51 In that study, four cytokines were increased as were four chemokines, with a similar chemokine pattern as that observed in our report in CCI alone. Our results also mirror the work of Semple and colleagues, using a similar Multiplex approach in a closed TBI model in mice.52 In contrast, cytokine predominance was observed using Multiplex in a stab wound model of TBI in mice.31 These distinct patterns may be the result of differences between penetrating and blunt TBI, including injury severity and sampling location. Though
these differences are intriguing, the vast majority of clinical TBI is secondary to closed-head injury.53 Also, other important factors could produce differences between clinical studies and our model, such as 1) assessment of RNA versus protein, 2) use of microdialysis or CSF versus brain tissue, 3) timing of the evaluation, and 4) brain regions studied.
Nevertheless, the importance of chemokine response in TBI has been gaining appreciation. Glabinski and colleagues16 reported acute increases in MCP-1 after TBI in mice. In a clinical study, Kossmann and colleagues54reported that the chemokine, IL-8, was markedly increased in CSF after severe TBI (sTBI), which has been corroborated.33 A murine analog of IL-8, KC, was increased in the brain in our model. Studies in experimental TBI show selective chemokine messenger RNA expression55–57 as well as the presence of MIP-2,25,57 IP-10,21 KC,32 and MIP-1α25 in the brain. IP-10 is increased in human cerebral contusions after TBI.30 Buttram and colleagues12 found increased CSF levels of MIP-1α and IL-8 after sTBI in children. Roles have been described for IP-10 in microglial recruitment,58 MCP-1, and MIP-1α in Wallerian
degeneration,59 MCP-1 in macrophage recruitment,60 IP-10 in T-cell recruitment,21 KC in neutrophil recruitment,61 and C-X3-C chemokine ligand 1 in leukocyte recruitment.62 Assessment of these