It is well known that TBI can induce an inflammatory reaction [1, 27, 28]; nevertheless, data supporting a role for blood-borne white blood cells as a causal factor in secondary brain injury following head trauma are highly controversial. There are many possible ways in which leukocytes could contribute to secondary brain damage, including the release of free radicals, the activation of proteases, the production of pro-inflammatory chemokines and cytokines, alterations in cerebral blood flow, and/or increases in vascular permeability [28–31]. These mechanisms could be mediated either via the interaction of leukocytes with the endothelium or via the leukocytes that migrate into the tissue. In the current study, we investigated both possibilities by visualizing leukocytes both in the intravascular space using in vivo microscopy and in brain tissue using immunohistochemistry. Of note, all data were obtained from the traumatic penumbra, which is the area of the brain in which secondary brain damage occurs within the first 24 h following brain trauma [18, 21, 24, 32]. Our results demonstrate that both increased LEI and the formation of leukocyte-platelet aggregates are initiated in the microcirculation of the penumbra within the first few hours following TBI. Nevertheless, these effects seem to occur predominantly in superficial vessels, and only to a much lower degree in deeper microvessels. Additionally, leukocytes migrate into the post-TBI brain only after the tissue becomes necrotic. The inhibition of LEI had no effect on secondary lesion expansion following CCI.
Intravascular leukocyte-endothelium interactions in superficial vessels
Using intravital microscopy, we investigated and quantified LEI up to 13.5 h following CCI. Under normal physiological conditions, LEI was limited to some rolling leukocytes in venules, which is in line with observations published previously by our group  and others . Immediately following trauma, however, the number of rolling leukocytes increases significantly and - even more importantly - leukocytes begin to adhere to the venular endothelium. Most interestingly, these events occurred before secondary lesion expansion and hence could have potentially mediated secondary brain damage (for example, by disrupting the BBB or by initiating inflammatory cascades).
Elegant studies have suggested a correlation between post-trauma leukocyte accumulation in the brain and secondary brain damage [11, 12, 29, 34, 35]. However, in those studies, it was unclear whether the detrimental effect was caused exclusively by leukocyte accumulation or by an associated phenomenon such as leukocyte-endothelium adhesion initiating inflammatory cascades or an up-regulation of the adhesion mediator ICAM-1 and subsequent brain edema formation. Moreover, post-trauma ICAM-1 expression has been correlated with increased permeability of the BBB despite being independent of leukocyte accumulation in the brain [13–16, 30, 36]. In view of the possibility that ICAM-1 might play a leukocyte-independent role in secondary brain damage [2, 31], we used an antibody directed against a structure located on the leukocytes themselves to directly investigate the role of intravascular LEI following TBI. Hence, we used an anti-CD18 antibody that is directed against the beta unit of the lymphocyte function-associated antigen 1 (LFA-1; β-chain CD18 and α-chain CD11a), which binds to ICAM-1 and mediates (among other effects) leukocyte adhesion to the endothelium [2, 37, 38]. Thus, by blocking the interaction between LFA-1 and ICAM-1, we inhibited leukocyte-endothelium interactions. Using this antibody, we reduced leukocyte adherence by approximately two-thirds compared to an IgG control antibody. However, this did not affect the progression of secondary lesion expansion, indicating that leukocyte adherence to the cerebrovascular endothelium does not play an important role in the pathophysiology of secondary lesion expansion following CCI within the first 24 h. We focused on leukocyte adherence, which has been shown to initiate intracellular signaling and disruption of the BBB [2, 39, 40]. In contrast, rolling leukocytes interact with the endothelium only very briefly and have not been assigned a role in either initiating inflammatory cascades or opening the BBB.
Effect of aggregates on secondary brain damage following traumatic brain injury
To date, leukocyte-platelet aggregates have been reported to occur primarily in relation to endothelial stress, for example, due to increased levels of oxidized lipoprotein, inflammation, or diabetes [41–43]. Activated platelets up-regulate their expression of P-selectin, which then binds to its natural ligand, P-selectin-glycoprotein-ligand-1 (PSGL-1), on neutrophils and monocytes .
Using intravital microscopy, the formation of leukocyte-platelet aggregates was observed both after subarachnoid hemorrhage (SAH)  and after TBI . Following SAH, an antibody directed against P-selectin significantly reduced the formation of leukocyte-platelet aggregates and the adherence of aggregates to the endothelium . Similarly, in our study, inhibiting leukocyte adherence to the endothelium led to a reduction in the adherence of aggregates, which confirms that the aggregates were composed-at least in part-by leukocytes. Nevertheless, unlike the effect of inhibiting P-selectin following SAH, inhibiting LEI did not affect aggregate formation itself. Because aggregates were observed almost exclusively in venules, their effect on the cerebral microcirculation - and in particular, their contribution to vessel occlusion - might not be of primary importance. However, post-TBI microvessel occlusions in tissues outside of the brain have been reported, for example, in the lung . Despite the fact that the reduction of the adherence of aggregates to the venular endothelium did not affect secondary lesion expansion, it remains unclear whether directly inhibiting aggregate formation would have a beneficial systemic/pulmonary effect following TBI. Accordingly, future studies of the role of P-selectin in aggregate formation and the role of post-trauma aggregates both in the brain and in other organs would be needed to clarify these questions.
Intravascular leukocytes and aggregates in deeper brain levels
Although we also observed rolling and adherent leukocytes and aggregates at a depth of up to 250 μm in the brain using 2-photon microscopy, these events were much less prevalent than in superficial vessels. This effect becomes even more prominent when the average vessel volume is taken into account. The vessel volume investigated in superficial venules (being approximately 85,500 μm3) than in deeper regions of the brain (492,100, 173,200, and 125,800 μm3 at depths of 0-50, 100-150, and 200-250 μm, respectively).
Several factors may account for this observation. First of all, the diameter of deep vessels (which are primarily capillaries with some arterioles and venules) is much smaller than the diameter of superficial venules. According to the Bernoulli and Venturi Law, a decrease in diameter is accompanied by an increase in blood flow velocity. Therefore, the much faster blood flow velocity in deeper vessels might reduce LEI by increasing shear stress and minimizing cell-cell interactions. Secondly, the post-trauma inflammatory reaction is caused by the contusion itself and is therefore predominantly present in the vessels that drain blood from the site of injury. Hence, leukocyte activation and aggregate formation was scarcely present in arterioles but was present mainly in superficial venules and, to a limited degree, in deeper tissue, most likely in draining capillaries and post-capillary venules.
Leukocyte migration into vulnerable tissue
Our second aim was to investigate whether leukocytes accumulate in the region of interest (that is, the penumbra) before the tissue becomes necrotic. Because we quantified the expansion pattern of secondary brain damage surrounding the primary contusion at high temporal and spatial resolution [17, 21], we could compare the progression of neuronal cell death and leukocyte accumulation within the tissue over time. Following CCI, leukocytes accumulated predominantly in the contusion core. Although significant numbers of granulocytes and monocytes migrated into the tissue during the first 48 h, neither B lymphocytes nor T lymphocytes appeared in the brain. The peak accumulation of leukocytes occurred at 24 to 48 h post-trauma, which is in agreement with previous studies, including a clinical report  and several animal experiments using either CCI [15, 48] or a weight-drop paradigm [48–50]. Other white blood cells such as monocytes, macrophages, T cells, and B cells appear predominantly 5 to 6 days after trauma [47, 50, 51], which is in agreement with our results. In contrast, Fee et al. reported the presence of activated CD4-positive T cells at the site of traumatic injury within 24 h of aseptic cold injury (ACI) . The authors described a clear correlation between CD4-positive T lymphocytes in the tissue and increased post-traumatic brain damage. This early accumulation of T cells is in contrast to both our results and results from Holmin et al.  and might be explained - at least in part - by the differences in the trauma models and experimental methods that were used in the respective studies.
In our experiments, leukocytes migrated into the tissue only after neuronal cell death had occurred; thus, their accumulation does not seem to play a role in secondary lesion growth following CCI.
The role of leukocytes following controlled cortical impact versus other brain injury models
Leukocyte-endothelium adherence and the subsequent migration of leukocytes are known to play a role in secondary brain injury following stroke. This role has been studied extensively, particularly with respect to LFA-1 and Mac-1, which both use the CD18 binding site [6, 52–54]. The different result in our study can be most likely attributed to the post-injury differences in the pathophysiology that occur between stroke and TBI and by differences in the processes that are initiated selectively by the mechanical injury that results from CCI . Despite some similarities in the progression of secondary brain damage following both types of injury, the relative importance of individual factors seems to differ. For example, the inflammatory response, including leukocyte-endothelium interactions, clearly plays a more important role in secondary brain injury following stroke than it plays in the wake of TBI [1, 55].
Another study reported that leukocyte infiltration into the tissue was correlated to histological outcome after fluid percussion injury (FPI) . The accumulation of monocytes and macrophages within the brain was significantly reduced three days after FPI by the administration of anti-CD11 antibodies, which bind to a fraction of the CD11/CD18 integrin and thereby inhibit leukocyte-endothelium adhesion; most importantly, this reduction in leukocyte accumulation was accompanied by a reduction in lesion volume. However, it remains unclear whether this beneficial effect was due to a reduction in either leukocyte adherence and/or leukocyte migration into the tissue. Again, this discrepancy with our results could be attributed primarily to differences in kinetics initiated by the different experimental models . Firstly, FPI leads to diffuse brain injury that includes hemorrhagic contusions and reaches distant brain regions relative to the primary impact site [56–58]. In contrast, CCI produces a relatively restricted contusion with rapidly developing central necrosis [56, 59] in which the diffuse part  might be much less important. Hence, in TBI models that produce a much more extensive injury pattern (for example, CCI, in which the maximum lesion volume is reached within 24 h ), leukocytes may not contribute significantly to secondary lesion growth compared to models in which the initial injury is less severe and develops over longer periods of time (for example, FPI or ischemic brain injury). This is also supported by the finding that post-CCI leukocyte accumulation in the vulnerable tissue peaks at 24 to 48 h (as shown both in the current study and by others [15, 48]), which is 1 to 2 days earlier than the time of peak leukocyte accumulation following FPI [11, 51].
Following CCI, leukocytes begin to migrate into the injured brain only after neuronal cell death has already occurred. Moreover, leukocytes accumulate predominantly in the contusion core (that is, in tissue that is already necrotic), but barely in the traumatic penumbra, where secondary injury occurs. Inhibiting the adhesion of leukocytes and aggregates to the cerebrovascular endothelium does not reduce the progression of secondary lesion growth. Consequently, our data suggest that blood-borne leukocytes do not mediate secondary lesion expansion following contusional TBI.