Blast exposure elicits blood-brain barrier disruption and repair mediated by tight junction integrity and nitric oxide dependent processes

Mild blast-induced traumatic brain injury (TBI) is associated with blood-brain barrier (BBB) disruption. However, the mechanisms whereby blast disrupts BBB integrity are not well understood. To address this issue BBB permeability to peripherally injected 14C-sucrose and 99mTc-albumin was quantified in ten brain regions at time points ranging from 0.25 to 72 hours. In mice, repetitive (2X) blast provoked BBB permeability to 14C-sucrose that persisted in specific brain regions from 0.25 to 72 hours. However, 99mTc-albumin revealed biphasic BBB disruption (open-closed-open) over the same interval, which was most pronounced in frontal cortex and hippocampus. This indicates that blast initiates interacting BBB disruption and reparative processes in specific brain regions. Further investigation of delayed (72 hour) BBB disruption revealed that claudin-5 (CLD5) expression was disrupted specifically in the hippocampus, but not in dorsal striatum, a brain region that showed no blast-induced BBB permeability to sucrose or albumin. In addition, we found that delayed BBB permeability and disrupted CLD5 expression were blocked by the nitric oxide synthase inhibitor N(G)-nitro-L-arginine methyl ester (L-NAME). These data argue that latent nitric oxide-dependent signaling pathways initiate processes that result in delayed BBB disruption, which are manifested in a brain-region specific manner.

In this study, we used a sensitive and highly quantifiable blood-borne radiolabeled tracer approach to assess the time course and magnitude of BBB disruption in multiple brain regions after blast-induced mTBI in mice. We chose time points to assess apparent mechanical damage (0.25 h), repair (3 and 24 h), and a hypothesized secondary process (72 h). We also assessed the effects of blast on tight junction integrity and the roles of nitric oxide and prostaglandin in mediating blast-induced BBB disruption.

Results
Blast exposure induces aberrant BBB permeability to radiolabeled sucrose and albumin. As an initial examination of the effects of blast exposure on BBB integrity, using well established methods 25 we quantified specific BBB permeability to peripherally co-administered 14 C-sucrose and 99m Tc-albumin that under normal conditions cross the BBB poorly [30][31][32] . These molecules are often used to measure BBB permeability to small (~0.34 kDa) and larger (~66.44 kDa) blood-borne molecules, as well as for assessing aspects of paracellular and transcellular BBB disruption, respectively 33,34 .
In keeping with previous reports 23,24,35 , blast overpressures were produced using a well-established pneumatic shock tube approach that accurately replicates battlefield-relevant shock waves generated by detonation of high explosives in the open field ( Supplementary Fig. S1). A single blast exposure provoked BBB disruption as measured by 14 C-sucrose entry into the CNS at 0.25, 24, and 72 h post-blast ( Fig. 1a; p ≤ 0.01). Pair-wise Dunnett's post hoc analysis confirmed a significant increase in 14 C-sucrose CNS entry at 0.25 h (p ≤ 0.001, n = 8), 24 h (p ≤ 0.05; n = 8), and 72 h (p ≤ 0.05; n = 9) post-blast compared to sham controls (n = 7). The same mice also exhibited BBB permeability to 99m Tc-albumin ( Fig. 1b; p ≤ 0.05). However, post hoc analysis revealed that blast-induced 99m Tc-albumin CNS entry was significantly elevated only at the shortest, 0.25 h time point (p ≤ 0.05) compared to sham controls. These results are in keeping with a level of BBB disruption due to a mild to moderate blast exposure that elicits both BBB injury and neuro-reparative processes. For the smaller sucrose molecules, BBB disruption persisted for at least 72 h, while aberrant BBB permeability to the larger albumin molecules was restored to normal levels within 24 h.
To assess the effects of repetitive injury, animals received two blasts (5 min inter-blast interval) that were of equivalent intensity to the 1X blast-exposed mice. These exposures occurred during a single, brief isoflurane Figure 1. Blood-brain barrier disruption to large and small molecules following blast. (a) A significant increase in brain/serum ratios of 14 C-sucrose, (b) and 99m Tc-albumin was observed in whole brain after 1X blast. (c) A significant increase in 14  treatment in order to minimize the number of anesthesia sessions needed for repetitive exposure (See Methods). 2X blast exposure evoked BBB permeability to 14 C-sucrose that persisted for at least 72 h ( Fig. 1c; p ≤ 0.001). Post hoc analysis confirmed a significant increase in 14 C-sucrose at 0.25 h (p ≤ 0.001; n = 5), 24 h (p ≤ 0.01; n = 11) and 72 h (p ≤ 0.05; n = 13) post-blast compared to sham controls (n = 22). 2X blast also induced BBB permeability to 99m Tc-albumin ( Fig. 1d; p ≤ 0.001). Post hoc analysis confirmed a significant increase in 99m Tc-albumin permeability at 0.25 h (p ≤ 0.001), as well as at 72 h post-blast (p ≤ 0.05) compared to sham controls.
The 2X blast results demonstrated that the CNS is able to restore normal BBB permeability to the larger albumin molecules within 24 h, while blast-induced BBB disruption affecting the sucrose radiotracer persisted for at least 72 h. However, unlike single blast, repetitive blast exposure initiated an additional latent process resulting in a delayed, second round of BBB permeability to 99m Tc-albumin that emerged by 72 h post-blast (Fig. 1d). Currently, little is understood regarding the mechanisms by which mild blast-induced neurotrauma gives rise to delayed BBB disruption. Therefore, we focused on the 72 h opening to 99m Tc-albumin occurring after 2X blast.
Blast elicits brain region-specific BBB disruption. To investigate the vulnerability of distinct brain regions to repetitive blast-induced BBB disruption, we quantified aberrant 14 C-sucrose and 99m Tc-albumin BBB permeability in ten brain regions: frontal cortex, occipital cortex, parietal cortex, striatum, midbrain, thalamus, cerebellum, pons/medulla (brainstem), and hippocampus. Eight of the ten anatomical regions displayed significant blast-induced permeability to blood-borne 14 C-sucrose within 0.25 h compared to sham controls (Table 1). Under these conditions, blast exposure provoked significant BBB disruption that persisted in four of ten brain regions at 3 h (p ≤ 0.05), three of ten regions at 24 h (p ≤ 0.05), and four of ten regions at 72 h (p ≤ 0.05). Of note, the hippocampus and frontal cortex exhibited biphasic permeability to 14 C-sucorse at 0.25 and 72 h, but not at 24 h. 2X blast exposure significantly increased BBB permeability to 99m Tc-albumin in the hippocampus, frontal cortex, and pons/medulla compared to sham controls (Table 2). For these brain regions, normal BBB integrity against 99m Tc-albumin entry was restored within 3 to 24 h; and was followed by a second cycle of BBB disruption at 72 h. Similar to the sucrose data (Table 1), the hippocampus and frontal cortex exhibited biphasic permeability to albumin at 0.25 and 72 h. Of special note was an increase in BBB opening at 72 h that occurred in the Whole Brain 87 ± 15 *** 34 ± 11 * 36 ± 11 * 28 ± 11 * Table 1. Temporal profile of regional BBB permeability to 14 C-sucrose after 2X blast mTBI. Blood-brain barrier (BBB); mild traumatic brain injury (mTBI). Values represent mean difference ± SEM normalized to sham control animals (n = 22). One-way ANOVA; post-hoc Dunnett's; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.  Table 2. Temporal profile of regional BBB permeability to 99m Tc-albumin after 2X blast mTBI. Blood-brain barrier (BBB); mild traumatic brain injury (mTBI). Values represent mean difference ± SEM normalized to sham control animals (n = 22). One-way ANOVA; post-hoc Dunnett's; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. pons-medulla (and whole brain) for 99m Tc-albumin but not for 14 C-sucrose (Table 2). Thus, blast exposure elicited a multiphase CNS response involving anatomically distinct manifestations of BBB disruption, initiating both immediate and latent processes that differentially modulated permeability to smaller and larger blood-borne molecules.
Delayed BBB disruption is associated with brain region-specific disturbances in claudin-5 expression. One mechanism by which delayed BBB disruption could have occurred is through altered expression of the brain endothelial cell (BEC) tight junction protein, claudin-5 (CLD5). Our group has previously shown aberrant CLD5 expression in the acute phase following blast mTBI 23 . If disturbed CLD5 expression plays a role in mediating delayed blast-induced BBB disruption, we expected to find both a time and brain region-specific disturbance in the pattern of CLD5 expression mirroring the functional BBB integrity pattern seen above. The hippocampus displayed biphasic permeability to both 14 C-sucrose and 99m Tc-albumin following blast exposure (Tables 1 and 2). Conversely, the striatum displayed the least vulnerability to blast-induced BBB disruption (Tables 1 and 2). Therefore, we focused our attention on these regions. CLD5 immunoreactivity in hippocampal CA1 appeared markedly reduced 72 h post-blast (n = 4/4) compared to sham controls ( Fig. 2a-d); n = 0/4). The intensity of CA1 CLD5 immunoreactivity was quantified using Imaris software (see Methods) by an operator who was blind to the experimental conditions. 2X blast provoked a 43.9% loss of CLD5 immunoreactivity in the CA1 region of hippocampus at 72 h compared to sham controls ( Fig. 2e; p ≤ 0.05). In contrast to the hippocampus, CLD5 immunoreactivity in the dorsal striatum of these same mice used was not significantly altered by 2X blast exposure at any time point (Fig. 2f-j).
The loss of CLD5 immunoreactivity in hippocampus at 72 h could be due either to a loss of overall CLD5 protein levels or to mislocalization/diffusion of CLD5 away from the prominent tight junction domains where CLD5 is highly concentrated. Western blot analysis of CLD5 levels revealed no significant changes in hippocampal, or striatal protein lysates of 2X blast mice versus sham controls ( Supplementary Fig. S2). Thus, as has been reported previously regarding CLD5 expression after brain injury [36][37][38] , it is more likely blast caused extant CLD5 proteins to mislocalize and diffuse within the endothelial cytoplasm, making them more difficult to detect by immunofluorescence. Taken together these results indicate that blast exposure caused brain region-specific disturbances in the expression of tight junction-localized CLD5. Interestingly, we found no changes in CLD5 immunoreactivity at 0.25 h post-blast, even though there was a significant increase in blast-induced permeability to 14 C-sucrose and 99m Tc-albumin ( Fig. 1, Tables 1 and 2). This suggests that the mechanisms mediating BBB disruption immediately after blast are distinct from second phase BBB disruptions arising 72 h later.
Blast causes perivascular reactive gliosis in association with aberrant claudin-5 expression. As with hippocampus, a delayed increase in permeability to both 14 C-sucrose and 99m Tc-albumin at 72 h post-blast was also observed in frontal cortex (Tables 1 and 2). In keeping with this, we similarly observed decreased CLD5 immunoreactivity in frontal cortex of blast-exposed mice (  (an endothelial cell-specific marker) showed that BECs appeared intact ( Fig. 3j and n); even though their CLD5 immunoreactivity was markedly reduced in blast-exposed mice. Additional GFAP immunostaining confirmed perivascular reactive gliosis at 72 h post-blast (Fig. 3k,l; n = 4/4) compared to sham controls (Fig. 3o,p; n = 0/4). In spite of robust morphological disturbances in the BECs and perivascular astrocytes, we found no overt alterations in neuronal somato-dendritc morphology at 72 hours post-blast ( Supplementary Fig. S3). This is broadly in keeping with previous findings in this animal model in other brain regions where it takes a month, or more for significant neuronal loss to become evident 24 .
Taken together, these data confirm the potential significance of disturbed CLD5 expression during delayed BBB disruption and highlight the comparative vulnerability of the BBB to blast exposure. These data also show that aberrant CLD5 expression is closely associated with reactive gliosis, thus suggesting that perivascular neuroinflammation may play a role in delayed aberrant CLD5 expression provoked by blast exposure.

Effects of cyclooxygenase or nitric oxide synthase inhibition on delayed BBB disruption.
Broadly speaking, numerous findings point to at least two primary neuroinflammatory signaling cascades, the prostaglandin and/or nitric oxide pathways, which could mediate aspects of blast-induced delayed BBB   [39][40][41] . We first tested the potential significance of prostaglandin-related signaling pathways by inhibiting cyclooxygenase with indomethacin (INDO). To accomplish this, we administered INDO at 48, 54, and 71 h post-blast and then measured BBB permeability to 14 C-sucrose and 99m Tc-albumin at 72 h. As expected, blast exposed mice (n = 5) exhibited a significant increase of 77.4% in 14 C-sucrose permeability at 72 h ( Fig. 4a; p ≤ 0.05) compared to INDO sham controls (n = 5). However, INDO treatment did not significantly reduce the effects of blast on BBB permeability to 14 C-sucrose. Blast exposure also increased BBB permeability to 99m Tc-albumin (Fig. 4b), with a significant 156% increase in 99m Tc-albumin at 72 h post-blast (p ≤ 0.05) compared to INDO sham controls. As with 14 C-sucrose, INDO treatment only partially reduced BBB permeability to 99m Tc-albumin. These results suggest that under these experimental conditions prostaglandin-related pathways play a limited role in mediating second phase BBB disruption caused by blast exposure.
We next treated mice with N(G)-nitro-L-arginine methyl ester (L-NAME), a well-characterized pan-nitric oxide synthase (NOS) inhibitor [42][43][44][45][46] . L-NAME was injected at 48, 54, and 71 h post-blast or sham treatment and brain BBB permeability was assessed at 72 h. Blast exposed mice (n = 5) exhibited a significant increase of 55% in 14 C-sucrose permeability at 72 h ( Fig. 4c; p ≤ 0.01) compared to L-NAME sham controls (n = 5). Moreover, L-NAME significantly reduced 14 C-sucrose permeability by 35.8% compared to vehicle-treated/blast exposed animals ( Fig. 4c; p ≤ 0.05). 2X blast also increased 99m Tc-albumin permeability at 72 h (Fig. 4d). A significant 93% increase in 99m Tc-albumin was measured in whole brain at 72 h post-blast (p ≤ 0.05) compared to L-NAME sham controls. L-NAME significantly reduced 99m Tc-albumin permeability by 96% compared to vehicle-treated/ blast exposed animals ( Fig. 4d; p ≤ 0.05). These data argue that the mechanisms governing delayed blast-induced BBB disruption involve one or more nitric oxide signaling pathways that are engaged over a period of 48 to 72 h following injury.
L-NAME reverses blast-induced disturbances in claudin-5 expression. The ability of L-NAME to rescue delayed blast-induced BBB disruption led us to hypothesize that L-NAME may block blast-induced disturbances in CLD5 expression. Representative CLD5 confocal microscopy results show that CA1 hippocampal CLD5 immunoreactivity was reduced in 72 h post-blast mice (n = 4/4) compared to a sham controls (Fig. 5a,b); n = 0/4). CLD5 immunostaining appeared normal in blast-exposed animals treated with L-NAME ( Fig. 5c; n = 4/4). Quantification of CLD5 immunoreactivity by a condition-blind observer revealed a significant 54% decrease in CLD5 immunofluorescence intensity in 72 h post-blast mice compared to sham controls ( Fig. 5d; p ≤ 0.05). L-NAME administration to blast mice significantly rescued CLD5 immunofluorescence by 44% compared to vehicle-treated/blast exposed animals ( Fig. 5d; p ≤ 0.05).
We also examined hippocampal CA1 CLD5 immunostaining in blast exposed mice treated with INDO. In contrast to the effects of L-NAME, INDO treatment failed to normalize CLD5 expression in hippocampus ( Supplementary Fig. S4). Overall, these data significantly strengthen the idea that nitric oxide-related signaling pathways mediate second-phase BBB disruption and CLD5 disturbances caused by blast.

L-Name attenuates TNFα in BBB-vulnerable brain regions after blast.
TNFα is known to alter CLD5 expression 47 . To test whether blast induces brain region-specific changes in TNFα, we measured TNFα in hippocampus and striatum at 72 h. Hippocampal brain lysates from blast exposed mice (n = 5) exhibited a significant increase of 276.2 ± 118.4 (mean ± SEM) in TNFα protein expression (pg/mg) compared to sham controls ( Fig. 6a; p ≤ 0.05; n = 5). No significant differences were observed in the striatum of the same mice (Fig. 6a).
A complex interplay exists between nitric oxide and TNFα [48][49][50][51] ; therefore, in a separate experiment we examined the effects of L-NAME treatment on blast-induced TNFα expression in the hippocampus. A significant difference in TNFα protein expression was observed at 72 h after 2X blast ( Fig. 6b; p ≤ 0.05). Blast exposed mice (n = 5) exhibited an increase of 168.8 ± 82.9 in TNFα protein expression compared to sham controls (n = 5), and an increase of 235.3 ± 63.9 compared to L-NAME sham controls (n = 5). L-NAME administration after blast (n = 5) significantly decreased TNFα protein expression by 219.8 ± 65.7 compared to vehicle-treated/blast exposed animals ( Fig. 6b; p ≤ 0.05), suggesting that the mechanism by which blast provokes delayed BBB disruption involves nitric oxide-mediated TNFα elevation. Moreover, these data also suggest that the mechanistic pathways regulated by NOS and TNFα may play a role in accounting for differences in the vulnerability of different brain regions (in this instance, the hippocampus versus the striatum) to delayed cycles of blast-induced BBB disruption.

Discussion
In keeping with other reports 21,23,52 , we found that a single blast exposure provokes rapid, transient BBB disruption, permitting aberrant entry of both sucrose and albumin radiotracers into multiple brain regions. By 24 h, BBB integrity was restored to albumin, but not sucrose. After two consecutive (2X) mild blast exposures, the BBB reopens to albumin at 72 hours. We found that tight junction disintegrity correlates anatomically and temporally to this second phase of BBB disruption. These data report that blast exposure causes immediate and sustained brain permeability to small molecules, such as sucrose and is accompanied by additional latent processes that elicit delayed, brain region-specific permeability to larger molecules. Complementary findings of biphasic BBB opening has been reported in feline 53 and murine models of stroke 54,55 . Furthermore, biphasic BBB opening has been observed in rats after controlled cortical impact TBI 56,57 .
Taken together, these findings suggest that several different forms of brain trauma set in motion multiple physiological and/or pathological processes with interacting phases of BBB repair and delayed disruption. Thus, as is well-established in managing the pathologic processes attending stroke, these results highlight the importance of elucidating the specific pathophysiological cascades evoked by blast, as well as the timing of these cascades. Both considerations are crucial in the search for appropriate therapeutic targets; and in establishing when such interventions would likely to be of benefit rather than potential harm.
Previously we have reported that inhibiting prostaglandin synthesis with indomethacin can reduce lipopolysaccharide-induced BBB permeability 39 . Other groups have shown that inhibition of nitric oxide production with L-NAME can reduce stroke-induced BBB permeability 58 . In this study, we show that secondary BBB disruption following blast is significantly attenuated by pan-specific NOS signaling inhibition with L-NAME. The results suggest that NOS pathways play an important role in mediating tight junction integrity after neurotrauma and have potential therapeutic implications for reducing BBB dysfunction and subsequent pathophysiology after blast injury. It has been shown that tight junction remodeling can occur after BEC perturbation [59][60][61][62][63] , which could explain why albumin CNS influx is attenuated by 24 hours. Initial BBB opening is likely dominated by mechanical disruption of endothelial membranes caused by a putative surge of vascular overpressure as previously suggested 64 and is consistent with reported vascular pathology 12 . It may also be possible that the BBB continues to oscillate between open and closed into the subacute phase, and this may be due to ongoing aberrant nitric oxide production induced by blast.
It is possible that NO production could subserve BEC compensatory mechanisms that cause tight junction protein mislocalization, and upregulates caveolin-1-dependent transcytosis 46 . This could partially explain why L-NAME reduces albumin influx, suggesting that future studies to address the causal mechanism(s) by which blast opens the BBB to larger molecules. The nitric oxide synthase pathway is activated after blast TBI 52,65 and has been implicated in neurodegeneration associated with TBI 66,67 . Moreover, nitric oxide inhibition has been shown to improve sensorimotor function 68 , reduce neuronal cell death 69 , and prevent edema formation after TBI 70 , supporting the importance of nitric oxide pathways in response to brain injury 71 .
Tight junction protein expression has been shown to change after blast injury 22,23 and correlates with TBI-induced BBB disruption 64,72 . Our findings of aberrant CLD5 expression 72 hours after blast, without a significant change in CLD5 protein expression as measured by immunoblotting, could suggest that CLD5 may not be degraded as a result of blast, but rather diffused within BECs. Such redistribution has been demonstrated in rodent stroke models 36,37 and in primary BECs after blast injury 38 . CLD5 −/− mice exhibited BBB disruption to small molecules, such as sucrose, with no effect on larger molecules such as albumin 73 . This raises the possibility that aberrant CLD5 expression alone may not fully account for the delayed phase of blast-induced albumin influx into the CNS.
Perivascular astrocytic end-feet surround the abluminal side of BECs and are necessary for the formation and maintenance of the tight junctional barrier and BBB integrity 74,75 . In an animal model of CNS inflammatory disease, endothelial growth factors released by activated astrocytes have been shown to reduce tight junction integrity, inducing BBB breakdown 76,77 . Thus, our findings of marked perivascular reactive gliosis and brain region-specific increases in TNFα argue that intercellular interactions between astrocytic neuroimmune responses and the BECs they enshroud, play important roles in mediating the anatomical and temporal manifestations of blast-induced BBB disruption.
Veterans with mild TBI show structural and functional CNS abnormalities [78][79][80][81][82] . Future studies will focus on vulnerable brain regions at prolonged time points to assess the pathophysiology accompanying BBB disruption. Vulnerable regions of the BBB may be those that are more susceptible to chronic low-grade inflammation and associated with long-term neurodegeneration. The results herein build upon very recent findings of regional blast-induced BBB disruption 83 also suggest that inhibiting nitric oxide may block the cyclic nature of BBB disruption such that carefully targeted and timed therapies may help prevent long-term blast-related pathophysiology, cognitive, and behavioral disturbances.

Methods
Animals. Male C57BL/6 (n = 189) mice (Jackson Laboratories) aged 8-12 weeks (27-32 g) were used for this study. All mice had ad libitum access to food and water, were maintained under a 12 h day/night cycle, and group housed. Animals were randomly assigned to sham, drug, or mTBI groups. All animals were housed and handled in accordance with protocols approved by the Veterans Affairs Puget Sound Health Care System's Institutional Animal Care and Use Committee (IACUC) and all experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All experiments reported are in compliance with the ARRIVE guidelines.
Blast exposure. All mice were allowed to acclimatize to the animal facility for at least one week prior to blast exposure. In preparation for blast exposure, animals were first anesthetized with 5% isoflurane, followed by maintenance with 2% isoflurane using a non-rebreathing anesthesia machine at a flow rate of 1 L/min oxygen. A flexible custom facemask designed to fit over the nose and mouth was attached to the mouse harness and provided anesthesia during blast exposure or sham treatment. In keeping with well-established procedures 23,35 , mice were positioned in the shock tube with their ventral body surface facing the oncoming shock wave as previously described. A pneumatic shock tube was used to expose mice to either a single blast exposure over the course of 6-7 min, or two consecutive blast exposures over the course of 12-14 min. Two consecutive blast exposures consistently produced a significant effect at 72 h (p ≤ 0.05), compared to a more variable response after a single blast. After the first blast exposure, the mouse was removed from the shock tube and re-adjusted in the harness prior to the second exposure. Sham control animals were mounted in the restraint harness and held under anesthesia for the same amount of time as the respective blast-exposed mice.
The blast exposure used in these experiments had a peak static pressure of 20.56 psi (±0.21), a positive phase duration of 5.37 ms (±0.070), and a shock wave velocity of 1.4 Mach. In keeping with a mild to moderate blast exposure 176/189 animals in this study survived the blast exposure (1X or 2X) and by inspection displayed normal ambulation, visually guided grasping responses when elevated by the tail near a foothold, and grooming behavior within 1 hour after exposure.
Radiolabeled tracer preparation. Following established procedures 39 , albumin (Sigma, St. Louis MO) was labeled with 99m Tc (GE Healthcare, Piscataway, NJ). A mixture of 240 mg/ml stannous tartrate and 1 mg/ml albumin was adjusted to pH 3.0 with HCl. One millicurie of 99m Tc-NaOH 4 was added to this mixture and allowed to incubate for 20 min. The 99m Tc-albumin was purified on a column of G-10 Sephadex (GE Healthcare) in 0.1 ml fractions of phosphate buffer (0.25 M). Radioactivity in the purified 99m Tc-albumin peak was more than 90% acid precipitable in an equal volume of 1% bovine serum albumin (BSA) and trichloroacetic acid (30%). 5 × 10 6 cpm/ mouse of purified 99m Tc-albumin fraction was combined with 1 × 10 7 dpm/mouse of 14 C-sucrose (Perkin Elmer, Waltman, MA) in a final volume (0.2 ml/mouse) of lactated Ringer's solution containing 1% BSA.

Radiolabeled tracer injections.
At 0.25, 24, and 72 h after single blast, and at 0.25, 3, 24, and 72 h after 2X blast, or after time-matched sham procedures, mice were anesthetized with urethane (4 g/kg; 0.2 ml; ip), and the jugular veins exposed. The mice were given an intravenous co-injection of 99m Tc-albumin (5 × 10 6 cpm) + 14 C-sucrose (1 × 10 7 dpm) in 0.2 ml of lactated Ringer's solution with 1% BSA for 10 min. Blood was collected from a cut in the descending abdominal aorta. The vascular space of the brain was then washed free of blood by opening the thorax, clamping the descending thoracic aorta, severing both jugular veins, and perfusing 20 ml of lactated Ringer's solution through the left ventricle of the heart in less than 1 min. After washout, the mouse was immediately decapitated, the brain was removed, and dissected into ten brain regions that were individually weighed. Olfactory bulb was excluded due to variations in collection efficiency. Brains with visible blood after washout were excluded from analysis (n = 6). Serum was obtained by centrifuging the carotid artery blood for 10 min at 4000 × g. Levels of 99m Tc radioactivity in the serum and brain regions were determined in a gamma counter. The same brain regions and sera were then solubilized, and the level of 14 C radioactivity was determined in a beta counter. Brain tissue radioactivity was calculated by dividing the cpm (gamma), or dpm (beta) in the brain region by the weight of the brain region. Serum radioactivity was calculated by dividing the cpm, or dpm in the serum by the milliliters of serum counted. The brain tissue radioactivity was then divided by the corresponding serum radioactivity and the results given in units of microliters per gram of brain tissue. For shams, no significant differences were observed among the brain/serum ratios at the various points and so were combined and used as a single reference value (n = 22).
Confocal microscopy. Dissected mouse brains were immediately post-fixed in 4% paraformaldehyde in PBS at 4 °C and then equilibrated with cryoprotectant (30% sucrose in PBS) overnight at 4 °C. Sagittal sections embedded in OCT (Tissue-Tek, Torrance, CA), and after decay of 99m Tc, were cut at 50 µm thickness from bisected brains with a CM1850UV cryostat (Leica, Buffalo Grove, IL). Antigen retrieval was performed with 50 mM sodium citrate (pH 9.0) and then heating at 80 °C for 30 min. Immunostaining was performed according to our previously published methods 14 . Floating tissue sections were cover slipped with a drop of Prolong ® Gold Antifade Reagent with or without DAPI (Invitrogen, Eugene, OR). The following antibodies were applied overnight at 4 °C: mouse monoclonal anti-claudin-5 labeled with Alexa 488 (Thermo, Rockford, IL; 1:100), rabbit monoclonal anti-microtubule-associated protein 2 (MAP2; Cell Signaling, Danvers, MA; 1:2000), rabbit polyclonal anti-IBA1 (Wako, Osaka, Japan; 1:1000), chicken polyclonal anti-GFAP (Millipore, Billerica, MA; 1:1000), and rabbit polyclonal CD31 (Abcam, Cambridge, MA; 1:500). Goat secondary antibodies labeled with Alexa 647 and Cy3 were applied for 2 h (Jackson Immunoresearch, West Grove, PA; 1:1000). Confocal microscopy was performed with a Leica TCS SP5 II microscope with Leica objectives (10X and 20X with 0.3 and 0.7 numerical aperture, respectively). Confocal microscopic images were acquired with the Leica Application Suite and processed using image adjustments limited only to linear contrast and brightness adjustments applied identically to data from blast-and sham-treated animals in each experiment.
Immunoblotting. Left brain hemispheres from mice in the 99m Tc-albumin experiments were immediately flash frozen in liquid nitrogen and stored at −80 °C for ten 99m Tc-albumin half-lives (~60 h). Hippocampus and striatum were dissected and sonicated in radioimmunoprecipitation assay (RIPA) buffer with protease/phosphatase inhibitors (Thermo; Rockford, IL 1:100). Samples were then centrifuged for 15 min at 12000 × g at 4 °C and supernatant collected and standardized with a bicinchoninic acid (BCA) assay. Brain extracts of equivalent total protein content were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (20 µg total protein/lane) under reducing conditions, transferred to nitrocellulose, and probed with rabbit polyclonal claudin-5 (Invitrogen, Eugene, OR; 1:1000) and rabbit monoclonal β-actin (Cell signaling, Danvers, MA; 1:10000) as loading controls. Blots were then probed with anti-rabbit horseradish peroxidase-conjugated secondary antibody (Jackson ImmunoResearch; 1:5000) and visualized by enhanced chemiluminescence (GE Healthcare, Piscataway, NJ). Optical densities of the immunoreactive protein bands were quantified using Image-Quant with background subtraction and normalization to loading control. Cytokine measurements. Tumor necrosis factor alpha (TNFα) levels were measured using a murine Bio-Plex Pro ™ assay kit (Bio-Rad Laboratories, Inc.; Hercules, CA) in brain lysates previously prepared for immunoblotting. All brain samples were diluted 1:3 in sample diluent provided in the kit and processed according to the manufacturer's protocol. Plates were read on a Bio-Plex 200 (Bio-Rad Laboratories, Inc., Hercules, CA). Values (pg/mg) were acquired by normalizing each sample to their respective BCA concentration obtained for immunoblotting.
Error bars represent the standard error of the mean (SEM). Comparison of more than two groups was made using one-way analysis of variance (ANOVA) followed by Dunnett's post-hoc test (for comparing time points to a collective sham group) or Newman-Keuls post-hoc test (for comparing experimental groups to sham control groups, and drug treatment groups). Data availability. Authors will provide published data and relevant protocols upon request to the Principal Investigator by phone or email.