All procedures on pigs were approved by the Johns Hopkins University Animal Care and Use Committee and by the Animal Care and Use Review Office of the US Army Medical Research and Materiel Command for Award Number W81XWH-19-C-0022 (Fort Detrick, MD). In conducting research using animals, the investigators adhered to the Animal Welfare Act Regulations and other Federal statutes relating to animals and experiments involving animals and the principles set forth in the current version of the Guide for the Care and Use of Laboratory Animals, National Research Council.

Because there can be sex differences in the response to TBI (43, 44) and TBI in the young and in military personnel is more prevalent in males (45), the study was conducted in male pigs. A total of 48 pigs weighing 28 ± 2 kg and approximately 3 months of age were used in the overall study. The experimental protocols for the TBI + HS experiment and the TBI alone experiment are delineated in Figure 1. The pigs were sedated with intramuscular injection of Telazol (50 mg/ml tiletamine and 50 mg/ml zolazepam, 4.4 mg/kg each component), ketamine 2.2 mg/kg and xylazine 2.2 mg/kg. Isoflurane (4% in 30% O₂) was administered via face mask to produce an anesthetic depth for oral intubation of the trachea. After a surgical plane of anesthesia was achieved, as assessed by the lack of limb withdrawal to hoof pinching and by looseness of muscle tone in the jaw, anesthesia was maintained with 2% isoflurane in approximately 30% O₂ with mechanical ventilation of the lungs. The antibiotic Baytril 10 mg/kg (100 mg/ml) was injected intramuscularly. Surgery was conducted using aseptic techniques. Through a 5-cm neck incision, an external jugular vein was isolated by blunt dissection. The vein was ligated and a catheter was advanced toward the heart and secured with another ligature. For arterial catheterization, we chose the axillary artery because occlusion of the carotid artery could limit cerebral blood flow after TBI and catheterization of the femoral artery can limit use of the hindlimb. An incision was made in the axilla, and the axillary artery was isolated, ligated, and cannulated with a flexible polyvinyl catheter that minimized kinking. The arterial and venous catheters were tunneled subcutaneously to the back of the neck, where they exited through a small incision. Pigs were able to bear weight on the forelimb and ambulate on the day after surgery.

Pigs were injected with fentanyl 20 µg/kg bolus and followed by continuous infusion of 20 µg/kg/h through the jugular vein catheter while maintained on isoflurane anesthesia. They were placed in a stereotactic head holder that was placed on the base of the impactor frame to prevent movement of the head during impact. The CCI model was similar to that described by others (38). A 20 mm diameter craniotomy was made over the right frontal cortex centered 1 cm anterior to the bregma and 1 cm lateral to the midline. The dura remained intact. Once the craniotomy was complete, the isoflurane concentration was decreased to 1%. At 5 min before the TBI, the isoflurane was stopped while fentanyl continued to be infused. TBI was produced with a CCI device (Custom Design and Fabrication, Inc., Sandston, VA, USA) that used a linear electric motor accelerator to control an impactor with a 15 mm diameter and a rounded tip. We used a 10 cm inward displacement of the dura at a velocity of 5 m/s for a duration of 50 ms. After CCI, gel foam was inserted into the craniotomy and sealed with tissue glue. The scalp incision was closed, and lidocaine hydrochloride cream (2.0%) was topically applied to all surgical incisions. The pigs remained sedated by infusing 0.1 mg/kg midazolam while continuing the infusion of 20 µg/kg/h fentanyl. For postoperative analgesia, they received a subcutaneous injection of sustained-release buprenorphine (0.12 mg/kg). They typically became responsive and were extubated at 3–5 h after TBI and were placed in a recovery cage for additional physiology monitoring for the first 4 h after TBI. When they were able to stand, they were returned to their pen in the central housing facility and continued to be observed for full mobility, which usually occurred within 6 h after TBI. Pigs were observed at least twice per day over the 4-day recovery period and were assessed for appetite, urination, defecation, mobility and level of awareness.

In the first experiment with TBI + HS, hemorrhage was produced by withdrawing 30 ml/kg of blood starting 5 min after TBI. Because very low mean arterial pressure (MAP) will decrease coronary blood flow and cardiac contractility, hemorrhage sometimes needs to be briefly interrupted to prevent loss of cardiac function (42). In our protocol, blood withdrawal was temporarily stopped when MAP fell to 35–40 mmHg and resumed when MAP began to recover above 40 mmHg. When the targeted amount of blood had been removed, blood withdrawal ceased and MAP was allowed to spontaneously increase above 40 mmHg. Thus, we used a variable rate of blood withdrawal to achieve a fixed volume of hemorrhage while attempting to avoid MAP < 35 mmHg and minimize the risk of cardiogenic shock.

In the first experiment, pigs were randomized to groups undergoing TBI followed by 2 h of hemorrhagic hypotension and then fluid resuscitation with 60 ml/kg lactated Ringer's (LR) solution (n = 7), 10 ml/kg PNPH (n = 8), or 20 ml/kg PNPH (n = 8). In the second experiment, pigs undergoing TBI without hemorrhagic hypotension were randomized to fluid treatment with 10 ml/kg LR solution (n = 6) or 10 ml/kg PNPH (n = 9) starting 2 h after TBI. A sham-operated group underwent catheterization and craniotomy, but without CCI or hemorrhage (n = 5). Because the craniotomy itself can potentially cause inflammation in the meninges and in cortex, brains from a naïve group without surgery were also collected (n = 5).

Fluid treatment commenced 2 h after TBI in those with and without induced hemorrhagic hypotension. Because PEGylation conveys hyper-oncotic properties to the molecule, smaller infusion volumes of PNPH are needed to restore MAP than LR. In addition, rapid fluid resuscitation can cause a rapid increase in preload and afterload, which can lead to pulmonary edema when cardiac contractility is already compromised by low perfusion pressure. Because cardiac output can be reduced and cerebral edema worsened with rapid LR infusion, others recommend using stepwise increases in infusion rates (39). Therefore, we performed fluid resuscitation with stepwise increasing infusion rates. In the group resuscitated with LR after hemorrhage, infusion started at a rate of 10 ml/min for 5 min (50 ml), then 20 ml/min for 22.5 min (additional 450 ml), and lastly at 60 ml/min until 60 ml/kg was infused (twice the hemorrhage volume). In the group without hemorrhage and treated with LR, infusion was stopped when a total of 10 ml/kg was infused. For the PNPH infusion groups, resuscitation commenced with a priming dose of LR infused at a rate of 10 ml/min for 5 min followed by PNPH infusion at a rate of 2.5 ml/min for 10 min and then at 5 ml/min until the targeted dose of 10 or 20 ml/kg was achieved. In the pigs that underwent hemorrhage, an additional LR infusion was performed at the end of the infusion period until the total of PNPH + LR equaled 30 ml/kg (the hemorrhage volume). If mean arterial pressure dropped below 40 mmHg during HS or early resuscitation, epinephrine was infused to maintain coronary perfusion pressure and avoid irreversible shock.

Throughout the first 4 h after TBI or sham surgery, arterial blood pressure was recorded and arterial blood samples (0.5 ml) were obtained hourly. The samples were analyzed for pH, blood gases, Hb concentration, metHb, COHb, and glucose with an ABL800Flex blood gas machine (Radiometer Inc., Brea, CA USA). In the group receiving PNPH, the remaining blood sample was centrifuged to obtain plasma, which was then analyzed for Hb concentration. Rectal temperature also was monitored.

On day 1, 2, 3 and 4 of recovery, a 1 ml blood sample was withdrawn from the arterial catheter for analysis of total blood Hb concentration, metHb, and COHb. The plasma Hb was also analyzed in the PNPH groups. If the arterial catheter was not patent, then samples were drawn from the jugular vein. Values from arterial and venous blood were combined for statistical analysis because the Hb concentration, metHb, and COHb from the 2 sites are expected to be similar. On day 4, piglets were deeply anesthetized and perfused transcardially with cold phosphate-buffered saline (PBS) followed by cold 4% paraformaldehyde. The thoracic aorta was clamped to direct fixative to the upper body. The quality of ongoing and final tissue fixation was judged by neck and facial muscle firmness. The heads were removed and placed in 4% paraformaldehyde to allow for in situ post-fixation of the brain within the skull overnight. Then the brains were carefully dissected and placed in a solution of 4% paraformaldehyde and 30% sucrose in PBS at 4°C. Using the optic chiasm and pons as landmarks, the brain was cut into 3 blocks. The anterior third block was cut into 50 µm sections with a sliding microtome and serially allocated into three 12-well plates and repeating the order 1–36.

Evaluation of neuropathology was performed with blinding to treatment group. Histologic sections from the anterior third of the brain, encompassing the frontal lobes, were used to estimate the ipsilateral and contralateral hemisphere volume of structurally normal appearing cerebral cortex and the volume of peri-lesion subcortical white matter. Every 12th serial coronal section was stained with 0.1% Luxol fast blue followed by 0.1% cresyl violet staining after dehydration in a series of increasing alcohol concentrations. Luxol fast blue staining was used to identify white matter. Cresyl violet was used to counterstain tissue sections for lesion identification and neocortex loss. Using Image J software, we traced the area of each hemisphere excluding the contusion injury (cavity plus pale cresyl violet stained tissue) and the area of white matter in each hemisphere. We calculated the volume by summing section area × 600 µm spacing between sections along the entire axial length of the anterior third of the cerebral hemisphere. The percentage of contusion injury volume was calculated 100 (left–right)/left hemisphere volume. The percentage of white matter volume loss was calculated as 100 (left–right)/left white matter volume. For this analysis, 15–20 brain sections were used to calculate the percentage of contusion injury volume and the percentage of white matter volume loss for each brain.

Immunostaining for amyloid precursor protein (APP) was used as a surrogate marker for impairment of axonal transport because it is known to accumulate when transport function in axons is blocked (46). Neuronal somatodendritic integrity was assessed by immunostaining for the cytoskeletal protein microtubule-associated protein-2 (MAP-2). Assessment of microglia morphology was performed with immunostaining for Iba-1. Immunohistochemistry was performed on 50 µm free-floating sections under moderate shaking. Before staining, the sections were incubated 30 min in 0.3% hydrogen peroxide to quench endogenous peroxidases. After three washing steps in 0.1 M phosphate buffer (pH 7.4), non-specific antibody binding sites were blocked with using 10% normal goat serum. Different free-floating sections were incubated overnight at 4 °C with anti-MAP-2 (1 : 500, monoclonal mouse-IgG; Millipore, MAB3418), anti-APP (1 : 500, monoclonal mouse-IgG; Millipore, MAB348), or anti-Iba-1 (1 : 500, polyclonal rabbit-IgG; Wako, 019-19741) in 5% normal goat serum. After several washes, sections were incubated for 2 h at room temperature with secondary antibodies. For APP and MAP-2 staining, biotinylated anti-mouse-IgG, 1 : 500, Vector, BA9200 was used, and the streptavidin/horseradish peroxidase detection was performed according to the manufacturer's recommendations. These sections were incubated with the substrate diaminobenzidine (DAB, D3939; Sigma-Aldrich Company, St. Louis, MO, United States) for 10 min at room temperature. For Iba-1 staining, Alexa Fluor 594 sary antibody (1:500, Invitrogen, #A-11037) was used. Immunofluorescent images were obtained with a Leica microscope (model DMi8 with THUNDER Imager 3D software).

Stereological measurements were made by using a Nikon Eclipse 90i microscope (Nikon, Tokyo, Japan) attached to a Qimage Retiga-2000R camera, which was connected to a workstation with Stereo Investigator software (Version 10; MicroBrightField, Williston, VT, USA). Every 12th section was examined, placing the analyzed sections 600 μm apart. Using a 2× objective, we traced the entire area of subcortical white matter for outlining the region of interest to be used in the stereological analysis. Then we counted the APP-positive particles under a 40× objective. We identified particles as being a brown dot that first came into focus within the optical dissector counting frame. The counting frame was 150 × 150 μm with a grid size of 3,000 × 3,000 μm. The dissector height was set at 10 μm with an upper and lower guard zone of 1 μm. This analysis used 11–14 coronal sections and 98–176 sampling fields to ensure that at least 100 particles were counted per brain (range 118–1086) to obtain the estimate of the total number of APP-positive particles within the entire subcortical white matter area.

Processes stained for MAP-2 had a more fragmented appearance after TBI, indicative of disruption of microtubules. As a measure of dendrite integrity, we counted the number of MAP-2–immunopositive processes with a length >50 µm within the plane of sectioning of the 50 µm-thick sections. This 50 µm threshold length was based on a preliminary survey of the images in the TBI brains where many of the fragmented dendrites were <50 µm in length. For this analysis, adjacent images of the entire cortical mantle of the gyrus lateral to the contused gyrus were obtained at 40× and the image tiles were stitched together to obtain a single composite image. For each brain, 10–12 sections spaced 600 µm apart were analyzed and the number of long processes per coronal section were averaged. Counting was performed on approximately 100 fields of view (0.25 mm² area each) per section spread over a 5 × 5 mm area of the cortical mantle. For each brain, approximately 1,000 fields were used to obtain the number of MAP-2 immunostained processes >50 µm per section.

Iba-1 immunofluorescent images were obtained on 3 coronal sections for each brain. On each section, 3 fields of view at 40× power were analyzed in the peri-lesion area. For each brain, an average value was obtained from the 9 fields of view. The total number of Iba-1–positive cells were counted to quantify the number of microglia/macrophages in the peri-lesion area. In addition, we quantified three different kinds of morphologies of these cells. Those with a small cell body area and long processes were regarded as ramified microglia. Those with a larger cell body area and shorter processes were regarded as hypertrophic microglia. Those which almost had no long, thin processes were regarded as fully activated bushy microglia. Because the hypertrophied and bushy morphologies were not always distinguishable, these two morphologies were combined for statistical purposes.

All data were analyzed by GraphPad Prism version 6 statistical software. Measurement data were expressed by mean ± standard deviation. Statistical analysis included all pigs that completed the protocol and survived for brain perfusion and fixation: The number of pigs in each group was: 5 naïve, 5 sham surgery, 5 TBI + HS + 60 ml/kg LR, 4 TBI + HS + 10 ml/kg, PNPH, 5 TBI + HS + 20 ml/kg PNPH, 6 TBI alone + 10 ml/kg LR, and 6 TBI alone + 10 ml/kg PNPH. For histologic analysis, the naïve and sham groups were combined into a single control group of 10 pigs since all of the histologic values were similar in each group. The TBI + HS experiment and the TBI alone experiment were analyzed with separate one-way analysis of variance (ANOVA) for each experiment. If the F value was significant, then we used the Holm-Sidak procedure to keep the family-wise error at <0.05 for multiple comparisons. In the TBI + HS experiment, there were 5 comparisons of interest. We compared the combined Naïve + Sham group with (1) TBI + HS + LR, (2) TBI + HS + 10 ml/kg PNPH and (3) TBI + HS + 20 ml/kg PNPH; and we compared TBI + HS + LR with (4) TBI + HS + 10 ml/kg PNPH and (5) TBI + HS + 20 ml/kg PNPH. In the TBI alone experiment, there were 3 comparisons of interest. We compared the combined naïve + Sham group with (1) the TBI + LR group and (2) the TBI + 10 ml/kg PNPH group; and we compared the TBI + LR group with (3) the TBI + 10 ml/kg PNPH group. For physiologic measurements, some missing values occurred because of difficulty in drawing arterial blood samples or because of technical problems with the blood gas analyzer. In addition, some sham-operated pigs regained consciousness more quickly than the TBI pigs and were too active to obtain MAP and blood sample measurements at the 120–240 min time points, resulting in n = 2–3 at these times. Because of the small sample size, the sham group was not included in the statistical comparisons for MAP and blood sample analysis.

In the TBI + HS experiment, 33 pigs were enrolled in the TBI + HS study and assigned to an intention-to-treat group prior to the day of surgery. Of these, 24 survived and were used for neuropathologic analysis. These included: (1) 5 naïve pigs perfused with fixative to provide normal brain histology, (2) 5 survivors from sham surgery, (3) 5 survivors from TBI + HS + LR resuscitation (mortality = 2 of 7 pigs: one pig died during surgery from presumed arrhythmia, 1 pig died 4 h after TBI), (4) 4 survivors from 10 ml/kg PNPH resuscitation (mortality = 4 of 8 pigs: one pig died during hemorrhage before the start of resuscitation; two pigs were hypotensive during resuscitation and died at 2.5–4 h after TBI; one pig completed the resuscitation protocol, but its arterial catheter accidently got pulled out when the pig was waking up from anesthesia and moving its legs, which resulted in a second episode of hypotension; this pig was humanely euthanized), (5) 5 survivors from 20 ml/kg PNPH resuscitation (mortality = 3 of 8 pigs: one pig died during hemorrhage before the start of resuscitation; two pigs were hypotensive during resuscitation and died at 2.5–3 h after TBI). Thus, among the groups destined to undergo TBI + HS, 14 of 23 pigs completed the protocol.

In the TBI alone experiment, 15 pigs were enrolled and 12 survived and were used for neuropathologic analysis. These included: (1) 6 survivors from TBI + LR infusion (no mortality) and (2) 6 survivors from TBI + 10 ml/kg PNPH infusion (mortality = 3 of 9: two pigs had severe hypotension leading to cardiac arrest during infusion; one pig failed to fully regain consciousness).

The time course of physiologic data is presented for the survivors from each group in Figure 2 and the individual data are shown at each time point in Supplementary Figures S1–S5. All TBI + HS groups subjected to HS had a similar level of MAP at the time of TBI and had a similar level of hypotension during the 2 h of HS (Figure 2A). In all TBI + HS groups, MAP recovered to their pre-injury levels within 30–60 min of the start of fluid resuscitation. All groups had further increases in MAP to approximately 100 mmHg as pigs awakened at variable times. Thus, the different resuscitation fluid composition and volumes were well matched to yield a similar recovery of MAP. This allowed comparisons of potential neuroprotective effects to be independent of the recovery of MAP.

The concentration of hemoglobin (Hb) decreased during the fluid resuscitation with LR at 150 and 180 min after TBI (30 and 60 min of fluid infusion) compared to the PNPH groups, which were infused at a slower rate (Figure 2B). PNPH is nearly saturated with carbon monoxide (CO) to prevent metHb formation during storage. During the infusion of PNPH, CO is released from PNPH and equilibrates with native Hb. The total blood COHb (plasma plus red blood cell Hb) increased progressively during PNPH infusion but remained below 4% (Figure 2C). COHb recovered to normal levels by Day 1. PNPH infusion also resulted in a significant increase in metHb by 240 min after TBI compared to the LR group (Figure 2D), but the elevation was only about 1% metHb and resolved by 1–2 days. Together, the increase in COHb and metHb will act to decrease the fraction of total Hb that was carrying oxygen (FHbO2). Although there was a trend for FHbO2 to decrease at 210 and 240 min when COHb and metHb were at their peak levels, the decrease in FHbO2 was variable and did not significantly differ from values in the group resuscitated with LR at these time points (Figure 2E). The concentration of Hb in the plasma increased to 0.71 ± 0.16 g/dl after infusion of 10 ml/kg PNPH and then declined to 0.21 ± 0.03 g/dl on Day 1 and to 0.03 ± 0.02 g/dl on Day 2 (Figure 2F). After 20 ml/kg PNPH infusion, plasma Hb concentration increased to 0.84 ± 0.40 g/dl and declined to 0.52 ± 0.07 g/dl on Day 1 and to 0.13 ± 0.04 g/dl on Day 2 as PNPH was slowly cleared from the circulation. Thus, a significant amount of PNPH was retained in the circulation over the first two days of recovery.

Representative coronal brain sections stained with Luxol fast blue followed by cresyl violet are shown from a naïve pig and TBI + HS pigs resuscitated with LR and PNPH (Figure 3A). At 4 days of recovery, cavitation with some remnants of hemorrhage were evident in the cerebral cortex and underlying white matter. The areas used for measuring intact hemisphere volume and white matter volume are illustrated by the red and yellow traces, respectively, in Figure 3B. Contusion injury area measurements included the cavity and tissue with atypical sparse cresyl violet staining. The remaining area of histologically normal tissue in the injured hemisphere was subtracted from the area of the contralateral hemisphere and summed across the frontal lobe to estimate the contusion lesion volume, which was normalized by the contralateral hemisphere volume. Because the Sham group had minor differences between the volume of each hemisphere volume, pigs in this group were combined with the Naïve group as a single control group for statistical analysis. One-way ANOVA indicated a significant overall effect of treatment among the four groups [F (3,20) = 10.463; P < 0.001]. Contusion injury volume averaged 13%, 9%, and 8% of the homotypic contralateral hemisphere volume in the TBI + HS groups resuscitated with 60 ml/kg LR, 10 ml/kg PNPH, and 20 ml/kg PNPH, respectively (Figure 3C). Contusion injury volume exhibited considerable variability and no significant differences were noted among the 3 TBI + HS groups with TBI + HS with the Holm-Sidak procedure.

White matter loss, relative to the homotypic contralateral white matter volume, averaged 26% in the frontal lobe in the TBI group resuscitated with LR, 16% in the group resuscitated with 10 ml/kg PNPH, and 9% in the group resuscitated with 20 ml/kg PNPH (Figure 3D). The overall effect of group intervention was significant by one-way ANOVA [F (3,20) = 13.233; P < 0.001]. Multiple comparisons with the Holm-Sidak procedure indicated that the level in the 20 ml/kg PNPH group was significantly less than the level in the LR group (P = 0.006) but was not significantly different than the measurements in the Naïve + Sham group (P = 0.18). This result suggests that PNPH protects white matter in a gyrencephalic brain where there is an abundance of subcortical white matter.

Axonopathy was assessed by APP staining as interpreted by accumulation of punctate immunoreactivity in the white matter seen as brown particulate inclusions on DAB immunohistochemistry (Figure 4A). Using 11–14 equally spaced coronal sections per brain, unbiased stereological analysis was performed over the entire area of subcortical white matter. An example of the area included in the analysis is shown by the outlined red trace in Figure 4B. One-way ANOVA indicated a significant overall effect of group intervention [F (3,20) = 7.898; P < 0.001]. The Holm-Sidak procedure for multiple comparisons indicated a consistent increase in the estimated APP particles, from 1,749,836 ± 908,968 in the combined group Sham + Naïve group to 4,066,307 ± 1,244,399 in the TBI + HS group resuscitated with LR (P < 0.001, Figure 4C). In contrast, the estimated number of APP particles in groups resuscitated with 10 ml/kg PNPH (2,385,524 ± 715,346) and 20 ml/kg PNPH (2,201,756 ± 257,790) were not significantly different from the Sham + Naïve group (P > 0.5). Moreover, APP particle accumulation in the 10 ml/kg and 20 ml/kg PNPH groups were significantly less than that in the LR group (P < 0.05). These results indicate a significant rescue of neuronal axonal transport with PNPH resuscitation from TBI + HS.

To assess neuronal dendrite integrity in neocortical gray matter of frontal lobe, we examined the number of long MAP-2–positive processes in the plane of the section. In Figure 5A, arrows point to the MAP-2–positive dendritic segments with >50 µm length. Counting was performed on 10–12 equally spaced coronal sections throughout the neocortex of the gyrus adjacent to the contusion as outlined in Figure 5B. One-way ANOVA indicated a significant overall effect of group intervention [F (3,20) = 4.615; P = 0.013]. The number of long dendritic processes per coronal section was reduced from 403 ± 75 in the Naïve + Sham group to 237 ± 95 in the TBI + HS + LR group (P < 0.05), whereas the number of long dendrites in the groups resuscitated with 10 ml/kg PNPH (406 ± 121) and 20 ml/kg PNPH (368 ± 139) were not significantly different from the Sham + Naïve group (Figure 5C). Furthermore, the value in the 10 ml/kg PNPH group was significantly greater than that seen with LR resuscitation (P < 0.05).

We next performed assessment of Iba1-positive cells to quantify microglial cell number and morphologic appearance in the cerebral cortex adjacent to the contusion. In each of 3 sections, we counted three fields of view as shown as Figure 6A, and obtained an average value from the total of 9 fields for each brain. We quantified the number of Iba-1–immunopositive cells, the number of immunopositive cells with a small soma and ramified process morphology, and the number of immunopositive cells with a large soma and shorter processes (hypertrophic microglia) or essentially no processes (bushy microglia). Examples are shown in Figure 6B. One-way ANOVA indicated a significant overall effect of group intervention for the total Iba-1–immunopositive cell number [F (3,20) = 5.187; P = 0.008], for the number of ramified microglia [F (3,20) = 7.360; P = 0.002], and for the number of hypertrophic and bushy microglia [F (3,20) = 35.002; P < 0.001]. For TBI + HS with LR resuscitation, the total Iba-1 immunopositive–cell number was significantly increased compared with Naïve + Sham group (P < 0.01). This increase was modestly but significantly reduced with 20 ml/kg PNPH resuscitation compared to LR resuscitation (P = 0.04) and not significantly different from the level in the Naïve + Sham group (Figure 6C). TBI + HS produced an overall decrease in the number of microglia with ramified processes without significant differences between the LR group and the two PNPH groups (Figure 6D). The decrease in ramified microglia was associated with an increase in the number of microglia with hypertrophic and bushy appearance in all three TBI + HS groups compared to the Naïve + Sham group. Treatment with 20 ml/kg PNPH treatment modestly reduced the number of hypertrophic and bushy Iba1-positive microglia when compared with LR-treated TBI piglets (P = 0.04; Figure 6E).

The time course of physiologic data is presented for the survivors from each group in the TBI alone experiment in Figure 7, and the individual data are shown at each time point in Supplementary Figures S6–S10. Before TBI, MAP was approximately 80 mmHg and remained stable during the first 2 h after TBI in the absence of hemorrhagic shock in the groups destined to receive 10 ml/kg LR or 10 ml/kg PNPH at 2 h (Figure 7A). Afterwards, MAP gradually increased in both TBI groups as the pigs regained consciousness. Some pigs receiving PNPH had brief episodes of arterial hypotension that were restored by infusion of epinephrine. However, infusion of epinephrine sometimes resulted in an overshoot in MAP, which may have accounted for the elevated MAP seen at 150 min and 180 min time points after TBI. The oncotic effect of PNPH might also be a factor in this early increase in MAP. At the remaining time points, MAP in both TBI groups was similar to that in the sham group.

During the first 4 h after TBI, both groups had similar levels of whole blood Hb (Figure 7B), FHbO2 (Figure 7C), and metHb (Figure 7D). PNPH infusion resulted in the expected increase in whole blood COHb, but the peak level was only 2% COHb and was eliminated by day 1 (Figure 7E). Plasma Hb concentration increased to 0.51 ± 0.06 g/dl after 10 ml/kg PNPH infusion and declined to 0.11 ± 0.06 and 0.02 ± 0.01 g/dl on days 1 and 2, respectively (Figure 7F).

Representative cresyl violet/Luxol fast blue-stained images of coronal brain sections are shown from a naïve pig and pigs with 10 ml/kg LR and 10 ml/kg PNPH infusion started 2 h after TBI (Figure 8A). Contusion lesion volume displayed an overall group effect [F (2,19) = 7.513; P = 0.004], but there was no significant difference (P = 0.17) between the TBI groups receiving LR and PNPH (Figure 8B). The volume of the subcortical white matter on the impacted side relative to the contralateral side displayed an overall group effect [F (2,19) = 6.826; P = 0.006]. Compared to the Naïve + Sham group, the TBI group receiving LR exhibited a significant loss of white matter volume ipsilateral to the injury (P = 0.022) (Figure 8C). In the TBI group receiving PNPH, the loss of white matter volume was significantly attenuated compared to the group receiving LR (P = 0.027). The value in the PNPH group did not significantly differ from the value in the Naïve + Sham group (P = 0.99).

Examples of APP immunoreactive particles are shown in Figure 9A for sections from naïve, TBI + LR, and TBI + PNPH pigs, and the area of subcortical white matter used in the stereological analysis is shown in Figure 9B. The quantification of APP particles in subcortical white matter in the TBI alone experiment (Figure 9C) showed a significant overall effect of group intervention [F (2,19) = 7.513; P = 0.004]. The Holm-Sidak procedure for multiple comparisons indicated that the estimate of the number of APP particles in the TBI + LR group (3,826,066 ± 1,977,136) was significantly greater than the number in the Naïve + Sham group (1,749,836 ± 908,968, P = 0.007). The number in the TBI + LR group was also significantly greater than that in the TBI + PNPH group (1,365,371 ± 429,906, P = 0.007). Interestingly, the value in the TBI + PNPH group was not significantly different from that in the Naïve + Sham group. Thus, the benefit of PNPH treatment on both white matter volume and APP accumulation seen in the TBI + HS experiment persisted in the absence of hemorrhagic shock.

Images of MAP-2–positive dendritic segments with >50 µm length in the gyrus adjacent to the contusion are shown in Figure 10A for brains from the TBI alone experiment, and an example of the area of the gyrus used in the analysis is shown in Figure 10B. As occurred with TBI + HS, one-way ANOVA indicated a significant overall effect of group intervention on dendritic integrity [F (2,19) = 40.465; P < 0.001]. The Holm-Sidak procedure for multiple comparisons revealed a marked decrease in the number of intact dendritic processes with a length >50 µm in the TBI group treated with LR (139 ± 35 per coronal section; P < 0.001) compared to the Naive + Sham group (403 ± 75 per coronal section) (Figure 10C). Treatment with PNPH greatly attenuated this decrease in long dendrites (349 ± 36 per coronal section; P < 0.001 from LR group), and the values were only marginally different from the Naïve + Sham group (P = 0.09). Again, the benefit of PNPH treatment after TBI did not require the presence of HS.

With TBI alone, the density of Iba-1–positive cell counts showed no overall effect among the three groups [F (2,19) = 1.620; P = 0.22] (Figure 11A). Analysis of the number of Iba-1–positive cells classified as having ramified processes yielded an overall group effect [F (2,19) = 27.810; P < 0.001] accompanied by significant decreases in the LR and PNPH groups compared to the Naïve + Sham group (Figure 11B). The difference between the LR and PNPH groups was not significant. On the other hand, the number of hypertrophic/bushy Iba-1–positive cells at 4 days after TBI alone displayed an overall group effect [F (2,19) = 42.107; P < 0.001] in which the significant increase seen with LR infusion (P < 0.001) compared to the Naive + Sham group was significantly attenuated with PNPH treatment (P = 0.011 vs. LR group).