Main

In animal models of cerebral ischemia, neutrophils accumulate within cerebral blood vessels and then extravasate into the brain parenchyma(1–4). There is evidence that they contribute to ischemia/reperfusion injury in adult animals as neutrophil depletion is markedly protective(2, 3, 5). Even more exciting from a clinical perspective are the reports showing that ischemic brain injury can be reduced by prevention of neutrophil-endothelial adhesion with strategies initiated after the primary HI insult(6–10). To date all these observations have been determined in only adult stroke models.

Compared with adults, neonates have a diminished ability to mount a neutrophil response, and neonatal neutrophils have a diminished tendency to extravasate blood vessels(11, 12). Accordingly, it is important to determine the role of neutrophils in perinatal ischemic brain injury. The timing and anatomic distribution of neutrophil accumulation in the immature brain is unknown. From the standpoint of designing antineutrophil strategies for clinical investigation, it is also important to elucidate any differences in the manner neutrophils contribute to perinatal stroke.

The scarce reports that refer to the presence of neutrophils in post-ischemic neonatal brain suggest that neutrophil extravasation into the parenchyma is rare(13, 14). However, this does not exempt neutrophils from a potentially pathogenic role, as neutrophils can release inflammatory mediators from within blood vessels(15), or can impair flow when blood pressure is reduced(16–20). There have been no animal studies to determine whether neutrophils contribute to brain injury in the newborn brain apart from a single observation in newborn piglets that confirmed accumulation of neutrophils in the lumen of cerebral blood vessels 4 h after cardiac arrest(21).

Blood vessels are particularly susceptible to reperfusion injury when the return of oxygenated blood generates free radicals. Free radicals, produced in part by endothelial cell derived xanthine oxidase, activate neutrophil adhesion and lead to vascular injury(15, 22, 23). As an example, after intestinal ischemia, neutrophilic infiltration and microvascular injury can be reduced by inhibition of xanthine oxidase with allopurinol(22, 24). High dose allopurinol protects the brain of the newborn rat from HI injury, even when allopurinol is administered during early reperfusion(25, 26). It is possible that allopurinol exerts its neuroprotective effect via neutrophils; alternatively, allopurinol's mechanism of action may be independent of neutrophils.

This study was designed to investigate whether neutrophils contribute to HI brain injury in the neonatal rat. The first objective was to determine the anatomic distribution and time sequence of neutrophil accumulation in the immature rat. The second objective was to determine whether neutropenia could reduce post HI brain swelling. In addition, we wished to compare the neuroprotective effect of neutropenia with that produced by allopurinol in this model and determine whether allopurinol could provide additional protection to the neutropenic rats.

METHODS

The HI immature rat model. Dated, pregnant, verified antigen-free Wistar rats were purchased from Charles River Inc., Wilmington, MA. HI injury to the right cerebral hemisphere of 7-d-old rats of either sex was produced by right common carotid artery ligation and exposure to hypoxia in 8% oxygen as described previously(26, 27). Rats were anesthetized with 3% halothane for induction and 1% for maintenance. The right common carotid artery was permanently ligated with 4-0 surgical silk through a midline neck incision. The wound was sutured, and the animals were allowed to recover with their dams in room air for 6 h. The duration of anesthesia was generally less than 5 min. At the end of the recovery from surgery, rats were placed in airtight jars, four rats per jar. The jars were partially submerged in a circulating water bath maintained at 36.8 °C to provide a stable thermal environment, and exposed to a continuous flow of 8% oxygen-92% nitrogen gas mixture for 2.25 h. After hypoxia the rats were returned to their dams. This insult produces swelling of the cerebral hemisphere ipsilateral to the ligated carotid in more than 90% of rats and cavitary infarction in about 50%(26). The contralateral hemisphere is not injured, except in severely damaged brains where a small strip of cortex (±2 mm wide) adjacent to the interhemispheric fissure will show some damage(28). This area was excluded from the determination of brain swelling and neutrophil accumulation described below. All animal procedures were performed with the approval of our institutional animal care committee.

Neutrophil depletion. To confirm our ability to deplete neutrophils without adversely affecting other hematologic parameters, 10 rat pups were injected intraperioneally with 50 μL of a rabbit polyclonal anti-rat neutrophil serum (ANS) prediluted 1:1 with PBS, pH 7.4 (donated by Neil D Granger, Ph.D.). Six hours after intraperitoneal injection of the antiserum, blood was collected into heparinized tubes via cardiac puncture under halothane anesthesia. An automated total nucleated cell count and platelet counts were performed (Celldyne 400, Sequoia Turner Co., Mountain View, CA). A differential count was performed on a blood smear, and the total white blood cell count was determined by correcting the automated count for NRBCs. The absolute neutrophil count per mm3 was calculated, as was the number of NRBCs per 100 white cells. The results of the 10 ANS-treated pups were compared with 21 normal littermates who did not receive ANS.

Neuroprotection experiments. Seven-day-old rats (n = 59) were weighed, then the right common carotid artery was ligated. The pups were randomized into four treatment groups (the Allopurinol, Neutropenic, Both, and None groups) so that rats allocated to the Neutropenic and Both groups were rendered neutropenic by injection of ANS at the time of surgery, whereas the Allopurinol and None groups received normal rabbit serum (Sigma Chemical Co., St. Louis, MO). Allopurinol was administered at 15 min of recovery from HI to the Allopurinol and Both groups (see below).

Preliminary studies showed that, by 6 h, ANS had depleted circulating neutrophils by about 95% (Table 1), so 6 h after receiving ANS, the pups were placed into glass jars and exposed to hypoxia. Each jar contained four rats, consisting of one from each assigned group. At 15 min of recovery from hypoxia, half the neutropenic (Both group) and half the nonneutropenic rats (Allopurinol group) were injected with allopurinol 135 mg/kg, s.c. (Sigma Chemical Co., adjusted to pH 11.2 with NaOH). The other groups received an equal volume of saline, pH 11.2. Thus, the Allopurinol group was treated with allopurinol and normal rabbit serum; the Neutropenic group was treated with ANS and saline; the None group received saline and normal rabbit serum; the Both group received ANS and allopurinol. After 42-h recovery from HI, the pups were reweighed, then decapitated, and right hemisphere swelling was determined as discussed below.

Table 1 Effect of ANS on circulating blood 6 h after administration
Full size table

Measurement of brain water content and brain swelling. Immediately after removal of the brain, the posterolateral half of each cerebral hemisphere was removed and weighed in a tared glass vial on a microanalytical balance. This region was chosen because it is frequently damaged in this model. After drying at 80 °C for 72 h, the vials were reweighed, and the percentage water content (%WC)(26) and percentage dry weight (%DW) of each sample was determined according to the formula (%DW = 100 -%WC). Swelling follows the influx of water into the right hemisphere. This is reflected in an increase in%WC and a decrease in%DW. Right hemisphere swelling was calculated from the relative change in%DW of the swollen right (R) compared with the normal left (L) hemisphere. Swelling was expressed as a percentage by the formula (%DWL -%DWR/%DWL) × 100.

Growth was calculated from the percentage change in weight from time of carotid ligation until sacrifice (approximately a 48-h interval). At decapitation, blood was collected from the severed neck vessels of all rats, and blood smears were made for differential white blood cell counts.

The effect of HI on circulating blood elements. To determine the effect of HI on the circulating blood components, blood was sampled from the neck of 10 normal rat pups and littermates after 0, 2, 4, and 18 h of recovery from HI (n = 8-10/group). Automated cell counts and differential counts were performed as described above.

Hemispheric MPO activity. MPO activity was measured as a biochemical estimate of neutrophil accumulation. Groups of untreated 7-d-old rats (n = 5-14/group) were subjected to the HI insult, and then killed at time points 0, 5, and 30 min, and 2, 4, 8, 18, and 42 h of recovery. We also included a group of six normal rat pups as controls. Brain samples from each hemisphere were taken from the same location as the samples for water content analysis. They were quick frozen in isopentane -70 °C, and then stored in Eppendorf tubes at -80 °C until analysis. Weighed tissue samples were homogenized on ice in 10 mM KH2PO4, pH 7.4 (100 mg to 1 mL) with an Omni 1000 homogenizer. The homogenate was centrifuged at 10 000 × g at 4 °C for 20 min, and the supernatant was discarded. The pellet was resuspended in 10 mM KH2PO4 (with 0.5% hexadecyltrimethylammonium bromide, pH 6.0) and frozen at -20 °C for up to 24 h. Tubes were thawed on ice and rehomogenized, sonicated for 5 min, and recentrifuged. MPO activity was measured in the supernatant according to the method described by Simpson et al.(29).

Briefly, 100 μL of supernatant were added to a reagent mixture containing 200 μL of 400 mM KH2PO4, 100 μL of 16 mM 3,3,5,5-tetramethylbenzidine in N,N-dimethylformamide, 500 μL of 15 mM KH2PO4, and 500 μL of normal saline. The mixture was equilibrated at 37 °C for 5 min. The reaction was started with 100 μL of 3 mM hydrogen peroxide and allowed to incubate for 3 min. The reaction was stopped with 3 mL of 0.2 M sodium acetate (pH 3.0). Absorbance was measured at 655 nm. The units of activity were expressed per g of brain wet weight. A unit represented the amount of activity that produced an absorbance change of 1 at 655 nm at pH 7.4 and 37 °C/min. When MPO standard (Sigma Chemical Co.) was added to brain homogenate, we were able to recover 88.3 ± 3.4% activity.

MPO-IR neutrophil counts. Other groups of 7-d-old rats(n = 5-14 rats/group) were subjected to the HI insult, and then killed at time points 0, 5, and 30 min, and 2, 4, 8, 18, and 42 h of recovery. A group of six normal rat pups were included as non-HI controls. After decapitation, the brains were removed intact from the skull and immersion fixed in FAM (1:1:8; 37% formaldehyde:1% acetic acid:100% methanol) for 1 wk. Paraffin-embedded 6-μm coronal sections were cut at the level of the anterior commissure (anterior level) and at the level of the infundibulum(posterior level).

The deparaffinized sections were rehydrated in PBS, exposed to 3% H2O2 for 2 min, then blocked in normal sheep serum for 20 min. The sections were exposed to 1:500 dilution of anti-human MPO rabbit antibody(Dako, Carpentiera, CA) for 1 h, then sheep anti-rabbit IgG-linking antibody for 30 min, and finally, the labeling antibody, rabbit peroxidase-labeled antiperoxidase, for 30 min. We used 3-amino-9-ethylcarabozole (Biomeda Corp., Foster City, CA) as a red chromagen and hematoxylin as counterstain. In addition, brain sections from each animal were stained with hematoxylin and eosin to permit assessment of ischemic injury. We did not perform neutrophil counts on the hematoxylin and eosin-stained sections because in the neonatal rats there are many other nucleated red and white cells within the vasculature that could be confused with neutrophils, and in the parenchyma neuronal nuclear changes and fragmentation (pyknosis and karryohexis) make neutrophil identification difficult without specific immunostains.

The neutrophil count was performed at 20× magnification, and the number and distribution per hemisphere in the 6-μm sections were recorded. The neutrophils present in the choroid plexus of the third and lateral ventricles were not included in the hemisphere count but were counted separately. We counted at least eight sections (four anterior and four posterior level) of the animals at each time interval.

Comparison of stains for neutrophil detection. Because the possibility existed that some neutrophils might not stain with the anti-MPO antibody as cells degranulated, we also stained adjacent sections of a few selected brains with a different anti-neutrophil marker. Accordingly, the six normal brains and brains from rats that had recovered 2, 4, and 8 h from HI(n = 2-3 brains per recovery interval) were selected for comparison of staining method. We included the 4- and 8-h time intervals, as they coincided with the peak in MPO activity. Sections of 6 μm (four to five sections per brain) were obtained from the same tissue blocks used for the MPO immunoperoxidase stain. These slides were stained by the same immunoperoxidase method for the MPO stain except that we used the polyclonal ANS as the primary antibody. The neutrophil count was obtained as described above, and the results of the two methods were compared.

ANS-IR neutrophil detection. In the above staining comparison we established that MPO staining failed to detect more than 90% of neutrophils beyond 2 h of recovery. So to obtain a true neutrophil count at the later recovery intervals, we selected all the brains from the 8-, 18-, and 42-h recovery (n = 7-8 brains per time point) and stained two anterior and two posterior sections for ANS immunoreactivity. All the brains showed histologic evidence of injury. The counts for the anterior and posterior sections were pooled and expressed as an average number of neutrophils per hemisphere.

Statistical methods. All data are presented as mean ± SD. For time course studies we controlled for litter and experimental effects by randomizing pups from each litter across time points and treatments. When necessary to meet distributional assumptions, significance tests were carried out after appropriate transformation. When several groups or treatments were compared by ANOVA, all post hoc comparisons were done using appropriate multiple comparison adjustments (i.e. Dunnett when comparing many treatments with a single control, Bonferroni when testing multiple linear contrasts). All post hoc p values reported in the text or figure legends have been adjusted for multiple comparisons.

RESULTS

Anti-neutrophil depletion. Administration of 50 μL of the diluted ANS reduced the absolute antineutrophil count by 95% in 6 h (see Table 1). The antiserum depleted all granulocytes, including the small number of monocytes. There was no significant effect on lymphocyte, platelet, or NRBC count. There were 52 ± 87 NRBCs per 100 white cells in the normal 7-d-old rats compared with 87 ± 88 in the neutropenic rats.

Brain swelling. Neutropenia and allopurinol reduced brain swelling by about 70%, p < 0.01, two-way ANOVA. There was no difference between the two in the amount of reduction. The combination of allopurinol and neutropenia (Both group) was not significantly different from the Allopurinol group or Neutropenic group alone (Fig. 1).

Figure 1
figure 1

The protective effect of neutropenia and allopurinol on post-HI brain swelling. Graph shows brain swelling: two-way ANOVA was used to assess the effects of neutropenia and allopurinol on brain swelling. We found that allopurinol and neutropenia both significantly reduce brain swelling and that there was no difference between the two in the amount of reduction. Further, the combination of neutropenia and allopurinol did not reduce brain swelling beyond that of neutropenia or allopurinol alone. *p< 0.01.

Full size image

The right hemisphere swelled 18.4 ± 2.49% in the None group(n = 15), and only 4.6 ± 0.9% in the Allopurinol group(n = 15), 5.4 ± 1.98% in the Neutropenic group (n = 15), and 4.5 ± 1.4% in the Both group (n = 14), p< 0.01 for each group compared with the None group. A single animal died during hypoxia in the untreated group. All other rats survived.

Growth and blood results. The blood obtained from the rats at sacrifice (Table 2) showed that those pups treated with ANS (Neutropenic group and Both group) were still significantly neutropenic at 42-h recovery; less than 1% of circulating white cells were neutrophils. Allopurinol did not produce neutrophil depletion.

Table 2 Effect of ANS on growth and blood cells at 42 h of recovery from HI
Full size table

The number of NRBCs per 100 white blood cells was also not different among the treatment groups. No correlation was found between the number of nucleated red cells in the peripheral blood at sacrifice and brain swelling in the 12 rats of the None group, r = -0.1, p = 0.75.

The neutropenic rats lost approximately 1% body weight, p < 0.0001, compared with the nonneutropenic. In contrast, of the nonneutropenic rats, the Allopurinol group gained 15.5 ± 5.7% weight, and the None group 18.5 ± 10.3%.

Hemispheric MPO activity. Despite injury being largely confined to the right hemisphere, there were no significant differences in MPO activity between the two hemispheres. A well defined increase in MPO activity, about 10 times more than baseline levels, was measured between 4 and 8 h of recovery in both hemispheres (Figs. 2 and 3).

Figure 2
figure 2

Brain MPO activity. Line graph shows MPO activity in the left (contralateral hemisphere, □―□) and right hemisphere(ipsilataral hemisphere, ♦―♦) for the first 4 h after HI. The normal range (mean ± 2 SD) of MPO activity is indicated by the open triangle. At 15 min of recovery the right hemisphere exhibited MPO activity that was significantly higher than normal. Both hemispheres had higher than normal levels at 1, 3, and 4 h of recovery. *p < 0.05.

Full size image
Figure 3
figure 3

Brain MPO activity. Line graph shows MPO activity in the left (contralateral, □―□) and right hemisphere (ipsilataral,♦―♦) for intervals up to 42 h in duration of recovery from HI. MPO activity was significantly higher than normal in both hemispheres between 1 and 8 h of recovery. After 8 h, only the left hemisphere had higher than normal activity. *p < 0.05 vs normal activity.

Full size image

From the onset of recovery from the HI insult, MPO levels trended higher than the baseline values of 144.8 ± 72.7 mU/g brain wet weight. They were significantly elevated by 15 min in the ipsilateral right hemisphere. Between 1 and 8 h of recovery MPO activity was significantly higher than normal in both hemispheres. At 1 h of recovery, levels were 758 ± 234 and 583 ± 234 mU/g in the right and left sides, respectively. This activity level persisted until 4 h, after which the levels more than doubled to peak 8 h into recovery at 1749 ± 555 on the right and 1512 ± 530 mU/g on the left side. At 18 h of recovery, the levels fell to 370± 97 on the right and 488 ± 100 on the left hemisphere. At 42 h of recovery, levels were 425 ± 198 on the right and 527 ± 227 mU/g on the left. MPO activity was significantly above normal at 18 and 42 h for the left side only.

Timing and distribution of MPO-IR neutrophils. We found significantly more MPO-IR neutrophils in the right hemisphere than the left across the full 42-h recovery period, p = 0.02, two-way ANOVA (Fig. 4). At zero recovery from HI, both hemispheres had four times more neutrophils than normal (normal = 1.78 ± 1.5 neutrophils per hemisphere). At 0-h recovery in the right hemisphere, there were 7.8 ± 7.0 cells, and at 5 min of recovery there were still significantly more neutrophils in the right hemisphere only, p = 0.02. Thereafter the counts remained within the normal range. Neutrophils were almost exclusively intravascular at all time periods examined (Fig. 5).

Figure 4
figure 4

MPO reactive neutrophil count. Chart shows number of neutrophils per hemisphere in a 6-μm section. Neutrophils were stained with the anti-MPO primary antibody during recovery from HI. Solid bars represent the right hemisphere (ipsilateral) and cross-hatched bars represent the left hemisphere (contralateral). In normal rat pups (n = 48 hemisphere sections) the number of neutrophils per hemisphere was 1.78 ± 1.5 (mean± SD) cells. At 0 h of recovery, there were significantly more neutrophils than is normal in both hemispheres. At 5 min of recovery, there were significantly more neutrophils in the right hemisphere only.*p < 0.05.

Full size image
Figure 5
figure 5

(A) Low power photomicrograph at 2 h of recovery from HI. MPO-IR neutrophils (arrows) in the vessels of the choroid plexus of the third ventricle (v). Injury in the ipsilateral medial habenular nucleus (*) is demonstrated by nuclear pyknosis and pallor of the cytoplasm. There are no neutrophils in the damaged parenchyma. (B) High power view of A shows the multilobed nucleus that is characteristic of the neutrophil. (C) MPO-IR neutrophil (arrow) at 18 h of recovery within a capillary. The damaged cortex shows extensive nuclear pyknosis. (D) ANS-IR neutrophils (arrows) at 18 h of recovery from HI in the leptomeningeal blood vessels over the cortex. Note the neutrophil within a penetrating vessel. (A-C) Immunoperoxidase with hematoxylin counterstain. (D) Immunoperoxidase with light hematoxylin counterstain. Calibration bar = 50 μm.

Full size image

Eighty-three percent of the neutrophils were located in the meninges, and only 17% were found in the brain parenchyma, mainly in vessels of the cortex. In addition, in normal rats (n = 10) there were 4.7 ± 12.3 neutrophils in the choroid plexus of the third and lateral ventricles. During recovery, the choroid plexus count followed a similar course to the hemisphere count. The highest choroid plexus count occurred at the onset of recovery, when 12.3 ± 13.2 cells were counted per hemisphere 6-μm section. Thereafter, the count fell in parallel with the hemisphere count, and no MPO-stained neutrophils could be detected in the choroid plexus at 4 and 8 h of recovery.

Very few neutrophils extravasated beyond the lumen (Fig. 5C) into the brain parenchyma, except at 42 h of recovery, when isolated neutrophils were seen within infarcted brain. Even in these instances, we could not be certain in some cases that neutrophils had indeed extravasated beyond the vessel, as vascular margins were no longer distinct. At all times neutrophils were more plentiful in the blood vessels of the leptomeninges and choroid plexus.

Comparison between MPO and ANS stains. In normal rats and at and 2 h of recovery, the number of cells stained with the two antibodies was not different. At 4 and 8 h of recovery, 10 times more neutrophils were detected with the ANS than were detected with the MPO stain, p ≤ 0.0001 (Fig. 6). For example, at 8 h of recovery, 6.1± 2.6 cells stained with the ANS in the right hemisphere compared with only 0.5 ± 0.76 with the anti-MPO stain, p ≤ 0.0001. The neutrophils were located in the same places as found with the MPO stain. Cells that stained with the ANS were also intravascular, with the majority being identified within the meningeal blood vessels (Fig. 5D). Microglial cells were not stained by either antibody.

Figure 6
figure 6

Comparison between anti-neutrophil stains (ANS vs MPO). Bar graph shows number of ANS-IR neutrophils(cross-hatched) and MPO-IR neutrophils (solid bar) at intervals of recovery from HI. Adjacent sections of six normal brains and two to three brains representing the 2-, 4-, and 8-h recovery interval were stained with both primary antibodies, and the counts were compared. The (mean ± SD) cells per hemisphere section is shown representing 48 normal hemispheres and 8-15 sections for the other post-HI recovery time intervals. Three-way ANOVA was used to assess differences between hemispheres across time and stain (MPO vs ANS). The right side stained more heavily than the left(p = 0.02). Both hemispheres stained more heavily with the ANS stain than the MPO stain at 4 and 8 h of recovery (*p ≤ 0.0001), but not differently at 2 h or in normal tissue.

Full size image

ANS reactive neutrophil count (Fig. 7). In the six normal 7-d-old rats there were 3.4 ± 3.0 ANS-IR neutrophils per 6-μm hemisphere section. In the same normal animals, the number of MPO-IR neutrophils was 1.78 ± 1.5 and was not significantly different from the number of ANS-IR cells counted (p = 0.38, paired t test). At 8 h of recovery ANS-IR neutrophils were still within the normal range in both hemispheres but at 18 h of recovery there were 11.0 ± 7.1 ANS-IR cells, which was significantly more than normal, in the right hemisphere only(p = 0.02). More neutrophils stained in the ipsilateral hemisphere than in the contralateral hemisphere. By 42 h of recovery neutrophil counts had returned to normal levels.

Figure 7
figure 7

ANS-IR neutrophil counts. Bar graph shows number of neutrophils stained with ANS per left (contralateral hemisphere, shaded bar) and right (ipsilateral hemisphere, open bar) at the 8-, 18-, and 42-h recovery interval. All brains counted were injured. The normal ANS-IR neutrophil count(solid horizontal line) and 2 SD above the mean is shown by the dotted line. Neutrophils were more plentiful in the right hemisphere than the left,p = 0.04 (two-way repeated measures of ANOVA). There was no difference between either hemisphere and normal tissues.

Full size image

DISCUSSION

This study shows that neutrophils do contribute to HI brain swelling in the 7-d-old rat. Neutrophil depletion before the HI insult reduced brain swelling at 42 h of recovery by approximately 70%, which was comparable to the protective effect of allopurinol. The protective effect with allopurinol on brain swelling supports our previous observation that “rescue” high dose allopurinol reduces early and late measures of HI brain injury. In the present study, we were unable to assess if the combination of allopurinol with neutropenia has additive effects as each treatment alone was so protective that there was very little margin for further reduction in brain swelling.

An unexplained, but interesting, finding in the neutropenic rats was the lack of growth. The lack of weight gain was probably not caused by dehydration, as water content assessed in the portion of the undamaged left hemisphere analyzed was normal. All the neutropenic animals were treated with ANS, and in subsequent studies with a different antineutrophil antibody, we found a similar reduction in growth (our unpublished findings).

To quantitate neutrophil accumulation we measured total hemispheric MPO activity as this is a standard indirect measure of neutrophil infiltration(1). Brain MPO activity was elevated in both hemispheres at the onset of recovery. The MPO levels in the right hemisphere were approximately four times normal by the third recovery hour, then doubled again at 4 h of recovery. This 10-fold increase in activity occurred well before brain edema is fully developed in the immature rat model used(30). The increase in MPO activity was earlier and relatively lower than found in adult ischemia models(3, 31). Matsuo et al.(3) showed that MPO activity and brain swelling followed a similar time course to the neutrophil count with little increased activity until 12 h of recovery. Then they found that MPO activity and brain swelling increased in parallel and peaked between 24 and 72 h of recovery.

We found no significant overall hemisphere difference for MPO activity. This was unexpected, as brain damage (swelling) is confined to the hemisphere ipsilateral to ligation only. We were careful to sample only those regions of the contralateral hemisphere that escape injury. However, as neutrophils were almost exclusively located within blood vessels, one must consider the likely possibility that the changes in MPO activity depict changes in MPO circulating in the blood vessels of the brain. This would be a function of the amount of blood in the hemisphere as well as the concentration of circulating MPO. Although we did not detect increased numbers of neutrophils in the peripheral blood at 4 h of recovery, it is conceivable that the systemic hypoxic insult could have increased MPO secretion from the total neutrophil population, including those adherent neutrophils that would have escaped measurement. Accordingly, the increased MPO activity in the contralateral (left) hemisphere at 18-42 h of recovery (Fig. 3) could be a reflection of diminished blood volume in the injured and infarcted ipsilateral hemisphere.

To count neutrophils within hemisphere sections, we initially performed MPO immunostaining. MPO is present in the neutrophil's azurophil granules, and other investigators have found good correlation between MPO activity and neutrophil accumulation with this immunostain(3, 32). We unexpectedly found that the sharp rise in MPO activity at 4-8 h of recovery (Fig. 3) was mirrored by a reduction in the MPO-IR cell count (Fig. 4).

To resolve this paradox we considered that neutrophils are capable of secreting MPO(33–35) and suspected that degranulating neutrophils might not be detected with the MPO immunostain. Also, peripheral neutrophil counts showed that there was not a reduction in circulating neutrophils. Therefore, we used the polyclonal ANS as an alternative stain for neutrophils and found that, at 0 and 2 h of recovery(before the rise in MPO activity), the two immunostains detected the same number of cells. However at 4-8 h of recovery, 10 times more neutrophils were detected with the ANS immunostain.

Having established that the MPO immunostain failed to detect over 90% of neutrophils after 2 h of recovery, we restained all the brains at 8, 18, and 42 h of recovery with the ANS immunostain. With these two staining methods we found that neutrophils were increased at the onset of recovery when neutrophil counts were about four times normal in both hemispheres. Counts were still increased in the ipsilateral hemisphere at 5 min or recovery. At later time points the number of neutrophils was not different from that of normal tissue. Significantly more neutrophils stained in the injured hemisphere compared with the noninjured hemisphere (p = 0.04; Fig. 7).

With regard to the distribution of neutrophils in the injured hemisphere, a striking observation was the lack of neutrophil extravasation during the evolution of the infarct. This is in sharp contrast to adult animal models of cerebral ischemia where neutrophils extravasate in large numbers(4, 10). Mastsuo et al.(3) found 40-80 infiltrating neutrophils per 5-μm hemisphere section at 12-72 h of recovery from transient middle cerebral artery occlusion in adult rats. In our study in the 7-d-old rat, the majority of neutrophils (83%) were located in the leptomeningeal blood vessels on the surface of the brain or within the choroid plexus. Neutrophils within the vessels of the leptomeninges are critically positioned to influence the integrity of penetrating blood vessels and red cell flow into the cortex below.

Although there were slightly more neutrophils in the ipsilateral hemisphere, the numbers were relatively similar in contrast to the amount of swelling which occurred only in the ipsilateral (right) hemisphere. Even hemispheric MPO activity was not different. These observations suggest that intravascular neutrophils or their secretory products are not harmful in their own right. Rather, to produce injury, neutrophils or their secretory products might need to interact with an injured endothelium. For example, neutrophil-derived MPO can catalyze a reaction with endothelial-derived hydrogen peroxide(36, 37) to produce hypochlorous acid(35, 38) and other mediators of tissue injury(39). Hydrogen peroxide is more likely to be produced in the ipsilateral hemisphere during reperfusion. Another way activated neutrophils can cause injury is by plugging blood vessels when flow is reduced(16, 19, 40, 41) or by promoting vasoconstriction. This is achieved through liberation of superoxide(42), which binds with nitric oxide(43, 44).

We measured changes in circulating blood elements, including neutrophils and NRBCs to ensure that the intravascular neutrophils counted were not merely a reflection of changes in the peripheral blood. However, circulating neutrophil counts did not change in response to the insult during the first 18 h of recovery (Table 3). We also reported the number of nucleated red cells in the peripheral blood because they were so plentiful, nearly 70 for every 100 white cells. It is conceivable that these large nucleated cells may contribute to red cell entrapment by neutrophils. Interestingly, we found that the nonnucleated red cells were nearly all relatively large reticulocytes(45). We found no correlation between the number of nucleated red cells (at the time of sacrifice) and brain swelling (data not shown). With these counts we also determined that neutrophil depletion did not affect the number of circulating NRBCs. In contrast, in human newborns the numbers of NRBCs in the peripheral blood increase sharply within hours after brain-damaging HI(46). A rise in NRBCs after the insult did not occur in the immature rat.

Table 3 Circulating white cells and NRBC after HI
Full size table

This study shows that neutrophils contribute to HI brain injury in the 7-d-old rat as neutropenia induced before HI reduced brain swelling. Neutrophil counts rose in a bimodal sequence first during HI and the first 5 min of recovery and again at 18 h of recovery. MPO activity peaked at 4-8 h of recovery and may have been affected by the amount of MPO circulating in the blood vessels of the brain. Unlike adult models of cerebral ischemia, neutrophils did not extravasate from blood vessels to infiltrate the brain parenchyma. From within blood vessels neutrophils may impair blood flow or interact with injured endothelial cells during reperfusion to aggravate vascular injury. We conclude that endeavors to reduce neutrophil-mediated damage to the developing brain post-HI are warranted. The early accumulation of neutrophils within cerebral blood vessels suggests that anti-neutrophil strategies should be targeted intravascularly and initiated as early during the insult or recovery period as possible.