Heat Shock Protein

We examined whether experimental pneumococcal meningitis induced the 72-kd heat shock protein (HSP72), a sensitive marker of neuronal stress in other models ofcentral nervous system (CNS) injury. Brain injury was characterized by vasculitis, cerebritis, and abscess formation in the cortex of infected animals. The extent of these changes correlated with the size of the inoculum (P < 0.003) and with pathophysiologic parameters of disease severity, i.e., cerebrospinalfluid (CSF) lactate (r = 0.61, P < 0.0001) and CSF glucose concentrations (r = -0.55, P <0.0001). Despite the presence of numerous cortical regions having morphologic evidence of injury, HSP72 was not detected in most animals. When present, only rare neurons wereHSP72 positive. Western blot analysis of brain samples confirmed the paucity of HSP72 induction. The lack of neuronal HSP72 expression in this model suggests that at least some of the events leading to neuronal injury in meningitis are unique, when compared with CNS diseases associated with HSP72 induction. (AmJPathol 1992, 141:53-60) (CNS) injury associated with meningeal inflammation, including vasculitis, focal CNS necrosis, neuronal loss, and inflammation of brain tissue and cranial nerves.3 Pathophysiologic studies in animals with experimental meningitis have further identified changes that could contribute to brain injury. These include alterations of cerebral blood flow,4 intracranial hypertension,5 disturbances of cerebrospinal fluid hydrodynamics,6 and the development of brain edema.5 Some of these changes have also been identified as important factors for outcome in clinical studies.7 However, little is known about the precise role of these pathophysiologic alterations in causing neuronal injury in meningitis. Heat shock protein 72 (HSP72) is a highly conserved, inducible stress protein that appears to be a sensitive marker of neuronal stress in several models of CNS injury. HSP72 has been shown to be consistently induced by both focal and global ischemia,8 10 seizure,11 and trauma.12 Based on the hypothesis that some of the molecular mechanisms responsible for neuronal injury and HSP72 induction in these other forms of brain injury may also be important in meningitis, we studied the expression of HSP72 in an animal model of bacterial meningitis. Streptococcus pneumoniae was used as the infecting organism because, among the three common pathogens of bacterial meningitis, it is associated with the most serious prognosis.13 Disease severity was modulated by varying the size of the bacterial inoculum. CNS injury was assessed by histology and by measurements of lactate and glucose concentrations in the cerebrospinal fluid (CSF). These CSF parameters of pathophysiologic derangement were chosen because both clinical14'15 and

We examined whether experimental pneumococcal meningitis induced the 72-kd heat shock protein (HSP72), a sensitive marker of neuronal stress in other models of central nervous system (CNS) injury. Brain injury was characterized by vasculitis, cerebritis, and abscess formation in the cortex of infected animals. The extent of these changes correlated with the size of the inoculum (P < 0.003) and with pathophysiologic parameters of disease severity, i.e., cerebrospinalfluid (CSF) lactate (r = 0.61, P < 0.0001) and CSF glucose concentrations (r = -0.55, P <0.0001). Despite the presence of numerous cortical regions having morphologic evidence of injury, HSP72 was not detected in most animals. When present, only rare neurons were HSP72 positive. Western blot analysis of brain samples confirmed the paucity of HSP72 induction. The lack of neuronal HSP72 expression in this model suggests that at least some of the events leading to neuronal injury in meningitis are unique, when compared with CNS diseases associated with HSP72 induction. (AmJPathol 1992, 141:53-60) (CNS) injury associated with meningeal inflammation, including vasculitis, focal CNS necrosis, neuronal loss, and inflammation of brain tissue and cranial nerves. 3 Pathophysiologic studies in animals with experimental meningitis have further identified changes that could contribute to brain injury. These include alterations of cerebral blood flow,4 intracranial hypertension,5 disturbances of cerebrospinal fluid hydrodynamics,6 and the development of brain edema.5 Some of these changes have also been identified as important factors for outcome in clinical studies.7 However, little is known about the precise role of these pathophysiologic alterations in causing neuronal injury in meningitis.
Heat shock protein 72 (HSP72) is a highly conserved, inducible stress protein that appears to be a sensitive marker of neuronal stress in several models of CNS injury. HSP72 has been shown to be consistently induced by both focal and global ischemia,8 10 seizure, 11 and trauma.12 Based on the hypothesis that some of the molecular mechanisms responsible for neuronal injury and HSP72 induction in these other forms of brain injury may also be important in meningitis, we studied the expression of HSP72 in an animal model of bacterial meningitis.
Streptococcus pneumoniae was used as the infecting organism because, among the three common pathogens of bacterial meningitis, it is associated with the most serious prognosis.13 Disease severity was modulated by varying the size of the bacterial inoculum. CNS injury was assessed by histology and by measurements of lactate and glucose concentrations in the cerebrospinal fluid (CSF). These CSF parameters of pathophysiologic derangement were chosen because both clinical14'15 and Bacterial meningitis can cause substantial neuronal dysfunction as manifested clinically by the occurrence of coma during the acute disease, and of mental retardation, learning disabilities, and focal neurologic deficits as long-term sequelae.1'2 The pathogenesis of these sequelae is not completely understood. Histopathologic studies document a spectrum of central nervous system experimental studies16 have shown that they are predictive of disease severity and outcome in meningitis.

Infecting Organism
A type 3 encapsulated S. pneumoniae originally isolated from a clinical specimen was grown on blood agar plates, resuspended in 0.9% NaCI, and stored at -700C.
To infect animals, the thawed inoculum was diluted directly to the desired concentration in 0.9% NaCI. Infection was produced by either 1 04, 105, or 1 06 colony forming units/injection.

Model of Meningitis
Meningitis was induced in male Sprague-Dawley rats weighing approximately 350 g, using the model de- The studies were approved by the Committee on Animal Research of the University of California San Francisco. The animals were anesthetized by intraperitoneal injection of ketamine (100 mg/kg) and xylazine (7 mg/kg). Animals were then placed in the ventral position with their head hanging vertically over an edge and the cisterna magna was punctured percutaneously with a handheld 25-gauge butterfly needle (Butterfly, Abbott Hospitals Inc., North Chicago, IL). Animals were infected by slow injection of 0.1 ml of the inoculum into the cisterna magna and were then allowed to awaken from the anesthesia. Twenty four hours later, the animals were reanesthetized and 50-100 ul of CSF was removed for study. Animals were randomly assigned to immediate sacrifice (n = 12) or to antibiotic therapy with ceftriaxone (100 mg/kg IM) every 12 hours for 1 (n = 10), 3 (n = 9), or 5 (n = 7) days, at which time another CSF sample (50-100 ul) was removed, and the animals were sacrificed with an overdose of pentobarbital.

CSF Examination
CSF was cultured in tenfold dilutions on blood agar plates incubated for 24 hours at 370C in room air with 5% CO2. CSF lactate and glucose concentrations were determined with a two-channel analyzer (YSI Model 2300 G/L, Yellow Springs Instruments, Yellow Springs, OH). In the first ten infected animals, CSF white blood cell counts were determined in a Neubaur hematocytometer. All animals had a pleocytosis of :'1000 cell/mm3, and this measurement was not performed in the remainder of the animals.

Brain Examination
Immediately after sacrifice, animals were perfused via the left ventricle of the heart with 100 ml of saline followed by 500 ml of 4% paraformaldehyde (PFA) dissolved in 0.1 mol/l phosphate buffer (PB) (pH 7.4). Brains were removed and posifixed for 24 hours in 4% PFA in PB and cut into 30-50 um sections on a vibratome. Sections were mounted on gelatinized glass slides for staining. After dehydration, sections were either stained in cresyl violet or acid fuchsin, the latter followed by toluidine blue for routine histopathology. Slides were rehydrated, and coverslips were fixed with Permount.
Immunocytochemistry was performed on unmounted sections washed in Tris(hydroxymethyl)aminomethane (TRIS) buffer (pH 7.6), 3 x 5 minutes, then incubated in 1% H202 in TRIS buffer for 30 minutes, washed for 5 minutes in TRIS buffer, followed by 15 minutes in TRIS A (TRIS buffer plus 0.1% Triton X), followed by 15 minutes in TRIS B (TRIS A plus 0.005% bovine serum albumin [BSA]), and then preincubated in 10% normal horse serum in TRIS B for 1 hour. After 15-minute washes in TRIS A and TRIS B, the sections were incubated with a mouse monoclonal antibody directed against HSP72 (Amersham, Chicago, IL), diluted 1:20,000 in TRIS B overnight at 40C on a rotary shaker. After additional washes, sections were incubated with a biotinylated horse anti-mouse IgG antibody (Vectastain Kit, Vector Laboratories, Burlingame, CA), diluted 1:200 in TRIS B for 45 minutes. After washing for 15 minutes each in TRIS A and TRIS B, sections were incubated in avidin-biotin horseradish peroxidase solution prepared according to the manufacturers instructions (Vectastain) for 1 hour. Sections were washed again in TRIS for 3 x 5 minutes, reacted with a solution containing 0.05% of diaminobenzadine tetrahydrochloride plus 0.1 % glucose oxidase plus 0.2% amonium chloride plus 0.8% glucose in TRIS, washed several times, mounted, dehydrated and coverslipped. Staining always included positive and negative controls. Positive controls consisted of brain sections obtained from animals injected intraperitoneally with 10 mg/kg of kainic acid, a potent inducer of HSP72,18 24 hours before sacrifice. Negative controls underwent all the steps except for the exposure to the primary antibody.
All sections were read blindly by one of us (DHL). Routine histopathology sections were assessed for subarachnoid or ventricular inflammation. Parenchymal involvement was graded as follows: normal = no vasculitis (defined as granulocytic accumulation around vessels without brain tissue destruction), cerebritis (granulocyte accumulation in brain tissue without destruction), or abscesses (granulocyte accumulation and tissue destruction); 1 + = foci of cerebritis or vasculitis, but without abscesses; 2+ = two or less abscesses per section (with or without additional foci of vasculitis or cerebritis); 3 + = three to five abscesses per section (with or without vasculitis or cerebritis); 4 + = more than five abscesses (with or without vasculitis or cerebritis) per section. The presence of HSP72-LI was assessed descriptively.

HSP72 Detection by Western Blot Analysis
Sodium dodecylpolyacrylamide (10%) gel electrophoresis (SDS-PAGE) was performed on brain samples from the cortex of a control rat, a rat sacrificed 48 hours after induction of seizures by intraperitoneal injection of 10 mg/ kg kainic acid, and a rat sacrificed 24 hours after induction of meningitis by 1 o6 colony forming units (cfu) of S. pneumoniae. In the rat with meningitis, the contralateral hemisphere was fixed in 4% PFA and histologic examination documented the presence of several abscesses (at least 3 + severity). Brain samples were solubilized in Laemmli sample buffer, and the proteins separated by SDS-PAGE. Approximately 20 ug of protein per sample (determined by Bio-Rad Protein assay (Bio-Rad, Richmond, CA) following the manufacturer's instructions) were electrophoretically transferred to nitrocellulose membranes and probed with the monoclonal antibody to HSP72 described earlier. Visualization of bound antibody was achieved by a commercial kit, using alkaline phosphatase-mediated color development (Mouse Immunodetection Kit, Boehringer Mannheim, Indianapolis, IN).

Statistical Analysis
CSF results are given as mean + SD and results between groups were compared by one-way analysis of variance, followed in the case of significant differences by Student's t-tests, corrected for multiple comparisons. The association between the histologic findings and the inoculum size was assessed by Chi-square analysis, and the correlation of histologic findings with CSF lactate and glucose concentrations was examined by Spearman rank correlations.

Characterization of Meningitis
All infected animals had positive CSF cultures 24 hours after infection. CSF titers were not significantly different whether animals were infected with 1 04, 1 05, or 106 cfu ( Table 1), indicating that bacterial growth in the CSF in all three groups had reached the stationary phase at the time of examination. However, CSF lactate concentrations were significantly higher in animals infected with the highest inoculum compared with animals infected with the lowest inoculum (P < 0.05) (Table 1). Similarly, CSF glucose concentrations declined progressively with increasing inoculum sizes (P < 0.05) ( Table 1). This suggests that the cumulative exposure to bacteria during the infection time, rather than the final titer, was critical for the degree of disease severity.
Treatment with 100 mg/kg ceftriaxone IM every 12 hours resulted in sterile CSF cultures within 24 hours of therapy. CSF chemical changes, however, normalized only after at least 3 days of therapy (data not shown).

Histopathology
After 24 hours of infection, all infected animals had pronounced infiltration of granulocytes in the subarachnoid space (Figure 1 A). In most animals infected with the two higher inocula, inflammation was also found in the ventricles (Figure 1 B), which in some animals was associated with focal ventricular hemorrhage. Density of the subarachnoid and ventricular infiltrates declined progressively during antibiotic therapy and after 5 days granulocytes in the subarachnoid space had completely disappeared. No inflammation was present in animals injected with sterile saline.
The most prominent finding on routine histopathology was multiple foci of tissue destruction in the cortex of most animals infected with the two higher inocula ( Table  2). Since no obvious quantitative effect was observed for the histology, whether animals were evaluated after 24 hours of infection or after various durations of antibiotic therapy, the rating of the histologic findings was combined for the different durations of therapy. The severity of focal lesions increased significantly with increasing inocula (P < 0.003; Table 2) and also correlated significantly  with CSF lactate concentrations (r = 0.61; P < 0.0001) and inversely with CSF glucose concentrations (r = -0.55; P < 0.0001). The appearance of these foci was uniform and characterized by the accumulation of large numbers of granulocytes with a sharp margin between pus and normally appearing brain tissue. A vessel could be identified traversing the center of many of these abscesses and appeared to represent the origin of the inflammatory process, since we found all stages of inflammation from focal vasculitis (Figure 1 C, D) with intact sur-rounding brain tissue to large abscesses with central vessels (Figure 1 E). Some of the foci had a prominent hemorrhagic component, while others extended along a vessel spanning a large portion of the depth of the cortex. The intensity of the granulocytic infiltrate decreased over time and gliosis was evident by 5 days (Figure 1 F). No tissue destruction was seen in saline controls. Other than a hemorrhagic abscess in one animal extending into the subcortical white matter, no deep structures of the brain showed signs of tissue destruction.  Figure 2E) and was specific for HSP72 ( Figure 2F

Heat Shock Protein 72 Expressio
Remarkably few neurons were positive for munocytochemical examination, and there icant association between HSP72 immun( disease severity ( Table 3). Treatment dic obvious effect on the detection of HSP72 rons, since positive cells were present b( sacrificed after 24 hours of infection as we imal after 5 days of therapy. Therefore HSP72 presented in Table 3 has been po( rations of therapy. In the group of animalc the lowest inoculum, only one animal had cortical neurons in association with an at 2A). Even in the groups with more se) HSP72 positivity was found only in a minoril a few hippocampal or cortical neurons in abscesses, with the exception of one anim many HSP72 positive neurons in the CA3 a of one hippocampus ( Figure 2B). Most ar tensive tissue destruction showed no HSP D). No saline control animal showed H cells.
In support of the results obtained by chemistry, Western blot analysis did not de samples of a rat with severe meningitis readily indentified in a positive control anir acid-induced seizures (Figure 3).
To address the possibility that the proc gitis might have inhibited the expression examined an animal that was infected for then given kainic acid to induce seizures.2 we found evidence of meningitis with focal,  tect cells from potentially fatal stress.2627 HSP72 is not detectable in unstressed cells, including neurons, but its concentration readily increases in cells exposed to conditions that are believed to lead to altered conformation or targeting of intracellular proteins.25 HSP72 has become recognized as a sensitive marker of neuronal stress after stroke, trauma, or seizures.8 1112 Since some of the mechanisms important in these forms of injury, particularly ischemia, have also been implicated in the patho-physiology of meningitis, we expected to find HSP72 induction in this disease.
Despite evidence of severe CNS injury, HSP72 expression was rarely observed in this model. The process of meningitis did not block the detection of HSP72, since kainic acid readily induced HSP72 in an animal with meningitis. Expression of HSP72 was assayed at 24 hours and beyond, because several considerations made it unlikely that a transient expression of the protein before the  28 Second, in other studies of CNS injury in which HSP72 was induced by a temporally defined stimulus, such as ischemia or seizures, HSP72 is consistently found after 24 to 48 hours. 8 11'29 Finally, in our model, the meningitic process, in contrast to a temporally defined stimulus, evolves over time and reaches its maximal intensity shortly after institution of antibiotic therapy30 (i.e., after 24 hours of disease in the present study). We cannot exclude that some injured neurons expressed HSP72 mRNA, but were unable to express the protein due to a loss of translational capacity. Nonetheless, it is likely that the paucity of immunocytochemical staining for HSP72 was largely due to a lack of induction of the protein.
In light of the potency of cerebral ischemia to induce HSP72, the lack of HSP72 expression in the current study indicates that the ischemic threshold necessary for induction of the protein is not reached. This is supported by cerebral blood flow studies in patients and experimental animals with bacterial meningitis that generally show only moderate reductions of flow.4 '16'3133 To what extent milder forms of ischemia that do not lead to HSP72 induction still contribute to CNS injury in meningitis is presently unknown.
Two forms of injury that may play a role in meningitis have recently been shown to not be inducers of HSP72. First, hydrogen peroxide and other oxidants have recently been shown to be cytotoxic in tracheal epithelial cell cultures without inducing the production of HSP72.34 Acute inflammation is the hallmark of brain tissue destruction in the present model and granulocytes may play an essential role in mediating injury in experimental meningitis.21'23 Granulocytes release proteolytic enzymes and oxidants, including oxygen-derived radicals and hydrogen peroxide that can be toxic to host tissue.35 It is tempting, in light of this evidence, to speculate that if CNS injury is caused by granulocytes, this may also occur without HSP72 induction. Secondly, the excitotoxin glutamate,? while being cytotoxic in primary neuronal cell cultures, does not induce HSP72 production in this setting. 37 We have recently found that glutamate is markedly elevated in the CSF during meningitis in rabbits.' Thus, excitotoxicity from glutamate may be another component of the pathophysiology of meningitis that leads to CNS injury without HSP72 expression.
In conclusion, there are two important observations in this study. First, we noted a strong correlation between the morphologic injury and evidence of pathophysiologic alterations; and second, HSP72 induction is not a prominent feature in this model of pneumococcal meningitis, in spite of inflammation-induced focal brain injury and metabolic dysfunction. This suggests that mechanisms mediating CNS injury in this model are different from those in other forms of injury that lead to HSP72 expression, presumably as a result of intracellular protein perturbation. Our study underscores the need to clarify the molecular mechanisms by which inflammation causes CNS injury during meningitis.