Morphologic alterations of the blood brain barrier with experimental meningitis in the rat Temporal sequence and role of encapsulation

Morphologic alterations of the blood-brain

barrier with experimental meningitis in the rat.

Temporal sequence and role of encapsulation.

V J Quagliarello, … , W J Long, W M Scheld

J Clin Invest.

1986;77(4):1084-1095. https://doi.org/10.1172/JCI112407.

The cerebral capillary endothelium is unique and functions as an effective blood-brain

barrier (BBB) owing to its intercellular tight junctions and rare pinocytotic vesicles. To

assess how bacterial meningitis alters the BBB, rats were inoculated intracisternally with

three encapsulated meningeal pathogens (Escherichia coli K1+, Streptococcus

pneumoniae type III, Haemophilus influenzae type b) and an unencapsulated mutant strain

(H. influenzae Rd). After defined infection durations, the morphologic alterations of the

cerebral capillary endothelium were quantitatively assessed by transmission electron

microscopy. Results revealed a significant increase in pinocytotic vesicle formation (P less

than 0.001) early after meningitis induction (4 h) that was sustained with longer infection

durations (10 h, 18 h) for all encapsulated strains tested. In addition, there was a

progressive increase in completely separated intercellular junctions with increasing

infection duration, (P less than 0.05). 4 h after induction of meningitis with H. influenzae Rd,

cerebrospinal fluid (CSF) bacterial concentrations, cerebral capillary morphologic changes,

and functional BBB permeability to circulating 125I-albumin were similar to those observed

with H. influenzae type b. However, prolonging the H. influenzae Rd infection to 18 h

allowed for CSF clearance of the organism, thereby precluding the significant increase in

separated junctions or progression of functional BBB permeability seen with the

encapsulated H. influenzae type b. These data suggest a uniform morphologic explanation

for altered BBB permeability […]

Research Article

Find the latest version:

Morphologic

Alterations

of the Blood-Brain

Barrier

with

Experimental

Meningitis in

the Rat

Temporal

Sequence

andRole of Encapsulation

Vincent J.Quagliarello, WilliamJ.Long, and W. Michael Scheld

DepartmentsofInternalMedicine andNeurosurgery, University ofVirginiaSchoolofMedicine, Charlottesville, Virginia22908

Abstract

Thecerebral capillary endothelium is unique and functions as aneffective blood-brainbarrier(BBB) owing to itsintercellular tightjunctionsand rarepinocytoticvesicles.Toassess how

bac-terial meningitisalters the BBB, rats wereinoculated

intracis-ternally with three encapsulated meningeal pathogens (Esche-richiacoli K1+,Streptococcus pneumoniae type III,

Haemo-philus

inpluenzae

type b) and an unencapsulated mutant strain (H.

influenzae

Rd). After defined infection

durations,

the

mor-phologic alterationsofthe cerebralcapillary endotheliumwere

quantitatively assessed by transmission electron microscopy.

Results revealed asignificant increaseinpinocytoticvesicle for-mation(P <0.001) earlyafter meningitis induction (4 h)that wassustainedwithlonger infectiondurations(10 h, 18 h) forall

encapsulated strainstested. In

addition,

therewas aprogressive increaseincompletelyseparated intercellularjunctions with

in-creasing

infection

duration, (P

<

0.05).

4 h after induction of

meningitis with H.

influenzae

Rd,

cerebrospinal

fluid (CSF) bacterial

concentrations,

cerebral

capillary morphologic

changes,

andfunctionalBBB

permeability

to

circulating 125I-albumin

were

similartothose observedwith H.

influenzae

typeb.However, prolonging the H.

influenzae

Rd infectionto 18 h allowed for

CSFclearanceof the

organism, thereby

precluding

the significant

increase inseparated

junctions

or

progression

offunctionalBBB

permeability

seenwiththe

encapsulated

H.

influenzae

typeb. These data suggest auniform

morphologic explanation

for altered BBB

permeability

in

meningitis

with a

reproducible

temporal

sequence.

Encapsulation

doesnotappear essential for BBB

in-jury,

but mayfacilitate its

_{progression by allowing}

the

organism

toevadehostclearance.

Introduction

Bacterial meningitis

remains a common disease with a

high

morbidity

and

mortality despite

effective bactericidalantibiotic

therapy (1). One

hypothesis

is that altered function ofthe

blood-This studywaspresented in partatthe annualmeetingof the American Federation for Clinical Research,Washington,DC, May 1985, and has appearedas anabstract(1985.Clin. Res. 33:416A).

Addressreprintrequeststo Dr.Scheld,Divisionof InfectiousDisease,

Box385,UniversityofVirginiaSchoolofMedicine,Charlottesville,VA 22908.

Receivedfor publication20May1985.

brainbarrier(BBB)'secondary toinfection in the subarachnoid

spacefacilitates irreversibleneuronalinjuryprior to, and despite,

bacteriologic cure. Hence, defining the precise mechanism of

infection-inducedBBBinjury iscrucial to a better understanding

ofthisdisease.

Anatomically, investigations pioneeredbyReese and

Kar-novsky(2)have localized thecerebral capillary endothelium as

themajor siteresponsible for theBBB. This is attributedtoits unique characteristics as an endothelium compared with

sys-temiccapillaries. Systemic capillaries (i.e., thoseofglands and

visceralmucosa)arefenestrated and allow intercellularpassage

of macromolecules and intact circulatingcells;pinocytotic

ve-siculartransportiscommonbutmitochondriaarerare. In con-trast,cerebralcapillariesexhibitrarepinocytosis,and theplasma membranes ofadjacent endothelial cells are fused togetherin

the form of continuous intact intercellular tightjunctions, or

zonulae occludentes (2, 3). This imparts on it the characteristics ofahigh-resistanceendothelium (4, 5),and allowsittofunction

as aneffective barriertomacromoleculartransport and cell

ex-udation.Additionally,the highmitochondrialdensity of the

ce-rebral endothelium (6) facilitates selective energy-dependent

ac-tive transport of

Na'

and K+ at the antiluminal membrane(7),

and maintains homeostasis of the perineuronal interstitialfluid.

Hence, only lipid-soluble substances or those transported by carrier-mediated diffusion (e.g., glucose, essential amino acids)

traversethe BBB under normal circumstances (8-10).

Becausethe cerebrospinal fluid (CSF) is in direct continuity with the extracellular space bathing the cerebral capillaries, meningeal infection is a logical precipitant of BBB injury. This injury may consist of disruption ofenergy-dependent active transport systems, aswell asbarrier permeability arising from increasing pinocytosis or separation oftightjunctions.

Despite decades of investigation of the BBB and its alteration with experimental injury, there remains a distinct paucity of data regarding its responsetoinfection. In thisinvestigation,we

havedevelopedamodel of experimental meningitisintherat

to identify the natural historyof the morphologic alterations

inducedintheBBB.Specificallyourgoalswere:(a)to assessthe influence of experimental bacterial meningitis on pinocytotic vesicle (PV) formation and intercellular junctionintegrityinthe cerebralcapillaryendotheliumusingtransmission electron

mi-croscopy; (b)to compare the morphologic changesseen with three encapsulated meningeal pathogens (Streptococcus pneu-moniae typeIII,EscherichiacoliKl +,Haemophilus influenzae

type b)-all threepossessing capsulesof differentcomposition;

1.Abbreviations usedinthispaper:BBB,blood-brainbarrier,cfu, colony-forming unit(s);CSF,cerebrospinal fluid; HRP,horseradishperoxidase;

PRP, polyribosephosphate; PV, pinocytotic vesicle(s); SJ, separated junction(s);TJ,tightjunction(s);WBC,white blood cell.

1084 V.J.Quagliarello, W.J.Long,andW.M. Scheld

J.Clin. Invest.

© TheAmerican Societyfor Clinical Investigation, Inc.

(c) to identify the temporal sequence of these morphologic

changes byquantitating theirappearanceatinfection durations

(4, 10, and 18 h) thatspan the natural history of the disease;

and(d) correlate the morphologic alterations in H. influenzae

typebmeningitis with functional BBB permeability tocirculating

'25I-albumin, andcontrastthesechanges with themorphologic

andfunctional alterations induced byanunencapsulated mutant, H.

influenzae

Rd.

Methods

Challenge organisms. The encapsulated bacterial strains selected forstudy

werecarefully chosentorepresent relevant human meningeal pathogens with different capsular compositions. The strain of S.pneumoniae type IIIused wasderived fromaclinical(CSF) isolate and has been previously characterized in thislaboratory (1 1). 8-h late log phase organisms grown intrypticase soy broth(DifcoLaboratories, Detroit, MI) supplemented with5%defibrinated sheep bloodat370Cwerecentrifuged (250 g for 5 mintosedimenterythrocytes), the supernatant recentrifuged (2,500 g for 15 min), theensuing pelletwashed inphysiologicsalinetwice, and resuspended in 10mlof 0.9% NaCl. The E.coli Kl+ isolate (strain C94, kindlyprovided by Dr. G. McCracken,Dallas, TX) previously charac-terized in thislaboratory, wasgrownto8-hlog phaseintrypticase soy broth, centrifuged, washed, and resuspendedasdescribed above. The H.

influenzaetype b (Eagan) strainaswellastheunencapsulated H.

influ-enzae Rdmutant wereobtained from the laboratory of E. Richard Moxon

(Oxford University). BothH. influenzaestrainsweregrown in brain-heartinfusion broth, supplemented with 5% Fildes enrichment media (BBLMicrobiology Systems, Cockeysville, Md.) and afteran8-h sub-cultureofanovernightgrowth, werecentrifuged, washed, and resus-pended inphosphate-bufferedsaline. All finalsuspensionsof challenge

organisms weredesigned to achieve aninoculum of I06>65

colony-formingunits(cfu)in ordertoelicitafatal meningitis. Log-phase or-ganisms were usedtofacilitatemaximalcapsuleexpressioninthe chal-lengeinocula.

Inductionofmeningitis. For the purpose of this investigation, itwas necessary todevelop a new model of experimental meningitis in the rat. AdultWistar rats(150 g) were anesthetized withacombinationof Ke-tamine(Parke-Davis,Morris Plains, NJ)/Xylazine (Miles Laboratories,

Shawnee,KS)at adoseof 100 and7mg/kg,respectively,by the intra-muscular route. 75Mlof CSF was removed via intracisternal puncture

usingamicromanipulator fitted toa25-gauge butterfly needle (Abbott Inc., NorthChicago, Ill.) calibrated for fluid increments of 25Al.After removalof CSF, 50 Ml ofthe challenge organism(lO'-'cfu) was injected

intracisternally into each experimental group, with 50

_Al

of saline (or phosphate-buffered saline) injected into simultaneous matched controls. Eachexperimental group consisted of at least three animals. After in-oculation, meningitis was allowed to progress for defined durations (4, 10, and 18 h) and the CSF was resampled (75Al)for determination of quantitative culture and white blood cell counts (WBC). Simultaneous blood samples were also obtained for quantitative cultures. 10-fold di-lutions of CSF and blood were made in physiologic saline, and bacterial concentrations were determined by surface colony growth on blood agar

(S.pneumoniaeIII,E.coliKl+)and chocolate agar (H.influenzaetype b and Rd) plates. CSF WBC concentrations were determined with a

hemocytometerby standardmethods.

Cerebral capillary isolation. Immediately after confirmation of men-ingitis for each challenge organism at each time point (i.e., 4, 10, 18 h), theanimals were killed by decapitation, and the intact brains were re-moved rapidly and placed in cold M199 media (Hank's salts, Gibco, Grand Island, NY) containing 20 mM Hepes, 100 U/ml penicillin, and 10Mg/ml streptomycin at pH 7.4. The meninges were removed with sterile paper tissue, and the brainstem, hindbrain, and choroid plexus werediscarded using sterile instruments. The cerebral cortices from each

experimental groupwerepooledin cold M199 mediaand mincedwith threepairsof sterile

scissors,

followed

_by

_{homogenization}

with 20 strokes

usingaloose-fitting(0.25-mmclearance)homogenizerat400 rpm after methodsdeveloped by Betz and Goldstein (8). The homogenate was

dilutedto 200 ml with M199 andcentrifugedat 1,000g for 10 minat

4VC. The supernatant wasdecanted, thepellets wereresuspended by

shaking for 60swith 175 ml of 15% dextran(molwt60,000-90,000,

Sigma Chemical Co., St. Louis, Mo.) inM199 media, and the suspension centrifugedat4,000 g for 10 min at 4VC.Thesupernatantcontaining

fat and dextran was discarded and thepelletcontaining cerebralcapillaries

resuspended in 10 ml of M 199. The cerebral capillaries were collected on a53-gmnylonmesh, washed with 150 ml of M199, andimmediately

fixedwith 2%glutaraldehyde in 0.1 Mphosphatebuffer, pH 7.3. Preparationfortransmissionelectronmicroscopy. The fixedcapillaries

were washed threetimes, (5 min per wash) in 0.066 M phosphatebuffer, pH 7.2, andcentrifuged for 5 minat4,000 g. Progressivedehydration

wasperformed by rinsing samples in 40%, 60%, 80%, and 100% ethanol solutions, followedby furtherdehydration in a 1:1 solution of100%/

ethanol/propylene oxide for 10min, 100%propylene oxide for 10min,

andfinally ina 1:1 solution ofpropylene oxide/epoxy resin(standard Epon812 mixture) overnight. Subsequently, the capillarieswereplaced in 100% epoxy resin for 1 h, baked for 48-72 h in plastic capsules, and sectionedon anultramicrotome (MT-5000 DuPont-Sorvall,Wilmington, DE). Multiple thin sections (-70 nm) were stained with 10% uranyl acetateandexamined by transmission electron microscopy.

Horseradish peroxidase (HRP) visualization. The method used was amodification basedontheprocedure described by Graham and Kar-novsky (12). To qualitatively assess HRP uptake into PV andpenetration through separated intercellular junctions, selected experimental groups wereinjected 30 min prior to sacrifice with 60 mg type II HRP (Sigma Chemical Co.) intravenously, with the capillaries subsequently isolated asabove.Afterfixation, the tissue was rinsed for 3hat 0C in several

changesof 0.1Mphosphate buffer, pH 7.3, with 3.5% sucrose. The tissue was thenincubated for 30min at 25°C in 0.05% diamino-benzidine, tetra-HClsalt(Sigma Chemical Co.) in 0.1 M phosphate buffer, pH 6.0, followed bya45-minincubation at 25°C in 0.05% diamino-benzidine in 0.1 Mphosphate buffer, pH 6.0, with 0.01% H202. The capillaries werethenrinsed in 0.1 M phosphate buffer, pH 7.3, at 0°C for 5-10 min and processed for transmission electron microscopy as above.

Quantitative analysis of the morphology of cerebral capillary endo-thelium. For each experimental and simultaneous control group, 30 ran-domintactcapillary cross-sections were quantitated blindly as to the numberof PV, intact closed junctions, and separated junctions that were visualized. Separated junctions were defined as areas where plasma membranesof adjacent endothelial cells lining a capillary lumen were completely separated. Closed junctions were defined as areas where these

adjacentmembranes werecontinuously or partially joined. PV were de-finedasintactintracytoplasmic membrane-bound vesicles (70-180 nm); thoseincompletely forming at the luminal membrane or fusing with the antiluminal membrane were not quantitated. Only capillary sections <15 Mm indiameter with intact luminal and antiluminal membranes weresubjected to quantitative analysis. Because each thin section con-tained.50 capillary cross-sections, every effort was made to confine

quantitationtothe samethin section in order to avoid counting the same

capillarysection twice.

Quantitativefunctionalassessment ofBBB permeability to)25I-bovine

serumalbumin (BSA). To correlate morphologic with functional alter-ations of the BBB, rats were injected intravenously with10-20 MCi(0.15 ml) of'251-BSA(ICN Radiochemicals, Irvine, CA) -30min after

Statistical analysis. The number of PV in experimental and control

groups were compared using the Student ttest(two-tailed unpaired). The proportion ofjunctions completely separated in experimental groups werecompared with controls using the Fisher exact test (two-tailed). The meanpercentage of'251-BSApenetration in experimental and control groups was compared using the Student t test (two-tailed, unpaired).

Results

Qualitative

effects

ofexperimental meningitis in the rat

on

the morphology ofcerebral capillary endothelium:

PVformation and

TJ integrity

Morphologic assessmentof PV formationbythe cerebral

cap-illaryendothelium revealed a uniform host response to experi-mental meningitis with the encapsulated pathogens tested (S. pneumoniae type III, E. coliK1,H.

influenzae

type b). Namely,

therewas anincreasedcytoplasmic content of PV, visibly active

formation

attheluminalmembrane, and demonstrable fusion ofvesicleswith the antiluminal membrane not seen in control preparations inoculated with saline. Depicted in Fig. 1 are ex-amples of PVformationin cerebral capillary endothelium after 4hofE. coliKl+ meningitis compared with a control

prepa-ration.

Whenalterations in the integrity ofthe intercellular junctions wereassessed, amorphologic spectrum wasrecognizedranging

fromcompletelyclosedtocompletely separatedwith infection.

Examplesof typical intercellular junctionalcomplexesvisualized

are

depicted

in Fig. 2.

Fig.

2aandb

depict

examples of

com-pletelyseparated intercellular junctions in ratcapillaries after

inoculation with H. influenzae Rd and

E.

coli Kl +. Note

that despite complete discontinuityin theencasementofthe lumen

bytheendothelialcelllayer,theluminalandantiluminal

mem-branesareintactwith little cytoplasmic rarefaction. Ajunctional

complex that appearstobe intransition (i.e., partially separated) isshowninFig. 2 c, from a capillary cross section 18 h

postin-oculation with E.

coli

Kl+. Evident isa

junctional

complex between

adjacent

endothelial cell membranes that is

almost,

but

notcompletely, separated.

Toqualitatively assesswhetherthese

morphologic

changes

werefunctionally significant, selected groupswereinjected

in-travenously 30 min

prior

to

being

killed with60 mg of HRP

(40,000 molwt)to assesslocalization ofa macromolecule the

approximate size of albumin. As

depicted

in

Fig. 3,

after 18 h

of experimentalE.coli

meningitis,

one seesthe presenceof

per-oxidase

activity

within

cytoplasmic

vesiclesaswellas

penetrating

throughaseparated intercellular

junction.

Quantitative

effects ofmeningitis

on

PVformation

and

TJ

integrity: temporal

sequence

Inordertocompare

quantitatively

the

degree

of

morphologic

alterations induced, 30 random endothelial cellsections

(<15

,um indiameter) wereexamined

blindly

in each

experimental

(and its respective

control)

group. The number of

intracyto-plasmicPVand

completely separated

intercellular

junction (SJ)

wereassessed and quantitated

by

blinded

analysis

4,

10,

and 18hafter inoculation.

E.coli KJ+ meningitis. Asshown in

Fig.

4a,a

statistically

significantincrease in PVformationis notedas

early

as4 h after

inoculation with E. coli Kl+

compared

with saline-matched controls ([mean+SE] numberof PV/100-nm capillary length,

5.57±0.57 vs. 2.76±0.37; P <0.001). Although quantitatively

greatestat4

h,

thisincreasedPVformation remained

significant

at 10 h(P<0.02) and 18 h (P<0.01)postinoculation.

In contrast, the percentage of junctional complexes

com-pletelyseparated (representing the number of separated junctions per total number of junctions present in the same 30 capillary

sections) increased with increasing infection duration, and reachedstatisticalsignificance 10 h postinoculation. For example, 4 h postinoculation with E. coli, 9% of the junctional complexes (4/44) were completely separated as opposed to 0% (0/48) of

controls (P> 0.05). However, when infection was allowed to progressfor 10h, 12% of junctions were separated (6/50) and weresignificantly increased (P = 0.022) when compared to con-trols(0/53). By 18 h postinoculation, the junctional separation with infection (10/54, 19%) reached greater significance (P

<0.002) comparedtocontrols(0/51) (Fig. 4 b).

This temporal sequence of morphologic changes (i.e., early and sustained increase in PV formation, progressive increase in junctional separation) with increasing infection duration was

associated with sustained CSF bacterial concentrations and WBC pleocytoses (Table I).

S.pneumoniae type IIImeningitis. Similar morphologic al-terations of the cerebral capillary endothelium were noted after intracisternal inoculation with S. pneumoniae type III. As shown in Fig. 5 a, a statistically significant increase in PV formation

asearlyas4 h postinoculation with S. pneumoniae ([mean±SE] number ofPV/100-nm capillary length, 6.38±0.67 vs. 2.76±0.37 incontrols; P < 0.001) is evident. Although the mean number of PV per capillary section was quantitatively greatest at 4 h postinoculation, PV formation remained significantly increased wheninfectionwasallowed to progressfor 10 h (P<0.05) and 18 h (P<0.01).Fig. 5 b,displayingthechanges in intercellular junction integrity, demonstratesasimilarincrease incompletely

separated endothelial cell junctions that again reaches statistical

significance 10 hpostinoculation. Thatis,4 h postinoculation

withS. pneumoniae III, 7% (3/43)ofthejunctionalcomplexes

quantitatedwerecompletelyseparated asopposedto0%(0/48) ofcontrols (P > 0.05). However, when infection was allowedto

progressfor 10 h, 18%(10/56)ofjunctionswereseparated and were statistically increased (P<0.002) compared to controls (0/53). Further progression of infectionto18 h showed that 17% (9/43) junctions separated, a value that again was highly statis-tically significant (P < 0.003) when compared to simultaneous

saline-injectedcontrols (0/51).

Thesemorphologic changeswerealsoassociated with similar

sustained CSF bacterial concentrations and WBC pleocytoses withincreasing infection durations (Table II).

Correction of morphologic andfunctional alterations of the BBB in

experimental

H.

influenzae

meningitis. After

confirming

this remarkably similar temporal sequence of morphologic

changesin the cerebralcapillaryendothelium with E. coli Kl + andS.pneumoniaetype IIImeningitis,wecontrasted the

mor-phologicalterations induced bytheencapsulatedH. influenzae

type b and the spontaneous unencapsulated mutant H. influ-enzaeRd.

Asshown in Fig. 6,the sequence of

morphologic

changes

seenwith the

encapsulated

H.

influenzae

typebwassimilar to the othertwoencapsulated

pathogens. Specifically,

4 h

postin-oculation with H.

influenzae

typeb there was astatistically sig-nificant increase in PV formation

compared

with controls

([mean±SE]numberofPV/I00-nm capillary length,5.24±0.52

vs.

2.72±0.35;

P <

_0.001).

This increased PV formation with

1. _0i*. ''I r,

-,.P%*- .7

W.;..- _{.". Al.}

'. I. ."

"aide

*e

.

1'.., 4vv

0

,-Figure 1. PV formation incerebralcapillaryendothelium.(A) Control capillarycrosssection 18 h after salineinoculation. Note the visibly intact intercellularjunction (white arrow) but the virtual absence of vesicles.E,intraluminal erythrocyte. X 15,000.(B) Capillary cross

sec-tion4hpostinoculation withE.coliK1+showingvesicle formation atthe luminal membrane(arrowheads)and within thecytoplasm

(ar-rows). E,intraluminalerythrocyte;N,nucleus. x60,000.

16

A

X

,;

....'

v*X#S *_

*

.X.f.ia

Ad

*

*-eSCz _w

t.,.:s

Or

Ha.. i

.. < __

_

A_

IF._

-_

K jX

_e[

*

X

. A,

A.,

S

.

|

.kz

-AN,,,, 8 .t'9w _ _

*

to£ w

X,

=

_

WS

B

1088 V. J.Quagliarello, W.J.Long,andW.M.Scheid

'n..,

V.-C

...RI .,.

c

Figure 2. Morphologic alterations of intercellular junctions in cerebral capillary endothelium. (Opposite) (A) Capillary cross section 18 h post-inoculation with H.influenzae Rd. Note the complete separation be-tweenadjacent endothelial cell membranes (arrow), which are other-wise continuously intact around the lumen (L). A visibly intact junc-tion(arrowheads) is also present. Note the surrounding basement membrane(bm), which appears damaged by the infection (X 20,000). (B) Capillary cross section4hpostinoculation with E. coli KI+. Note thewide separation between intact cell membranes of two endothelial

Figure 3.HRPvisualization. Capillary cross section after 18 h ofE.

coliKl+meningitis.Note theintracytoplasmicvesicles(150-180 nm)

filledwith peroxidase, includingonethat appearstobefusing with the antiluminal membrane(arrowheads).Moreover, noteperoxidase

cells(black arrows) that are otherwise joined by an intact intercellular junction(white arrow) at their other ends. Scattered cytoplasmic de-bris is present within the intercellular space but the basement mem-brane(arrowheads) is intact. E, erythrocyte (X 15,000.) (Above) (C) Capillarycrosssection 18hpostinoculation withE.coli Kl+. Note the junctional complex (arrow) with a widened intercellular cleft but without complete separation;bm, basement membrane; L, lumen (X30,000).

productlining the endothelial lumen andpenetratingthroughan in-tercellularjunction (arrow). E, erythrocyte; bm,basement membrane (x30,000).

Time (h) post inoculation

20 *

10

0O

-0 4 10 18 Time (h)post inoculation

Figure 4.Morphologicchangesin cerebralcapillary endothelium with experimentalE.coliKl+meningitis. (a)PV formationvs.time after induction of meningitis. (b)Percent (%)SJvs.timeafter induction of meningitis. (i)E.coli;(o) control; (time 0 o) morphology ofcerebral endothelium priortoinoculation.P<0.05 comparedtocontrols(see

textfor actual Pvalues).tThe number ofSJ/totalnumber of junc-tionsvisualized in 30 random endothelial cell sections.

infection remained significant when infection wasallowed to

progressfor 10 h(P<0.001) and 18h(P<0.01). When ex-aminingintercellularjunction integrity,thepercentage

ofjunc-tionalcomplexes completely separated increasedwithlonger in-fectiondurations,andagain reached statistical significanceafter 10 hofmeningitis. That is,4h postinoculation with H. influenzae

typeb,4% of thejunctional complexes (2/50)werecompletely separatedasopposedto0%(0/45)of controls(P>0.05). How-ever, when infection progressed for 10 h, 12% ofjunctions (7/58)wereseparatedandweresignificantlyincreased(P<0.02) when comparedtocontrols(0/52). After 18 h of infection, 15%

TableI. CharacteristicsofExperimental

E.coliKJ+Meningitis in Rats

Duration

of CSF bacterial WBC/ml

infection Inoculum concentration* ofCOF

h n log0, fOu/ml X10J

4 E.coliK1+ (3) 6.84±0.55 5.07±3.76

Saline (3) 0 0.05±0.03

10 E.coliKI+(4) 5.45_0.68 2.38±1.03

Saline (3) 0 0

18 E.coliKl+(6) 5.19±1.06 9.33±1.99

Saline (6) 0 0.33±0.25

Values given asmean±SE.

*_{P >0.05comparing 4-, 10-, and 18-h CSF}

_bacterial

_{concentrations.}

a*

4-mean±S E ₃

No.of7PV

Onm 2 |

capillary length

0 4 10 18

Time(h)post inoculation

b 20 * *

separated 1₀

junctions4

0 4 l0 18

Time (h)post inoculation

Figure 5.Morphologic changes in cerebral capillary endothelium with experimental S. pneumoniaetypeIIImeningitis. (a) PV formationvs. timeafter induction of meningitis. (b) Percent (%) SJvs.time after in-duction of meningitis. (i) S. pneumoniae; (o) control; (time00) mor-phology of cerebral endothelium priortoinoculation.P<0.05 com-pared (seetextfor actual P values).tThe number ofSJ/total number ofjunctions visualized in 30 random endothelial cell sections.

ofjunctionswereseparated(8/52)adifferencewhichremained significant (P<0.05) whencomparedwith controls (1/46). These morphologic changeswereassociated withastatistically signif-icant increase in CSF bacterial concentrations comparing the 4-h

([mean±SE] loglo

cfu/ml of CSF, 6.26±0.34), with 18-h ([mean±SEJ logocfu/ml, 8.06±0.35)infection(P<0.05) (Table III).Inaddition,unlikethe other twoencapsulated pathogens, 100% of the animalswereconcomitantlybacteremicat all time

TableII. Characteristics ofExperimental

S. pneumoniaeTypeIIIMeningitisinRats

Duration

of CSFbacterial WBC/ml infection Inoculum concentrations of CSF

h n log0, fu/ml X10

4 S.pneumoniae (4) 6.28±0.18 3.11±2.39

Saline(3) 0 0.05±0.03

10 S.pneumoniae

5.91_0.97

1.88+0.43

Saline(3) 0 0

18 S.pneumoniae (4) 6.50±0.28 2.63±1.33

Saline (6) 0 0.33±0.25

Valuesgiven as mean±SE.

*P >0.05comparing4-, 10-, and 18-h CSF bacterial concentrations.

1090 V. J.Quagliarello, WJ.Long,and W M. Scheld a

mean±SE

No ofPV/

/100 nm

capillary

length

b

separated

a

4-mean±SE 3-No. ofPV 0n

capillary length

0 4 10 18

Time(h)postinoculation

b

20-separated

_1E

l

junctions+

0 4 10 18

Time(h) post inoculation

Figure 6.Morphologicchanges in cerebral capillaryendothelium with experimental H.influenzaetype b and H.influenzaeRdmeningitis. (a)PVformationvs.timeafter induction ofmeningitis(b) Percent (%) SJvs.time afterinduction ofmeningitis.(i)H.influenzaetype b; (m) H.influenzae Rd; (o) control; (time 0 o) morphology of cerebral endo-thelium prior to inoculation. P<0.05 compared to controls (see text for actualPvalues). f The numberof SJ/total number ofjunctions visualized in 30 random endothelial cell sections.

points with high bacterial concentrations (mean

log1o

cfu/ml

blood>4.0) (Table IV).

Functional correlation with the observed morphologic changeswasobtainedbyassaying CSF penetration of'251-BSA

TableIII. CharacteristicsofExperimentalH. influenzae

Typebvs. H.influenzae RdMeningitis in Rats

Duration

of CSFbacterial WBC/ml

infection Inoculum concentration of CSF

h n log,0cfu/ml x10

4 H.influenzaetypeb(4) 6.26±0.34 2.20±1.07 H.influenzae Rd (3) 5.25±0.35* 4.02±1.69

Saline (4) 0 0.01±0.01

10 H.influenzaetype b(4) 7.88±0.39 0.63±0.63

Saline (3) 0 0

18 H.influenzaetype b(4) 8.06±0.35 2.56±0.51 H.influenzaeRd(4) 3.32±0.32t§ 7.98±2.46

Saline (3) 0 0.02±0.02

Valuesgivenasmean±SE.

*_{P>0.05compared with4-hCSFH.influenzaetype b}

concentra-tion.

*P<0.01compared with 18-h CSF H. influenzae type b

concentra-tion.

§P <0.05compared with4-hCSFH.influenzaeRdconcentration.

TableIV. Incidence and MagnitudeofBacteremia withExperimental Meningitis intheRat

Duration Proportion Blood

of ofanimals bacterial

Challenge organism infection bacteremic concentration

h % log0 cfu/ml

E.coli Kl+ 4 2/3 (66) 1.25±0.74*

10 1/4(25) 0.43±0.42 18 1/6 (17) 0.71±0.71

S. pneumoniaeIII 4 4/4(100) 2.00±0.25 10 4/4 (100) 1.33±0.12 18 2/4(50) 0.90±0.52

H.influenzaetype b 4 4/4 (100) 4.77±0.08 10 4/4(100) 5.12±0.11 18 4/4(100) 4.53±0.35

H.influenzaeRd 4 0/3(0) 0

18 0/4(0) 0

* Values

given

asmean±SEM.

at 4and 18hpostinoculation. Asshown in TableV,4h post-inoculation with H. influenzaetype b, there was asignificant

increase in

1251-BSA

penetration into CSF ([mean±SE] %

125I-BSApenetration,3.80±0.75) comparedtocontrols(0.26±0.08;

P < 0.01) thatcorrelated with the observed increased PV

for-mation. Wheninfectionprogressed to18htherewas afunctional progression of '25I-BSA permeability ([mean±SEJ % 1251-BSA

penetration, 8.10±1.20) thatwassignificantly greater thanboth

the control group ([mean±SE] % 125I-BSA penetration,

1.25+0.27;

P < 0.01) as well as the 4-h H.

influenzae

type b

infection (P < 0.05). This functional progression ofalbumin

permeabilitycorrelated with themorphologicpresence of both increasedPV formation and thedevelopmentofastatistically

significantincrease in SJ(Table V).

Comparison observations made aftermeningitisinducedby

H. influenzaeRd aredepicted inFig.6 andTables IIIand V. Asshown, afterreceiving virtuallyidentical inocula,therewas no significant difference in CSF bacterial concentration (P

Table V CorrelationofMorphologic and Functional AlterationsoftheBBB in H. influenzaeMeningitis

CSF Changein CSF bacterial penetration of BBB Inoculum concentration '"I-albumin morphology

n log,0 fu/ml %

4h Saline (3) 0 0.26±0.08 H.influenzaeRd(3) 5.25±0.35 4.12±1.25* TPV

H.influenzaetype b(4) 6.26+0.34 3.80±0.75* tPV 18 h Saline (4) 0 1.25+0.27

H.influenzaeRd(4) 3.32±0.32§ 4.64+0.80* T PV

H.influenzaetypeb(4) 8.06±0.35t 8.10±1.20*t TPV+ T SJ

Valuesgivenas

mean±SE.

*_P<0.05 comparedtocontrol.

*P<0.05comparedto H.influenzae typeb at4 handH.influenzaeRd at 18

h.

>0.05), leukocyte count (P > 0.05), cerebral capillary mor-phology, or functional BBB penetration of'25I-BSAwhen com-paredwith H. influenzaetype b 4 h postinfection. That is, there was asignificant increase in PV formation ([mean±SE] number of PV/100-nm capillary length, 5.40±0.81 vs. 2.72±0.35; P

<0.01); but the percentage of SJ (0/46, 0%) was no different than controls (0/46). This correlated with a mean CSF '25I-BSA penetration (4.21±1.26%) that was significantly greater than controls (P<0.05) but no different than the functional pene-tration seen with H. influenzae type b at 4 h (3.80±0.75%).

However,

noneof the animals with H.

_influenzae

Rd

meningitis

were bacteremic at 4 h (Table IV), contrasting with 100% of

thosewith H. influenzae type b meningitis.

When infection was allowed to progress for 18 h with H.

influenzaeRd(Fig. 6), one sees a sustained increase in PV

for-mationwhen compared with controls (P < 0.01) but, in contrast toH. influenzaetype b, the percentage of SJ (2/47, 4%) did not

quantitativelyprogress to reach statistical significance (P > 0.05) compared to controls (1/48, 2%). Similarly, functional penetra-tionof'251-BSAwas notsignificantly different 18 h

postinocu-lation ([mean±SE]% 1251 penetration, 4.64±0.80) than it was 4 hpostinfection(P > 0.05)-see Table V. This morphologic and

physiologic distinction of H. influenzae Rd meningitis (when compared to H. influenzaetype b) 18 hpostinoculation, was

associated withasignificantlylower CSF bacterial concentration

(mean

log1o

cfu/ml, 3.32) when compared with H. influenzae

typeb (mean

log1o

cfu/ml, 8.06;P<0.01)despite identical in-ocula(TableIII). TheCSF WBCconcentrations were not

sig-nificantly different fromH. influenzaetypeb;however, the ab-senceof bacteremia withH. influenzaeRd wasagain distinctly different from the100%incidence ofhigh grade bacteremianoted

withH. influenzae typeb.

Discussion

Inthis investigation, we observed auniform host response to

experimental meningitis at the level ofthe cerebral

capillary

endothelium. Specifically,therewas anearlyandsustained

in-creasein PVformationaswellas a

progressive

increasein

sep-aration of intercellular

junctions

observed

during

the natural

history ofthe infection. Both morphologic changesappearto

contributetofunctional

penetration

ofalbuminacrossthe BBB.

Inthepreantibiotic era,bacterial

meningitis

wasadisease

withdevastatingmortalityclaimingthelives of

virtually

all those

afflicted (1). However,

despite

theadvent ofeffective antimicro-bialagents thereremainsafinitecase

fatality

rate

_(e.g.,

28% in

S.

pneumoniae meningitis)

(13),

with

permanent

neurologic

se-quelae

affecting

manysurvivors

(14-16).

One

potential

expla-nationisthatthe

pathologic

consequencesofthedisease

beyond

the

leptomeninges,

butwithin the centralnervoussystem, prog-ressdespite

bacteriologic

cure

(17).

The

BBB,

the

major

portion

of which is

anatomically

localizedtothe cerebral

capillary

en-dothelium,

represents a critical

extrameningeal

site known to

be

functionally

altered in

meningitis (18-20).

Although

many critical functions of this

unique

endothelium

(e.g.,

abluminal transport of

Na'

andK+, carrier-mediateddiffusion of hexoses

andaminoacids)are

subject

to

injury,

we

sought

to

investigate

the

morphologic

alterations

rendering

itmore

permeable during

infection. Hence, we

quantitatively

assessed

by

transmission electronmicroscopythe

morphologic changes

inPVformation and intercellular junction integrity of the cerebral capillary

en-dothelium using an experimental rat model of meningitis.

Anin vivo model was developed for this investigation because morphologic and physiologic investigations suggest that

per-meability ofthe cerebralmicrovasculatureis responsive to both neural(21, 22)and hormonalinfluence (23). Hence, any

resul-tantchangeinitsmorphology and functional permeability might

depend

on acritical interactionbetween the inciting injury (e.g.,

infection), the inflammatory host response (e.g., PMN

exuda-tion),

and humoral regulatoryprocesses (e.g., adrenal cortical

function, circulating catecholamines). Thus, an in vivo model possesses distinctadvantagesoverinvitro techniques in facili-tatingthisdynamic interplay.The cerebral microvasculature was chosen formorphologicassessmentinthisstudy becauseit rep-resents the dominant site of the BBB. The circumventricular organs (e.g.,choroid plexus, areapostrema) theoretically con-tributetotheBBBdespitefenestrations in their capillary supply because of the presence ofepithelialtight junctions. However, the surface area of the cerebral microvasculature is 5,000-fold greaterthanthesurfaceareaofcapillariessupplying the

circum-ventricular organs (24, 25), rendering the former more pertinent for thisinvestigation.

Qualitatively, ourresults revealed two distinct morphologic changes in the cerebral capillary endothelium resulting from meningeal infection. Namely, there was an increase in PV

for-mation and complete separation of intercellular junctions that allowed for functional penetration of horseradish peroxidaseto

the abluminal side of the endothelium (Figs. 1-3). When these morphologic changes were quantitatively assessed byblinded analysis, a remarkably consistent temporal sequence was ob-served in meningitis induced by the encapsulated pathogens.

For example, after intracisternal inoculation with E. coli Kl +, there was a statistically significant increase in PV formation (P <0.001) as early as 4 h later that was sustained when infection was allowed to progress for 10 h (P<0.02) and 18 h (P<0.01). Similar statistically significant increases in PV formationwere

observedsoonafterinoculation (i.e., 4 h) with the encapsulated S. pneumoniae type III as well as H. influenzae type b; these increases were likewise sustained when infection progressed for 10-18 h (Figs. 4-6). Hence, despite different capsule composition the morphologic alterations of PV formation were similar for allthreeencapsulated pathogens and correlated with sustained CSF bacterial concentrations and WBC pleocytosis (Tables

I-III).

Transendothelial vesicular transport has been postulatedas

apotential mechanism of alteredpermeabilityof theBBB, al-though its significance remains conjectural. As previously noted, thecerebralcapillaryendothelium containsrarePVasopposed tosystemic capillaries (e.g., muscle) in which theyareabundant.

In muscle capillaries, these vesicles have been shownto func-tionally endocytose and exocytosemicroperoxidase.The precise mechanism of transcellular vesicular transport is unknown but hypothesizedtobe eitherbyBrownian motion of vesicles freely mobile in thecytoplasm, orvia transcellular channels forming from transiently fused vesicles across the capillary wall (26). Hence, theuniquepaucity of vesicles within cerebral endothe-lium has been proposedasapotentialcontribution to its function as a barrier to macromolecular transport. Interestingly, other investigations have demonstrated qualitative increases in pi-nocytosis in brain endothelium afterhypertensive (27),ischemic (28), and convulsiveinjury (29),but it remains unclear whether theseare truequantitativeandstatistically significantincreases.

poralsequenceof

morphologic changes,

and theirquantitative

functional significance regarding BBB permeability to various molecularweight proteins.

Morphologically, ourdataregardingincreasedcytoplasmic

vesicular content during experimental meningitiscould be in-terpreted in two ways. First, although luminal endocytosis is

increasedandcytoplasmicvesicularcontentincreased,abluminal

exocytosismay be nonexistent and hence functionaltranscellular transportinsignificant. Indeed, investigationsina mousemodel suggest that under normal conditions pinocytosed macromol-ecules by the cerebral capillary endothelium are destined for lysozomal fusionanddegradation rather than abluminal

exo-cytosis(30).Furthermore, thesetwocritical processes necessary forvesiculartransport(i.e.,endocytosisandexocytosis)may be independently regulated (31); hence, thesameinjurytothe

en-dothelium mayaffectonewithout the other.However,onecould also suggest that there could be greaterluminal

endocytosis

than abluminalexocytosisresulting in agreater

cytoplasmic

vesicle contentmorphologicallyaswellasfunctional transcellular pro-tein transport. This interpretation is supported bythe experi-ments with H.

influenzae

meningitis correlating the observed

BBBmorphologic changes with functional

permeability

to

cir-culating

1251I-BSA.

Asshown(Fig. 6,TableV),4hpostinoculation withH.

influenzae

type b and H.

influenzae

Rd when only cy-toplasmic vesicular content was significantly increased, there was a significant increase in CSF penetration of

'251I-BSA

(P

<0.01) comparedwithcontrols.Similarly, 18 hpostinoculation

with H. influenzaeRd whenincreasedPVformation remained prominent (butjunctional separation still insignificant) there was a similar quantitative CSF penetration of 125I-BSA

([mean±SE] 4.64±0.80%) that remainedsignificant(P<0.05)

comparedwithcontrols. Hence themorphologicappearanceof increased PV formation appears to correlate with functional permeabilityoftheBBBtocirculating albuminsuggestingthat

pinocytosis maybe an independent mechanism of

macromo-lecular transport in meningitis.

The temporal sequenceof intercellularjunctionseparation

with experimental meningitis was distinctly different than that for PVformation, butwaslikewise reproducible regardless of

the encapsulated pathogen tested. That is, a longer infection

duration (10 h)of experimental meningitiswasrequired foreach encapsulatedstrain beforethe percentage of separatedjunctions

reachedstatistical significance when compared to saline-matched, controls. Furthermore, the percentage of separated junctions progressivelyincreased with longer infection durations reaching

avalueof17-19%at 18 hpostinoculation (Figs. 4-6).

Interest-ingly,however, is that despite the animals being clinically

mor-ibund 18 h after inoculation, the majority of the intercellular

junctions visualizedremained closed (i.e., <20% were separated). One explanation for this observation is that each junctional complex may vary in its degree of membrane fusion between

adjacent endothelial cells. Hence, some junctions may be in-herently "less tight" than others and thus more susceptible to separation upon infectious injury. This hypothesis is supported by freeze-fracture studies of epithelial tissue in which the TJ region was found to consist of a complex network of anasta-mosing linear strands appearing as ridges or grooves in the com-plementary fracture faces (32). The number and continuity of the anastomosing strands appeared to correlate with the

tran-sepithelialresistance (i.e., "tightness") of the tissue. For example, mammalian

proximal

convoluted renal tubule

_representing

a weakbarrier with a low electrical resistance

₍₆

_ohm-cm2),

had

few ordiscontinuous anastomosing strands. In contrast, toad bladder epithelium, inherently tight physiologically (transepi-thelial resistance of 1,000-2,000 ohm-cm2) hadnumerous

an-astamosing strands thatwerecontinuous.Hence,onecould

pos-tulate thatwithin the same capillary tissue preparation, although theresistance of junctional complexes are comparatively high

(5), there may bevariability among different capillaries as well as within the same capillary. Thus, given the same degree of injury, somejunctions may completely separate while others may remainpartiallyorcompletely intact.

Asecond possibility is that junctional separation may require acomplexinjury pattern and demand both luminal and antilu-minal stress for completion. Hence, junctions found to separate in ourstudies may require notonly antiluminal injury from bacterialsurface ligands or extracellular toxins resultant from the meningitis. Simultaneous luminal injury from a sustained bacteremiaoranoxic injury (perhaps from localized vasospasm) maybe necessarycontributing factors, and may bemoresevere in focal areasof the cerebral vasculature.

Supporting the hypothesis that junctional separation might require a complex injury pattern are our studies comparing

morphologic andfunctional BBB alterations in H. influenzae type bvs.H. influenzae Rdmeningitis (Tables IV and V; Fig.

6). The Rd strainwasselected forcomparison because it

elab-orates nodetectable polyribosyl phosphate (PRP) capsule either byenzyme-linked immunosorbent assay (ELISA)or immuno-fluorescent techniques. Other spontaneous capsule-deficient mutantsexist (e.g., S2strain; Sec 1 strain)but both elaboratea

finite, albeit small amount of PRP (4-45 ng/109 cells) (33). Hence, to assess whether BBBinjurycouldoccurdespiteabsence of detectablecapsule, the Rd strain wasused for comparison. Ideally, onewoulddesire a strain for comparison that possessed identical subcapsularsurfacecomponents(e.g.,outermembrane

protein,lipopolysaccharide) totheencapsulatedstrain and dif-fered only in its lack of encapsulation. Unfortunately, even

ge-netically transformed strains(e.g.,transformation of Rd strain

with H. influenzae type b DNA) possess heterogeneous lipo-polysaccharidepatternsonsodiumdodecyl

sulfate-polyacrilam-ide gel electrophoresis thatdiffer from the recipient (e.g., Rd)

strain (34), precludingsuchacomparisonatpresent.

Aspreviously noted, whenmeningitiswasinduced withthe Rdstrain, identical morphologic changes ofthe cerebralcapillary

endotheliumwerenoted(Fig. 6) when comparedtoH. influenzae type b 4 hafter inoculation.Namely, there wasanincreasedPV

formationbutnosignificant junctional separation.Thesechanges

correlated withsignificantCSFpenetration of

'25I-albumin

(Ta-ble V) andwerepresentdespiteabsenceof detectable bacteremia

withthe Rdstrain compared with the 100% incidence and high-gradebacteremia (mean±SE log10cfu/ml, 4.77±0.08) seen 4 h

postinoculation withH.

influenzae

typeb(Table IV). These data suggest thatincreased vesicular transport as a host response to H. influenzaeinfection does not require the presence of either

PRPcapsuleor bacteremia.

When meningitis was allowed to progress for 18 h with the

Rdstrain, the morphologic changes of the cerebral capillary en-dothelium remained the same (i.e., an increased PV formation but only4%separatedjunctions; P > 0.05) and functional BBB

permeabilitywas unchanged (4.64% '25I-BSApenetration) com-paredtothe4-hinfection (4.12%

'25I-BSA

penetration; P> 0.05). Again there was no detectable bacteremia and the CSF bacterial

unen-capsulated

strain. However, 18hafterinoculation with the

en-capsulated

H.

influenzae

type b therewasasignificantincrease

in

separated junctions

(8/52, 15%) aswell as sustained PV

for-mation

(Fig. 6);

aphenomenon previously noted in the encap-sulated E.

coli

Kl +and S.

pneumoniae type

III

meningitis

ex-periments.

This

development

ofa

significant

increase in sepa-rated

junctions

18 h

postinoculation

was associated with a

progression

of BBB

permeability

to

125I-BSA

([mean±SE]

8.1±

_1.2%)

thatwas

significantly

greater than the permeability

change

induced

by

H.

influenzae Rd at 18 h (P

<

0.05) as well

as H.

_influenzae

type

b at 4 h

(P

<

0.05)-see Table V.

This

morphologic progression (i.e.,

significant junctionalseparation) andincreased functionalBBB

permeability

wereassociated with

a

_significant

increasein the CSFbacterial concentration

(com-pared

with the 4-h CSF

concentration)

aswellas the presence ofa

high-grade

bacteremia

(mean±SE

log1o

cfu,4.53±0.35) in

100% oftheanimals. Theseobservationssuggestarole for the PRP

capsule

eitheras an

independent

effector ofTJinjuryor a

factor

facilitating high

CSFandblood bacterial concentrations

necessary for

junctional

separation.

Different surface

compo-nents

(e.g.,

capsule, lipopolysaccharide,

outer membrane

pro-teins)

maycontribute

independently

or

synergistically

tothe

en-dothelial cell

injury

observed.

In summary,

experimental

meningitis

in this model appears

toelicit uniform

morphologic

alterations oftheBBBwitha

re-producible temporal

sequence. PVformationappearstobean

early

host responsetoinfectionin thesubarachnoidspace,

fol-lowed

by

a progressiveincreasein

complete separation

of

inter-cellular

junctions

astheinfectionprogresses. Both

morphologic

hostresponses appeartocontributetofunctional alterationsin BBB

_permeability

to

circulating

macromolecules.

Encapsulation

as aneurovirulence factor

for

H.

influenzae

does not appear

essential for

morphologic

or functional BBB

injury

but may

facilitate their

_{progression by allowing}

the

organism

to evade hostclearance.Future

investigation

with thismodel will allow

elucidationof the

precise

role of various bacterial surface

com-ponents, the

inflammatory

host response, andintraluminal

fac-tors

_(e.g.,

local

_hypoxia,

hormonal

influences)

in

precipitating

this

_injury.

Acknowledgments

The authors thank Dr.A.Lorris Betz (University of Michigan) for in-valuable assistance with techniques for isolation ofcerebral microvessels andDr.JohnPovlishok (Medical College of Virginia) for review ofthe

photoelectronmicrographs.Joyce Henderson prepared the manuscript. This studywassupportedinpartbyatraininggrant(T32A107046) from the National Institutes of Health. Dr.Scheld is the recipient ofa ClinicalInvestigatorAward(K08-00517) from the National Institute of Allergy andInfectious Diseases andaresearchgrantfromthe Thomas Jeffress and Kate Miller JeffressTrust.

References

1. Swartz,M. N. 1984. Bacterial meningitis-more involvedthan

justthemeninges.N.Engl.J.Med.311:912-914.

2. Reese, T.S.,and M. J.Karnovsky. 1967.Fine structural localization ofabloodbrain barriertoexogenousperoxidase.J. Cell Biol. 34:207-217.

3. Brightman, M. W., and T. S.Reese. 1969. Junctionsbetween intimately opposedcellmembranes inthevertebrate brain.J.Cell Biol.

40:648-677.

4. Bradbury, M. W. 1984. The structure and function of the blood brain barrier. Fed. Proc. 43:186-190.

5. Crone, C., and S. P. Oleson. 1982. Electrical resistance ofbrain

microvascular endothelium. Brain Res. 241:49-55.

6. Oldendorf, W. H., and W. J. Brown. 1975. Greater number of capillary endothelial mitochondria in brain than in muscle. Proc. Soc. Exp.Biol.Med. 149:736-738.

7. Betz, A. L.,J. A. Firth, and G. W. Goldstein. 1980. Polarity of the blood brain barrier: distribution of enzymes between the luminal andantiluminal membranes of brain capillary endothelialcells. Brain Res. 192:17-28.

8. Betz, A. L., J. Csejtey, and G. W. Goldstein. 1979. Hexosetransport and phosphorylation by capillaries isolatedfrom rat brain. Am.J. Physiol. 236:C96-C102.

9. Betz, A. L., and G. W. Goldstein. 1978. Polarity of the blood-brain barrier: neutral amino acid transport into isolated blood-brain capillaries. Science(Wash. DC).202:225-227.

10. Goldstein, G. W., and A. L. Betz. 1983. Recent advances in understandingbrain capillary function. Ann.Neurol. 14:389-395.

11.Scheld, W. M., and M. A. Sande. 1983. Bactericidal versus bac-teriostatic antibiotic therapy of experimental pneumococcal meningitis in rabbits. J. Clin. Invest. 71:411-419.

12.Graham, R. C., and M. J. Karnovsky. 1966. The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mousekidney: ultrastructural cytochemistry by a new technique. J. His-tochem. CyHis-tochem. 14:291-302.

13. Centers for Disease Control. 1979. Bacterial meningitis and me-ningococcemia-United States, 1978. Morbid. Mortal. Week. Rep. 28: 277-279.

14.Alvon, V., V. Naveh, M. Gardos, and A. Greenman. 1971. Neu-rological sequelae of septic meningitis. Isr. J. Med. Sci. 15:512-517.

15.Ferry, P. C., J. L. Culbertson, J. A. Cooper, A. B. Silton, and S. H. W. Sell. 1982. Sequelae of Haemophilus influenzae meningitis: preliminary report of a long term follow-up study. In Haemophilus

in-fluenzae:Epidemiology, Immunology, and Prevention of Disease. S. H. Sell, and P. F. Wright, editors. Elsevier Biomedical, New York.

111-117.

16.Dodge, P. R., H. Davis, R. D. Feigen, S. J. Holmes, S. L. Kaplan, D. P.Jubelirer, B. W. Stechenberg, and S. K. Hirsh. 1984. Prospective evaluation of hearing impairment asasequelae of acutebacterial men-ingitis. N.Engl. J. Med. 311:869-874.

17.Scheld, W. M., R. G. Dacey, H. R. Winn, J. E. Welsh, J. A. Jane, and M. A. Sande. 1980. Cerebrospinal fluid outflow resistance in rabbits withexperimental meningitis: alterations with penicillin and methyl-prednisolone. J. Clin. Invest. 66:243-253.

18.Cooper,A.J.,H. M.Beaty, S. I. Oppenheimer, C. J.Goodner, andR.G.Petersdorf. 1968. Studiesonthepathogenesis ofmeningitis. VII. Glucose transport and spinal fluid production in experimental pneumococcal meningitis.J.Lab. Clin. Med. 71:473-481.

19. Spector, R., and A. V. Lorenzo. 1974. Inhibition ofpenicillin

transport from the cerebrospinal fluid after intracisternal inoculation of bacteria. J. Clin.Invest. 54:316-325.

20.Prockop, L. D., and R. A. Fishman. 1968. Experimental pneu-mococcal meningitis: permeability influencing the concentration of sugars andmacromolecules in cerebrospinal fluid. Arch. Neurol. 19:449-463.

21. Estrada, C., E. Hamel, and D. N. Krause. 1983. Biochemical evidence for cholinergic innervation of intracerebral blood vessels. Brain Res.266:261-270.

22. Herbst, T. J.,M. E.Raichle,andJ. A.Ferrendelli. 1979. ,B-Ad-renergicregulation of adenosine3',5'-monophosphateconcentration in brain microvessels.Science(Wash. DC).204:330-332.

23. Long, J. B.,and J. W.Holaday. 1985.Blood-brain barrier: en-dogenous modulation by adenal-cortical function. Science(Wash.DC).

227:1580-1583.

24. Pardridge, W. M. 1983. Neuropeptides and the blood-brain bar-rier.Annu. Rev.

Physiol.

45:73-82.

25. Crone, C. 1971. The blood-brain barrier-facts and questions. In Ion Homeostasis oftheBrain.B. K.Siesjo,and S.C. Sorenson, editors. Munksgaard, Copenhagen: 52-62.

26.Simionescu, N., M. Siminoescu, and G. E. Palade. 1978. Structural basis ofpermeabilityinsequential segments of the microvasculature of thediaphragm. II. Pathways followed by microperoxidase across the en-dothelium. Microvasc. Res. 15:17-36.

27. Westergaard, E. 1977. The blood brain barrier to horseradish peroxidase under normal and experimental conditions. Acta Neuropathol. 39:181-187.

28.Petito, C. K. 1979. Early and late mechanisms of increased vas-cularpermeability following experimental cerebral infarction. J. Neu-ropathol. Exp. Neurol. 38:222-234.

29.Petito, C. K., J. A. Schaefer, and F. Plum. 1977. Ultrastructural characteristics of the brain and blood-brain barrier in experimental sei-zures.BrainRes. 127:251-267.

30.Broadwell,R.D.,andM.Salcman.1981.Expandingthedefinition ofthe blood-brain barriertoprotein. Proc. Nail. Acad. Sci. USA. 78: 7820-7824.

31.Williams,S.K.,andR.C.Wagner. 1981.Regulationof micro-pinocytosisincapillaryendothelium by multivalent cations.Microvasc. Res. 21:175-182.

32.Claude, P., andD. A.Goodenough. 1973.Fracturefacesof

zonu-lae occludentes from"tight"and"leaky"epithelia.J.Cell Bio. 58:390-400.

33. Zwahlen, A.,J. A.Winkelstein,andE. R. Moxon. 1983. Partic-ipation of complement in host defense against capsule deficient

Hae-mophilus influenzae. Infect.Immun.42:708-715.

34. Zwahlen, A., J.A.Winkelstein, and E. R. Moxon. 1983. Surface determinants of Haemophilus _influenzae pathogenicity: comparative

virulence of capsular transformants innormalandcomplement-depleted