brain-derived neurotrophic factor in astrocytes, oligodendrocytes, and microglia/macrophages after...

d

Neurobiology of Disease 7, 574–585 (2000)

doi:10.1006/nbdi.2000.0318, available online at http://www.idealibrary.com on

Brain-Derived Neurotrophic Factor in Astrocytes,Oligodendrocytes, and Microglia/Macrophagesafter Spinal Cord Injury

Karen D. Dougherty, Cheryl F. Dreyfus, and Ira B. Black1

Department of Neuroscience and Cell Biology, UMDNJ-Robert Wood Johnson MedicalSchool, Piscataway, New Jersey 08854

Received December 9, 1999; revised June 27, 2000; accepted June 27, 2000

Recent studies suggest that the injured adult spinal cord responds to brain-derived neurotrophicfactor (BDNF) and neurotrophin-3 (NT3) with enhanced neuron survival and axon regeneration.Potential neurotrophin sources and cellular localization in spinal cord are largely undefined. Weexamined glial BDNF localization in normal cord and its temporospatial distribution after injury invivo. We used dual immunolabeling for BDNF and glial fibrillary acidic protein (GFAP) in astrocytes,adenomatous polyposis coli tumor suppressor protein (APC) for oligodendrocytes or type III CDHreceptor (OX42) for microglia/macrophages. In normal cord, small subsets of astrocytes andmicroglia/macrophages and most oligodendrocytes exhibited BDNF-immunoreactivity. Followinginjury, the number of BDNF-immunopositive astrocytes and microglia/macrophages increaseddramatically at the injury site over time. Most oligodendrocytes contained BDNF 1 day and 1 weekfollowing injury, but APC-positive cells were largely absent at the injury site 6 weeks postinjury.Glial BDNF-immunolabeling was also examined 10 and 20 mm from the wound. Ten millimetersfrom the lesion, astrocyte and microglia/macrophage BDNF-immunolabeling resembled that atthe injury at all times examined. Twenty millimeters from injury, BDNF localization in all three glialsubtypes resembled controls, regardless of time postlesion. Our findings suggest that in normaladult cord, astrocytes, oligodendrocytes, and microglia/macrophages play roles in local trophinavailability and in trophin-mediated injury and healing responses directly within and surroundingthe wound site. © 2000 Academic Press

Key Words: neurotrophins; GFAP; OX42; APC; astrocytosis.

INTRODUCTION

Functional deficits that follow spinal cord or brainlesions are largely due to failure of damaged axonsand their parent neuronal perikarya to exhibit strongregenerative responses. However, recent investiga-tions indicate that treatment of injured CNS tissuewith brain-derived neurotrophic factor (BDNF) andneurotrophin-3 (NT3) may foster postinjury functionalrecovery by enhancing neuron survival and axon re-generation. This raises a number of questions, sincethese trophins are present in normal brain and spinal

1 To whom correspondence and reprint requests should be ad-ressed. Fax: (732) 235-4990. E-mail: [email protected].

574

cord. Consequently, temporospatial localization oftrophins in the normal and injured central nervoussystem (CNS) has been the focus of many recent in-vestigations.

In the normal adult spinal cord and brain, neuro-nal and nonneuronal cells produce neurotrophinsand/or exhibit receptor-mediated responses to neu-rotrophins, including brain-derived neurotrophicfactor (BDNF), nerve growth factor (NGF), and neu-rotrophin-3 (NT3) (Schnell et al., 1994; Nakahara etal., 1996; Koliatsos et al., 1993; Cho et al., 1998);the trophins may potentially act in an autocrine orparacrine fashion. We and others have shown thatin brain, oligodendrocytes and microglia/macro-phages express NGF, BDNF, and NT3 in vitro and in

0969-9961/00 $35.00Copyright © 2000 by Academic Press

All rights of reproduction in any form reserved.

b

575Increased Glial BDNF after Spinal Injury

vivo (Byravan et al., 1994; Dai et al., 1997; Elkabes etal., 1996). In the adult cord, cellular neurotrophinsources are still being defined. Previous investiga-tions in our laboratory have shown that BDNF islocated within a subset of astrocytes in normal adultspinal cord in vivo (Dreyfus et al., 1999).

After spinal injury (SCI), BDNF and NT3, deliveredy injection (Schnell et al., 1994), gelfoam (Ye & Houle,

1997), transformed fibroblasts (Nakahara et al., (1996);Tuszynski et al., 1996) or Schwann cells, alone or com-bined with embryonic spinal cord (Bregman et al.,1998; Bregman et al., 1997; Tuszynski et al., 1998) re-duce atrophy of adult neurons projecting to the cordand promote axonal regeneration. Furthermore, in-jured adult CNS axons can grow for long distanceswithin peripheral nerve grafts containing an endoge-nous supply of trophins (Aguayo et al., 1991). Thesefindings demonstrate that regenerative failure of adultneurons and axons is not due to intrinsic limitationssuch as unresponsiveness to trophins.

Previous reports indicate that increased expressionof trophin mRNA occurs at sites of both cell body andaxon terminal degeneration in the adult rat CNS (Gosset al., 1998; Nieto-Sampedro et al., 1999; Arendt et al.,1995); these reactions may stimulate repair and hadbeen assumed to occur within neurons. However, sev-eral recent studies indicate that trophin increases oc-cur largely within reactive astrocytes, a basic featureof the CNS injury response (Goss et al., 1998; Nieto-Sampedro et al., 1999; Arendt et al., 1995; Krenz &Weaver, 2000). These studies clearly demonstrate in-creased glial trophins after brain injury. Taken to-gether with our findings and those of others showingthat astrocytes and microglia/macrophages in normaladult spinal cord and brain contain trophins, thesereports suggest that trophins increase in spinal gliaafter injury.

In the present work, we examined BDNF-immuno-reactivity in different glial subtypes (in normal adultspinal cord) and at varying times after injury. Wecombined immunocytochemical labeling for BDNFwith that for glial fibrillary acidic protein (GFAP) tolabel astrocytes, adenomatous polyposis coli tumorsuppressor protein (APC) to label oligodendrocytes orOX42 to label microglia/macrophages. Our work sug-gests that in the normal adult spinal cord, small pro-portions of astrocytes and microglia/macrophagesand many oligodendrocytes contain BDNF. Followinginjury, BDNF contents change in a cell-specific fash-ion.

MATERIALS AND METHODS

Subjects and Surgical Procedures

A total of 15 adult female Sprague–Dawley rats,weighing 250–275 g at the beginning of the experi-ment were used for these experiments (Harlan, Indi-anapolis, IN). All procedures involving experimentalanimals were performed at Ohio State University inthe laboratory of B. T. Stokes and conducted accordingto guidelines of the Ohio State University InstitutionalAnimal Care and Use Committee. Every effort wasmade to minimize animal suffering and to use a min-imum number of rats. Rats were anesthetized with amixture of ketamine and xylazine. A dorsal midlineincision was made over the midthoracic spinal cordand the T8 and T9 vertebrae were exposed. For thoserats in the control, laminectomy group (n 5 6), apartial laminectomy was performed at T8. For thoseanimals in the injury group, the spinal cord at T8–T9was exposed and compressed a distance of 1.1 mmfrom the 0-mm dural position using the Ohio StateUniversity device (For detailed description see Stokeset al., 1992). Following surgery, wounds were closed,topical anesthetic (Flurizolidone) was applied to thewound area, and the animals were returned to theircages. During the postoperative recovery period, allsubjects were given Gentocin (1 mg/kg). For injuredsubjects, bladders were manually expressed twicedaily. Food and water were available ad libitum and ratchow diet was supplemented with vitamin C tabletsand bread cubes. Subjects were allowed to survive for1 day (n 5 5), 1 week (n 5 5), or 6 weeks (n 5 5)following surgery. For each of these time points, agroup of two laminectomized and three injured ratswas prepared and handled identically until sacrifice.

Section Preparation

Rats were deeply anesthetized with sodium pento-barbital (Sigma Chemical Co., St. Louis, MO) and per-fused through the heart with phosphate buffered sa-line (PBS, pH 7.4) followed by 4% paraformaldehydein phosphate-buffered saline (PBS, pH 7.4). Spinalcords were removed and postfixed in the perfusate for1 h at 4°C. They were then placed into PBS and main-tained at 4°C for overnight transport. Upon arrival inour laboratory, cords were removed from PBS and the2 cm of tissue directly rostral to the point of laminec-tomy or injury in the thoracic region were taken fromthe cord. These tissue samples were cut into two 1-cmtissue blocks rostral to the lesion (or laminectomy): (1)

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

am(ItamCms1cnps

ci

D

ms10Po0(tAX1rgccotet

I

amdfdadm

576 Dougherty, Dreyfus, and Black

A

0–10 mm; (2) 11–20 mm from the lesion. The lesionwas included in the first block at 0 mm. All locationsrefer to distances from the lesion; designated as 0 mm.Blocks were immersed at 4°C in a graded series ofsucrose solutions (15 and 30% in PBS) for cryoprotec-tion. These were rapidly embedded and frozen in asingle piece of embedding compound. Sagittal sec-tions (15 mm) were cut using a Leitz cryostat andmounted onto gelatin (1%)-coated slides. All tissuesections were stored at 220°C until immunolabelingwas conducted.

Antibodies

A polyclonal antiserum raised in rabbit against pep-tides derived from human BDNF (a gift from D. R.Kaplan, McGill University) was used for immunoflu-orescent and immunoperoxidase labeling. Specificityof this antiserum has been characterized extensively(Friedman et al., 1998). In the present study, specificityof this anti-BDNF antiserum was further examined intissue sections from two laminectomized rats and oneinjured rat from each of the experimental injurygroups (n 5 5). These tissue sections were immunola-beled as described below, except that either (1) theprimary antiserum was preadsorbed by incubationwith full-length BDNF (4 mg of BDNF/ml of primaryantiserum at a dilution of 1:16,000) and visualizedusing 3,39 diaminobenzidine (DAB) as a chromagen or(2) the primary antiserum was omitted from the firstincubation. Further, for comparison, a second poly-clonal antiserum for BDNF raised in chicken (Pro-mega, Madison, WI) was also used to label tissuesections from these rats using the immunoperoxidaseprocedure (1:2000 dilution). This antiserum, which israised in chicken against full-length human BDNF,has been previously tested for specificity by Sayers etl. (1998). For immunolabeling of GFAP and OX42,onoclonal antibodies raised in mice were employed

obtained from Boehringer-Mannheim, Indianapolis,N, and Serotec, UK, respectively). The specificity ofhese antibodies has been previously validated (Elk-bes et al., 1996; Dreyfus et al., 1999). The anti-APConoclonal antibody (Oncogene Research Products,ambridge, MA), which is directed against the N ter-inus of the protein, has previously been shown to

electively label adult oligodendrocytes (Bhat et al.,996). To verify secondary antibody specificity in theurrent investigation, tissue sections from one lami-ectomized control and one injured rat in each of theostinjury groups underwent immunolabeling as de-cribed, except that the relevant polyclonal or mono-

Copyright © 2000 by Academic Pressll rights of reproduction in any form reserved.

lonal primary antibody was omitted from the initialncubation solution.

ual-Immunofluorescent Labeling

Tissue sections were thawed and air-dried for 15in before being placed into PBS. Sections were then

equentially incubated in: (a) a solution containing0% normal horse serum, 10% normal goat serum,.1% Triton X-100, and 0.5% bovine serum albumin inBS (1 h), (b) an empirically derived optimal dilutionf the primary antiserum for BDNF (1:500) containing.1% Triton X-100 and 1% normal horse serum in PBS48 h at 4°C), (c) an empirically derived optimal dilu-ion of the primary antibody against GFAP (1:1000),PC (1:500), or OX42 (1:200), containing 0.25% Triton-100 and 1% normal goat serum (24 h, 4°C), (d) a:100 dilution of fluorescein-conjugated (FITC) anti-abbit IgG (2 h, in dark) (Vector Laboratories, Burlin-ame, CA), and (e) a 1:100 dilution of Texas red-onjugated anti-mouse IgG (Vector) (in dark). All in-ubations were conducted at room temperature unlesstherwise stated and spaced with washes in PBS. Sec-ions were then briefly rinsed in deionized water, cov-rslipped with Vectashield-mounting medium (Vec-or), and stored at 220°C.

mmunoperoxidase Labeling of BDNF

Sections for immunoperoxidase labeling of a singlentigen were sequentially incubated in: (a) the pri-ary antiserum at an empirically derived optimal

ilution (1:16,000 for the Kaplan antiserum and 1:2,000or the Promega antiserum) for 24 h at 4°C, (b) 1:500ilution of either anti-rabbit or anti-mouse biotinyl-ted IgG (Vector) for 1 h, (c) 1:100 dilution of peroxi-ase–avidin complex (Vectastain Elite kit, Vector) 30in, (d) 3,39-diaminodenzidine and H2O2 in PBS.

Data Analysis

Sections processed using the anti-rabbit polyclonalantiserum (Kaplan) paired with monoclonal antibod-ies against glial markers were analyzed thoroughly.Sections in which morphological preservation was op-timal and dual immunolabeling was robust were se-lected for further examination. Portions of tissue at theinjury site and 20 mm rostral to the injury site wereexamined using a Leitz Aristoplan microscopeequipped with bright-field and fluorescence optics.Fluorescence was revealed with an I3 filter cube forFITC, a TX filter cube for Texas red, or a G/R cube that


permitted visualization of both fluorochromes. Finalfigures were generated from 35-mm slides or photo-graphic prints captured with either a Kodak R2035film scanner or a UMAX flatbed scanner, respectively,attached to a Macintosh Power Mac G3 9500 com-puter. Composites were assembled using Adobe Pho-toShop 5.0 and printed on a Fujix Pictography 3000printer.

RESULTS

BDNF Immunoperoxidase Labeling andAntiserum Specificity

Patterns of immunolabeling yielded by the rabbit orchicken polyclonal anti-BDNF antisera were indistin-guishable in normal and injured cord.

To document specificity of the anti-rabbit polyclonalantiserum, we examined sections of the thoracic cord1 week after injury using immunoperoxidase (Fig. 1).In a representative section, 10 mm rostral to thewound site, many star-shaped cells in the gray matterexhibited intense BDNF-immunoreactivity and posi-tive processes extended into the dorsal white matter(Fig. 1A). In contrast, sections exposed to primaryantiserum preadsorbed with full length BDNF exhib-ited no detectable immunoreactivity (Fig. 1B). Second-ary antibodies were all specific, since sections incu-bated in the absence of primary antisera contained nolabeled profiles, using both immunoperoxidase andimmunofluorescence (not shown).

BDNF-Immunopositive Astrocytes IncreaseFollowing Injury

In agreement with our previous findings in normaladult spinal cords (Dreyfus et al., 1999), laminecto-mized control rat cords contained few but detectableGFAP-positive astrocytes with BDNF-immunolabel-ing (Figs. 2A and 2B). After injury, both astrocytosisand increased colocalization of BDNF and GFAP wereapparent. At the wound site, the number of BDNF-IRastrocytes increased over time following injury (Figs.2A–2F, Table 1). This increase extended 10 mm rostralto the injury, but was not evident at 20 mm (Figs.2G–2J, Table 1). At 6 weeks postinjury, BDNF-IR as-trocytes exhibited enlarged somata and elaborate pro-cesses, typical of “reactive” astrocytes (Figs. 2E and2F). These cells comprised a large proportion of theglial scar surrounding the wound site. This pattern ofBDNF localization within astrocytes closely resembles

the pattern of astrocytic trophin upregulation reportedin other regions of the CNS following injury (Arendt etal., 1995; Goss et al., 1998; Funakoshi et al., 1993).

Oligodendrocytes Contain BDNF in Both Controland Injured Cord

In laminectomized controls, the majority of APC-IRoligodendrocytes exhibited prominent BDNF-immu-nolabeling (Figs. 3A and 3B). The cells were charac-teristically round or ovoid and appeared to be bipolar,conforming to previously described spinal oligoden-drocytes (Shuman et al., 1997; Bjartmar, 1998). One dayand 1 week following spinal cord injury, most APC-positive oligodendrocytes at the wound site exhibitedrobust BDNF immunolabeling similar to that observedin laminectomized controls. However, 1 week follow-ing injury, these cells exhibited gross morphologicaldisruption with loss of characteristic radial and axiallinear orientation (Figs. 3C and 3D). Six weeks afterinjury, only rare APC-labeled profiles were evident atthe wound site, either in the cavity or in the sparedrim of axons, while many BDNF-IR profiles werepresent (Figs. 3E and 3F). In contrast, at all timesexamined, many BDNF-IR oligodendrocytes were ob-served 10 and 20 mm rostral to the injury (Figs. 3G and3H, Table 1).

Microglia/Macrophages Containing BDNF-Immunolabeling Increase after Injury

In laminectomized controls, a minor subset ofOX42-IR microglia/macrophages also containedBDNF immunolabeling (Figs. 4A and 4B). OX42-pos-itive cells in laminectomized rats exhibited very smallcell bodies and fine, ramified processes. At the woundsite, the number of BDNF-IR microglia/macrophagesincreased over time following injury (Figs. 4C–4F,Table 1). This increase extended 10 mm rostral to thewound but was not evident 20 mm rostral to thewound (Figs. 4G–4J, Table 1). At the injury site, mi-croglia/macrophages were either round or ovoid withlarge nuclei and a small rim of cytoplasm, whichappeared to fill the wound cavity; others exhibitedenlarged somata and short fibrous processes withinthe rim of tissue surrounding the syrinx. (Figs. 4E and4F).

DISCUSSION

Our observations indicate that small subsets ofBDNF-IR astrocytes and microglia/macrophages are


FIG

.1.

Poly

clon

alan

tise

rum

spec

ifica

llyla

bels

cellu

lar

popu

lati

ons

inth

eth

orac

icsp

inal

cord

1w

eek

afte

rin

jury

,us

ing

imm

unop

erox

idas

e.R

epre

sent

ativ

ese

ctio

nsar

epr

esen

ted

inth

isan

dal

lsub

sequ

entfi

gure

s.O

new

eek

afte

rin

jury

,num

erou

sin

tens

ely

labe

led

BD

NF-

IRpr

ofile

sar

epr

esen

t.(A

)T

enm

illim

eter

sro

stra

lto

the

inju

rysi

te,

man

yst

ar-s

hape

dce

llsin

the

gray

mat

ter

(arr

ows)

cont

ain

BD

NF-

imm

unor

eact

ivit

y.Pr

oces

ses

are

also

pres

ent

that

exte

ndin

tod

orsa

lw

hite

mat

ter

(arr

owhe

ad).

(B)

No

BD

NF-

IRpr

ofile

sar

ese

en10

mm

from

the

inju

rysi

tew

hen

tiss

ueis

incu

bate

dw

ith

prim

ary

anti

seru

mpr

ead

sorb

edw

ith

full-

leng

thB

DN

F.B

ar,1

00m

m.


Copyright © 2000 by Academic PressAll rights of reproduction in any form res
erved.

mtwap


evident in the normal adult spinal cord. One dayfollowing injury, increased numbers of astrocytes andmicroglia/macrophages became BDNF-immunoposi-tive. One and 6 weeks postinjury, BDNF-IR astrocytes

FIG. 2. Following injury, an increased proportion of GFAP-positiized rats, few GFAP-IR profiles also contain BDNF-immunolabe

hrough the white matter. (E and F) Six weeks after injury, many lareek after injury, many astrocytes also contain BDNF-immunolabe

way from the wound, the proportion of dual-labeled astrocytes is srofile; arrow, BDNF/GFAP-IR profile. Bar, 100 mm.

and microglia/macrophages comprised the major pro-portion of these cell populations 0 to 10 mm rostral tothe injury site. In contrast, the majority of oligoden-drocytes contained BDNF-immunolabeling in both

ocytes contains BDNF-immunoreactivity. (A and B) In laminecto-C and D) One day after injury, BDNF-IR astrocytic fibers extendocytes near the wound are BDNF-immunoreactive. (G and H) Onemm from the wound. (I and J) Six weeks after injury and 20 mm

d resembles that in controls. Green, BDNF; Red, GFAP; *, GFAP-IR

ve astrling. (ge astrling 10mall an


nedtpnaspisiBs

D1

Sa

oprcpltlS


A

normal and injured spinal cord up to 1 week postin-jury. By 6 weeks, the injury-site was largely devoid ofall APC-labeled oligodendrocytes, while 10 mm fromthe injury, many oligodendrocytes were BDNF-immu-nopositive. Twenty millimeters from the injury site,localization of BDNF-immunoreactivity in each glialsubtype resembled that in controls at all postinjurytimes examined.

Methodologic Considerations

The specificity of our antiserum was previouslycharacterized by Friedman et al. (1998) using immu-

oblot and preadsorption techniques. In the presentxperiments, incubation of tissue with two distinctlyifferent primary antisera (one raised against a pep-

ide sequence, the other raised against the full-lengthrotein) yielded virtually identical patterns of immu-olabeling. Further, omission of primary antiserumbolished all labeling detectable at the light micro-copic level. Finally, preadsorption of the anti-rabbitrimary antiserum with full-length BDNF yielded no

mmunolabeled profiles in laminectomized control tis-ue (not shown). Thus, the patterns of immunolabel-ng observed accurately reflected the distribution ofDNF. Among injured rats, incubation with pread-orbed antiserum also completely blocked BDNF-im-

TABLE 1

Proportion of Glial Subpopulations Containing BDNF-Immunolabeling at Varying Distances from Wound

Glial marker Treatment/survival time

Distance from wound(mm)

0 mm 10 mm 20 mm

GFAP LAM 1 1 11 day 11 11 11 week 111 111 16 weeks 111 111 1

APC LAM 111 111 1111 day 111 111 1111 week 11 111 1116 weeks 2 111 111

OX42 LAM 1 1 11 day 11 11 11 week 111 111 16 weeks 111 111 1

Note. 1, small proportion of dual-labeled cells; 11, moderateproportion of dual labeled cells; 111, major proportion of dual-labeled cells; 2, no detectable APC-labeled cells present; LAM,laminectomized control subjects sacrificed at all time points aftersurgery.


munolabeling at 10 and 20 mm from the injury site.Previous investigations indicate that trophin produc-tion increases following CNS injury (Funakoshi et al.,1993; Goss et al., 1998; Arendt et al., 1995). Moreover, atleast four different isoforms of BDNF are present innormal tissue and these are differentially expressedfollowing seizures and neuronal excitation (Metsis etal., 1993; Timmusk et al., 1993). Isoform expressionremains to be defined in the normal and injured spinalcord.

BDNF-IR Glia in Normal Adult Spinal Cord

We previously found that BDNF-IR and NT-3-IRastrocytes occurred in spinal cord white matter (Drey-fus et al., 1999). Extending this work, the present find-ings indicate that oligodendrocytes and microglia/macrophages also exhibit BDNF-immunolabeled pro-tein in the normal adult spinal cord. Taken together,our results suggest that most oligodendrocytes andsmall subsets of astrocytes and microglia/macro-phages in the normal adult cord synthesize trophinsor take up trophins themselves. Further investigationis necessary to determine whether BDNF is producedendogenously or taken up from other sources by eachof these glial subtypes. Our results are consistent withprevious investigations in brain showing increasedglial production of trophin mRNA surrounding aninjury. Several lines of evidence support the notionthat glia produce trophins. Astrocytes, oligodendro-cytes and microglia/macrophages contain both neu-rotrophin mRNA and immunolabeled protein in vitroand in vivo (Byravan et al., 1994; Zhou & Rush, 1994;

reyfus et al., 1999; Goss et al., 1998; Elkabes et al.,996; Dai et al., 1997; Arendt et al., 1995).

patiotemporal Alterations in BDNF-IR Profilesfter Injury

One day and 1 week following injury, the majorityf spinal oligodendrocytes remain BDNF-immuno-ositive, though their morphology becomes dis-upted. The normal axial and radial orientation of theells is lost after injury. Six weeks postinjury, APC-ositive profiles are absent at the wound site. It is

ikely that these cells have died, since previous inves-igations documented oligodendrocyte apoptosis fol-owing brain and spinal injury (Crowe et al., 1997;human et al., 1997; Barres et al., 1993). However, it is

also possible that these cells survive at the wound sitebut no longer express the antigen recognized by theanti-APC antibody. Recent work indicates that while

cBS*


massive oligodendrocyte death occurs acutely at thewound site, adjacent oligodendrocytes, associatedwith intact axons are spared, even in the presence ofinfiltrating microglia/macrophages (Frei et al., 2000).In our study, at all postinjury times examined, both 10and 20-mm rostral to injury, many BDNF-immunopo-sitive oligodendrocytes were present, resembling con-

FIG. 3. Most APC-positive oligodendrocytes contain BDNF-immuytes are APC/BDNF-IR. (C and D) One week after injury, oligodenDNF-IR. (E and F) Six weeks after injury, no APC-IR profiles are pix weeks after injury, 10 and 20 mm away from the wound, many A, APC-IR profile; arrow, BDNF/APC-IR profile. Bar, 100 mm.

trols. While it has been shown that oligodendrocytesdistant from the site of injury degenerate, our resultsindicate that many of those that remain continue toeither take up or produce BDNF.

At the injury site, the number of BDNF-IR profilesincreased dramatically 1 and 6 weeks after injury.These profiles mostly consisted of astrocytes and

ling. (A and B) In a laminectomized control rat, most oligodendro-s near the wound are morphologically disrupted but many are still

at the injury site, though many BDNF-IR profiles can be seen. (G–J)sitive oligodendrocytes are also BDNF-IR. Green, BDNF; Red, APC;

nolabedrocyteresentPC-po


lwrOB


A

microglia/macrophages which constituted the glialscar and filled the wound syrinx. Our findings sug-gest that glial trophin content increases in thewound region. Previous studies examining astro-cyte mRNA suggested that elevated trophins fol-lowing CNS insult is attributable to increased gene

FIG. 4. Following injury, an increased proportion of OX42/BDaminectomized controls, few OX42 profiles are BDNF-IR. (C and D)

hite matter near the injury site. (E and F) One week after injury, moostral to the wound, many OX42/BDNF-IR cells are present in a subX42-positive cells are not BDNF-IR, in a subject sacrificed 6 wDNF/OX42-IR profile. Bar, 100 mm.


expression surrounding the injury site (Goss et al.,1998; Gall et al., 1991; Arendt et al., 1995; Ballarin etal., 1991). We are presently examining whether ourobservations in injured cord represent a quantita-tive increase in trophin production, or de novo syn-thesis by new populations.

microglia/macrophages is observed at the lesion. (A and B) Inay after injury, several OX42/BDNF-IR profiles are observed in the2-positive microglia/macrophages are BDNF-IR. (G and H) 10 mm

crificed 1 week after injury. (I and J) 20 mm rostral the wound, mostfter injury. Green, BDNF; Red, OX42; *, OX42-IR profile; arrow,

NF-IROne dst OX4ject sa

eeks a


Increased numbers of BDNF-IR astrocytes and mi-croglia/macrophages were most dramatic at the pri-mary wound site, extended to at least 10 mm, butapproximated controls 20-mm rostral to injury. Con-sequently, increased glial trophin content is largelylimited to the site of the primary wound and thepenumbra, or surrounding tissue which is subject todelayed, secondary damage. Our findings are conso-nant with previous reports describing increased tro-phin content and mRNA limited to the regions sur-rounding brain lesions (Oderfeld-Nowak et al., 1992;Bakhit et al., 1991; Funakoshi et al., 1993; Goss et al.,1998; Arendt et al., 1995). The present and previousreports suggest that increased trophin expression iscontrolled by factor(s) unique to the injury site, poten-tially related to ischemia, excitotoxicity, blood–brainbarrier disruption, reperfusion, and/or inflammation.The proximodistal gradient of changes we observedsuggests that biochemical characterization of the in-jury site may reveal mechanisms regulating BDNFlevels. Moreover, the current findings suggest thatcommon mechanisms may affect BDNF content inastrocytes and macrophages/microglia, since in-creased proportions of both glial subtypes exhibitedBDNF at the injury site.

Implications of Increased Glial Trophin Content

The roles of reactive astrocytes and transformedmicroglia/macrophages in healing and wound repairare not fully understood. For example, it has beensuggested that reactive astrocytes and microglia/mac-rophages may enhance neuronal plasticity by phago-cytizing cellular debris (Rudge et al., 1989; Politis &Miller, 1985; Kesslak et al., 1986). However, reactiveastrocytes may act as barriers to axonal sprouting byexpression of inhibitory molecules (McKeon et al.,1997; Liuzzi & Lase, 1987); microglia/macrophagesrelease cytotoxins known to cause cell death. Similarinhibition is exhibited by oligodendrocyte myelin fol-lowing injury (Bandtlow et al., 1990; Schwab & Caroni,1988). Moreover, it is possible that reactive glia takeup and sequester trophins that would otherwise aid inthe survival of remaining neurons. Similarly, thesecells may produce and release trophins, which mightraise neuronal susceptibility to excitotoxic cell deathby increasing N-methyl-d-aspartate receptor-medi-ated intracellular calcium concentrations, initiatingapoptotic signaling cascades (Kerr et al., 1999; Lin etal., 1999). In contrast, enhanced glial trophin produc-tion may help survival and repair of remaining neu-rons. We are presently examining the potentially ben-

eficial and deleterious effects of increased trophins atthe injury site.

Several lines of evidence suggest that increasingtrophin content following CNS injury may improverecovery. Treatment with neurotrophins followingspinal cord injury enhances spinal and cortical neuro-nal survival and process regrowth (Jiang et al., 1997;Ye & Houle, 1997; Tuszynski et al., 1998; Schnell et al.,1994; Nakahara et al., 1996; Bregman et al., 1998; Breg-man et al., 1997). However, penetration of exogenoustrophins is limited (Dittrich et al., 1996; Shuman et al.,1997) and cellular origins and responses to neurotro-phins in normal and lesioned cord are obscure. Ourexperiments indicate that astrocytes, oligodendro-cytes, and microglia/macrophages are sources ofthese proteins both in normal cord and in injured cordover extended postinjury times. Future therapeuticapproaches may involve antagonizing inhibitory glialagents and enhancing naturally occurring trophinelaboration.

CONCLUSIONS

In conclusion, we have found that astrocytes, oligo-dendrocytes and microglia/macrophages exhibit dis-tinct BDNF-immunolabeling patterns in both normaland injured spinal cord. Moreover, astrocytic and mi-croglial immunolabeling increases following injuryand the proportion of these cells which are BDNF-IRalso increases. Our findings suggest that each glialsubtype plays a role in regulating local trophin func-tion in the normal cord via trophin production and/oruptake. Moreover, enhanced glial BDNF is evident atthe site of spinal injury, paralleling previous observa-tions of elevated nerve growth factor in the injuredbrain. Our findings further indicate that glial alter-ations in trophin content are limited to the regiondirectly surrounding the wound. Consequently, glialtrophins may play important roles in the response toinjury and in the potential for functional recovery. Weand others have begun to define the molecular signalsthat regulate neurotrophin expression (Zafra et al.,1992; Wu et al., 1996; Lackland et al., 1996; Schwartz &Nishiyama, 1994; Lu et al., 1991). These studies suggestthat a variety of neuronal signals, including gluta-mate, acetylcholine, and cytokines, regulate glial tro-phin gene expression. Elucidation of mechanisms reg-ulating spinal trophin production and responsivenessmay delineate new therapeutic approaches.


A

B

B

B

B

B

B

B

B

B

C

C

D

D

D

E

F

F

G

G

H

J

K

K

K

K

L

L

L

L


A

ACKNOWLEDGMENTS

The authors thank Dr. Lyn Jakeman for expert preparation of thespinal cord tissue used in these experiments and Dr. David Kaplanfor generously providing BDNF primary antiserum. The Christo-pher Reeve Paralysis Foundation and NICHD HD23315 supportedthis work.

REFERENCES

Aguayo, A. J., Rasminsky, M., Bray, G. M., Carbonetto, S., McKer-racher, L., Villegas-Perez, M. P., Vidal-Sanz, M., & Carter, D. A.(1991) Degenerative and regenerative responses of injured neu-rons in the central nervous system for adult mammals. Phil. Trans.R. Soc. London [Biol.] 331, 337–343.rendt, T., Bruckner, M. Kk, Krell, T., Pagliusi, S., Kruska, L., &Heumann, R. (1995) Degeneration of rat cholinergic basal fore-brain neurons and reactive changes in nerve growth factor ex-pression after chronic neurotoxic injury—II. Reactive expressionof the nerve growth factor gene in astrocytes. Neuroscience 65,647–659.

akhit, C., Armanini, M., Bennett, G. L., Wong, W. L. T., Hansen,S. E., & Taylor, R. (1991) Increase in glia-derived nerve growthfactor following destruction of hippocampal neurons. Brain Res.560, 76–83.

allarin, M., Ernfors, P., Lindefors, N., Persson, H. (1991) Hip-pocampal damage and kainic acid injection induce a rapid in-crease in mRNA for BDNF and NGF in the rat brain. Exp. Neurol.114, 35–43.

andtlow, C., Zachleder, T., & Schwab, M. E. (1990) Oligodendro-cytes arrest neurite outgrowth by contact inhibition. J. Neurosci.10, 3837–3848.

arres, B. A., Jacobson, M. D., Schmid, R., Sendtner, M., & Raff,M. C. (1993) Does oligodendrocyte survival depend on axons?Curr. Biol. 3, 489–497.

hat, R. V., Axt, K. J., Fosnaugh, J. S., Smith, K. J., Johnson, K. A.,Hill, D. E., Kinzler, K. W., & Baraban, J. M. (1996) Expression ofthe APC tumor suppressor protein in oligodendroglia. Glia 17,169–174.

jartmar, C. (1998) Morphological heterogeneity of cultured spinaland cerebral rat oligodendrocytes. Neurosci. Lett. 247, 91–94.

regman, B. S., McAtee, M., Dai, H. N., & Kuhn, P. L. (1997)Neurotrophic factors increase axonal growth after spinal cordinjury and transplantation in the adult rat. Exp. Neurol. 148, 475–494.

regman, B. S., Broude, E., McAtee, M., & Kelley, M. S. (1998)Transplants and neurotrophic factors prevent atrophy of matureCNS neurons after spinal cord injury. Exp. Neurol. 149, 13–27.

yravan, S., Foster, L. M., Phan, T., Verity, A. N., & Campagnoni,A. T. (1994) Murine oligodendroglial cells express nerve growthfactor. Proc. Nat. Acad. Sci. USA 91, 8812–8816.

ho, H. J., Kim, J. K., Park, H. C., Kim, D. S., Ha, S. O., & Hong, H. S.(1998) Changes in brain-derived neurotrophic factor immunore-activity in rat dorsal root ganglia, spinal cord, and gracile nucleifollowing cut or crush injuries. Exp. Neurol. 154, 224–230.

rowe, M. J., Bresnahan, J. C., Shuman, S. L., Masters, J. N., &Beattie, E. C. (1997) Apoptosis and delayed degeneration afterspinal cord injury in rats. Nature Med. 3, 73–76.ai, X. D., Lercher, L., Yang, L., Shen, M., Black, I. B., & Dreyfus,C. F. (1997) Expression of neurotrophins by basal forebrain (BF)oligodendrocytes. Soc. Neurosci. Abstr. 23, 331.


ittrich, F., Ochs, G., Grosse-Wilde, A., Berweiler, U., Yan, Q.,Miller, J. A., Toyka, K. V., & Sendtner, M. (1996) Pharmacokineticsof intrathecally applied BDNF and effects on spinal motoneurons.Exp. Neurol. 141, 225–239.reyfus, C. F., Dai, X. D., Lercher, L., Racey, B., Friedman, W. J., &Black, I. B. (1999) Expression of neurotrophins in the adult spinalcord in vivo. J. Neurosci. Res. 56, 1–7.

lkabes, S., DiCicco-Bloom, E. M., & Black, I. B. (1996) Brain micro-glia/macrophages express neurotrophins that selectively regulatemicroglial proliferation and function. J. Neurosci. 16, 2508–2521.

rei, E., Klusman, I., Schnell, L., & Schwab, M. E. (2000) Reactions ofoligodendrocytes to spinal cord injury: Cell survival and myelinrepair. Exp. Neurol. 163, 373–380.

unakoshi, H., Frisen, J., Barbany, G., Timmusk, T., Zachrisson, O.,Verge, V. M. K., & Persson, H. (1993) Differential expression ofmRNAs for neurotrophins and their receptors after axotomy ofthe sciatic nerve. J. Cell Biol. 123, 455–465.all, C. M., Murray, K., & Isackson, P. J. (1991) Kainic acid-inducedseizures stimulate increased expression of nerve growth factormRNA in rat hippocampus. Mol. Brain Res. 9, 113–123.oss, J. R., O’Malley, M. E., Zou, L., Styren, S. D., Kochanek, P. M.,& DeKosky, S. T. (1998) Astrocytes are the major source of nervegrowth factor upregulation following traumatic brain injury inthe rat. Exp. Neurol. 149, 301–309.efti, F., Hartikka, J., & Knusel, B. (1989) Function of neurotrophicfactors in the adult and aging brain and their possible use in thetreatment of neurodegenerative diseases. Neurobiol. Aging 10,515–533.

iang, H., St. Ulme, D., Dickens, G., Chabuk, A., Lavarreda, M.,Lazarovici, P., & Guroff, G. (1997) Both p140-trk and p75-NGFRnerve growth factor receptors mediate nerve growth factor-stim-ulated calcium uptake. J. Biol. Chem. 272, 6835–6837.

esslak, J. P., Nieto-Sampedro, M., Globus, J., & Cotman, C. W.(1986) Transplants of purified astrocytes promote behavioral re-covery after frontal cortex ablation. Exp. Neurol. 93, 377–390.

err, B. J., Bradbury, E. J., Bennett, D. L., Trivedi, P. M., Dassan, D.,French, J., Shelton, D. B., McMahon, S. B., & Thompson, S. W.(1999) Brain derived neurotrophic factor modulates nociceptivesensory inputs and NMDA-evoked responses in the rat spinalcord. J. Neurosci. 19, 5138–5148.

oliatsos, V. E., Clatterbuck, R. E., Winslow, J. W., Cayouette, M. H.,& Price, D. L. (1993) Evidence that brain-derived neurotrophicfactor is a trophic factor for motor neurons in vivo. Neuron 10,359–367.

renz, N. R., & Weaver, L. C. (2000) Nerve growth factor in glia andinflammatory cells of the injured rat spinal cord. J. Neurochem. 74,730–739.

ackland, J., Wu, H., Black, I. B., & Dreyfus, C. F. (1996) Differentialregulation of adult and neonatal cortical astrocyte function bydepolarizing signals. Soc. Neurosci. Abstr. 22, 547.

evine, E. S., Dreyfus, C. F., Black, I. B., & Plummer, M. R. (1995)Brain-derived neurotrophic factor rapidly enhances synaptictransmission in hippocampal neurons via postsynaptic tyrosinekinase receptors. Proc. Natl. Acad. Sci. USA 92, 8074–8077.

in, S., Wu, K., Len, G., Xu, J., Levine, E. S., Suen, P., Mount, H. T.,& Black, I. B. (1999) Brain-derived neurotrophic factor enhancesassociation of protein tyrosine phosphatase PTP1D with theNMDA receptor subunit NR2B in the cortical postsynaptic den-sity. Mol. Brain. Res. 70, 18–25.

iuzzi, F., & Lasek, R. J. (1987) Astrocytes block axonal regenerationin mammals by activating the physiological stop pathway. Science237, 642–645.

S

S

S

S

S

S

T

T

W

Y

Z

Z


Lu, B., Yokoyama, M., Dreyfus, C. F., & Black, I. B. (1991) NGFgene expression in actively growing brain glia. J. Neurosci. 11,318 –326.

McKeon, R. J., Silver, J., & Large, T. H. (1997) Expression of full-length trkB receptors by reactive astrocytes after chronic CNSinjury. Exp. Neurol. 148, 558–567.

Metsis, M., Timmusk, T., Arenas, E., & Persson, H. (1993) Differen-tial usage of multiple brain-derived neurotrophic factor promot-ers in the rat brain following neuronal activation. Proc. Natl. Acad.Sci. USA 90, 8802–8806.

Nakahara, Y., Gage, F. H., & Tuszynski, M. H. (1996) Grafts offibroblasts genetically modified to secrete NGF, BDNF, NT-3, orbasic FGF elicit differential responses in the adult spinal cord. CellTrans. 5, 191–204.

Nieto-Sampedro, M., Manthrope, M., Barbin, G., Varon, S., & Cot-man, C. W. (1999) Injury-induced neuronotrophic activity in adultrat brain: Correlation with survival of delayed implants in thewound cavity. J. Neurosci.

Oderfeld-Nowak, B., Bacia, A., Gradkowska, M., Fusco, M., Vantini,G., Leon, A., & Aloe, L. (1992) In vivo activated brain astrocytesmay produce and secrete nerve growth factor-like molecules.Neurochem. Int. 21, 455–461.

Politis, M. J., & Miller, J. E. (1985) The roles of reactive gliosis andmitosis on tropic factor production in traumatized nervous sys-tem tissue. Brain Res. 346, 186–189.

Rudge, J. S., Smith, G. M., & Silver, J. (1989) An in vitro model ofwound healing in the CNS: Analysis of cell reaction and interac-tion at different ages. Exp. Neurol. 103, 1–16.

ayers, S. T., Khan, N., Ahmed, Y., Shahid, R., & Khan, T. (1998)Preparation of brain-derived neurotrophic factor- and neurotro-phin-3-secreting cells by infection with a retroviral vector. J. Mol.Neurosci. 10, 143–160.

chnell, L., Schneider, R., Kolbeck, R., Barde, Y. A., & Schwab, M. E.(1994) Neurotrophin-3 enhances sprouting of corticospinal tract

during development and after adult spinal cord lesion. Nature367, 170–173.

chwab, M. E., & Caroni, P. (1988) Oligodendrocytes and myelin arenonpermissive substrates for neurite growth and fibroblastspreading in vitro. J. Neurosci. 8, 2381–2393.

chwartz, J. P., & Nishiyama, N. (1994) Neurotrophic factor geneexpression in astrocytes during development and following in-jury. Brain. Res. Bull. 35, 403–407.

human, S. L., Bresnahan, J. C., & Beattie, M. S. (1997) Apoptosis ofmicroglia and oligodendrocytes after spinal cord contusions inrats. J. Neurosci. Res. 50, 798–808.

tokes, B. T., Noyes, D. H., & Behrmann, D. L. (1992) An electro-mechanical spinal injury technique with dynamic sensitivity.J. Neurotrauma 9, 187–195.

immusk, T., Palm, K., Metsis, M., Reintam, T., Paalme, V., Saarma,M., & Persson, H. (1993) Multiple promoters direct tissue-specificexpression of the rat BDNF gene. Neuron 10, 475–489.

uszynski, M. H., Weidner, N., McCormack, M., Miller, I., Powell,H., & Conner, J. (1998) Grafts of genetically modified Schwanncells to the spinal cord: Survival, axon growth, and myelination.Cell Trans. 7, 187–196.u, H., Black, I. B., & Dreyfus, C. F. (1996) Differential regulation ofneurotrophin expression in basal forebrain astrocytes. Soc. Neu-rosci. Abstr. 22, 548.

e, J. H., & Houle, J. D. (1997) Treatment of the chronically injuredspinal cord with neurotrophic factors can promote axonal regen-eration from supraspinal neurons. Exp. Neurol. 143, 70–81.

afra, F., Lindholm, D., Castren, E., Hartikka, J., & Thoenen, H.(1992) Regulation of brain-derived neurotrophic factor and nervegrowth factor mRNA in primary cultures of hippocampal neu-rons and astrocytes. J. Neurosci. 12, 4793–4799.

hou, X. F., & Rush, R. A. (1994) Localization of neurotrophin-3-likeimmunoreactivity in the rat central nervous system. Brain Res.643, 162–172.


brain-derived neurotrophic factor in astrocytes, oligodendrocytes, and microglia/macrophages after...

Documents