Intraspinal transplantation of microglia


Material Information

Intraspinal transplantation of microglia studies of host cellular responses and effects on neuritic growth
Alternate title:
Studies of host cellular responses and effects on neuritic growth
Physical Description:
vii, 141 leaves : ill. ; 29 cm.
Rabchevsky, Alexander George, 1966-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Tissue Transplantation -- veterinary   ( mesh )
Spinal Cord -- transplantation   ( mesh )
Microglia -- transplantation   ( mesh )
Microglia -- physiology   ( mesh )
Graft Rejection   ( mesh )
Neurons -- growth & development   ( mesh )
Nerve Regeneration   ( mesh )
Astrocytes -- transplantation   ( mesh )
Rats   ( mesh )
Department of Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Neuroscience -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1995.
Bibliography: leaves 125-139.
Statement of Responsibility:
by Alexander George Rabchevsky.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 028281863
oclc - 50706796
System ID:

Full Text








I would like to begin by thanking the members of my dissertation
advisory committee, Dr. Paul Reier, Dr. Wolfgang Streit, Dr. David Muir, and
Dr. Joel Schiffenbauer for enduring some grueling meetings, and still
showing up to the next ones.
I would especially like to thank my thesis advisor, Dr. Paul J. Reier, for
having given me the opportunity to work in his laboratory with the freedom
and latitude to explore my ideas, and for supporting most of my endeavors.
I thank Dr. "Jake" for introducing me to the microglial cell, and for
showing me potential ways to characterize its involvement in CNS pathology
and regeneration. I also thank him for all of his help and advice, and for
letting me pester him constantly about new ideas.
Dr. David Muir provided me with the first exposure to the world of a
culture dish, and for this I am totally......confused! I appreciate his taking the
time to help me design the in vitro studies included in this thesis, and for the

use of his lab and equipment when I first got started.
I thank Dr. Dan "The Man" Theele for being a good friend and
confidant, as well as an inspirational help to me when things were going
rough (especially when I had no room to complain).
I thank Drs. William Luttge, Don Walker and Gregory Schrimsher for
there suggestions and guidance in the statistical analysis of the data.
Sharon Walter deserves all the credit and appreciation for doing the
molecular analysis of IL-1 mRNA expression of LPS-stimulated microglia.

Dr. Colin Sumner and Mrs. Tammy Gault are deeply appreciated for

their advice and technical assistance in teaching me to culture microglia.
Dr. Gerry Shaw is appreciated for helping me with the "kinesin
project" early on B.M. (before microglia), and for generously supplying his
neurofilament antibodies used in these studies.
Stacy Wall and Eric Rick were instrumental in enabling me to get this
thesis to print out the way I wanted it to. And thanks goes out to all the
Neuroscience secretaries (or office assistants?) who know my Xerox number
by heart because I asked for their help daily!
Minnie Smith and Barbara O'Steen deserve recognition for
maintaining stability in the dynamic Reier lab, and they made my life a lot
easier by helping out with any and all favors I asked of them. Pat Shinholster
also warrants praise for her kindness and concern regarding problems I
encountered along the way to graduate freedom.
I would like to acknowledge the "Rick Hansen Man in Motion World
Tour Society Fund" in Canada for providing me with a two year studentship
award allowing me to develop techniques potentially useful in the future.
Thanks also go to W.L. Gore & Associates, Inc. (Flagstaff, AZ) for supplying
me the Gore-tex 'PTFE' tubing, and Ed Morales for making the polymeric

Lastly, I would like to thank my Dad and sister, George and Natasha,
for supporting me the past five years over a long distance. And, of course, I
owe everything else to Gisele, the love of my life!



ACKNOWLEDGEMENTS...................................................................................... ii

A BSTR A C T ............................................................................................................... v i


1. BA CK G RO U N D ................................................................................................... 1

Spinal Cord Injury and Gliosis................................................................... 1
Overview of Microglial Ontogeny and Responses to Injury................ 2
In Situ Grafting of Glia Derived from Tissue Culture........................... 7
Microglial Involvement in Neural Regeneration................................. 9
Synopsis of the Evolution of this Dissertation........................................ 9

NEURITE GROWTH-PROMOTING ACTIVITY........................................ 12

Introduction.................................................................................................. 12
Methods and Materials............................................................................... 14
R esults............................................................................................................ 20
D iscussion...................................................................................................... 35


Introduction.................................................................................................. 43
Methods and Materials............................................................................... 46
R esults............................................................................................................ 53
D iscussion...................................................................................................... 91


MICROGLIA INTO THE ADULT RAT SPINAL CORD.......................... 104

Introduction................................................................................................ 104
M ethods and M aterials............................................................................. 105
Results.......................................................................................................... 107
D iscussion.................................................................................................... 113

5. O V ERV IEW ...................................................................................................... 116

REFEREN CES ......................................................................................................... 125

BIO G RA PH ICA L SK ETCH ................................................................................... 140


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Alexander George Rabchevsky

August, 1995

Chairman: Dr. Paul J. Reier
Major Department: Neuroscience

Investigations into the cellular biology of microglial cells suggest these
cells not only initiate the rejection of neural grafts into the central nervous
system (CNS) but also secrete molecules that are potentially toxic to neurons.
Studies of cytokine and growth factor production by activated microglia/brain
macrophages, on the other hand, suggest they may directly or indirectly
promote neuritic elongation in the injured CNS. In view of these contrasting
perspectives, this dissertation utilizes a transplantation approach to
investigate the influence on neuronal growth of tissue cultured microglia
placed into the injured rat spinal cord. The specific focus is on the
histological and immunohistochemical demonstration of neuritic ingrowth
into various grafts containing matrices of gelfoam that were impregnated

with cultured microglia or astrocytes.
Prominent neuritic ingrowth was observed in the microglia-
impregnated implants that displayed intense OX-42-immunoreactivity (IR).

In contrast, astrocyte implants showed less neuritic growth-promoting effects.
In addition to neuritic ingrowth into the microglia-seeded gelfoam implants,

there was an infiltration of host cellular elements, many of which were
laminin-IR and thus thought to be microvascular, Schwann cell and/or
mesechymal infiltrates. Neuritic growth also was seen in activated microglial
environments of the host outside the polymeric tube implants. Lastly, some
control (i.e., cell-free) implants demonstrated a slight ingrowth of neurites,
but this only occurred coincident with the infiltration of host cellular
elements which consistently involved OX-42-IR microglia/brain

Collectively, this evidence argues against a neurotoxic role played by
microglia in the injured CNS. Accordingly, these cells may at least partially
counteract the events that contribute to poor regeneration in the mammalian
CNS. It remains to be determined, however, whether the neuritic growth-
promoting effect observed is directly related to grafted microglia or
secondarily associated with host cellular elements that could be more directly
conducive to the outgrowth of neuronal processes.


Spinal Cord Injury and Gliosis

After injury to the spinal cord there is an immediate extravasation of

hemorrhagic elements into the lesioned area. This is followed by neutrophil
invasion and neuronophagia (Means and Anderson, 1983). Macrophages and
microglial cells accumulate at the lesion site during the first week. Astrocytic
processes soon delineate an area of necrosis and form an intense astroglial
lining that, in many instances, develops into a compact scar which can be
invaded by Schwann cells (Dusart et al., 1992), as well as mesodermal
elements that establish a dense collagenous matrix.
Astrocytes, together with other non-neuronal cellular elements (i.e,.
fibroblasts), form densely interwoven scars that fill the space vacated by dead
or dying cells that are thought to inhibit regeneration in the central nervous

system (CNS) (Reier et al., 1983; 1988; Feringa et al., 1984; Bernstein et al., 1985;
Liuzzi and Lasek, 1987). Studies characterizing successful axonal regrowth
through scar tissue in the adult mammalian spinal cord demonstrate that
neurites will invade traumatized tissue domains, but only in conjunction
with nonneuronal elements, most notably ependymal and mononuclear cells
(Guth et al., 1983; 1985). One mononuclear cell type endogenous to the CNS
which becomes activated after CNS damage and whose effects on neuronal
elongation and glial scar formation is poorly understood is the microglial cell.

Overview of Microglial Ontogeny and Responses to Injury

During embryonic development, microglia have an ameboid
morphology, are highly motile, and display phagocytic function thought to
play a role in scavenging other dying cells during development (Ashwell,
1990). However, during the course of CNS development ameboid microglia

disappear and ramified microglia, having a small soma with long branching
processes, increase in number. It has been speculated that some of the
ameboid microglia eventually differentiate into ramified microglia (Ling,
1981; Ling and Wong, 1993). The proposed origins of microglia, however, are
still controversial. Whether they are ectodermally or mesodermally derived
has been the focus of much debate (see Theele and Streit, 1993). Currently, the
most plausible theory is that they initially derive from a precursor cell related
to the monocyte/macrophage lineage associated with embryonic hemopoietic
organs, such as the bone marrow, spleen, or liver.
Microglia have many phenotypic characteristics that vary depending
on the stage of development, location within the CNS, and on whether the
CNS has been perturbed by injury or disease. Studies also have demonstrated
that microglia in vitro display polymorphic phenotypes that usually depend
on the culturing conditions. This diverse morphology of microglia is
thought to be closely associated with their functional state.
Ramified microglia in the normal, adult rat CNS are uniformly
dispersed, unlike other glial cells, and are thought to be in a functional
"resting state" (Figs. 1-1A-C). These glia represent endogenous cells of the

CNS involved in ionic homeostasis and are thought to act as "pinocytotic
filters" sampling the extracellular fluids for foreign antigens and other
diffusible substances (Ward et al., 1991). When the CNS is injured or

Fig. 1-1. Photomicrographs displaying the heterogeneity of microglial
phenotypes found within the rat spinal cord. All panels represent high
magnifications of horizontal cryosections immunofluorescently stained with
OX-42 antibody, recognizing the CR3 receptor expressed on microglia and
macrophages (X320). A shows highly ramified, parenchymal microglia, B
and C demonstrate the different microglial profiles found in white matter
regions, D shows an injured area in the rat spinal cord that is highlighted by
an increased hypertrophy of microglia and their transformation into brain
macrophages. Note the polymorphic stages of activation (i.e., process-bearing
to ameboid; arrows).



affected by disease, microglia become reactive and are thought to be
functionally activated (Streit and Kreutzberg, 1988; Streit et al., 1988).
Microglial activation involves cellular hypertrophy and retraction of
cytoplasmic processes (Fig. 1-1D). In regions of necrosis, activated microglia
can transform into brain macrophages that proliferate and phagocytose
cellular debris (Streit and Kreutzberg, 1988; Graeber et al., 1989). Microglial
activation also is characterized by marked changes in phenotype. This
includes upregulation of existing surface antigens as well as de novo
expression of such molecules (Graeber et al., 1988; Streit et al., 1989).
It is difficult to demonstrate conclusively the invasion of central
nervous (CNS) tissue by blood monocytes because specific markers that
distinguish macrophages from activated microglia do not exist. However, it
appears that the increase in cell number after CNS damage is related to a
recruitment and proliferation of microglial cells, as well as of blood
monocytes (Schelper and Adrian, 1986; Graeber et al., 1989; Morshead and
VaderKooy, 1990; Andersson et al., 1991; Marty et al., 1991). Andersson et al.
(1992) report monocytic recruitment even in the absence of blood-brain
barrier disruption. Mononuclear phagocytes clear cellular debris and secrete
cytokines/growth factors which are involved in immunological responses to
tissue repair. They also can express molecules that are potentially toxic to
neural cells and may thus negatively affect the potential for regeneration.
Neurotoxicity and Regeneration-Inhibiting Effects of Microglia
In tissue culture, stimulation of microglia with lipopolysaccharide
(LPS), interferon-gamma (IFN-y), and/or zymosan A results in the
elaboration of various noxious and cytotoxic agents (Banati et al., 1993),
including reactive oxygen intermediates (Colton and Gilbert, 1987), tumor
necrosis factor-alpha (TNF-a) (Frei et al., 1987; Sawada et al., 1989), glutamate

(Piani et al., 1991), and low molecular weight factors with neurotoxic activity
(Giulian et al., 1993a,b). The progressive destruction of axotomized retinal
ganglion cells has been shown to be enhanced or suppressed with
macrophage-stimulating (MSF) or -inhibiting factors (MIF), respectively
(Thanos et al., 1993). The retention of retinal ganglion cells by MIF treatment
can lead to the regrowth of axons in vivo and in vitro. Although the
mechanisms by which these factors produce their effects are poorly
understood, the suggestion has been made that suppression of brain
macrophages is essential for regeneration to occur.
Neurotrophic Properties of Microglia
In contrast to the proposed adverse nature of microglia/brain
macrophages, it has been shown that changes in axonal growth-promoting
properties of the injured CNS may be produced by mononuclear phagocytes
that invade lesioned areas and then modify nonpermissive substrata (David
et al., 1990). Microglia are known to secrete trophic molecules such as
transforming growth factor-beta (TGF-p) (Constam et al., 1992; Lindholm et al.,
1992; Finch et al., 1993; Kiefer et al., 1993), basic fibroblast growth factor (bFGF)
(Shimojo et al., 1991), and nerve growth factor (NGF) (Mallat et al., 1989;
Lindholm et al., 1992). This has prompted investigations into the putative
neurotrophic properties of cultured microglia.
Studies in which neurons from various CNS regions have been co-
cultured with purified microglia and/or microglia-conditioned medium
concluded that both contact with microglia or growth in microglia-
conditioned medium promoted neuronal survival and neuritic arborization
(Nakajima et al., 1989,1993; Nagata et al., 1993a,b; Zhang and Federoff, 1993;
Chamak et al., 1994). It remains to be determined, however, whether
microglia produce similar effects in the injured CNS.

In Situ Grafting of Glia Derived from Tissue Culture

Over the years, neural tissue transplantation has served as an effective
tool for investigating the nature of various cellular interactions in the
injured CNS (Reier et al 1983; 1986; Houle and Reier, 1988; Jakeman and
Reier, 1991). In this regard, glial cells from both the CNS and PNS have been
transplanted into the CNS as whole (i.e., peripheral nerve) or dissociated
tissue, as well as in vitro-derived cells. Using tissue culture as a source of
donor cells allows the greatest control over transplant composition because
such cells can be manipulated to some extent using specific culturing
protocols and isolation techniques. Cultured oligodendrocytes transplanted
into hypomyelinated rodents have been shown to form myelin sheaths with
some myelination occurring at a distance from the transplantation site (Gout
et al., 1988; Rosenbluth et al., 1990). Other groups have transplanted mouse
glia to repopulate ethidium bromide lesions in rats and showed that even
xenogeneic donor oligodendrocytes can remyelinate host axons (Crang and
Blakemore, 1991).
Transplantation of astrocytes into the adult rat spinal cord results in
the extensive migration of injected or graft-derived astrocytes (Goldberg and
Bernstein, 1988; Bernstein and Goldberg, 1989; Wang et al., 1995). Cultured
astrocytes have also been transplanted to examine their effects on
remyelination (Franklin et al., 1991) and scar formation (Smith and Silver,
1988). Under certain circumstances, astroglia may provide a matrix that will
support axonal growth in vitro and during axonal development and
regrowth in mammalian CNS tissue (Matthews et al., 1979; David et al., 1984;
Fishman and Kelly, 1984; Noble et al., 1984; Fallon, 1985; Guth et al., 1985;
Miller et al., 1986; Smith et al., 1986; Assouline et al., 1987). Recently,

cultured astrocytes injected into hemisected adult rat spinal cords alone or
presented in a gelfoam matrix were shown to reduce scarring, as measured by
glial fibrillary acidic protein (GFAP) specific to astrocytes, and to increase the

intensity of neurofilament staining (Wang et al., 1995). However, the
prelabeled, grafted astrocytes migrated out of the Gelfoam implants as early as
one week after implantation raising the possibilities that they either produced
their effects early after grafting or that their emigration created a more
permissive environment for neuritic growth.
Experimentally demyelinated CNS axons exert a powerful attractant
and mitotic effect on transplanted Schwann cells (Blakemore, 1984; Crang and
Blakemore, 1989). The invasion of myelinating Schwann cells into the CNS
depends on the concurrent loss of both astrocytes and oligodendrocytes as
demonstrated by the inability to detect Schwann cells after they are injected
into tissue where astrocytes are present (Blakemore et al., 1986). The capacity

of cultured Schwann cells to myelinate CNS axons in a glial deficient
environment has been demonstrated when these cells were transplanted into
ethidium bromide lesions in spinal cords irradiated with X-rays (Blakemore
et al., 1987a,b). Enriched Schwann cell grafts also can reduce both post-
traumatic cystic cavitation and astrocytic scar formation when injected into
the site of an acute compression lesion (Martin et al., 1991).

Considerable axonal growth has been demonstrated through cultured,
peripheral Schwann cells (combined with their collagen substratum) after
placement into the aspiration cavities of adult rat spinal cords (Paino and
Bunge, 1991; Paino et al., 1994). Recently, permselective, polymeric tubes were
filled with cultured Schwann cells seeded in MatrigelTM (Collaborative
Research, Inc.; Bedford, MA) to reconstruct complete fimbria-fornix and/or
spinal cord resection cavities in rats (Hoffman and Aebischer, 1993; Xu et al.,

1995). These studies demonstrated both the regrowth of cholinergic axons,
and the rostral/caudal ingrowth of spinal axons through the implants,
respectively. It was concluded that Schwann-cell-seeded guidance channels
provided directionally oriented biosynthetic bridges that induce CNS
regeneration via a direct association between Schwann cells and regenerating
axons. Functional assessment following transplantation of these grafts awaits
further study.

Microglial Involvement in the Neural Regeneration

At present, microglial-neuronal interactions can be construed as a two-
edged sword, due in part to two very different experimental approaches that
have been employed in vivo and in vitro (see Streit, 1993). Therefore, to
address the disparate results of these fundamentally different paradigms for
characterizing microglial function, the primary focus of this thesis is on the
examination of the cellular and neuritic responses to cultured microglia
transplanted into the injured, adult rat spinal cord. Unlike the
transplantation paradigms outlined above, cultured microglia have not yet
been transplanted into the CNS to study their cellular interactions, migratory
capabilities, differentiation or effects on the injured CNS. This is in spite of
the demonstration that macrophages injected into peripheral wounds
increase the rate of healing and reduce scarring (Danon et al., 1989).

Synopsis of the Evolution of this Dissertation

For general perspective, this thesis evolved from preliminary
experiments that were designed to investigate the involvement of microglia

in the process of neural allograft rejection in the CNS. Recent work in
cellular neuroimmunology has culminated in a new concept that views
microglia as a network of intrinsic, immunocompetent cells of the CNS
(Graeber and Streit, 1990). This sheds new light on the longstanding belief
that the CNS is an immunologically privileged system, and may explain
reasons for rejection. Immunohistochemical studies examining the
expression of major histocompatibility complex (MHC) molecules in both
normal and injured CNS tissue have shown that the principle cell type
expressing MHC antigens is the microglial cell (Matsumoto et al., 1986;
McGeer et al., 1988; Kono et al., 1989; Streit et al., 1989; Gehrmann et al., 1992;
Morioko et al., 1992; Sedgwick et al., 1993). There is also compelling evidence
that the immunocompetent microglia are the antigen-presenting cells (APCs)
of the CNS and are thus responsible for initiating antigen-directed immune
reactions mediated by T-lymphocytes that result in graft rejection (Frei et al.,
1987; Hickey and Kimura, 1988; Streit et al., 1988; Poltorak and Freed, 1989;
Graeber and Streit, 1990; Lawrence et al., 1990).
Therefore, experimental strategies were developed to deplete cells
capable of MHC expression (i.e., microglia) from allogeneic neural suspension
grafts. During the development of the microglial depletion/isolation
procedures, the question arose as to what effect such depletion would have on
axonal elongation either from embryonic CNS grafts, the host, or both.
Review of the literature underscored that nothing was actually known about
the effects of microglia on neuritic outgrowth in the injured CNS.
Accordingly, the experimental studies to be described in subsequent
chapters represent the first attempts at investigating the role of
microglia/brain macrophages in spinal cord regeneration. A primary issue in
this regard relates to the mode of delivery. In the first set of experiments

(Chapter 2) this entailed the use of biodegradable polymeric tubes containing a
gelfoam matrix that could be seeded with microglia derived from tissue
culture. While some interesting observations were obtained pertaining to a
possible supportive role of microglia, there was pronounced inflammation
occurring around the polymeric carrier tubes that clouded interpretation of
the control results. Another series of experiments (Chapter 3) was thus
conducted in which a more stable support tube was used with which
inflammation was not associated. The results of both investigations
collectively challenge the view, noted earlier, that microglia exert a totally
adverse effect on regeneration. In addition, some findings serendipitously
indicated that the microglial implants might modulate astroglial responses to
spinal cord injury, as well as the distribution of at least one ECM molecule
(i.e., laminin). Finally, the question of reliable prelabelling of implanted cells
was investigated (Chapter 4) as this is one of the more challenging technical
issues associated with the transplantation of cells that have endogenous
counterparts in the host CNS. Thus, some theoretical insights pertaining to
neuron-glial and glial-glial interactions have emerged that set the stage for
future experimental analyses (Chapter 5).



Microglia are known to secrete a variety of cytokines and growth
factors, such as interleukin-1- (IL-1-) and IL-6-like molecules (Giulian &
Baker, 1985; Giulian et al., 1986; Remick et al., 1988; Hetier et al., 1988; Frei et
al., 1989; Woodroofe et al., 1991; Ganter et al., 1992), tumor necrosis factor
(TNFa) (Frei et al., 1987; Remick et al., 1988; Woodroofe et al., 1991),
transforming growth factor-beta (TGF-p) (Constam et al., 1992; Lindholm et al.,
1992; Finch et al., 1993; Kiefer et al., 1993), basic fibroblast growth factor (bFGF)
(Shimojo et al., 1991), and nerve growth factor (NGF) (Mallat et al., 1989;
Lindholm et al., 1992). Since all these substances play interactive roles in
tissue repair and nerve regeneration, several investigations have begun to
explore neurotrophic and/or neurotropic properties of microglia under a
variety of experimental conditions. Some results have demonstrated that
microglia or microglia-conditioned medium can promote the survival of
CNS neurons, as well as enhance their elaboration of neuritic extensions
(Nakajima et al., 1989; 1993; Nagata et al., 1993a,b; Zhang and Federoff, 1993;
Chamak et al., 1994). Whether these effects can be attributed to the secretion of
cytokines/growth factors or to the deposition of extracellular matrix (ECM)
molecules by microglia remains to be determined.

However, the presence and abundance of microglia/brain macrophages
at sites of insult within the CNS has led others to propose that these cells may

exert more destructive effects (Banati et al., 1993). For example they could
exacerbate lesions through the production of putative cytotoxic molecules
such as superoxide anions, glutamate and nitric oxide (Colton & Gilbert, 1987;
Piani et al., 1991; Chao et al., 1992). It also has been reported that activation of
microglia can lead to the in vitro production of low molecular weight,
neurotoxic substances (Giulian, 1990; Giulian et al., 1993a,b).
In light of the conflicting views about microglial involvement in both
CNS pathology and regeneration, the present study was designed to test
whether cultured microglia exhibit any neurite growth-promoting actions
after being grafted into the injured rat spinal cord. Whether microglia can
have a modulatory effect on other glial (e.g., astrocytic) responses to trauma
was another issue of particular interest. In addition, how microglia may
influence the distribution of the ECM molecule laminin, commonly found
around blood vessels and along the external glial limiting membrane of the
normal CNS (Sanes, 1983), was also investigated. The reasoning stems from
the association of laminin with astrocytes after invasive mechanical lesions at
the interface between the cicatrix and blood vessels, and at reconstituted
subpial surfaces (Liesi et al., 1984; Bernstein et al., 1985). Our results suggest
that both grafted, as well as endogenous microglia that are present in regions
of inflammation can be compatible with growing neurites and that their
modulation of the microenvironment and gliotic responses may account for
these observations. Portions of this study have been previously summarized
(Rabchevsky et al., 1994; 1995).

Methods and Materials

Isolation of Microglia
Neonatal Sprague-Dawley rat brains were stripped of surrounding
meninges in Solution "D" (containing 0.137 M NaCl, 5.4 mM KC1, 0.02 mM
NaH2PO4, 0.02 mM KH2P04, 5.5 mM Dextrose, 58.5 mM Sucrose, 1x106 u.
Penicillin G-1.Og Streptomycin (Sigma) and 2.5mg Fungizone (Gibco) in 10 ml
distilled H20, pH 7.6). The tissue was then minced with a scalpel and
resuspended in 20 ml of Solution "D" to which 0.15 % trypsin (Worthington)
was added,and gently agitated on an orbital shaker at 370C for 20 min. DNase
(Sigma; 400 kunitz) was then added and the suspension was shaken for
another 10 min. An equal volume of DMEM (pH 7.2), supplemented with
10% fetal bovine serum (FBS), was used to quench enzymatic activity. The
suspension was subsequently filtered through a 130 gm Nitex filter. The
filtrate was pelleted (400 x g, 10 min.), resuspended with 10 ml complete
media, then filtered through a 40 gm Nitex filter. The cell concentration was
adjusted to ~-1x106 cells/ml of complete medium. Then 10 ml of suspension
was added to individual poly-l-lysine-coated 75 cm2 flasks before placing them
into a humidified incubator set at 370C, 8% C02 92% atmosphere for 3 days.
The conditioned medium was replaced with complete medium, and 1 week
later microglia were isolated from conditioned medium every 3-4 days by
simply pelleting floating cells collected in the supernatants.
Characterization of "Floating Cells"
After plating, approximately 90% of floating cells were microglia as
evidenced by labeling with Griffonia simplicifolia I-B4 isolectin (GSI-B4;
Sigma) (Streit, 1990) or Dil-labeled acetoacetylated LDL (DiI-ac-LDL;
Biomedical Technologies, Inc.) (Giulian et al., 1986). Removing non-adherent

cells after 1 hr. @ 370C yields >95% enrichment. Viability of the cells before
transplantation was assessed with Trypan Blue to be >95%.
Experimental groups
Two strategies were employed to introduce cultured microglia into the
spinal cords, each of which utilized a gelfoam (Upjohn Co.; Kalamazoo, MI)
matrix that could be impregnated with enriched microglia. One method was
the implantation of gelfoam alone (GF), and the other presented the matrix
within biodegradable polymeric tubes (GF+PT) to provide a potential conduit
through which neurites could grow. Control implants which consisted of
cell-free (i.e., medium-only) gelfoam (GF-DMEM) paralleled these two
experimental groups.. The microglia-impregnated gelfoam grafts (GF-M-
DBM; gelfoam and microglia-derived brain macrophages) were implanted
into injured, adult rat spinal cords for 2 (N=3) and 3 (N=4) weeks with two
control implants at each interval. The gelfoam-filled polymeric tubes
saturated with microglia (GF+PT-M-DBM) were also examined at 2 (N=2) and
3 (N=3) weeks after transplantation. These were paralleled by control tube
implants (GF+PT-DMEM) of 2 (N=3) and 3 (N=2) weeks.
Microglial Impregnation of Gelfoam/Polymeric Implants
Polymeric tubes (Medisorb Technologies International, L.P.; 5050 DL),
consisting of 1:1 polylactic:polyglycolic acid (-3 mm x 1 mm), were UV
sterilized and filled with sterile gelfoam. Placing the tubes upright within a
sterile petri dish, a suspension of microglia (-50,000 cells/pl) was slowly
injected with a sterile 22 gauge Hamilton syringe until the gelfoam core was
saturated. Tubes were kept in this position for several minutes before
inverting them and repeating this procedure. The saturated tubes were then
placed on their sides and immersed in complete medium for subsequent
incubation at 37C (>4 hrs.) before transplantation. Gelfoam-only implants

(~l-mm3) were seeded with microglia similarly and incubated in complete

medium until transplantation. Control, cell-free gelfoam was soaked only in
complete medium in both types of implants.
Surgical Procedures
Adult female (150-300g) Sprague-Dawley rats (N=21) were anesthetized
with Ketamine (90 mg/kg, i.p.) and Xylazine (10 mg/kg, i.p.) before a
laminectomy was performed at vertebral T1l. A midline dural incision was
made for approximately 2-4 mm length. Suction was then used to create a

dorsolateral cavity -3 mm long. After hemostasis was achieved, the
experimental (N=12) or control (N=9) implants were put into the damaged
spinal cord after which the dural incision was sutured with 10-0 Ethicon.
Gelfoam was laid on top of the dura before the muscle and skin openings
were dosed. Bladders were expressed as required.
Tissue Processing
After 2-3 weeks, the rats were overdosed with 4% chloral hydrate before
transcardial perfusion with phosphate-buffered saline (PBS) containing
heparin and sodium nitrate, followed by 4% paraformaldehyde in PBS (PF).
The fixed spinal cords were immediately dissected and post-fixed in 4% PF for
30 min. before transferring them to 30% sucrose/PBS for several days at 40C.
Specimens were then embedded in OCT mounting media, frozen in liquid
nitrogen-cooled isopentane, and stored at -700C until cryosectioning (20-
30am) and mounting onto gelatin-coated slides.
Microglial cultures were maintained either on glass coverslips or in 35
mm plastic petri dishes. Before staining, cultures were washed twice with
PBS and then fixed with 4% PF for 20 min., or acetone for 10 minutes.

Immunofluorescent Cytochemistry
In Situ. Before applying primary antibodies to air-dried sections on
slides, the sections were incubated for 30 min. with PBS containing 3% serum
from the species in which the secondary antibodies were raised. Then,
primary antibodies diluted in PBS containing 1% of the same serum were
applied overnight at 4C [OX-42 (Serotec; 1:200); mouse anti-neurofilament
(NaP4-phosphorylated; 1:400); mouse anti-GFAP (Sigma; 1:400); rabbit anti-
laminin, (Sigma; 1:500)]. After several washes with PBS, secondary antibodies
in PBS were applied for 2 hours at room temperature [biotinylated anti-
mouse/rabbit (Vector; 1:300; 1:500); goat anti-mouse IgG-TRITC (Jackson;
1:400)]. If there was a tertiary antibody [streptavidin-FITC/TRITC (Vector;
1:300)], the procedure was repeated, incubating with this antibody in PBS for at
least 1 hour at room temperature. Control slides were incubated with
PBS/1% serum without primary antibodies.
When all incubations were complete, the slides were washed several
times with PBS and then coverslipped with glycerin/PBS (Citifluor;
Canterbury, U.K.) mounting medium for viewing under a Zeiss microscope
equipped with epifluorescent illumination (Axiophot filters: #01-UV; #09-
FITC; #14-RITC). Hoechst 33342 (Sigma) nuclear dye was added to the
mounting media (5gM) in order to get a better estimate of cellular profiles.
Coverslips were sealed onto the slides with Cutex nail hardener (#01).
In Vitro. After fixation, the dishes/coverslips were washed with PBS
before incubating the cultures with primary antibodies or GSI-B4-FITC lectin
(Sigma; 10 gg/ml) in PBS for 2 hr. at room temperature. The
dishes/coverslips were then washed several times with PBS and incubated
with secondary antibodies in PBS for 2 hrs. The remaining procedures were
done as described above.

Fig. 2-1. Immunofluorescent staining of microglia/brain macrophages in
vitro. A. Acetone-fixed microglia cultured on a glass coverslip (6 DIV) were
stained with OX-42, an antibody recognizing the CR3 complement receptor
expressed on macrophages (X320). B-C represent images of 4% PF-fixed
microglia impregnated in gelfoam (same field viewed through UV and FITC
filters, respectively) (X320). B. Hoechst 33342 dye in the mounting media
demonstrates the viability of the cells within the gelfoam matrix after 7 days
in vitro. C. GSI-B4-FITC lectin stain identifies these cells as microglia.



Characterization of Microglia-Impregnated Gelfoam In Vitro
Isolated microglia were cultured and histochemically identified with
OX-42 antibody against their CR3 receptor (Graeber et al., 1988; Fagan and
Gage, 1990) (Fig. 2-1A) and with GSI-B4 isolectin (Streit et al., 1988; Streit, 1990)
(Fig. 2-1C). They were cultured on pieces of gelfoam to establish that the cells
were adhering and surviving in the matrix (Figs. 2-1B,C). Additionally,
microglia grew well in vitro on the polymeric material comprising the tube
walls and remained attached to it for up to 28 days in vitro (DIV) while the
polymer was undergoing degradation. In retrospect, however, one limitation
of the gelfoam was that uniform impregnation could not be routinely
achieved since cultured cells were not always found in the central regions of
the gelatinous matrices.
For the purposes of this report, the term "microglia-derived brain
macrophage" (M-DBM) is used as a more accurate description of the nature of
cultured microglial cells because they are typically isolated from primary
mixed brain cultures as brain macrophages. These cells form in response to

the tissue debris generated during the culture preparation. Thus, using the in
vivo terminology, purified microglia in vitro are microglia which have
already undergone activation to the macrophage state (Streit, 1993).
Neuritic Ingrowth and OX-42-IR Distribution in
GF-DMEM and GF+PT Control Implants
Immunostaining for the phosphorylated, heavy subunit of
neurofilament (NaP4) was used to demonstrate the extension of neuritic
processes into M-DBM-seeded or control GF matrices. Immunoreactivity
(NF-IR) was clearly seen in axons of surrounding host white and gray matter,

Fig. 2-2. Control, cell-free gelfoam as observed 3 weeks after implantation. A.
Horizontal section through a spinal cord implanted with gelfoam alone
shows scant inflammation and sparse OX-42-IR in the matrix (X40). B. Anti-
neurofilament staining of the same implant (inset in A) at the gelfoam-host
border (arrowheads) shows scant neurofilament staining (arrows), despite
host cellular infiltration (X120). C. Adjacent section to A, showing that
laminin-IR is primarily restricted to the host cord surrounding the implant
and to the periphery of the gelfoam.



Fig. 2-3. Control biodegradable polymeric implants filled with gelfoam soaked
in culture medium 2-3 weeks after implantation. A. Hoechst 33342 staining
of a horizontal section through the lumen of a gelfoam-only tube
demonstrates the accumulation of cells around the implant, especially along
the walls and within the rostral and caudal parts of the matrix after 2 weeks
(X40). B. An adjacent section to A immunostained with OX-42 antibody.
Numerous immunoreactive elements are seen along the tube walls and a
prominent infiltration of OX-42-IR cells is indicated in the matrix. C. A
section at the rostral tube implant-spinal cord interface (arrowheads outline
implant walls) stained with neurofilament antiserum after 3 weeks (X80).
Some neuritic processes (arrows) are seen extending into the gelfoam matrix
which was shown by Hoechst 33342 dye and OX-42 staining to be heavily
infiltrated with host cells. Note that while the wall thickness appears much
less than in A and B, this is due to a cryosectioning artifact in which the
integrity of the walls in many of the implants cracked and separated (AB)
after thawing onto gelatin-coated slides.



and in the soma of neurons surrounding the implants which were perhaps

damaged during transplantation (Drager and Hofbauer, 1984).
None of the GF-DMEM implants had exhibited prominent neuritic
ingrowth. Fibers were seen in the host tissue juxtaposed to the implants, but
few neurites were seen in the peripheral margins of the matrices (Fig. 2-2C).
After two and three weeks, OX-42-IR elements were primarily restricted to
cellular infiltrates at the periphery of these implants, and a more pronounced
occupation of the graft matrix, which included other unidentified cells
(Hoechst dye), was apparent after three weeks (Fig. 2-2A).
Control GF+PT implants, however, presented a more complex result in
that a modest degree of neuritic ingrowth was seen in some cases. When
present, these NF-IR profiles were primarily restricted to the rostral and
caudal ends of the polymeric tubes (Fig. 2-3C). By two weeks, the implants
were heavily infiltrated by host-derived cellular elements, many of which
exhibited OX-42-IR (Figs. 2-3A,B). These infiltrates completely filled the tubes
after three weeks, and such cell-rich control implants showed the most robust
ingrowth of neurofilament-IR processes that were usually seen in
conjunction with OX-42-IR cells.
Neuritic Ingrowth and OX-42-IR Distribution in
GF and GF+PT Implants Seeded with M-DBM
Unlike the control gelfoam implants, the matrices impregnated with
cultured microglia displayed a relatively homogeneous distribution of OX-42-
IR cells throughout the matrix after two and three weeks, except at the center
of the implants which tended to be relatively refractory to the cell seeding
procedure (Fig. 2-4A). The OX-42-IR profiles within the matrix had a unique,
swirling pattern reflecting M-DBM aggregates formed during the seeding

Fig. 2-4. Intraspinal implants of gelfoam impregnated with cultured
microglia examined 2 weeks after grafting. A. Horizontal section through an
implant demonstrating OX-42-IR elements dispersed throughout the matrix
(X40). B. An adjacent section to A demonstrating the lack of GFAP-IR cells in
an implant; however, the gelfoam insert was completely lined by a wall of
GFAP-IR processes (arrows). C and D. Microglial-impregnated gelfoam
implants double-stained with anti-OX-42 (Q) and anti-laminin (D) antibodies.
While some of the laminin-IR appeared to be associated with microvascular
elements in the surrounding host tissue, there was a very similar distribution
between the immunoreactive profiles within the microglia-impreganted
matrices (arrows) (X120).


Fig. 2-5. Higher magnification of the graft-host border in Fig. 2-5 (C,D)
demonstrating the sparse, but consistent colocalization (arrows) of OX-42-IR
cells (A) with laminin-IR elements within the matrix (B). Note the lack of
OX-42-IR in the host vasculature (asterisk), serving as an internal control. C.
Adjacent section demonstrates pronounced neuritic ingrowth (arrows) into
the microglia-impregnated gelfoam (X320).

procedure. Neuritic ingrowth was seen throughout these impregnated

matrices in register, with many of these aggregates (Fig. 2-5C).
After two and three weeks, GF+PT implants seeded with M-DBM also

showed homogeneous distributions of OX-42-IR cells throughout their
matrices (Fig. 2-7A), and numerous NF-IR profiles were seen crossing at the
rostral and caudal host-tube interfaces (Figs. 2-6B, 2-7B). This neuritic
ingrowth was consistently more robust than that seen in any of the control
GF+PT implants, even after three weeks. However, while the NF-IR
processes were found in all orientations throughout the matrices (Fig. 2-7B),
some were seen coursing along the inner walls of the tube (Fig. 2-6B), similar
to those seen in control GF+PT implants with heavy infiltrates.
Surrounding Host Inflammatory Responses and NF-IR Distributions
Nissl stains and Hoechst 33342 dye revealed that implants of gelfoam,
with or without seeded M-DBM, induced relatively little inflammation. In
striking contrast, all the GF+PT implants demonstrated a robust host
inflammatory response characterized by a multilayered disposition of cells
surrounding the exterior of each tube that could be revealed by OX-42
immunostaining (Figs. 2-3A,B). Many of these cells also appeared to stream
into the gelfoam matrix at the open ends of the tubes, and they lined the
inner walls as well. Interestingly, NF-IR profiles were often seen coursing
through regions of inflammation in the surrounding tissue that were
partially characterized by OX-42-IR cells (Figs. 2-8A-C).
Astroglial and Laminin Distributions
A narrow, but well-defined, zone of astroglial cells and processes,
denoted by GFAP-IR, was seen encapsulating the control GF-DMEM implants;
however, no GFAP-IR elements were extended into the matrix. Anti-
laminin immunohistochemistry revealed characteristic staining of blood

Fig. 2-6. Horizontal sections of polymeric tube implants which contained
microglial-impregnated gelfoam as seen 2 weeks after being implanted into
the injured rat spinal cord (arrowheads indicate inner tube walls). A.
Photograph of GF+PT implant impregnated with prelabeled microglia
(Hoechst 33342) that are seen scattered throughout the matrix (white dots).
Adjacent sections stained with GSI-B4 lectin confirmed that the nuclear
profiles were virtually all microglia (not shown). B. The same field shown in
A that was immunostained with anti-neurofilament (NF). Arrows indicate
NF-IR profiles that were seen extending into the gelfoam interior at the
implant-host interface (X60). C. Higher magnification of graft-host border in
an adjacent section demonstrates few GFAP-IR processes penetrating the
gelfoam (X80), and no GFAP-IR elements were found in the matrices or
juxtaposed to the exterior wall surfaces.

Fig. 2-7. Shown are adjacent horizontal sections through a polymeric tube
implant containing gelfoam impregnated with cultured microglia (3 weeks
after implantation). A. Intense OX-42-IR is seen throughout the microglia-
impregnated gelfoam matrix (X50). B. Anti-neurofilament staining showed
pronounced neuritic ingrowth from both rostral and caudal segments of the
spinal cord and numerous neuritic processes coursing along the outer walls
of the degrading tube (arrows).


-4 -iti

vessels and subpial surfaces in the spinal cord, and intense laminin-IR also

was seen immediately surrounding the gelfoam (Fig. 2-2C). Such
immunoreactivity, however, was scarcely detected in the interior of these
implants. Similarly, GFAP-IR was exclusively restricted to the host spinal

cord surrounding the control GF+PT implants (Fig. 2-6C). These astroglia
were separated from the exterior surfaces by OX-42-IR cells intrinsic to the
inflammatory responses around these tubes as noted above.
Similar to control GF-DMEM, the M-DBM-seeded GF implants were
also surrounded by a wall of astroglial cells and their processes, and no
GFAP-IR elements were seen within these matrices (Fig. 2-4B). In contrast,
however, to control implants intense laminin-IR was seen throughout these
grafts in register with many of the OX-42-IR cellular profiles (Figs. 2-4C,D).
While some of the laminin-IR appeared to be associated with microvascular
elements, higher magnification revealed some colocalization of OX-42-IR
cells with laminin-IR elements (Figs. 2-5A,B).


Permissive vs. Non-Permissive Effects of Microglia on Axonal Elongation
This study represents an initial investigation of the influences of
microglia-derived brain macrophages (herein referred to as microglia for
simplicity) on axonal elongation in the injured spinal cord using a
transplantation approach in which cells isolated in vitro were introduced into
lesioned cavities in situ. The hypothetical framework of this study is based
on the contrasting effects that microglia have been proposed to have in the
injured CNS. In particular, the interest was in determining whether
microglia would adversely affect axonal elongation based on the speculation

Fig. 2-8. Characterization of inflammatory regions in longitudinal, horizontal
sections through spinal cords implanted with control tubes after 2 weeks. A.
Neurofilament staining was seen along the outer walls of the degrading
polymeric tubes (arrows) that were marked by intense OX-42-IR (see Fig. 2-3b),
and virtually no neurites were seen entering the Gelfoam (X120). B. An area
of inflammation (OX-42-IR) in the dorsal columns rostral to the tube
(arrowheads) demonstrates the presence of neurofilament staining (K within
this region in an adjacent section (X20).


of their putative neurotoxic effects (Colton & Gilbert, 1987; Piani et al., 1991;
Chao et al., 1992; Giulian et al., 1993a,b; Banati et al., 1993). In principle, any
substantial degree of neuritic elongation in the presence of these cells would
constitute a challenge to this hypothesis.
In that regard, it was found first that a qualitatively greater ingrowth of
NF-IR profiles was seen in gelfoam matrices that had been seeded with
microglia prior to grafting. This occurred irrespective of whether or not this
material was partially encased in biodegradable polymeric tubes. Some
neuritic elongation also was observed, unexpectedly, into control gelfoam
(i.e., unseeded) when placed into the spinal cord in conjunction with the

polymer casing. The tubular prosthetic was used to provide a more
structured environment for analyzing axonal growth patterns analogous to
what has been done in studies of other grafted cell types into the injured CNS
(Kromer and Cornbrooks, 1985; Hoffman et al., 1993; Xu et al., 1995).
However, the extent of NF-IR elements entering some control GF+PT
implants was modest compared to the cell-seeded counterparts, and the

distribution of profiles was usually restricted to the open ends of the tubes
infiltrated by host elements. Interestingly, many NF-IR fibers appeared to
coincide with the presence of OX-42-IR cells. It is tempting to speculate,
therefore, that this result indicates endogenous, activated microglia also
might be permissive to axonal elongation. Thus, the results obtained with

cell-seeded matrices are not necessarily attributable to a unique property of
microglia in vitro.
This is further borne out by the observation of NF-IR profiles in
regions of dense OX-42-IR in the spinal cord immediately surrounding the
polymeric tubes. The accumulation of these cells most likely reflects an
inflammatory response elicited by the degradation of the tubes; no similar

inflammation was observed with any of the GF implants. It has been shown
that changes in axonal growth-promoting properties of the CNS near lesion
sites may be produced by brain macrophages that modify nonpermissive
substrata (David et al., 1990). Thus, the indication of axonal elongation
around the degrading tubes would be consistent with this view; however,
further studies are necessary to demonstrate the specific temporal
relationships between neuritic elongation and the establishment of the OX-
42-IR cell microenvironment within the degrading polymeric tubes
(Rabchevsky and Reier, in progress).
Distribution of GFAP- and Laminin-Immunoreactive Elements

One of the hallmark features of CNS injury is the re-establishment of
an astroglial limiting membrane along regions of white and gray matter that
become exposed to non-CNS tissue microenvironments. The magnitude of
this glial response can vary relative to a variety of lesion conditions (Feringa
et al., 1980; 1984; Liesi et al., 1984; Bernstein et al., 1985). Nevertheless, even a
single layer of astrocytes has been thought to be sufficient enough to impede
axonal elongation (Smith et al., 1986; McKeon et al., 1991). In the present
study, a well-defined region of glial reactivity was indicated by GFAP
immunostaining that essentially outlined the border of these implants. In
the case of GF grafts, the glial zone was in proximity to the border of the
matrix, whereas in the case of the GF+PT implants, regions of gliosis were

separated from the implants by OX-42-IR cells.
Given the presence of astroglial reactivity in these transplant
paradigms, it is of interest that evidence for neuritic ingrowth was observed.
This leads to the possibility that microglia may modify either the geometry of
a glial scar or certain properties of astroglia that are incompatible with axonal
elongation (Smith et al., 1986; McKeon et al., 1991). Again, speculation needs

to be tempered since it cannot be determined from these studies whether fiber
extension into the gelfoam matrices preceded development of the astroglial
wall. Also, ultrastructural studies would serve to define the degree of
integrity that such glial regions actually exhibit. Finally, it would be of
interest to determine whether ingrowth continues at later time-points than

examined in the present investigation.
There are many conceivable ways whereby astroglial responses could be
modified by activated microglia. For example, it has been suggested that these

cells may regulate astroglial proliferation and scar formation through the
production of IL-1 and TGF-01 (Giulian et al., 1986; Lindholm et al., 1992;
Sievers et al., 1993). Another way could be related to the deposition of basal
lamina (BL) which, in part, could be examined by way of laminin-IR. After
CNS injury, a BL is formed along exposed regions of the brain or spinal cord
which essentially mirrors the distribution of GFAP-IR processes (Matthews
and Gelderd, 1979; Feringa et al., 1980; Liesi et al., 1984; Kromer and
Cornbrooks, 1985). This BL, associated with the subpial endfeet of astrocytes,
may interfere with axonal growth through a lesion either by being a physical
barrier or conversely a favorable substratum to which axons prefer to remain
adherent (Reier et al., 1983; 1988). It is possible that the astrocytes, developing
into scar tissue, do not represent a physical barrier but a poor substratum
because of a deficiency in surface molecules, ECM molecules, or soluble
substances that must be provided to their external milieu for regeneration to
occur (Schwartz et al., 1989). It thus becomes of interest that all control GF
and GF+PT implants did not show laminin-IR in the matrix environment,
whereas there was a pronounced distribution of laminin in matrices that
were originally seeded with microglia in vitro. Thus, part of the neuritic
extension into the gelfoam matrices may be due to the deposition of this ECM

molecule known to have pronounced neurite growth-promoting properties
in vitro (Rogers et al., 1983; Lander et al., 1985).
Potential Growth-Promoting Influences of Other Cell/Tissue Types
The presence of laminin-IR in cell-seeded or host microglia infiltrated
gelfoam implants raises the possibility that the ingrowth of NF-IR elements
may not be directly related to axon-microglial interactions but, instead, to a
microglial-induced infiltration of other cell types that have growth-
promoting properties. While there were some examples of colocalization of
OX-42- and laminin-IR, this was not of a magnitude sufficient enough to
account for the degree of laminin-IR observed in the grafts. Furthermore, it
has been reported that laminin is not produced by microglia grown in vitro
(Chamak et al., 1994). Likewise, it is unlikely that laminin deposition
reflected some redistribution of mature astroglia since no GFAP-IR elements
were observed in any of the grafted matrices. Because this colocalization was
unexpected, especially in the absence of any GFAP-IR cells which are known
to express laminin, further controls may be necessary to repeat these results
and to possibly identify the isoforms expressed (Liesi and Risteli, 1989).
By exclusion, endothelial and Schwann cells represent the most likely
laminin-producing cell types that could be recruited from the host into the
gelfoam matrices. Whether microglia exert an angiogenic effect as do their
peripheral counterparts (Mustoe et al., 1987) is presently unknown. However,
because no distinction can be made in culture between parenchymal and
perivascular microglia, many of the OX-42-IR elements in the grafts may have
been of a perivascular nature. However, while laminin immunostaining
revealed distinct vascular elements in the host spinal cord, the pattern of
laminin-IR in the gelfoam matrices was not as obvious. The ingrowth of
vascular elements, however, remains a tangible possibility, and this cellular

terrain could also serve as one route of Schwann cell entry. This emphasis on
laminin, however, does not exclude the possible involvement of other non-
laminin-producing elements with neurite growth-promoting properties (e.g.,
immature mesenchymal cells) or the deposition of other ECM molecules.

Closing Comment
The present transplantation study has provided some examples of
neuritic elongation in the presence of microglia/brain macrophages of both

donor and host origin. While the specific role of these cells in promoting
axonal elongation cannot be ascertained from the circumstantial
observations, these initial findings argue against the proposed negative effects
of microglia in CNS injury. In addition, these observations suggest that
microglia may be instrumental in mobilizing cells that have more well-
defined neurite growth-promoting properties. In that case, the continued
expression of growth-promoting properties is a further indication that
microglia do not cancel or suppress these effects.



Axonal regeneration in the adult mammalian central nervous system
(CNS) has long been viewed as being an abortive process (Ramon y Cajal,
1928). Over the years, this failure of regrowth has been attributed to a variety
of mechanisms (McKeon et al., 1991; Reier et al., 1983; 1988; Schwab, 1990).
Two of the more prominent, and not necessarily mutually exclusive, views is
that the CNS either lacks cell types that can exert either neurotrophic or
neurotropic influences or that the cellular composition of the CNS is non-

permissive to axonal elongation. The fact that neurons intrinsic to various
regions of the brain, spinal cord, or retina have the inherent ability to regrow

damaged processes is now well recognized as the result of peripheral nervous
system (PNS)-to-CNS grafting experiments (Aguayo et al., 1982; David and
Aguayo, 1981; Richardson et al., 1980). This effect of PNS tissue has been
ascribed primarily to the Schwann cell. More recent emphasis has been on
the use of purified Schwann cell populations to promote axonal regeneration
in the injured spinal cord (Paino and Bunge, 1991; Paino et al., 1994).
Thus, neuritic elongation can be evoked when the cellular
microenvironment of the CNS is replaced by Schwann cells and other
cellular components of the PNS. One CNS component that has been
traditionally viewed as being incompatible with regeneration is the astrocyte

(Reier et al., 1983; 1988). Though the adverse effects of this cell on axonal

outgrowth is still not universally accepted, there are several lines of evidence
showing poor fiber extension in an astrogliotic environment both in situ and
in vitro (Reier et al., 1983; 1988; Liuzzi and Lasek, 1987; Fawcett et al., 1989).
Oligodendrocytes and the white matter which these cells produce represent
another collective constituent that can render the CNS nonpermissive to
regeneration (Schnell and Schwab, 1990; Schwab, 1990).
Another major cellular component of the CNS is the microglial cell;
however, its role in relation to neuritic outgrowth has not been extensively
studied. In this respect, differing views have emerged from in vivo and
tissue culture studies to the extent that one perspective is that these cells may
actually be compatible with fiber outgrowth (Nakajima et al., 1989; 1993;
Nagata et al., 1993a,b; Streit, 1993; Zhang & Federoff, 1993; Chamak et al., 1994),
whereas the other notion is that these cells may produce substances that can
be toxic to neurons (Colton & Gilbert, 1987; Giulian, 1990; Piani et al., 1991;
Chao et al., 1992; Banati et al., 1993). This issue was addressed in the preceding
chapter by virtue of an intraspinal grafting approach in which microglia-
derived brain macrophages were isolated in vitro and then subsequently
seeded into a gelfoam matrix that was placed into intraspinal lesion cavities
either alone or partially invested by a biodegradable polymeric tube. The
findings of that investigation suggested that some neuritic growth had
occurred within the gelfoam matrices thereby challenging the proposed
neurotoxic impact of these cells.
It was unclear from that initial set of experiments, however, whether
grafted microglia had a direct effect on axonal elongation. They potentially
could have had a secondary influence by mobilizing other host cell or tissue
types with demonstrated neurite growth-promoting properties. For example,

it is conceivable that an influx of Schwann cells or vascular elements, as well
as some appropriate ECM molecules (i.e., laminin) may have occurred.
Some of the interpretational difficulty could be attributed to the
biodegradable polymeric tubes that appeared to spawn a surrounding, robust
inflammatory response. To control for this complication more rigorously,
the present study was carried out using Gore-texTM implants constructed of
modified Teflon, polytetrafluoroethylene (PTFE). Such PTFE tubes have been
successfully used in PNS regeneration studies and only elicit a minimal
surrounding tissue reaction (Valentini et al., 1989; Young et al., 1984).
In view of the suggested growth-promoting effects of microglia seen in
the previous experiments, another objective of the present study was to
determine how co-grafts of microglia and astrocytes would affect neuritic
elongation. Additionally, these experiments included PTFE tube implants
that were impregnated with cultured microglia stimulated with
lipopolysaccharide (LPS) prior to transplantation. Reasons for including the
LPS-activated microglia group were based on the neuritic growth seen in
areas of endogenous microglia/brain macrophage activation in the previous
experiments, as well as the possibility that cultured microglia, although
already in a macrophage state, may not be activated to secrete factors
influencing cell survival and/or differentiation. In addition, it may
determine whether stimulated, cultured microglia produce a neurite-
inhibitory effect in vivo. Portions of this study have been previously
summarized (Rabchevsky et al., 1994; 1995).

Methods and Materials

Isolation of Cultured Glial Cells
After several weeks of isolating microglia as described previously (see
Materials and Methods, Chapter 2), 12 mM lidocaine was added to complete
medium (DMEM-10% FBS) and the cultures were put on an orbital shaker set
at 500 r.p.m. for 10 minutes to remove the majority of microglia from the
underlying astrocyte layer (Nakajima et al., 1989). The supernatants were
collected for microglial isolation, the cultures were again fed with complete
medium, and the flasks were put on an orbital shaker set at 250 r.p.m. for 14
hours (370C). The floating cells were removed and the flasks once again fed
with complete medium containing 5mM EDTA for 10 minutes while slowly
agitating. The supernatants were aspirated and the flasks rinsed several times
before DMEM containing 0.15% trypsin was added to remove adherent
astrocytes. The cells were collected and an equal volume of complete
medium was added to quench enzymatic activity. The astroglial suspensions
were pelleted and resuspended in complete medium for impregnation of
PTFE tubes. The mixed glial suspensions were made by combining
resuspended microglial and astrocyte pellets. Before and after
transplantation, samples of the suspensions were plated and cultured for
immunocytochemical characterization.
Activation of Cultured Microglia
Microglia were isolated and grown in complete medium before
stimulating them with bacterial lipopolysaccharide (LPS) which was chosen to
activate microglia because it does not induce them to secrete neurotoxinss'
(Giulian et al. 1993a) yet causes them to release cytokines (Giulian et al., 1986).
The chosen measure of activation was the production of IL-1 mRNA in

stimulated versus non-stimulated cultures following the protocol of Hetier et
al. (1988). Confluent cultures of microglia were grown in 60 mm plastic petri
dishes containing complete medium for several days. The cells were then fed
with fresh medium containing 2 or 10 gg/ml LPS and exposed for 4 or 24 hrs.
Then, the control and stimulated microglia were removed from the petri
dishes by replacing the LPS medium with complete medium containing 12
mM lidocaine followed by vigorous agitation (500 rpm) for several minutes
(Nakajima et al., 1989). Nearly all adherent microglia were removed.
Supernatants were collected and placed into conical centrifuge tubes
containing an equal amount of complete media before pelleting (400 x g, 10
min.). The microglia were resuspended in PBS and added to 1.5 ml
Eppendorf tubes (~2 x 106 cells/ml) before pelleting (2000 x g, 5 min.). They
were then stored at -700C until RNA extraction. The extraction of total RNA
and the Northern Blot analysis were performed by Sharon Walter of Dr.
Wolfgang Streit's laboratory in the University of Florida Neuroscience
Impregnation of Gelfoam-PTFE Tube Implants
Polytetrafluoroethylene (PTFE) tubes (Gore Technologies; Flagstaff, AZ)
measuring ~3 mm x 1.5 mm (inner diameter) were UV sterilized and filled
with sterile gelfoam. The porosity of the PTFE channel walls was 30 Arn and
the wall thickness was ~ 500 gm. After placing the tubes upright within a
sterile petri dish, a concentrated suspension of cultured glial cells (-50,000
cells/gl) was slowly injected with a sterile 22 gauge Hamilton syringe to
saturate the gelfoam-filled lumen. The tubes were kept in this position for
several minutes before inverting them and repeating this procedure. The
saturated tubes were then placed on their sides and immersed in complete

media for subsequent incubation before transplantation (4-48 hrs.). Control
PTFE implants contained gelfoam matrices soaked only in complete medium.
Impregnation of LPS-Stimulated Microglia
PTFE tubes were saturated with cultured microglia and then stimulated
with LPS prior to transplantation as mentioned above. After allowing the
cells to adhere to the gelfoam and the surrounding petri dish surface (>60
min.), they were incubated in complete medium containing 2 jig/ml of LPS.
The cells were left at 37C until implantation 4-12 hrs. later at which time the
LPS-containing medium was replaced with fresh complete medium.
Surgical Procedures
Control and experimental animal groups are outlined in Table 3-1.
Adult female (150-300g) Sprague-Dawley rats (N=51) were anesthetized
(Ketamine: Xylazine-90:10 mg/kg, I.P.) before a laminectomy was performed
at vertebral T10. A midline dural incision was made for approximately 3-5
mm length. A suction micropipet was then used to create a dorsolateral
cavity ~4 mm long. Hemostasis was achieved by replacement of dry gelfoam
pieces into the cavity. Implants were then put into the damaged spinal cord
after which the dura was sewn together with 10-0 Ethicon suture. A piece of
gelfoam was laid on top of the dura before the muscles and skin were sutured.
Bladders were expressed as required.
Tissue Processing
After survival times of 2, 3 or 5 weeks, the rats were overdosed with
4% chloral hydrate before transcardial perfusion with phosphate-buffered
saline (PBS) containing heparin and sodium nitrate, followed by 4%
paraformaldehyde in PBS (PF). For plastic semi-thin sectioning, some of the 5
week survivors were perfused with a double aldehyde solution containing
5.0% glutaraldehyde and 4.0% paraformaldehyde in 0.1 M PBS.




(N=13) (N=16) (N=8) (N=4) (N=5)

Table 3-1. The animals were divided into 4 experimental groups and 1 control
group. The control group received PTFE tube implants containing gelfoam
soaked only in culture medium. The other 4 groups each received implants
containing matrices seeded with microglia and/or astrocytes derived from
primary cultures of rat CNS. The LPS-microglia group consisted of seeded
implants which were exposed to LPS prior to implantation in order to "activate"
the grafted cells. The mixed glial implants served to characterize the effects that
astroglia may have on putative growth-promoting properties of microglia.
Animals were sacrificed 5 weeks after implantation and perfused with 4%
paraformaldehyde. Cryosections (-10 mm) were immunohistochemically
examined for cellular and neuritic profiles within and around the implants.
Some animals were perfused with a dual aldehyde fixative for analysis of plastic
embedded PTFE tubes (Group I, N=3; II, N=4; V, N=1).

The spinal cords which were perfused with 4% PF were immediately
dissected and post-fixed in 4% PF for 30 min. before putting them in 30%
sucrose/PBS for several days at 40C. Specimens were then embedded in OCT
mounting media and frozen in liquid nitrogen-cooled isopentane. They were
stored at -700C until cryosectioning (10-15 gim) and mounting them onto
gelatin-coated slides that were again stored at -700C until

immunohistochemistry was performed.
The spinal cords which were perfused for plastic embedding were
dissected the day after perfusion and post-fixed in the same solution for
several days. The segments containing the tubes were then divided into two
pieces by a transverse section midway through the implant, and the tissue
blocks were osmicated, dehydrated, and embedded in Epon for transverse

sections (2 gm).
Microglia isolated from mixed glial cultures were maintained either on
glass coverslips or in 35 mm plastic petri dishes. Before immunostaining,
cultures were washed twice with PBS and then fixed with 4% PF for 20 min.,

or acetone for 10 minutes.
Immunofluorescent Cytochemistry
In Situ. Before applying primary antibodies to air-dried sections on
slides, the sections were incubated for 30 min. with PBS containing 3% serum
from the species in which the secondary antibodies were raised. Then,
primary antibodies diluted in PBS containing 1% of the same serum were
applied overnight at 40C [OX-42 (Serotec; 1:200); mouse anti-neurofilament
(NaP4-phosphorylated; 1:400); mouse anti-GFAP (Sigma; 1:400); rabbit anti-
laminin, CGRP, 5-HT (Sigma; 1:500)]. After several washes with PBS,
secondary antibodies in PBS were applied for 2 hours at room temperature
[biotinylated anti-mouse/rabbit (Vector; 1:300; 1:500); goat anti-mouse IgG-

TRITC (Jackson; 1:400)]. If there was a tertiary antibody [streptavidin-
FITC/TRITC (Vector; 1:300)], the procedure was repeated, incubating with this
antibody in PBS for at least 1 hour at room temperature. Control slides were
incubated with PBS/1% serum without primary antibodies.
When all incubations were complete, the slides were washed several
times with PBS and then coverslipped with glycerin/PBS (Citifluor;
Canterbury, U.K.) mounting medium for viewing under a Zeiss microscope
equipped with epifluorescent illumination (Axiophot filters: #01-UV; #09-
FITC; #14-RITC). Hoechst 33342 (Sigma) nuclear dye was added to the
mounting media (5pM) in order to get a better estimate of cellular profiles.
Coverslips were sealed onto the slides with Cutex nail hardener (#01).
In Vitro. After fixation, the dishes/coverslips were washed with PBS
before incubating the cultures with primary antibodies or GSI-B4-FITC lectin
(Sigma; 10 gg/ml) in PBS for 2 hr. at room temperature. The
dishes/coverslips were then washed several times with PBS and incubated
with secondary antibodies in PBS for 2 hrs. The remaining procedures were
done as described above.
Ouantitative Analysis of Immunoreactive Profiles in Implants
To assess the content and/or infiltration of immunoreactive elements
in the PTFE tubes, three slides (ea. containing ~6 cryosections) from all
implants were stained with OX-42, laminin (LAM) and/or neurofilament
(NF) antibodies. The tubes, which were all ~ 3 mm in length, were examined
microscopically and divided into rostral (R) and caudal (C) segments. Then,
the R and C portions of the tubes were again divided into 4 ordinal modes
that were used to rank the content of immunoreactive profiles seen at both
ends of the tubes. An example of this method for ordinal ranking is



ROSTRAL (r) 0 1 2 3 3 2 1 0 CAUDAL (c)

rLAM=2 cLAM=1


"Kruskal-Wallis" anova Test by Ranks

rOX-42 H (4,N=579) = 110.4606 P < 0.0001

cOX-42 H (4,N=579) = 219.4768 P < 0.0001

rLAM H (4,N=567) = 85.99577 P < 0.0001

cLAM H (4,N=567) = 220.6777 P < 0.0001

rNF H (4,N=559) = 95.73747 P < 0.0001

cNF H (4,N=559) = 180.9473 P < 0.0001
Test does not support the "Null Hypothesis" that these experimental groups are from the
same population (Statistica by StatSoft).

Fig. 3-1. These schematics represent an example of the method used to assign
ordinal rankings to the immunohistochemical profiles observed in each stained
cryosection from the PTFE implants. The ranks were determined by establishing
the existence of at least three immunoreactive profiles (r and c) seen within an
ordinal mode (0-3) for each stain. After compiling all the ordinal data, shown
schematically in the upper-most figure, a "Kruskal-Wallis" a nova test by ranks
was performed to establish whether these data were from the same experimental

illustrated in Fig. 3-1. After the ordinal data were collected from all R and C
segments of each implant, the Kruskal-Wallis a nova test by ranks
determined that the independent samples (i.e., rOX-42, cOX-42, rLAM, cLAM,
rNF, cNF) were from different populations (Fig. 3-1). After summarizing the
ordinal data for each variable within the implants, the Mann-Whitney U test
was performed to determine the significant differences among these various


General Features of PTFE Implants
As noted earlier, in our previous study of intraspinal microglial
implants (Chapter 2) we had used a biodegradable polymer tube as a way to
provide a more structured orientation of the biomatrix (i.e., cell-seeded
gelfoam). A complication of those tubes was that they induced a robust
inflammation probably as a consequence of degradation; this was not the case
with the PTFE tubes that were used in the present study. Neither at 2, 3 or 5
weeks post-implantation was there evidence of either profuse cellular
accumulations along the exterior surface of these tubes (Fig. 3-2A,B) or
linearly oriented neurofilament immunoreactive (NF-IR) profiles coursing
alongside (see Chapter 2). While some mononuclear cells were seen in this
region, the density of these cells did not appear to be strikingly greater than
that of more distant host neuropil. Neurons in the immediate vicinity of
these implants appeared unaffected by the presence of this prosthetic material
(Fig. 3-2B), although positive staining for phosphorylated neurofilament
indicated some form of perturbation (Fig. 3-2C). Ramified microglia were

Fig. 3-2. Cytological features of host tissue surrounding the PTFE tube
implants. These photomicrographs demonstrate the cellular elements
immediately surrounding the implants after 5 weeks. AB. Nissl stained
preparations show the relative lack of inflammation induced by the PTFE
implants. A. Microglia-impregnated PTFE implant seen in longitudinal
section demonstrating an example of the variability in the impregnation of
gelfoam matrices (X40). B. At higher magnification note that despite the
presence of some inflammatory cells along edges of the tube, viable
motoneurons are seen immediately adjacent to the wall (X120). C. Higher
magnification of neurons immunofluorescently stained with a
neurofilament antibody (RITC) found immediately adjacent to the tube walls
(marked by arrowheads) (X320). D. Same field of view as B seen through an
FITC filter to demonstrate perineuronal positioning of OX-42-IR microglia,
some ensheathing apparently healthy neurons (arrows).



it :.7t

. AN

I'L tI

Fig. 3-3. Characterization of PTFE tubes containing gelfoam saturated with
cultured microglia prelableled with DiI-ac-LDL, 3 weeks post-implantation.
AB are the same field of view seen through UV and RITC fluorescent filters,
respectively, showing the distribution of cellular elements in these implants.
The top of each photograph represents the PTFE walls juxtaposed to host
spinal cord tissue while the gelfoam core is seen in the lower half of each
frame. A. Nuclear profile (Hoechst 33342) demonstrates the distribution of
cells in the matrix and that many cellular elements entered the wall channels,
especially from the exterior surface (X80). B. Scattered DiI-ac-LDL-labeled
microglia are seen throughout the matrix. Some elements in the Gelfoam
were not labeled, suggesting that these may be derived from the host. Note
that while labeled cells do line the interior surfaces of the tube walls, few were
seen entering the porous channels to exit the matrix.

49, ~

.. 'I V

found in the spinal gray matter surrounding these cells (Fig. 3-2D), and while
some appeared activated (i.e., hypertrophied) their density and distribution
did not appear strikingly different from normal parenchyma.
Because of the 30 gm porosity of the tube walls, many cellular elements
from the surrounding spinal cord were observed coursing through the walls
toward the abluminal surface (Fig. 3-3A). To determine whether any of these
migrating cells were grafted cells, some PTFE tubes contained microglia that
were prelabeled with DiI-ac-LDL (N=2). While these implants demonstrated
many cells traversing the tube walls after three weeks, very few showed
positive DiI-ac-LDL labeling (Fig. 3-3B).
Many of the inwardly migrating cells were microglia/brain
macrophages as defined by OX-42-IR (Fig. 3-4A). While a detailed
characterization of the other infiltrating cells was not performed, it was found
that none of the elements within the channel walls exhibited GFAP-IR.
Some regions of intense laminin-IR were seen along the external walls of the
PTFE tubes, and thin extensions of this ECM molecule were often observed
within the walls per se (Fig. 3-4B). Interestingly, there were also instances in
which NF-IR profiles were seen entering these channel walls, many times
paralleling the laminin-IR elements (not shown).
Cellular and Neuritic Composition of Control PTFE Implants
While there was minor surrounding inflammation elicited by PTFE,
the tubes containing cell-free gelfoam soaked in culture medium still
demonstrated an infiltration of host cellular elements after 5 weeks (Figs. 3-
5A, 3-6A). The amount of infiltration and the preferential rostral or caudal
entrance into these tubes varied from animal to animal, and after dissection
no correlation could be made with regard to the tube placement and the

Fig 3-4. High magnifications of the porous PTFE walls demonstrating the
infiltration of cellular elements five weeks after implantation. These
photographs represent adjacent sections showing that while many of the cells
coursing through the channels of the tube walls are OX-42-IR brain
macrophages (A), there was also the presence of laminin-IR profiles within
the lacunae (B). Note the similar distribution of OX-42-IR and laminin-IR
along the outer walls of the tube.

*1 ?I'L

- ^ *

Fig. 3-5. Sections through control PTFE implants after 5 weeks. A. Nissi stain
of a longitudinal, horizontal section through an implant showing an example
of only slight infiltration of host cellular elements beyond the gelfoam-host
border (X80). The host spinal cord is to the left side of the photograph,
demonstrated by the central canal seen at the bottom left. B. Transverse,
plastic section (2 gm) through a control PTFE implant (arrowheads delineate
inner wall) demonstrates vascularization close to the edges of the lumen,
accompanied by occasional Schwann cell-myelinated axons (arrows) in the
presence of mononuclear phagocytes and other cells (X320). CD are higher
magnifications of the graft-host border in A showing the same field of view
seen through different fluorescent filters to demonstrate the coexistence
(arrows) of infiltrating OX-42-IR cells (CD) and neuritic ingrowth (D);
arrowheads demarcate the graft-host border (X120).




incidence of robust cellular infiltration. Semi-thin, plastic sections of the
control PTFE tubes indicated that by 5 weeks some of the gelfoam matrix had
been degraded. At this time, the interior of the PTFE tubes contained
numerous mononuclear phagocytes and connective tissue elements that

formed a loose trabecular lattice that appeared to provide a supporting matrix
for the infiltrating cells (Fig. 3-5B). However, the increased cellularity and
accompanying vascularization of these control implants was predominantly
restricted to the inner walls of the PTFE tubes (Fig. 3-5B).
Immunohistochemical staining of these tubes showed prominent OX-
42-IR presented by the cellular infiltrates (Figs. 3-5C, 3-6B). Staining for
neurofilament revealed a modest ingrowth of neuritic profiles, the
distribution of which was in register with OX-42-IR. A modest degree of
laminin-IR also was noted in some tubes that coincided with OX-42-IR (Figs.
3-6B,C). The majority of laminin-IR, however, was seen at the interface
between host spinal gray/white matter and the gelfoam within the walls of
the tube. This region also exhibited a wall of GFAP-IR processes that to some
extent was in register with the laminin-IR in the host tissue (Figs. 3-6C,D).

A correlate of the NF-IR elements, noted above, was observed in semi-
thin plastic sections. Myelinated axons were seen occurring as small bundles
having a rather random distribution within the tube lumen (Fig. 3-5B) but
never in cell-free regions. At higher magnification (not illustrated), these
myelinated sheaths were seen enveloped in some cases by cytoplasm that
occasionally contained a nuclear profile as is characteristic of axon-Schwann
cell relationships in peripheral nerves.
Microglia-Impregnated PTFE Implants

At 2 and 5 weeks post-implantation, the cell-seeded tubes displayed a
rather dense OX-42-IR at the rostral and caudal parts of the matrices

Fig. 3-6. Sections through a control PTFE tube which demonstrated robust
cellular infiltration seen 5 weeks post-implantation. The top of each
photograph represents the graft (gelfoam)-host border. AB and CD represent
same fields of view, respectively, seen through AMCA (A), FITC (B.D) or
RITC (C) fluorescent filters. A. Prominent infiltration of endogenous cells
was seen in some control implants, most of which were OX-42-IR (.B) (X80). C.
Adjacent section to B demonstrating laminin-IR in register with infiltrating
brain macrophages and other cellular elements (X80). Note that the majority
of staining, however, is confined to the graft-host border and the host spinal
cord. D. In the presence of penetrating laminin-IR elements, no GFAP-IR
cells were ever seen in any of the control implants (arrows indicate astroglial
lining restricted to the host issue).


Fig. 3-7. Sections through PTFE implants impregnated with microglia. A-C
represent sections through one implant examined 2 weeks post-
transplantation. The lateral portions of the photographs represent the walls
of the tube. A. OX-42-IR is demonstrated throughout the matrix, and there is
penetration of neurofilament-IR processes (B) into the microglial
environment (X60). Penetration of laminin-IR elements (C) also paralleled
the neuritic ingrowth (X40). Laminin-IR was even more apparent
throughout the matrices after 5 weeks, seen in a cross section (D) of a
microglia-impregnated tube (X40). Note that the entire implant was invested
by an intense laminin-IR lining.

(Figs. 3-7A, 3-8A). While most tubes had a relatively homogeneous
distribution of OX-42-IR cells, the concentration of such cells in the center of
some tubes was considerably reduced due to difficulties in obtaining complete
cell-seeding routinely (Fig. 3-2A). Anti-neurofilament staining demonstrated
neurites with linear or branched trajectories distributed within the OX-42-IR
domains (Figs. 3-7A,B; 3-9A,B). Qualitatively, the degree of neuritic ingrowth
appeared to be substantially greater than that noted in control tubes, and
neurofilament-IR processes were distributed throughout the matrices in all
By 5 weeks after implantation, transverse sections of the spinal cord
showed that the area occupied by the PTFE tubes essentially represented a
hemisection lesion (Figs. 3-7D, 3-10A). The entire tube was invested by a
laminin-IR lining that was in register with GFAP-IR. However, there was a
cellular boundary between the exterior tube surface and the astrocytic
processes surrounding it that was highly OX-42-IR (Fig. 3-4A). In contrast to
control tubes (see above), a greater degree of laminin-IR also was seen within
the cell-seeded matrices (Figs. 3-7C, 3-8B, 3-9D). Inside the tubes, this ECM
molecule was distributed along the abluminal surface, but this did not
coincide with any GFAP-IR as such was undetectable in the matrices (Fig. 3-
8C). While there were examples of colocalization, the distribution of
laminin-IR did not appear to be in consistent register with neuritic ingrowth
(compare Figs. 3-7B,C to 3-9C,D).
Plastic sections demonstrated a greater cellularity than seen in the
control implants, most notably within the core of the gelfoam (Fig. 3-10A). In
addition to mononuclear phagocytes, there were numerous multinucleated
giant cells in these cell-seeded tubes (Figs. 3-10B,C), along with vascular and
connective tissue elements that appeared to be embedded in a collagenous

Fig. 3-8. Longitudinal, horizontal sections through microglia-impregnated
PTFE tubes studied at 5 weeks post-transplantation. The extreme top and
bottom of the photographs represent the tube walls and the host-graft
(gelfoam) border is to the left A. Microglia-impregnated gelfoam was
inundated with OX-42-IR profiles (X80). However, there were many other
cellular elements in these matrices that were not microglia/brain
macrophages, as revealed by cells (Hoechst 33342) that were not OX-42-IR (not
shown). B.. Adjacent section to A demonstrates that unlike control PTFE
implants, these gelfoam matrices were filled with laminin-IR profiles, many
of which appeared to be associated with microvascular elements. C. Adjacent
section to B demonstrating the lack of any GFAP-IR cells in the matrix but the
presence of an astroglial lining surrounding the implant (arrows).


Fig. 3-9. AB and C,D represent the same fields of view, respectively, seen
through RITC (A,C) and FITC (B,D) fluorescent filters. Higher magnification
of a microglia-saturated gelfoam matrix (Fig. 3-8A) stained with
neurofilament antibody (A) shows neuritic processes coursing through the
OX-42-IR microenvironment (B) (X320). C and D demonstrate the occasional
colocalization of neuritic processes (0) with laminin-IR elements (D) (X320).

ground substance (Fig. 3-10B). As in control implants, Schwann cell
myelinated axons also were seen in plastic sections throughout these cell-
seeded PTFE tubes. These fibers were randomly distributed and did not
appear to be consistently related to any cell or tissue type present in these
grafts (Fig. 3-10B).
PTFE Implants Containing LPS-Activated Microglia

Some PTFE tubes saturated with cultured microglia were incubated in
complete medium containing 2gg/ml LPS for 4-12 hours before implantation.
This incubation period was based on the level of IL-1 mRNA expression
which is a standard measure of microglial activation in vitro (Fig 3-11A). In
addition, inspection of the petri dishes containing the saturated tubes
revealed morphological alterations of microglia adhering to the dishes. This
correlated with LPS-induced levels of IL-1 mRNA expression (Figs. 3-11B,C).
Implants saturated with LPS-stimulated microglia displayed similar
immunohistochemical profiles as the unstimulated microglial tubes after 5
weeks. Thus, a comparable pattern of laminin-IR and NF-IR distribution was
observed in OX-42-IR rich areas of the gelfoam (Figs. 3-12A,B). However, a
notable difference appeared to be a more pronounced infiltration of various
host cellular elements into the LPS-stimulated microglial matrices.
Astrocyte-Seeded PTFE Implants
There were actually two subsets of implants within the astrocyte
implant group; those that were impregnated with a mixed population of
cultured microglia and astrocytes, and those that were impregnated with an
enriched suspension of astrocytes. Cultures of the transplanted cells were
made with the remaining suspensions after surgery in order to
immunocytochemically identify the cellular constituents. While the mixed

Fig. 3-10. Transverse, plastic sections (2 gm) through PTFE tubes impregnated
with microglia 5 weeks after implantation. A. Prominent cellularity is
demonstrated within these implants compared to controls (Fig. 3-5B),
especially within the core of the gelfoam (X10). B and C demonstrate the
increased concentration of macrophages and blood vessels within these
matrices (X320). While there were groups of Schwann cell-myelinated axons
found within regions of macrophage accumulation (arrows), most areas
contained large, multinucleated cells (arrowheads) and microvasculature
associated with mononuclear cells in the absence of myelinated axons (C).


j qi

- ,,'<1 -:". **-qb.

Fig. 3-11. Characterization of cultured microglia stimulated with
lipopolysaccharide (LPS). A. Northern blot demonstrating the expression of
IL-1 mRNA by cultured microglia after stimulation with various
concentrations of LPS. B. The lectin-conjugate GSI-B4-FITC was used to
demonstrate phenotypic changes in microglia that were cultured in the
presence of 2 jig/ml LPS for 4 hours (X200). Note the elongated, highly
branched morphology of these activated microglia as compared to
unstimulated cultured microglia (see Fig. 4-2). C. After 24 hours of exposure
to LPS, microglia had retracted their processes and assumed a more flattened,
ameboid morphology. This may correlate with their decreased IL-1 mRNA
expression (A., lane C) (X320). Note: Sharon Walter performed the RNA
extraction and created the Northern Blot shown.

i 1.9%
A a
s m
.qltfltt dI 1uBmiii~ltl i 0. lM^M--f
-L duI- U i I -. ,i m
TT!!-St v w Jql~mUIg.'lnm m unllw m

glial population contained ~60% microglia (GSI-B4), astrocyte enriched
populations had >80% GFAP-IR cells (Fig. 3-13A-D).
The mixed glial implants demonstrated a robust ingrowth of neurites
and laminin-IR elements similar to microglial grafts. In addition, OX-42-IR
seemed equally as robust as in the microglia-impregnated implants.
Interestingly, however, no GFAP-IR cells were detected within these matrices.

In light of the in vitro characterizations of the graft suspensions (Fig 3-
13C,D), a more unexpected finding was the scant GFAP-IR cells seen within

the enriched astrocyte implants (Fig. 3-14A). Those cells that did express
GFAP were almost exclusively associated with laminin-IR elements, some of
which did not resemble characteristic microvascular constituents (Figs 3-
14B,C). However, many of the GFAP-IR profiles did colocalize with laminin-
IR vessels, especially those near or entering the wall channels (Figs. 3-15AB).
In many regions devoid of GFAP-IR cells, there were areas in the gelfoam that
contained numerous OX-42-IR cells (Fig. 3-15C). Despite the lack of many
GFAP-IR cells in these implants, plastic sections confirmed that they
contained numerous islands of astroglial cells extending processes
throughout the gelfoam matrix, and few neurites were seen in these grafts
(Figs. 3-16B,C).
Interestingly, most of the neuritic growth observed in astrocyte grafts
was into the caudal end of the PTFE tubes. Like all other cell-impregnated
PTFE tubes, the astrocyte implants were penetrated by microvasculature and
other elements, most notably brain macrophages (Figs. 3-16A-C). However,
few Schwann cells were seen in plastic sections, and those observed were
myelinating axons in astrocyte-free regions of gelfoam.

Fig. 3-12. Microglia-impregnated PTFE implants 5 weeks after activation with
LPS and subsequent grafting. The top and bottom of the photographs
represent the lateral margins of the gelfoam matrix, and the host-graft
(gelfoam) border is to the left. A. Sections through the lumens of microglial
implants exposed to LPS before transplantation demonstrated similar
penetration of laminin-IR profiles to that seen in non-stimulated microglial
implants (X80). B. Neurofilament staining of an adjacent section
demonstrates that neuritic penetration was also coincident with the
appearance of laminin-IR elements. However, there seemed to be an overall
increase in the cellularity within these implants compared to non-activated
microglial grafts, as evidenced by the more compact aggregation of nuclear
profiles (Hoechst 33342) and OX-42-IR cells in regions of these matrices.


Fig. 3-13. Astrocyte and mixed glial transplant characterizations. A and B
represent identical fields of a mixed glial suspension cultured and doubled-
stained for GSI-B4 (FITC) and GFAP (RITC), respectively (X320). C and D
represent an enriched astrocyte suspension cultured and stained for GFAP
(RITC). While CD show high levels of GFAP expression, some regions of the
culture contained neither GFAP-IR nor lectin-positive cells (_D). Interestingly,
the nuclei (Hoechst 33342) of these non-labeled cells very much resembled
neighboring GFAP-IR cells (not shown).

Quantitative Evaluation of PTFE Control and Experimental Implants
An effort was made to put the qualitative observations described above
into a quantitative context. As noted in Material and Methods, this entailed
compiling the ordinal measurements of immunoreactive (IR) profiles seen
within the rostral (R) and caudal (C) segments of cryosections through each
PTFE tube implant after 5 weeks (Fig. 3-1). In brief, after each implant was
immunohistochemically stained with OX-42, neurofilament (NF) or laminin
(LAM) antibodies, an ordinal rank (0-3) was given to the amount of
immunoreactivity seen in R and C portions of the matrices based on the
observation of at least 3 immunoreactive profiles within an ordinal mode (0-
3). A summary of all the data is schematically represented in Fig. 3-17. Non-
parametric statistics were then used to determine whether there were
significant differences in the immunohistochemical profiles among the
various implant groups (Fig. 3-19).
In summary, the control tube implants demonstrated the presence of
OX-42-IR cells accompanied by neurofilament- and laminin-IR profiles that
were all significantly less than those seen in the cell-seeded implants (Figs. 3-
18,19). This is illustrated in Fig. 3-20 showing that the infiltrating OX-42-IR
cells were accompanied by neurofilament- and laminin-IR profiles, especially
prominent in the rostral end of the implant. For comparison, Figs. 3-19,20
illustrate that the R and C segments of the microglia-impregnated tubes
contained significantly more immunoreactive profiles throughout their
matrices compared to the control tubes after 5 weeks. However, there were
no significant differences between the LPS-stimulated versus non-stimulated
microglial implants, except in the rankings for caudal laminin-IR (Fig. 3-19).
Interestingly, the mixed glial implants demonstrated the highest
ordinal rankings with the least variances of any group, and were found to

Fig. 3-14. PTFE implants impregnated with cultured astrocytes 5 weeks after
transplantation. The top and bottom of A represent the lateral walls of the
implant (host tissue not seen). However, the lateral walls of the implant are
located to the left and right in B.C. A. Longitudinal, horizontal section
through an astrocyte-seeded PTFE tube stained for GFAP demonstrates the
unexpectedly scarce detection of GFAP-IR cells within these implants, and the
apparent emigration of some these cells through the channels of the wall
(arrows; X80). B and C both represent the same field of view seen through
FITC and RITC fluorescent filters, respectively, in order to demonstrate the
coexistence (arrows) of GFAP-IR elements (BQ) almost exclusively with
laminin-IR profiles (Cj) that, in many instances, did not resembled
microvasculature (X120).


Fig. 3-15. High magnifications of astrocyte-impregnated PTFE implants. The
tops of AB show the lateral gelfoam-PTFE wall border (arrows) while the core
of the matrix is shown lower. A,B represent the same fields of view seen
though different fluorescent filters demonstrating remarkable coexistence of
GFAP-IR elements (A) with laminin-IR profiles (jB) resembling
microvasculature (X120). Note that some of the GFAP-IR elements are seen
penetrating the lacunae of the walls. Interestingly, these astrocyte implants
also displayed faint OX-42-IR (Q) in regions devoid of GFAP-IR cells (X120).


have significantly higher rankings than the microglial implants (Fig. 3-19).
Nevertheless, both microglial and mixed glial implants had significantly
higher neurofilament-IR (NF-IR) rankings (R and C) than did astroglial
implants (Fig. 3-19). This is again illustrated in Fig. 3-20 which shows that
this ingrowth was diametrically opposite to that seen in the control tubes.
Neurotransmitter Characterization of Elongated Neurites in IPTFE Tubes
Preliminary immunohistochemical investigations were performed to
determine the neurotransmitter content of the ingrowing neurites which
might give some clues as to their origin. The two antibodies chosen to

characterize these processes were directed against calcitonin gene-related
peptide (CGRP) recognizing ascending primary afferent fibers and 5-HT which
labels descending serotonergic fibers.
Both stains demonstrated the normal distributions of these fibers
within the host spinal cord surrounding the PTFE implants. In no instances
were any 5-HT-IR processes seen penetrating into the gelfoam matrices or
into the lacunae of the walls in control or experimental (i.e., cell-seeded)

tubes. However, some were found to course along the edges of the tube outer

On the contrary, CGRP-IR fibers penetrated both control and cell-seeded
gelfoam matrices, but they only represented a relatively minor population of

all the NF-IR processes seen in the implants, especially the grafts containing
microglia. There did not appear to be any preferential ingrowth with respect to
rostral or caudal ends, but qualitatively there seemed to be more CGRP-IR
processes in the dorsal-most longitudinal sections.

Fig. 3-16. Transverse, plastic sections (2gm) through astrocyte-impregnated
PTFE implants after 5 weeks. A. Low power photomicrograph shows regions
in the graft containing large cell bodies (arrows) accompanied by various
other cellular phenotypes (X10). B. Higher power reveals islands of
cohesively packed astrocytes accompanied by macrophages and other
elements in the center of these grafts (X320). C. All along the periphery of the
gelfoam matrices was a trabecular meshwork of astrocytes that, in many
instances, resembled an external glial limiting membrane. It was difficult to
find neurites entering these grafts, and the few observed (arrow) were not
found in astrocyte-rich regions (X320).

S* **:. -

# ._ owl

Aa.- 0


In the preceding investigation (Chapter 2), an initial attempt was made
to determine whether activated microglia/brain macrophages derived from
tissue culture (for simplicity herein referred to as microglia) would be
permissive or nonpermissive to neuritic outgrowth in the injured spinal
cord. The findings countered the view that reactive microglia in the injured
CNS have neurotoxic effects since neuritic profiles were observed extending
into microglia-seeded gelfoam matrices. Outgrowth also was indicated in
regions of inflammation, characterized by accumulations of OX-42-IR cells,
seen in the host CNS immediately adjacent to biodegradable polymeric tubes
housing some of the gelfoam matrices. The present findings are consistent
with the previous observations regarding neuritic fiber penetration of cell-
seeded matrices. Qualitatively and quantitatively, it was observed that
gelfoam matrices seeded either with microglia alone or a combination of
those cells and cultured astrocytes induced a greater degree of fiber ingrowth
than seen either into control gelfoam implants or gelfoam containing
enriched astroglial populations. Lastly, the LPS-activation of microglia in
vitro prior to their transplantation failed to render these cells nonpermissive
to axonal elongation.
PTFE vs. Biodegradable Polymeric Implants
One objective of this study was to overcome some of the complications
introduced in the earlier investigation by the polymeric tubes that were used
to introduce gelfoam matrices in a more oriented fashion. In particular, it
appeared that a local inflammatory response had occurred as a consequence of
polymer degradation. This may have caused an influx of endogenous
microglia-derived brain macrophages and other cell types such that it was

Figure 3-17


stral 0 1 21 31 31 2 1 0 Caudal







7 45

40 65

48, 23

62' 24
IN =157)


4 1

31 38

31 37

115 105
(N = 1811

3 4

30 21

40 71

103 80
N = 176


15 3

1 24

15, 18

79 68
(N = 1101


0 0

0 0

0 0

70 70
IN 70)

0 0

0 0

0 0

68 68
IN -68)


* q Y Y I

0 0

0 0

0 0

54 54
(N 54)

p p p

18, 41

31 53

54 29

57 37
IN 1601

41 69 16 24 18 2 1 0 23 0
39 54 51 44 24 34 6 6 20 11
I I i I
44 23 38 52 25 46 12 12 6 12
28 7 69 54 43 28 49 50 3 29
IN = 1521 (N 1741 IN 110) (N =68) (N = 52)

9 0

7 7

13 20

81 83
(N = 110)

3 0

7 0

12, 0

30 52
IN 52)

Figure 3-18










Fig. 3-17,18. Representational summary of the immunohistochemical profiles
that were observed within all the PTFE implants and given an ordinal ranking
according to the procedure illustrated in Fig. 3-1. In brief, each implantation
group had rostral (R) and caudal (C) sections of their tubes examined for each of
the variable stains, and each tallied number in Fig. 3-17 represents individual
sections given a rank (0-3; N=total sections counted). The shaded numbers are
the ordinal rank most often ascribed to the variable immunostain and thus
represents the rank assigned to each implantation group (R and C). These are
highlighted in Fig. 3-18, which also displays the median (M) ranks along with the
quartile variances. The assigned ranks for the implantation groups are
schematically shown in Fig. 3-20.