Acta Neuropathol (1992) 85:93 - 100
Acta Heuropa|hologlca 9 Springer-Verlag1992
Histopathological effects of intracerebral injections of human recombinant tumor necrosis factor- in the rat* James L. Wright 1 and Randall E. Merchant 1,2 Departments of 1Anatomy,MCV Station, Box 709, and 2Surgery (Division of Neurosurgery),Virginia CommonwealthUniversity, Medical College of Virginia, Richmond,VA 23298-0709, USA Received November5, 1991/Revised, accepted June 11, 1992
Summary. Human recombinant tumor necrosis factor-a (rTNF-a) was administered to normal Fischer 344 rats by stereotaxic intracerebral (IC) injection. Animals received a single injection of either 6 x 104 U rTNF-a or excipient in their right parietal lobe. Others received three consecutive daily injections of either 6 • 104 U rTNF-a or excipient to examine effects of higher accumulative doses. Histological examination of the brain revealed that both single and multiple IC injections of rTNF-a triggered an immigration of circulating leukocytes into the site of TNF-a injection. After one injection, this cell population was composed mainly of macrophages and neutrophils. Maximal leukocytic influx occurred by 48 h and was composed mostly of neutrophils which were limited to the injection site and perivascular space. Quantitation of the inflammatory reaction by measurement of tissue myeloperoxidase levels supported these histological observations. One day after multiple rTNF-a injections, leukocytic adhesion to endothelium, vascular cuffing and leukocytic infiltration into the neuropil was observed at levels comparable to those seen 3 days following a single rTNF-c~ injection. We conclude that while one or more IC injection(s) of 6 x 104 U rTNF-c~ was well tolerated in normal rats, at this dose the cytokine triggers a pronounced leukocytic infiltration at the site of injection. These results support a role for TNF-~ as a mediator in inflammatory responses within the central nervous system. Key words: Tumor necrosis factor-a - Cytokines Inflammation - Neutrophil - Leukocytes
* Supported in part by a grant from the A. D. Williams Research Fund and by giftsfrom the KelloggFoundation,the Lind Lawrence Fund, and the family and friends of Christine Armstrong, Jack Harvey, Christopher Wemple, and Pearl Ylonen. Correspondence to: R. E. Merchant (address see 1 above)
It has become increasingly clear in recent years that most of the events of acute inflammation such as increased vascular permeability and leukocytosis can be attributed to cytokines, like tumor necrosis factor-~ (TNF-c0, produced by inflammatory cells [15, 24]. Originally described as the agent responsible for hemorrhagic necrosis of sarcoma, it is now known that the major role of TNF-~ is as a pro-inflammatory agent which is responsible for its antitumor activity. Pro-inflammatory actions which are most effective against tumor are those which involve the vascular and leukocytic elements of inflammation. In vitro, the effects of TNF-c~ on leukocytes include increased neutrophil adherence to endothelial cells and induction of an adhesion-dependent migration of neutrophils through endothelial monolayers [9, 23, 26]. Human recombinant TNF-c~ (rTNF-c0 has been also shown to be chemotactic for neutrophils, and will induce their migration across polycarbonate and nitrocellulose filters along a TNF-~ gradient [18]. In addition, the cytokine stimulates neutrophil respiratory burst and degranulation leading to a release of reactive oxygen intermediates [38]. Although the pro-inflammatory actions of TNF-~ are generally beneficial to the host, TNF-c~ can have deleterious effects when systemic or local concentrations are high.With regards to the central nervous system (CNS), rTNF-c~ has been reported to be pyrogenic through its ability to induce prostaglandin synthesis by the hypothalamus [6, 28]. In vitro, rTNF-c~ has been shown to be directly toxic for oligodendrocytes and the myelin sheath of mouse embryonic spinal cord explants [30, 34], evidence which confirms suspicions of the factor's role in demyelination associated with multiple sclerosis (MS) and AIDS-related encephalopathy [4, 20]. In patients with malaria, high serum levels of TNF-a have been linked to cerebral hemorrhagic necrosis which is seen in severe cases [12]. High levels of serum TNF-a have also been strongly associated with septic shock and the fatal outcome in cases of meningitis [35].When rabbits were injected with 104 U rTNF-~x into the cisterna magna, an induction of a significant leuko-
94
cytic migration into the subarachnoid space resulted, mimicking the leukocytic response seen in an experimental model of gram-positive meningitis [33]. Previous studies from our laboratory have examined the host response in the normal rat brain and in intracerebral (IC) gliomas of rats following a local or intravenous (IV) injection(s) of human recombinant interleukin-2 (rIL-2) and murine recombinant interleukin-l[3 (rIL-1]3), and systemic injections of human rTNF-a [11, 13, 29, 37]. Although antitumor effects were seen in animal models of glioma receiving IV rTNF-a, these effects were offset by systemic toxicity, specifically hemorrhagic enterocolitis [13]. It is our hope to limit the systemic toxicity of rTNF-a as well as intensify the inflammatory reaction to the cytokine by its IC administration into growing gliomas. However, to advance rTNF-a as a possible agent for use in nonspeciflc active immunotherapy for glioma, it is important to define more clearly the differences between inflammatory-potentiating actions of rTNF-a on normal brain and glioma, and to demonstrate a level of rTNF-a which, while having little or no inflammatory actions on normal brain, can still initiate a potentially therapeutic inflammatory response in glioma. Because of the ramifications of neurological damage, clinicians are presented with the problem of localizing the area affected by treatment of CNS neoplasms. In nonspecific immunotherapeutic regimens applied to the CNS it is, therefore, important to investigate whether the inflammatory response, once initiated, will be limited to the site of injection, or spread to the surrounding neuropil. To compare empirically the effects of rTNF-a in normal to tumor-bearing brain, and to investigate the regions involved in the inflammatory response, we combined quantitative methods to measure inflammation with extensive histological examination of rats receiving IC injections of human rTNF-a.
Materials and methods
Animals Female Fischer 344 rats (Harlan Sprague Dawley, Inc.) weighing 140-160 g were used. Animals were housed in individual cages and maintained on a 12-h light/dark cycle with food and water supplied ad libitum. Weights and body temperatures were monitored throughout the study.
Cytokine Lyophilized human rTNF-a was provided by the Cetus Corporation (Emeryville, Calif.). Purity was > 99 % as determined by SDS-PAGE, whith a specific activity of 24 • 106 U/mg protein and an endotoxin content of 0.02 ng/ml (as determined by the Limulus amoebocyte lysate assay). Human rTNF-a in a bulking agent of 2.6 % mannitol and 0.9 % sucrose was reconstituted with sterile, endotoxin-free water. Excipient was composed of 2.6 % mannitol (Sigma) and 0.9% sucrose (Sigma) in sterile, endotoxin-free water.
Injection protocol Under pentobarbital anesthesia, all animals had a 22G stainless steel guide cannula (Plastic Products, Inc.) stereotactically implanted in the right parietal bone 1 mm posterior and 4 mm lateral to bregma. Guide cannulas were fastened to the cranium with dental acrylic which in turn was anchored to three ~00 stainless steel screws of 0.75-mm diameter (Small Parts, Inc.) screwed into the parietal and frontal bones. These cannulas passed through the cranium but did not pierce the dura mater and were kept patent with a stainless steel wire of 0.4-ram diameter. Three days following cannula implantation, injections were made with a Hamilton syringe fitted with a length of polyethylene tubing and connected to a 28G blunt-tip injection cannula. Animals were first anesthetized by methoxyflurane inhalation and the injection cannula was inserted through the guide cannula into the parietal lobe at a depth of 3.5 ram. After 3-min accommodation, a 5 @ volume of either rTNF-ct or excipient (see below) was injected over a 10-rain period. Following a subsequent 5-rain accommodation period, the injection cannula was slowly withdrawn.
Single rTNF-a injections Using the injection cannula, animals received either 6 • 104 U rTNF-a or excipient in 5 gl. Rats were later killed after 4 (n = 4), 6 (n = 7) and 12 (n = 6) h and 1 (n = 4), 2 (n = 6), 3 (n = 14), 5 (n = 11), and 7 (n = 8) days. Each group consisted of n rats and was divided evenly between those receiving rTNF-a and excipient (rats killed 6 h following injection were divided so that four received rTNF-a and three received excipient; rats killed 5 days following injection were divided so that six received rTNF-a and five received excipient). Sham control animals (n = 3) were implanted with a guide cannula and then killed 3 days later to examine the effects of cannula implantation on leukocytic infiltration. Brain tissue was processed for histological examination as described below.
Multiple rTNF-a injections Three days after cannula placement, a regimen of three daily injections was started. For each injection, animals were anesthetized by methoxyflurane inhalation and were then given a 5-gl volume of either 6 x 104 U rTNF-c~ (n = 4) or excipient (n = 3). The injection cannula was inserted through the guide cannula and injections were made within the cortex of the parietal lobe at a depth of 1.5 mm. Sham control animals (n = 2) were implanted with a guide cannula and an injection cannula was inserted and withdrawn on each of the 3 days of the study. Animals were killed 24 h following the third injection of either rTNF-a or excipient, and brain tissue was processed for histological examination as described below.
Quantitation of inflammatory reaction Rats given a 5-~1 volume of either 6 x 104 U rTNF-a or excipient. and killed after 4 (n = 9), 6 (n = 6) and 12 (n = 8 ) h and 1 (n = 12), 2 (n = 17), 3 (n = 6) and 5 (n = 6) days following injection (rats killed 4 h following injection were divided so that five received rTNF-a and four received excipient; rats killed 2 days following injection were divided so that nine received rTNF-a and eight received excipient). Tissue samples were assayed for myeloperoxidase (MPO) content as described below.
95
Effects of rTNF-a dose on inflammatory reaction Three days following cannula implantation, animals were injected with varying doses of rTNF-c~ or excipient in 5 ~1 and killed 48 h later. They were divided evenly between those receiving rTNF-a and those receiving excipient at the following doses: 1.5 • 104 U (n = 10), 3 x 104 U (n = 12), 6 • 104 U (n = 16), and 2.4 • 105 U (n = 10). Tissue samples were assayed for MPO content as described below.
readings were compared against a standard curve constructed with known quantities of bovine serum albumin to give relative g proteirdml homogenate. Units MPO activity/g homogenate were calculated on a Lotus 1-2-3 spreadsheet.
Statistics Percent weight and temperature changes were converted to arcsine values prior to statistical analysis using Student's t-test.
Tissue processing Rats were transcardially perfused first with physiological saline followed by a 0.1 M phosphate-buffered fixative containing 2.5 % paraformaldehyde and 2 % glutaraldehyde. Brains were removed and held in buffer until processed further. Tissue from the mid-injection site was dehydrated and embedded in JB-4 glycomethacrylate (Polysciences). Sections were stained with hematoxylin and eosin phloxine (H&E, Polysciences). A semiquantitative system of histological grading was devised for plastic-embedded specimens stained with H&E. Sections were examined for four traits and graded from 0-3 for location and intensity of each particular trait. These four traits were (l) leukocytic adherence to luminal endothelium, (2) leukocytic cuffing around cerebral vasculature, (3) leukocytic infiltration into injection site, and (4) leukocytic infiltration throughout neuropil. Rats receiving rTNF-c~ were compared to those receiving excipient by averaging results from combined observations.
Myeloperoxidase assay Animals were anesthetized with pentobarbital and then transcardially perfused with a 50 mM phosphate buffer (pH 6.0) for 90 s. Brains were removed, and a coronal section encompassing the injection site (5 mm in width) was taken. Sections were bisected into left and right hemispheres, the cortex was removed by blunt dissection, and remaining tissue was deposited into vials of approximately 1.3 ml of 0.5 % hexadecyltrimethylammonium bromide (HTAB, Sigma). Tissue was frozen at - 7 0 ~ until further use. Tissue was thawed and homogenized in a teflon pestle homogenizer (Thomas Scientific) attached to a variable speed drive. Following homogenization, samples were sonicated for 10 s in an ice-water bath and then freeze-thawed twice more. Following a second sonication in an ice-water bath, 500 ~1 of each sample was extracted and conserved for determination of protein levels as described below. The remainder of the samples were centrifuged in a microcentrifuge (Eppendorf) at 12,000 g for 15 rain, and 10 ~tl of the supernatant was transferred in triplicate to a 96-well plate (Coming). To each well, 300 ~1 of 50 mM PBS (pH 6.0) was added containing 0.2% o-dianisidinedihydrochloride (Sigma) and 0.0005 % hydrogen peroxide (Sigma). After incubation at 23 ~ for 60 rain, the absorbance of samples was recorded at a wavelength of 460 nm in a spectrophotometer (Biotek). Absorbance readings were compared to a standard curve constructed from known quantities of human MPO (Sigma) to give relative units MPO/ml supernate.
Protein content assay (BCA assay) Samples were taken as noted above, diluted five fold with 50 mM phosphate buffer and transferred in f0 ~1 triplicates to a 96-well plate (Coming). Each well received 200 ~1 of a "BC~2' solution (Sigma) containing 1 part copper (II) sulfate pentahydrate solution to 50 parts bicinchonic acid solution. After incubation at 23 ~ for 20 rain, the absorbance of each sample was measured in a spectrophotometer (Biotek) at 570 nm wavelength. Absorbance
Results N o r m a l rats t o l e r a t e d single a n d multiple I C injections of rTNF-c~ w i t h o u t observable distress o r neurological toxicity. T h e dose-limiting side effects of h e m o r r h a g i c enterocolitis and destruction o f intestinal villi that we h a d seen with systemic administration of h u m a n r T N F - a in the rat [13] were n o t observed. A n i m a l s t o l e r a t e d c a n n u l a i m p l a n t a t i o n and single injections of 6 x 104 U r T N F - ~ or excipient w i t h o u t observable distress and w i t h o u t significant effect on weight. T e m p e r a t u r e changes were such that 10 min following administration of m e t h o x y f l u r a n e anesthesia, t e m p e r a t u r e s in b o t h groups decreased f r o m baseline levels b y o v e r 2.5 ~ O v e r the next 6 h, t e m p e r a t u r e s increased steadily to over 0 . 5 ~ a b o v e the baseline and r e t u r n e d and r e m a i n e d n e a r baseline levels f r o m 12 h to 7 days following injection (data n o t shown). N o significant difference was seen b e t w e e n the variation of t e m p e r a tures in r T N F - a and excipient-injected animals at any time point. Histological e x a m i n a t i o n r e v e a l e d that all injections passed t h r o u g h the external capsule and e n t e r e d the basal ganglia. E x p e r i m e n t a l animals were e x a m i n e d and g r a d e d for the p r e s e n c e of (1) leukocytic a d h e r e n c e to luminal e n d o t h e l i u m , (2) leukocytic cuffing o f cerebral vessels, (3) leukocytic infiltration into the injection site, and (4) leukocytic infiltration t h r o u g h o u t s u r r o u n d i n g neuropil. T h e e x a m i n a t i o n o f the cerebral vasculature of rats receiving an I C excipient injection revealed small n u m bers of m o n o n u c l e a r cells a d h e r i n g to luminal e n d o t h e lium in the i m m e d i a t e vicinity of the injection site only, but by 12 h t h r o u g h 7 days postinjection, n o f u r t h e r evidence of leukocytic a d h e r e n c e was observed. I n r T N F - a - i n j e c t e d rats, a m o d e r a t e level o f m o n o n u c l e a r leukocytic a d h e r e n c e to luminal e n d o t h e l i u m t h r o u g h out the injection h e m i s p h e r e was seen at 4 h postinjection. T h e extent o f this a d h e r e n c e increased, so that at 12 and 24 h vessels t h r o u g h o u t the ipsilateral and contralateral h e m i s p h e r e s s h o w e d n u m e r o u s a d h e r e n t m o n o c y t e s (Fig. 1). B y 48 h postinjection, neutrophils r e p l a c e d m a c r o p h a g e s as the p r e d o m i n a n t a d h e r e n t cell type in vessels o f b o t h h e m i s p h e r e s , and the n u m b e r o f vessels displaying these l e u k o c y t e s was greater t h a n seen at previous time points as well (Fig. 2). T h r e e to 7 days following injection o f r T N F - a , leukocytic a d h e r e n c e declined and b e c a m e almost u n d e t e c t a b l e and w h e n observed, was only in the i m m e d i a t e vicinity of the injection site.
96
Fig. 1. Monocytic adherence to vessel in contralateral basal ganglia 12 h following recombinant tumor necrosis factor-c~ (rTNF-c~) injection, x 400 Fig. 2. Vessel in neuropil surrounding injection site, 48 h following rTNF-a, showing adherence (small arrow) and cuffing (large arrow) of neutrophils. Also apparent are a few, scattered neutrophils outside of the perivascular space within the neuropil (circle). x 400 Fig. 3. Infiltrate composed mainly of erythrocytes within injection site of excipient-injected animals; 48 h postinjection, x 85 Fig. 4. Massive neutrophil infiltration into injection site of rTNF-a recipients; 48 h following injection. • 65 Fig. 5. Edge of neutrophil infiltration site in rTNF-c~ recipients. The limit of neutrophil migration into the tissue is rather abrupt and corresponds with the resumption of normal neuropil cytoarchitecture, x 300
In e xcipient-injected animals, a low level of vascular cuffing was seen f r o m 4 to 4 8 h in the injection hemisphere only c o m p o s e d of m a c r o p h a g e s and some neutrophils, and no cuffing was detectable by 3 days.
Vascular cuffing in r T N F - ~ recipients exceeded that seen in those receiving an excipient injection (Fig. 2). O n day 2, this cuffing was observed throughout the injection hemisphere and consisted mainly of neutrophils with considerably less neutrophilic infiltration into surrounding neuropil, while on day 3 a combination of neutrophils and macrophages were observed around vessels throughout the hemisphere.
97
Leukocytes within the injection site of excipientinjected animals were composed of low to moderate levels of neutrophils and macrophages after 4 to 48 h (Fig. 3). The cell composition within the injection site changed to predominately macrophages at 3 days postinjection, and at 5 and 7 days, no leukocytes were observed. In rats receiving rTNF-(,, however, a large area of densely concentrated neutrophils was observed at the injection site 48 h following administration (Fig. 4). The area of neutrophil infiltration was limited to the edematous tissue surrounding the injection site with only scattered neutrophils in the surrounding normal neuropil. Upon closer examination of these infiltrated areas, a fairly sharp transition was observed between infiltrated tissue and surrounding normal neuropil (Fig. 5). The size of this reaction clearly surpassed that seen in excipient animals at this or any other time point. To quantitate the inflammatory response observed in histological studies following single injections into normal brain, brain samples were analyzed for MPO content at time points ranging from 4 h to 3 days following injection with either 6 x 104 U rTNF-~ or excipient. As evidenced by the MPO/protein assay, no significant difference was seen between animals receiving rTNF-(* or excipient at any time point, although a trend towards increased tissue MPO was seen at 48-h post-rTNF-(, injection (Fig. 6). Histological examination of sham controls for multipie-injection models indicated an increase in collagen fibers and ground substance as well as proliferating fibroblasts and moderate edema in and around the needle entry site. Leukocytes were limited to a small number of macrophages in the immediate area of the entry site. Both rTNF-~ and excipient recipients shared these findings, but in addition, a compact area of dense neutrophil accumulation was observed at the cannula implantation site. Histological descriptions for all multiple injection animals are superimposed on these findings. Twenty-four hours following the last of the three daily injections of 6 x 104 U rTNF-(,, there was extensive leukocytic adhesion to endothelium and cuffing around cerebral vessels of the injection site and within the basal ganglia similar in appearance to that seen 3 days following a single rTNF-(, injection. This was accompanied in the majority of animals by a dense infiltration of macrophages and lymphocytes into the neuropil at the injection site (Fig. 7). Animals given multiple IC injections of excipient as well as sham controls showed some leukocytic infiltration, but in comparison to multiple rTNF-(x recipients, leukocytic adhesion to and cuffing around cerebral vessels was greatly reduced or absent. Discussion
This study examined the effects of single and multiple injections of human rTNF-(~ directly into normal brain tissue to assess its role in mediating inflammatory responses within the CNS and potential as an immuno-
120 c
O_ 03
[] []
rTNF Excip.
90
>, >
~5
60
0n 09
30
c
L
4h
6h
12h
24 h
d
3d
Fig, 6. Quantification of inflammatory response to ICinjection of either excipient or rTNF-c~. Although assay supports histological data, showing greater tissue MPO content 48 h following rTNF-a injection, no significant differences are seen between rTNF-c~ and excipient groups
Fig. 7. Large-scale infiltration of macrophages (large arrows) and lymphocytes (small arrows) into injection site 24 h after the last of three daily injections of rTNF-0t, x 450
therapeutic agent against glioma. While much has been done in recent years to elucidate the wide range of activities of TNF-~ in systemic infections [24, 27], only lately has attention focused on its effects in the CNS [13, 28, 33]. In these latter studies, IV or intracerebroventricular (ICV) injections of rTNF-c~ were made to examine in more detail its suspected toxicity for neural tissue, since the neurotoxic effects of TNF
Acta Heuropa|hologlca 9 Springer-Verlag1992
Histopathological effects of intracerebral injections of human recombinant tumor necrosis factor- in the rat* James L. Wright 1 and Randall E. Merchant 1,2 Departments of 1Anatomy,MCV Station, Box 709, and 2Surgery (Division of Neurosurgery),Virginia CommonwealthUniversity, Medical College of Virginia, Richmond,VA 23298-0709, USA Received November5, 1991/Revised, accepted June 11, 1992
Summary. Human recombinant tumor necrosis factor-a (rTNF-a) was administered to normal Fischer 344 rats by stereotaxic intracerebral (IC) injection. Animals received a single injection of either 6 x 104 U rTNF-a or excipient in their right parietal lobe. Others received three consecutive daily injections of either 6 • 104 U rTNF-a or excipient to examine effects of higher accumulative doses. Histological examination of the brain revealed that both single and multiple IC injections of rTNF-a triggered an immigration of circulating leukocytes into the site of TNF-a injection. After one injection, this cell population was composed mainly of macrophages and neutrophils. Maximal leukocytic influx occurred by 48 h and was composed mostly of neutrophils which were limited to the injection site and perivascular space. Quantitation of the inflammatory reaction by measurement of tissue myeloperoxidase levels supported these histological observations. One day after multiple rTNF-a injections, leukocytic adhesion to endothelium, vascular cuffing and leukocytic infiltration into the neuropil was observed at levels comparable to those seen 3 days following a single rTNF-c~ injection. We conclude that while one or more IC injection(s) of 6 x 104 U rTNF-c~ was well tolerated in normal rats, at this dose the cytokine triggers a pronounced leukocytic infiltration at the site of injection. These results support a role for TNF-~ as a mediator in inflammatory responses within the central nervous system. Key words: Tumor necrosis factor-a - Cytokines Inflammation - Neutrophil - Leukocytes
* Supported in part by a grant from the A. D. Williams Research Fund and by giftsfrom the KelloggFoundation,the Lind Lawrence Fund, and the family and friends of Christine Armstrong, Jack Harvey, Christopher Wemple, and Pearl Ylonen. Correspondence to: R. E. Merchant (address see 1 above)
It has become increasingly clear in recent years that most of the events of acute inflammation such as increased vascular permeability and leukocytosis can be attributed to cytokines, like tumor necrosis factor-~ (TNF-c0, produced by inflammatory cells [15, 24]. Originally described as the agent responsible for hemorrhagic necrosis of sarcoma, it is now known that the major role of TNF-~ is as a pro-inflammatory agent which is responsible for its antitumor activity. Pro-inflammatory actions which are most effective against tumor are those which involve the vascular and leukocytic elements of inflammation. In vitro, the effects of TNF-c~ on leukocytes include increased neutrophil adherence to endothelial cells and induction of an adhesion-dependent migration of neutrophils through endothelial monolayers [9, 23, 26]. Human recombinant TNF-c~ (rTNF-c0 has been also shown to be chemotactic for neutrophils, and will induce their migration across polycarbonate and nitrocellulose filters along a TNF-~ gradient [18]. In addition, the cytokine stimulates neutrophil respiratory burst and degranulation leading to a release of reactive oxygen intermediates [38]. Although the pro-inflammatory actions of TNF-~ are generally beneficial to the host, TNF-c~ can have deleterious effects when systemic or local concentrations are high.With regards to the central nervous system (CNS), rTNF-c~ has been reported to be pyrogenic through its ability to induce prostaglandin synthesis by the hypothalamus [6, 28]. In vitro, rTNF-c~ has been shown to be directly toxic for oligodendrocytes and the myelin sheath of mouse embryonic spinal cord explants [30, 34], evidence which confirms suspicions of the factor's role in demyelination associated with multiple sclerosis (MS) and AIDS-related encephalopathy [4, 20]. In patients with malaria, high serum levels of TNF-a have been linked to cerebral hemorrhagic necrosis which is seen in severe cases [12]. High levels of serum TNF-a have also been strongly associated with septic shock and the fatal outcome in cases of meningitis [35].When rabbits were injected with 104 U rTNF-~x into the cisterna magna, an induction of a significant leuko-
94
cytic migration into the subarachnoid space resulted, mimicking the leukocytic response seen in an experimental model of gram-positive meningitis [33]. Previous studies from our laboratory have examined the host response in the normal rat brain and in intracerebral (IC) gliomas of rats following a local or intravenous (IV) injection(s) of human recombinant interleukin-2 (rIL-2) and murine recombinant interleukin-l[3 (rIL-1]3), and systemic injections of human rTNF-a [11, 13, 29, 37]. Although antitumor effects were seen in animal models of glioma receiving IV rTNF-a, these effects were offset by systemic toxicity, specifically hemorrhagic enterocolitis [13]. It is our hope to limit the systemic toxicity of rTNF-a as well as intensify the inflammatory reaction to the cytokine by its IC administration into growing gliomas. However, to advance rTNF-a as a possible agent for use in nonspeciflc active immunotherapy for glioma, it is important to define more clearly the differences between inflammatory-potentiating actions of rTNF-a on normal brain and glioma, and to demonstrate a level of rTNF-a which, while having little or no inflammatory actions on normal brain, can still initiate a potentially therapeutic inflammatory response in glioma. Because of the ramifications of neurological damage, clinicians are presented with the problem of localizing the area affected by treatment of CNS neoplasms. In nonspecific immunotherapeutic regimens applied to the CNS it is, therefore, important to investigate whether the inflammatory response, once initiated, will be limited to the site of injection, or spread to the surrounding neuropil. To compare empirically the effects of rTNF-a in normal to tumor-bearing brain, and to investigate the regions involved in the inflammatory response, we combined quantitative methods to measure inflammation with extensive histological examination of rats receiving IC injections of human rTNF-a.
Materials and methods
Animals Female Fischer 344 rats (Harlan Sprague Dawley, Inc.) weighing 140-160 g were used. Animals were housed in individual cages and maintained on a 12-h light/dark cycle with food and water supplied ad libitum. Weights and body temperatures were monitored throughout the study.
Cytokine Lyophilized human rTNF-a was provided by the Cetus Corporation (Emeryville, Calif.). Purity was > 99 % as determined by SDS-PAGE, whith a specific activity of 24 • 106 U/mg protein and an endotoxin content of 0.02 ng/ml (as determined by the Limulus amoebocyte lysate assay). Human rTNF-a in a bulking agent of 2.6 % mannitol and 0.9 % sucrose was reconstituted with sterile, endotoxin-free water. Excipient was composed of 2.6 % mannitol (Sigma) and 0.9% sucrose (Sigma) in sterile, endotoxin-free water.
Injection protocol Under pentobarbital anesthesia, all animals had a 22G stainless steel guide cannula (Plastic Products, Inc.) stereotactically implanted in the right parietal bone 1 mm posterior and 4 mm lateral to bregma. Guide cannulas were fastened to the cranium with dental acrylic which in turn was anchored to three ~00 stainless steel screws of 0.75-mm diameter (Small Parts, Inc.) screwed into the parietal and frontal bones. These cannulas passed through the cranium but did not pierce the dura mater and were kept patent with a stainless steel wire of 0.4-ram diameter. Three days following cannula implantation, injections were made with a Hamilton syringe fitted with a length of polyethylene tubing and connected to a 28G blunt-tip injection cannula. Animals were first anesthetized by methoxyflurane inhalation and the injection cannula was inserted through the guide cannula into the parietal lobe at a depth of 3.5 ram. After 3-min accommodation, a 5 @ volume of either rTNF-ct or excipient (see below) was injected over a 10-rain period. Following a subsequent 5-rain accommodation period, the injection cannula was slowly withdrawn.
Single rTNF-a injections Using the injection cannula, animals received either 6 • 104 U rTNF-a or excipient in 5 gl. Rats were later killed after 4 (n = 4), 6 (n = 7) and 12 (n = 6) h and 1 (n = 4), 2 (n = 6), 3 (n = 14), 5 (n = 11), and 7 (n = 8) days. Each group consisted of n rats and was divided evenly between those receiving rTNF-a and excipient (rats killed 6 h following injection were divided so that four received rTNF-a and three received excipient; rats killed 5 days following injection were divided so that six received rTNF-a and five received excipient). Sham control animals (n = 3) were implanted with a guide cannula and then killed 3 days later to examine the effects of cannula implantation on leukocytic infiltration. Brain tissue was processed for histological examination as described below.
Multiple rTNF-a injections Three days after cannula placement, a regimen of three daily injections was started. For each injection, animals were anesthetized by methoxyflurane inhalation and were then given a 5-gl volume of either 6 x 104 U rTNF-c~ (n = 4) or excipient (n = 3). The injection cannula was inserted through the guide cannula and injections were made within the cortex of the parietal lobe at a depth of 1.5 mm. Sham control animals (n = 2) were implanted with a guide cannula and an injection cannula was inserted and withdrawn on each of the 3 days of the study. Animals were killed 24 h following the third injection of either rTNF-a or excipient, and brain tissue was processed for histological examination as described below.
Quantitation of inflammatory reaction Rats given a 5-~1 volume of either 6 x 104 U rTNF-a or excipient. and killed after 4 (n = 9), 6 (n = 6) and 12 (n = 8 ) h and 1 (n = 12), 2 (n = 17), 3 (n = 6) and 5 (n = 6) days following injection (rats killed 4 h following injection were divided so that five received rTNF-a and four received excipient; rats killed 2 days following injection were divided so that nine received rTNF-a and eight received excipient). Tissue samples were assayed for myeloperoxidase (MPO) content as described below.
95
Effects of rTNF-a dose on inflammatory reaction Three days following cannula implantation, animals were injected with varying doses of rTNF-c~ or excipient in 5 ~1 and killed 48 h later. They were divided evenly between those receiving rTNF-a and those receiving excipient at the following doses: 1.5 • 104 U (n = 10), 3 x 104 U (n = 12), 6 • 104 U (n = 16), and 2.4 • 105 U (n = 10). Tissue samples were assayed for MPO content as described below.
readings were compared against a standard curve constructed with known quantities of bovine serum albumin to give relative g proteirdml homogenate. Units MPO activity/g homogenate were calculated on a Lotus 1-2-3 spreadsheet.
Statistics Percent weight and temperature changes were converted to arcsine values prior to statistical analysis using Student's t-test.
Tissue processing Rats were transcardially perfused first with physiological saline followed by a 0.1 M phosphate-buffered fixative containing 2.5 % paraformaldehyde and 2 % glutaraldehyde. Brains were removed and held in buffer until processed further. Tissue from the mid-injection site was dehydrated and embedded in JB-4 glycomethacrylate (Polysciences). Sections were stained with hematoxylin and eosin phloxine (H&E, Polysciences). A semiquantitative system of histological grading was devised for plastic-embedded specimens stained with H&E. Sections were examined for four traits and graded from 0-3 for location and intensity of each particular trait. These four traits were (l) leukocytic adherence to luminal endothelium, (2) leukocytic cuffing around cerebral vasculature, (3) leukocytic infiltration into injection site, and (4) leukocytic infiltration throughout neuropil. Rats receiving rTNF-c~ were compared to those receiving excipient by averaging results from combined observations.
Myeloperoxidase assay Animals were anesthetized with pentobarbital and then transcardially perfused with a 50 mM phosphate buffer (pH 6.0) for 90 s. Brains were removed, and a coronal section encompassing the injection site (5 mm in width) was taken. Sections were bisected into left and right hemispheres, the cortex was removed by blunt dissection, and remaining tissue was deposited into vials of approximately 1.3 ml of 0.5 % hexadecyltrimethylammonium bromide (HTAB, Sigma). Tissue was frozen at - 7 0 ~ until further use. Tissue was thawed and homogenized in a teflon pestle homogenizer (Thomas Scientific) attached to a variable speed drive. Following homogenization, samples were sonicated for 10 s in an ice-water bath and then freeze-thawed twice more. Following a second sonication in an ice-water bath, 500 ~1 of each sample was extracted and conserved for determination of protein levels as described below. The remainder of the samples were centrifuged in a microcentrifuge (Eppendorf) at 12,000 g for 15 rain, and 10 ~tl of the supernatant was transferred in triplicate to a 96-well plate (Coming). To each well, 300 ~1 of 50 mM PBS (pH 6.0) was added containing 0.2% o-dianisidinedihydrochloride (Sigma) and 0.0005 % hydrogen peroxide (Sigma). After incubation at 23 ~ for 60 rain, the absorbance of samples was recorded at a wavelength of 460 nm in a spectrophotometer (Biotek). Absorbance readings were compared to a standard curve constructed from known quantities of human MPO (Sigma) to give relative units MPO/ml supernate.
Protein content assay (BCA assay) Samples were taken as noted above, diluted five fold with 50 mM phosphate buffer and transferred in f0 ~1 triplicates to a 96-well plate (Coming). Each well received 200 ~1 of a "BC~2' solution (Sigma) containing 1 part copper (II) sulfate pentahydrate solution to 50 parts bicinchonic acid solution. After incubation at 23 ~ for 20 rain, the absorbance of each sample was measured in a spectrophotometer (Biotek) at 570 nm wavelength. Absorbance
Results N o r m a l rats t o l e r a t e d single a n d multiple I C injections of rTNF-c~ w i t h o u t observable distress o r neurological toxicity. T h e dose-limiting side effects of h e m o r r h a g i c enterocolitis and destruction o f intestinal villi that we h a d seen with systemic administration of h u m a n r T N F - a in the rat [13] were n o t observed. A n i m a l s t o l e r a t e d c a n n u l a i m p l a n t a t i o n and single injections of 6 x 104 U r T N F - ~ or excipient w i t h o u t observable distress and w i t h o u t significant effect on weight. T e m p e r a t u r e changes were such that 10 min following administration of m e t h o x y f l u r a n e anesthesia, t e m p e r a t u r e s in b o t h groups decreased f r o m baseline levels b y o v e r 2.5 ~ O v e r the next 6 h, t e m p e r a t u r e s increased steadily to over 0 . 5 ~ a b o v e the baseline and r e t u r n e d and r e m a i n e d n e a r baseline levels f r o m 12 h to 7 days following injection (data n o t shown). N o significant difference was seen b e t w e e n the variation of t e m p e r a tures in r T N F - a and excipient-injected animals at any time point. Histological e x a m i n a t i o n r e v e a l e d that all injections passed t h r o u g h the external capsule and e n t e r e d the basal ganglia. E x p e r i m e n t a l animals were e x a m i n e d and g r a d e d for the p r e s e n c e of (1) leukocytic a d h e r e n c e to luminal e n d o t h e l i u m , (2) leukocytic cuffing o f cerebral vessels, (3) leukocytic infiltration into the injection site, and (4) leukocytic infiltration t h r o u g h o u t s u r r o u n d i n g neuropil. T h e e x a m i n a t i o n o f the cerebral vasculature of rats receiving an I C excipient injection revealed small n u m bers of m o n o n u c l e a r cells a d h e r i n g to luminal e n d o t h e lium in the i m m e d i a t e vicinity of the injection site only, but by 12 h t h r o u g h 7 days postinjection, n o f u r t h e r evidence of leukocytic a d h e r e n c e was observed. I n r T N F - a - i n j e c t e d rats, a m o d e r a t e level o f m o n o n u c l e a r leukocytic a d h e r e n c e to luminal e n d o t h e l i u m t h r o u g h out the injection h e m i s p h e r e was seen at 4 h postinjection. T h e extent o f this a d h e r e n c e increased, so that at 12 and 24 h vessels t h r o u g h o u t the ipsilateral and contralateral h e m i s p h e r e s s h o w e d n u m e r o u s a d h e r e n t m o n o c y t e s (Fig. 1). B y 48 h postinjection, neutrophils r e p l a c e d m a c r o p h a g e s as the p r e d o m i n a n t a d h e r e n t cell type in vessels o f b o t h h e m i s p h e r e s , and the n u m b e r o f vessels displaying these l e u k o c y t e s was greater t h a n seen at previous time points as well (Fig. 2). T h r e e to 7 days following injection o f r T N F - a , leukocytic a d h e r e n c e declined and b e c a m e almost u n d e t e c t a b l e and w h e n observed, was only in the i m m e d i a t e vicinity of the injection site.
96
Fig. 1. Monocytic adherence to vessel in contralateral basal ganglia 12 h following recombinant tumor necrosis factor-c~ (rTNF-c~) injection, x 400 Fig. 2. Vessel in neuropil surrounding injection site, 48 h following rTNF-a, showing adherence (small arrow) and cuffing (large arrow) of neutrophils. Also apparent are a few, scattered neutrophils outside of the perivascular space within the neuropil (circle). x 400 Fig. 3. Infiltrate composed mainly of erythrocytes within injection site of excipient-injected animals; 48 h postinjection, x 85 Fig. 4. Massive neutrophil infiltration into injection site of rTNF-a recipients; 48 h following injection. • 65 Fig. 5. Edge of neutrophil infiltration site in rTNF-c~ recipients. The limit of neutrophil migration into the tissue is rather abrupt and corresponds with the resumption of normal neuropil cytoarchitecture, x 300
In e xcipient-injected animals, a low level of vascular cuffing was seen f r o m 4 to 4 8 h in the injection hemisphere only c o m p o s e d of m a c r o p h a g e s and some neutrophils, and no cuffing was detectable by 3 days.
Vascular cuffing in r T N F - ~ recipients exceeded that seen in those receiving an excipient injection (Fig. 2). O n day 2, this cuffing was observed throughout the injection hemisphere and consisted mainly of neutrophils with considerably less neutrophilic infiltration into surrounding neuropil, while on day 3 a combination of neutrophils and macrophages were observed around vessels throughout the hemisphere.
97
Leukocytes within the injection site of excipientinjected animals were composed of low to moderate levels of neutrophils and macrophages after 4 to 48 h (Fig. 3). The cell composition within the injection site changed to predominately macrophages at 3 days postinjection, and at 5 and 7 days, no leukocytes were observed. In rats receiving rTNF-(,, however, a large area of densely concentrated neutrophils was observed at the injection site 48 h following administration (Fig. 4). The area of neutrophil infiltration was limited to the edematous tissue surrounding the injection site with only scattered neutrophils in the surrounding normal neuropil. Upon closer examination of these infiltrated areas, a fairly sharp transition was observed between infiltrated tissue and surrounding normal neuropil (Fig. 5). The size of this reaction clearly surpassed that seen in excipient animals at this or any other time point. To quantitate the inflammatory response observed in histological studies following single injections into normal brain, brain samples were analyzed for MPO content at time points ranging from 4 h to 3 days following injection with either 6 x 104 U rTNF-~ or excipient. As evidenced by the MPO/protein assay, no significant difference was seen between animals receiving rTNF-(* or excipient at any time point, although a trend towards increased tissue MPO was seen at 48-h post-rTNF-(, injection (Fig. 6). Histological examination of sham controls for multipie-injection models indicated an increase in collagen fibers and ground substance as well as proliferating fibroblasts and moderate edema in and around the needle entry site. Leukocytes were limited to a small number of macrophages in the immediate area of the entry site. Both rTNF-~ and excipient recipients shared these findings, but in addition, a compact area of dense neutrophil accumulation was observed at the cannula implantation site. Histological descriptions for all multiple injection animals are superimposed on these findings. Twenty-four hours following the last of the three daily injections of 6 x 104 U rTNF-(,, there was extensive leukocytic adhesion to endothelium and cuffing around cerebral vessels of the injection site and within the basal ganglia similar in appearance to that seen 3 days following a single rTNF-(, injection. This was accompanied in the majority of animals by a dense infiltration of macrophages and lymphocytes into the neuropil at the injection site (Fig. 7). Animals given multiple IC injections of excipient as well as sham controls showed some leukocytic infiltration, but in comparison to multiple rTNF-(x recipients, leukocytic adhesion to and cuffing around cerebral vessels was greatly reduced or absent. Discussion
This study examined the effects of single and multiple injections of human rTNF-(~ directly into normal brain tissue to assess its role in mediating inflammatory responses within the CNS and potential as an immuno-
120 c
O_ 03
[] []
rTNF Excip.
90
>, >
~5
60
0n 09
30
c
L
4h
6h
12h
24 h
d
3d
Fig, 6. Quantification of inflammatory response to ICinjection of either excipient or rTNF-c~. Although assay supports histological data, showing greater tissue MPO content 48 h following rTNF-a injection, no significant differences are seen between rTNF-c~ and excipient groups
Fig. 7. Large-scale infiltration of macrophages (large arrows) and lymphocytes (small arrows) into injection site 24 h after the last of three daily injections of rTNF-0t, x 450
therapeutic agent against glioma. While much has been done in recent years to elucidate the wide range of activities of TNF-~ in systemic infections [24, 27], only lately has attention focused on its effects in the CNS [13, 28, 33]. In these latter studies, IV or intracerebroventricular (ICV) injections of rTNF-c~ were made to examine in more detail its suspected toxicity for neural tissue, since the neurotoxic effects of TNF
