Immune-Based Therapy for Spinal Cord Repair: Autologous Macrophages and Beyond
MICHAL SCHWARTZ1 and ETI YOLES2
ABSTRACT
Spinal cord injury is a devastating condition of the central nervous system (CNS), often resultingin severe loss of tissue, functional impairment, and only limited repair. Studies over the last fewyears have shown that response to the insult and spontaneous attempts at repair are multiphasicprocesses, with varying and sometimes conflicting requirements. This knowledge has led to novelstrategies of therapeutic intervention. Our view is that a pivotal role in repair, maintenance, heal-ing, and cell renewal in the CNS, as in other tissues, is played by the immune system. The modeand timing of intervention must be carefully selected, however, as the capacity of the CNS to toler-ate local repair mechanisms is limited. Studies have shown that the spontaneously evoked early in-nate response to CNS injury is characterized by invasion of neutrophils and is unfavorable for cellsurvival. This is followed by a response of the resident innate immune cells (microglia), which how-ever cannot supply all the needs of the damaged tissue; moreover, once evoked, and for as long asthe damage persists, the microglial response remains beyond the capacity of the CNS to tolerate it.Immune-based clinical intervention is most effective in improving functional and morphological re-covery when delayed for a certain period. Effective intervention might be in the form of (1) localinjection of “alternatively activated” macrophages, (2) systemic injection of dendritic cells specificto CNS antigens, or (3) T-cell–based vaccination. The treatment of choice depends on the severityof the insult, the site of injury, the therapeutic window, and safety considerations.
1Department of Neurobiology, Weizmann Institute of Science, Rehovot, Israel.2Proneuron Biotechnologies, Kiryat Weizmann, Ness Ziona, Israel.
INTRODUCTION
SPINAL CORD INJURY (SCI) often has a devastating out-come. This is because axons severed by the primary
insult often degenerate all the way back to the cell bod-ies, causing their eventual death, and the degenerativeprocess, abetted by neurotoxins emitted by the degen-
erating tissues, continues to progress through the adja-cent tissues, causing further emission and spread oftoxic substances. The escalating losses incurred as a re-sult of this “secondary degeneration” cause further dam-age to the cell bodies and axons that survived the pri-mary insult (Crowe et al., 1997; Park et al., 2004). Thecumulative loss, moreover, is normally irreversible, ow-
ing to the virtual absence of regeneration of new axonsfrom the cell bodies of damaged neurons, and little orno neurogenesis.
The role of immune cells in degenerative conditions hasnot yet been fully elucidated. The immune system isknown to protect the body’s tissues from damage and healthem after injury, irrespective of whether the potential oractual aggressors are mechanical, chemical, or biologicalin nature, although different situations might vary in termsof the types and numbers of participating immune cells,as well as in their timing and mode of action (Lawrence,1998). From an immunological point of view, however,the central nervous system (CNS) enjoys a privileged sta-tus (Moalem et al., 2000). Partly because of immune priv-ilege, the traditional belief, based on interpretations ofwell-documented observations, is that immune activity inthe damaged CNS is harmful and must therefore be elim-inated or suppressed (Neumann and Wekerle, 1998). Nev-ertheless, the CNS does possess resident immune cells(microglia), which become activated in response to acuteadverse conditions and remain activated if the adversecondition becomes chronic. The question then arises:should their activation be suppressed, or is it an appro-priate and desirable response (Nakajima and Kohsaka,2001; Nguyen et al., 2002; van Beek et al., 2003)?
Studies carried out in our laboratories have shown thatsuitably activated macrophages can contribute positivelyto the homeostasis of the CNS, and that recruitment ofadaptive immunity from the periphery is part of a phys-iological repair mechanism that needs, however, to berigorously controlled in terms of the timing and intensityof the cellular response. In the following, we summarizethe data showing how (1) macrophages, (2) T cells, and(3) dendritic cells contribute to CNS repair.
MACROPHAGE THERAPY
Up to about 10 years ago, activated macrophages weredefined simply as cells that secrete inflammatory medi-ators and kill intracellular pathogens. Data accumulatedover the last decade suggest, however, that monocytesare multi-talented cells that are capable of expressing dif-ferent functional programs in response to distinct micro-environmental signals (Nagorsen et al., 2005). Microbialproducts and cytokines profoundly affect the differenti-ation of monocytes towards two phenotypic extremes.Microbial products are associated with the “classical” ac-tivation of monocytes, which turns them into potent ef-fector cells that kill microorganisms and tumor cells. Incontrast, macrophages (which promote angiogenesis andare involved in tissue remodeling and repair) are oftenviewed as “alternatively” activated macrophages (Bom-
stein et al., 2003; Gordon, 2003; Hauben et al., 2003;Klusman and Schwab, 1997; Mantovani et al., 2002;Mosser, 2003; Rothwell and Strijbos, 1995; Butovsky etal. 2005; Butovsky et al. 2006).
The results of our early studies, in which we comparedthe characteristics of regenerative and non-regenerativeneural tissues, prompted us to examine the possibility thattimely and appropriate boosting of the local immune re-sponse has a beneficial effect on post-traumatic CNS re-covery. Initial experiments in rats with completely tran-sected optic nerves or spinal cords demonstrated that localapplication of macrophages preincubated with fragmentsof sciatic nerve (a peripheral nerve, and capable of re-generation) promotes motor recovery (Lazarov-Spiegler etal., 1996; Rapalino et al., 1998), the first signs of whichare detectable not earlier than 4 weeks after the insult. Sim-ilar results were obtained by Yin et al. (2003), who showedthat macrophages activated by intravitreal injections ofZymosan, a yeast cell-wall preparation, induced regener-ative growth of axotomized rat retinal ganglion cell (RGC)axons into the distal optic nerve stump. The same effectwas obtained with media conditioned by incubation withthe activated macrophages (Yin et al., 2003). Likewise,macrophages stimulated by group B–Streptococcus exo-toxin after an optic nerve injury showed increased phago-cytosis of inhibitory debris; they also exhibited less densereactive gliosis, which enabled axons to regrow throughthe glial scar (Ohlsson et al., 2004).
The experiments of local injection of “alternatively”activated macrophages were repeated in a model of se-vere spinal cord contusion, using blood-borne monocytesactivated by preincubation with autologous skin (Bom-stein et al., 2003). In these and subsequent experiments,the macrophage phenotypes were characterized, andparameters such as the site of injection, the dosage reg-imen, and the therapeutic window were assessed.
Macrophage Characteristics
Skin-coincubated macrophages that promote spinalcord recovery express the cell-surface markers CD80,CD86, and CD54, as well as class II major histocompat-ibility complex molecules (MHC-II), all of which arecharacteristic of antigen-presenting cells (APCs). In ad-dition, they secrete brain-derived neurotrophic factor(BDNF), but not the cytokine tumor necrosis factor(TNF)–�. The abundant presence of BDNF coupled withthe absence of TNF-� is suggestive of beneficial neuro-protection (Bomstein et al., 2003). Similarly, it wasshown that rat cortical neurons cultured in conditionedmedia from human monocyte-derived macrophages(MDM) exhibit increased neuronal protein synthesis,neurite outgrowth, mitogen-activating protein kinase ac-
IMMUNE-BASED THERAPY FOR SPINAL CORD INJURY
361
tivity, and synaptic function. The neurotrophic propertiesof human MDM-conditioned media were significantlyenhanced by stimulation human peripheral nerve and,though to a more limited extent, by pre-stimulation withCD40 ligand. The positive effects of MDM secretion onneuronal function resemble those obtained by secretionof BDNF (Shibata et al., 2003).
The cellular features of blood-borne macrophages co-cultured with fragments of skin are in many aspects rem-iniscent of those of microglia activated by adaptive im-munity, and unlike those of the classically activatedmacrophages or microglia (activated by lipopolysaccha-ride [LPS] of bacterial cell walls), as discussed below(Butovsky et al., 2005, 2006a; Shaked, 2005).
Experimental activation of rat macrophages by LPS ischaracterized by a dramatic increase in TNF-� produc-tion (Young et al., 2001), and local injection of zymosaninto the healthy rodent spinal cord induces uncontrolleddestructive inflammation (Popovich et al., 2002). Inreaching their conclusions about the effects of activatedmicroglia, however, those authors made no distinctionbetween the microglial activation that is related spe-cifically to inflammation and other types of microglialactivation. A recent study by our group demonstratedthat well-controlled levels of T cell–derived cytokinesrender microglia of a phenotype that is beneficial toneural tissue (Butovsky et al., 2005; Shaked et al., 2005).Yet, uncontrolled levels of cytokines such as IFN-�might lead to cytotoxic effects (Butovsky et al., 2005;Butovsky et al., 2006a, 2006b). Moreover, and again incontrast to activation by LPS or zymosan, injection oflarge doses of skin-coincubated macrophages into un-damaged spinal cords of rats caused no apparent anom-alies in clinical or histopathological parameters (internaldata from animal safety studies by Harlan Biotech Israeland Proneuron Biotechnologies). It is therefore mis-leading to suggest that “macrophage activation” neces-sarily implies inflammation (Hauben and Schwartz,2003; Schwartz and Hauben, 2002). Studies by several
research groups have pointed out that activated macro-phages promote CNS regeneration (Ohlsson et al., 2004;Shibata et al., 2003).
When and Where Are Macrophages Needed?
In studies aimed at establishing the optimal time forintervention using macrophages it became clear that, asin any other tissue, repair and restoration in the CNS aredependent not only on context but also on timing. Thefollowing time windows were examined, each represent-ing a different post-injury physiological stage (Fig. 1):(a) 3-4 days after SCI, a period characterized by a de-cline in primary infiltration of neutrophils participatingin inflammation and a high incidence of apoptotic cells(Leskovar et al., 2000; Liu et al., 1997; Popovich et al.,1997; Yong et al., 1998); (b) 7–10 days after SCI, a pe-riod of maximal proliferation and/or accumulation ofED1-positive cells (activated microglia/macrophages), Tcells, and progenitor glial cells (Leskovar et al., 2000;Liu et al., 1997; McTigue et al., 2001); (c) 14 days afterSCI, when the numbers of ED1-positive cells and T cellsare still very high, while cytokines and chemokines in theinjured tissue are decreasing or disappearing (Lee et al.,2000; Leskovar et al., 2000; Liu et al., 1997; McTigue etal., 2001); and (d) 21 days after SCI, by which time manyof the injury-induced biochemical and cellular activitiesin the spinal cord have peaked and begun to return to nor-mal (Leskovar et al., 2000; McTigue et al., 2001).
Table 1 records the effects of macrophage treatmentadministered to rats at different times after contusive SCI.On the indicated days post-injury, each rat received a sin-gle injection, at the caudal border of the lesion, of0.2–0.25 � 106 macrophages, which had been purifiedfrom spinally contused donor rats and preincubated withskin. Rats treated on days 8–9 after injury showed sig-nificant recovery compared to those in the control group(one-tailed Fisher exact test, p � 0.014). Timing of thetreatment was found to be important for successful func-
SCHWARTZ AND YOLES
362
FIG. 1. Phases in the immune response to spinal cord injury.
tional recovery of the damaged CNS. Because the im-mediately post-traumatic microenvironment of the spinalcord is at an acute pro-inflammatory stage, it is an unfa-vorable time for survival and differentiation of the donormacrophages. At the chronic stage, on the other hand,formation of glial scars in the injured site inhibits re-generation of neuronal axons. Thus, we believe that theoptimal timing of transplantation is 1–2 weeks after theinjury. This schedule is in line with the optimal time fortransplantation for stem-cell and other cellular therapies(Iwanami et al., 2005; Koshizuka et al., 2004).
The efficacy of the local treatment with activatedmacrophages, in terms of recovery of motor functions.was found to be dependent on the site of their injection;injection close to the caudal margin of the contusion sitewas beneficial (approximately 60% of the treated animalsshowed motor scores of �6, up to 8–9 in the BBB scale;the control animals showed BBB values below 6, andonly 30% achieved values of 6), whereas injections oneor three segments below that level resulted in no signif-icant improvement in recovery (Bomstein et al., unpub-lished data).
Effect of Macrophage Implantation on Post-SCI Histopathology
Post-traumatic syringomyelia (cysts in the central partof the spinal cord) develops in more than 20% of patientswith SCI, especially after injuries to the thoracic or lum-bar spine. Clinical symptoms (radicular pain, spasticity,sensory loss, hyperhydrosis, and weakness) sometimesappear several months to many years after the injury, andsurgical treatment is usually needed to drain the cysts.
The pathogenesis of cyst formation is not fully under-stood; however, studies point to the possible participa-tion of an inflammatory response (Fitch et al., 1999).
Figure 2 shows representative micrographs of longitu-dinal sections of rat spinal cords that were untreated ortreated with activated macrophages after contusive SCIand excised 5–6 months later. Cavities caudal and rostralto the site of injury are seen in both sections, but are muchsmaller in the macrophage-treated spinal cord.
Quantitative analysis of areas occupied by cysts mea-sured in spinal cord sections (three sections per rat) offive macrophage-treated and nine control rats discloseda statistically significant difference (t-test for unequalvariances, p � 0.024) in mean cyst areas between the twogroups. An inverse significant (p � 0.012) relationshipwas found between cavity size and motor scores in theanalyzed animals.
The experiments summarized above, together withother recently reported studies, suggest that the effect ofwell-controlled activated macrophages on the injuredCNS tissue is reminiscent of wound repair. Thesemacrophages clear the lesion site of cellular debris andproduce the growth factors and enzymes known to beneeded for wound repair.
ROLE OF ADAPTIVE IMMUNITY IN CNS REPAIR: T CELLS
Macrophages found to be beneficial bear a resem-blance to antigen-presenting cells (APCs) (Bomstein etal., 2003), and express a unique phenotype that differsfrom that of LPS-activated macrophages. In the course
IMMUNE-BASED THERAPY FOR SPINAL CORD INJURY
363
TABLE 1. RECOVERY OF RATS AFTER SPINAL CORD CONTUSION: EFFECT OF MACROPHAGE
TREATMENT ADMINISTERED AT DIFFERENT TIMES AFTER INJURY
Day of Number of Statisticaltreatment (after Number of recovered rats differencespinal cord rats in (% animals with (Fisher exact t-contusion) Treatment group BBBc score �5) test, one-tailed)
4 days Control 13 2 (15.4%) NoneMacrophages 7 1 (14.3%)
8–9 days Control 94a 24 (25.5%) p � 0.014Macrophages 18 10 (55.6%)
14 days Control 15b 3 (20.0%) NoneMacrophages 14 3 (21.4%)
20–21 days Control 8b 1 (12.5%) NoneMacrophages 10 0 (0.0%)
aThe large number of controls is attributable to the wide use of this particular control treatment (vehicle injection caudal to injurysite at 8–9 days) in a number of different experiments.
bThese numbers include non-injected and PBS-injected rats.cBBB, locomotion scale of Basso, Beattie and Bresnahan (Basso et al., 1995).
of our studies of the role of the immune system in thepost-injury recovery of the CNS, we discovered that theperipheral immune system (traditionally viewed as beingaffected only in a passive way by CNS injury) in factplays an active role in CNS repair and is an integral partof it. This discovery led us to embark on the series ofstudies that culminated in our formulation of the conceptof “protective autoimmunity” (Moalem et al., 1999). Ac-cording to this concept, T cells directed to specific CNSantigens, by locally controlling the activity of the resi-dent microglia, play a central role in the physiologicalprocesses of CNS protection and repair. Moreover, den-dritic cell–based treatment can recapitulate the macro-phage effect (Hauben et al., 2000a). In the following, weoutline the steps that led us to this view (Schwartz andKipnis, 2002).
Our unexpected findings in connection with macro-phages prompted us to consider the almost heretical pos-sibility that the injured CNS, like any other tissue, can be healed by the systemic immune system. We postulatedthat just as the innate immune system (represented bymacrophages and microglia) appears to be recruited as abeneficial player after CNS injury, the adaptive immunesystem is recruited as a means of augmenting and regu-lating the local immune response. Our studies in rodentmodels confirmed that systemic injection of T cells di-rected to CNS antigens such as myelin basic protein(MBP), unlike T cells directed to non-self antigens, pro-mote recovery from both optic nerve injury and contusiveSCI (Hauben et al., 2000a,b; Moalem et al., 1999). Othershave shown that vaccination of adult rats with spinal cordhomogenate can also promote regeneration of RGCs aftermicrocrush lesion of the optic nerve (Ellezam et al., 2003).
In subsequent studies, recovery from SCI was foundto be impaired in immune-deficient rats (Hauben et al.,2002), and was critically affected by the timing and in-tensity of the autoimmune T-cell response and its regu-lation (Kipnis et al., 2002a,b; Shaked et al., 2004). Theseand other studies demonstrated that the observed T-celleffect is a spontaneous physiological response, but thatits effect is limited, either because it is insufficient or be-
cause it is not optimally controlled. Autoimmune effec-tor cells with neuroprotective activity in acute CNS in-juries were identified as CD4� T-helper (Th1) cells—that is, the very cells which, under different conditionsof timing and intensity (Kipnis et al., 2002a,b; Moalemet al., 1999), cause an autoimmune disease. In all cases,recovery could be boosted by passive or active vaccina-tion with CNS antigens residing in the site of injury, pro-vided that both antigen and adjuvant were carefully se-lected (Fisher et al., 2001; Hauben et al., 2000a,b; 2001).If the vaccination led to an excessive T-cell responsecausing severe symptoms of experimental autoimmuneencephalomyelitis (EAE; attributable to either the anti-gen or adjuvant used), it outweighed the neuroprotectiveeffect (Hauben et al., 2001; Mizrahi et al., 2002).
In investigating potential therapeutic applications, wefound that a T-cell–dependent neuroprotective effectcould be obtained by vaccination with a weak agonist ofself-antigens (e.g., an altered peptide ligand, such as asegment of MBP comprising amino acids 87–99) resid-ing in the site of damage (Hauben et al., 2001).
The observed T-cell–dependent recovery is manifestedby better preservation of tissue, survival of more axons,and reduced scar formation, suggesting that the thera-peutic activity affects the secondary degeneration thatfollows a partial injury (Butovsky et al., 2001; Haubenet al., 2003). In a model of complete spinal cord tran-section, unlike in cases of severe contusion, passive trans-fer of T cells directed to MBP did not lead to improvedrecovery of motor scores (Hauben et al., 2000a,b). Thefailure to benefit from the T-cell–based therapy in casesof a complete transection might, however, reflect a tech-nical problem related to the transection paradigm, ratherthan a failure to arrest retrograde degeneration or to in-duce sprouting following a complete transection.
Further studies partially uncovered the mechanism un-derlying the T cell effect. After an insult to the CNS,homing of T cells to the affected site appears to be anessential first step, in which the specificity of the T cellsthat accumulate at the site of damage is dictated by theparticular self-antigens that reside there. Upon their ar-
SCHWARTZ AND YOLES
364
FIG. 2. Spinal cord sections (20 mm long, stained with Sudan black) from an untreated control rat (top) and from a rat treatedwith activated macrophages (bottom).
rival at the site of injury, these autoimmune T cells en-gage the resident microglia in a dialog that results in ac-tivation of the microglia in a way that the CNS tissue canapparently tolerate (Butovsky et al., 2005; Shaked et al.,2005; Butovsky et al., 2006a, 2006b; Ziv et al., 2006).Accordingly, the activated microglia demonstrate an abil-ity to act as APCs, produce growth factors, and scavengeneurotoxins such as excessive quantities of glutamate(Butovsky et al., 2001, 2005; Shaked et al., 2005). Suchactivities distinguish these microglia from the classicalmicroglia activated by microorganisms. More recentfindings strongly suggested that microglia which are ac-tivated by cytokines derived from T cells (both controlledlevels of Th1 and Th2) not only support neuronal sur-vival, but also promote neurogenesis and oligodendroge-nesis, as well as axonal sprouting, from adult neural stemcells (Butovsky et al., 2006a, 2006b).
The ability of an injured nerve to manifest a T-cell–me-diated response specific to CNS antigens depends on therelationship between the availability of autoimmune T cells and their control by naturally occurringCD4�CD25� regulatory T cells (Kipnis et al., 2004a,b,c;Kipnis et al., 2002a,b). It was further found that the reg-ulatory T cells are themselves controlled by brain-derivedpeptides (Kipnis et al., 2004a,b,c). The therapeutic win-dow for T-cell–based vaccination apparently allows a de-lay in treatment of at least 1 week for passive vaccina-tion and approximately 5 days for active vaccination(Hauben et al., 2000a,b; 2001), and can therefore ac-commodate immediate treatment modalities, such assteroids or blocking of neutrophil infiltration, even if theyare unsuited to the tissue’s subsequent needs (Ibarra etal., 2004).
DENDRITIC CELLS AS ANTIGEN-PRESENTING CELLS IN CNS REPAIR
A considerable body of literature assigns a key role todendritic cells (DCs) in promoting and modulating im-mune responses (Knight et al., 2002; Link et al., 2001).DCs are immune cells whose principal function is anti-gen presentation. They have an extraordinary capacity tostimulate naive T cells, control the quality of the T cellresponse, and initiate primary immune responses (Mell-man and Steinman, 2001). Their effects vary from con-ferring active autoimmunity to conferring immune toler-ance, and they are capable of bringing about changes inT-cell polarization (Dittel et al., 1999; Turley, 2002; Xiaoet al., 2001).
The diverse activities of DCs in immune regulation area function of the diversity of DC subsets and lineages aswell as the functional plasticity of DCs while still im-
mature (Liu, 2001). The state of maturation of these cells,as well as their numbers and the context in which theyare activated, determines the nature of the resulting im-mune response. Three distinct stages of DC differentia-tion were recently described, and it was suggested thattolerance is conferred when the DCs are semi-mature,whereas only fully mature DCs are immunogenic. Thedecisive signal that induces a T-cell–mediated immuneresponse seems to be the expression of CD86 (B7-2) andMHC-II molecules concurrently with the release of proin-flammatory cytokines from the DCs, in particular inter-leukin (IL)–12, IL-6, and TNF-� (Lutz and Schuler,2002).
We postulated that the use of DCs pulsed with a myelinpeptide, such as a peptide of MBP, might provide a wayto harness the immune system and exploit its functionsfor both protection and regeneration of the injured spinaltissue (Hauben et al., 2000a,b; Rapalino et al., 1998). Inan attempt to secure a beneficial outcome while reduc-ing the risk of accompanying autoimmune disease, wealso tried pulsing DCs with an altered peptide ligand, asegment of MBP (amino acids 87–99) in which the aminoacid lysine in position 91 is replaced by alanine. Thismodified peptide (A91) was found to cross-react with theoriginal encephalitogenic peptide, activating weak self-reacting T cells, and thereby inducing autoimmunitywithout the risk of inducing EAE (Gaur et al., 1997).
After severe contusive SCI, injured rats treated withDCs pulsed with either MBP peptide 87–99 or A91showed significantly improved recovery. Recovery (val-ues of motor scores in the BBB scale of 8 and 9 relativeto control values of 3–4) was detectable as early as 11days after the injury and was evident in 70–80% of thetreated rats. The recovery of rats treated with non-pulsedDCs did not differ significantly from that of PBS-treatedrats or of rats treated with DCs pulsed with ovalbumin,a non-CNS antigen.
Histological analysis of spinal cords excised 3.5months after severe contusive SCI and local injection ofDCs pulsed with MBP peptide 87–99 or PBS injectionrevealed significantly better tissue preservation in theDC-treated rats than in the controls, manifested by lesscavitation and smaller sites of injury. Lesion sites in thetreated rats were significantly smaller (by two- to three-fold) than in the controls. Quantification of cyst areas re-vealed significant differences, suggesting that treatmentwith DCs pulsed with the MBP peptide ameliorates sy-ringomyelia. Furthermore, as shown by Mikami et al.(2004), DCs implanted into the injured spinal cords ofadult mice activated endogenous neural stem/progenitorcells for mitotic de novo neurogenesis. These DCs pro-duced neurotrophin-3 and activated endogenous mi-croglia in the injured spinal cord. Behavioral analysis
IMMUNE-BASED THERAPY FOR SPINAL CORD INJURY
365
disclosed significantly better recovery of locomotor func-tions in DC-implanted mice than in controls (Mikami etal., 2004). Similar inhibitory effects of cavity and scarformation, as well as extensive axonal regeneration, wereinduced by olfactory ensheathing cells transplanted intothe injured spinal cord (Ramer et al., 2004).
Systemic Treatment with Dendritic Cells
Because the DCs were found to be mature and theirmechanism of action is T-cell dependent, it was of in-terest to determine whether their beneficial effect on re-covery could be reproduced by their systemic adminis-tration. Male SPD rats were subjected to severe contusiveSCI and immediately afterward received A91-pulsed DCsby intravenous (i.v.) injection. Starting 15 days after theinjury, a significant effect on recovery was observed inconsecutive assessments on an open-field locomotor test(Hauben et al., 2003b). The extent of recovery was notsignificantly greater than in rats injected via other routesof treatment, but more rats showed significant recoverythan with other routes. Injection of A91-pulsed DCs givenas late as 12 days after SCI was not less effective thanan immediate injection, but no benefit was obtained whenthe injection was delayed for 28 days. On day 11 (i.e.,the day before rats received a delayed injection), theirmobility scores reflected complete paralysis. The ob-served improvement led us to suggest that the DC-in-duced effect on functional recovery reflects both neuro-protection and sprouting (Gothilf and Schwartz,unpublished data).
Dendritic Cell Studies
Locally or systemically administered DCs loaded withmyelin-derived peptide or with altered peptide ligandhave a beneficial effect on recovery from SCI. Interest-ingly, the therapeutic window was found to be wide, sim-ilar to that found in the case of macrophage implantationor T cell vaccination, suggesting that the mechanisms un-derlying all three of these interventional techniques mightconverge into a common pathway.
CONCLUSION
The CNS has a unique relationship with the immunesystem. It was believed that the blood–brain barrier prevents peripheral immune cells from entering the CNS, and that the consequences of destruction of theblood–brain barrier are therefore detrimental. Moreover,activated microglia, the brain-resident immune cells, areusually found in sites of neurodegeneration and weretherefore thought to be part of the pathology. Apparently
in line with this perception was the observation that anti-inflammatory drugs reduce the functional deficit afterSCI if administered within few hours after the injury (Griset al., 2004). If such treatment was given later, however,it was not effective and even had adverse results (Robertset al., 2004). Injection of skin-activated macrophages orskin-pulsed DCs was more effective if given after a de-lay of a few days rather than immediately after SCI (Bom-stein et al., 2003). This apparent inconsistency can be ex-plained in terms of the varying needs of the tissue atdifferent phases of recovery. It appears to be possible toexploit the different timing and duration of these two po-tentially conflicting immune effects in a therapy thatcombines two strategies—i.e., by administering anti-in-flammatory treatment at an early stage after SCI and ac-tivated macrophages or DCs later (Fig. 1).
The different phases of neuronal loss and physiologi-cal attempts at repair after SCI correlate well with the na-ture of the infiltrating immune cells and the local immuneresponse observed over time (Okano et al., 2003). Neu-trophils are the main participants during the hyper-acutephase, activated macrophages/microglia in the subacutephase, and the T-cell–mediated adaptive immune re-sponse then takes over to orchestrate the ongoing activ-ities of macrophages, microglia, cytokines, and growthfactors. During the first phase of the innate immune re-sponse, neutrophils (which secrete cytotoxic elementssuch as NO, free radicals, and COX-2) reach peak activ-ity several hours after the injury and disappear withinabout 2 days. Although early inflammatory responsesprobably contribute to secondary degeneration, delayedinflammatory events are likely to be reparative. Innateimmune activity at the sub-acute stage is characterizedby an accumulation of activated microglia/macrophagesat the lesion site. These cells, if properly activated, canremove toxic elements and dying cells and restore home-ostasis. They also serve as APCs, presenting antigens totheir specific autoimmune T cells that home to the dam-aged tissue (Nakajima and Kohsaka, 2004). The T cells,activated by this encounter, secrete growth-promoting cy-tokines and neurotrophic factors (Moalem et al., 1999;Butovsky et al., 2006b; Ziv et al., 2006). Only those T cells that encounter relevant antigens presented to themby APCs can be locally activated to produce the neu-rotrophic factors needed for neuronal survival, and thecytokines and growth factors needed for shaping the mi-croglial response. Microglia surrounding injured neuronswere found to express some co-stimulatory moleculesfrom the B7 family at 1–4 days after injury; later, the ex-pression became particularly intense and peaked at day14 (Bohatschek et al., 2004). These late microglialchanges were accompanied by invasion of T-lympho-cytes and a marked increase in the expression of proin-
SCHWARTZ AND YOLES
366
flammatory cytokines such as IL-1�, TNF-�, and inter-feron (IFN)–� (Raivich et al., 1998). In the absence ofproperly activated T cells, which serve as a source ofcytokines, the local microglia might not develop intoAPCs, or at least not at the right time. The ability to spon-taneously manifest a T-cell-mediated autoimmune re-sponse is rigorously controlled by naturally occurringCD4�CD25� regulatory T cells, which in turn are con-trolled by neuropeptides in the brain (Kipnis et al.,2004a,b,c). Therefore, the beneficial T-cell–dependent ac-tivity can be boosted by vaccination with relevant anti-gens or by weakening of the suppressive activity of thenaturally occurring regulatory CD4�CD25� T cells (Kip-nis et al., 2004a,b,c; Kipnis et al., 2002a,b). It can alsobe boosted by homeostasis-driven lymphopenia inducedby weak lymphoid irradiation (Kipnis et al., 2004a,b,c).
It thus appears that several immune-based manipula-tions can yield a beneficial response. In choosing a strat-egy, both the therapeutic window and safety parametersmust be taken into account. The main characteristics ofDCs and skin-activated macrophages are their expressionof MHC-II and co-stimulatory molecules, which are nec-essary for their function as APCs. Injection of DCs orskin-activated macrophages into the damaged spinal cordat the optimal time after injury triggers the autoreactiveT cells to secrete cytokines, which induce the residentcells to support tissue repair.
Intravenously administered DCs can act like an adju-vant, augmenting the injury-induced T cell response inthe periphery and hence increasing their accumulation atthe lesion site. Clinically, the optimal therapeutic timewindow for such intervention in cases of complete tran-section of the spinal cord is 7–10 days after the injury.The time between harvesting of autologous macrophagesand their injection after activation by skin does not ex-ceed 2 days, and the time required to prepare autologousDCs is approximately 7 days. From a practical point ofview, therefore, activated macrophages have the advan-tage over dendritic cells in terms of clinical applicabilityat the optimal time window. If the SCI is partial, activevaccination or systemic injection of antigen-specific den-dritic cells might be superior to macrophage injection.This possibility is currently under investigation. Our re-sults suggest, moreover, that the degenerative environ-ment, which impairs neuronal survival and regrowth, alsoimpairs neurogenesis and oligodendrogenesis (Butovskyet al., 2005; Butovsky 2006a, 2006b). Therefore, im-munomodulation combined with stem-cell therapy mightbenefit from a significant synergistic effect: the im-munomodulation would not only augment survival andregrowth, but would also support the homing and differ-entiation of neural stem cells, which would then furtherboost recovery by contributing to tissue restoration.
ACKNOWLEDGMENTS
We thank S.R. Smith for editing the manuscript. M.S.holds the Maurice and Ilse Katz Professorial Chair inNeuroimmunology. This work was supported in part byProneuron Biotechnologies Ltd. (Kiryat Weizmann,Ness-Ziona, Israel).
REFERENCES
BASSO, D.M., BEATTIE, M.S. and BRESNAHAN, J.C.(1995). A sensitive and reliable locomotor rating scale foropen field testing in rats. J. Neurotrauma 12, 1–21.
BETHEA, J.R., and DIETRICH, W.D. (2002). Targeting thehost inflammatory response in traumatic spinal cord injury.Curr. Opin. Neurol. 15, 355–360.
BOHATSCHEK, M., KLOSS, C.U., PFEFFER, K., BLUETH-MANN, H., and RAIVICH, G. (2004). B7.2 on activated andphagocytic microglia in the facial axotomy model: regula-tion by interleukin-1 receptor type 1, tumor necrosis factorreceptors 1 and 2 and endotoxin. J. Neuroimmunol. 156,132–145.
BOMSTEIN, Y., MARDER, J.B., VITNER, K., et al. (2003).Features of skin-coincubated macrophages that promote re-covery from spinal cord injury. J. Neuroimmunol. 142,10–16.
BUTOVSKY, O., HAUBEN, E., and SCHWARTZ, M. (2001).Morphological aspects of spinal cord autoimmune neuropro-tection: colocalization of T cells with B7–2 (CD86) and pre-vention of cyst formation. FASEB J. 15, 1065–1067.
BUTOVSKY, O., ZIV, Y., SCHWARTZ, A., et al. (2006a).Microglia activated by IL-4 or IFN-gamma differentially in-duce neurogenesis and oligodendrogenesis from adultstem/progenitor cells. Mol. Cell. Neurosci. 31, 149–160.
BUTOVSKY, O., LANDA, G., KUNIS, G., et al. (2006b). In-duction and blockage of oligodendrogenesis by differentlyactivated microglia in an animal model of multiple sclerosis.J. Clin. Invest. (in press).
BUTOVSKY, O., TALPALAR, A.E., BEN-YAAKOV, K., andSCHWARTZ, M. (2005). Aggregated beta-amyloid rendersmicroglia cytotoxic but T cell–derived IFN-gamma and IL-4 make them protective. Mol. Cell. Neurosci. (in press).
BUTOVSKY, O., TALPALAR, A.E., BEN-YAAKOV, K., andSCHWARTZ, M. (2005). Activation of microglia by aggre-gated beta-amyloid or lipopolysaccharide impairs MHC-IIexpression and renders them cytotoxic whereas IFN-gammaand IL-4 render them protective. Mol. Cell. Neurosci. 29,381–393.
CROWE, M.J., BRESNAHAN, J.C., SHUMAN, S.L., MAS-TERS, J.N., and BEATTIE, M.S. (1997). Apoptosis and de-layed degeneration after spinal cord injury in rats and mon-keys. Nat. Med. 3, 73–76.
IMMUNE-BASED THERAPY FOR SPINAL CORD INJURY
367
DITTEL, B.N., VISINTIN, I., MERCHANT, R.M., andJANEWAY, C.A., JR. (1999). Presentation of the self anti-gen myelin basic protein by dendritic cells leads to experi-mental autoimmune encephalomyelitis. J. Immunol. 163,32–39.
FISHER, J., LEVKOVITCH-VERBIN, H., SCHORI, H., et al.(2001). Vaccination for neuroprotection in the mouse opticnerve: implications for optic neuropathies. J. Neurosci. 21,136–142.
FITCH, M.T., DOLLER, C., COMBS, C.K., LANDRETH,G.E., and SILVER, J. (1999). Cellular and molecular mech-anisms of glial scarring and progressive cavitation: in vivoand in vitro analysis of inflammation-induced secondary in-jury after CNS trauma. J. Neurosci. 19, 8182–8198.
GAUR, A., BOEHME, S.A., CHALMERS, D., et al. (1997).Amelioration of relapsing experimental autoimmune en-cephalomyelitis with altered myelin basic protein peptidesinvolves different cellular mechanisms. J. Neuroimmunol.74, 149–158.
GORDON, S. (2003). Alternative activation of macrophages.Nat. Rev. Immunol. 3, 23–35.
GRIS, D., MARSH, D.R., OATWAY, M.A., et al. (2004). Tran-sient blockade of the CD11d/CD18 integrin reduces sec-ondary damage after spinal cord injury, improving sensory,autonomic, and motor function. J. Neurosci. 24, 4043–4051.
HAUBEN, E., and SCHWARTZ, M. (2003). Therapeutic vac-cination for spinal cord injury: Helping the body to cure it-self. Trends Pharmacol. Sci. 24, 7–12.
HAUBEN, E., BUTOVSKY, O., NEVO, U., et al. (2000a). Pas-sive or active immunization with myelin basic protein pro-motes recovery from spinal cord contusion. J. Neurosci. 20,6421–6430.
HAUBEN, E., NEVO, U., YOLES, E., et al. (2000b). Au-toimmune T cells as potential neuroprotective therapy forspinal cord injury. Lancet 355, 286–287.
HAUBEN, E., AGRANOV, E., GOTHILF, A., et al. (2001).Posttraumatic therapeutic vaccination with modified myelinself-antigen prevents complete paralysis while avoiding au-toimmune disease. J. Clin. Invest. 108, 591–599.
HAUBEN, E., MIZRAHI, T., AGRANOV, E., andSCHWARTZ, M. (2002). Sexual dimorphism in the sponta-neous recovery from spinal cord injury: a gender gap in ben-eficial autoimmunity? Eur. J. Neurosci. 16, 1731–1740.
HAUBEN, E., GOTHILF, A., COHEN, A., et al. (2003). Vac-cination with dendritic cells pulsed with peptides of myelinbasic protein promotes functional recovery from spinal cordinjury. J. Neurosci. 23, 8808–8819.
IBARRA, A., HAUBEN, E., BUTOVSKY, O., and M., S.(2004). The therapeutic window after spinal cord injury can
accommodate T cell–based vaccination and methylpred-nisolone in rats. Eur. J. Neurosci. 19, 2984–2990.
IWANAMI, A., KANEKO, S., NAKAMURA, M., et al. (2005).Transplantation of human neural stem cells for spinal cordinjury in primates. J. Neurosci. Res. 80, 182–190.
KIPNIS, J., MIZRAHI, T., HAUBEN, E., SHAKED, I., SHE-VACH, E., and SCHWARTZ, M. (2002a). Neuroprotectiveautoimmunity: naturally occurring CD4�CD25� regulatory Tcells suppress the ability to withstand injury to the central ner-vous system. Proc. Natl. Acad. Sci. USA 99, 15620–15625.
KIPNIS, J., MIZRAHI, T., YOLES, E., BEN-NUN, A., andSCHWARTZ, M. (2002b). Myelin specific Th1 cells are nec-essary for post-traumatic protective autoimmunity. J. Neu-roimmunol. 130, 78–85.
KIPNIS, J., AVIDAN, H., CASPI, R.R., and SCHWARTZ, M.(2004a). Dual effect of CD4�CD25� regulatory T cells inneurodegeneration: pro- and anti-inflammatory cytokines de-termine microglial activity. Proc. Natl. Acad. Sci. USA 101,Suppl 2, 14663–14669.
KIPNIS, J., AVIDAN, H., MARKOVICH, Y., et al. (2004b).Low-dose gamma-irradiation promotes survival of injuredneurons in the central nervous system via homeostasis-dri-ven proliferation of T cells. Eur. J. Neurosci. 19, 1191–1198.
KIPNIS, J., CARDON, M., AVIDAN, H., et al. (2004c).Dopamine, through the extracellular signal-regulated kinasepathway, downregulates CD4�CD25� regulatory T-cell ac-tivity: implications for neurodegeneration. J. Neurosci. 24,6133–6143.
KLUSMAN, I., and SCHWAB, M.E. (1997). Effects of pro-in-flammatory cytokines in experimental spinal cord injury.Brain Res. 762, 173–184.
KNIGHT, S.C., BURKE, F., and BEDFORD, P.A. (2002). Den-dritic cells, antigen distribution and the initiation of primaryimmune responses to self and non-self antigens. Semin. Can-cer Biol. 12, 301–308.
KOSHIZUKA, S., OKADA, S., OKAWA, A., et al. (2004).Transplanted hematopoietic stem cells from bone marrowdifferentiate into neural lineage cells and promote functionalrecovery after spinal cord injury in mice. J. Neuropathol. Exp.Neurol. 63, 64–72.
LAWRENCE, W.T. (1998). Physiology of the acute wound.Clin. Plast. Surg. 25, 321–340.
LAZAROV-SPIEGLER, O., SOLOMON, A.S., ZEEV-BRANN, A.B., HIRSCHBERG, D.L., LAVIE, V., andSCHWARTZ, M. (1996). Transplantation of activatedmacrophages overcomes central nervous system regrowthfailure. FASEB J. 10, 1296–1302.
LEE, Y.L., SHIH, K., BAO, P., GHIRNIKAR, R.S., and ENG,L.F. (2000). Cytokine chemokine expression in contused ratspinal cord. Neurochem. Int. 36, 417–425.
LESKOVAR, A., MORIARTY, L.J., TUREK, J.J., SCHOEN-LEIN, I.A., and BORGENS, R.B. (2000). The macrophage
SCHWARTZ AND YOLES
368
in acute neural injury: changes in cell numbers over time andlevels of cytokine production in mammalian central and pe-ripheral nervous systems. J. Exp. Biol. 203, 1783–1795.
LINK, H., HUANG, Y.M., MASTERMAN, T., and XIAO,B.G. (2001). Vaccination with autologous dendritic cells:from experimental autoimmune encephalomyelitis to multi-ple sclerosis. J. Neuroimmunol. 114, 1–7.
LIU, X.Z., XU, X.M., HU, R., et al. (1997). Neuronal and glialapoptosis after traumatic spinal cord injury. J. Neurosci. 17,5395–5406.
LIU, Y.J. (2001). Dendritic cell subsets and lineages, and theirfunctions in innate and adaptive immunity. Cell 106,259–262.
LUTZ, M., and SCHULER, G. (2002). Immature, semi-matureand fully mature dendritic cells: which signals induce toler-ance or immunity? Trends Immunol. 23, 445.
MANTOVANI, A., SOZZANI, S., LOCATI, M., ALLAVENA,P., and SICA, A. (2002). Macrophage polarization: tumor-as-sociated macrophages as a paradigm for polarized M2mononuclear phagocytes. Trends Immunol. 23, 549–555.
MCTIGUE, D.M., WEI, P., and STOKES, B.T. (2001). Prolif-eration of NG2-positive cells and altered oligodendrocytenumbers in the contused rat spinal cord. J. Neurosci. 21,3392–3400.
MELLMAN, I., and STEINMAN, R.M. (2001). Dendritic cells:specialized and regulated antigen processing machines. Cell106, 255–258.
MIKAMI, Y., OKANO, H., SAKAGUCHI, M., et al. (2004).Implantation of dendritic cells in injured adult spinal cord re-sults in activation of endogenous neural stem/progenitor cellsleading to de novo neurogenesis and functional recovery. J.Neurosci. Res. 76, 453–465.
MIZRAHI, T., HAUBEN, E., and SCHWARTZ, M. (2002).The tissue-specific self-pathogen is the protective self-anti-gen: the case of uveitis. J. Immunol. 169, 5971–5977.
MOALEM, G., LEIBOWITZ-AMIT, R., YOLES, E., MOR, F.,COHEN, I.R., and SCHWARTZ, M. (1999). Autoimmune Tcells protect neurons from secondary degeneration after cen-tral nervous system axotomy. Nat. Med. 5, 49–55.
MOALEM, G., YOLES, E., LEIBOWITZ-AMIT, R., et al.(2000). Autoimmune T cells retard the loss of function in in-jured rat optic nerves. J. Neuroimmunol. 106, 189–197.
MOSSER, D.M. (2003). The many faces of macrophage acti-vation. J. Leukoc. Biol. 73, 209–212.
NAGORSEN, D., DEOLA, S., SMITH, K., et al. (2005). Po-larized monocyte response to cytokine stimulation. GenomeBiol. 6, R15.
NAKAJIMA, K., and KOHSAKA, S. (2001). Microglia: acti-vation and their significance in the central nervous system.J. Biochem. (Tokyo) 130, 169–175.
NAKAJIMA, K., and KOHSAKA, S. (2004). Microglia: neu-roprotective and neurotrophic cells in the central nervoussystem. Curr. Drug Targets Cardiovasc. Haematol. Disord.4, 65–84.
NEUMANN, H., and WEKERLE, H. (1998). Neuronal controlof the immune response in the central nervous system: link-ing brain immunity to neurodegeneration. J. Neuropathol.Exp. Neurol. 57, 1–9.
NGUYEN, M.D., JULIEN, J.P., and RIVEST, S. (2002). In-nate immunity: the missing link in neuroprotection and neu-rodegeneration? Nat. Rev. Neurosci. 3, 216–227.
OHLSSON, M., MATTSSON, P., WAMIL, B.D.,HELLERQVIST, C.G., and SVENSSON, M. (2004).Macrophage stimulation using a group B-streptococcus exo-toxin (CM101) leads to axonal regrowth in the injured opticnerve. Restor. Neurol. Neurosci. 22, 33–41.
OKANO, H., OGAWA, Y., NAKAMURA, M., KANEKO, S.,IWANAMI, A., and TOYAMA, Y. (2003). Transplantationof neural stem cells into the spinal cord after injury. Semin.Cell. Dev. Biol. 14, 191–198.
PARK, E., VELUMIAN, A.A., and FEHLINGS, M.G. (2004).The role of excitotoxicity in secondary mechanisms of spinalcord injury: a review with an emphasis on the implicationsfor white matter degeneration. J. Neurotrauma 21, 754–774.
POPOVICH, P.G., GUAN, Z., MCGAUGHY, V., FISHER, L.,HICKEY, W.F., and BASSO, D.M. (2002). The neu-ropathological and behavioral consequences of intraspinalmicroglial/macrophage activation. J. Neuropathol. Exp. Neu-rol. 61, 623–633.
POPOVICH, P.G., WEI, P., and STOKES, B.T. (1997). Cellu-lar inflammatory response after spinal cord injury in Sprague-Dawley and Lewis rats. J. Comp. Neurol. 377, 443–464.
RAIVICH, G., JONES, L.L., KLOSS, C.U., WERNER, A.,NEUMANN, H., and KREUTZBERG, G.W. (1998). Im-mune surveillance in the injured nervous system: T-lympho-cytes invade the axotomized mouse facial motor nucleus andaggregate around sites of neuronal degeneration. J. Neurosci.18, 5804–5816.
RAMER, L.M., AU, E., RICHTER, M.W., LIU, J., TET-ZLAFF, W., and ROSKAMS, A.J. (2004). Peripheral olfac-tory ensheathing cells reduce scar and cavity formation andpromote regeneration after spinal cord injury. J. Comp. Neu-rol. 473, 1–15.
RAPALINO, O., LAZAROV-SPIEGLER, O., AGRANOV, E.,et al. (1998). Implantation of stimulated homologousmacrophages results in partial recovery of paraplegic rats.Nat. Med. 4, 814–821.
ROBERTS, I., YATES, D., SANDERCOCK, P., et al. (2004).Effect of intravenous corticosteroids on death within 14 daysin 10008 adults with clinically significant head injury (MRCCRASH trial): randomised placebo-controlled trial. Lancet364, 1321–1328.
IMMUNE-BASED THERAPY FOR SPINAL CORD INJURY
369
ROTHWELL, N.J., and STRIJBOS, P.J. (1995). Cytokines inneurodegeneration and repair. Int. J. Dev. Neurosci. 13,179–185.
SCHWARTZ, M., and HAUBEN, E. (2002). Differing viewson spinal cord repair. Science 296, 1400.
SCHWARTZ, M., and KIPNIS, J. (2002). Autoimmunity onalert: naturally occurring regulatory CD4�CD25� T cells aspart of the evolutionary compromise between a “need” anda “risk.” Trends Immunol. 23, 530–534.
SHAKED, I., PORAT, Z., GERSNER, R., KIPNIS, J., andSCHWARTZ, M. (2004). Early activation of microglia asantigen-presenting cells correlates with T cell-mediated pro-tection and repair of the injured central nervous system. J.Neuroimmunol. 146, 84–93.
SHAKED, I., TCHROESH, D., GERSNER, R., et al. (2005).Protective autoimmunity: interferon-gamma enables mi-croglia to remove glutamate without evoking inflammatorymediators. J. Neurochem. 92, 997–1009.
SHIBATA, A., ZELIVYANSKAYA, M., LIMOGES, J., et al.(2003). Peripheral nerve induces macrophage neurotrophicactivities: regulation of neuronal process outgrowth, intra-cellular signaling and synaptic function. J. Neuroimmunol.142, 112–129.
VAN BEEK, J., ELWARD, K., and GASQUE, P. (2003). Ac-tivation of complement in the central nervous system: roles
in neurodegeneration and neuroprotection. Ann. NY Acad.Sci. 992, 56–71.
XIAO, B.G., HUANG, Y.M., YANG, J.S., XU, L.Y., and LINK,H. (2001). Bone marrow-derived dendritic cells from exper-imental allergic encephalomyelitis induce immune toleranceto EAE in Lewis rats. Clin. Exp. Immunol. 125, 300–309.
YIN, Y., CUI, Q., LI, Y., et al. (2003). Macrophage-derivedfactors stimulate optic nerve regeneration. J. Neurosci. 23,2284–2293.
YONG, C., ARNOLD, P.M., ZOUBINE, M.N., et al. (1998).Apoptosis in cellular compartments of rat spinal cord aftersevere contusion injury. J. Neurotrauma 15, 459–472.
YOUNG, S.H., YE, J., FRAZER, D.G., SHI, X., and CASTRA-NOVA, V. (2001). Molecular mechanism of tumor necrosisfactor–alpha production in 1 � 3-beta-glucan (zymosan)–acti-vated macrophages. J. Biol. Chem. 276, 20781–20787.
ZIV, Y., RON, N., BUTOVSKY, O., et al. (2006). Immune cellscontributed to the maintenance of neurogenesis and spatiallearning abilities in adulthood. Nat. Neurosci. 9, 268–275.
Address reprint requests to:Dr. Michal Schwartz
Department of NeurobiologyWeizmann Institute of Science
