Gen. Pharmac. Vol. 30, No. 3, pp. 265–273, 1998 ISSN 0306-3623/98 $19.00 1 .00Copyright 1998 Elsevier Science Inc. PII S0306-3623(97)00356-XPrinted in the USA. All rights reserved

REVIEW

Growth Factors and Gangliosides asNeuroprotective Agents in Excitotoxicity and Ischemia*

David Hicks,† Valerie Heidinger,Saddek Mohand-Said, Jose Sahel and Henri Dreyfus

Laboratoire de Physiopathologie Retinienne,INSERM CJF 92-02, Centre Hospitalier et Universitaire

Regional, 1 Place de l’Hopital, 67091 Strasbourg Cedex FranceTel: (33) 388 24 33 15; Fax: (33) 388 24 33 14; E-mail: hicks@neurochem.u-strasbg.fr

ABSTRACT. 1. At least two different groups of molecules can be considered neurotrophic factorsbecause they exert a variety of effects upon neural cells. The first consists of the numerous families ofpolypeptide growth factors known to take part in almost all stages of neural cell growth and functioning,including development, differentiation, survival and pathology. The second group also is characterizedby extensive complexity of multiple forms, and consists of the sialic acid–containing glycosphingolipidsor gangliosides. These molecules also take part in the transfer of information from the extracellular mi-lieu to the cell interior, and, similarly to growth factors, are participants in such aspects as development,differentiation and functioning.

2. In this short overview, we consider the existing data on the neuroprotective effects of growthfactors [e.g., basic fibroblast growth factor (bFGF), epidermal growth factor (EGF) and brain-derivedneurotrophic factor] and one species of ganglioside (GM1) against retinal ischemia in vivo and cerebralexcitotoxicity in vitro.

3. We used three different experimental models to investigate their relevance to ischemic and exci-totoxic conditions in the retina and have shown that: (a) both bFGF and EGF show highly effectiveneuroprotection for rat retinal neurons exposed to toxic levels of glutamate or its nonphysiological ago-nist kainate in vitro (b) retinal glial cells suffer morphological perturbations after glutamate or kainatetreatment, and this effect depends on neuron–glial interactions; (c) these glial changes can also be cor-rected by posttreatment with either bFGF or EGF in vitro; (d) with the use of an in vivo animal modelinvolving anterior chamber pressure-induced ischemia in adult rats, either pretreatment by intraperito-neal injection of GM1 or posttreatment by intraocular injection of the same ganglioside significantlyreduces histological damage to inner nuclear regions.

4. Hence both groups of trophic molecules show interesting features for retinal ischemic treatment.gen pharmac 30;3:265–273, 1998. 1998 Elsevier Science Inc.

KEY WORDS. Retina; amacrine cells; Muller glia; cell culture; animal models; neurotrophic factors

INTRODUCTION tivation of functional excitatory amino acid receptors prevalent inthe retina and CNS. According to the generally accepted pathologi-Ischemic and reperfusion injury to the retina, as for the other re-cal mechanism, a decrease in blood-borne nutriments would lead togions of the central nervous system (CNS), are major concerns inreductions in energy-dependent processes such as glutamate uptakecurrent ophthalmological care. Such complications are frequentlyby the glial cells present between the vascular system and the neu-observed in such varied pathologies as diabetic retinopathy, hyper-rons. This increase in extracellular levels of glutamate would thentension and vascular occlusion, where a diminution in the supply ofstimulate the widespread glutamate receptors, inducing depolariza-oxygen and glucose to neural tissue leads to serious perturbations intion and calcium entry on a large scale, these events leading to pas-neuronal and glial function and survival. One critical step of thesive water influx and imbalance in intracellular calcium homeostasisischemic process entails excessive activation of glutamatergic recep-(Frandsen and Schousboe, 1993). Increased intracellular calciumtors, leading to massive entry of calcium and culminating in cellconcentrations would then trigger a series of biochemical reactions,death. This process has been termed excitotoxicity (Choi, 1988,including stimulation of phospholipase A2. This enzyme hydrolyzes1992; Rothman and Olney, 1986) because it includes abnormal ac-membrane lipids to free fatty acids including arachidonic acid,which in turn generate dangerous free radical species that would

*Part of this work was presented at a meeting on “Basic mechanisms related lead among other consequences to lipid peroxidation. Neuronalto stress, ischemia and neuroprotection in the retina” held in Coimbra, Por- stress would lead to further glutamate release and exacerbation oftugal (11–12 April 1997) and organized by A. P. Carvalho and N. N. Os- the cascade, resulting eventually in irreversible damage to the sensi-borne.

tive neuronal population.†To whom correspondence should be addressed.Received 1 May 1997; accepted 13 July 1997. Several possible therapeutic approaches are envisageable. Gluta-

266 D. Hicks et al.

mate antagonists (Weber et al., 1995) could alleviate development these injury-, photic- and hereditary-related increases in retinalbFGF expression, ischemic insults have been reported to downregu-of the pathology through reducing initial ionic imbalances. Such ap-

proaches have the inconvenience that they might also interfere late bFGF expression as well as that of one of its receptors (Hayashiet al., 1996). In this last study, both vascular endothelial growth fac-with normal physiological transmission. Antioxidants and free-radi-

cal scavengers could reduce cellular damage by trapping highly reac- tor and platelet-derived growth factor were upregulated. Two inde-pendent studies examined the protective effects of bFGF injectionstive oxygen species before they engender breakdown of cellular

components (Gupta and Marmor, 1993). A third avenue would be on anterior chamber pressure-induced ischemia in the adult rat.Zhang et al. (1994) injected bFGF simultaneously with induction ofthrough the use of the various naturally occurring trophic factors ex-

hibiting neurotrophic activity and constituting the object of this the ischemic insult and observed significant protection against in-ner retinal thinning and ganglion cell necrosis after 7 and 14 daysshort overview.of reperfusion, although ganglion cell losses were no longer reducedin the posterior retina at 14 days. Unoki and LaVail (1994) injected

POLYPEPTIDE GROWTH FACTORSbFGF either prior to or subsequent to ischemia and observed a tran-

The number of diffusible protein factors being isolated from differ- sient protection of inner retinal layers at 7 but not at 14 days. Theent tissues has increased dramatically. The reasons for such belated two studies concur in demonstrating significant beneficial effects ofrecognition lie in the facts that these molecules often exist in mi- this growth factor at 1 week postischemic survival times, with re-nute amounts and operate over only short distances, making their duced or negligible protection at longer times.isolation and characterization very difficult by classical biochemical Most polypeptide growth factors have been shown to confer someapproaches. Most of these factors, irrespective of the fact that they degree of protection against excitotoxic or ischemic damage in vivooften have been initially discovered in nonneural cells, turn out to or in vitro. Transforming growth factor b exists in three isoforms,have widespread effects within the CNS and retina. The general and only isoforms 1 and 3 exhibited neuroprotective effects for

chick or rat brain neurons (Prehn et al., 1993). Epidermal growthmechanism of action of these signaling molecules is their releasefrom their sites of synthesis to bind to specific membrane-bound re- factor (EGF) revealed protective effects for cerebellar neurons (Abe

and Saito, 1992) and CNTF protected hippocampal neurons [partic-ceptors on the surface of their target cells, which promotes dimeriza-tion and phosphorylation of these receptors. In turn, the receptors ularly when acting in synergy with ganglioside GM1 (see next sec-

tion) (Skaper et al., 1992)] in vitro. Within the retina, both CNTFare now able to activate intracellular second-messenger cascadesthat will ultimately result in modification of gene expression and and brain-derived neurotrophic factor transiently reduced ischemic

damage in vivo (Unoki and LaVail, 1994), whereas repeated injec-generate some kind of cellular response (cell division, differentia-tion, migration, etc.) (Schlessinger and Ullrich, 1992). This simplis- tions of another neurotrophin, nerve growth factor, into ischemic

cat eyes led to increased functional recovery of ganglion cells (Sili-tic schema is in reality extremely complicated, with the existence ofmultiple isoforms for many growth factors and their corresponding prandi et al., 1993).

Because excitotoxicity and ischemia can be conveniently modeledreceptors, both high- and low-affinity receptors, soluble receptors,and so forth. In this review, we will not attempt to discuss in any in vitro for both brain-derived [reviewed in Frandsen and Schousboe

(1993)] and retinal neurons (Abrams et al., 1989; Facci et al., 1990;depth the potential roles and mechanisms of these different factors,but instead give a brief survey of the literature regarding their use Kitano et al., 1996) and, furthermore, growth factor effects are easily

monitored under culture conditions (Morrison et al., 1986), weas neuroprotective agents against excitotoxicity and ischemia in theCNS and, where known, in the retina, both in vivo and in vitro. chose to study potential neuroprotective growth factor effects

against excitotoxic neuronal damage with the use of monolayer cul-One of the most well studied growth factors from the point of viewof its potential neuroprotective effects is basic fibroblast growth fac- tures prepared from newborn rat retinas according to previously pub-

lished procedures (Hicks and Courtois, 1992). As shown in Figuretor (bFGF) [reviewed in Burgess and Maciag (1989)]. It constitutesone of the prototypical members of the FGF family, which currently 1, in such cultures maintained for 1 week either in serum-containing

medium or in chemically defined medium to reduce glial contami-comprises 14 members (Birnbaum et al., 1997). Its effects on neu-ronal survival and neurite elongation in vivo and in vitro have been nation, several neuronal populations can be identified through the

use of specific antibody markers: rod photoreceptors are labeled byknown for many years (Anderson et al., 1988; Morrison et al., 1986),and the FGF family has been well studied within the retina [reviewed anti-opsin antibody (Hicks and Barnstable, 1987), amacrine cells

are labeled by anti-syntaxin antibody (Barnstable et al., 1985) andin Hicks et al. (1991)]. FGF mRNA and protein levels undergo markedupregulation subsequent to mechanically or chemically induced ganglion cells are labeled by anti-neurofilament antibody (Drager et

al., 1984). In addition, the total neuronal population can be labeledtrauma within both the CNS (Finklestein et al., 1988) and the ret-ina (Wen et al., 1995). In addition, Gao and Hollyfield (1996) dem- by anti-neuron-specific enolase (NSE) (Hicks and Courtois, 1992).

Compared with untreated control cultures, parallel cultures exposedonstrated upregulation in retinal bFGF levels in both hereditary andphotic-induced degeneration. Retinal localization of bFGF expres- to increasing doses of glutamate exhibit increasing losses in NSE-

immunoreactive neurons, with 100 mM glutamate leading to thesion is still rather controversial, but Wen et al. (1995) demonstratedby in situ hybridization techniques that bFGF mRNA is present loss of 50% of such cells and 1 mM to 80% loss (Fig. 2). This popula-

tion includes all the amacrine cells, whereas rods are unharmed.throughout the retinal nuclear layers and that this expression increasesglobally around the wound area after local wounding but that the Similar treatment with the use of nonphysiological glutamate ago-

nist kainate led to small decreases in total neuronal numbersgreatest increase is observed in the inner nuclear layers [as was alsothe case for ciliary neurotrophic factor (CNTF) mRNA]. Such data (z14%), and this decrease was entirely accounted for by the ama-

crine cells (Fig. 3), as has been demonstrated by others (Abrams etsuggest that increased endogenous bFGF or CNTF production orboth are responsible for the resistance of such mechanically injured al., 1989). Pretreatment of sister cultures with either bFGF or EGF

(both used at 500 pM, corresponding to relatively physiological con-retinas to subsequent phototoxic insults (Faktorovich et al., 1992),a phenomenon that may be likened to “conditioning lesion” studies centrations) led to highly significant protection against excitotoxi-

city (1 mM for both excitatory amino acids). In the glutamate treat-in the brain (Nieto-Sampedro and Cotman, 1985). In contrast with

Neurotrophic Factors and Ischemia 267

FIGURE 1. Retinal neuronal cultures after 5 days in defined medium in vitro. [same field in (a–c). (a) Nomarski image of typical fieldshowing numerous rounded and extensive fiber outgrowth. Rare glial cells also are observed. (b) Amacrine cells immunostained withanti-syntaxin antibody. [Short arrow indicates labeled cell in (a) and (b).] (c) Photoreceptors labeled with anti-opsin antibody. [Longarrow indicates labeled cell in (a) and (c).] (d) NSE antiserum labels all neurons (Heidinger et al., Brain Res., 1997).

ment, NSE-immunoreactive cell loss was blocked by about 80% by shown that growth factor protection (at least that in one possiblemechanism of action, that of preventing nitric oxide–induced toxic-either factor, irrespective of culture medium composition (Fig. 2).

In the case of kainate treatment, growth factor pretreatment led to ity) is observed when such agents are added prior to but not subse-quent to the pathological insult (Maiese et al., 1993). Interestingly,retention of neurite-bearing amacrine cells (Fig. 3) (Heidinger et

al., 1997). Because growth factor treatment was performed before compared with other figures quoted in the literature for bFGF pro-tection against excitotoxicity (Mattson et al., 1995), retinal neu-excitotoxic treatment, we do not know if these factors would reduce

cell damage when subsequently added to excess levels of excitatory ronal survival seems to be greatly enhanced by this growth factor(z250% increase compared with 1 mM glutamate alone, versusamino acids. Previous studies on cultured cortical neurons have

FIGURE 2. Neuronal loss after glutamic acid exposure, and protection afforded by EGF or bFGF pretreatment. For each experimentalgroup (no growth factor, EGF or bFGF, with or without serum), numbers of NSE immunoreactive cells in control cultures were ex-pressed as 100%. The percentage of surviving neurons was calculated with respect to control values. For untreated cultures in chemi-cally defined medium (CDM), 100% represented an average of 6065 cells per field. There were no significant differences in cell num-bers between control [either in CDM or 10% fetal calf serum (FCS)] and cells treated only with growth factors (either EGF or bFGF).***P,0.001, significance values for glutamic acid–treated cells in comparison with their control untreated cells; sssP,0.001 andsP,0.05, significance values for EGF- or bFGF-pretreated cells exposed to glutamic acid in comparison with their controls (pretreatedwith growth factors but without glutamic acid exposure) (Heidinger et al., Brain Res., 1997).

268 D. Hicks et al.

also been established [reviewed in Landis (1994) and Duffy andMacVicar (1996)]. Much less has been published on glial responsesto ischemia or excitotoxicity in vitro, particularly with respect to ret-inal Muller glial cells. Using purified Muller glial cell cultures ex-posed to concentrations of either glutamate or kainate up to 10 mM,we observed no change in glial viability or morphology. However,when retinas were incubated in excess levels of excitatory aminoacids for 30 min prior to enzymatic digestion and seeding into cul-ture, we observed that, compared with control retinal cultures, thelarge majority of glia in treated cultures assumed a rounded globularappearance, very different from the normal epithelioid glial mor-phology in vitro. Such an aspect is reminiscent of cytoskeletal col-lapse. When either bFGF or EGF was included in the culture medium,(i.e., added subsequently to excitotoxic treatment), the number ofrounded glia was drastically reduced, with glial morphology beingfully restored to that observed in control untreated cultures (Fig. 4).Hence in this case, growth factors are capable of exerting beneficialeffects on glial recovery even after insult, possibly through helpingto restore stability of the cytoskeleton (Heidinger et al., in press).

The mechanism(s) of action of growth factors in the neuroprotec-tive effects reported by us and by others are still not clear. As inmost in vivo and in vitro models in which growth factors prove to beof benefit, the agent is delivered considerably before the toxic insult;this would indicate that de novo synthesis of some other molecule isnecessary to confer protection. Furthermore, Abe and Saito (1992)showed that neuroprotective effects of EGF were abolished in thepresence of protein synthesis inhibitors. In hippocampal neuronalcultures, growth factor treatment has been demonstrated to induceincreased expression of antioxidant enzymes such as superoxide dis-mutase and glutathione reductase, both important detoxifyingagents (Mattson et al., 1995). Another possible pathway leading toneuroprotection is the upregulation of calcium-binding proteins(Mattson et al., 1991). Although the role of nitric oxide in corticalneuronal excitotoxicity in vitro has been demonstrated, as well asthe ability of growth factors to reduce nitric oxide–induced celldeath (Maiese et al., 1993), the mechanism whereby this wasacheived was not described. We showed that growth factors alsolead to significant increases in ganglioside levels in retinal Mullerglial cells in vitro (Hicks et al., 1996). As described in the next sec-tion, although the mechanism by which these lipids alleviate isch-emic damage is not known either, growth factor neuroprotectionmay also be partly due to this route.

GANGLIOSIDES

These sialic acid–containing glycosphingolipids in general are pres-ent in relatively high proportions in neural cell membranes (about5–10% total membrane lipids), where they affect the functional dy-namics of the cellular membranes in several ways: they contributeFIGURE 3. Effects of EGF on kainic acid-induced amacrine cellto the structural rigidity of membranes, they take part in the transferdeath: (a) control untreated cultures; (b) cultures pretreated forof information between neighboring cells or between cell surfaces24 h with 500 pM EGF only; (c) cultures treated with 1 mM kai-

nic acid for 24 h. (d) cultures pretreated with 500 pM EGF fol- and the extracellular environment (neurotransmitters, hormones)lowed by 1 mM kainic acid exposure for 24 h. Bar510 mm (Hei- and they modulate neuronal development and differentiation (Ha-dinger et al., Brain Res., 1997). komori, 1981; Ledeen, 1985; Nagai, 1995).

In experimental models of cerebral ischemia and excitotoxicity,monosialoganglioside GM1 partly reduces neuronal damage (Laz-

z130% increase for hippocampal neurons with the use of 100 mM zaro et al., 1994; Mahadik et al., 1989; Manev et al., 1990). In reti-glutamate), perhaps owing to such factors as differences in receptor nal models of excitotoxicity, GM1 used at high doses (2.1024 M)density. was shown to exhibit limited neuroprotection (Facci et al., 1990).

Although neuronal loss is the principal long-term handicap in The mechanisms underlying this neuroprotection are not com-ischemia, glial involvement in the development of the pathology, as pletely understood but may be related to protective effects on neu-

ronal function, expressed as maintenance of (Na1,K1) ATPase ac-well as glial suffering in the presence of such toxic conditions, has

Neurotrophic Factors and Ischemia 269

FIGURE 4. Quantitative measure of cytoskeletal collapse in retinal Muller glial cells incubated in kainic acid alone before culture orwhen either EGF or bFGF are added at the time of culture after kainic acid treatment (right hand group of columns) (Heidinger etal., in press).

tivity (Mahadik et al., 1989); to blockade of excitatory amino overall ischemic loss in retinal thickness was reduced by 70%, andischemic-induced density losses in ganglion and inner nuclear layeracid–mediated neurotoxicity (Garofalo and Cuello, 1994; Magal et

al., 1990); to membrane stabilization (Karpiak and Mahadik, 1987); (INL) cells were decreased by 45 and 40%, respectively (Figs. 5 and 6).However, in clinical situations such as acute ischemic retinal celland to response potentialization toward neurotrophic factors (Doh-

erty et al., 1985) produced locally in increased quantities as a bio- damage, therapeutic approaches would obviously need to be at-tempted after the initiation of the ischemic insult. Given the bio-chemical response to injury (Nieto-Sampedro and Cotman, 1985).

GM1 also increases regeneration of crush-injured axons of the mam- logical effects of gangliosides stated earlier, it is plausible that theycould be effective even after initiation of ischemic damage. More-malian optic nerve (Sautter et al., 1991).

We showed that GM1 administered intraperitoneally before the over, in view of both the impairment of blood supply to the eye inthis condition and of the deleterious side effects of systemic adminis-experimental induction (elevated anterior chamber pressure) of ret-

inal ischemia in adult rats provides a significant protective effect tration of gangliosides (Simone et al., 1993), an intravitreal injec-tion might be more appropriate while limiting the risk of side ef-(Weber et al., 1996). This effect was witnessed as improvements in

retinal cell densitites and layer thicknesses subsequent to ischemic fects. To estimate the protective role of postischemic intravitrealinjection of monosialoganglioside GM1, we performed histopatho-and reperfusion injury, using the histomorphometric method de-

scribed by Hughes (1990). After both 8- and 15-day survival times, logical studies on ischemia-induced lesions in the rat retina, with or

FIGURE 5. Measurements of thethickness of different retinal layers(mean6SD) in control animals(white column, n55), saline-in-jected animals (black column, n54)or GM1-injected rats (hatched col-umn, n54) after 8 days postischemia(significance values above the blackcolumn correspond to comparisonsbetween normal and ischemic condi-tions, whereas significance valuesabove the hatched column refer tocomparisons between GM1-injectedand saline-treated groups, *P,0.05).Abbreviations: ILM, inner limitingmembrane; INL, inner nuclear layer;IPL, inner plexiform layer; OLM,outer limiting membrane; ORL, outerretinal layer (Weber et al., 1996).

270 D. Hicks et al.

FIGURE 6. Measurements of thedensity of outer and inner nuclearand ganglion cell layers (mean6SD)in control animals (white column,n55) and saline-injected animals(black column, n54) or GM1-injected rats (hatched column,n54) after 15 days postischemia.See Figure 5 for explanation of sig-nificance values (*P,0.05) andother abbreviations. GCL, ganglioncell layer; ONL, outer nuclearlayer (Weber et al., 1996).

without the use of GM1 (Mohand-Said et al., 1997). As in our pre- has been implicated in decreasing the damage caused by elevated in-tracellular Ca21 levels due to excitatory amino acids. The effects ofvious study, histological examination of ischemic animals lacking

ganglioside treatment showed that the overall thickness of the ret- gangliosides may conceivably be at those amplification steps afterthe initiation of the ischemic process (Favaron et al., 1988; Lom-ina was much less than that in normal animals. The decrease in reti-

nal thickness was mostly due to a loss of inner retinal cells and pro- bardi and Moroni, 1992; Magel et al., 1990), hence the reason forthem being effective even after the primary insult of ischemic dam-cesses, with less obvious changes in the outer retina (Fig. 7). The

two retinal layers that were the most reduced in thickness were the age in our experimental model of acute retinal ischemia. The dimi-nution of ischemic-induced loss by GM1 was essentially observed forinner plexiform layer (IPL) (reduction by 75% of normal, P,0.05)

and the INL (reduction by 35% of normal, P,0.05). Ganglion cell the inner retinal layers, the neurons in these regions being the mostvulnerable to excitotoxicity, as has been shown by the majority ofdensity was reduced by 60% of normal (P,0.05), and INL density

by 45% of normal (P,0.05) (Fig. 8). reports on retinal excitotoxicity (Abrams et al., 1989; Lucas andNewhouse, 1957). The greater damage to inner retinal neurons inIn animals treated with GM1, we observed clear protection of the

retina from ischemic damage: the GM1-treated animals showed a the rat could be attributed to the greater proximity of these cells tothe retinal circulation or to the widespread distribution of gluta-well-preserved architecture with thicker inner retinal layers than

those of the saline-injected animals. Measurements of layer thicknesses matergic receptors in the inner retina or to both (Brandstater et al.,1994; Hamassaki-Brito et al., 1993). The rationale for the choice ofand cell densities of the GM1-treated animals quantitatively con-

firmed that GM1 alleviated the ischemia-induced injury (Fig. 8). the particular ganglioside species (GM1) and the dose (1025 M)used in these experiments was based on data obtained from studiesThe IPL and INL thicknesses were significantly greater than those

of the saline-treated group (P,0.05), the ischemic-induced thick- conducted on the brain (Lazzaro et al., 1994). However, mammalianretinal ganglioside composition is substantially different from that ofness losses being diminished by 17% and 44%, respectively. The cell

densities were also significantly greater than those of the saline the brain, with, for example, much less GM1 and much more disia-loganglioside GD3 being found in the former (Dreyfus et al., 1996).group, with ischemic-induced ganglion and INL cell density losses

being reduced by 37% and 27%, respectively. At 15-day survival Further experiments to determine the degree of protection of othergangliosides will thus be of great interest.times, for all retinal thicknesses and densities, there was a statisti-

cally significant difference between the saline-treated and GM1-treated postischemic groups (P,0.05), as observed at 8 days of re-

CONCLUSIONSperfusion. The losses due to ischemia were blocked by 8% to 35%,which was about 10% less than after 8 days of reperfusion. This brief overview has presented data on the beneficial effects of

two general types of trophic factors—namely, growth factors andIn retinal or cerebral ischemia, the mechanism(s) by which GM1permits protection against ischemic injury is not clear, although it gangliosides—on retinal excitotoxicity and ischemia in vivo and in

Neurotrophic Factors and Ischemia 271

FIGURE 7. Light micrographs of the posterior pole of adult rat retina. Normal, uninjected rat (A). One week after retinal ischemiain saline-injected rat (B) and in GM1-injected rat (C). Two weeks after retinal ischemia in saline-injected rat (D) and in GM1-injectedrat (E). Bar510 mm (Mohand-Said et al., 1997).

vitro. Although in neither case is the mechanism of protection very successful (Prehn et al., 1993; Unoki and LaVail, 1994). Ganglio-sides may present an advantage in this respect because they incorpo-clear (and it must be stated that their physiological roles are not

fully established), the fact that these molecules exhibit significant rate directly into target membranes and retain their biological activ-ity (Tettamanti et al., 1981). Under roughly similar experimentalneuroprotective features warrants further examination of their use-

fulness in this respect. Although most studies with growth factors conditions—anterior chamber pressure-induced ischemia in adultrats, with similar durations and intensities of experimental ischemiaexhibited protective effects only when they were applied before the

toxic paradigm, indicating a necessary de novo synthesis of other and survival times (Mohand-Said et al., 1997; Unoki and LaVail,1994; Zhang et al., 1994)—gangliosides may show more prolongedneuroprotective compounds (antioxidant enzymes, calcium-binding

proteins, gangliosides), in certain cases, posttreatments have been beneficial effects, which may be due to their lipophilic behavior fa-

FIGURE 8. Measurements of the thickness of different retinal layers (A) and of the cell density of ONL, INL and GCL (B)(mean6SD) in control animals (white column, n55), saline-injected controls (black column, n54) and GM1-treated rats (hatchedcolumn, n54) 8 days after ischemia (*P,0.05); significance values above the black columns correspond to comparisons between nor-mal and ischemic conditions, whereas significance values above the hatched columns refer to comparisons between GM1-injected andsaline treated groups. Abbreviations as in previous Figures 6 and 7 (Mohand-Said et al., 1997).

272 D. Hicks et al.

totoxicity and calcium homeostasis in cultured neurons. J. Neurochem.voring their incorporation into the cells surrounding the area of in-60, 1202–1211.jection, whereas soluble growth factors would tend to diffuse away

Gao H. and Hollyfield J. G. (1996) Basic fibroblast growth factor: increasedwhen administered under the same form. gene expression in inherited and light induced photoreceptor degenera-

Future areas of investigation include a more detailed dissection of tion. Exp. Eye Res. 62, 181–189.Garofalo L. and Cuello A. C. (1994) Nerve growth factor and the monosia-the molecular mechanisms underlying these trophic rescue effects,

loganglioside GM1: analogous and different in vivo effects on biochemi-more comprehensive dose-response studies and possible synergisticcal, morphological, and behavioral parameters of adult cortically le-

influences (Skaper et al., 1992). Animal trials with other ganglio- sioned rats. Exp. Neurol. 125, 195–217.side species more representative of adult mammalian retina, espe- Gupta L. Y. and Marmor M. F. (1993) Mannitol, dextromethorphan and

catalase minimize ischemic damage to retinal pigment epithelium andcially GD3 (Dreyfus et al., 1996), would be very interesting. In addi-retina. Arch. Ophthalmol. 111, 384–388.tion, the role of the retinal Muller glial cell in the initiation,

Hakomori S. I. (1981) Glycosphingolipids in cellular interaction, differenti-development and possible management of excitotoxicity and isch- ation and oncogenesis. Annu. Rev. Biochem. 50, 733–764.emia is an exciting field that deserves more study. Hamassaki-Britto D., Hermans-Borgmeyer I., Heinemann S. and Hugues

T. E. (1993) Expression of glutamate receptor genes in the mammalianThis overview was part of the meeting on Basic Mechanisms Related to Stress, retina: the localization of GluR1 through GluR7 mRNAs. J. Neurosci.Ischemia and Neuroprotection in the Retina held at the University of Coimbra, 13, 1888–1898.Portugal, on 11–12 April 1997. This workshop was organized to discuss the prog- Hayashi A., Koroma B. M., Imai K. and De Juan E. Jr. (1996) Increase ofress made by the different member laboratories and to outline potential further col- protein tyrosine phosphorylation in rat retina after ischemia-reperfusionlaborative projects within the network Retinal Ischemia and was made possible by injury. Invest. Ophthalmol. Visual. Sci. 37, 2146–2156.the Human Capital and Mobility Programme from the European Community.

Heidinger V., Hicks, D., Sahel J. and Dreyfus H. (1997) Peptide growth fac-The authors would like to thank the EC for their financial assistance for the dura-tors but not ganglioside protect against excitotoxicity in rat retinal neu-tion of this network; without it none of the collaborative work undertaken wouldrons in vitro. Brain Res. 767, 279–288.have been possible.

Heidinger V., Dreyfus H., Sahel J., Christen Y. and Hicks D. (in press) Exci-totoxic damage of retinal glial cells depends upon normal neuron-glialinteractions. Glia.

References Hicks D. and Barnstable C. J. (1987) Different rhodopsin monoclonal anti-Abe K. and Saito H. (1992) Protective effect of epidermal growth factor on bodies reveal different binding patterns on developing and adult rat ret-

glutamate neurotoxicity in cultured cerebellar neurons. Neurosci. Res. ina. J. Histochem. Cytochem. 35, 1317–1328.14, 117–123. Hicks D. and Courtois Y. (1992) Fibroblast growth factor stimulates photo-

Abrams L., Politi L. E. and Adler R. (1989) Differential susceptibility of iso- receptor differenciation in vitro. J. Neurosci. 12, 2022–2033.lated mouse retinal and photoreceptors to kainic acid toxicity. Invest. Hicks D., Bugra K., Faucheux B., Jeanny J.-C., Laurent M., Malecaze F., Masca-Ophthalmol. Visual Sci. 30, 2300–2308. relli F., Raulais D., Cohen Y. and Courtois Y. (1991) Fibroblast growth fac-

Anderson K. J., Dam D., Lee S. and Cotman C. W. (1988) Basic fibroblast tors in the retina. In: Progress in Retinal Research, (Edited by Osborne N. andgrowth factor prevents death of lesioned cholinergic in vivo. Nature 332, Chader G.), Vol. 11, pp. 333–374. Pergamon Press, Oxford.360–361. Hicks D., Guerold B. and Dreyfus H. (1996) Stimulation of endogenous gan-

Barnstable C. J., Hofstein R. and Akagawa K. (1985) A marker of early ama- glioside metabolism by neurotrophic factors in cultured retinal Mullercrine cell development in rat retina. Dev. Brain Res. 20, 286–290. glia. Glia 16, 316–324.

Birnbaum D., Lovec H., Coulier F., Goldfarb M. and Lopez-Ferber M. (1997) Hughes W. F. (1990) Quantitation of ischemic damage in the rat retina.Les FGF: une famille en croissance. Med. Sci. 13, 392–396. Exp. Eye Res. 53, 573–582.

Brandstater J. H., Hartveit E., Sasso-Pognetto M. and Wassle H. (1994) Ex- Karpiak S. E., Li Y. S. and Mahadik S. P. (1987) Gangliosides (GM1 andpression of NMDA and high-affinity kainate receptor subunit mRNAs in AGF2) reduce mortality due to ischemia: protection of membrane func-the adult rat retina. Eur. J. Neurosci.6, 1100–1112. tion. Stroke 18, 184–187.

Burgess W. and Macaig T. (1989) The heparin-binding (fibroblast) growth Kitano S., Morgan J. and Caprioli J. (1996) Hypoxic and excitotoxic damagefactor family of proteins. Annu. Rev. Biochem. 58, 575–606. to cultured rat retinal ganglion cells. Exp. Eye Res. 63, 105–112.

Choi D.W. (1988) Glutamate neurotoxicity and diseases of the nervous sys- Landis D. M. D. (1994) The early reactions of non-neuronal cells to braintem. Neuron 1, 623–634. injury. Annu. Rev. Neurosci. 17, 133–151.

Choi D. W. (1992) Excitotoxic cell death. J. Neurobiol. 9, 1261–1276. Lazzaro A., Seren M. S., Koga T., Zanoni R., Schiavo N. and Manev H.Doherty P., Dickson J. G., Flanigan T. P. and Walsh S. (1985) Ganglioside (1994) GM1 reduces infarct volume after focal cerebral ischemia. Exp.

GM1 does not initiate, but enhances neurite regeneration of nerve Neurol.125, 278–285.growth factor dependent sensory neurons. J. Neurochem. 44, 1259–1265. Ledeen R. W. (1985) Gangliosides of the neuron. Trends Neurosci 8, 169–174.

Drager U., Edwards D. L. and Barnstable C. J. (1984) Antibodies against fila- Lombardi G. and Moroni F. (1992) GM1 ganglioside reduces ischemia-mentous components in discrete cell types of the mouse retina. J. Neu- induced excitatory amino acid output: microdialysis study in the gerbilrosci. 4, 2025–2042. hippocampus. Neurosci. Letters 134, 171–174.

Lucas D. R. and Newhouse J. P. (1957) The toxic effect of sodium l-gluta-Dreyfus H., Gurold B., Fontaine V. and Hicks D. (1996) Simplified ganglio-side profile of photoreceptors compared to other retinal neurons. Invest. mate on the inner layers of the retina. Arch. Ophthalmol. 58, 193–201.

Magal E., Louis J.-C., Aguilera J. and Yavin E. (1990) Gangliosides preventOphthalmol. Visual Sci. 37, 574–585.Duffy S. and MacVicar B.A. (1996) In vitro ischemia promotes calcium in- ischemia-induced downregulation of protein kinase C in fetal rat brain.

J. Neurochem. 55, 2126–2131.flux and intracellular calcium release in hippocampal astrocytes. J. Neu-rosci. 16, 71–81. Mahadik S. P., Hawver D. B., Hungund B. L., Li Y. S. and Karpiak S. E.

(1989) GM1 ganglioside treatment after global ischemia protectsFacci L., Leon A. and Skaper S. D. (1990) Excitatory amino acid neurotoxic-ity in cultured retinal neurons: involvement of N-methyl-d-aspartate changes in membrane fatty acids and properties of Na1, K1-ATPase and

Mg2-ATPase. J. Neurosci. Res. 24, 402–412.(NMDA) and non-NMDA receptors and effect of ganglioside GM1.J. Neurosci. Res. 27, 202–210. Maiese K., Boniece I., DeMeo D. and Wagner J. A. (1993) Peptide growth

factors protect against ischemia in culture by preventing nitric oxide tox-Faktorovich E. G., Steinberg R. H., Yasumura D., Matthes M. T. and LaVailM. M. (1992) Basic fibroblast growth factor and local injury protect pho- icity. J. Neurosci. 13, 3034–3040.

Manev H., Favaron M., Vicini S. and Guidotti A. (1990) Ganglioside-medi-toreceptors from light damage in the rat. J. Neurosci. 12, 3554–3567.Favaron M., Manev M., Alho H., Bertolino M., Ferret B., Guidotti A. and ated protection from glutamate-induced neuronal death. Acta Neurobiol.

Exp. 50, 381–394.Costa E. (1988) Gangliosides prevent glutamate and kainate neurotoxic-ity in primary neuronal cultures of neonatal rat cerebellum and cortex. Mattson M. P., Rychlik B., Chu C. and Christakos S. (1991) Evidence for cal-

cium reducing and excitoprotective roles for the calcium binding proteinProc. Natl. Acad. Sci. USA 85, 7351–7355.Finklestein S. P., Apostolides P. J., Caday C. G., Prosser J., Philips M. F. and (calbindin D28k) in cultured hippocampal neurons. Neuron 6, 41–51.

Mattson M. P., Lovell M. A., Furakawa K. and Markesbery, W. A. (1995)Klagsbrun M. (1988) Increased basic fibroblast growth factor (bFGF) immu-noreactivity at the site of focal brain wounds. Brain Res. 460, 253–259. Neurotrophic factors attenuate glutamate-induced accumulation of per-

oxides, elevation of intracellular Ca21 concentration, and neurotoxicityFrandsen A. and Schousboe A. (1993) Excitatory amino acid-mediated cy-

Neurotrophic Factors and Ischemia 273

and increase antioxidant enzyme activities in hippocampal neurons. P. (1993) Serum and CSF anti-GM1 antibodies in patients with Guil-lain-Barre syndrome and chronic inflammatory demyelinative polyneu-J. Neurochem. 65, 1740–1751.

Mohand-Said S., Weber M., Hicks D., Dreyfus H. and Sahel J. A. (1997) In- ropathy. J. Neurol. Sci. 114, 49–55.Skaper S. D., Negro A., Dal Toso R. and Facci L. (1992) Recombinant hu-travitreal injection of ganglioside GM1 after ischemia reduces retinal

damage in rats. Stroke 28, 617–622. man ciliary neurotrophic factor alters the threshold of hippocampal pyra-midal neuron sensitivity to excitotoxin damage: synergistic effects of mo-Morrison R. S., Sharma A., DeVellis J. and Bradshaw R. A. (1986) Basic fi-

broblast growth factor supports the survival of cerebral cortical neurons nosialogangliosides. J. Neurosci. Res. 33, 330–337.Tettamanti G., Venerando B., Roberti S., Lhigorno V., Sonnino S., Ghidoniin primary culture. Proc. Natl. Acad. Sci. USA 83, 7537–7541.

Nagai Y. (1995) Functional roles of gangliosides in bio-signalling. Behav. R., Orlando P. and Massari P. (1981). The fate of exogenously adminis-tered brain gangliosides. In: Gangliosides in Neurological and Neuromuscu-Brain Res. 66, 99–104.

Nieto-Sampedro M. and Cotman C.W. (1985) Growth factor induction and lar Function, Development and Repair (Edited by Rapport M. M. and GorioA.), pp. 225–240. Raven Press, New York.temporal order in central nervous sytem repair. In: Synaptic Plasticity (Ed-

ited by Cotman C. W.), pp. 407–455. Guilford Press, New York. Unoki K. and LaVail M. M. (1994) Protection of the rat retina from isch-emic injury by brain-derived neurotrophic factor, ciliary neurotrophicPrehn J. H. M., Peruche B., Unsicker K. and Krieglstein J. (1993) Isoform

specific effects of transforming growth factor b on degeneration of pri- factor, and basic fibroblast growth factor. Invest. Ophthalmol. Visual Sci.35, 907–915.mary neuronal cultures induced by cytotoxic hypoxia or glutamate.

J. Neurochem. 60, 1665–1672. Weber M., Bonaventure N. and Sahel J. A. (1995) Protective role of excit-atory amino acid antagonists in experimental retinal ischaemia. GraefesRothman S. M. and Olney J. W. (1986) Glutamate and the pathophysiology

of hypoxic-ischemic brain damage. Ann. Neurol. 19, 105–111. Arch. Clin. Exp. Ophthalmol. 233, 360–365.Weber M., Mohand-Said S., Hicks D., Dreyfus H. and Sahel J. A. (1996)Sautter J., Schwartz M., Duvdevani R. and Sabel B. A. (1991) GM1 ganglio-

side treatment reduces visual deficits after graded crush of the rat optic Monosialoganglioside GM1 reduces ischemia-reperfusion-induced injuryin the rat retina. Invest. Ophthalmol. Visual Sci. 37, 267–273.nerve. Brain Res. 565, 23–33.

Schlessinger J. and Ulrich A. (1992) Growth factor signaling by receptor ty- Wen R., Song Y., Cheng T., Matthes M. T., Yasumura D., LaVail M. M. andSteinberg R. H. (1995) Injury-induced upregulation of bFGF and CNTFrosine kinases. Neuron 9, 383–391.

Siliprandi R., Canella R. and Carmignoto G. (1993) Nerve growth factor mRNAs in the rat retina. J. Neurosci. 15, 7377–7385.Zhang C., Takahashi K., Lam T. T. and Tso M. O. M. (1994) Effects of basicpromotes functional recovery of retinal ganglion cells after ischemia. In-

vest. Ophthalmol. Visual Sci. 34, 3232–3245. fibroblast growth factor on retinal ischemia. Invest. Ophthalmol. VisualSci. 35, 3163–3168.Simone I. L., Annunciata P., Maimone D., Liguori M., Leante R. and Livrea