FIG. 16. Photographs from the same brain region of two rhesus infants, one treated with a high subcutaneous dose of NaCl (a), the other with an equimolar (4 g/kg) subcutaneous dose of MSG (b). a. Infant H. Red blood cells in the ventricular cavity (arrows) is the only obvious abnormality. Many blood vessels appear poorly dilated, an effect possibly related to hemodynamic disturbances associated with hypernatremic dehydration (X 110). b. Infant I. A massive edema and necrotizing reaction affecting mixed glial and neuronal components has spread throughout the infundibular nucleus and stalk region and laterally beyond the boundaries of the infundibular nucleus. A few red blood cells (arrows) are present in the ventricular cavity revealing that intracranial bleeding also occurred in this brain (X 130).

483

FIG. 17. Comparison of a minimal and maximal MSG effect in the hypothalamus of the infant mouse. a. The hypothalamus of a 10 day old infant mouse given MSG, 0.5 g/kg orally by feeding tube. The necrotizing reaction is confined to a few neurons in the floor and at the angles of the 3rd ventricle (median eminence). The infundibular (arcuate) nucleus proper is hardly involved (X 300). b. A massive spreading lesion induced by administering a large dose (4 g/kg) of MSG to a 1 day old infant mouse. The reaction is not confined to the median eminence or even to the infundibular (arcuate) nucleus, but has spread dorsolaterally to include the entire mediobasilar portion of this neonatal brain (X 75).

blood elements or by a shift of glutamate from other tissue compartments into blood. The brain contains a higher natural glutamate concentration than other tissues of the body (38). An efflux of glutamate from brain to blood may have been at least partially responsible for the elevation of blood glutamate. In contrast, it has been found in infant mice that an exogenous load of MSG (2 g/kg) results in a massive but selective transfer of glutamate from blood into a particular region of brain, the infundibular (arcuate) nucleus (39). Presumably unless tissue concentrations of glutamate in the infundibular nucleus rise above some threshold level, a lesion will not develop. We would postulate that the glutamate content of the infundibular nucleus of infant H did not exceed such a threshold level despite the increased blood concentration of glutamate. Since infant H was judged to be dehydrated at the onset of the experiment, it is possible that at least some of the rise in blood glutamate following an injection. of hypertonic NaCl reflected, not an increased amount of total blood glutamate, but rather a hemoconcentration phenomenon. This, in turn, may have promoted an efflux of glutamate from brain into blood since various ions are known to be shifted from intracellular compartments into blood under conditions of hypernatremic dehydration (37). The role of biological variation must also be consid

GLUTAMATE-INDUCED BRAIN DAMAGE IN INFANT PRIMATES

485

ered. Assuming a blood glutamate concentration of 20 mg% to be the threshold value for occurence of brain lesions in the average healthy primate infant, an occasional individual might, nevertheless, be resistant to damage from blood glutamate levels in this range.

Our data do not support the view that only subprimates are susceptible to MSG-induced neurotoxicity. The lesions in our primate infants A, B and I following relatively high subcutaneous doses of MSG were so similar to those consistently observed in mice treated with high doses of MSG that it seems unrealistic to deny primate susceptibility to the MSG effect. Since lesions we detected in infants treated orally with lower doses were also similar in localization and identical in cytopathological detail to those we routinely find in infant mice treated orally with low doses of MSG, a causal link between low oral doses of MSG and necrosis of neurons in the infant primate hypothalamus also seems likely. However, since the lesions in these infants were quite small, careful consideration should be given to possible explanations other than MSG toxicity.

Because of the very nature of the changes involved in these lesions (neuronal necrosis) we believe the fixation artefact thesis of Reynolds et al. can reasonably be discarded. In particular, it is difficult to explain clumping of nuclear chromatin and nuclear pyknosis in terms of fixation vagaries. It would be more plausible to consider such tiny necrotic lesions as manifestations of the necrobiosis process (spontaneous death of neurons during natural development). However, necrobiosis is not a process selectively affecting clusters of neurons in the infundibular nucleus but tends rather to affect neurons individually in scattered distribution throughout the developing central nervous system. Further, in our experience the perikaryon of a neuron undergoing necrobiosis is not swollen and vacuous, rather the entire cellular profile is uniformly dark and condensed. Five hours following MSG treatment some cells in the lesion site appear dark and condensed but within the same lesion there are always others with acutely swollen cell bodies and pyknotic nuclei. Also, we have noted that while necrobiotic cellular profiles are frequently encountered in the fetal central nervous system, they are exceedingly rare in the 1-week-old infant primate brain. And finally, it is inconsistent with the necrobiosis interpretation that neither we nor others (10, 15) could find acutely swollen infundibular neurons with pyknotic nuclei in brains of infant monkeys not treated with MSG.

Although it has been established that rodents receiving MSG in infancy have endocrine abnormalities and obesity as adults (6, 17-20) the late occurring sequelae of MSG-intake by infant primates has not been studied. Sheehan and Kovacs (31), reporting on human adults found that neurons in a subventricular portion of the infundibular nucleus, which they termed the subventricular nucleus, were hypertrophied in women who were post-menopausal, parapartum or who suffered from post-partum hypopituitarism. In the latter condition a decreased number of neurons in the subventricular nucleus was also described. This suggests an involvement of these particular neurons in neuroendocrine regulatory function. Probably, since the subventricular nucleus of Sheehan and Kovacs was badly damaged as were more dorsolateral portions of the infundib

486

J. W. OLNEY, L. G. SHARPE, AND R. D. FEIGIN

ular nucleus in infants A, B and I, these infants would have manifested endocrine abnormalities in adulthood. It would be of value, but may require very sensitive tests, to determine whether infants having sustained only minimal lesions localized about the subventricular zone would have an increased liability to develop endocrine disturbances as adults.

According to an NAS Subcommittee (26), “in considering the safety of added MSG in baby foods, one must remember that the levels added are small-not higher than 0.6 g% ...". This means that one small jar of baby food (130 g) would provide about 0.78 g of MSG or 0.13 g/kg of body weight for a human infant weighing 6 kg. Based on our findings that an oral dose of 1 g/kg in the primate or 0.5 g/kg in the mouse (7) is sufficient to destroy hypothalamic neurons, this leaves a 4- to 8-fold margin of safety for a human infant eating one jar, a 2- to 4-fold margin if two jars are eaten and so forth. This is substantially less than the 100-fold margin generally recommended to accommodate contingencies such as species or individual differences in susceptibility to the mechanism of a toxic compound. Although age at time of exposure is undoubtedly an important variable, there is no reliable way to determine how a human infant compares in sensitivity to the MSG effect with any particular animal species at any particular developmental age.

In support of their assumption that human infants are invulnerable to MSG-induced brain damage, the NAS Subcommittee (26) pointed to absence of behavioral manifestations in human infants givén intravenous infusions of protein hydrolysates providing 0.3 g/kg/day of free glutamic acid. Our demonstration that MSG destroys hypothalamic neurons in monkeys as well as mice (7) at intake doses lower than those required to produce acute behavioral manifestations points to a serious flaw in this line of reasoning. The subcutaneous injection of protein hydrolysate (0.2 cc) produces, in 10 day old infant mice, a hypothalamic lesion unaccompanied by behavior disturbances (22). It may be important that there are at least three neurotoxic amino acids present in a protein hydrolysate (glutamate, aspartate and cysteine) and these are capable of acting in concert, by an additive mechanism, to destroy hypothalamic neurons (7, 22). It is relevant to the food safety issue that commercially available preparations such as bouillon cubes and broths (40), which might be used in the home to prepare infant meals, contain unspecified amounts of added MSG, plus added cysteine and hydrolysed protein (more free glutamate, aspartate and cysteine).

Although the focus of attention in the MSG controversy has been on the potential risk of exposure to exogenous sources of MSG, the significance of our findings can better be appreciated within a broader context. Elsewhere we have shown in infant mice that several acidic amino acids which occur naturally in brain (glutamic, aspartic, cysteic, and cysteine sulfinic acids) cause the MSG type of hypothalamic lesion (9). One non-acidic amino acid (L-cysteine), at relatively low doses (0.8-1.2 g/kg), induces a more disseminated neurodegenerative reaction throughout other brain regions, possibly by conversion intracerebrally to its acidic analogues (41). Although the effect of the acidic compounds may not transmit across the placenta (at least of the rodent) 、the

GLUTAMATE-INDUCED BRAIN DAMAGE IN INFANT PRIMATES

487

neurotoxicity of cysteine does (24). If any of these natural metabolites were to accumulate, even transiently in local regions of the developing brain, nerve cells would be deleted, residual traces of tissue damage would be minimal or absent but functional deficits, perhaps of the "minimal brain dysfunction" variety might develop. The natural occurrence in brain of these amino acids, the additive neurotoxic mechanism they share and the fact that at least one of them transmits its effects across the placenta to produce a disseminated neurodegenerative syndrome must all be taken into consideration in evaluating the potential significance of our demonstration that susceptibility to the MSG type of neurotoxic mechanism is not limited to subprimates.

Acknowledgments: The assistance of Dr. Richard M. Torack, Department of Neuropathology, Washington University School of Medicine, St. Louis, Missouri, in interpreting both our control and experimental material is gratefully acknowledged.

REFERENCES

1. LUCAS, D. R., AND NEWHOUSE, J. P.: The toxic effect of sodium L-glutamate on the inner layers of the retina. AMA Arch. Ophth., 58: 193, 1957.

2. POTTS, A. M., MODRELL, K. W., AND KINGSBURY, C.: Permanent fractionation of the ERG by sodium glutamate. Am. J. Ophth., 50: 900, 1960.

3. FREEDMAN, J. K., AND POTTS, A. M.: Repression of glutaminase I in rat retina by administration of sodium L-glutamate. Invest. Ophthal., 2: 252, 1963.

4. COHEN, A. I.: An electron microscopic study of the modification by monosodium glutamate of the retinas of normal and "rodless" mice. Am. J. Anat., 120: 319, 1967.

5. OLNEY, J. W.: Glutamate-induced retinal degeneration in neonatal mice. Electron microscopy of the acutely evolving lesion. J. Neuropath. Exp. Neurol., 28: 455, 1969.

6.

7.

8.

9.

: Brain lesions, obesity and other disturbances in mice treated with monosodium glutamate. Science, 164: 719, 1969.

AND HO, O. L.: Brain damage in infant mice following oral intake of glutamate, aspartate or cysteine. Nature, 227: 609, 1970.

: Glutamate-induced neuronal necrosis in the infant mouse hypothalamus: an electron microscopic study. J. Neuropath. Exp. Neurol., 30: 75, 1971.

Ho, O. L., AND RHEE, V.: Cytotoxic effects of acidic and sulphur containing amino acids on the infant mouse central nervous system. Exp. Brain Res., 14: 61, 1971. 10. ABRAHAM, R., DOUGHTERY, W., GOLDBERG, L., AND COULSTON, F.: The response of the hypothalamus to high doses of monosodium glutamate in mice and monkeys. Cytochemistry and ultrastructural study of lysosomal changes. Exp. Molec. Path., 15: 43, 1971.

11. AREES, E., AND MAYER, J.: Monosodium glutamate-induced brain lesions. Electron microscopic examination. Science, 170: 549, 1970.

12.

13.

SANDREW, B., AND MAYER, J.: MSG-induced brain lesions in infant mice following subcutaneous injection. Fed. Proc., 30: 521, 1971.

AND MAYER, J.: Monosodium glutamate-induced brain lesions in mice. Paper presented at 47th Annual Meeting of Am. Assoc. Neuropath. in Puerto Rico, June, 1971.

14. BURDE, R. M., SCHAINKER, B., AND KAYES, J.: Monosodium glutamate: acute effect of oral and subcutaneous administration on the arcuate nucleus of the hypothalamus in mice and rats. Nature, 233: 58, 1971.

15. REYNOLDS, W. A., LEMKY-JOHNSTON, N., FILER, L. J., JR., AND PITKIN, R. M.: Monosodium glutamate: absence of hypothalamic lesions after ingestion by newborn primates. Science, 172: 1342, 1971.

16. EVERLY, J. L.: Light microscopic examination of MSG-induced lesions in brain of fetal and neonatal rats. Anat. Rec., 169: 312, 1971.