FIG. 10. Detail from figure 5. A cell manifesting relatively early degenerative changes of the kind typically seen in an MSG-induced lesion. This cell is identified as a neuron by the several axosomatic synapses impinging upon its surface. One synapse (arrow) is shown at higher magnification in figure 10b (x 10,500). b. Detail from figure 10a. An axon terminal in synaptic contact with the somal surface of a degenerating neuron. Increased densification and smudging of the post-synaptic membrane, as is apparent here, represents one stage in the deafferentation of a neuron affected by MSG. The next stage (9) involves displacement of the axon terminal so that it is no longer aligned with the post-synaptic density (X 48,000). FIG. 11. A nerve cell from the rostro-subventricular region (shown by light micrography in fig. 4f) of NaCl control infant G. The distribution of chromatin in the nucleus appears normal as do mitochondria, the endoplasmic reticulum and all features of the two axosomatic synapses terminating on the surface of this neuron (X 32,000).

76-300 O 73 pt. 4A - 9

GLUTAMATE-INDUCED BRAIN DAMAGE IN INFANT PRIMATES

479

lesser extent in infant B, occurs routinely in infant mice given high doses (4-6 g/kg) of MSG either subcutaneously or orally (MSG-treated mice do not vomit). Figures 17a and b provide examples of typical reaction patterns seen in the infant mouse hypothalamus following low (fig. 17a) compared to high (fig. 17b) doses of MSG. Our findings for infants C, D and E compared with those for A, B and I indicate that comparable changes in reaction pattern also occur in the infant primate treated from low to high extremes over the same dose

range.

Blood glutamate values over the 5 hour post-treatment period for all infants except infant A are given in table 1. The highest and most sustained elevations of blood glutamate occurred in infants treated subcutaneously with high doses (infants B and I) which correlates well with the severity of pathology in these infants' brains (figs. 3 and 16b). Presumably, had we measured blood glutamate in infant A, a similarly high sustained elevation would have been recorded. None of the 3 infants treated orally (infants C, D and E) developed more than a moderate elevation which is consistent with the low doses administered to infants C and D and with the vomiting observed in infant E following a higher dose. The one glutamate curve which was inconsistent with expectation was that for infant H which started from a relatively high base line value of 12 mg% and rose to 25 mg% 22 hours after a large subcutaneous dose of NaCl. The other two control infants treated orally with lower doses of NaCl in dilute skim milk had no increase in blood glutamate concentrations. The high base line value for infant I (20 mg%) probably reflects a hemoconcentration phenomenon in that this infant had poor skin turgor and appeared dehydrated at the beginning of the experiment.

DISCUSSION

In contrast to Reynolds et al. (15) and Abraham et al. (10) who reported no differences between brains of control and MSG-treated primate infants, we found evidence for a neuron-necrotizing process in the hypothalamus of each of 6 infant rhesus monkeys treated with MSG and in none of 3 NaCl-treated controls. A comparison of these studies with ours reveals no important differences in age and sex of treated infants or in dose and routes of administration of MSG. The discrepancy in findings, however, may relate to certain other features of research design and methodology which do distinguish the present study from the prior two (15, 10).

Reynolds, et al. included 6 MSG-treated rhesus infants but no rhesus controls in their study. Two of their rhesus infants, each treated with low oral doses of MSG, were examined by electron microscopy. However, since only a spot sampling technique was employed, the possible occurrence of small lesions in these brains was not actually ruled out. The other 4 rhesus infants were examined only by light microscopy after preparation of brains by a method (formalin, paraffin, cresyl violet) which others have found unsatisfactory for evaluating even large MSG-induced lesions in infant rodent brain (14). The uniformly negative data obtained by these authors, even in infants given doses of MSG as

GLUTAMATE-INDUCED BRAIN DAMAGE IN INFANT PRIMATES

481

high as 4 g/kg, may also relate to the emetic properties of MSG. In discussing their data at a recent meeting (25), W. A. Reynolds mentioned that some of their infants vomited an unknown portion of the administered dose.

Abraham et al. (10) supported their negative findings with a single light micrograph (H and E stain) from a rhesus infant sacrificed 24 hours following oral intake of an emetic dose (4 g/kg) of MSG. Since their description of methods for studying primate infants was confined almost entirely to the few remarks accompanying this illustration, it is difficult to evaluate the technical adequacy of their approach. However, because of the remarkable efficiency with which degenerate elements are removed from the scene of an MSG-induced lesion (minimal lesions are cleared from mouse brain within 12 to 18 hours), it is essential to examine the brain earlier than 24 hours. This is particularly true if, due to vomiting, the infant retained very little MSG and, therefore, sustained only a minimal lesion. But Abraham et al. also reported (10) arcuate lesions in only 60% of infant mice treated subcutaneously with 4 g/kg doses of MSG. In our experience subcutaneously administered MSG predictably destroys neurons in 100% of infant mice treated at any dose exceeding 0.5 g/kg. The 4 g/kg dose administered by Abraham et al., in fact, destroys not only the arcuate nucleus but kills neurons in the medial habenular nucleus, the subfornical organ, midline portions of the cingulate cortex, the anterior hippocampus. (dentate gyrus) and the superior colliculus (6). This distribution of lesions following high doses of MSG and the predictability that an arcuate lesion will occur even at low doses have both been confirmed in other laboratories (11, 13, 14, 16, 25). The high yield of negative findings by Abraham et al. in either mice or monkeys despite high doses of MSG reportedly given, therefore, suggests some repeating defect (s) in their treatment approach or in their timing or technique of tissue collection, preparation and examination.

It is puzzling that blood glutamate concentrations rose to 25 mg% in infant H after a subcutaneous dose of NaCl and no hypothalamic cytopathology developed, whereas, hypothalamic lesions did develop in MSG-treated infants C and D whose blood glutamate concentrations rose only to 20 mg%. Since no exogenous glutamate was administered to infant H, any real increase in blood glutamate content must have occurred either by generation of glutamate by

FIG. 12, 13 and 14. All are from a rostro-subventricular lesion site (shown by light micrography in fig. 4b) in the brain of infant C, treated orally with MSG, 1 g/kg. 12. Two degenerating neurons (top and center) are shown, one (N) which is being encircled by the processes of a phagocytic cell (P). In the surrounding neuropile some elements appear normal and well preserved, others are exhibiting such acute degenerative changes as are characteristically seen in MSG-induced lesion sites (× 3,900). 13. A degenerating cell which is clearly a neuron because of the axon terminal (A) in synaptic contact with its surface. This degenerating cell appears identical to neurons undergoing degeneration in the brains of MSGtreated mice. The densities present in mitochondria of this cell appeared in degenerating cells of each MSG-treated monkey in our series. We did not describe such mitochondrial densities in mouse (8) because they were encountered only infrequently in that species (X 18.000). 14. Compare the appearance of this degenerating neuron from the lesion in infant C's brain with that of a degenerating neuron we previously illustrated (fig. 15, Ref. 8) in the brain of a mouse treated orally with MSG. The vacuolated endoplasmic reticulum (v), spherical mitochondria (m) and clumped chromatin (cc) which is consolidating centrally to produce the typical appearance of a pyknotic nucleus are features these degenerating cells from 2 different species clearly have in common (× 23,000).

FIG. 15. a. Survey electron microscopic view from the rostro-subventricular region of NaCl control infant F's brain. There are no degenerate cellular components present. In several portions of the field there are discontinuities in tissue density because this section was cut tangential to one large vascular complex and transversely across another so that much of the field contains perivascular spaces and adventitial elements. Numerous neurons (N) are also present in small clusters in this subventricular location. One neuron (boxed region) from a small cluster is shown at higher magnification in figure 15b (x 1,000). b. Neuron from boxed region of figure 15a. Cytoplasmic and nuclear components of this cell and the surrounding neuropile all appear normal. We interpret a miscellaneous vacuole (arrow) in an otherwise well preserved and normal appearing cell as an insignificant finding in either control or experimental tissue (X 13,500).