Natural groupings of primate spinothalamic neurons based on cutaneous stimulation. Physiological and anatomical features

1. Two hundred and twenty-one spino thalamic tract (STT) neurons in the lumbar spinal cord of anesthetized monkeys were studied. The majority of the recordings were in laminae IV-VI. Thirteen of these neurons were intracellularly injected with horseradish peroxidase and histologically reconstructed. 2. A standard series of four mechanical cutaneous stimuli, which ranged in intensity from innocuous brushing to tissue-damaging pinching, were used to test the mechanical responsiveness of STT neurons. The mean alterations in discharge rate produced by these test stimuli when delivered to a neuron's excitatory receptive field were used as response measures. 3. Univariate and bivariate analyses of these response measures failed to reveal natural groupings of STT neurons. To assess whether natural groupings dependent upon shared multivariate response patterns were present, a K-means cluster analysis of the responses was performed. 4. Because an assumption about the type of coding used by the STT system had to be made prior to clustering, two independent analyses were performed. One approach assumed a labeled line coding model; response magnitudes were determined within the context of the neuron under study (within neuron analysis). The other approach assumed a population coding model; response magnitudes were determined within the context of the STT population (across-neuron analysis). 5. The within-neuron analysis suggested that the STT sample could be partitioned into four groups. The smallest group (n = 18, 8%) responded primarily to brushing but often had a convergent nociceptive input; this group was referred to as type I. A second group (n = 31, 14%) had strong responses to low-intensity stimuli, particularly pressure, and modestly larger responses to noxious stimuli; this group was referred to as type II. The clustering in these two groups was relatively weak, reflecting some heterogeneity in response pattern. 6. The largest within-neuron group (n = 108, 49%) was most responsive to noxious stimuli but had a saturating response function; because of their apparent role in coding intermediate intensity stimuli, this group was referred to as type III. The fourth group (n = 64, 29%) responded best to the most intense stimulus used; this group was referred to as type IV. 7. The across-neuron analysis also suggested that the STT sample could be partitioned into four groups. The largest group (n = 122, 55%) had relatively weak responses to all the cutaneous stimuli; this group was referred to as type A. 8. All of the remaining across-neuron groups had mean responses at or above the mean for all cutaneous stimuli. One group (n = 26, 12%) had relatively strong responses to low-intensity stimuli and was referred to as type B. Another group (n = 9, 4%) had relatively strong responses to intermediate intensity stimuli and was referred to as type C. The largest of these groups (n = 64, 29%) responded best to high-intensity stimuli and was referred to as type D. 9. Clear anatomical correlations were found with the across-neuron classes but not with those derived from the within-neuron analysis. Type A neurons had dendritic trees, which tended to arborize in ventral laminae (IV-VI) and to have slowly conducting ascending axons. Type B, C, and D neurons had faster conducting ascending axons and dendritic trees that tended to arborize in dorsal laminae. Type D neurons had relatively profuse branching within superficial laminae (I and II), whereas type C and B neurons had modal branching within lamina III. Dendritic branches were also found within the lateral white matter and in lamina X. 10. Although our results do not allow a conclusion to be drawn about the type of coding used by the STT system, they strongly suggest that a population-based model of intensity coding may be appropriate.