Report

Interhemispheric Asymmetries of the Modular Structure in Human Temporal Cortex

See allHide authors and affiliations

Science  15 Sep 2000:
Vol. 289, Issue 5486, pp. 1946-1949
DOI: 10.1126/science.289.5486.1946

Abstract

Language-relevant processing of auditory signals is lateralized and involves the posterior part of Brodmann area 22. We found that the functional lateralization in this area was accompanied by interhemispheric differences in the organization of the intrinsic microcircuitry. Neuronal tract tracing revealed a modular network of long-range intrinsic connections linking regularly spaced clusters of neurons. Although the cluster diameter was similar in both hemispheres, their spacing was about 20 percent larger in the left hemisphere. Assuming similar relations between functional and anatomical architecture as in visual cortex, the present data suggest that more functionally distinct columnar systems are included per surface unit in the left than in the right area 22.

Neuropsychological, electrophysiological, and noninvasive imaging studies indicate that language comprehension and production are accompanied by activation of certain cortical areas. Several of these areas are activated only unilaterally in the dominant hemisphere (1, 2). One of the regions exhibiting robust unilateral activation during processing of language-related signals is located in the posterior part of the first temporal gyrus and the posterior temporal plane (Fig. 1A) (3–6). This region corresponds to a discrete cytoarchitectonical entity, the posterior part of area 22 in the Brodmann classification (7), or to area TA1 after von Economo and Koskinas (8) and is distinguishable in both hemispheres on the basis of common cytoarchitectonic features. However, the temporal plane [(9), but see (10)], in particular, area 22 (11), is significantly larger, and cell size and density are superior in the dominant hemisphere (12, 13). In the dominant hemisphere, area 22 is involved in word detection and generation (3–6), which have been attributed to its ability to process rapid frequency changes (14,15), whereas the homolog area in the nondominant hemisphere appears to be specialized in the discrimination of melody, pitch, and sound intensity (16,17).

Figure 1

(A) Cytoarchitectonic subdivision of the upper portion of the temporal lobe (see shaded area in the inset) according to the classification of Brodmann (7). This parcellization is closely related to the classification of von Economo and Koskinas (8), with area 41 corresponding to TC, area 52 to TD, area 42 to TB, the posterior (post) portion of area 22 to TA1, and the anterior (ant) portion of area 22 to TA2. a, anterior; p, posterior. (B toE) Patterns of postmortem DiI labeling in area 22 of the human temporal cortex. (B) Retrogradely labeled supragranular pyramidal neuron, about 5 mm from the dye crystal. Filled arrow, axon; open arrow, horizontal axon collateral. Scale bar, 125 μm. (C) Cluster of retrogradely DiA-labeled neurons (between the arrows) in the supragranular layers about 4 mm from the dye crystal. Scale bar, 250 μm. (D) Clusters of retrogradely labeled cells and anterogradely labeled axon terminals (arrows) in a tangential section through supragranular layers of left area 22. The dye crystal was about 1.5 mm beyond the left edge of the micrograph. Scale bar, 1 mm. m, medial; p, posterior. (E) DiI-labeled axons (open arrows) converging onto the same spot in a tangential section through supragranular cortical layers about 4.5 mm from the dye crystal. Filled arrow, terminal branching of one of the axons; asterisks, dendrites of two retrogradely labeled neurons. Scale bar, 100 μm.

The low spatial resolution of noninvasive methods prevents the analysis of local processing architectures in the human brain. However, certain aspects of microcircuitry can be studied postmortem with neuronal tracing techniques. In the visual cortex, close relations could be established between the layout of long-range intrinsic connections and the functional organization of cortical modules. These connections exhibit a high degree of topological specificity (18,19) and reciprocally link discrete clusters of supragranular pyramidal cells that are spaced at regular intervals (20,21) and share similar functional properties (22–24).

We examined the layout of long-range intrinsic connections in the posterior part of area 22 (Fig. 1A) of both hemispheres using postmortem implantation of carbocyanine dyes (25–27). To establish the extent to which interhemispheric structural differences already exist on the level of primary sensory areas, we also examined intrinsic connections in the primary auditory cortex (area 41, Fig. 1A) of a limited sample of subjects.

In posterior area 22, we found a dense plexus of tangential connections (Fig. 1D) bridging distances of up to 7 mm. Stained structures consisted of anterogradely labeled axons (Fig. 1E) and retrogradely labeled neurons (Fig. 1, B and C). Most of these neurons, especially at distances >3 mm from the site of dye implantation, were concentrated in supragranular layers and exhibited the morphological features of pyramidal cells (Fig. 1, B and C). Layer IV contained only very few labeled axons and retrogradely stained neurons. A small number of long-range axons and retrogradely labeled cells were present in layer V, and the latter were in register with labeled neurons in supragranular layers. As in the visual cortex, retrogradely labeled cells and anterogradely stained axonal terminal arbors tended to be superimposed and to form regularly spaced clusters around the injection site (Fig. 1, C to E). For interhemispheric comparison of this architectural feature, we quantified the size of the labeled clusters and their spacing (27). The size of the labeled clusters was similar in both hemispheres (Table 1 and Fig. 2C). However, the center-to-center distance of the clusters was significantly larger in the left hemisphere in all seven subjects (Table 1 and Fig. 2C). The magnitude of this difference ranged from 12 to 30% (Table 1). Comparative data from the primary auditory cortex in two of the subjects, which both showed a significant asymmetry in posterior area 22, did not reveal interhemispheric differences in spacing, suggesting that the observed differences in posterior area 22 are not reflecting a general interhemispheric asymmetry of microcircuitry in the temporal lobe (Table 2 and Fig. 2D). However, this issue needs further investigation by analysis of other components of the central auditory system.

Figure 2

(A) Same horizontal section as inFig. 1D, with dashed outlines demarcating patches. (B) Same as (A), with lines indicating measured variables. Black arrows, patch diameters; gray arrows, interpatch distances. (C) Box plots of interpatch distances and patch sizes (ordinate) in the posterior part of area 22. Boxes represent the 25th and the 75th percentile; horizontal lines indicate the 10th and 90th percentile. Measurements from the same cases are grouped together; light gray boxes indicate data from left hemispheres, and dark gray boxes indicate data from right hemispheres. Data are normalized to the respective mean value in each brain. (D) Box plots of the interpatch distances and patch sizes in the primary auditory cortex (area 41). Conventions as in (C). (E) Box plots of distances between patches and injection site in area 22. (F) Box plots of the number of labeled clusters per dye injection in area 22.

Table 1

Interpatch distances and patch sizes in area 22.

View this table:
Table 2

Interpatch distances and patch sizes in the primary auditory cortex.

View this table:

Clustering could be a reflection of interdigitating subsystems of selectively interconnected columns or of a single system of discrete, gridlike connections. We implanted both dyes, DiI and DiA, in the same block of tissue (n = 10) at distances ranging from 1.5 to 3 mm. Each crystal labeled several clusters of neurons over a distance of up to 5 mm from the injection site. In most cases (n = 8), clusters labeled by different dyes were nonoverlapping and interleaved (Fig. 3A). In only two cases, a few of the clusters overlapped (Fig. 3B). The overlap included the complete area of single clusters, indicating that the two injections labeled the same connectivity system. This suggests the first scenario, i.e., interdigitating subsystems of selectively interconnected columns. Because cluster size was similar in the two hemispheres, the wider spacing implies that more subsystems can be included per surface unit in the left than in the right area 22. This interpretation is supported by the interhemispheric comparison of the number of clusters labeled by a single dye implantation and the distance of clusters from the implantation site (Fig. 2, E and F). The average distance of the clusters from the implantation site was similar in both hemispheres (Fig. 2E). Accordingly, the number of labeled clusters per injection was slightly smaller in the left than in the right hemisphere (Fig. 2F), suggesting that left area 22 contains fewer elements of an individual connectivity system per unit area.

Figure 3

(A and B) Camera lucida drawings of the patchy patterns of long-range intrinsic connections in tangential sections through area 22 after simultaneous injection of DiI and DiA. Dark gray, DiI; light gray, DiA; asterisks, implantation sites. m, medial; a, anterior; d, dorsal. (A) Pattern with interdigitating patches (arrows). (B) Pattern with two double-labeled patches (arrows). (C) Estimation of the number of different subsystems of interconnected columns in left and right area 22. The error bars on the top of each column give the standard error of the mean. Interhemispheric differences were significant (Mann-Whitney U, P = 0.01).

Calculations based on the measured cluster size and spacing suggest that the left area 22 can contain about 30% more distinct subsystems within the same volume of tissue (Fig. 3C). Because area 22 is larger in the dominant than in the nondominant hemisphere (11), interdigitation of more columnar subsystems can be achieved without reducing the number of columns constituting a particular subsystem. Thus, if these subsystems are tuned to different features, as in the visual cortex, more feature domains could be represented in the left than in the right area 22 without reduction of the number of neurons devoted to a particular feature. This, then, is likely to permit a more fine-grained analysis of feature domains, and we propose that this is related to special functions of auditory processing such as are required, e.g., for the analysis of speech.

An increase in cluster spacing has also been noted in higher areas of the visual cortex, and it has been proposed that this might reflect the need to represent more complex feature constellations in higher areas (28).

Neuropsychological studies in patients with temporal lesions or callosotomy suggest that the nondominant hemisphere has a limited capacity to handle language and to substitute for the loss or inaccessibility of the language areas in the dominant hemisphere (29). We propose that this limitation may be related to the different connectivity patterns identified in the present study.

Because early lesions in language areas of the dominant hemisphere can be compensated by the nondominant hemisphere (30), it is conceivable that the observed interhemispheric differences in the layout of intrinsic connections are at least in part due to use-dependent modifications of circuitry during early development. Future studies will have to determine the exact time course of postnatal maturation of this connection system.

  • * To whom correspondence should be addressed: E-mail: galuske{at}mpih-frankfurt.mpg.de

REFERENCES AND NOTES

View Abstract

Navigate This Article