Abstract
We used diffusion tensor imaging to assess callosal morphology in 35 pairs of monozygotic twins, of which 17 pairs were concordant for handedness and 18 pairs were discordant for handedness. Functional hemispheric language dominance was established for each twin member using fMRI, resulting in 26 twin pairs concordant and 9 twin pairs discordant for language dominance. On the basis of genetic models of handedness and language dominance, which assume one “right shift” (RS) gene with two alleles, an RS+ allele biasing toward right-handedness and left cerebral language dominance and an RS− allele leaving both asymmetries to chance, all twins were classified according to their putative genotypes, and the possible effects of the gene on callosal morphology was assessed. Whereas callosal size was under a high genetic control that was independent of handedness and language dominance, twin pairs with a high probability of carrying the putative RS+ allele showed a connectivity pattern characterized by a genetically controlled, low anisotropic diffusion over the whole corpus callosum. In contrast, the high connectivity pattern exhibited by twin pairs more likely to lack the RS+ allele was under significantly less genetic influence. The data suggest that handedness and hemispheric dominance for speech production might be at least partly dependent on genetically controlled processes of axonal pruning in the corpus callosum.
INTRODUCTION
Individual variations in brain structures result from a complex interaction between genetic factors and environmental influences. Brain volume, for example, is under substantial genetic control with heritability estimates at around 90% (Baare et al., 2001; Bartley, Jones, & Weinberger, 1997). In contrast, the overall gyral patterning of the cortex is astonishingly dissimilar in identical twins (Bartley et al., 1997), suggesting underlying environmental influences. Heritability of the cross-sectional area of the corpus callosum, the main fiber tract connecting the two cerebral hemispheres, is high, with estimates ranging between 79% and 94% in adults (Hulshoff Pol et al., 2006; Scamvougeras, Kigar, Jones, Weinberger, & Witelson, 2003; Pfefferbaum, Sullivan, Swan, & Carmelli, 2000) and remaining stable over a lifetime (Pfefferbaum, Sullivan, & Carmelli, 2004). Diffusion tensor imaging of the corpus callosum has yielded a heritability estimate of fractional anisotropy (FA), which is a measure of the coherence of the white matter tracts, at around 60% (Chiang et al., 2009).
Because the corpus callosum plays the major role in interhemispheric communication, its size and morphological structure may also influence functional hemispheric asymmetries. Some studies have shown, for example, that the corpus callosum is larger in left-handers than in right-handers (Habib et al., 1991; Witelson, 1985, 1989), which may explain why left-handers are more likely to be right-hemisphere dominant for language or to have a greater degree of right hemispheric involvement. Other studies, though, have found no differences in callosal morphology between handedness groups (Preuss et al., 2002; Jancke, Staiger, Schlaug, Huang, & Steinmetz, 1997). Handedness is at best an indirect measure of cerebral asymmetry for language, but even when more direct measures are used, such as fMRI or dichotic listening, the results have been conflicting, with some studies suggesting higher connectivity in more lateralized brains (Josse, Seghier, Kherif, & Price, 2008; Westerhausen et al., 2006) and others the reverse (Gootjes et al., 2006; Yazgan, Wexler, Kinsbourne, Peterson, & Leckman, 1995; Hines, Chiu, McAdams, Bentler, & Lipcamon, 1992).
In this article, we examine hemispheric asymmetry and callosal morphology in monozygotic twins. Some 20–25% of monozygotic twins are of opposite handedness (Annett, 2002; McManus, 1980) with some even showing reversed cerebral asymmetries (Badzakova-Trajkov, Haberling, & Corballis, 2010; Sommer, Ramsey, Mandl, & Kahn, 2002). Although this may seem counter to the idea that handedness is genetically inherited, genetic models can account for discordant handedness in twins by postulating a chance influence. The simplest such models propose a single gene with two alleles, one disposing to right-handedness and the other leaving the direction of handedness to chance (Annett, 2002; McManus, 2002). There is abundant evidence that right-handedness and left cerebral dominance for language are correlated, suggesting that the same laterality gene is involved in both (Annett, 2002; McManus, 2002). In support of these two-allele models, Geschwind, Miller, DeCarli, and Carmelli (2002) compared monozygotic and dizygotic twins and found the genetic influence on structural asymmetries of frontal, temporal, and parietal brain volumes to be much higher in right-handed twin pairs than in pairs in which at least one twin was left-handed. Nevertheless, asymmetry of the planum temporale, as a structural marker for language lateralization, has been shown to be only weakly influenced by genes in twin studies, with some evidence for a diminished asymmetry in the left-handed twin member only (Eckert et al., 2002; Steinmetz, Herzog, Schlaug, Huang, & Jancke, 1995). However, a recent study investigating the planum temporale in 274 subjects has shown a reduced asymmetry in subjects with familial sinistrality whereas no immediate handedness effect was evident (Tzourio-Mazoyer et al., 2010).
Here, we choose Annett's (2002) version of the genetic model to generate estimates of the incidence of the laterality gene in monozygotic twins of differing combinations of handedness and cerebral dominance, because she has applied her model specifically to twins (Annett, 2003) and it provides a reasonable fit to twin data on handedness and cerebral asymmetry (Badzakova-Trajkov, Haberling, & Corballis, 2010). In her terminology, the laterality gene is termed the “right shift” (RS) gene, with one allele (RS+), which biases toward right-handedness and left cerebral dominance, and the other (RS−), which leaves both asymmetries to chance. This gives rise to three genotypes, labeled RS++, RS+−, and RS−−. The RS+ allele operates in a dominant–recessive fashion on cerebral dominance, so that RS++ and RS+− genotypes will show typical left-hemisphere dominance and RS−− genotypes have the direction of cerebral dominance assigned at random. For handedness, though, the RS+ operates additively. In adapting the model for twins, Annett (2003) assumes a 14% incidence of left-handedness (slightly higher than in the singly born), with a shift of 1.72 standard deviations to the right in RS++ individuals and 0.73 standard deviations to the right in RS+− individuals. Although the genetic influence in RS− individuals is assumed to be neutral, she assumes that the ratio of right- to left-handedness in RS− individuals is 65:35 because of environmental pressures toward right-handedness.
Table 1A reproduces Annett's predictions of the genotypes for each handedness combination per 1000 pairs of twins. Following Annett, we assume that all RS++ and RS+− pairs and 25% of RS− pairs will have left hemispheric representation of language. Thus, we can compute the probabilities of the different genotypes for twin pairs that are right-handed and left cerebrally dominant for language (RS+ group) and for twin pairs where at least one member is left-handed or right cerebrally dominant (RS− group), as shown in part B of Table 1. From Table 1, we can convert the percentage to the probability of the RS+ allele in the RS+ and RS− group (see part C of Table 1). The expected probability of carrying at least one copy of the RS+ allele in twins who are both right-handed and left cerebrally dominant for language is .971, whereas in twins who at least one member is left-handed or right cerebrally dominant for language it is .460. Comparison of these groups should therefore provide an assessment of the effects of the RS+ gene on callosal morphology.
Calculation of the Probability of the RS+ Allele in the RS+ and RS− Groups, Respectively
A. Estimates of Handedness Combinations per 1000 Pairs of Twins According to Annett's Table 3 (2003) . | . | ||||
---|---|---|---|---|---|
Genotype . | Handedness Combinations . | ||||
Right–Right . | . | Right–Left . | Left–Left . | ||
RS++ | 305 | 20 | 0 | ||
RS+− | 369 | 113 | 8 | ||
RS− | 78 | 84 | 23 | ||
B. Genotypic Probabilities for the RS+ and RS− Groups | |||||
Group | RS++ | RS+− | RS− | Total | |
RS+ | .305 | .369 | .020 | .694 | |
RS− | .020 | .121 | .165 | .306 | |
C. Probability of the RS+ Allele in the RS+ and the RS− Group | |||||
Phenotype | RS++ | RS+− | RS− | p(RS+) | |
RS+ group | .439 | .532 | .029 | .971 | |
RS− group | .065 | .395 | .539 | .460 |
A. Estimates of Handedness Combinations per 1000 Pairs of Twins According to Annett's Table 3 (2003) . | . | ||||
---|---|---|---|---|---|
Genotype . | Handedness Combinations . | ||||
Right–Right . | . | Right–Left . | Left–Left . | ||
RS++ | 305 | 20 | 0 | ||
RS+− | 369 | 113 | 8 | ||
RS− | 78 | 84 | 23 | ||
B. Genotypic Probabilities for the RS+ and RS− Groups | |||||
Group | RS++ | RS+− | RS− | Total | |
RS+ | .305 | .369 | .020 | .694 | |
RS− | .020 | .121 | .165 | .306 | |
C. Probability of the RS+ Allele in the RS+ and the RS− Group | |||||
Phenotype | RS++ | RS+− | RS− | p(RS+) | |
RS+ group | .439 | .532 | .029 | .971 | |
RS− group | .065 | .395 | .539 | .460 |
In a further analysis, we removed twin pairs with left-handers in the immediate family from the RS+ group, which should further increase the probability of carrying the RS+ allele (McManus, 1995). Although familial sinistrality has been criticized as a reliable measure in that it depends on family size and the accuracy of the handedness reports, we show it to enhance the difference in callosal morphology between the putative RS+ and RS− groups. In a recent study, FS+ subjects showed an increase in gray matter, especially in the right hemisphere, which led to an absence of hemispheric gray matter asymmetry typically found in FS− subjects (Tzourio-Mazoyer et al., 2010). This might be explained by a reduced synaptic pruning in subjects with familial left-handedness. Annett (1991) further suggested that cerebral asymmetry might arise through pruning of neurons in the corpus callosum. We therefore hypothesize that the corpus callosum might be smaller and/or the FA values lower in the group with the highest likelihood of carrying the RS+ allele.
METHODS
Participants
A total of 35 monozygotic twin pairs (14 male/male, 21 female/female; mean age = 24.7 years, SD = 8.8 years, range = 15–49 years) took part. These were a subgroup of monozygotic twins drawn from a larger pool of volunteers who participated in an earlier study (Badzakova-Trajkov, Haberling, & Corballis, 2010). A short zygosity questionnaire was filled out by each twin member separately, asking about their physical resemblance, the difficulty for family and friends to tell them apart, and their hair and eye color in childhood (Christiansen et al., 2003). In seven twin pairs, where zygosity could not be established beyond doubt based on the answers and the physical appearance, additional DNA testing was carried out. DNA was extracted from mouth swab samples, and the individuals were genetically typed using the multiplex PCR kit AmpFISTR Identifiler (Applied Biosystems, Carlsbad, CA). The kit compares the STR (short tandem repeat) profiles on 15 highly polymorphic loci. Twins were considered as monozygotic when no differences in the 15 loci were detected (Yang, Tzeng, Tseng, & Huang, 2006).
On the basis of writing hand, 16 pairs were both right-handed (RR pairs; 7 male/male, 9 female/female; mean age = 22.6 years, SD = 7.2 years), 1 pair were both left-handed (LL pair; 1 female/female; age = 21 years), and 18 pairs were of opposite handedness (RL pairs; 7 male/male, 11 female/female; mean age = 26.8 years, SD = 10.0 years). Twin pairs with discordant handedness were deliberately over-represented in the study. Each twin also filled out a handedness inventory, in which they indicated the preferred hand in 12 activities: writing, throwing a ball, holding a racquet, lighting a match, cutting with scissors, threading a needle, sweeping with a broom (top hand), shoveling, dealing cards, hammering, holding a toothbrush, and unscrewing a lid (Annett, 2002). They gave two ticks for the preferred hand or one tick for each hand if there was no preference. A laterality index (LI) was then calculated from the formula 100 × (R − L)/(L + R), where R and L represent the number of ticks for the right and left hands, respectively. Although the primary criteria for handedness was the writing hand, as has been suggested by McManus (2002), all left-handed twin members had LIs < 40 and all right-handed twin members had LIs > 50, resulting in a classification scheme similar to that used in previous research (e.g., Whitehouse & Bishop, 2009).
Information concerning familial left-handedness was collected by asking the participants to state the handedness of their immediate family members. Familial sinistrality was defined as the presence (FS+) or absence (FS−) of at least one left-hander among the parents or siblings of the subject.
Word Generation Task
To assess language dominance, all participants undertook a word generation task adapted from the Controlled Oral Word Association Test (Lezak, 1995) during fMRI. Participants were asked to silently generate as many words as possible, starting with a designated letter (F, A, S, B, and M), which was randomly and centrally projected onto a screen (in Courier New black font, size 50). They were instructed not to use proper names or the same words with different endings. The letters were presented for 30 sec, followed by a 30-sec baseline period that consisted of a black cross, resulting in an acquisition time of 5 min. Before the experiment, all participants completed a comparable short version of the task, naming the words overtly, to obtain a behavioral performance measure.
Image Acquisition and Processing
MRI scanning was performed on a 1.5-T Siemens Avanto scanner (Erlangen, Germany). A T1-weighted structural image was acquired using a 3-D magnetization prepared rapid gradient echo sequence with 176 axial slices parallel to the AC–PC line, ensuring whole-brain coverage. The following parameters were used: repetition time (TR) = 11 msec, echo time (TE) = 4.94 msec, flip angle = 158°, field of view (FOV) = 256 × 256 mm2. Slice thickness was 1 mm, and the interslice gap was 0, resulting in an isotropic voxel resolution of 1 × 1 × 1 mm. For the diffusion weighted images, a single-shot spin echo sequence along 30 diffusion gradient directions with a b0 of 1000 sec/mm2 was used with the following parameters: TR = 6601 msec, TE = 101 msec, FOV = 230 mm, in-plane resolution = 1.8 × 1.8 mm, slice thickness = 3 mm. In addition, one image without diffusion weighting was acquired. The sequence was repeated twice, resulting in an acquisition time of approximately 7 min. The EPI acquisition had the following parameters: TR = 2500 msec, TE = 50 msec, flip angle = 90°, FOV = 192 × 192 mm2, matrix size = 64 × 64, 29 slices parallel to AC–PC line, slice thickness = 3 mm, interslice gap: 25% = 0.8 mm.
Diffusion Image Preprocessing and Analyses
The tractography-based segmentation of the corpus callosum was carried out using the FDT toolbox implemented in the FMRIB software library (FSL; www.fmrib.ox.ac.uk/fsl/). On the basis of the probability that a voxel in a seed mask connects to a predefined target mask, the anatomical connectivity pattern of a brain region can be determined (Behrens, Johansen-Berg, et al., 2003). This method was applied to the callosal fiber tracts as follows.
For each participant, the corpus callosum was manually outlined on the midsagittal slice of the individual T1 images by applying the following steps. First, the structural T1-weighted images were registered to an optimized individual target brain in the Talairach coordinate system (Kochunov et al., 2002) available at the BrainMap website (brainmap.org/ale/index.html). A rigid body transformation with 6 degrees of freedom (3 translations, 3 rotations) was applied without scaling the brain to maintain individual brain sizes. Second, the images were segmented into gray and white matter using the automated segmentation tool implemented in FSL (Zhang, Brady, & Smith, 2001). Third, the corpus callosum was manually outlined on the white matter segmentation image. To assess reliability, a second independent measurement of the callosal area for 15 randomly chosen participants was made by a different rater, resulting in an intraclass correlation of .96. In addition, total brain volume was calculated with SIENAX, an algorithm for automated brain extraction implemented in FSL (Smith et al., 2002).
To segment the corpus callosum into three distinct subdivisions, specific cortical target masks were created on the Montreal Neurological Institute (MNI) template. The most anterior mask comprised the frontal lobe but spared the motor cortex. The motor cortex was derived from the Juelich histological atlas implemented in FSL and included M1 and premotor cortex. A further mask included the temporal and parietal lobes, whereas the most posterior mask comprised the occipital lobes. To ensure a good alignment of these masks, the individual T1 images were nonlinearly registered to the MNI template and the resulting registration matrices were inversed and applied to the target masks, transferring them back into each participant's native space.
The diffusion data were corrected for eddy currents and head motion, and the two acquisitions were averaged to improve signal-to-noise ratio. Diffusion tensors were fitted at each voxel, and FA maps were generated. Then, the probability distribution of fiber direction at each voxel was calculated using previously described methods (Behrens, Woolrich, et al., 2003). For each participant, probabilistic tractography was run from each voxel in the corpus callosum to each target mask by drawing 5000 random samples with a step length of 0.5 mm and a curvature threshold of 0.2. The resulting value in each callosal voxel represents the number of samples reaching the relevant target mask and therefore the probability of the connections between the two. The analyses were run separately for the right and left hemispheres. Using a similar threshold as previous research, each output mask was thresholded at 10% of the maximum connectivity value to the relevant target mask (Zarei et al., 2006). Then, tracts going to the left and right hemispheres were compared, and only overlapping areas were considered for further analyses (Westerhausen, Gruner, Specht, & Hugdahl, 2009). The resulting masks were binarized and multiplied with the FA maps to extract the mean FA values. A distribution of the FA values in the twin pairs is depicted in Figure 1.
Correlation of the FA values between one twin member and the other, with annotations of handedness and cerebral asymmetry. Twin members were allocated to the x and y axes randomly. First two letters correspond to handedness (RR = both right-handed, RL = one twin right-handed and the other left-handed, LL = both left-handed), last two letters to language dominance (TT = both typical left hemispheric dominant, AT = one twin left hemispheric dominant and one twin atypical right hemispheric dominant, AA = both twins atypical right hemispheric dominant).
Correlation of the FA values between one twin member and the other, with annotations of handedness and cerebral asymmetry. Twin members were allocated to the x and y axes randomly. First two letters correspond to handedness (RR = both right-handed, RL = one twin right-handed and the other left-handed, LL = both left-handed), last two letters to language dominance (TT = both typical left hemispheric dominant, AT = one twin left hemispheric dominant and one twin atypical right hemispheric dominant, AA = both twins atypical right hemispheric dominant).
Functional Imaging Processing and Analyses
The functional images were analyzed using SPM5 software (Wellcome Department of Imaging Neuroscience, London, UK; www.fil.ion.ucl.ac.uk). First, the standard preprocessing steps (realignment, coregistration, normalization, and smoothing) were applied. The functional scans were realigned to the first image of the session, and the mean of the functional volumes was calculated. The T1-weighted structural image was then coregistered with the previously obtained mean of the functional volumes. Then, all images were normalized into standardized stereotactic space (MNI) and spatially smoothed with an anisotropic Gaussian filter of 9 × 9 × 9 mm of FWHM. For each participant, the functional volumes were subjected to a fixed-effects analysis using the general linear model that was applied at each voxel across the whole brain. The model was set up as a boxcar function with the two alternate conditions, letter versus baseline. The resulting function was convolved with a canonical hemodynamic response function, and movement regressors were also included in the model. For the group analysis, a second-level random effects analysis was performed by applying a one-sample t test to the contrast images of the first-level analyses. A family-wise error correction was applied at p < .05, with a contiguity threshold of 10 voxels.
To establish language dominance, LIs were calculated by comparing the activity between the left and right Broca's regions using the formula LI = (L − R)/(L + R), where L and R represent activations in the left and right hemispheres, respectively. The laterality toolbox available on the SPM website was used to calculate the LIs (Wilke & Lidzba, 2007). LIs based on one single fixed statistical threshold do not yield robust or reproducible results (Jansen et al., 2006). Therefore, the toolbox applies a bootstrap algorithm to calculate about 10,000 indices based on the asymmetry of the voxel values at different thresholds yielding a robust mean LI ranging between −1 for extreme left and 1 for extreme right lateralization. For each participant, the weighted mean LI for Broca's area was computed. Broca's area was defined using the WFU Pick Atlas toolbox (Maldjian, Laurienti, Kraft, & Burdette, 2003) and included BA 44 and BA 45. The masks were smoothed with a 6-mm Gaussian filter to control for interindividual variability. In addition, the toolbox integrates a mask weighting factor that represents the relation of the volumes of the masks on the left and on the right to rule out influences of different mask sizes. A positive index within the ROI corresponded to a left hemispheric dominance (LI > 0.1) and a negative index to a right hemispheric dominance (LI < −0.1; Fernandez et al., 2003). Because of the small numbers in the bilateral group (LIs ranging from −0.1 to +0.1), those two subjects were assigned to the right hemispheric group, because their hemispheric dominance pattern is also atypical in nature.
The word generation paradigm induced significant leftward activations in the inferior frontal gyrus, including pars opercularis and pars triangularis, insula, precentral gyrus, SMA, and inferior temporal gyrus (see Figure 2; Table 2). Additional right hemispheric clusters were observed in the inferior and middle occipital gyrus. Overall, a one-sample t test revealed that the LIs were leftward asymmetric in Broca's area (M = 0.61, SE = 0.051, t(69) = 12.1, p < .001). Classification of the twin pairs according to the LIs of both twin members revealed that, overall, 26 pairs were concordant left dominant for language processing whereas 9 pairs (6 pairs of opposite handedness, 2 right-handed twin pairs, 1 left-handed twin pair) showed opposite hemispheric dominance. As discussed elsewhere, these results do not differ significantly from expectations derived from Annett's RS theory of the inheritance of handedness and cerebral asymmetry (Badzakova-Trajkov, Haberling, & Corballis, 2010).
Activation map for the Word Generation Task displayed laterally on the cortical surface of rendered brain. In addition, activations overlaid on slices of a template of 152 brains are shown.
Significantly Activated Brain Regions for the Word Generation Task
Anatomical Location . | BA . | MNI Coordinates . | t . | ||
---|---|---|---|---|---|
x . | y . | z . | |||
Word Generation | |||||
Left hemisphere | |||||
Insula | 13 | −30 | 24 | 3 | 11.86 |
Inferior frontal gyrus (p. opercularis) | 44 | −45 | 9 | 27 | 11.61 |
Inferior frontal gyrus (p. triangularis) | 45 | −45 | 24 | 24 | 10.65 |
SMA | 6 | −3 | 12 | 51 | 12.84 |
Caudate nucleus | −15 | −6 | 24 | 8.68 | |
Superior frontal gyrus | 10 | −27 | 60 | 9 | 6.66 |
Inferior temporal gyrus | 37 | −45 | −48 | −12 | 6.42 |
Putamen | −21 | 6 | 12 | 6.31 | |
Right hemisphere | |||||
Insula | 13 | 33 | 24 | 3 | 8.44 |
Caudate nucleus | 18 | 0 | 24 | 8.49 | |
Middle occipital gryus | 18 | 33 | −93 | 3 | 8.23 |
Hippocampus | 33 | −42 | 6 | 7.57 |
Anatomical Location . | BA . | MNI Coordinates . | t . | ||
---|---|---|---|---|---|
x . | y . | z . | |||
Word Generation | |||||
Left hemisphere | |||||
Insula | 13 | −30 | 24 | 3 | 11.86 |
Inferior frontal gyrus (p. opercularis) | 44 | −45 | 9 | 27 | 11.61 |
Inferior frontal gyrus (p. triangularis) | 45 | −45 | 24 | 24 | 10.65 |
SMA | 6 | −3 | 12 | 51 | 12.84 |
Caudate nucleus | −15 | −6 | 24 | 8.68 | |
Superior frontal gyrus | 10 | −27 | 60 | 9 | 6.66 |
Inferior temporal gyrus | 37 | −45 | −48 | −12 | 6.42 |
Putamen | −21 | 6 | 12 | 6.31 | |
Right hemisphere | |||||
Insula | 13 | 33 | 24 | 3 | 8.44 |
Caudate nucleus | 18 | 0 | 24 | 8.49 | |
Middle occipital gryus | 18 | 33 | −93 | 3 | 8.23 |
Hippocampus | 33 | −42 | 6 | 7.57 |
Brodmann's area (BA), MNI coordinates for the peak activation voxel, and t values are also shown. The Anatomy toolbox was used to determine the anatomical location. Brodmann's areas were derived from the Talairach Daemon.
Statistical Analysis
To assess the degree to which twin members resemble each other, intraclass correlation coefficients were computed. Intraclass correlations compare the variations within a pair with the variation between pairs and allow inferences about possible genetic influences. To compare intraclass correlations between different groupings of twins, a Fisher r-to-z transformation was applied and the significance of the difference between the two correlation coefficients was assessed. To detect potential differences in the mean values between different groupings of twins, twin pair was treated as a random factor, group as between-twin factor, and gender and age as covariates were performed. Subjects (twins) were treated as nested within each twin pair. For the analysis of the individual segments, a Bonferroni correction for multiple comparisons was applied to the alpha level by dividing it by the number of applied tests (i.e., divided by four, given that four callosal segments were investigated), resulting in a new alpha level of .013.
Because of previous reports on gender differences in language lateralization (Shaywitz et al., 1995) and callosal morphology (Davatzikos & Resnick, 1998; Habib et al., 1991), an ANCOVA on the LIs with gender as between-twin factor, age as covariate, and twin pair as a random factor was conducted. Lateralization in male twin pairs (M = 0.65, SE = 0.082) did not differ significantly from that in female twin pairs (M = 0.59, SE = 0.067), F(1, 32) = 0.395, p = .534. Further analyses revealed no gender differences on callosal volumes or FA values. Although a trend for larger total callosal volumes in male (M = 705, SE = 21.8) than in female twin pairs (M = 652, SE = 17.8), F(1, 32) = 3.56, p = .068, was observed, the effect disappeared if callosal size was taken as a proportion of the total brain volume, F(1, 32) = 0.99, p = .327. That is, the observed gender differences in callosal size might be attributed to differences in overall brain size, as has been suggested previously (Jancke et al., 1997).
RESULTS
Comparing Groups Defined by Handedness and Cerebral Asymmetry
As explained above, we combined handedness measures and LIs for Broca's area to group the twin pairs according to their likelihood of carrying the RS+ allele. Of the 35 twin pairs examined, 14 were both right-handed and left cerebrally dominant and therefore constituted the RS+ group (handedness inventory scores: M = 88.8, SD = 15.3; LI language M = 0.75, SD = 0.16), whereas the remaining 21 included at least one left-handed or right cerebrally dominant member and were placed in the RS− group (handedness inventory scores: M = 23.0, SD = 80.5; LI language M = 0.52, SD = 0.51).
Intraclass correlations separately for the RS+ and the RS− group are shown in Table 3. For measures of callosal volume, they are high and reveal no differences between the two groups. For callosal FA, though, the intraclass correlations were significantly higher for the RS+ group (r = .882, p < .001) than the RS− group (r = .519, p = .005, z = 2.12, p = .034) over the whole corpus callosum.
Intraclass Correlations for the Two Genetic Groups (RS+/RS−)
. | RS+ Group (n = 14) . | RS− Group (n = 21) . | z . | p . |
---|---|---|---|---|
Area | ||||
Corpus callosum | .931, p < .001 | .932, p < .001 | −.02 | .984 |
Corpus callosum/white matter ratio | .972, p < .001 | .936, p < .001 | 1.1 | .271 |
Anterior | .535, p = .017 | .466, p = .013 | .24 | .810 |
Middle | .569, p = .011 | .506, p = .007 | .23 | .811 |
Parieto-temporal | .808, p < .001 | .802, p < .001 | .04 | .968 |
Posterior | .683, p = .002 | .542, p = .004 | .60 | .549 |
FA | ||||
Corpus callosum | .882, p < .001 | .519, p = .006 | 2.12 | .034** |
Anterior | .797, p < .001 | .325, p = .067 | 1.97 | .049* |
Middle | .456, p = .039 | .269, p = .109 | .57 | .569 |
Parieto-temporal | .760, p < .001 | .503, p = .008 | 1.16 | .246 |
Posterior | .552, p = .014 | .511, p = .007 | .15 | .881 |
. | RS+ Group (n = 14) . | RS− Group (n = 21) . | z . | p . |
---|---|---|---|---|
Area | ||||
Corpus callosum | .931, p < .001 | .932, p < .001 | −.02 | .984 |
Corpus callosum/white matter ratio | .972, p < .001 | .936, p < .001 | 1.1 | .271 |
Anterior | .535, p = .017 | .466, p = .013 | .24 | .810 |
Middle | .569, p = .011 | .506, p = .007 | .23 | .811 |
Parieto-temporal | .808, p < .001 | .802, p < .001 | .04 | .968 |
Posterior | .683, p = .002 | .542, p = .004 | .60 | .549 |
FA | ||||
Corpus callosum | .882, p < .001 | .519, p = .006 | 2.12 | .034** |
Anterior | .797, p < .001 | .325, p = .067 | 1.97 | .049* |
Middle | .456, p = .039 | .269, p = .109 | .57 | .569 |
Parieto-temporal | .760, p < .001 | .503, p = .008 | 1.16 | .246 |
Posterior | .552, p = .014 | .511, p = .007 | .15 | .881 |
The z transformation and associated p values are also shown.
*Nominal significance level at p < .05, but not significant after Bonferroni correction.
**Significant at p < .05.
The mean volumes and FAs are shown in Table 4. The RS+ and RS− groups did not differ significantly in volumes, but the FA values were significantly lower for the corpus callosum as a whole for the RS+ group (M = 0.538, SE = 0.008) than for the RS− group (M = 0.563, SE = 0.007), F(1, 30) = 4.77, p = .037. This effect was also nominally significant for the anterior corpus callosum but did not reach significance with Bonferroni adjustment.
Mean and Standard Deviation of the Callosal Area and the FA Values
. | All Twins (n = 35) . | RS+ Group (n = 14) . | RS− Group (n = 21) . | RS+/FS− (n = 9) . | RS−/FS+ (n = 26) . |
---|---|---|---|---|---|
Area | |||||
Corpus callosum | 673 ± 86.1 | 666 ± 80.2 | 678 ± 90.4 | 678 ± 91.5 | 671 ± 85.0 |
Anterior | 209 ± 48.1 | 200 ± 46.8 | 215 ± 48.6 | 210 ± 51.0 | 208 ± 47.5 |
Middle | 174 ± 44.9 | 159 ± 46.5 | 184 ± 41.4 | 159 ± 52.5 | 179 ± 41.2 |
Parieto-temporal | 198 ± 47.2 | 192 ± 57.5 | 202 ± 39.0 | 200 ± 65.2 | 197 ± 39.9 |
Posterior | 67 ± 20.9 | 65 ± 24.5 | 69 ± 18.2 | 69 ± 30.0 | 67 ± 18.2 |
FA | |||||
Corpus callosum | .55 ± .034 | .54 ± .040 | .56 ± .027** | .53 ± .035 | .56 ± .031** |
Anterior | .59 ± .051 | .58 ± .055 | .61 ± .046* | .58 ± .056 | .61 ± .047* |
Middle | .52 ± .048 | .51 ± .050 | .53 ± .046 | .51 ± .047 | .53 ± .047 |
Parieto-temporal | .64 ± 042 | .63 ± .048 | .65 ± .935 | .62 ± .040 | .65 ± .039* |
Posterior | .72 ± .046 | .72 ± .055 | .72 ± .040 | .70 ± .054 | .72 ± .046 |
. | All Twins (n = 35) . | RS+ Group (n = 14) . | RS− Group (n = 21) . | RS+/FS− (n = 9) . | RS−/FS+ (n = 26) . |
---|---|---|---|---|---|
Area | |||||
Corpus callosum | 673 ± 86.1 | 666 ± 80.2 | 678 ± 90.4 | 678 ± 91.5 | 671 ± 85.0 |
Anterior | 209 ± 48.1 | 200 ± 46.8 | 215 ± 48.6 | 210 ± 51.0 | 208 ± 47.5 |
Middle | 174 ± 44.9 | 159 ± 46.5 | 184 ± 41.4 | 159 ± 52.5 | 179 ± 41.2 |
Parieto-temporal | 198 ± 47.2 | 192 ± 57.5 | 202 ± 39.0 | 200 ± 65.2 | 197 ± 39.9 |
Posterior | 67 ± 20.9 | 65 ± 24.5 | 69 ± 18.2 | 69 ± 30.0 | 67 ± 18.2 |
FA | |||||
Corpus callosum | .55 ± .034 | .54 ± .040 | .56 ± .027** | .53 ± .035 | .56 ± .031** |
Anterior | .59 ± .051 | .58 ± .055 | .61 ± .046* | .58 ± .056 | .61 ± .047* |
Middle | .52 ± .048 | .51 ± .050 | .53 ± .046 | .51 ± .047 | .53 ± .047 |
Parieto-temporal | .64 ± 042 | .63 ± .048 | .65 ± .935 | .62 ± .040 | .65 ± .039* |
Posterior | .72 ± .046 | .72 ± .055 | .72 ± .040 | .70 ± .054 | .72 ± .046 |
They are listed for the total sample, separately for the RS+ and RS− group, and separately for the RS+/FS− group and the RS−/FS+ group, respectively.
*Nominal significance level at p < .05, but not significant after Bonferroni correction.
**Significant at p < .05.
Comparing Groups Additionally Defined by Familial Sinistrality
Familial sinistrality was added as an additional factor to enhance the homogeneity in the RS+ group. In a first step, the twins were regrouped such that in the first group twin pairs who were both right-handed and left cerebral dominant for language and who had no immediate family member that is left-handed were included (RS+/FS− group, n = 9). The second group contained the remaining twin pairs of the RS+ group but who had a history of familial sinistrality (RS+/FS+ group, n = 5). The third group (RS− group, n = 21) remained the same, given that in this group a distinction between FS+ and FS− is not meaningful. A preliminary analysis run on the FA values over the whole corpus callosum revealed a significant group effect, F(2, 30) = 3.32, p = .050. Post hoc analysis with Bonferroni adjustment revealed that the RS+/FS− group showed lower FA values over the whole corpus callosum (M = 0.531, SE = 0.010) than the RS− group (M = 0.563, SE = 0.007, p = .015), whereas the RS+/FS+ group did not differ in the FA values compared with the RS− group (M = 0.551, SE = 0.015, p = .457). Therefore, in a second step the RS+/FS+ twin pairs were placed in the RS− group and compared with the RS+/FS− group, which contains the twins most likely to be homozygous for the RS+ allele, as shown in Table 4. Subsequent analysis revealed that the RS+/FS− group showed lower FA values over the whole corpus callosum (M = 0.531, SE = 0.010) than the new RS−/FS+ group (M = 0.561, SE = 0.006), F(1, 31) = 6.15, p = .019. The same effect was evident in the prefrontal segment, F(1, 31) = 4.20, p = .049, and the parieto-temporal segment, F(1, 31) = 6.30, p = .018, but did not reach significance after Bonferroni adjustment.
Could Frequency of Interhemispheric Transfer Play a Role?
The foregoing analyses offer support for the view that the RS+ gene plays a role in reducing callosal connectivity between the cerebral hemispheres. An alternative possibility, though, is that callosal connectivity is increased in those with opposite hemispheric asymmetry for hand control and language, perhaps because activities such as writing require interhemispheric transfer. The sheer frequency of interhemispheric transfer might therefore have increased connectivity. To test this possibility, all twins were reclassified into those having hand control and language localized in the same hemisphere and those having localized those two functions in opposite hemispheres. The first group consisted of 46 right-handers with left cerebral language representation and 5 left-handers with right cerebral language representation, whereas the second consisted of 4 right-handers with right cerebral language and 15 left-handers with left cerebral language. Then, all twins were reclassified into those with consistent and those with inconsistent asymmetries for hand control and language, as shown in Table 5. If the frequency of the interhemispheric transfer increases connectivity, we might expect that twin members who have the two functions localized in opposite hemispheres should exhibit higher FA values in the corpus callosum than their cotwins who support both functions unilaterally. Thus, a paired t test on the FA values in the 15 discordant twin pairs was carried out, but no significant effect was evident (p = .450).
Number of Twin Pairs Allocated to the Different Genotypes and Their Cerebral Connectivity Pattern
Phenotype . | Connections . | Total . | Allele . | ||
---|---|---|---|---|---|
Both Intra . | Both Inter . | Discordant . | |||
RR/LL | 14 | – | – | 14 | RS+ |
RL/LL | – | – | 12 | 12 | RS− |
LL/LL | – | – | – | – | RS− |
RR/LR | – | – | 1 | 1 | RS− |
RL/RL | – | 2 | – | 2 | RS− |
RL/LR | 4 | – | – | 4 | RS− |
LL/RL | – | – | 1 | 1 | RS− |
RR/RR | – | – | 1 | 1 | RS− |
RL/RR | – | – | – | – | RS− |
LL/RR | – | – | – | – | RS− |
Total | 18 | 2 | 15 | 35 | RS− |
Phenotype . | Connections . | Total . | Allele . | ||
---|---|---|---|---|---|
Both Intra . | Both Inter . | Discordant . | |||
RR/LL | 14 | – | – | 14 | RS+ |
RL/LL | – | – | 12 | 12 | RS− |
LL/LL | – | – | – | – | RS− |
RR/LR | – | – | 1 | 1 | RS− |
RL/RL | – | 2 | – | 2 | RS− |
RL/LR | 4 | – | – | 4 | RS− |
LL/RL | – | – | 1 | 1 | RS− |
RR/RR | – | – | 1 | 1 | RS− |
RL/RR | – | – | – | – | RS− |
LL/RR | – | – | – | – | RS− |
Total | 18 | 2 | 15 | 35 | RS− |
Allocation to the genetic groups is also shown. First two letters correspond to handedness (RR = both right-handed, RL = one twin right-handed and the other left-handed, LL = both left-handed), last two letters to language dominance (LL = both typical left hemispheric dominant, LR = first twin left hemispheric dominant and second twin atypical right hemispheric dominant, RL = first twin atypical right hemispheric dominant and second twin left hemispheric dominant, RR = both twins atypical right hemispheric dominant).
DISCUSSION
For callosal volume, the high intraclass correlations suggest a strong genetic component, regardless of handedness or cerebral dominance. This finding fits well with an earlier study reporting intraclass correlations of .87 for total callosal area in monozygotic twins (Scamvougeras et al., 2003). Diffusion tensor imaging allows the assessment not only of the macrostructure of callosal tracts but also of certain characteristics of their microstructure. FA, for example, reflects the degree of the alignment or integrity of the underlying tissue structure. That is, high anisotropy might indicate a greater number and/or greater density of axons and a stronger proportion of myelin material (Chepuri et al., 2002). Consequently, the intraclass correlations observed for the FA values were generally lower than those for the volumetric measurements, suggesting dynamic adaptations of the fiber tracts. For example, the FA values in the body of the corpus callosum have been shown to correlate with bimanual coordination skills (Johansen-Berg, Della-Maggiore, Behrens, Smith, & Paus, 2007), and in professional musicians the number of hours practiced on the piano in childhood correlates with FA values in the body of the corpus callosum (Bengtsson et al., 2005).
Using Annett's (2003) predictions for the probability of the RS+ allele in different phenotypes, the twins were grouped according to their likelihood of carrying the RS+ allele. Although no effect on callosal volume was found, those allocated to the RS+ group showed lower FA values throughout the whole corpus callosum than those allocated to the RS− group. The relationship between callosal volume and number of fibers remains somewhat unclear given that increased callosal area might be associated with decreased fiber density. In macaques, a lack of correspondence between number of fibers and callosal areas has been reported (Lamantia & Rakic, 1990b), whereas in men callosal area was found to be correlated with the number of fibers with thin axons but not with large ones (Aboitiz, Scheibel, Fisher, & Zaidel, 1992). In contrast, indices of anisotropy are thought to be sensitive to the density and orientation of axons (Beaulieu, 2002). Moreover, a recent imaging study showed that the FA values in the corpus callosum are related to axonal density, intraaxonal volume fraction within the white matter, and axonal diameter (Alexander et al., 2010). The low FA values in those allocated to the RS+ group might therefore indicate diminished interhemispheric connectivity. Indeed, the lowest FA values were observed in the twin pairs of the RS+ group with no left-handedness in the immediate family. In right-handers without familial sinistrality, the planum temporale asymmetry is most pronounced (Tzourio-Mazoyer et al., 2010). Concurrently, we found low connectivity in the parietotemporal segment in this group, which could indicate a reduced interhemispheric connectivity related to the enhanced planum temporale asymmetry. Differences in interhemispheric communication might influence individual cognitive processing styles. For example, FS+ individuals outperform FS− subjects in an episodic memory task where the participants are required to recall previously seen words, but not in an implicit memory task where participants have to complete word fragments (Christman & Propper, 2001). The episodic memory task involves integration of processing between the hemispheres whereas the implicit memory task is supported unilaterally. That is, in individuals with familial sinistrality, the interhemispheric communication might be enhanced, which is advantageous for some memory processes but not for all.
In terms of Annett's RS theory, the data suggest that the putative RS+ gene exerted some control over interhemispheric connectivity and that in the absence of this control connectivity was subject to a random component. Annett (1991) suggested that left hemispheric dominance for language arises from axonal pruning of the corpus callosum. That is, the development of the corpus callosum is initially characterized by an overproduction of neurons, axons, and synapses that are subsequently eliminated. In monkeys, for example, 70% of the callosal axons are eliminated between birth and adulthood (LaMantia & Rakic, 1990a). The mechanism of selection and elimination of axons is poorly understood and seems to depend on a number of factors, such as sensory input, competition among axonal systems for chemotropic substances, hormones, and expression of molecules that identify appropriate targets (Innocenti & Price, 2005).
Leftward asymmetry for processing auditory stimuli is present from birth onward, probably because of a stronger excitability of the left auditory areas but without discrimination between linguistic and nonlinguistic stimuli (Dehaene-Lambertz, Hertz-Pannier, & Dubois, 2006). For example, in 4-month-old infants, higher ERPs were recorded over the left than the right hemisphere for both tones and syllables (Dehaene-Lambertz, 2000). Concurrently, fMRI activation to forward and backward speech stimuli were greater in the left than the right temporal lobe in 3-month-old infants (Dehaene-Lambertz, Dehaene, & Hertz-Pannier, 2002). The development of language functions in childhood and early adolescence is accompanied by the thinning of cortical language areas. For example, better performance in generating words in the verbal fluency task was accompanied by the thinning of gray matter in language areas in 9- to 23-year-old children and young adults (Porter, Collins, Muetzel, Lim, & Luciana, 2011). This decrease in gray matter is accompanied by an increase in white matter. Both processes have been attributed to axonal pruning. That is, nonpreferred connections are removed whereas those that support frequent information transfer are retained. Callosal maturation is far from completed by birth with the structure increasing in size well into adulthood, probably because of ongoing myelination and increase in axonal size (Keshavan et al., 2002). This late white matter maturation might interplay with the thinning of the cortical gray matter areas, resulting in the establishment and maintenance of hemispheric dominance.
The human brain is characterized by a variety of functional hemispheric asymmetries with at least some of them evolving independently (Badzakova-Trajkov, Haberling, Roberts, & Corballis, 2010). Different combinations of these hemispheric asymmetries seem to exert unique influences on the interhemispheric connectivity (Haberling, Badzakova-Trajkov, & Corballis, 2011). Nevertheless, we found no evidence that differences in callosal connectivity could be attributed to more frequent interhemispheric coordination between language and hand motor functions, although the possibility that callosal connectivity might be influenced by different patterns of hemispheric asymmetry cannot be excluded.
Our data suggest that axonal pruning in the corpus callosum might be influenced by the same genetic factors that also control handedness and cerebral dominance for language. On this view, individuals who carry the RS+ allele might exhibit enhanced axonal pruning in the corpus callosum, reducing interhemispheric connectivity and leading to right-handedness and left hemispheric dominance for language. Individuals who lack this allele, in contrast, might be right- or left-handed but generally exhibit high connectivity patterns through the corpus callosum. It has been suggested that symmetric or bihemispheric functional networks rely on the exchange of information to coordinate processing (Witelson & Nowakowski, 1991; Galaburda, Rosen, & Sherman, 1990), and thus, the enhanced interhemispheric connectivity found in left-handers (Westerhausen et al., 2004) and in those with atypical hemispheric dominance (Haberling et al., 2011) could reflect an adaptation to the underlying functional dominance pattern. According to this hypothesis, however, the twin member with hemispheric asymmetries localized in opposite hemispheres should exhibit high FA values, whereas his or her cotwin who supports those functions unilaterally should exhibit low FA values. Our data do not support this notion, given that all subjects assumed to lack the RS+ allele showed high FA values through the corpus callosum independently of their individual hemispheric dominance patterns. Thus, although the high interhemispheric connectivity found in this group might enable the enhanced information exchange that is crucial for an efficient functioning of the hemispheres in cases of atypical hemispheric dominance, it does not determine the hemispheric dominance pattern in itself. That is, the diminished axonal pruning of the corpus callosum creates the basis for a random establishment of either lateralized or bilateral networks.
In conclusion, the data reported here suggest that the genetic mechanisms involved in determining handedness and hemispheric dominance for language act on the level of the corpus callosum by exerting control over the interhemispheric connectivity and therefore influencing the degree to which unilateral and bilateral networks are established.
Acknowledgments
This work was supported by a grant to M. C. C. from the Marsden Fund (07-UOA-015) administered by The Royal Society of New Zealand.
Reprint requests should be sent to Isabelle S. Häberling, Research Centre for Cognitive Neuroscience, Department of Psychology, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand, or via e-mail: i.haberling@auckland.ac.nz.