Animal care and drug administration were performed under veterinary control according to European Union Directive 86/609/EEC and with approval from the Ethics Committee of the University Miguel Hernández of Elche and the Genèralitat Valènciana, València, Spain.

The experimental rats used in this study are the same as those used in a previous work on anterior commissure development (Lucia et al., 2018), thereby allowing us to optimize the number of rats sacrificed. Here we present an abbreviated methodology, with a more detailed explanation when necessary, and include differences in the areas studied and sample size. Material was stored at 4°C in 0.05% sodium azide in 0.1M phosphate buffer (PB; pH 7.3–7.4).

As in our previous study, rats were housed in animal quarters with controlled temperature (22–24°C) and light/darkness cycle (14/10 h, respectively). Young females (250–300 g weight) were mated at the embryonic day (E) 0. Hypothyroidism was induced by adding 0.02% methimazole (MMI, Sigma-Aldrich Co., St. Louis, MO, United States) to drinking water. Only male pups were used in this study since gender-based structural differences in the CC have been previously reported. For instance, the percentage of myelinated axons is lower in the genu (Juraska and Kopcik, 1988; Kim and Juraska, 1997) and splenium (Mack et al., 1995) in female rats, and the callosal axon diameter and g-ratio values are higher in male rats (Pesaresi et al., 2015).

Four groups of MMI treated rats were studied, as well as a control group (Figure 1). The MMI_E10 group was treated from E10 to postnatal day (P) 170, and then sacrificed (at 150, four rats were sacrificed for EM; see below). MMI_P0–21 and MMI_P0 groups were treated from P0 to P21 and from P0 to P150, respectively, both being sacrificed at P150. The MMI_P0–21 + T4_P15–21 group had the same treatment as MMI_P0–21 pups with the addition of L-thyroxine (T4) (2.4 μg/100 g of body weight; Escobar-Morreale et al., 1995) per day from P15 to P21 (sacrificed at P150). T4 was administered subcutaneously using osmotic mini-pumps (ALZET^®, model 2001; Alza Corporation, Mountain View, CA), which have a continuous pumping rate of 1 μl/h, used for 7 days during T4 treatment from P15 to 21 (see Methodological Considerations in the Discussion for further details). All MMI-treated rats received 1% KClO₄ with the MMI drinking solution until P21 in order to inhibit thyroidal iodine uptake and thyroid hormone synthesis during fetal and lactating periods. In the text, MMI_P0–21 and MMI_P0–21 + T4_P15–21 groups are referred to collectively as transient hypothyroid rats, and individually when significant differences between these two groups were found. Similarly, MMI_P0 and MMI_E10 groups are referred to as chronic hypothyroid rats. Control (C) rats were sacrificed at P150, P180 and P365. Four litters per group were used and culled to 8 pups per litter. After weaning, dams and female pups were anesthetized by 1.5–2% isoflurane inhalation (Laboratorios Dr. Esteve, S.A., Barcelona, Spain) in O₂ (0.9 L/min) and decapitated.

Using isoflurane anaesthesia, blood samples (∼1 ml) were taken from the heart ventricle using a heparinized syringe, at P15, P21, and P50 (8 rats/group; 2 per litter). Blood was centrifuged and plasma kept at –20°C. After extraction and purification of the plasma samples, total thyroxine and tri-iodothyronine plasmatic concentrations were obtained by radioimmunoassay (see details in Lucia et al., 2018).

The general procedure for our EM study has been previously described in detail (Lucia et al., 2018). Isoflurane anaesthetized rats at P150 (4 rats/group; 1 per litter) were perfused with saline and fixed with 4% paraformaldehyde, 1% glutaraldehyde, 0.1M sucrose, 0.002% CaCl₂ in 0.1M phosphate buffer (pH:7.3–7.4; PB) for 10 min. The heads were kept at 4°C during 2 h, the brains were then carefully removed and postfixed overnight, at 4°C.

The brains were sectioned into 250 μm slices in the mid-sagittal plane using a vibratome to obtain transversally cut callosal axons. From each hemisphere the two most central slices were OsO₄ postfixed (30 min at room temperature) and uranyl acetate stained (1 h at 4°C in the dark), ethanol dehydrated, immersed in propylene oxide (Lab Baker, Deventry, Holland) and embedded overnight in Araldite (Durcupan, Fluka, Buchs SG, Switzerland). Semithin toluidine blue-stained sections (1.5 μm thick) and lead citrate-contrasted ultrathin sections (60–70 nm thin) were obtained using an ultra-microtome (Ultracut UC-6, Leica, Heidelberg, Germany). Images were obtained with a transmission electron microscope (FEI Tecnai Spirit G2, Eindhoven, Netherlands), equipped with a digital camera (Morada, Soft Imaging System, Olympus). Tissue sections and perfused brains not used were stored in 0.05% sodium azide in PB at 4°C, for future studies.

The mid-sagittal transversal area of CC was obtained from outlines obtained by combining photographs from semithin sections (Figures 2E,F), and 30, 40, and 30% which was divided into three zones: the anterior 30% containing the genu, a middle 40% containing the body and the posterior 30% containing the splenium (Figure 2F).

Evenly spaced EM photomicrographs (16,500× magnification) were taken of the anterior, middle and posterior zones of each CC (9 photomicrographs on each zone; Figure 2F, dots). Unmyelinated and myelinated axon data, such as density, inner diameter and myelin thickness were obtained using the Cellgraph system (Microptic S.L., Barcelona, Spain). Unmyelinated and myelinated axon number in anterior, middle and posterior zones was calculated from mean axon density (measured in EM micrographs) and the corresponding mid-parasagittal area (measured in semithin sections). The outer diameter of myelinated axons was calculated as the sum of the axon inner diameter and 2 times its myelin thickness (see details in Berbel et al., 1994; Lucia et al., 2018). For myelinated axons, the g-ratio, and the conduction velocity (5.5 times the outer diameter) were calculated accordingly (Sanders and Whitteridge, 1946; Waxman and Bennett, 1972; Caminiti et al., 2013; Innocenti, 2017).

The MRI data acquisition parameters and images analyzed here are the same as those obtained for our previous study (Lucia et al., 2018). However in this study specific acquisition planes and region of interest (ROI) have been redefined, and fully described below. In summary, for MRI data acquisition, isoflurane anesthetized rats were scanned using a custom-made MRI compatible holder, positioned on the magnet isocenter, and monitored using a MRI compatible control unit (MultiSens Signal conditioner, OpSens, Quebec, Canada). Rats were maintained alive throughout the study period, with scans being taken periodically from P8 onward (Figure 1). Scans were obtained with a horizontal 7 Tesla scanner with a 30?cm diameter bore (Biospec 70/30v; Bruker Medical, Ettlingen, Germany), equipped with a 675 mT/m actively shielded gradient coil (Bruker Medical; BGA 12-S) of 11.4?cm inner diameter. A ¹H rat brain receive-only phase array coil with integrated combiner and preamplifier (no tune/no match) was used, in combination with an actively detuned transmit-only resonator (Bruker BioSpin MRI) and Paravision software (Bruker Medical).

The coronal and sagittal planes of study were determined from preliminary T2-weighted (T2w) images acquired in the three orthogonal planes using rapid acquisition relaxation enhanced sequence (RARE) with the following parameters: RARE factor 8, 15 slices, slice thickness 1 mm, field of view (FOV) 40 mm × 40 mm, matrix 256 × 256, effective echo time (TE_eff) 56 ms, repetition time (TR) 2,000 ms, 1 average for 1 min 4 s total acquisition time (Hennig and Friedburg, 1988; Perez-Cervera et al., 2018). Using these anatomical images, final coronal MRI images were acquired using RARE sequence with the following parameters: RARE factor 8, 25 slices, slice thickness 0.5 mm, FOV 20 mm × 20 mm, matrix 200 × 200 (voxel size 100 μm × 100 μm × 500 μm), TE_eff 56 ms, TR 3,728 ms, 4 averages for 12 min 26 s total acquisition time (Lucia et al., 2018).

MRI images were obtained from 8 rats per group (2 rats per litter) at ages from P8 to P365, with the rats of each group individually scanned at the ages indicated in Figure 1 and images analyzed using the ImageJ software (National Institute of Health, Bethesda, MD, United States). The ROI included the central portion at ± 0.7 mm from the midline in the anterior (located in the genu), middle (located in the body) and posterior (located in the splenium) zones of the CC (Figures 2A–D, vertical white lines in Figures 2B–D). T₂r was calculated as the ratio between the signal intensity of T2w in the ROI to that of T2w in a ROI of the lateral ventricle cerebrospinal fluid. T₂r was obtained in order to compare T2w signals between callosal zone, age and experimental groups. During postnatal development, T2w values of the lateral ventricle were similar between MMI and C rats (mean T2w values were 12,565 ± 1242 in transient and 12,161 ± 1146 in chronic hypothyroid, and 12,327 ± 961 in C rats; all T2w values ranging from 10,431 to 14,518).

Due to brain shrinkage found in MMI rats, anterior, middle and posterior anatomical landmarks were used to identify MMI and C sections located at equivalent anterio-posterior planes. In MMI and C rats at P150, the anterior landmark was the point where the optic nerves begin to form the optic chiasm (at 0.9 mm from Bregma in C rats; Paxinos et al., 2015); the middle landmark was the point where the rostral hippocampus begins to appear (at -1.6 mm from Bregma in C rats); and the posterior landmark was the caudal end-point of the hippocampal commissure (at -4.5 mm from Bregma in C rats; Figures 2B–D). At these levels, the motor, somatosensory, auditory and visual cortices were present (Supplementary Figure S1). In MRI scans at P150, contralateral distances between four selected homotopic contralateral cortical areas projecting through the CC were measured in four rats per group, specifically the projections of anterior (motor), middle (somatosensory), and posterior (auditory and visual) cortices. Contralateral distances were measured between mid-zones of pial surface in the selected neocortical areas (Supplementary Figure S1, white dots). As in the anterior commissure (Lucia et al., 2018), these measurements have a potential antero-posterior error of about 250 μm per hemisphere, given the MRI scan thickness of 500 μm. An estimation of myelinated axon conduction delay between selected homotopic areas was calculated from conduction velocity and mean contralateral distance between homotopic areas (Supplementary Figure S3 and Supplementary Table S6).

For statistical analysis (see details in Lucia et al., 2018), we used SYSTAT software (Systat Software, Inc., Chicago, IL, United States). Two-way ANOVA followed by either Tukey’s (equal variances) or Games-Howell’s (unequal variances) tests to identify significant differences (P ≤ 0.05) between means among age and experimental groups were used for mean frequency distribution analysis of MRI and EM data. One-way ANOVA followed by either Tukey’s test or the Student-Newman-Keuls method was used for the analysis of plasma concentration of thyroid hormones. The coefficient of determination (R²) calculated from non-linear and linear regression functions, provides a measure of how well estimated values correspond with observed values.

In transient hypothyroid and C rats, body weight rapidly increased to 400 ± 20 g by P150 and then more slowly to 446 ± 13 g in C rats at P365. Weight was significantly lower (P < 0.001) in chronic hypothyroid rats with an average value of 75 ± 25 gr at P150 (Figure 3A).

At P15, plasma concentration levels of total T4 (tT4) and total T3 (tT3) in MMI rats had an average value of 7.2 and 0.05 ng/ml respectively, and were significantly lower (P < 0.001) than the control averages of 40.1 ng tT4/ml and 0.26 ng tT3/ml. Normal levels of tT4 and tT3 were reached at P21 in MMI_P0–21 + T4_P15–21 rats and at P50 in MMI_P0–21 rats. Levels of tT4 and tT3 in P50 chronic hypothyroid rats were significantly lower (P < 0.001) than in transient hypothyroid and C rats of the same age (Figures 3B,C).

As already mentioned, these data correspond to rats used in a previous study (Lucia et al., 2018).

At early ages, anterior, middle and posterior CC appeared lighter and hardly distinguishable from the adjacent neuropil in T2w images from both MMI and C rats. Anterior CC showed increased border definition in C rats at P30, and in middle and posterior CC at P40 (arrows in Figures 4A, 5A, 6A). Anterior and middle CC remained hardly distinguishable in MMI_P0 at P40 and in MMI_E10 rats at P60 and P150 (see arrowheads in Figures 4A, 5A), as did posterior CC, in transient hypothyroid rats at P60 and in chronic hypothyroid rats at P150 (arrowheads in Figure 6A and Supplementary Tables S1–S3).

In C rats, T₂r decreased rapidly from P8 (on average, T₂r = 0.99 ± 0.21) to P40 in anterior (T₂r = 0.35 ± 0.01), to P60 in middle (0.33 ± 0.01), and to P75 in posterior CC (0.34 ± 0.02). This was followed by a more slow decline, reaching T₂r = 0.25 ± 0.01, 0.26 ± 0.01 and 0.27 ± 0.03, in anterior, middle and posterior CC respectively at P365. T₂r in MMI rats also decreased rapidly from P8 to P40 but remained higher than controls (Figures 4B, 5B, 6B, Supplementary Figure S2, and Supplementary Tables S1–S3). Anterior CC at P40 and P75 showed significant T₂r differences between MMI and C rats (P < 0.001). At P150, differences decreased (P < 0.05) between transient hypothyroid and C rats, but still remained higher (P < 0.001) in chronic hypothyroid (Figure 4C and Supplementary Table S1). In middle CC, significant T₂r differences were found between MMI and C rats at P40 (P < 0.001). At P75, no differences were found between MMI_P0–21 + T4_P15–21 and C rats; and a significant difference (P < 0.05) was found between MMI_P0–21 and C rats. At P150, differences remained (P < 0.001) between chronic hypothyroid rats with respect to the other groups; and a significant difference (P < 0.05) was found between MMI_P0–21 and MMI_P0–21 rats (Figure 5C and Supplementary Table S2). In posterior CC, significant T₂r differences (P < 0.01) were found between MMI and C rats at all ages (Figure 6C and Supplementary Table S3), and remained higher in transient and chronic hypothyroid rats in anterior and posterior CC (Supplementary Figure S2).

The ultrastructure of commissural axons and glial processes in anterior, middle and posterior CC at P150 observed in this study (Figure 7) was similar to that previously described in C (Gravel et al., 1990; Berbel et al., 1994) and chronic hypothyroid MMI-treated rats (Berbel et al., 1994). The decreased myelinated axon density in anterior, middle and posterior CC of chronic hypothyroid (Figures 7J,O) is noteworthy compared to C (Figures 7A,C) rats. The CC mid-sagittal area was 2,539,483 ± 479,729 μm² in C rats; it significantly decreased to 1,933,471 ± 195,946 μm², on average, in transient hypothyroid (P < 0.01), and to 957,134 ± 10,480 μm², on average, in chronic hypothyroid rats (P < 0.001), representing a 23.9 and 62.3% reduction, respectively compared to C rats (Supplementary Table S4).

In agreement with previous data (Berbel et al., 1994), the total axon number in the CC was similar in C and MMI rats at P150, ranging from 12,706,226 axons in MMI_P0 to 14,913,049 axons in MMI_E10 rats (13,362,275 in C rats) (Figure 8A and Supplementary Table S4). Compared to C rats, unmyelinated CC axon number significantly increased in transient hypothyroid and MMI_P0 (P < 0.05), and in MMI_E10 (P < 0.001) rats (Figure 8B), while myelinated axon number decreased in transient (P < 0.05) and in chronic (P < 0.001) hypothyroid rats (Figure 8C). In anterior and middle CC, unmyelinated axon number remained higher in MMI_E10 (P < 0.001; Figures 8D,E) and in posterior CC, in MMI rats (P < 0.001; Figure 8F than in other groups. In anterior, middle and posterior CC, myelinated axon number decreased significantly (P < 0.001) in transient and chronic hypothyroid rats compared to C rats (Figures 8G–I and Supplementary Table S4). In transient hypothyroid rats, the average decrease of the myelinated axon was 29.7% in anterior and middle CC and 67.7% in posterior CC compared to C rats. In chronic hypothyroid rats, it was 82.5% in the anterior and middle CC and 96.0% in the posterior CC compared to C rats (Figures 8J–L and Supplementary Table S4).

The biological significance of MRI in the development of the CC was explored by generating the corresponding regression function between T₂r and quantitative EM data. Using the EM data from the CC in C rats at P150 and at different postnatal ages (Berbel et al., 1994), we found a significant fit between T₂r and myelinated axon number in the anterior (R² = 0.993; Figures 9A,B), middle (R² = 0.858; Figures 9C,D) and posterior CC (R² = 0.954; Figures 9E,F). No fit was found between T₂r and other quantitative EM data such as myelin thickness (R² = 0.582). The regression functions between T₂r and myelinated axon number in the three regions of the CC were validated by comparing the estimated myelinated axon number for MMI_E10 rats with values obtained from already published EM data (MMI_E10 group in Berbel et al., 1994). No significant differences were found between estimated MMI_E10 values and those published (Figures 10A,C,E). Regression functions were used to compare the estimated myelinated axon number for C, transient and chronic hypothyroid rats at different postnatal ages, and P150 values were compared with the EM values. Estimated and EM values were similar, except for MMI_P0 anterior (233,500 axons vs. 378,900 estimated axons) and middle CC (300,300 axons vs. 528,100 estimated axons; Figures 10B,D,F and Supplementary Table S4).

Mean unmyelinated axon diameter decreased in posterior CC of all groups, except for MMI_E10 rats where posterior was similar to anterior, middle CC. The highest unmyelinated axon diameters were observed in anterior CC of C and transient hypothyroid rats (on average, 0.24 ± 0.07 μm) with MMI_E10 diameters being significantly lower (0.16 ± 0.04 μm; P < 0.001; Figures 11A,D,G,J,M). In middle and posterior CC, unmyelinated axon diameter was similar between groups Figures 12A,D,G,J,M, 13A,D,G,J,M and Supplementary Table S5).

In anterior and middle CC, myelinated axon inner diameter decreased (P < 0.001) in chronic hypothyroid rats (0.49 ± 0.14 and 0.47 ± 0.13 μm, respectively) compared to transient and C rats (on average, 0.56 ± 0.21 μm) (Figures 11A,D,G,J,M, 12A,D,G,J,M and Supplementary Table S5). In posterior CC, myelinated axon inner diameter decreased (P < 0.05) in transient and chronic (on average, 0.47 ± 0.13 μm) hypothyroid compared to C (0.50 ± 0.16 μm) rats (Figures 13A,D,G,J,M and Supplementary Table S5). Myelinated axons with an inner diameter ≥ 1.0 μm made up 8.6% in anterior, 7.9% in middle and 2.7% in posterior CC of C rats. A decreased inner diameter was found respectively in transient (3.8, 4.4 and 0.8%) and chronic hypothyroid (1.1, 0.3, and 0.0%) rats (Figures 11A,D,G,J,M, 12A,D,G,J,M, 13A,D,G,J,M).

Anterior CC myelin thickness (on average, 0.082 ± 0.022 μm) was similar throughout all groups (Figures 11B,E,H,K,N and Supplementary Table S5). Middle CC was significantly decreased (P < 0.01) in transient and chronic hypothyroid rats (on average, 0.078 ± 0.021 μm) compared to controls (0.085 ± 0.027 μm; Figures 12B,E,H,K,N and Supplementary Table S5). Posterior CC average myelin thickness was significantly lower (P < 0.01) in transient hypothyroid rats (0.075 ± 0.019 μm) compared to other groups (on average, 0.085 ± 0.022 μm; Figures 13B,E,H,K,N and Supplementary Table S5). No significant fit was found between myelin thickness and axon inner diameter in anterior, middle and posterior CC (R² ranging from 0.061 to 0.199). Notwithstanding, the slope of the anterior, middle and posterior CC regression function was higher in C rats (3.9°, 4.1°, and 3.8°, respectively) than in transient (on average, 2.1°, 2.6°, and 2.4°) and chronic hypothyroid rats (on average, 1.6°, 2.8°, and 3.5°; Figures 11C,F,I,L,O, 12C,F,I,L,O, 13C,F,I,L,O).

In anterior, middle and posterior CC, the average g-ratio (0.74 ± 0.06, 0.73 ± 0.06, and 0.73 ± 0.06, respectively) decreased significantly (P < 0.01) in chronic hypothyroid rats compared to transient hypothyroid and C rats (on average, 0.76 ± 0.06 in anterior, 0.77 ± 0.06 in middle and 0.75 ± 0.06 in posterior CC) (Figure 14 and Supplementary Table S5). The average slope of the regression functions for anterior, middle and posterior CC was lower in C (9.5°, 9.7°, and 12.7°) than in transient (13.3°, 11.3°, and 16.2°) and chronic (18.2°, 16.3°, and 16.0°) rats (Figure 14).

In anterior CC, the conduction velocity of myelinated axons decreased in transient (3.89 m/s; P < 0.05) and chronic (3.60 m/s; P < 0.001) hypothyroid rats compared to C rats (4.09 m/s; Supplementary Table S5). In middle CC, conduction velocity decreased in chronic hypothyroid rats (3.47 m/s) compared to the other groups (on average, 3.96 m/s; P < 0.001), while in posterior CC, it decreased in transient and chronic hypothyroid (3.47 m/s; P < 0.01) rats compared to C (3.70 m/s; Figure 15 and Supplementary Table S5). The average distance estimated between primary motor and sensory cortical areas decreased significantly in MMI rats (Supplementary Figures S1, S6 and Supplementary Table S6), commensurate with brain shrinkage (Figures 4A, 5A, 6A and Supplementary Figure S1). In summary, the estimated conduction delay between the homotopic cortical zones decreased in transient (range 5.3–16.8%, excluding auditory zones) and chronic (range 5.7–20.6%) hypothyroid rats when compared to controls (Supplementary Figure S6 and Supplementary Table S6).

The present study follows on from our previous study of the anterior commissure (Lucia et al., 2018), and has similar strengths and limitations. Its main strength is the combined use of non-invasive in vivo MRI analysis of the CC with an ex vivo EM qualitative and quantitative study of the number and diameter of unmyelinated and myelinated axons. Myelinated axon percentage could be estimate as a function of T₂r values and was able to confirm altered CC postnatal development in late T4-treated transient hypothyroid rats.

The quantitative EM measures are based on average axon data from 36 samples on each CC region (4 rats and 9 sampling areas per rat). The values obtained for MMI_E10 and C rats at P150 were similar to those obtained in previous studies (Gravel et al., 1990; Berbel et al., 1994). Although the three antero-posterior regions aim to define callosal axons from motor, somatosensory, auditory and visual areas, different subpopulations of unmyelinated and myelinated axons are being averaged, and this could mask changes occurring in one or the other of these subpopulations. Further studies using electrophysiology, behavior and functional imaging, among other methodologies, will be needed to study identified subpopulations of callosal neurons.

The management of thyroid hormone concentrations administered to MMI pups from P15 to P21 is critical in the developing cerebral cortex because it is related to the amount of active hormone that target cells require for correct gene expression. Several events play a role such as: (i) postnatal angiogenesis which increases the surface for the transfer of thyroid hormones, (ii) the expression of thyroid hormone transporters carrying free thyroid hormones from serum to target neural cells, and (iii) the proliferation of type 2 deiodinase expressing astrocytes that mediate the deiodination of T4 to T3. Thyroid hormones levels in the cerebral cortex of early postnatal hypothyroid rats remain unknown. Despite the possible differences between males and females, the L-T4 dose administered in this study was adjusted from previously published data on 6 groups of young female rats thyroidectomized when weighing 120–150 g (approximate age P50–75) and 30 days later, subcutaneously infused with 6 different doses of L-T4 (using ALZET^® osmotic minipumps) spaced over 12–13 days with tT4 and tT3 being measured in plasma and different tissues, including the cerebral cortex (Escobar-Morreale et al., 1995). For our study, circulating tT4 and tT3 levels were measured at P15 and P21 (onset and end of L-T4 treatment). At P15, MMI_P0–21 + T4_P15–21 rats were hypothyroid while at P21, tT4 and tT3 levels in plasma were similar to C rats. Possible age and gender differences affecting thyroid hormone concentrations during development of the cerebral cortex of rats should be addressed in further studies. The adjustment of doses and their timing for T4 treatment of neonates diagnosed with congenital hypothyroid must be derived from the very important studies that correlate a child’s neurocognitive outcome with T4 treatments (Rovet, 2005), given that the levels of thyroid hormones in the cerebral cortex cannot be obtained.

There is a common consensus that water trapped in myelin and the reduction of the extra-cellular space are the main factors decreasing T₂r values, which reflect the degree of maturation of the commissures in mammals (Le Bihan, 2014; Jelescu et al., 2016; MacKay and Laule, 2016). Supporting this, T₂r was found related with the percentage and number of myelinated axons and axon density in the AC (Lucia et al., 2018), both of which contribute to the reduction of extra-cellular space. The tight association between T₂r values and number of myelinated axons is clearly demonstrated in this work, indicating its usefulness in assessing myelination in children, by itself or in conjunction with myelin water fraction or magnetization transfer.

In the CC of C and MMI rats, T₂r rapidly decreased from P8 to P20, and at a slower pace from P20 to P100 after which it leveled off, showing that at some point during cortical maturation, T₂r becomes insensitive to white matter and extra-cellular changes. In fact, myelinated axon number continues to increase after P100 (more significantly in C rats, Figures 10B,D,F) in parallel with an increased number of glial cells (Gravel et al., 1990; Berbel et al., 1994). This is noteworthy since MRI techniques currently available may not be sensitive to changes in myelinated axon number in later periods of cortical maturation (especially to those of small diameter showing reduced myelin thickness), thereby limiting their early detection of white matter associated diseases. T₂r images of children aged 6–48 months show a T₂r sensitivity inflection point from 20 months onward, missing the myelination progression in subcortical white matter of the parietal and occipital lobe (Liu et al., 2018). In MRI scans, demyelination of callosal commissural and subcortical white matter axons in patients suffering toxic leucoencephalopathy only became detectable when the disease was at an advanced stage (Phan-Ba et al., 2012). In the case of deficient myelination, often seen using MRI in some young patients with monocarboxylate transporter 8 deficiency (Namba et al., 2008; Gika et al., 2010), current MRI techniques may not be able to follow the progression of demyelinating diseases, that become evident beyond 5–6 years of age (Vaurs-Barrière et al., 2009; Gika et al., 2010). As already discussed in the case of the anterior commissure (Lucia et al., 2018), this raises the important issue in MRI analysis of differentiation between delayed myelination and hypomyelination (van der Knaap and Wolf, 2010) in infants and children suffering diseases affecting white matter development. Enhanced approaches such as MRI through Mn²⁺ injection (MEMRI; Chan et al., 2012; Deng et al., 2019) and diffusion kurtosis image (DKI; Cheung et al., 2009) which is highly sensitive and directionally specific, should add valuable information to the study of developmental ultrastructural changes in white matter and the plasticity of callosal connections in normal and hypothyroid rats.

In MMI and C rats, the lowest T₂r values are found in anterior CC and the highest in posterior CC, reflecting the antero-posterior gradient of maturation in the CC. Anterior (motor) and middle (somatosensory) callosal axons grow equally in diameter and myelinate earlier than posterior (auditory and visual) axons (Supplementary Figure 3 and Supplementary Table S5). In humans and monkeys, myelination begins in the body of the CC, then the splenium and continues in the genu (LaMantia and Rakic, 1990; Luders et al., 2010). In monkeys, this gradient parallels the growth in diameter of callosal axons, which in turn depends on the speed at which contralateral neocortical areas communicate (Caminiti et al., 2009; Innocenti, 2011). These data suggest a similar gradient in rats, humans and monkeys, given that the frontal telencephalic region in rats is principally a motor area (Krubitzer et al., 2011) with commissural neurons sending their axons through the anterior CC (Isseroff et al., 1984). Although the antero-posterior gradient is maintained in MMI rats, the maturation of the CC is delayed in MMI in comparison with C rats with the posterior zone being the most affected region (Supplementary Figure 2 and Supplementary Table S5). The maturation delay of the posterior CC can be explained by the severe alterations observed in visual and auditory neurons observed in hypothyroid rats (Ruiz-Marcos et al., 1979; Uziel et al., 1985). Our data strongly support the MRI data from children with early postnatal thyroid hormone insufficiency that show cognitive alterations involving language, motor, auditory and visual functions affecting telencephalic commissural connections (Rovet et al., 1992; Clairman et al., 2015).

Although an in-depth review on how development and evolution can drive the transfer of information within and between cerebral hemispheres in mammals is beyond the scope this study, we would point out that callosal connections are fundamental to the understanding of how emergent specialized cortical areas in more evolved mammals are connected (Kaas, 2013). The number and size of connected areas, the spread of information within these areas and the conduction delay to synchronize signals are all fundamental factors (Innocenti, 2017). These factors in turn depend on axon length and intermodal distances (Rushton, 1951; Waxman and Bennett, 1972; Waxman, 1975), size of terminal arbors and axon boutons, and the tangential and radial distribution of callosal neurons (Innocenti, 2011, 2017; Innocenti and Caminiti, 2017). Additional data suggests that the size (and conduction velocity) of myelinated callosal axons decreases as hemispheric specialization and asymmetry increases (Caminiti et al., 2009).

Significant differences in number and diameter of unmyelinated and myelinated axons were found between C and transient and chronic hypothyroid rats, which shows that the transfer of information between cerebral hemispheres in MMI rats is abnormal. Factors such as abnormal radial and tangential neuron location, terminal arbor size and bouton number of callosal axons are significantly affected in the cerebral cortex of transient and chronic hypothyroid rats (Berbel et al., 2014).

Although g-ratio was similar between MMI and C rats and to values obtained in other species (Waxman and Swadlow, 1976; Stikov et al., 2015), the distance between homotopic callosal areas decreased in MMI rats, with a corresponding decrease in conduction delays estimated from these distances and the outer diameter of myelinated axons. The size of terminal arbors and number of boutons in callosal axons of MMI rats might be also affected. The formation of microcircuits in the cerebral cortex is dependent on the activation of the sonic hedgehog signaling pathway (Harwell et al., 2012), which is regulated by T3 (Gil-Ibañez et al., 2017). In fact, the number and density of excitatory and inhibitory boutons in the somatosensory cortex of early postnatal hypothyroid rats (Navarro et al., 2015), and the number and length of terminal branches of thalamic axons, and the number of boutons in these branches were significantly reduced in chronic hypothyroid rats (Ausó et al., 2001) were observed.

Unmyelinated axon inner diameter in the anterior and posterior CC was less in chronic hypothyroid compared to transient hypothyroid and C rats (Supplementary Table S5). However, myelinated axon inner diameter in the anterior and middle CC was smaller in chronic hypothyroid compared to transient and C rats, while in the posterior CC, it was smaller in chronic and transient hypothyroid rats compared to controls. Furthermore, the estimated conduction velocity decreased more in the posterior CC of MMI rats than in middle and anterior CC, compared to C rats. These data strongly suggest a delay in CC transfer of information in chronic and transient hypothyroid rats, being more pronounced in the posterior CC.

The estimated conduction delay of myelinated axons between homotopic callosal neurons based on morphological characteristics is speculative because it is not based on electrophysiological recordings. Conduction velocity is estimated on precise EM data from myelinated axons, but other data such as: (i) type of myelinated axon (e.g., motor, somatosensorial, visual, and auditory callosal axons) and (ii) the individual length of myelinated axons, could not be fully ascertained.

Electrophysiological recordings showing properties of callosal axons from living cells are unfortunately not available in early postnatal transient hypothyroid rats. Our estimations are based on the methodology published in pioneering work (Rushton, 1951; Waxman and Bennett, 1972; Waxman, 1975) and more recent electrophysiological and anatomical combined studies (Roxin et al., 2005; Caminiti et al., 2009, 2013). We provide data that may help understand the role of thyroid hormones in the transfer of information between cortical hemispheres of rats at earlier postnatal ages.

The conduction delay of a myelinated axon reflects the time needed for an action potential to reach its target. Conduction delay is calculated as axon length divided by conduction velocity (= 5.5 × myelinated axon outer diameter). Velocity and length have opposite effects: if velocity (axon diameter) is greater then delay is less, conversely if length is greater then delay is greater. Our data suggests that in MMI rats the distances between hemispheres are more affected (they have smaller brains) than the myelinated axon conduction velocity. On average, the distance between cortical areas decreased 12.1 ± 5.0% in transient and 22.1 ± 6.6% in chronic hypothyroid rats, while the conduction velocity decreased 4.7 ± 2.5% in transient and 11.5 ± 2.9% in chronic hypothyroid rats with respect to controls (Supplementary Table S6).

The decreased number of myelinated axons in the CC of MMI rats will affect motor, somatosensorial, auditory and visual inputs, jeopardizing the spatial and social learning of MMI rats and consequently, its behavior.

As discussed in our previous study (Navarro et al., 2015), specific tests have been undertaken to determine the effects of cerebral cortex alterations in hypothyroid rats. These tests studied audiogenic seizure susceptibility and motor excitability (Van Middlesworth and Norris, 1980; Ng et al., 2001; Ausó et al., 2004; Wallis et al., 2008). Hearing loss is a common neurologic disease found in humans with different types of hypothyroidism such as cretinism (Trotter, 1960; DeLong et al., 1985; Morreale de Escobar et al., 2000), congenital hypothyroidism (Vanderschueren-Lodeweyckx et al., 1983; Rovet et al., 1996) and resistance to thyroid hormone also known as Refetoff syndrome (Refetoff et al., 1967). Deafness caused by hypothyroidism is often due to irreversible damage of the Corti’s organ (Knipper et al., 2000) and to abnormal development of the brain stem (Halpern et al., 1989; Ma et al., 1989). Impaired hearing (Navarro et al., 2015) and the abnormal radial distribution of callosal auditory neurons (Berbel et al., 1993) have been found in chronic hypothyroid rats. The delay observed in the maturation of the posterior CC in transient and chronic hypothyroid rats in the present work is an additional factor to be considered for understanding the physiopathology of hearing loss in late treated children with congenital hypothyroidism.

The study of learning, attention and memory deficits revealed alterations in the organization of the cerebral cortex, including callosal connections (Negishi et al., 2005; Venero et al., 2005; Gilbert et al., 2007; Opazo et al., 2008; Gilbert and Lasley, 2013). In mammals, prepulse inhibition of the acoustic startle response measures sensorimotor gating mechanisms (Graham, 1975; Blumenthal et al., 1996; Koch, 1999). Deficient sensorimotor gating, demonstrated by disrupted prepulse inhibition, exists in several neuropsychiatric disorders such as schizophrenia (Braff et al., 2001; Hamm et al., 2001). In hypothyroid rats from P0 to P40, reduced prepulse inhibition of the acoustic startle response was found to be stimulus intensity dependent (Navarro et al., 2015), and is associated with decreased electrophysiological activity of the somatosensory cortex (Graham, 1975; Blumenthal et al., 1996; Koch, 1999), which might reflect a degree of attention deficit, including impaired spatial learning (Wilcoxon et al., 2007). This is in agreement with elevated open arms test data (Navarro et al., 2015) where MMI pups typically walked to the end of the open arm and then fell, suggesting an orientation and memory loss concerning dangerous situations in which the somatosensory cortex could be involved. This possibility is supported by data showing abnormal cytoarchitecture and connectivity in the primary somatosensory cortex of MMI rats (Berbel et al., 2001; Ausó et al., 2001) and the impaired callosal connections identified in this paper.

The use of animal models to further our understanding of the physiopathology of neurocognitive and psychiatric diseases has its limitations. Most of these diseases are caused by alterations of cortical areas not found in lower vertebrates (e.g., the frontal and prefrontal cortices). The frontal and prefrontal areas of the rat are greatly reduced (as further discussed below) and consequently, the genu contains mostly motor and some somatosensory axons, instead of the prefrontal and frontal callosal axons that are found in the human’s genu. The abnormal development of the neocortex is associated with childhood and adolescent neurocognitive (Goodman et al., 2014; Wheeler et al., 2015) and psychiatric disorders such as autism spectrum disorders (ASD; Bauman and Kemper, 2005; Román et al., 2013), attention deficit-hyperactivity disorder (ADHD; Li Y. et al., 2014; Li Z. et al., 2014) and schizophrenia (Innocenti et al., 2003; Santos et al., 2012). In particular, alterations in cerebral white matter have been described in ASD (Piven et al., 1997; Alexander et al., 2007), ADHD (Hynd et al., 1991), bipolar disorder and schizophrenia (McIntosh et al., 2009; Santos et al., 2012). In all these disorders, events such as abnormal cell proliferation and migration, deficient axonal branching and growth, myelination and oligodendrocyte development, unbalanced excitatory/inhibitory input ratios and abnormal firing synchronization may occur.

Although there are important differences in the development of the cerebral cortex between humans and rats, similarities can be found in the basic events of corticogenesis such as axonal sprouting and pruning, myelination and synaptogenesis, as well as cortical functions that are driven by evolutionary preserved T3-regulated genes (Morte et al., 2010; Berbel et al., 2014; Chatonnet et al., 2015; Bernal, 2017). To this extent we can consider that the transiently hypothyroid rats in this study mimic the condition of late diagnosed congenital hypothyroidism in children, resulting in impaired cognitive development (O’Callaghan et al., 1995; Kester et al., 2004; Rovet and Simic, 2008; Williams and Hume, 2008; Willoughby et al., 2014).

In cultures of cortical cells derived from E14 mice several hundred genes transcriptionally regulated by T3 have been described as being involved in key events of cerebral cortex development (Morte et al., 2010; Berbel et al., 2014; Chatonnet et al., 2015; Gil-Ibañez et al., 2017; Bernal, 2017). Some of these genes have been found mutated in humans diagnosed with neurological and mental diseases (Berbel et al., 2014; Gil-Ibañez et al., 2017; Bernal, 2017). Cortical white matter abnormalities have been described in bipolar disorders and schizophrenia, and have been associated with an abnormal expression of neuregulin-1, which plays a crucial role in oligodendrocyte development and function (Fernandez et al., 2000; McIntosh et al., 2009; Uranova et al., 2018). Genes involved in the expression of neuron-glia adhesion proteins affecting myelination are under-expressed in schizophrenia and bipolar disorder (Tkachev et al., 2003) as well as in postnatal hypothyroid rats (Rodríguez-Peña et al., 1993; Rodríguez-Peña, 1999; Barradas et al., 2001; Sharlin et al., 2008). Furthermore, Erk1/2 expression is decreased in late fetal hypothyroid rats (Berbel et al., 2010). The Erk1/2 pathway activates autotaxin, which drives oligodendrocyte maturation (Liu et al., 2011; Lee and Petratos, 2016). All these data suggest that the decreased number and percentage of myelinated axons in transient and chronic hypothyroid rats might result not only from arrested axon growth, resulting in small diameter callosal axons, but also that congenital hypothyroidism might also affect the proliferation and maturation of oligodendrocyte precursors, albeit transiently. This would explain at least in part, the physiopathology of myelinated tracts in the cerebral cortex of late diagnosed children with congenital hypothyroidism, and the relevance of genes involved in the myelination of cortical white matter axons. Any factor altering the expression of T3-regulated genes associated with the development of telencephalic commissures, such as low T3 availability to target cells, must be considered as potentially increasing the risk of children developing neurologic and psychiatric diseases.