De-identified human fetal brain specimens (deemed non-pathological) of 15, 18, 21, 24 and 27 GW, respectively, were procured from the Departments of Pathology, Mediscan Systems Pvt. Ltd. Chennai, India, and Saveetha Medical College and Hospital (SMCH) Chennai, India. Autopsies with brain extraction were performed after due consent wa obtained from the next of kin. All procedures used in this study have been performed in accordance with Declaration of Helsinki and were approved by the Institute Ethics Committee (IEC) and Institute’s Review Board (IRB) of the Indian Institute of Technology, Madras (IITM) (IEC/2021-01/MS/06) and the respective medical centers mentioned above.

The specimens were obtained within a post-mortem interval (PMI) of 2–4 h, and as a part of larger study at Sudha Gopalakrishnan Brain Center (SGBC), IITM, postmortem in-skull MRIs (1.5 and 3 T) were obtained prior to the extraction (Verma et al., 2024). The brains were extracted after 2 weeks in fixative and stored in room temperature in 4% paraformaldehyde (PFA) in 0.01 M phosphate buffer (PB) until ready for histological processing.

All the fixed whole brain specimens were cryoprotected using gradient sucrose solution, starting with 10% sucrose in 4% PFA, followed by 20 and 30% sucrose in 0.01 M PB, with a common end point of specimens attaining complete equilibrium in the solution (i.e., the brain sample submerges to the bottom of the container).

Following cryoprotection, the whole brain was frozen into customized molds previously described (Kumarasami et al., 2023), which not only enables alignment of the frozen block but also minimizes tissue distortions. Briefly, the whole brain was frozen in −80°C in an isopentane bath surrounded by a dry ice chamber. The frozen blocks were stored in −80°C refrigerator until sectioning (Pinskiy et al., 2013).

The brains were then cryosectioned using a Leica 1,520 cryostat (Leica Biosystems, India) for the smaller brains and the Leica 3600XP cryomacrotome (Leica Biosystems, India) for the larger brains. Tape transfer technique has been used for sectioning at 20 μm thickness in the sagittal and coronal planes (Pinskiy et al., 2015) and the sections were stained for three different and consecutive series, Nissl, Hematoxylin and Eosin (H&E) and Immunohistochemistry (IHC).

The first series of sections were stained in 0.2% Thionine solution (in water) (Thionine Acetate, SRL Chemicals), to label the Nissl bodies, then differentiated using distilled water, dehydrated using graded alcohol solutions and cleared with xylene prior to coverslipping using DPX (Merck, India). The second series of sections was stained with H&E using Harris’ regressive protocol. Since the fetal brains have excessive water content, all sections were dried in a temperature-controlled chamber at 33°C, prior to coverslipping.

The third series underwent immunostaining for specific antibodies with protocol specifically modified for use in the developing brain (Pandurangan et al., 2024). Briefly, the sections underwent an antigen retrieval process using boric acid buffer (pH = 9.0) at 70°C for 20 min, after which the sections were allowed to cool down at room temperature (RT). After cooling, the slides were rinsed with 0.01 M Phosphate Buffer (PB) three times for 5 min each and treated with 0.3% H₂O₂ solution for 20 min, to quench the endogenous peroxidase. After the peroxidase blocking, the slides were again washed three times with 0.01 M PB and a protein block was performed with animal free blocking reagent [5X Cell Signaling Technology (CST) for 1.5 h at RT. Following the protein block, the slides were incubated with IgG anti-GFAP (E4L7M) (Cat. No. 80788) XP monoclonal antibody 1:1500] or CR [Rabbit, IgG anti-CR (E7R6O) (Cat. No. 92635) XP monoclonal antibody 1:1000], for overnight and 48 h, respectively. After incubation the slides were washed 4 times with 0.01 M PB, following which the slides were incubated with Goat, anti-rabbit, IgG, polyclonal antibody (HRP conjugate, ENZO, Cat. No. ADI-SAB-300-J) at 1:500 ratio for 1.5 h at RT. Sections were rinsed thrice with 0.01 M PB and stained with a substrate containing diaminobenzidine (SignalStain DAB Substrate Kit, CST). The slides underwent dehydration with graded alcohol before being coverslipped.

For CR immunostaining in 15–24 GW, we used the DAKO EnVision secondary kit for immunohistochemistry. The sections first underwent antigen retrieval process using EnVision FLEX Target retrieval solution, High pH, for 20 min at 70°C and then cooled to RT. After cooling, the slides were rinsed in EnVision FLEX wash buffer and treated with EnVision FLEX peroxidase-blocking reagent for 10 min, followed by another rinse with the wash buffer. Subsequently, the slides were incubated for 24 h with the anti-CR primary monoclonal antibody [Rabbit, IgG anti-CR (E7R6O) (Cat. No. 92635) XP monoclonal antibody, 1:1000]. Post-incubation, the slides were washed with the buffer and incubated with EnVision FLEX/HRP for 30 min. After this step, they were washed again and stained with DAB solution (EnVision DAB + chromogen). Finally, the slides were dehydrated using graded alcohol and coverslipped.

The stained sections were scanned and digitized using a large format scanner (TissueScope LE120, Huron Digital Pathology, Canada), with an in-plane resolution of 0.5 μm/pixel. These images underwent quality control tests and only those that passed the Quality Check (QC) process were considered for further analysis.

ICs of both hemispheres were included in our analysis. We identified the IC in 100 sections in the 15 GW specimen, 150 sections in the 18 GW and 21 GW specimens, and in over 180 sections in the 24 GW and 27 GW specimens (see Table 1). An additional sagitally cut 27 GW specimen was processed only for GFAP immunostaining. This second specimen is identified as “27 GW S2” and its results are described in Section 3.3.1.

On the basis of the previously published literature (Bayer and Altman, 2005; Bayer and Altman, 2003; Ding et al., 2022; Ding et al., 2016; Mansour et al., 2019; Olszewski and Baxter, 1982; Tardif et al., 2003), we identified and annotated IC and its subdivisions: the central nucleus (CNIC), the dorsal cortex (DCIC) and the external cortex (ECIC) along with the fiber tracts related to IC: the lateral lemniscus (LL). After the identification of the IC and based on cytoarchitectonical patterns, we annotated the different subparts of the IC, as shown in Figures 1. The adjacent H&E sections were referred to confirm the cells appearances, as well as to rule out any pathology. For comparisons reasons, we also used the marmoset atlas of Paxinos et al. (2012).

In order to create data volumes, the Nissl stained 2D images were first down-sampled to 60 μm per pixel, and then padded and stack aligned using the scale invariant features tool (SIFT) as a feature of ImageJ/Fiji (Lowe, 2004; RRID:SCR_002285). This aligned stack resulted in 3D volumes with 60 μm isotropic voxels, which allowed for reorientation of the brains using the following landmarks: the brain midline, the brainstem orientation, the anterior commissure (ac) and the posterior commissure (pc) line. The volumes of the whole brains (posterior view for better IC visualization) are shown in Figure 1, top row. The 3D reconstructed segmented volumes of the left ICs are displayed in the bottom row. The volumes of the left ICs (Table 2) were calculated using the segmented voxels (shown in brackets), and visualized in 3D viewers such as MATLAB (see Figure 1).

The 3D reconstruction of histological sections was used to evaluate the increase of the IC in the five considered stages in the second trimester. Figure 1 shows the whole brains 3D volumes with the IC highlighted, while the bottom row displays the left IC of all brain specimens ranging from 15 GW (extreme left) to 27 GW (extreme right). The overall IC volume increases approximately threefold from 15 GW (12.85 mm³) to 27 GW (34.27 mm³).

From 15 GW to 18 GW, the IC volume increases sharply by about 1.5 times, whereas there is a minimal rise of only 1.60 mm³ from 18 GW to 21 GW. A large increase of almost 60% is also observed between 21 GW and 24 GW, when the IC volume grows from 19.61 mm³ to 33.21 mm³, respectively. Finally, there is a small increase of only 1.06 mm³ from 24 GW to 27 GW.

As described above, the volumetric increase of the developing human IC is not linear, but occurs in two major stages; the first during the 15–18 GW period, and the second in the 21–24 GW period.

In Figure 2, the IC, its subparts, and the neighboring structures are shown both in sagittal (21 GW) and transverse (22 GW) sections. The CNIC appears as a densely packed structure with a roughly circular shape in the transverse section (Figure 2A), lateral to the periaqueductal gray (PAG). In sagittal view, it appears as an almond-shaped structure (Figure 2B), demarcated rostrally (anteriorly) by the superior colliculus (SC; Figure 2B). The DCIC is identified in both sagittal and transverse planes, superior (dorsal) to CNIC [Figures 2A,B; (Geniec and Morest, 1971; Tardif et al., 2003)]. A portion of the ECIC is present in the caudal (posterior) side of the CNIC separated by a thin part of LL. ECIC is lined by a prominent white matter tract laterally (LL), and by the nucleus of lateral lemniscus, dorsal (NLLd) in its inferior part (ventrally; Figure 2A). The ECIC is present ventrally to CNIC in the sagittal plane and it extends further laterally (about 1.1 mm from the midline; see Figure 2C).

At 15 GW, the MN is very thick and covers a large portion of the ventral IC. Two layers of cells arise from the MN, an outer and an inner layer, and are separated by a cell sparse layer (see Figures 3A,B). The inner layer of cells converges about 1.4 mm from the CA laterally, into the oval shape forming the CNIC (see Figure 3C). The MN gradually decreases in the older developmental stages and becomes almost negligible in 27 GW. We also observe a small PAG posterior to MN at 15 GW (Figure 3A), which increases in cytoarchitectural density at 18 GW, and gradually decreases in 21 (Figure 3I), 24 (Figure 3M) and 27 GW (Figure 3Q).

CNIC is the most prominent part in the adult IC, and it can be detected in the 15 GW specimen as an oval shape with a distinct dense cellular layer relative to the overall appearance of the nucleus (Figure 3C). A streak of thin white matter, identified as the LL, marks the ventromedial boundary of the CNIC whereas there is a diffuse dorsolateral boundary with the DCIC (Figure 3C). The shape of CNIC becomes more prominent with GW (Figures 3C,G,K,O,S). The cytoarchitectural boundaries of DCIC cannot be clearly demarcated at 15 GW. However, DCIC can be distinguished as a distinct group of cells capping the superior (dorsal) part of the CNIC at 18 GW (Figure 3G). It develops morphologically and cellularly at 21 GW (Figure 3K), at 24 (Figure 3O), and at 27 GW (Figure 3S). As the DCIC develops, the boundary of CNIC with DCIC becomes more distinct with increasing GW. ECIC can be seen as a group of cells inferior to CNIC, separated by LL, which also continues further laterally (Figures 3D,H,L,P,T). There are no visible cytoarchitectural patterns in ECIC at 15 GW, apart from a few cellular aggregations which may be migrating cells. In the subsequent stages, ECIC does not undergo any major changes cytoarchitecturally, and appears as a uniform cellular field that is sparse when compared with CNIC.

At 24 and 27 GW, we see a cell-sparse layer which separates IC from SC (the intercollicular area; Figures 3O,S; Nara et al., 1996).

Serial and high-resolution images of Nissl preparations (see Table 1) were used to study the maturation of cells in all three IC nuclei. The various stages of neuronal differentiation (Šimić et al., 2022) are described below:

At 15 GW, the IC includes only undifferentiated and uniformly stained cells (Figures 4A,F,K). At this stage, the neurons cannot be differentiated from the glia. At 18 GW, few cells have circular somata larger as compared with 15 GW (Figures 4B,G,L), and scattered in all IC nuclei. At 21 GW (Figures 4C,H,M), the neurons start to show a single nucleolus in the middle of the cell body, which distinguishes them from other classes of cells. At 24 GW (Figures 4D,I,N) and 27 GW (Figures 4E,J,O), neurons appear as flat enlarged cells (compared to the surrounding glia), with a prominent and dense single nucleolus and a sparsely stained nucleus (see Nara et al., 1996).

Qualitatively, the overall density and number of cells appeared highest in CNIC, followed by DCIC, and the lowest in ECIC.

In the older specimens of 24 GW and 27 GW, based on the cell body shape, we identified two distinct neuronal populations: triangular or polygonal looking shape (stellate-like) and round/oval with circular looking cell body shapes (see Figure 5). These neuronal populations may correspond to those described in the adult human (Mansour et al., 2019), and the rat CNIC (Malmierca et al., 1993). Nevertheless, further experiments and individual cells reconstructions are needed to better identify these neurons (see Malmierca et al., 1993). Stellate-like neurons have been primarily identified in CNIC, and round/oval neurons in DCIC and ECIC.

The literature about the neuroanatomical features of adult human IC provides information about different parts of the IC, and their neuronal architecture (Geniec and Morest, 1971; Mansour et al., 2019). We have used the existing literature of adult human IC to identify and compare the development of IC and its subparts in all our histological sections of all specimens. At a gross anatomical level, CNIC occupies the major part of the IC and a thin layer of cortex surrounds it. The CNIC appears as a dense ovoid patch of cells differentiating it from ECIC and PAG by the developing LL, as described earlier in adult IC (Mansour et al., 2019). The ovoid shape is clearly distinguishable from 21 GW, but the dense cellular organization of the CNIC and the developing LL can be identified in our youngest specimen of 15 GW.

In order to understand the spatial dynamic of the developing IC, we performed volumetric measurements, using isotropic voxels of 60 μm on aligned histological sections. This provided a precise measurement of the IC volume. We observed continuous growth of the developing IC from 15 GW to 27 GW, with the left hemisphere IC at 27 GW measuring 34.27 mm³, nearly three times its volume at 15 GW (12.85 mm³). This finding contrasts with previous literature (Nara et al., 1996), which reports an IC volume of approximately 5.5 mm³ at 27 GW. Several factors may account for this discrepancy. First, Nara et al. may have measured the CNIC volume only, and we included in our measurements all three IC nuclei. Second, our use of isotropic voxels of 60 μm accounts for the unique shape and size of the IC, providing a more accurate volume estimate compared to the Nara et al. method of calculating volume by multiplying the mean area by the nuclei length and considering it as a column. Last, we used cryosections for Nissl staining and volume calculation, with negligible shrinkage, whereas Nara et al. used celloidin blocks, which can shrink the tissue non uniformly by 10% (Nara et al., 1996).

The literature about the development of IC reports accelerated growth of IC from 18 GW (Nara et al., 1996; Sharma et al., 2009a). In the 18 GW specimen, we see an almost two-fold increase in the thickness of the tectum in the sagittal view, with a dense arrangement of radially arranged cells around the CA. We hypothesized this may be at least partially explained by an increased migration pattern around this age. Therefore, we have employed GFAP-IR, which is commonly used to study the developing brain (Rezaie et al., 2003; Sharma et al., 2009b; Verma et al., 2024). GFAP+ radial glial fibers originate from the ventricular zone from 15 GW onwards in a general pattern that resembles that of the developing cerebral cortex. However, the radial glial fibers density in the tectum peaks at 18 GW, and then it is gradually reduced. The development shown by GFAP labeling was confirmed by CR immunohistochemistry. The CR IR at 18 GW reveals a high amount of neuropil staining in all parts of the IC cortex, including the middle portions of CNIC, and compared to other developmental stages. Moreover, CR-positive neurons are spread randomly across the CNIC. This implies that the expression of CR is high at this stage, and it may play an important role in the migration of the neurons.

Nevertheless, the hypothesis discussed here needs further confirmation from more stages during the second trimester of development, as well as within each stage, in such way that statistical analysis becomes possible.

We have imaged all sections at 0.5 μm/pixel resolution (see Table 1), offering a detailed view of intracellular staining, which allowed differentiation between various developmental stages. Overall, the neuronal differentiation in the IC follows a similar trajectory as in other parts of the brain (Šimić et al., 2022). By 18 GW they reach an intermediate stage with multiple Nissl-stained nucleoli, resembling developing astrocytes. Finally, by 24 GW, the IC neuron bodies begin to take polygonal or round/oval shapes (Geniec and Morest, 1971; Mansour et al., 2019). Different from the adult (Mansour et al., 2019), we identified neurons with polygonal cell bodies only in CNIC, but not in the cortical part of the nucleus. One limitation of this study is that of the employed methods: Nissl staining is not the most suitable one to study detailed morphology of neurons. Other types of staining, such as Golgi staining, as well as quantitative analysis, are required to properly classify and describe these neurons. Moreover, it is possible that stellate neurons develop in different parts of the IC at different time points of development, which might depend on other factors, such as the development of white matter fiber tracts (LL; see also Figure 9).

GFAP-IR patterns described above provide information about the development and differentiation of astroglia in the 15–27 GW human fetal IC. Contrasting to the radial glial trajectory in the developing IC, we observed an increase of the astroglia development by the gestational age. The GFAP immunoreactive astrocytes are first observed at 18 GW. This developmental stage may be a point of transition when the glial cells start losing their fibers and begin to differentiate into astrocytes. This may be an initial phase of differentiation, as the astrocytes identified at this stage are spherical, and do not have any processes. As described above, the astrocytes become abundant by 24 GW and this continues at the 27 GW. In these developmental stages, the astrocytes appears to be more mature as they have smaller cell bodies and long processes, giving them a typical look as reported earlier in different parts in the developing brain, and in the adult brain (Kostović et al., 2018; Rezaie et al., 2003). This is supported by in situ data, which suggest that in the IC the astroglia start to differentiate after 16 GW (Sharma et al., 2009b). During 24 GW and 27 GW, the astrocytes tend to cluster around the ventromedial boundary of CNIC lining the LL, which may indicate the start of innervation of CNIC by LL. This was also observed using CR immunoreactivity in the 27 GW specimen (see also Figure 9). However, further evidence is needed to confirm it. Nevertheless, this is the first study to our knowledge that details the dynamics of both radial glia and astrocytes in the second-trimester human fetal IC.

Calcium-binding protein expression patterns are involved in neuronal differentiation, and they have been previously investigated in the IC (Tardif et al., 2003). We have CB and PV immuno-stained sections of the 22 GW coronally cut specimen, and the results are similar with those already reported in the literature (Sharma et al., 2009a). Hence, we report here only the CR expression patterns in the human fetal IC during the second trimester of development. As described above, there are several common and specific features in the general pattern of CR IR across all five developmental stages. The major common feature is the pattern of neuropil staining. In all developmental stages, the CNIC has negligible neuropil staining unlike the cortex. This particular pattern of CR staining has been reported in the adult IC (Tardif et al., 2003). Furthermore, at 27 GW, an intense staining in the NLLd was observed.

Finally, different patterns of neuronal staining are present in different parts of the IC nuclei, and at different gestational ages, which may indicate specific stages of neuron maturation. The spatial specificity of CR+ neurons may also suggest that topographically distinct neuron types or classes may develop and mature at different speeds in the human IC.

To summarize, the development of the human fetal IC during the second trimester can be described as occurring in two major stages. During the first stage, between 15 GW and 18 GW, the process of migration is prevalent, as well as the density of GFAP+ radial fibers. At this stage the cells are immature, and the growth of the IC may be attributed to the high density of radial glial fibers. The second stage of growth occurs between 21 GW and 24 GW, when the process of maturation of cells becomes prevalent, especially in the CNIC, and the morphological change of the IC may be attributed to this. During this stage, the migration process tapers down, and the immature cells start to develop into neurons of different classes, and into glial cells. Overall, the second trimester is therefore a crucial period for the neurotypical development of the human IC.