We are interested in understanding morphological changes during development in the human auditory cortex and decided to focus on the development and maturation of inhibitory interneurons, particularly somatostatin (SOM+) interneurons that shape cortical circuits and regulate plasticity, which remains underexplored. The present study examines the development of SOM+ interneurons in the postmortem human auditory cortex from the prenatal period to adulthood, within primary and association auditory cortices. We find that there is an increase in SOM+ interneurons in the infragranular and subsequently supragranular cortical layers between prenatal development and childhood. Soma size and aspect ratio increase between the prenatal development and adulthood, of varying magnitudes across subtypes and cortical layers. Neurite complexity is highest during fetal and early postnatal stages and decreases during adolescence. The density of SOM+ interneurons peaks in layers 2 and 3 during childhood and adolescence and decreases in adulthood, coinciding with known periods of auditory plasticity. Our findings provide critical insights into the structural and spatial changes in SOM+ interneurons that likely regulate the sensitive period of the human auditory cortex. Since alterations in SOM+ interneuron density are linked to various neurodevelopmental disorders, our results provide an important baseline for future functional and clinical investigations.

All procedures on post-mortem human brains used in this study have been approved by the Institutional Human Ethics committee at the National Brain Research Centre (BRIC - NBRC), Manesar, which are in accordance with international guidelines and those framed by the Indian Council of Medical Research and were also followed by collaborating institutes. All experiments were undertaken with the understanding and written consent of the next of kin or parents of each subject. Data regarding the identity of subjects and their case histories were anonymized and stored securely at BRIC - NBRC.

A total of 25 post-mortem human brain samples ranging from 30 gestational weeks to 45 years were used in this study (Supplementary Table 4). These cases were categorised into five groups based on age, namely, FE, PN, P, ADL, and AD. We ensured that the parents of the aborted fetuses were not suffering from any serious illness or psychological disorders and had no history of hearing or speech defects. Samples of the temporal lobe from the left hemisphere were collected only from those cases where there was no medical history of neurological or psychiatric disorders nor history of any hearing or speech deficits. We obtained sixteen of these samples (n = 2, Postnatal; n = 4, Pediatric; n = 5, Adolescence; n = 5, Adult) from the All India Institute of Medical Sciences (AIIMS), Delhi, and the remaining nine (n= 5, Fetal; n = 3, Postnatal; n = 1, Pediatric) from the Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh (Supplementary Table 4). Methods for dissecting formalin-fixed tissue and specifically, the human auditory cortex including the Heschl’s gyrus (area TC) and adjacent non-primary auditory areas (TA and TB) were performed as described by Pundir et al., (2012)³. Blocks containing TC, TA and TB were photographed in different orientations (dorsal, lateral, rostral, and caudal) to facilitate sectioning and identifying the locations of different auditory areas after staining. The blocks were kept in a cryoprotectant, that is, 30% sucrose in PBS at 4°C until they were completely immersed in the solution. Blocks were then frozen using dry ice and mounted in the optimal cutting temperature (OCT) media (Tissue Freezing Medium, Leica). This was followed by cryosectioning at -20ºC using a cryostat (Leica, CM3050 S) to obtain 7 sets of serial sections, each having a thickness of 50 µm^2,3. The first set of these serial sections was used for Nissl staining to distinguish between different cortical layers and to delineate auditory areas based on cytoarchitecture⁴¹.

Immunohistochemistry

Immunohistochemistry was used to detect the presence of SOM⁺ neurons in the auditory cortex (left hemisphere) from all ages used in our study. The sections were rinsed thrice using 0.01 M PBS and incubated in citrate buffer (pH 6.0) for 20 minutes at 80°C for antigen unmasking.  Sections were incubated in 2% H₂O₂ with 0.3% Triton-X 100 for 20 minutes in the dark to quench endogenous peroxidase activity. After rinsing sections thrice in PBS, they were incubated for 1 hour in a blocking solution containing 5% normal goat serum (S-1000, Vector Laboratories, Burlingame, CA, USA), and 1% Bovine serum albumin (BSA; A7906, Sigma-Aldrich, USA). This was followed by incubating the sections overnight (for 15-17 hours) at 4ºC in a solution containing the primary antibody against SOM (made in rat; 1:200; Cat# ab30788, Abcam, RRID: AB_778010 or Cat# S5331, USBiological). The sections were again rinsed thrice in PBS and incubated for 2 hours in a secondary antibody solution (Biotinylated anti-rat antibody raised in goat; 1:250 dilution, BA-9400, Vector Laboratories, Burlingame, CA, USA) at RT. After this step, sections were rinsed in PBS three times and incubated in a solution containing the avidin-biotin complex (ABC reagent; PK-6100, Vectastain Elite ABC HRP kit, Vector Laboratories, USA) for 2 hours. This was followed by three rinses in PBS, after which the sections were developed in a solution containing the chromogen (NovaRED substrate kit, peroxidase; SK-4800, Vector Laboratories, USA) according to the manufacturer’s instructions. Finally, the sections were rinsed in PBS, mounted on gelatin-coated slides, and air-dried overnight before coverslipping with DPX. The same protocol and time durations was followed for each step of the staining procedure for all age groups used in our study. This ensured that variability due to staining duration was minimized across age groups. A negative control was also performed by incubating a section of the human auditory cortex in a solution containing the secondary antibody while omitting the primary antibody against SOM (Supplementary Fig. 17).

Double labelling protocol

The immunohistochemistry protocol used for SOM detection (detailed above) was followed by Nissl staining to visualize SOM⁺ neurons in relation to their nuclei. Additionally, immunofluorescence staining was performed to show co-labelling of SOM⁺ interneurons with the neuronal marker NeuN. For immunofluorescence, sections were rinsed, subjected to antigen unmasking and endogenous peroxidase quenching, as mentioned earlier. Sections were then blocked and incubated in the primary antibody against SOM, as described above. After this step, sections were incubated for 4 hours at room temperature in the dark with Alexa Fluor 488 anti-rat secondary antibody (raised in goat; 1:500; Cat#A11006; Invitrogen, USA). Following PBS rinses, sections were blocked for 1 hour in 5% normal horse serum (S-2000, Vector Laboratories, Burlingame, CA, USA) and then incubated overnight at 4°C with the NeuN antibody (raised in goat; 1:2000; Cat#PA5-143586, Invitrogen, RRID: AB_2942815). After rinsing in PBS, sections were incubated with Alexa Fluor 594 secondary antibody (Anti-goat, raised in donkey; 1:500; Cat#A11058; Invitrogen, USA) for 4 hours at room temperature. Finally, sections were rinsed thrice in PBS, mounted on gelatin-coated slides, air-dried, and coverslipped with Vectashield® Plus Antifade mounting medium containing DAPI nuclear stain (H-2000, Vector Laboratories, USA).

Quantification of SOM⁺ Interneurons

Somatostatin-positive neurons at different stages of development were quantified using the optical fractionator method (Stereoinvestigator software; Microbrightfield, Williston, VT) attached to an Olympus microscope (BX51). This method employs stereology wherein Systematic Uniformly Random Sampling (SURS) is performed by placing uniform grids over the area of interest. A contour was drawn around the areas TA, TB, and TC of the auditory cortex in three serial sections at an interval of 7. The quantification of SOM⁺ neurons was performed using a 40X objective, with the counting frame set at 150 μm × 150 μm, the sampling grid area set at 173 μm × 173 μm, and an optical disector height of 6 μm with 2 μm guard zones. The thickness of sections was measured at each counting site.

Staining intensity across ages and cortical layers in auditory areas was quantified from images acquired under identical conditions (light intensity, brightness, contrast, exposure). Images were converted to grayscale for analysis. Regions of the tissue lacking cells were marked as background, whereas individual SOM⁺ neurons were outlined and categorized based on staining intensity as very light, intermediate (light to medium), or dark. Mean intensity values were measured for each category. Background staining had a mean intensity of 197 ± 15, whereas very lightly stained neurons had mean intensities of 166 ± 10 and were excluded from counts.

Only those SOM⁺ neurons which were darkly stained (85 ± 23) or demonstrated medium to low (135 ± 12) levels of staining compared to background staining, with one or more clearly distinguishable dendrites were counted and considered for analysis. Accordingly, all analyses were restricted to SOM⁺ interneurons with low to high levels staining.**** Whereas the very lightly stained SOM⁺ interneurons may represent a distinct category reflecting either developmental or functional differences, we decided to exclude them from our analyses to avoid introducing errors in the morphometric analysis that we performed. An estimated count using mean section thickness divided by estimated volume (Cavalieri method) was used to compare cell density across various age groups, layers, and areas^91,92. For comparing the overall distribution of SOM⁺ interneurons, digital images of sections stained for Nissl and somatostatin were acquired and adjusted for brightness and contrast but were not altered in any other way. 

Morphometric Analysis of SOM⁺ Interneurons

A three-dimensional reconstruction of SOM⁺ interneurons at all developmental ages in each cortical layer and area was performed using Neurolucida and Neurolucida Explorer (Version:11, MBF Bioscience, USA) at a magnification of 100X. We specifically chose darkly stained neurons wherein the somata were clearly visible and possessed intact processes for reconstruction, all of which were considered as neurites. However, it is possible that some of the neurites may have extended into the sections above and below this plane. The morphometric analyses included changes in neurite branching and length (provided by the Neuron summary module) and a Sholl analysis was also performed to quantify neuronal complexity, using the Neurolucida Explorer software. For the Sholl analysis, concentric circles at a fixed interval (10 μm) were placed around the neuronal soma and the number of the points wherein neurites intersected these circles provided a measure of neuronal complexity. A statistical analysis was performed on data collected from the Sholl radius greater than 10 μm, to avoid errors in neuronal complexity, since the soma size of SOM⁺ interneurons increased with age.

            Somatostatin-positive interneurons undergo significant physiological and functional maturation during the 2nd and 3rd postnatal weeks in the mouse cortex^93,94,95, with axonal outgrowth being particularly prominent during the postnatal period following migration, when dendritic and axonal features become more distinguishable. Since we did not employ markers such as Ankyrin-G, TRIM46, MAP2, and Tau to differentiate axons and dendrites^96,97,98,99, the difference between axons and dendrites was clear in some cases in our study and ambiguous in others. It was also not possible for us to use techniques such as filling individual neurons with neurobiotin⁹⁴, since these techniques require fresh or very lightly fixed brain tissue. We therefore decided to label all processes visualized using immunohistochemistry as neurites and perform the morphological analysis, rather than erroneously classifying them as dendrites or axons. 

Statistics and reproducibility

Statistical analyses were performed on the density and morphometric data following quantification and three-dimensional reconstruction of SOM⁺ interneurons using R (version 4.1.1) software. The data was tested to check whether it was normally distributed and possessed equal variance, using the Shapiro-Wilk and Levine tests, respectively. A One-Way ANOVA was used to compare developmental changes in the density and complexity of SOM⁺ interneurons, for normally distributed data with equal variance. However, if the data was not normally distributed, a Kruskal-Wallis test was performed to judge statistical significance.  In both cases, the Tukey’s or Dunn’s post-hoc tests were performed for multiple comparisons.  A P-value of less than 0.05 was considered statistically significant for all statistical tests. All statistically significant and non-significant test values are provided as Supplementary Data 1.

            To assess the relationship between density and morphometric data with age, correlation analyses was performed. As the data were not normally distributed, MIC was used to detect non-linear and complex associations that may not be captured by simple linear correlation methods. The MIC values range from 0 to 1, with higher values indicating stronger associations between variables¹⁰⁰. Therefore, MIC values below 0.3 were considered indicative of very weak associations, values between 0.3 to 0.6 as moderate associations, and values above 0.6 as very strong associations. The statistical significance of MIC was assessed using permutation-based tests, and the p-values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure.