The sections were incubated in blocking solution (SST & VIP: 10% normal horse serum, NPY: 10% normal goat serum; all solutions containing 0.3% Triton-X100 in 0.1 M PBS) for 1 hr at room temperature before undergoing primary antibody [Anti-mouse SST (H-11, Santa Cruz, RRID:AB_2271061, 1:500); Anti-mouse VIP (Ab30680, Abcam, (RRID:AB_778830, 1:500); Anti-rabbit NPY (N9528, Sigma-Aldrich, RRID:AB_260814, 1:15000); Anti Neuronal Nuclei (NeuN, MAB377 Clone A60; Millipore, RRID: AB_2298772. 1:800)] incubation at 4 °C for 42–46 h. A biotinylated IgG secondary antibody [SST, VIP & NeuN: horse anti-mouse (PK-6102, VECTASTAIN^® Elite^® ABC-HRP Kit,1:200, RRID: AB_2336821); NPY: goat anti-rabbit (PK-6101, VECTASTAIN^® Elite^® ABC-HRP Kit, 1:200, RRID: AB_2336820)] incubation was then conducted for 30 min, followed by treatment with Avidin-Biotin Complex reagent. 3, 3′-diaminobenzidine (DAB) substrate working solution (DAB kit SK-4100, RRID: AB_2336382) was then applied. Given that different isoforms of SST might be present in variable cells, we selected an SST antibody that detects the 14–amino acid active peptide and any larger precursors or peptides containing it, such as SST-28 (Bakhit et al., 1984).

For co-immunofluorescence staining of GABA and SST, we used sections with only 1 h post-fixation and processed them for 42-46 h incubation at 4 °C with primary antibodies [SST (H-11), 1:100 and anti-rabbit GABA (A2052, Sigma-Aldrich, RRID: AB_477652)] before application of the secondary antibodies [1:600; Alexa Fluor^® 488 (ab150109, RRID: AB_2571721) and Alexa Fluor^® 594 (ab150064, RRID: AB_2734146)] for 60 min at room temperature.

After cryosectioning, tissue sections were post-fixed in 4% buffered formalin for minimum 2 weeks and then mounted on double gel-subbed slides out of warm buffered 0.3% gelatine solution. They were air dried for 5-7 days before staining. Details of solutions and steps for staining were as described previously (Gallyas, 1979, modified by Worthy et al., 2017).

After cryosectioning, tissue sections were post-fixed in 4% buffered formalin for 1-2 weeks, then mounted on gel-subbed slides out of warm buffered 0.5% gelatine solution. They were air dried for 5-7 days before defatting overnight in 50:50 chloroform and 100% ethanol. Sections were rinsed in 100% ethanol for 5 min, then placed into xylene for 3 h. Sections were then taken through dehydration and rehydration steps (Powers and Clark, 1955; modified by Worthy and Burman, 2017) into distilled water. From there, sections were placed in filtered cresyl violet at 38-40 °C for 8-10 mins. Then slides were removed and went back through distilled water and increasing gradient of ethanol, followed by differentiation in acidified 70% alcohol to remove background and make both SC layers and cell nuclei more distinguishable.

An Aperio Scanscope AT Turbo (Leica Biosystems) was used to scan sections stained for SST, VIP and NPY using DAB immunohistochemistry ( × 20 magnification, resolution: 0.5 μm/pixel). Neuronal counts were obtained from at least three coronal SC sections 200 μm apart, approximately from the central part of the SC [including AP level +1.0 mm (Paxinos et al., 2012)]. Similar to our previous report (Chong et al., 2022), we placed counting frames (dimensions: 150 × 100 μm) at equal distances across the mediolateral extent of all SC layers (identified using Nissl, NeuN and myelin staining) except (stratum zonale) SZ and SAP which had no or very few cells (Figure 1). The top and right borders of each counting frame were considered the inclusion lines, and the left and bottom borders the exclusion lines and thus cells within the frame or those touching the inclusion lines were counted (Atapour et al., 2017; Chong et al., 2022). Only neurons with clear nuclei were counted, independent of size and shape.

In calculating the number of neurons per volume unit, both the section thickness (40 μm) and a shrinkage factor of 0.801 (Atapour et al., 2017, 2019) were considered. Average data from consecutive sections taken from middle of the SC were combined to calculate the mean neuronal density of SST+ neurons. The most caudal or rostral sections, which do not include all SC layers, were excluded.

For co-localization analysis, we assessed GABA expression in up to 200 SST+ positive neurons present in different layers/sections using images taken by Nikon C1 laser scanning confocal microscope. All statistical analyses were conducted using GraphPad Prism v10.3.1 (Graph-Pad Software, La Jolla, CA, USA). Simple linear regression or one-way ANOVA with post-hoc Tukey comparisons test was conducted for statistical comparison.

Immunostaining for SST revealed the presence of SST+ neurons across the SC layers (Figure 2). The cell bodies of SST+ neurons had different shapes, including many elongated cells. Although the full morphologies were not clearly visible by immunostaining, they showed bipolar and multipolar features (Figures 2C, D). The cell bodies measured approximately 6–20 μm at the longest axis and 3–9 μm at the shortest axis. Those observed in the intermediate layers were generally larger in size, consistent with the presence of large neurons in these layers (Bourne and Rosa, 2003; Chong et al., 2022). The mean density of SST+ neurons was estimated to be about 1,698 ± 364.5/mm³ based on stereological sampling in 3 cases. The density of SST+ neurons was highest in the superficial layers, creating a gradient across the dorsoventral axis of SC (Simple linear regression; R² = 0.38, p < 0.0001), with the SGS having the highest density [Figures 2C, E, SGS: 3,140, SO: 2,230, uSGI: 1,850, lSGI: 1,240, SAI: 910, SGP: 820/mm³, One-way ANOVA; F _{(5, 48)} = 2.94, p = 0.02, Tukey's multiple comparisons; SGS vs. lSGI: p < 0.01 and SGS vs. SAI or SGP: p < 0.001]. The SST+ neuronal density was also higher in the medial side of SC, which corresponds to the representation of the upper contralateral quadrant (Chen et al., 2019; Santana et al., 2023), revealing a decreasing mediolateral gradient for average neuronal density across all layers (Figure 2F, Simple linear regression; R² = 0.31, p < 0.0004). However, we found a similar distribution of these neurons across the rostrocaudal extent of SC, which roughly corresponds to the gradient of representation from central to peripheral vision (Simple linear regression; R² = 0.06, p = 0.20). This contrasts with our previous finding (Chong et al., 2022) of a significant increasing gradient of total neuronal density from rostral to caudal SC.

SST+ neurons constituted approximately 3 – 5% of total SC neurons, based on estimations of total neuronal density using NeuN staining (Chong et al., 2022). This proportion was relatively similar across layers (Figures 3A, B; SGS: 4.67%, SO: 4.55%, uSGI: 5.05%, lSGI: 4.08%, SAI: 3.44%, SGP: 3.15%).

Double immunofluorescence staining for SST and the neurotransmitter GABA confirmed the inhibitory nature of SST+ neurons (Figure 3C). Our analysis in 200 SST+ neuronal images taken by confocal microscopy revealed that they all co-express GABA. While we cannot exclude the possibility of a small population of excitatory SST+ neurons, it is clear that the SST+ neurons are primarily GABAergic.

Our results revealed no examples of VIP+ and NPY+ neurons in the SC (Figure 4A) despite the clear presence of neurons stained for both peptides in the cortex and subcortical structures including thalamic reticular nucleus, hippocampus and caudate nucleus, present in the same sections (Figure 4B). This observation was consistent across sections covering the entire SC in each of the three stained cases, suggesting lack of expression of these peptides in the SC of marmoset monkey.

Identifying neuronal subtypes is crucial for understanding circuitry, connectivity and function (Liu et al., 2023; Sans-Dublanc et al., 2021). SST expression in neurons of marmoset SC is consistent with findings in other mammalian species (Laemle et al., 1982; Harvey et al., 2001; Byun et al., 2016; Spangler and Morley, 1987; Mensah-Brown and Garey, 2006; De Souza et al., 2015). A more prominent SST expression in the superficial layers, confirmed by immunostaining (Harvey et al., 2001; Mensah-Brown and Garey, 2006) and gene expression data (Shimogori et al., 2018), are indicative of a possible role of these neurons in visual processing. Further validating this idea is the observation that, contrary to the higher density of overall neuronal populations in the caudal parts of SC (Chong et al., 2022), the rostrocaudal distribution of SST+ neurons is mostly uniform, suggesting that there may be relatively more SST+ neurons present in the anterior part of marmoset SC, which receives foveal visual input (Chen et al., 2019; Hafed et al., 2021). Moreover, our observations on the expression of SST in thalamic reticular nucleus neurons (Figure 4), along with the known role of these neurons in modulation of visual information processing (Bu et al., 2025) suggest a conserved role of SST+ neurons in visual processing across different mammalian species along with ample evidence of involvement of SST+ neurons of visual cortex to such processes (Rikhye et al., 2021; Song et al., 2020; Ma et al., 2010; Onorato et al., 2025).

Apart from the dorsoventral and rostrocaudal distribution of SST+ neurons discussed above, we also observed a mediolateral gradient for these neurons across SC. However, due to lack of quantitative data in other species (Byun et al., 2016; De Souza et al., 2015; Harvey et al., 2001; Mensah-Brown and Garey, 2006; Spangler and Morley, 1987), the functional implications of this gradient remain to be understood. Data in other species of primate indicate asymmetries in visual processing between the upper and lower visual fields (Jóhannesson et al., 2018; Zito et al., 2016), including roles such as attention (Palmieri et al., 2023).

While inhibitory nature of SST+ neurons in cortex, hippocampus and some other structures are well known (Yavorska and Wehr, 2016), to the best of our knowledge, there has been no study in primates investigating directly whether SST+ neurons in SC were inhibitory. In the cat SC, particularly in the SGS, the SST+ neurons have the morphology of local interneurons (Spangler and Morley, 1987). In the mouse SC, previous in-situ hybridization data (Harvey et al., 2001), along with more recent transcriptomic studies (Choi et al., 2023; Liu et al., 2023; Cheung et al., 2024) define SST as a specific GABAergic subtype in the SC. Here our study provides direct evidence for the inhibitory nature of SST+ neurons in the SC of marmoset monkey, while their role in visual processing and perception remains to be tested.

We observed an absence of NPY protein expression in the marmoset SC, despite its reported expression in the cat and rodent (Coveñas et al., 1990; Harvey et al., 2001). While NPY gene expression has been reported in the rat SC, mostly within the superficial layers (Harvey et al., 2001), marmoset SC shows very low level of gene expression for NPY only at birth, but not in adulthood (Shimogori et al., 2018; Kita et al., 2021) consistent with our immunohistochemical data. Similar to marmosets, studies in lemurs and squirrel monkeys also reported an absence of NPY+ cell bodies in the SC (Bons et al., 1990; Smith et al., 1985). The absence of NPY in primate SC, along with the increased neocortical NPY innervation, particularly in humans and great apes (Raghanti et al., 2014), may highlight the primate specializations in the circuitry roles involving this neuropeptide. NPY-containing neurons were present in cortex and other subcortical structures such as caudate and hippocampus (Figure 4), where they are known to influence a variety of physiological and cognitive functions (Thorsell et al., 2006).

The absence of VIP protein expression is consistent with lack of its gene expression in the marmoset SC (Shimogori et al., 2018; Kita et al., 2021). Limited data in other species, such as expression of both VIP protein and gene in rodent SC (Byun et al., 2016; Dussaillant et al., 1992; Ogawa-Meguro et al., 1992; Okamoto et al., 1990), vs. its absence in the cat SC (Borostyánkoi et al., 1999), make it clear that there is no uniform expression of these peptides in different mammalian species. Furthermore, it points to obvious differences in the cellular circuitry of the SC in the marmoset, and perhaps other primates, compared to rodents.

SST and VIP immunostaining requires optimal fixation (as specified in the methods section) to label reliably all of the cell populations in different structures. Difficulty of staining could be a contributing factor behind the significant lack of immunostaining data for VIP and SST in the literature. However, in the present study we found clearly labeled neurons in other structures, processed in the same batches and in many cases in the same sections.

Finding out molecular markers that define specific cell types in SC is essential for better understanding of its circuitry, including accurate biophysical models of cellular interactions underlying behavior. Our data point to the neurochemistry of the adult SC neurons in a non-human primate which has become important in studies of vision (Solomon and Rosa, 2014) that might be different during development and aging as well as chronic conditions associated with disease or injury. Future studies should characterize the full morphology of SST+ neurons using other techniques including cell filling and higher-resolution imaging to better understand layer-specific axonal and dendritic distributions and overall SC microcircuit organization.