For the purposes of H&E staining, we used two sexually mature female O. minor specimens, with a body weight of 160~180 g, obtained from commercial catches in the coastal regions of Jeollanam-do, South Korea, in May 2017. These specimens were identified as O. minor based on an examination of morphological characteristics [12]. Prior to experimentation, the two females were maintained for 3 months in filtered fresh seawater; during this time, they were fed with Manila clams. Comparison of the mitochondrial DNA (mtDNA) COI (cytochrome oxidase subunit I) barcode region [13] was performed to confirm species identity and determine haplotypes of the COI gene. Briefly, total DNA was isolated from the arm muscle of each animal using a DNeasy Blood & Tissue kit (QIAGEN, USA), following the manufacturer's protocol. For amplification of the partial COI gene, polymerase chain reaction (PCR) was performed as previously described [14], using the primers HCO2198 and LCO1490 [15]. The PCR product of approximately 630 bp was purified using a QIAquick PCR purification kit (QIAGEN, USA) and sequenced in an ABI 3730xl DNA Analyzer (Applied Biosystems, USA) using the same PCR primer set. The sequence data were edited and aligned using ATGC 4.0 software (Genetyx, Japan). For intact whole brain imaging and immunostaining, we used three sexually mature male O. minor specimens, with a body weight range of 220~290 g, captured in June 2017 or June 2018, and identified based on morphological characteristics. For each animal, sex was determined by the presence of eggs in the mantle cavity (for females) or a hectocotylized right third arm (for males).

For H&E staining, the animals were anesthetized by immersion in artificial seawater containing 2% ethanol for 8 min. When they became relaxed and immobile, cartilaginous tissues containing the brain were excised and fixed with 10% formalin in artificial seawater at 4℃ for 2 days. Paraffin-embedded serial sagittal or coronal sections at 10 µm thickness were obtained from each brain (Histore Inc., Korea). Sections were stained with H&E using a staining kit (ScyTek Laboratories, USA). Briefly, Mayer's Hematoxylin (Lillie's modification) was applied to completely cover the tissue section and incubated for 5 min. After rinsing twice with distilled water to remove any excess stain, a Bluing Reagent was applied for 10~15 s. Following two additional washes in distilled water, slides were dipped in absolute alcohol. Thereafter, the slides were incubated in adequate Eosin Y solution (modified alcoholic) for 2~3 min, rinsed, and dehydrated four times in absolute alcohol. The slides were then cleared and mounted with Histomount (National Diagnostics, USA).

Stereoscopic images of the fixed intact brain were taken after removing the cartilage surrounding the brain as completely as possible using forceps. Images of the intact brain and H&E slide preparations were obtained under a LEICA S6D stereoscope (Leica Microsystems, Germany) using an OPTINITY KCS-2000SS CMOS camera (Korea Lab Tech, Korea). White balance was adjusted using OptiView 3.7 software (Korea Lab Tech, Korea). To establish the coordinates of each image, the distance from the midline point of the posterior end was calculated according to the thickness of the sections. Illustrations of brain gross anatomy were sketched by tracing a stereoscopic image of the intact brain using PowerPoint software (Microsoft, USA). For delineations, 40 representative H&E images for each sagittal and coronal section were selected based on the appearance of distinguishable changes in structural morphology. The delineations were hand-drawn on a transparent film placed over each of the paper-printed H&E images. Outlines of the brain were sketched with thick black lines, the brain internal structures with thin black lines, and Hematoxylin-stained layers with cyan lines or planes. All sketches were scanned and used as templates for tracing (Penodegr Inc., Korea) using Adobe Illustrator software (Adobe, USA).

On the tracings, we indicated the name of each subregion with abbreviations, using Adobe Illustrator software (Adobe, USA). For efficient conveyance of information among cephalopod researchers, we adopted the terms used in previous studies. All nomenclatures and abbreviations used are based on those presented in the book “The anatomy of the nervous system of Octopus vulgaris” [9]. With reference to the subregions identified on non-serial, partial images of O. vulgaris brain sections, subregions of the O. minor brain were carefully identified by the appearance and disappearance of distinct structures along the serial H&E images. The organization of this atlas is based upon that of the Paxinos mouse brain atlas [16].

For immunofluorescence staining in the supraesophageal and subesophageal masses, 10-µm-thick sagittal paraffin sections were prepared using the same procedure as described for H&E staining. In case of the optic lobe, 30-µm-thick free-floating sagittal sections were prepared using a VT-1200S vibratome (Leica, Germany), after fixation with 4% paraformaldehyde overnight at 4℃, followed by dehydration and rehydration. Immunofluorescence staining of neurons was carried out as previously described using an anti-acetylated α-tubulin antibody (1:500; Sigma-Aldrich, USA) [17]. For fluorescent detection of antibody labeling, we used an Alexa Fluor 568 goat anti-mouse conjugate (1:1000; Molecular Probes, USA). DAPI (ThermoFisher, USA) counterstaining was performed for visualizing cell nuclei. Fluorescence images were obtained using a BX53 digital fluorescence microscope (Olympus, Japan).

We initially isolated the brain with surrounding cranial cartilage, which is located inside the head of O. minor (Fig. 1A). After complete removal of cartilage, the intact O. minor brain (Fig. 1B) showed a typical brain morphology of the Octopodiformes brain [8]. The gross anatomy of the brain indicated that the supraesophageal and subesophageal masses are distributed around the esophagus, and the optic and olfactory lobes are bilaterally attached (Fig. 1C). Their orientation was determined by the position adopted by the animal during normal horizontal locomotion, as reported in previous studies [8]. In detail, along the esophagus, the direction toward the mouth is anterior, and on the opposite side, it is posterior. The direction towards the midline, which divides the brain symmetrically, is medial, whereas the direction away from the midline is lateral. The supraesophageal mass is considered to be the dorsal part, and the subesophageal mass is the ventral part.

Prior to preparing tissue for construction of the O. minor brain atlas, we confirmed the species identity of the animals by comparing nucleotide sequences of the mtDNA COI gene, which is the most commonly used DNA barcode region for invertebrates, including O. minor [13]. Sequence identification of the mtDNA COI gene through NCBI GenBank database (http://www.ncbi.nlm.nih.gov/) revealed that the sequences of the two specimens used for this atlas (sagittal section and coronal section, respectively) showed over 99% sequence identity to that of O. minor (GenBank accession No. HQ638215.1), 87% similarity to O. kaurna (AY545188.1), and 85% similarity to O. hubbsorum (AB385874). Within the 630-bp region of the COI gene, 12 variable sites were recognized (Fig. 2A) as previously reported [14]. They exhibited a haplotype 1 pattern that is found in Korean populations, with a high haplotype frequency distribution value (0.7~1.0), compared with haplotype 6, which is found only in three populations in China (Fig. 2A).

We subsequently obtained serial sagittal sections covering the entire brain of the sequence-confirmed, sexually mature adult O. minor (Fig. 2B). To establish the coordinates of each section in this atlas, we selected the midline point of the posterior end as the fiducial point (asterisk in Fig. 2B). Sections were stained with H&E to visualize the cellular nuclei and to facilitate subregion discrimination (Fig. 2C). We obtained 122 sagittal images of the serial sections at 30~70-µm intervals (Supplementary Materials 1). Among the images, we carefully selected 40 images that sufficiently represented the complex structure of the O. minor brain. After preparing delineated illustrations of the selected images (Fig. 2D), we established the name of each subregion and denoted it accordingly (Supplementary Materials 2). We also prepared serial coronal sections covering the entire brain of the other sequence-confirmed sexually mature adult O. minor (Fig. 2E). As with the sagittal sections, coronal sections were stained using H&E (Fig. 2F). We obtained 112 coronal images of the serial sections at 40~80-µm intervals (Supplementary Materials 3). Forty representative images were selected for illustration (Fig. 2G) and subregion identification (Supplementary Materials 4). On the basis of delineation of representative H&E images, we identified more than 70 subregions, the abbreviations of which are listed in Table 1. We followed the nomenclatures and abbreviations used in Young's book [9], because it describes the largest number of subregions among the previous cephalopod studies and O. vulgaris belongs to the same genus with O. minor.

In the constructed atlas, the O. minor brain showed many compartmentalized structures which might be involved in particular brain functions such as motor control and cognition. A number of lobes were discriminated by the Hematoxylin-positive layer (Supplementary Materials 1 and 3). We observed that the subesophageal mass consisted of brachial, pedal, chromatophore, funnel, vasomotor, and other lobes (Fig. 3A, 3B, and Supplementary Materials 2 and 4). These are known to be lower and intermediate motor centers. According to previous studies that performed electrical stimulations and induced lesions in the brains of O. vulgaris, the anterior part of the subesophageal mass is involved in the actions of arms and suckers, and the posterior part is involved in the actions of the mantle and viscera [8,9]. Compared with the subesophageal mass, the supraesophageal mass showed a more complex structure, consisting of multiple lobes, including the vertical, basal, frontal, and buccal lobes (Fig. 3A, 3B, and Supplementary Materials 2 and 4). In O. vulgaris, the basal lobes are known to be higher motor centers that are responsible for fine motor control and coordination, and other supraesophageal lobes are implicated in cognitive processes such as feeding and defensive behaviors [8,9]. Although similar functions are expected in the brain of O. minor, the contribution of each subregion to the behavior of O. minor needs to be tested by additional experiments. The optic and olfactory lobes were present on each lateral side (Fig. 3B and Supplementary Materials 4). With statocysts, these periesophageal lobes seem to relay sensory information to other parts of the brain.

Some of the known lobes in the brains of O. vulgaris were difficult to detect in O. minor, and thus additional staining studies will be needed to locate these. In addition, it is possible that some subregions that may be involved in the inhabitance of O. minor in mudflats are unique or different from those of the O. vulgaris, with respect to the relative volume and microcircuitry level. Further studies addressing this issue would be of great interest.

From the H&E images of the supraesophageal and subesophageal mass, we found that the thickness and cell density of Hematoxylin-positive layers markedly varied depending upon the region (Fig. 4A). The optic lobe showed a distinctive structure, comprising islands of cells and a thin double Hematoxylin-positive layer near the surface (Fig. 4A). In the magnified images, the size of cellular nuclei in Hematoxylin-positive layer seemed to be diverse (image not shown). Therefore, we questioned the cellular identity of the dense Hematoxylin-positive cell layers of the O. minor brain, and sought to determine whether these layers contain neurons, as in the mammalian hippocampal formation [18], or non-neuronal cells, as in the Bergmann glial cell layer of the mammalian cerebellum [19]. To address this issue, we performed immunostaining using an anti-acetylated α-tubulin antibody, which was previously utilized for visualizing neuronal somata and axons in many other cephalopod species [20,21,22,23,24], followed by counterstaining with DAPI. The somata of nerve cells showed a strong immunoreactivity with DAPI staining, whereas immunoreactivity without DAPI staining was observed in the regions with neuronal fibers. We found that the proportion of neurons in the DAPI-positive cell layers, the neuronal distribution patterns, and the size of neuronal somata varied markedly depending upon the region (Fig. 4B). Compared with the lobes of the subesophageal mass, those of the supraesophageal mass showed a considerably more diverse staining pattern. Interestingly, most of the cells in the thick dense DAPI-positive cell layer of the vertical lobe (v.l.), which is also known to function in the learning and memory process in O. vulgaris [8], were found to be non-neuronal cells (Fig. 4B). The subesophageal mass and optic lobe (opt.) showed intrinsic neuronal distribution patterns with less diversity (Fig. 4B). These results indicate the heterogeneity of neuronal subtypes as well as the presence of diverse and complex neural circuits in each brain subregion. Additional staining studies to identify subtypes of neurons and non-neuronal cells, possibly glia, and determination of the physiological functions of non-neuronal cell layers will provide interesting areas of focus in future research.

Although individual differences in brain size according to body weight and tissue shrinkage after formalin fixation are factors that should be taken into consideration, this study, nevertheless, represents an initial step towards characterizing the brain of O. minor. To the best of our knowledge, this is the first brain atlas of an Octopodiformes species as well as that of an adult cephalopod. Similar to the brain atlases of other animals, this atlas will enable site-specific approaches to brain stimulation, gene knockdown, and gene expression analyses in the O. minor brain. Although our findings have raised many questions regarding the cellular identity of neurons and non-neuronal cells for further investigation, we anticipate that the O. minor brain atlas compiled in the present study will make a valuable contribution to comparative neuroscience research.