Introduction

α-Smooth muscle actin (αSMA) belongs to the actin superfamily that is conservatively expressed in eukaryotic cells and supports a wide range of cellular functions, such as cytoskeletal maintenance, myofiber contraction, cell division and migration, chromatin remodeling, transcriptional regulation and vesicle trafficking1,2,3,4,5,6. αSMA is the actin isoform best characterized in vascular smooth muscle cells forming the contractile thin filaments7,8. αSMA is encoded by the ACTA2 gene in human located on chromosomal site 10q22-q249. Mutations in this gene cause various vascular diseases10, such as cerebral arteriopathy11, arterial aneurysms12, multisystemic smooth muscle dysfunction syndrome13,14,15, Moyamoya disease16 and cortical dysgyria17. Notably, αSMA is also expressed in fibroblasts, which is involved in wound healing but also tumorigenesis18,19,20,21,22. In addition, αSMA expression has been reported in hair follicle dermis23, vascular pericytes24,25, mesenchymal stem cells26, cultured astrocytes27,28 and lens epithelial cells29. Thus, αSMA appears to play some biological roles in many non-myocytic cells.

The hippocampal formation supports complex cognitive functions in mammals30,31,32. The hippocampal trisynaptic circuit relays neuronal activity from the entorhinal cortex to dentate granule cells via the perforant path, further to CA3 pyramidal neurons via the granule cell axons or mossy fibers (MF), and finally to CA1 pyramidal neurons via Schaffer collaterals33. MF terminals form complex synapses with the thorny excrescences, a type of large-sized postsynaptic apparatus, on the dendrites of CA3 pyramidal neurons and hilar mossy cells. This giant MF synapse is featured by great neurotransmission capacity and dynamic plasticity34,35,36,37,38. The trisynaptic circuit is critically involved in learning and memory and other cognitive functions such as perception of space and time and is pathogenically related to neurological and psychiatric diseases39,40,41,42,43,44,45.

We incidentally noticed a distinct αSMA immunolabeling at the hippocampal MF terminals in adult and aged human brains, which pointed to an underrecognized neuronal expression of this actin isoform in the brain and a potential neurobiological implication of this protein for the hippocampal giant synapse. In the present study, we first cross-validated this novel axonal αSMA expression in the adult human brain using immunohistochemical and immunoblotting methods, including by assessing aSMA immunolabeling pattern in the periphery. Further, we explored the origin of this molecular pattern in phylogenic and ontogenic perspectives.

Materials and methods

Postmortem human brains and tissue preparation

Postmortem brain and peripheral tissue samples were used in the current study with the approval (2020KT-37, 4/10/2020; #2023-KT084, 6/21/2023) by the Ethics Committee of Central South University Xiangya School of Medicine in compliance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). Postmortem human brains were banked through a willed body donation program, with the donations conducted following written informed consent by the donor and/or the next-of-kin46. Research using postmortem human brain samples is not a human subject study (not involving live humans), therefore patient consent was not applicable to the present investigation.

Each brain was bisected after removal from the skull, with a half cut into 1 cm-thick coronal slices, fresh-frozen then stored at − 80 °C. The other hemibrain was immersed in formalin for 2–4 weeks, which was subsequently cut coronally into 1 cm-thick slices. Tissue blocks were obtained from the fixed slices and prepared into cryostat sections at 35 µm thickness and paraffin sections at 3–4 µm thickness. All brains from the adult and aged donors were neuropathologically evaluated for the presence of Alzheimer’s and Parkinson disease lesions according to the Standard Brain Banking Protocol proposed by China Brain Bank Consortium47. Postmortem fetal brains were banked under parent request with signed consent. The fetuses died during or after delivery due to congenital abnormalities, with autopsy carried out for pathological diagnosis including histological examination of the brain. A total of 45 brains lacking neuropathological changes were used in this study. Paraffin blocks of peripheral tissues from several adult cadavers were available in the lab and used in this study to assess ɑSMA immunolabeling pattern in various organs (Supplemental Table 1).

For immunoblot, a temporal lobe slice (~ 0.5 cm thick) passing the anterior hippocampus was dissected out, followed by focal dissections of the dentate gyrus (including the granule cell layer and hilus), subiculum (the pyramidal cell layer) and temporal neocortex (layers II-IV) under a surgical dissection scope. Frontal neocortical and cerebellar cortical samples were also dissected out in a similar way by avoiding the tissue near the pia. In addition, the basal arteries were dissected out and cut into segments. All the samples were placed separately in plastic tubes, weighed, labeled and stored at − 80 °C until use.

Animal brain sections

Brain sections from different mammalian species collected in previous research projects and preserved in a cryoprotectant in the principal investigator’s (XX Yan) lab were used in the current study. Specifically, brain sections from adult C57BL/6J mice (2–4 months of age, n = 6) were available from previous investigations in Alzheimer’s disease mouse models along with wildtype controls48. Brain sections from adult Sprague Dawley rats (2–6 months of age, n = 6) were from the control groups used in experimental study of cerebral stroke49. Brain sections from adult guinea pigs (1–3 years of age, n = 4) were from original study in examination of immature neurons50. Brains sections from domestic cats (4–6 years of age, n = 4) and rhesus monkeys (12–33 years of age, n = 6) were originally obtained for studying cerebral immature neurons and Alzheimer’s disease pathology48. In addition, temporal lobe blocks from perfused (4% paraformaldehyde in 0.1 M phosphate-buffered saline, PBS, pH7.2) rhesus monkey brains (3–4 years of age, n = 4) were provided by Kunming Institute of Zoology. These animals were used as control in research projects unrelated to the present study, approved by the Ethics Committee of Kunming Institute of Zoology, Chinese Academy of Sciences. All experimental methods were carried out in accordance with NIH guidelines on animal housing, maintenance and euthanasia procedures, and the ARRIVE guidelines. Coronal brain sections from the above species in adult age range and passing the mid-hippocampus were used for quantitative analyses in the current study.

Cultivation of cell lines and primary neurons

Pilot experiments were carried out to screen αSMA expression using nonneuronal and neuronal cells lines obtained commercially (see Supplemental Fig. 2), with the SH-SY5Y cells used in formal experiments (Procell Life Science & Technology Co. Ltd., Wuhan, China). Human embryonic cortical cells were obtained from a commercial supplier (Procell Life Science & Technology Co. Ltd.) (Supplemental Table 2). Rat primary cortical cell cultures were prepared using newborn pups from two litters following established protocol51.

Cell stocks were re-suspended, adjusted to optimal beginning densities (depending on experimental purpose) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 1X penicillin/streptomycin at 37 °C in Corning-Costar plates in humidified incubators with 95% air and 5% CO2 supply. The embryonic human cortical cells were cultured with a specialized medium developed by the supplier (Procell Life Science & Technology Co. Ltd.). Neonatal rat cortical cells (two litters of pups were used) were cultured in a serum-free Neurobasal™ media. Parallel cell culture wells with and without a polylysine-coated glass coverslip were each applied with 2 ml of medium and replaced every 2 days. Cells were harvested by gentle pipette blow, with cell pellets obtained by centrifugation, and with the supernatants used for immunoblotting assays. The coverslips were recovered and fixed by immersion in 2% paraformaldehyde for immunocytochemistry.

Small interfering RNA (siRNA) silencing

Lipofectamine-mediated transfection of ACTA2 siRNA (product #944, Invitrogen, Thermo Fisher Scientific Inc., USA) was carried out in SH-SY5Y cells, with the control cell groups transfected with a negative control siRNA and left untreated (blank group), respectively. A volume of 500 μl serum-free Opti-MEM Medium was transferred into two 1.5 ml tubes, with 80 pmol siRNA duplex (ACTA2 siRNA and negative control siRNA) and 12 μl Lipofectamine™ RNAiMAX added, respectively. The two tubes were mixed, then incubated for 15 min at room temperature. Stocked SH-SY5Y cells in complete growth medium were diluted to 4 × 105 cells/ml. The above siRNA duplex/RNAiMAX solution and 1 ml of the diluted cells were placed into a culture well and incubated for 24 h at 37 °C in a CO2 incubator to allow transinfection. The concentration of yielded cells was measured, with 3 × 105 cells/well seeded into 24 well-plates. Cells in the ACTA siRNA, negative control siRNA and blank control groups were cultured for 8–36 h (in pilot and formal experiments) in a quadrilateral manner in each experiment, with a 14 mm glass coverslip placed in one of the 4 wells in each cell treatment group. Western blot and immunocytochemical data from cells cultured for 16–20 h were presented in this study.

Immunohistochemistry, immunocytochemistry and immunofluorescence

Cryostat sections were immunohistochemically stained in a batch-processing manner, including the following experimental settings: (1) temporal lobe sections from different postmortem brains; (2) sections from different locations of the whole brain; (3) sections across the entire longitudinal axis of the hippocampus; (4) temporal lobe sections from different animal species and human; (5) temporal lobe sections from fetal, infant and youth human brains. The immunolabeling was performed using the avidin–biotin complex (ABC) method. Selected batches of sections were treated first in PBS with 5% normal horse serum, 5% H2O2 and 0.1% Triton-100 for 1 h at room temperature to block nonspecific reactivity. The sections were then reacted with a primary antibody (Supplemental Table 2) at 4 °C overnight. On the second day, the sections were reacted with the biotinylated pan-specific secondary antibody for 1 h, and further with the ABC reagent for another hr at room temperature. Immunoreaction product was visualized by reaction with 0.05% diaminobenzidine (DAB) and 0.003% H2O2. A few sections were included in each experiment through all steps but the primary antibody incubation to define background labeling. Immunolabeled sections were mounted on glass slides, allowed to air-dry, counterstained with hematoxylin in some cases, and coverslippered. For immunocytochemistry, cells grown on glass coverslips were fixed with 2% paraformaldehyde, rinsed with PBS for a few times, and stained on-slide under identical conditions following the same steps as described above.

Double immunofluorescence was carried out on-slide for human brain paraffin sections or cultured cells fixed on coverslips. The paraffin sections were dewaxed and rehydrated, then treated with 1 mM EDTA at 95 °C for 10 min. The glass slides were applied with drops of PBS containing 5% normal donkey serum and 0.1% Triton-100 and incubated for 1 h at room temperature. They were further reacted in PBS containing a pair of primary antibodies from different host species. Immunofluorescence was visualized with Alexa Fluor® 488 and 594 donkey anti-mouse, rabbit or goat secondary antibody. A brief Sudan black (0.1%) treatment was applied to the human brain sections to block autofluorescence. All the microscopic slides or coverslips were mounted with an anti-fading medium containing the nuclear dye 4’,6-diamidino-2-phenylindole (DAPI).

Western blot

Brain and basal artery samples and pellets of cultured cells were homogenized in a BeadBlaster microtube homogenizer in radioimmuno-precipitation (RIPA) buffer containing a cocktail of proteinase inhibitors. The lysates were centrifuged at 16,000×g for 15 min at 4 °C, with supernatants collected and assayed for protein concentration. Following optimization tests, tissue or cell lysates containing proper amounts of protein were run in SDS-PAGE gels prepared at desired concentrations (based on the molecular weight of the proteins to be blotted). Separated protein products were electrotransferred onto Trans-Blot pure nitrocellulose membranes. Membranes were immunoblotted or reblotted for αSMA, selected reference proteins and internal loading control markers. The bound proteins were visualized using horseradish peroxidase (HRP)-conjugated secondary antibodies, with the signal visualized using the Pierce™ ECL-Plus Western Blotting Substrate detection kit, and image-documented in a ChemStudio/PLUS device with multiple exposure setting.

Image acquisition

Immunolabeled bright-field sections were scan-imaged using the 20× objective on a Motic-Olympus microscope equipped with an automated stage and autofocusing image capture system (Motic China Group Co. Ltd., Wuhan, Hubei, China). Immunolabeled culture cells were scan-imaged at 40× with a Keyence imaging system. Immunofluorescent brain sections or cells were also scan-imaged with the Keyence system using the 40× magnification objective, with a Z-stack setting of 1 µm scanning depth. Images were examined on-screen on the interface of the Motic Digital Slide Assistant System Lite 2.0 or the BZ-X800 Analyzer, with the areas of interest extracted for figure preparation.

Optic densitometry

Microscopic images were first saved as TIFF file in grey-scale mode, with optic density (o.d.) measured using the OptiQuant software as described our previous studies52,53. Areas of interest were selected using the interconnecting (brain regions) or the rectangle (immunoblot) tool. Total o.d., expressed as digital light units per square millimeter (DLU/mm2), was measured over a given selected area, with specific o.d. calculated using a threshold cutoff. The cutoff o.d. was obtained from the sections parallelly processed during immunohistochemistry excluding the incubation of the primary antibody, or from the white matter area of same immunostained section, or outside the section area. For densitometry of immunoblotting data, the threshold cutoff o.d. was obtained from the blank area in the membrane image.

Data processing, statistical analysis and figure preparation

Microscopical and immunoblot data were processed in Excel spreadsheets. This included the organization of data into comparing groups and calculation of relative values, means and standard deviations. The processed data were imported into GraphPad spreadsheets (GraphPad Prism 10) for graph preparation and statistical analysis. One-way analysis of variance (ANOVA) along with Bonferroni multiple comparison posthoc tests, or two-tailed paired t-test, were carried out, with P < 0.5 set as the cutoff for significant overall or intergroup difference. Figures were prepared with Photoshop 2024 by assembling representative micrographs and graphs generated through data analyses. Brightness and contrast adjustments were applied to the final image file for proper visibility.

Results

ɑSMA antibodies distinctly labeled mossy fibers in human hippocampus

In an attempt to study vascular pathology, we noticed a novel MF labeling by a goat anti-αSMA antibody in the human hippocampus, which existed in addition to the expected smooth muscle cell labeling in vascular profiles, while this mossy fiber labeling was not detectable in rat brain (Fig. 1A and Suppl. Fig. 1). A rabbit anti-αSMA antibody was further used, which exhibited the same MF labeling in the human hippocampus (Fig. 1B). This apparent axonal labeling delineated the distribution of MF along the anterioposterior axis of the hippocampus (Fig. 1C).

Fig. 1
figure 1

Immunolabeling of α-smooth muscle actin (αSMA) in hippocampal mossy fibers (MF) with densitometry relative to age. Antibody information, counterstain, ages of the brain donors, scale bar and quantitative methodology are included in the figure. (A) Concurrent ɑSMA immunolabeling by the goat antibody in human temporal lobe and adult rat brain sections, with enlarged panels illustrating the antibody labeling in human (A1) but not rat (A2, A3) hippocampal MF. (B) Mossy fiber ɑSMA immunolabeling by a rabbit antibody, enlarged panels showing simultaneous labeling in vascular smooth muscle cells (B1), the transition pattern of MF labeling at the border of CA3/CA2 (B2), and a lack of fiber-like labeling in the prosubiculum (B3). (C) Distribution of ɑSMA labeled MF across the anteroposterior axis of the hippocampal formation, with sections passing the head, body and tail of the hippocampus indicated. (D) Densitometry of αSMA labeling in the MF field (hilus and CA3) relative to other temporal lobe areas. The specific optic density (o.d.) in the areas of interest are calculated using a threshold o.d. obtained from the white matter region or from an adjacent section immunostained by excluding the primary antibody. (E, F) Graphic and statistical analysis of data from 28 postmortem brains, donor’s ages 1–76 years. *: Significantly different according to posthoc test. Additional abbreviations: FG: fusiform gyrus; PHG: parahippocampal gyrus; Ent: entorhinal cortex; Sub: subiculum, Pro-S: prosubiculum, Pre-S: presubiculum; Para-S: parasubiculum; ML: molecular layer; GCL: granule cell layer; s.o.: stratum oriens; s.p.: stratum pyramidale; s.r.: stratum radiatum; s.l.m.: stratum lacunosum-moleculare; mf: mossy fiber; WM: white matter.

Full size image

To determine if the MF labeling of αSMA in the human hippocampus was a site-specific decoration throughout the entire brain, sections covering multiple cerebral and subcortical regions from four adults were stained and examined. Representative images covering major anatomical regions from a 60 year-old donor are provided as supplementary data. Thus, except for the hippocampal MF, the overall αSMA immunolabeling was microscopically evident in vascular profiles across major neuroanatomical structures (Suppl. Figs. 2–7). A faint neuropil-like labeling was visible in the caudate and hypothalamic subregions (Suppl. Figs. 4 and 5).

We additionally assessed αSMA immunolabeling in peripheral tissues as a measure to cross-validate the antibody specificity (Suppl. Fig. 8). In the heart, the immunolabeling was clearly located in the wall of small and larger blood vessels, with the cardiac myocytes exhibited a background-like reactivity (Suppl. Fig. 8A, A1, A2). In the liver, the vascular labeling was evident in the smooth muscle cells of the portal vein and hepatic artery (Suppl. Fig. 8B,B1,B2). This vascular small muscle fiber labeling was also mostly prominent in the spleen (Suppl. Fig. 8C,C1,C2), lung (Suppl. Fig. 8D,D1,D2), kidney (Suppl. Fig. 8E,E1,E2), skeletal muscle (Suppl. Fig. 8F,F1,F2), thyroid gland (Suppl. Fig. 8G,G1,G2) and small intestine (Suppl. Fig. 8H,H1,H2). The skeletal myocytes per se exhibited background reactivity, whereas the intestinal smooth muscle cells showed labeling intensity comparable to those in the blood vessels (Suppl. Fig. 8F,F1,F2;H,H1,H2). It should be noted that there existed dense labeling in the capsule of the thyroid glands, which should be related to fibroblasts (Suppl. Fig. 8G,G1,G2).

We used 28 postmortem human brains from donors died at ages from 1 to 76 year and without β-amyloid deposition, tauopathy and alpha-synuclein pathology to explore whether there existed microscopically overt age-related difference in the MF ɑSMA labeling. This MF labeling pattern was essentially comparable among all the brains examined (Suppl. Fig. 9). Densitometry was conducted using immunolabeled temporal lobe sections from these brains. It should be noted that we used the term “hilus” to define the area between the granule cell layer (GCL) and the pyramidal cell layer (stratum pyramidale, s.p.) of CA3; this area is also described as CA4 by many authors40. Using the background optic density (o.d.) obtained from sections immunolabeled by omitting the primary antibody reaction as a cutoff (Fig. 1D), the specific αSMA o.d. (mean ± S.D., same format below) were 1476.6 ± 505.4, 1482.6 ± 511.0, 637.3 ± 297.7, 643.2 ± 306.4, 549.4 ± 305.8 and 557.4 ± 348.6 digital light unit (DLU)/mm2 in the hilus, CA3, CA1, subicular subregions (hereafter, subiculum or Sub), and the neocortex of the parahippocampal gyrus (PHG) and fusiform gyrus (FG), respectively. There was an overall significant difference among the means (P < 0.0001, F = 38.5, df = 5, 162), with posthoc test indicating statistically significant differences of the means in the hilus and CA3 relative to other areas (Fig. 1E). Because immunolabeled vascular profiles (including smooth myocytes and pericytes) would contribute to the total αSMA o.d. in a given grey or white mater region, we also used the o.d. obtained in the white matter (WM) of the temporal lobe as a cutoff for calculating the specific o.d. in the sampling areas. A similar trend was observed among the set of regional densitometric data (Fig. 1F, values not repeated here). Thus, in both data processing conditions, the means of specific o.d. in CA1, Sub, PHG and FG did not show difference in posthoc tests (Fig. 1E,F), indicating that the overall density contributed by labeled vascular profiles was comparable between regions. It should be noted that there was no statistically significant difference between the means of specific o.d. in the hilus and CA3 area (Fig. 1E,F).

Double immunofluorescence was carried out in paraffin sections to verify the localization of αSMA labeling to the MF terminals in addition to vascular cells (Fig. 2). β-Secretase 1 (BACE1) is enriched in the hippocampal MF in mammalian brains54,55, and used as a reference marker in the present study. There was a complete colocalization of αSMA (goat antibody) and BACE1 immunofluorescence at the MF terminals (Fig. 2A1–A3). Sortilin is enriched in the somata and dendrites of hilar mossy cells and CA3 pyramidal neurons52,56,57, therefore used as a reference marker for the characteristic complex large spines on these neurons. αSMA labeled large MF terminals and sortilin labeled thorny excrescences were aligned in close proximity along the dendritic processes of hilar mossy cells and CA3 pyramidal neurons (Fig. 2B1–B3). In addition, αSMA labeling of smooth muscle cells in the tunica media (TM) was confirmed in the basal artery sections (Fig. 2C). The internal elastic lamina (IEL) showed autofluorescence through the blue and green emission channels. It should be noted that, by closer examination across the microscopic images, αSMA immunofluorescence was observed as small nucleated cellular profiles at capillary-like as well as small vascular profiles, representing pericytes (Fig. 2A3,B3)24,25.

Fig. 2
figure 2

Double immunofluorescent characterization of αSMA labeling in the human hippocampal mossy fibers and vascular cells. Information about antibody combination, fluorescent indicators, image panel arrangement and scale bars are provided. A complete colocalization of ɑSMA and β-secretase 1 (BACE1) labeling is seen at the mossy fiber terminals (mf) (A, A1, A2, A3). Note the unlabeled mossy cells (*) in the hilus. (B, B1, B2, B3) αSMA labeled MF terminals occur in close proximity to the thorny excrescences of sortilin labeled hilar mossy cells (*) and CA3 pyramidal neurons. (A3, B3) αSMA immunolabeling of the nucleated pericytes at capillaries and small vascular profiles. (C) Typical ɑSMA labeling in the smooth muscle cells in the tunica media (TM) in a section of the basal artery. The internal elastic laminae (IEL) exhibits autofluorescence through the blue and green fluorescence channels. Other abbreviations are as defined in Fig. 1.

Full size image

αSMA labeling in hippocampal mossy fibers exhibited an evolutionary trend among mammals

As mentioned earlier, we initially noticed a lack of αSMA labeling of the MF in the rat hippocampus (Suppl. Fig. 1). This led us to wonder whether αSMA expression in the hippocampal MF represents a human-specific or mammalian evolution phenomenon. To address this issue, temporal lobe sections from adult mice, rats, guinea pigs, domestic cats and rhesus monkeys were immunolabeled for αSMA by including human temporal lobe sections as positive assay control. Adjacent sections were stained for BACE1 as a reference marker. The two αSMA antibodies exhibited the same labeling pattern, therefore we present the set of images of the labeling with the rabbit antibody for illustration (Fig. 3; Suppl. Figs. 10, 11).

Fig. 3
figure 3

Evolutionary trend of αSMA labeling in hippocampal mossy fibers among mammals. Shown are low power and enlarged images of αSMA immunolabeling (rabbit antibody) with hematoxylin counterstain in temporal lobe sections from the species as indicated, along with BACE1 labeling in adjacent sections for comparison. αSMA labeling of the MF is not recognizable in mouse (A) and in rat as illustrated in Fig. 1A. However, lightly stained MF terminals (pointed by arrowheads) are seen in the CA3 area but not in the dentate hilus in guinea pig (C) and cat (E). In monkey (G) and human (I) temporal lobe sections, the MF terminals in both CA3 and the hilus are clearly labeled by the ɑSMA antibody. In comparison, BACE1 labeled MF are present in CA3 and the hilus in all species (B, D, F, H, J). Small and large blood vessels (pointed by arrows) are labeled by the αSMA antibody in all species. The quantitative data of aSMA labeling densities from the species are illustrated as panels (A1, C1, E1, G1 and I1), with the specific optic density (o.d.) calculated by subtracting the background cutoff measured outside the section covered area in the same image. Scale bars are indicated in the panels. Abbreviations are as defined in Fig. 1.

Full size image

Overall, the patterns of ɑSMA and BACE1 immunolabeling were consistent between all the stained sections from different brains of the same species. Thus, in mice and rats, αSMA labeling in the DG and CA3 was comparable to other cortical and hippocampal regions (Figs. 1A, A2, A3; 3A; Suppl. Fig. 10A–C, E–G). BACE1 immunolabeling was distinct in the MF field in the hilus and CA3 and exhibited a neuropil pattern over the hippocampal and cortical cellular layers (Fig. 3B). In guinea pigs and cats, light αSMA labeling was seen at the MF terminals (pointed by arrowheads) in the CA3 area, but not in the dentate hilus (Fig. 3C, E; Suppl. Fig. 10I–K, M–O). In comparison, BACE1 labeling marked the MF distribution in both the hilus and CA3 (Fig. 3D,F). In rhesus monkeys, the αSMA and BACE1 antibodies visualized the MF distribution in the hilus as well as CA3 (Fig. 3G,H and Suppl. Fig. 10Q–S), comparable to the patterns seen in humans (Fig. 3I,J). αSMA labeled vascular profiles were present in the sections from all species (Fig. 3, enlarged views, pointed by arrows; and Suppl. Fig. 10D, H, L, P, T). It should be noted that in the brain sections of newborn rats (animals used for primary neuronal culture studies), there existed a high background-like αSMA labeling across the cortical, hippocampal and subcortical regions. The vascular profiles were also labeled, whereas no MF labeling could be recognized (Suppl. Fig. 11).

Densitometry for the MF labeling of ɑSMA in CA3 and hilus was carried out using sections passing the dorsal or middle hippocampal levels from the examined mammalian species in adult age range, with the o.d. obtained from the CA1 area used as a reference (its mean defined as 100%). Thus, the relative ɑSMA o.d. levels in the CA3 MF field were 113.9 ± 18.3% (mean ± S.D.) in mice (n = 6), 107.8 ± 8.1% in rats (n = 6), 152.5 ± 32.0% in guinea pigs (n = 4), 153.5 ± 17.7% in cats (n = 4), 276.9 ± 111.2% in monkeys (n = 4) and 352.2 ± 34.0% in humans (n = 6). The was an overall statistically significant difference among the means [P < 0.0001, F = 26.9(5, 24), ANOVA)], with difference existing for the levels of monkeys and humans relative to the other species by posthoc tests (Fig. 3K). The means of ɑSMA labeling in the dentate hilus were 101.9 ± 25.1%, 101.5 ± 13.3%, 85.3 ± 26.4%, 112.3 ± 14.6%, 319.8 ± 66.3% and 349.4 ± 33.8% among the above species in the same listing order, with an overall difference among the groups [P < 0.0001, F = 70.7(5, 24), ANOVA)], for the monkey and human groups relative to other mammals (Fig. 3L).

Mossy fiber and neuronal αSMA labeling differentiated during human brain development

To understand αSMA expression in human hippocampal MF from a developmental perspective, we assessed αSMA immunolabeling in temporal lobe sections from brains of fetuses (24–38 gestational weeks, GW, n = 8), infants (28 days to 1 year-old, n = 7), and youths and young adults (8–22 years-old, n = 11) (Supplemental Table 1), using BACE1 and zinc transport-3 (ZnT3) as reference markers. αSMA labeled vascular profiles were observed in the sections of all brains, however, development related differences were seen in the immunolabeling of neuronal profiles, as detailed below.

The subregions of the hippocampus proper and DG could be well recognized in the temporal lobe sections from the fetal human brain at 30 GW. Overall, MF αSMA immunolabeling appeared to differentiate from an overall expression of this protein in neuronal somata in the fetal brains. Thus, in the 28 GW (not shown), 30 GW (Fig. 4A,B) and 33 GW (not shown) brains, αSMA labeled somata occurred across the cortical layers II-IV (more densely packed in its superficial part), the subicular and hippocampal pyramidal cell layers, and the dentate granule cell layer and hilus. However, little MF labeling could be identified. In the 36 GW (Fig. 4C,D) and 38 GW (not shown) cases, light MF αSMA labeling occurred in CA3 and DG in mix with cellular labeling. In the 28 days-old infant brain (Fig. 4E,F), αSMA labeling occurred more clearly in the MF, with an overall reduction in the labeling of somal profiles in the cortex and hippocampus. In the infant cases (6 month to 1 year of age), there was a further decrease in the overall αSMA reactivity across the cortex and hippocampal cellular layers, whereas the MF labeling in CA3 and hilus became highlighted (Fig. 4G,H). In the sections from the youth cases, heavy αSMA labeling occurred in the MF, whereas little labeling of neuronal somata was seen in the cortex and hippocampal formation (Fig. 4I,J). It should be noted that the two ɑSMA antibodies exhibited somewhat differential labeling appearance especially involving the somal profiles, such that the rabbit antibody labeling occurred largely in the cell nuclei while the goat antibody labeled the cytoplasm, especially obvious in the prenatal brains (Fig. 4A–D).

Fig. 4
figure 4

Developmental trend of neuronal and hippocampal mossy fiber αSMA immunolabeling in human brains. The left and right panel groups show low magnification images and enlarged fields of temporal lobe sections immunolabeled by the rabbit and goat antibodies to αSMA. Gestational weeks (GW) and ages of brain donors are indicated on the left. αSMA immunolabeling occurs in the cortical grey matter, heavier in layers II/III relative to IV-VI, and occurs in the hippocampal cellular layers in the 30GW, 36 GW cases (A, B, C, D), with labeled cellular profiles seen in the enlarged panels (A1, A2, B1, B2, C1, C2, D1, D2). This cellular labeling is reduced in the newborn case (E1, E2, F1, F2), sparsely seen in the infant case (G1, G2, H1, H2), and not present in the youth case (I1, I2, J1, J2). αSMA labeling of the MF emerges in the dentate gyrus and CA3 in mix with local cellular labeling in the fetal and newborn brains and appears as heavier and distinct terminal labeling against little cellular labeling in the infant and youth cases. The two antibodies exhibit noticeable differences regarding the cellular labeling, with the goat antibody having a high background reactivity over the sections. Quantitative values (mean ± S.D. %) from the 28 GW-28 days (d), 6 months (m) to 1 year (y) and 8–22 y groups, respectively, are the following: 100 ± 29.7, 61.2 ± 29.4 and 47.4 ± 15.5 for rabbit antibody labeling in the temporal neocortex; 59.7 ± 16.0, 70.1 ± 15.9 and 100 ± 19.9 for rabbit antibody labeling in MF field; 100 ± 17.3, 59.6 ± 9.9 and 59.0 ± 29.3 for goat antibody labeling in the temporal neocortex; 85.6 ± 13.4, 76.7 ± 5.9 and 100 ± 12.6 for goat antibody labeling in MF field. Statistics are as indicated in the graphs. Abbreviations are as defined in Fig. 1.

Full size image

In order to express the developmental trend of MF and cellular labeling quantitatively, densitometry was carried out over layers II-IV in the temporal neocortex and CA3/hilar MF filed. We grouped the 28 days-old (newborn) along with the late gestational cases together considering a much similar ɑSMA cellular labeling pattern (Fig. 4A–F). We used the mean o.d. (defined as 100%) from the prenatal/newborn group for normalization of the temporal cortical data and used the mean o.d. (100%) from the youth group for normalization of the MF labeling intensities (Fig. 4K,L). As detailed in the figure legend, there existed a decline in the ɑSMA labeling in the temporal neocortex and a trend of increase of labeling in the MF field while comparing the prenatal/newborn to infant and youth groups. However, the cellular labeling by the two antibodies affected density readouts and statistical results of the MF ɑSMA labeling in the prenatal/newborn relative to infant and youth groups. Nonetheless, the MF labeling intensity was significantly higher in the youth than the infant group (Fig. 4K,L).

To further understand MF development in human hippocampus, we examined BACE1 and ZnT3 immunolabeling in the temporal lobe sections (Fig. 5). Thus, in the 30 GW and 33 GW brains, BACE1 labeling appeared slightly stronger in the Sub, hippocampus and DG relative to the temporal cortex (Fig. 5A,B). BACE1 labeling of the MF appeared to differentiate out in the 36 GW and 28 day-old brains (Fig. 5C,D,D1), and became adult-like in the infant and youth brains (Fig. 5E–H,E1, also see Fig. 3J). ZnT3 immunolabeling was largely background-like across the temporal lobe in the 30 GW and 33 GW brains (Fig. 5I,J), with a weak MF labeling seen in the 36 GW brain (Fig. 5K). In comparison, ZnT3 labeling was apparently increased in the 28 day-old and youth brains across the temporal lobe regions, with the MF labeling being clearly visualized (Fig. 5L–P), including a dense band along the dentate inner molecular layer (iML) (Fig. 5L1,N1). We carried out densitometric analysis for BACE1 and ZnT3 immunolabeling in the temporal neocortex (layers II-IV) and in the CA3 and hilus areas together, with the density readouts normalized to the means of MF values from the youth group. Overall, there was a trend of increase in BACE1 and ZnT3 labeling in the neocortex with brain development. The MF labeling by BACE1 and ZnT3 did not show statistically significant differences between the infants and youths (Fig. 5Q,R).

Fig. 5
figure 5

Immunohistochemical labeling of β-secretase 1 (BACE) and zinc transporter 3 (ZnT3) in temporal lobe sections from prenatal, infant and youth human brains. Image panels show the developmental pattern of BACE1 (A, B, C, D, D1, E, F, G, H) and ZnT3 (I, J, K, L, M, N, O, P) immunolabeling among the temporal lobe sections at low magnification, with enlarged panels (D1, E1, L1, N1) show the details of MF labeling in the dentate gyrus and CA3. BACE1 labeling appears to increase progressively from the prenatal to infant cases, while ZnT3 labeling appears to increase mainly after birth. A clearly ZnT3 labeled band is located in the inner molecular layer (iML) of the dentate gyrus (L1, N1). Quantitative values (mean ± S.D. %) from the 28 GW-28 days (d), 6 months (m) to 1 year (y) and 8–22 y groups, respectively, are the following: 44.3 ± 9.6, 52.6 ± 16.0 and 82.4 ± 8.5 for BACE1 labeling in the temporal neocortex; 80.4 ± 17.9, 96.3 ± 3.6 and 100 ± 10.2 for BACE1 labeling in the MF field; 39.3 ± 23.2, 92.1 ± 13.9 and 92.7 ± 7.8 for ZnT3 labeling in the temporal neocortex; 45.2 ± 20.4, 93.7 ± 20.4 and 100 ± 8.1 for ZnT3 labeling in the MF field. Statistical reports are included in the graphs. Additional abbreviations are as defined in main Fig. 1.

Full size image

αSMA immunolabeling was detected in human and rat primary neuronal cultures

Given that ɑSMA immunolabeling occurred widely in neuronal somata in the prenatal human brains, in vitro experiments were carried out to explore ɑSMA expression in developing neurons. We initially screened a number of cell lines (Supplemental Table 2) with western blot and detected an endogenous ɑSMA expression in human and rat originated neuronal cell lines including SY-SH5Y, HT22 and R28 cells (data not shown). We further used double immunofluorescence to observe αSMA immunolabeling in primary human and rat neuronal cultures. ɑSMA was found to colocalize with neuronal markers including microtubule-associated protein 2 (MAP2), β-tubulin or neuronal nuclei antigen (NeuN) in neuronal somata and processes (Fig. 6A–D and Suppl. Figs. 12, 13). However, other types of cells were present in the primary cultures, including fibroblasts, astrocytes, oligodendrocytes, microglia and macrophages based on the co-labeling of ɑSMA and corresponding cell linage markers. Specifically, the large-sized fibroblasts exhibited the brightest αSMA immunofluorescence relative to other types of cells including neurons (Suppl Figs. 13 and 14), which was consistent with established data18,19,20,21,22. A fairly long culture time (7–14 days) was needed to observe the differentiation of the primary neurons, which was not optimal for siRNA interfering experimental study with these preparations.

Fig. 6
figure 6

Immunofluorescent characterization of αSMA expression in cultured human and rat primary cortical neurons. Shown are double immunofluorescence for αSMA and neuronal markers with DAPI counterstain in morphologically differentiated neurons fixed following 10 and 5 days in vitro (DIV). αSMA immunofluorescence is colocalized with that of microtubule associated proteins 2 (MAP2) (A, C), β-tubulin (B) and neuron-specific nuclear antigen (NeuN) (D) in the somata and processes of the primary neurons. Scale bars are indicated in the left panels.

Full size image

ɑSMA protein level was higher in the dentate gyrus relative to other human brain areas

Actins including β-actin and F-actin are abundantly expressed in the brain including in neurons58,59,60. To validate the specific labeling of the αSMA isoform from ACTA2 transcription by the antibodies, we carried out siRNA inferring experiments in cultured neurons. The ACTA2 siRNA #944 (Invitrogen) was identified to effectively inhibit αSMA protein expression in SY-SH5Y cells up to 48 h following transinfection in pilot experiments, with the levels of β-actin unaffected (see Supplemental original western blot image data). The cells grown to a moderate density (30%-50% confluent) by 16–20 h after seeding; this time point was therefore chosen to examine the gene silencing effect. Three separate experiments were carried out for immunoblotting and immunocytochemical study. Levels (mean ± SD) of αSMA protein were 6.5 ± 7.9% (mean ± SD, % of β-tubulin, same below), 105.9 ± 10.0% and 101.1 ± 12.2% in the ACTA2 siRNA treated, untreated (blank) and negative control siRNA treated SY-SH5Y cell groups, respectively, with a significant difference in the means (P < 0.0001, F = 90.6, One-way ANOVA), which existed in the silencing group relative to the other two by posthoc tests (Fig. 7A,B). No difference was detected for the protein levels of F-actin (86.8 ± 8.5%, 90.9 ± 22.6% and 82.8 ± 3.7%, same group order as the above αSMA data, P = 0.78, F = 0.25), and MAP2 (133.0 ± 4.7%, 132.7 ± 23.2% and 132.5 ± 8.4%, P = 0.99, F = 0.00) among the cell groups (Fig. 7A,B). A reduced αSMA immunolabeling was also visible in the ACTA2 siRNA treated relative to untreated and negative control cells (Suppl. Fig. 14).

Fig. 7
figure 7

Verification of ɑSMA antibody specificity with ACTA2 gene silencing method and assessment of ɑSMA protein levels in human dentate gyrus relative to control regions. (A) Western blot detection of αSMA, β-tubulin, F-actin and microtubule associated protein 2 (MAP2) in lysates from a representative set of cultured SY-SH5Y cells in the untreated (blank) (−/−), ACTA2 siRNA treated (+/−) and negative control siRNA treated (−/+) groups. The αSMA signal is greatly reduced in the ACTA2 siRNA treated cells, whereas that of the β-tubulin, F-actin and MAP2 proteins are comparable between the cell groups. (B) Quantitation from 3 separate culture experiments. (C) Microdissection of the dentate gyrus (DG), temporal neocortical (TC), frontal neocortical (FC) and basal artery (BA) samples from postmortem human brain. (D) Representative membrane images showing the optimization of the loading amounts of human frontal cortical and BA extracts to blot αSMA protein. (E, F) Representative immunoblot images showing the detection of αSMA relative to reference proteins in the regional brain and BA samples., with the amount of sample loading indicated. Note the increased signal in DG relative to FC, TC and cerebellar cortical (CBL) lysates (E, G), and similar levels of collagen IV and neuron-specific nuclear antigen (NeuN) in these regional samples (F). (H, I) Immunoblot images and quantification showing a higher αSMA protein level in the DG than TC lysates from an additional set of brains (n = 4) blotted with the rabbit and goat αSMA antibodies, respectively. Also note that the levels of GAPDH are reduced or become undetectable in the BA extracts with the dilution of sample loading (D, E, F, H). *: Significantly different per posthoc test.

Full size image

For immunoblotting human brain tissue lysates, initial experiments were carried out to determine the proper amount of sample loading to blot αSMA and reference proteins, using extracts from the cerebral and cerebellar cortical regions, hippocampal subregions, and basal artery (BA) as assay control (Fig. 7C,D). High amount of brain sample loading (> 50 µg) and diluted BA lysate loading (< 1 µg) were optimal for blotting αSMA protein (7D). In this setting, the levels of αSMA blotted by the rabbit antibody were higher in the DG than the FC, TC and CBL lysates (Fig. 7E,G). The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was not detectable in the diluted BA lysates (Fig. 7E). Levels of the vascular basal membrane protein collagen IV (blotted with higher sample loading) were lower in the brain than in the BA lysates (Fig. 7F). The neuron-specific nuclear antigen (NeuN) was detected in the brain regional samples with comparable levels, but not in the BA lysates (Fig. 7F). Using an additional set of samples, we confirmed a higher αSMA protein level (mean ± SD, % of GAPDH o.d.) in the DG than TC lysates as blotted by the rabbit antibody (232.1 ± 47.9% vs. 102.8 ± 15.4%, P = 0.023, paired two-tailed t-test) and by the goat antibody (82.9 ± 21.3% vs. 20.2 ± 9.7% of GAPDH o.d., P = 0.016, paired two-tailed t-test) (Fig. 7H,I).

Discussion

The immunohistochemical and immunofluorescent data obtained in the current study clearly showed the ɑSMA expression in smooth muscle cells among relatively large blood vessels and also in the pericytes around capillaries and small vasculature in the brain. ɑSMA immunolabeling in various peripheral organs was localized to smooth muscle cells in the blood vessels, and also in the smooth muscle cells in the gastrointestinal wall. ɑSMA labeling was further observed in fibroblasts, especially evident in the capsule of the glandulous organs or tissues. Together, these findings are consistent with the established cellular expression pattern of this actin isoform in bodily tissues7,8,18,19,20,21,22,24,25. We carried out siRNA silencing experiments, which confirmed the specific labeling of ɑSMA by the antibodies in a manner dependent on ACTA2 transcription.

Thus, the present study reveals a previous underrecognized ɑSMA expression in neuronal profiles. To the best of our knowledge, limited literature is available regarding neuronal expression of αSMA. Several previous studies reported αSMA expression in cultured rodent embryonic brain and neural crest cells, and in neuroblastoma cells61,62,63,64,65, although these observations were discussed largely in the context of lineage differentiation of neural progenitors to smooth muscle cells. In the current study, we found a transient αSMA expression in cerebral cortical and hippocampal neurons in prenatal human brains, which remained detectable until early infancy. We also found αSMA expression in cultured human and rat primary neurons as well as neuronal cell lines. Therefore, our findings indicate that αSMA is expressed constitutively in developing neurons in vivo and in vitro.

The hippocampal formation is greatly expanded from rodents to primates, and in human relative to nonhuman primates62,66. This structure plays a fundamental role in memory formation42,66,67,68,69, but is increasingly involved in complex cognitive functions with mammalian evolution45,70,71. The giant MF synapse lies at the center of hippocampal network, which is considered structurally and functionally the most sophisticated synapse in the nervous system. While this unique synapse has fascinated generations of neuroscientists including Ramón y Cajal, much of our knowledge regarding its anatomical, molecular and electrophysiological properties has been learnt from studies in lower mammalian species especially rodents38,72. Recent investigations in nonhuman primates and humans revealed substantial differences in the anatomical and physiological properties of the hippocampal trisynaptic system relative to rodents73,74.

The expression of αSMA in human hippocampal MF observed in the current study is remarkable. First, it occurs distinctively across the entire brain from early childhood to elderly ages, selectively highlighting the giant synapse. Second, there exists an evolutionary pattern from rodents to nonhuman primate and human. Thus, MF αSMA labeling is essentially undetectable in mice and rats, while a light labeling is seen in CA3 in guinea pigs and cats. In comparison, MF αSMA immunolabeling is distributed in both the CA3 and hilus fields in rhesus monkeys, comparable to the human pattern. Although further studies are needed to detail the developmental pattern of MF αSMA among mammals, it appears that mice and rats do not exhibit a developmentally regulated MF expression/differentiation of αSMA, while an overall neuronal expression of this actin appears to exist in newborn rat brains. It is notable that the evolutionary trend of MF αSMA expression is unique to ɑSMA, because BACE1 labeled MF terminals occur in the hilus and CA3 in all the above mammals. Third, MF αSMA labeling in human hippocampus emerges during late gestation and appears to establish an adult-like pattern by the first year of age. In fact, our current examination on BACE1 and ZnT3 expression also suggests that the first year of life or infancy appears to be a key period for the development of mossy fiber terminals in human hippocampus. In this context, the development of the trisynaptic circuit in humans would be likely shaped by early life experience.

As a “full-blown” pattern of mossy fiber ɑSMA expression occurs in primates especially in humans, there exists a challenge to experimentally elucidate the implication of this novel actin isoform expression for the hippocampal giant synapse in perspectives of its plasticity, electrophysiology or role in cognitive functions. For instances, it remains uncertain whether site-specific in vivo manipulation of ɑSMA expression at the MF terminals is feasible, and which mammalian species is suited for such experimental investigation. Rodents are the most commonly used animal models but they lack a human-like MF ɑSMA expression, raising a doubt whether manipulating this protein expression in hippocampal formation would yield physiologically relevant outcome, not to mention that the trisynaptic circuit may convey huge different hippocampal functions between rodents and human73,74,75,76,77. Nonetheless, the anatomically advancing pattern of MF αSMA expression from rodents to human points to a certain unprecedented aspect of hippocampal evolution at molecular and synaptic levels. Recent studies have revealed novel neuroimaging findings of cerebral malformations such as dysgyria among patients with ACTA2 mutations15,17,62,66. Since neurological deficits such as temporal lobe epilepsy are reported in the patients with ɑSMA mutations15,71, it may be worthwhile to assess whether hippocampus-dependent cognitive/behavioral functions may be affected among the patients.

Notably, in the online databases (https://mouse.brain-map.org/experiment/show?id=71919092, https://mouse.brain-map.org/experiment/show?id=68743877), the Acta2 mRNA in situ hybridization map of adult mouse brain shows widespread transcript signal across neuroanatomical structures of the central nervous system, with the highest densities noticeable in the granule cell and pyramidal cell layers of the hippocampal formation, the Purkinje cell layer of cerebellar cortex and layer II of the cerebral neocortex and paleocortex. Stereo-seq data of Acta2 mRNA transcriptomics in the adult human brain also exhibit broad cellular signal across the cerebral neocortical layers, with quantified enrichment change highest in microglia, followed by vascular cells, astrocytes, oligodendrocytes and neurons (https://www.proteinatlas.org/ENSG00000107796-ACTA2/brain). These data appear to be compatible with the current findings that neuronal and nonneuronal cells in the brain could express ɑSMA protein, such as during development and in vitro. However, it appears that developmental and evolutionary processes can dramatically influence messenger and protein expression as such highly selective cellular ɑSMA localizations are finalized in the adult brain and peripheral systems. This may relate to the fundamental question regarding the discrepancy between mRNA and protein expression noted in many biological systems, whereas the regulatory mechanism thereof remains poorly understood75,76,77.

In summary, we report a novel αSMA expression in human hippocampal mossy fibers that persists lifelong from early childhood. This pattern appears to be established phylogenetically through a trend of refinement from rodents to primates and ontogenically during late gestation to the infancy period. The present study also reveals that αSMA expression is expressed in developing neurons, which manifests as a transient neuronal expression in prenatal and early postnatal human brains and can be detected in cultured neurons.