Samples

We obtained postmortem human tissue samples from the Autism Tissue Program (ATP) (predecessor to Autism BrainNet) and the NIH NeuroBioBank. Brain banks used the Autism Diagnostic Interview-Revised (ADI-R) to confirm diagnosis. Control (CT) cases were free of neurological disorders, based on medical records and information gathered at the time of death from next of kin. This study includes twenty cases with autism and twenty age- and sex-matched control cases. Of those 38 were males (n = 19 CT + 19 AU) and 2 females (n = 1 CT + 1 AU), (Table 1). One subject with autism presented with seizures (Table 1). The average age of control cases was 24.9 years (from 7 to 56 years), and of cases with autism was 25.6 years (from 6 to 56 years). There was no difference in PMI between control and autism groups (31.8 hours for CT and 24.8 for AU, p > 0.05). For causes of death see Table 1. Blocks of Brodmann areas BA9, BA46 and BA47 were isolated based on Brodmann anatomy [4]. Tissue blocks were fixed in 10% buffered formalin and cryoprotected in a 30% sucrose. Tissue was embedded in Optimum Cutting Temperature (OCT) compound and frozen. 14μm-thick slide-mounted sections were cut in a cryostat and stored at -80°C until use. Occasionally, tissue for a given case was unsuitable for analysis, including: BA9 from subject UCD13AP86; BA46 from cases 210, 4899, 4999, and AN00764; and BA47 from cases 4337, 210, 4899 and 5574; no data was excluded on the basis of age, sex, or PMI. One section per block was Nissl-stained to confirm cortical areas based on von Economo histology (Fig 1), (Hashemi et al., 2017), and adjacent sections were used for immunostaining.

Immunostaining

Slide mounted tissue sections were enzymatically immunostained with an antibody against parvalbumin. Briefly, tissue was treated with chloroform:100% ethanol at 1:1 followed by sequential immersion in 100%, 96%, 90%, 70% and 50% EtOH, then diH₂0. Antigen retrieval was performed by exposing the tissue to 110°C in 1x Diva Decloaker (Biocare Medical) for 8 min., and slides were washed (all washes consisted of TBS twice followed by TBS + 0.05% tween once). Endogenous peroxidase blocking was performed with 3% H₂O₂. The slides were blocked with tris buffered saline (TBS) + 10% normal donkey serum (NDS) + 0.3% triton for 1 hour at room temperature, and then treated with an avidin-biotin blocking kit (Vector Labs). Primary antibody solution (rabbit anti-parvalbumin antibody, Abcam, 1:500) was added to each slide for 24 hours at 4˚C with Parafilm coverslips in a dark humidified box. Slides were then washed and incubated with the secondary antibody (biotinylated donkey anti-rabbit IgG, Jackson, 1:150) for 1 hour, then washed and incubated with ABC solution (Vector Labs) for 2 hours in a dark, humidified box. After washing, slides were developed with DAB substrate (brown; Vector Labs) for 1 minute, then washed again. Tissue was dehydrated through sequential immersion in 50%, 70%, 90%, 96%, and 100% EtOH for 3 minutes each, then cleared in Xylene for 6 minutes and coverslipped with Permount mounting medium.

Antibody characterization

Anti-Parvalbumin antibody: Immunogen: Full length native protein (purified) corresponding to Rat Parvalbumin. Purified parvalbumin from rat skeletal muscle; Sequence similarities: Belongs to the parvalbumin family. Contains 2 EF-hand domains; Host: Rabbit; Isotype: IgG; Antibody type: polyclonal; Manufacturer and Catalogue#: Abcam ab11427; Concentration: 1:500; Additional information: Reacts with mouse, rat, chicken, human, and sea urchin, and predicted to work with gerbil and common marmoset; RRID: AB_298032.

Imaging, quantification, and statistics

PV immunoreactivity in Ch cell cartridges across prefrontal areas BA9, BA46 and BA47 was located most clearly in layers II/III and layer V (Fig 2). The PV staining pattern was consistent across our cases, and qualitative analysis did not reveal a noticeable difference in the Ch cartridge expression of PV between autism and control cases.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Ch cell cartridges stained with an antibody against PV in Brodmann areas BA9 (A,D,E), BA46 (B,F,G), and BA47 (C,H,I) at low and high magnification. Scale bar in A-C: 500 μm; D,F,H: 10 μm; E,G, I: 5 μm.

https://doi.org/10.1371/journal.pone.0281477.g002

We captured brightfield images using the 100x objective of an Olympus microscope (BX61). Five cartridges per layer, per Brodmann Area, per case were captured, and cartridges were chosen based on quality and homogeneity of staining, cartridge completeness, and clearly distinguishable boutons. We then measured the length of the cartridges using the ImageJ plugin NeuronJ. To quantify the number of synaptic boutons (Fig 3), each cartridge was analyzed separately by two researchers, each blinded to the diagnoses, and the two duplicate counts were then averaged together. To measure bouton size, individual boutons were first identified within a cartridge by imposing a circularity threshold, which outlined round/oval shapes within an image. Subsequently, if any boutons were missed or unclearly marked, or if any other structure was erroneously identified as a bouton, borders were manually corrected using the selection brush tool until each bouton within the cartridge was distinctly delineated. The size of each bouton was then quantified using Feret’s diameter (ImageJ).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

A. Ch cell cartridges stained with PV in control and ASD prefrontal cortex. B-C: The length of cartridges (B) and the number of boutons per μm of cartridge (C) are similar in ASD and control cases, in both supra- and infragranular layers. Scale bar in A: 2 μm.

https://doi.org/10.1371/journal.pone.0281477.g003

Data collected from the supragranular (layer III) and infragranular (layer V) layers (5 cartridges per layer) were analyzed for each area (BA9, BA46, BA47). Data for a given layer of a given case was comprised of the average values from all cartridges analyzed within that layer of that case. The goal of the statistical analysis was to compare cartridge synaptic bouton number and density, bouton size, and cartridge length between autism and control cases, and to assess the relationship between anatomical parameters and other patient/sample characteristics (such as age, PMI, and time in formalin). We performed a repeated measures mixed linear regression model analysis that accounted for the nested (diagnosis within subject) and crossed factors (counts within layers within regions) and avoids pseudo replication.

The length of Ch cell cartridges is not changed in autism

The Ch cell cartridge consists of a string of synaptic boutons connected by a cytoplasmic bridge, and each cartridge is aligned adjacent to the pyramidal cell AIS that innervates (Gallo et al. 2020). A decreased cartridge length in autism could result in fewer synaptic boutons per cartridge and thus fewer synaptic sites, potentially interfering with the Ch cell inhibitory signaling ability. However, we found no difference in Ch cell cartridge length between control and autism cases, with cartridges measuring an average of 35 μm in length across cortical layers, brain regions, and diagnostic categories. It is important to note that the length measured may represent the length of overlapping superposed cartridges innervating the same pyramidal neuron AIS. Our results suggest that while the number of Ch cells and the Ch cell cartridge density are both decreased in the prefrontal cortex in autism [3–5], the maximum length of cartridges on existing Ch cells is not affected and thus is likely not a factor contributing to the decreased GABA_A receptor subunit α2 present in the pyramidal AIS in some prefrontal cortical areas in autism [17].

The number of Ch cell synaptic boutons is not changed in autism

Boutons are axon terminal structures containing sites of synaptic junctions. They are presynaptic points of neurotransmitter storage and release to facilitate cell communication [38]. Boutons contain the active zones where synaptic vesicles dock. Vesicle exocytosis is stimulated through Ca2+ signaling, releasing signaling molecules into the synapse where they are taken up at the adjacent postsynaptic density [39]. Within the boutons, synaptic vesicles undergo cyclic trafficking comprised of exo- and endocytosis in the course of neurotransmitter packaging and release [39], the dynamics of which may be influenced by activity-dependent plasticity [40]. Cell bouton number is linked to Ch cell firing potential and its ability to regulate pyramidal neuron activity [41–43]. Approximately four Ch cells innervate each pyramidal AIS [11,16], and each Ch cell cartridge forms three to eight synapses on the AIS [11,26,35–37]. Bouton number is proportional to the number of synapses [44]. The number and distribution pattern of boutons varies by species, developmental stage, cortical region, and the functional state of cells. For example, the mean number of Ch cell boutons per pyramidal neuron AIS decreases 32% in monkey prefrontal cortex from 3 months of age to adulthood [16]. Additionally, the pattern of bouton distribution is correlated with the size of the pyramidal AIS [18,31–34,45]. In general, synaptic boutons tend to distribute homogeneously with random intervals along the axon and increase in number with axon length, independent of the distance from the soma or the length between branching nodes [44]. Bouton spread facilitates the optimal contact with the target. Along with small basket and double-bouquet cells, Ch cells have among the lowest proportion of boutons farther than 200 μm from the soma, demonstrating a highly compact bouton field, as well as a unique pattern of bouton clustering along axon cartridges [44].

We did not find a difference in the number or density of Ch cell terminal boutons per cartridge in autism vs. control cases. Our results may indicate that the synaptic input to a single pyramidal cell is conserved in autism. However, the number of Ch cells is decreased by approximately half in the prefrontal cortex in autism, suggesting an increase in the number of innervating cartridges either from one or several adjacent Ch cells to a single pyramidal neuron AIS, or that we quantified cartridges from areas of unchanged Ch cell density. As the parvalbumin immunostain does not distinguish individual cartridges, further study would be necessary to elucidate whether bouton density per individual cartridge is also unchanged, and to fully understand Ch cell innervation of pyramidal neurons. Additionally, it remains unclear whether there is complete loss of innervation to some pyramidal neurons in cases with autism, or if there remains an equivalent number of pyramidal neurons receiving input from Ch cell cartridges. If innervation is conserved in some pyramidal neurons but entirely lost in others, net inhibitory signaling could still be altered. Nevertheless, the data presented here do not provide evidence that synaptic contact between Ch cell axons and pyramidal neurons is altered in autism.

The size of Ch cell boutons is decreased in autism

Ch cell terminal bouton size and innervation pattern are linked to synaptic transmission. Though the majority of terminal boutons contain a single synaptic junction, 5% of boutons analyzed in the rat frontal cortex contained two synaptic junctions [44]; while these findings mainly pertained to boutons with dendritic targets, it is feasible that decreased bouton size in autism may diminish the potential for double synapse formation. In addition, bouton size correlates to the amount of synaptic vesicles stored. Synaptic transmission depends on the continued availability of neurotransmitter-filled synaptic vesicles for triggered release from presynaptic boutons. Small boutons contain fewer synaptic vesicles [46], so it is possible that the reduced bouton size in autism reduces the strength of the inhibitory signal transmission of Ch cells. Bouton size is linked with axo-axonic inhibitory synapse activity. Studies performing knockdown of DOCK7 in Ch cells in the cortex resulted in a disorganized network of Ch cartridges and a reduction in the number and size of boutons, while ectopic expression of DOCK7 produced the opposite phenotypes [20]. DOCK7 affects Ch cell cartridge/bouton development by modulating the activity of ErbB4 [20,47,48]. ErbB4 is expressed by axon terminals in PV+ cells. Gain- and loss-of-function experiments demonstrated that ErbB4 promotes the formation of axo-axonic inhibitory synapses on pyramidal cells [47]. Interestingly, bouton size and activity are altered in schizophrenia in ways similar to our findings in autism [20,47].