Shades of gray in human white matter

Anatomists have long expressed interest in neurons of the white matter, which is by definition supposed to be free of neurons. Hypotheses regarding their biochemical signature and physiological function are mainly derived from animal models. Here, we investigated 15 whole‐brain human postmortem specimens, including cognitively normal cases and those with pathologic Alzheimer's disease (AD). Quantitative and qualitative methods were used to investigate differences in neuronal size and density, and the relationship between neuronal processes and vasculature. Double staining was used to evaluate colocalization of neurochemicals. Two topographically distinct populations of neurons emerged: one appearing to arise from developmental subplate neurons and the other embedded within deep, subcortical white matter. Both populations appeared to be neurochemically heterogeneous, showing positive reactivity to acetylcholinesterase (AChE) [but not choline acetyltransferase (ChAT)], neuronal nuclei (NeuN), nicotinamide adenine dinucleotide phosphate‐diaphorase (NADPH‐d), microtubule‐associated protein 2 (MAP–2), somatostatin (SOM), nonphosphorylated neurofilament protein (SMI‐32), and calcium‐binding proteins calbindin‐D28K (CB), calretinin (CRT), and parvalbumin (PV). PV was more richly expressed in superficial as opposed to deep white matter neurons (WMNs); subplate neurons were also significantly larger than their deeper counterparts. NADPH‐d, a surrogate for nitric oxide synthase, allowed for the striking morphological visualization of subcortical WMNs. NADPH‐d‐positive subcortical neurons tended to embrace the outer walls of microvessels, suggesting a functional role in vasodilation. The presence of AChE positivity in these neurons, but not ChAT, suggests that they are cholinoceptive but noncholinergic. WMNs were also significantly smaller in AD compared to control cases. These observations provide a landscape for future systematic investigations.


INTRODUCTION
White matter neurons (WMNs) were first examined in 1867 by Theodor Meynert through accidental discovery (Meynert, 1867).Santiago Ramon y Cajal in 1900 studied the general morphology of what he originally called "interstitial" neurons using primarily Nissl and Golgi-stained specimens (Kostovic & Rakic, 1980).A century later, modern-day anatomists continued to carry out extensive and systematic analyses of neurons in white matter (Figure 1), which is by definition supposed to be free of neurons.Interest in these oddly placed neurons resulted in hypotheses regarding their developmental origin, biochemical signature, physiological function, and putative role in disease states, mainly derived from animal studies (Barbaresi et al., 2014;Mortazavi et al., 2016Mortazavi et al., , 2017;;Rockland & Nayyar, 2012;Suarez-Sola et al., 2009).As a result, two distinct and topographically separate populations in the human cortex have been proposed: one arising from what is speculated to be remnants of fetal subplate neurons, and the other from the subventricular/intermediate zones in the fetus-developing into subcortical neurons of the adult, embedded deep within central white matter (Judaš et al., 2010).We chose to use the terms "subplate" versus "subcortical deep" WMNs to uphold this important distinction.Although no definable boundary between "subplate" and "deep" white matter exists, several distinctive features have been ascribed to each class of WMNs.First, these classes differ in anatomic density; subplate WMNs are believed to exist in higher density, whereas fewer WMNs in the deep white matter have been described (Meyer et al., 1992).These classes also differ in appearance, where some report sublate WMNs as indistinguishable from layer VI cells of the cortex (Meyer et al., 1992), while deep subcortical WMNs show more heterogenous morphologies (Garcia-Marin et al., 2010).
Neuronal NOS is activated in nitrergic neurons through an influx of calcium ions, and nicotinamide adenine dinucleotide phosphate (NADPH) acts as a co-substrate, thus allowing these neurons to be visualized by NADPH-diaphorase (NADPH-d) histochemistry (Blottner et al., 1995).
In the rat brain as well as in the macaque, intensely stained NADPH-d+ neurons were found within the corpus callosum, and some within the ependymal region of the lateral ventricle, where they may be in contact with cerebrospinal fluid (CSF).These observations suggest the possibility that these neurons are a major source of NOS during neural activity, assisting blood vessel relaxation and monitoring composition, pH, and osmolarity changes of nearby CSF (Barbaresi et al., 2014;Rockland & Nayyar, 2012).In previous studies, we demonstrated that subcortical deep WMNs were neurochemically diverse and subpopulations exhibit reactivity to NADPH-d, acetylcholinesterase (AChE) (Smiley et al., 1998), and nonphosphorylated neurofilament (SMI-32) (Bu et al., 2003).WMNs in human tissue that are AChE rich have been labeled cholinoceptive due to the high co-labeling percentage (∼70%) of the muscarinic m-2 receptor, a postsynaptic component of the cholinergic pathway.These overlapping targets suggest cholinergic involvement in the release of NOS from WMNs (Smiley et al., 1998).
It has also been shown that AChE-positive WMNs display a significant age-related decrease in neocortex by ∼50% per region, prompting questions about their vulnerability to age-related neurodegenerative disease pathology, like Alzheimer's disease (AD), which can occur, albeit in limited density and distribution, in cognitively unimpaired older adults (Balasubramanian et al., 2012;Bu et al., 2003).
The morphological, regional, and neurochemical signatures of WMNs, particularly those in the deep white matter, have not been comprehensively characterized in the adult human brain.Additionally, the possibility of colocalization of neurochemicals has not been explored in depth.In this study, we attempted to define these features with a focus on deep WMNs in 15 whole-hemisphere human postmortem specimens, ranging from age 27 to 92, stained for distinct markers of cortical neurons and analyzed with digital methods.Qualitative observations explored the relationship between the processes of these neurons and blood vessels of the white matter.We also examined a subset of cases with AD neuropathologic change to determine if this cell population is vulnerable to disease.

Participant characteristics and demographic information
Autopsied brains from 15 participants were originally obtained from the Laboratory of M.-Marsel Mesulam, MD at Harvard University, with the exception of Cases 7-8 and 13-14, which were identified from the Northwestern University Alzheimer's Disease Research Center (ADRC) Brain Bank.Written informed consents were obtained from all participants, and the study was approved by the Northwestern University Institutional Review Board and in accordance with the Helsinki Declaration (www.wma.net/en/30publications/10policies/b3/).All cases received extensive postmortem neuropathologic analyses.Two groups of cases were included: cases with AD neuropathologic change (ADNC) as the primary neuropathologic diagnosis (AD, N = 5), and cases that were cognitively normal (CN, N = 9) based on extensive chart review.AD cases carried an antemortem clinical diagnosis of Dementia of the Alzheimer's type (DAT), with the exception of one case that presented clinically with F I G U R E 1 Original chartings and notes prepared by Marsel Mesulam, MD, a student of the late Deepak Pandya, MD to illustrate the regional distributions of NADPH-d+ (a) and MAP2+ (b) in serial whole-hemisphere sections from Case #1, a 55-year-old cognitively healthy female.Each mark represents a single neuron.The charting was performed with a microscope stage electronically coupled to an x-y plotter.This is the method taught by Dr. Pandya who used it to reconstruct the anatomy of corticocortical connections in the macaque brain.
primary progressive aphasia, a language-based dementia syndrome (Gorno-Tempini et al., 2011;Mesulam, 2003).All AD cases met criteria for Braak neurofibrillary tangle staging II or III (Braak & Braak, 1995) and intermediate or high ADNC according to the National Institute on Aging/Alzheimer's Association guidelines (Montine et al., 2012).See Table 1 for case characteristics at the individual level.

Neuropathologic evaluation and tissue preparation
Following autopsy, the cerebral hemispheres were separated in the midsagittal plane, cut into 3-4 cm coronal slabs, fixed in formalin for 2 weeks or 4% paraformaldehyde for 36 hours, taken through sucrose gradients (10−40%) for cryoprotection, and stored in 40% sucrose with 0.02% sodium azide at 4 • C. Pathologic diagnoses of AD participants were rendered using criteria set by Montine et al. (2012).AD brains contained secondary and/or tertiary pathologies that included, for example, periventricular leukomalacia (Case #4) and white matter infarction in the striatum (Case #14); in these cases, analyses were restricted to regions that appeared unaffected.For all cases, fixed frozen coronal slabs were cut into 40-μm-thick sections.
Extensive double staining procedures were performed to determine the possibility of colocalization (e.g., SMI-AChE, NADPH-d-SOM, MAP-2-NADPH-d, etc.).Finally, in a novel approach, a subset of sections was stained immunohistochemically with an antihuman smooth muscle actin antibody (αSMA; mouse monoclonal; Dako M0851; RRID: AB_2223500; 1/1000) to visualize vasculature in the white matter, and then double stained with NADPH-d (see Table 1).

Digital image acquisition and analysis
A total of six cases (AD, N = 3, Cases 10-12; Control, N = 3, Cases 1, 5, and 9) stained for NADPH-d were analyzed quantitatively for neuronal size and density.A single slide from a whole-hemisphere frontal, temporal, and occipital region from each case was scanned at 20× magnification using the Olympus V200 slide scanner system (Olympus, Tokyo, Japan).Whole-slide images were uploaded into QuPath v.0.3.0, an open-source software used for Quantitative Pathology and Bioimage Analysis (Bankhead et al., 2017).First, the white matter region of each slide was annotated manually using the polygon tool and based on the junction between the white matter and cerebral gray matter.
This area was then further divided into regions labeled "subplate" and "deep."We adapted methods used by Sedmak and Judaš (2021), where we defined subplate neurons as those found in the gyral/sulcal white matter, up to 2 mm below the cortical gray matter; this distinct area

Statistical analysis
Student t-tests were used to compare neuronal size in AD versus normal controls (between and within groups) and utilized in ancillary analyses described above.A one-way ANOVA was used to compare neuronal density in deep and superficial white matter across regions (frontal, temporal, and occipital).All statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA).Significance level was set to p ≤ .05.

WMNs have a heterogenous histochemical signature
In accordance with prior work, two topographically distinct populations of neurons emerged in cognitively normal cases: one between cortical layer VI and superficial white matter, and the other in deep subcortical white matter.Both populations were neurochemically heterogeneous (see Table 1).Markers of these neurons included MAP-2, AChE, SOM, SMI-32, NADPH-d, PV, CB, and CRT, with varying patterns of overlap among these markers.PV staining was done only in Case #7, where it was found that PV-positive neurons were more frequent in superficial white matter compared to deep white matter (Figure 2a,b).NADPH-d allowed morphological visualization of subcortical WMNs (see Figures 2g, 3, and 6).Some of these neurons tended to extend processes in the direction of micro vessels, supporting the speculation of a functional role in vasodilation and regulation of blood flow.In Cases #1 and 15, double staining showed that Rather, the results above provide a descriptive landscape to be better characterized by future systematic approaches.

Morphological features of NADPH-d-positive WMNs
Inspection of dendritic, axonal, and somatic features led to the classification of three groups of NADPH-d-positive WMNs: (1) unipolar, (2) bipolar, and (3) multipolar, which includes isodendritic and stellate subtypes.These cell types were found in both subplate and deep subcortical WMNs.Unipolar neurons were defined as those with a round soma and only one process extending from the cell body at least within the plane of section.Bipolar neurons harbored a longer, narrower cell body, with a prominent apical dendrite and one elongated process extended by distances of 100+ microns (Figure 2h).The multipolar group was comprised in part by isodendritic cells, with large, triangular cell bodies and at least three processes extending from the soma (Figure 2g), as well as stellate neurons, which had rounder, smaller cell bodies and many processes that contained a branching dendritic tree.
In all types, there were dendrites that occasionally bore spines visible by light microscopy.Across frontal, temporal, and occipital regions, multipolar neurons were found to be most ubiquitous, comprising at least half of all neurons inspected, concordant with morphological studies in the macaque (Rockland & Nayyar, 2012).Analysis of a single case (Case #9) revealed no significant differences in the frequency of morphological cell types between subplate and subcortical deep WMNs.For example, of the subcortical deep or subplate WMNs analyzed, multipolar neurons comprised approximately 60% of the total population found in frontal cortex.

Differences in size and density between subplate versus subcortical deep WMNs
NADPH-d-positive subplate WMNs were significantly larger than deep subcortical neurons (Figure 3) across frontal, temporal, and occipital regions (p < .0001).By observation, subplate WMNs tended to show longer, more frequent, and darker histochemical reactions in the cell body compared to neurons in subcortical deeper white matter, though there was no significant difference found in an analysis of optical density completed in the cognitively normal group (N = 3).
Additional quantitation of a sample of WMNs (∼100-120 per group) from the frontal section of Case #9 revealed no significant differences between the number of processes extending from the soma of subplate (M = 2.82 processes) versus subcortical deep WMNs (M = 2.97 processes).There were no statistically significant differences in estimated density across regions in these control cases (Figure 4), although WMN density of both groups appeared higher in occipital compared to frontal white matter.Subplate WMNs showed mean densities of 299.06, 413.81, and 462.49counts/mm 2 across frontal, temporal, and occipital regions, respectively, where subcortical deep WMNs showed means of 508.21, 617.20, and 673.01 counts/mm 2 across frontal, temporal, and occipital regions, respectively.

Comparison of NADPH-d-positive WMNs in normal control versus AD cases
The WMNs did not have neurofibrillary tangles in PHF-1, ALZ-50 or

F I G U R E 6
Nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d) histochemistry in a whole hemisphere orbitofrontal section from the Case #9, a 27-year-old cognitively normal male.The image highlights the proximity of subcortical deep white matter neuron (WMN) cell bodies and their processes to neighboring blood vessels, as seen in the corpus callosum (a) and the superior corona radiata (b).Inset images captured at 20×.MFG, middle frontal gyrus; CG, cingulate Gyrus; gCC, genu of corpus callosum; rCC, rostrum of corpus callosum; CdM, medial caudate nucleus; IP, insular pole; IFG, inferior frontal gyrus; MOrG, medial orbital gyrus.
we found no statistically significant differences in estimated density between control and AD cases in our small sample.

NADPH-d-positive WMNs and adjacent vasculature
Prior work in the corpus callosum of the rat showed that characteristics of NO-producing neurons (NOS-immunopositive) were identical to that of NADPH-d+ neurons (Barbaresi et al., 2014).Similar to results reported in other animal models, NADPH-d+ deep WMNs in the corpus callosum and superior corona radiata displayed processes abutting vasculature as shown in control Case #6 (Figure 6).We employed a double staining procedure of NADPH-d and αSMA, a marker of smooth muscle cells in vessel walls necessary for contractile function (DeRuiter et al., 1997).In normal control specimens, we were able to visualize what appear to be soma in proximity of vessels and dendritic, and axonal processes extending toward vessels, with frequent contact (Figure 7).We observed from close inspection of NADPHd single staining in frontal and occipital sections of a normal control specimen (Case #9) what appeared to be punctate boutons rich in NADPH-d on the outer walls of blood vessels, even in the absence of neuronal cell bodies in the immediate neighborhood (Figure 8).Similar patterns of neuron-vessel proximity were observed in AD specimens (Figure 7).This WMN-vessel proximity was observed in several but not all cases and warrants further study considering the relevance of NO for vasodilation.

DISCUSSION
In this study, we add to the characterization of morphological, regional, and neurochemical signatures of WMNs in the adult human brain.We attempt to do so through novel immunostaining that provides an initial exploration of the colocalization of neuronal markers.Fifteen wholehemisphere postmortem specimens, spanning seven decades of age were analyzed qualitatively and quantitatively.We chose to focus our attention on WMNs in the subcortical deep white matter because they are so conspicuous and because they defy the strict interpretation of the distinction between gray and white matter.We observed that subcortical WMNs display heterogenous chemical and morphological signatures, as reflected in reactivity for MAP-2, AChE, SMI-32, PV, CB, CRT, SOM, and NOS (Bu et al., 2003;Gonchar et al., 2007;Kostovic & Rakic, 1980;Kostović et al., 1991;Tao et al., 1999).The most extensively studied markers were MAP-2 and NADPH-d, which were surprisingly found to be dissociated from each other.These two groups may turn out to have distinct functionalities related to blood flow and perhaps the monitoring of activity along white matter bundles, within which the WMNs are embedded.Indeed, NADPH-d-positive neurons express NOS, and thus presumably release NO, which is a gaseous messenger molecule.A good number of SOMpositive WMNs were also positive for NADPH-d and AChE.AChE-rich neurons were found to be negative for CHAT and thus are believed to be cholinoceptive rather than cholinergic.AChE positivity implies innervation by the nucleus basalis of Meynert, due to high instance of co-labeling of m2 receptors, implicating the cholinergic system in the regulation of NOS in NADPH-d-positive WMNs (Smiley et al., 1998).
We found that only very few NADPH-d-positive neurons were also MAP-2 positive, concordant with previous results (Meyer et al., 1992).
The total number of WMNs should therefore be calculated based on the addition of neurons positive for each of these two markers.
Staining of CBPs revealed no overlap with AChE or NADPH-d.PVpositive neurons were more frequent in the superficial white matter, though this was based on a small number of cases processed for PV immunohistochemistry.
In cognitively normal control specimens stained with NADPH-d (see diverging from the literature that has shown decreased densities of m2-positive WMN in visual cortex (Smiley et al., 1998), though this finding was not significant.
The principal differences we observed in the ADNC specimens at Braak stages V and VI were that the soma sizes of NADPH-d+ subcortical and subplate WMNs were significantly smaller in AD brains (N = 3; Cases 10, 11, and 12) compared to their cognitively healthy, age-matched peers (N = 3; Cases 1, 5, and 9).Perikaryal atrophy was previously reported in human brains with AD (Nassif et al., 2022) and in cases of frontotemporal degeneration with TDP-43 proteinopathy (Kim et al., 2018).NADPH-d+ neurons in our study were devoid of tau tangle pathology, concordant with previous work on NADPH-d+ cortical neurons in humans (Kowall & Beal, 1988).In all cases regardless of cognitive or pathological status, we observed proximity of WMN to blood vessels in subcortical structures (Figure 6).NO is known to be a potent vasodilator, and has been shown to interact with the vasculature in the corpus callosum of the rat (Barbaresi et al., 2014).The literature has previously highlighted this neuron-vessel relationship in subplate WMNs in humans (Estrada & DeFelipe, 1998).
Our findings suggest that such relationships exist in subcortical deep white matter as well.The double staining with NADPH-d and αSMA visualized WMN neuronal processes extending toward vascular profiles as if establishing contact.Additionally, NADPH-d-positive puncta were aligned with blood vessels, although the exact nature of the puncta remains to be clarified (Figure 8).NO has a specific activation range, impacting nearby vessel up to distances of 100-200 microns through volume conduction (Barbaresi et al., 2015;Estrada & DeFelipe, 1998;Smiley et al., 1998;Wood & Garthwaite, 1994), making our findings particularly relevant to the possible role of WMN in the modulation of blood flow within white matter tracts.
Findings were derived from a small sample, given the limited availability of whole-hemisphere tissue; thus, more cases are needed for extensive quantitative analysis.Well-validated histological methods were utilized, although more work using newer methods such as singlecell transcriptomics may address questions of function and origin.
Future studies on the pathophysiological involvement of NOS in cortical neurons and WMNs will be fruitful for our understanding of circuitry function in the human.

[
NeuN]; Millipore; MAB377; RRID: AB_2298772).Cholinergic function was assessed through choline acetyltransferase (ChAT) staining (human polyclonal antibody) as well as acetylcholinesterase (AChE) F I G U R E 2 Photomicrographs at varying magnifications highlighting white matter neuron (WMN) histochemical features in cognitively normal adults.Images (a) and (b) depict parvalbumin (PV)-positive subplate white matter (a) versus subcortical deep white matter neurons (b) underlying temporal cortex.PV was more richly expressed in neurons from superficial white matter compared to deep white matter.Double staining of microtubule-associated protein 2 (MAP-2) (shown in brown) and nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d) (shown in blue) in a temporal (c-e) and frontal section (f-h) reveals almost no colocalization between MAP-2 and NADPH-d.These stains allowed for the identification of several WMN morphological features, including isodendritic (f and g) and bipolar (h) neurons.Note, images (a) and (b) were acquired from Case #7, a 45-year-old male; (c-e) from Case #1, a 55-year-old female; and (f-h) from Case #15, a 75-year-old male.Images (a) and (b) were taken at 20×, (c) at 10×, (d) at 40×, (e) and (f) at 10×, (g) at 100×, and (h) at 40×.
has been shown to correspond to the fetal subplate compartment and eventual remnant throughout development.Neurons labeled as "deep" were those found below the 2 mm range.Each annotation layer was then analyzed using an artificial intelligence object classifier that was trained to detect NADPH-d-positive neurons and ultimately calculate (1) soma size and (2) estimated density.Neuronal density was calculated by quantifying the number of WMNs detected in each annotation layer and dividing by total (subplate + deep) white matter area to compute a number with units of counts/mm 2 .Ancillary analyses measured F I G U R E 3 Photomicrograph of nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d)-positive subplate white matter neuron (a) in the middle frontal gyrus (MFG) versus subcortical deep white matter neuron (b) in the orbitofrontal region from Case #9, a 27-year-old cognitively normal male.By observation, superficial subplate white matter neurons appeared larger than those found in deep white matter.Images taken at 10×. the intensity of NADPH-d histochemical positivity, the frequency of morphological cell types, and number of processes between subplate and subcortical deep WMNs.
very few NADPH-d-positive WMNs are also MAP-2 positive (<1%) (Figure 2c-h).SOM-AChE double staining (Case #1) revealed that roughly three fourths of SOM-positive neurons are also AChE positive.NADPH-d and SOM double staining (Cases #2 and #3) showed that nearly all observable SOM-positive neurons are also NADPH-d positive.Positivity of AChE staining (Cases #1 and #5) and a lack of ChAT staining (Case #7) suggested that AChE-rich WMNs are cholinoceptive but noncholinergic.In the subcortical deep white matter of Case #6, only ∼10% of SMI-positive neurons were AChE positive.Neurons immunopositive for CRT and CB were encountered occasionally in the white matter.In Cases #1 and #4, double staining with CB and AChE suggested almost no overlap between the two stains; the same was true in double staining with CB and NADPH-d (Case #2).Case #1 also showed that roughly two thirds of AChE-positive WMNs were also positive for MAP-2.Subcortical deep WMNs were challenging to detect in CV-stained sections because of background interference with strongly staining astrocytic glial nuclei; visualization with the neuronal nuclear protein NeuN (Case #8) allowed for better resolution.Of note, the results here do not represent the outcome of a systematic investigation based on all possible marker combinations in all available cases.
Thio-S preparations.Comparison of AD cases (N = 3; Cases 10, 11, and 12) to age-matched cognitively normal control cases (N = 3; Cases 1, 5, and 9) showed significantly larger NADPH-d subcortical deep white matter soma size in controls (M = 130.20 μm 2 ; SD = 57.86)compared to AD cases (M = 123.48μm 2 ; SD = 55.33)(Figure 5; p < .0001).Differences were most apparent in temporal regions by an average difference of 14.15 μm 2 between groups.Though one study did show reduced NADPH-d+ WMN densities in AD specimens (Kowall & Beal, 1988), F I G U R E 4 Density (counts/mm 2 ) of nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d)-positive white matter subplate neurons compared to subcortical deep neurons by region in cognitively normal cases (N = 3).Both classes of neurons show the lowest density of WMNs in frontal sections and highest density in occipital regions.These differences did not reach statistical significance but suggest a regional trend.Error bars indicate SD.F I G U R E 5 (a) Mean subcortical deep white matter soma size (per um 2 ) in cognitively normal cases versus Alzheimer's disease (AD) groups (N = 3, per group) where dots represent individual cases.Photomicrograph of nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d)-positive subcortical deep white matter neuron in an occipital section from (b) Case #5, a 43-year-old cognitively normal female and (c) an occipital section from Case #11, a 79-year-old female with amnestic dementia due to AD. Images taken at 10×.Error bars indicate SD; ****p < .0001.

F
Double staining of nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d) and α-smooth muscle actin (αSMA) highlighting the relationship between subplate neurons and blood vessels in the white matter in frontal (a and b) and temporal (c) regions from Case #13, an 88-year-old cognitively normal male.The same double staining in the frontal region (d) from Case #14, a 61-year-old male with an aphasic dementia syndrome (primary progressive aphasia) due to Alzheimer's disease, shows that this neuron-vessel relationship persists in the disease state.Images taken at 20×.F I G U R E 8 Nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d) single staining in frontal (a and b) and occipital sections (c and d) from Case #9, a 27-year-old cognitively normal male, shows an affinity of NADPH-d-positive punctate boutons to the outer walls of blood vessels, even in the absence of neuronal cell bodies.This phenomenon is not well understood, though it suggests a contribution to vascular function and vasodilation-hemodynamic responses.Images (a) and (c) taken at 10×; images (b) and (d) taken at 40×.