Dystrophic Serotonin Axons in Postmortem Brains from Young Autism Patients

Serotonin and Autism

Since Schain and Freedman (1961) first suggested a link between autism and serotonin, more than 500 papers have subsequently been published in support of this association (Carneiro et al.,2008; Goldberg et al.,2009; Janusonis,2008; McNamara et al.,2008; West et al.,2009; Kolevzon et al.,2010; see Whitaker-Azmitia, 2005). When reviewing the neurochemical correlates of autism, Lam et al. (2006) stated, “overall, serotonin appears to have the most empirical evidence for a role in autism.”

Prenatal exposure to drugs like alcohol (Eriksen et al.,2002; Zhou et al.,2003) and cocaine (Akbari et al.,1992) increase 5-HT concentrations and are associated with humans with an increased incidence of autism (Davis et al.,1992; Nanson,1992). Patients (30%–40%) with autism have been reported to have elevated blood levels of 5-HT and tryptophan (Hanley et al.,1977; Anderson et al.,1987; Spivak et al.,2004; Hranilovic et al.,2007; Weiss et al.,2006; Coutinho et al., 2007; Carneiro et al.,2008). Interestingly, PET studies of autistic boys aged 2–5 years indicated that peripherally injected alpha-[¹¹C]methyl-L-tryptophan uptake is unilaterally lower in the frontal cortex and thalamus relative to controls (Chugani et al.,1997; Chugani et al.,1999). Finally, autism has been linked with 5-HTT gene polymorphisms (Devlin et al., 2005; Cho et al., 2007; Coutinho et al., 2007; Cross et al., 2008; Singh et al., 2008; Wassink et al., 2007).

A number of theories have been proposed as to how pathology of the 5-HT system may lead to autism (Chandana et al.,2005) and animal models can be made to exhibit signs of autism via manipulation of their 5-HT systems early on in development (Whitaker-Azmitia,2001,2005; Janusonis et al.,2006; Boylan et al.,2007; McNamara et al.,2008).

Heterogeneity of Autism and Serotonin Function

The main characteristics of autism are impaired social interaction and communication as well as restricted and repetitive behavior, which usually begin before a child is 3 years old (DSM IVr). The extreme social deficits of autism (as discussed above) are associated with the loss of brain 5-HT activity (Chugani et al.,1999). For example, PET studies have reported that patients with the most severe language problems had the least amount of tryptophan uptake in the left cortex (Chandana et al.,2005). Autistic children may also miss key milestones in speech and motor control and ∼30% actually regress backwards after a certain age (Tuchman and Rapin,1997).

Children with autism can present with different clinical symptoms and comorbid disorders. Individuals diagnosed with autism spectrum disorders (ASDs), particularly those between the ages of 2 and 18, may concurrently exhibit medical disorders such as epilepsy, immune dysregulation, and gastrointestinal and sleep disorders (Ming et al.,2008). Changes in 5-HT also contribute to seizures (Bagdy et al.,2007), anxiety (Lowry et al.,2008), obsessive-compulsive disorders (Goddard et al.,2008), impulsivity (Pattij and Vanderschuren,2008), mood disorders (Aan Het Rot et al.,2009), aggression, and violence (Sieve,2008). Even some of the somatic dysfunctions associated with autism have serotonergic components, such as immune and GI changes (Hansen,2003). Behavioral abnormalities and psychiatric disorders are also commonly observed in children with autism, most of which are also associated with changes in serotonin. These include profound inflexibility to changes in routine, stereotypy, aggression, self-injurious behaviors, and pronounced sensitivity to sensory stimuli (particularly touch and sound) (see Reynolds and Lane,2008; Jones et al.,2009). Some subjects even meet DSM criteria for a second psychiatric disorder such as depression, anxiety, oppositional, defiant, obsessive-compulsive, and/or bipolar disorder (Baghdadli et al.,2003; Simonoff et al.,2008).

Numerous reports have suggested that 5-HT drugs may be helpful in alleviating some of the symptoms of autism. However, treatment of autistic children with selective serotonin reuptake inhibitors (SSRIs) has been shown to have severe adverse effects in a proportion of the patients. For example, treatment of 35 young adults with the SSRI Clomipramine for a 12-week period resulted in 13 cases of clinically significant adverse effects (37%). These included three cases of seizures, three of clinical agitation, three of serious constipation, and three of significant weight gains (Brodkin et al.,1997). Use of the SSRI Citalopram as a treatment for autism was associated with increased energy levels, impulsiveness, decreased concentration, hyperactivity, stereotypy, diarrhea, insomnia, and pruritus (King et al.,2009). These results suggest that it is risky to treat autistic children with 5-HT or other potent psychoactive drugs in lieu of anatomical information of autism's underlying serotonin neuropathology.

Neuropathology of Autism Spectrum

In this report, we limit the scope of neuropathology to any significant structural or cellular deviation from the normal condition of the CNS, and neurodegeneration as a progressive dystrophic process that damages neuronal structures and obstructs neuronal function to an extent sufficient to cause cell death. Most neuro-developmental disorders within the autism spectrum exhibit neuropathology. In autism, macrocephaly (a head circumference greater than the 97th percentile of head sizes in a population) is the most commonly observed neuropathology (Kanner,1943; McCaffery and Deutsch,2005). Head circumferences of patients with autism have been found to increase by more than two standard deviations within the first few years of life, indicating a postnatal acceleration in brain overgrowth between the ages of 2 and 4. A marked slowing of growth rate is observed thereafter (Courchesne et al.,2001; Courchesne et al.,2003). Autistic children have higher brain weights relative to controls but the opposite trend is observed in autistic adults, who have lower brain weights in comparison with age- and sex-matched controls (Bauman and Kemper,2005). An effect of age is also seen in measures of brain volume. Young, preadolescent autistic children have greater brain volumes than older adolescents with autism (Courchesne et al.,2001). Bailey et al. (1998) studied brain tissue from six patients with autism and found cortical thickening and irregular cortical lamination in four cases. Individuals with autism also tend to have dominant right sides (Whitehouse and Bishop,2008).

In addition to global changes in brain features such as size and volume, studies in autism report neuropathology in specific brain regions. One study of 2–3-year olds with autism reported an ∼40% increase of white matter volume in the cerebellum (Courchesne et al.,2001), whereas another study found a larger amygdala volume in autistic children than in controls, though this changed with time and no difference was observed after adolescence (Schumann et al.,2004; Schumann and Amaral,2006).

Age also affects the kinds of cellular neuropathology observed in autism. For example, the nucleus of the diagonal band of Broca (NDB) of the septum in autistic children 13 years old (or younger) has unusually large neurons relative to controls. In autistic patients older than 21, NDB neurons are found to be small, pale, and markedly decreased in number (Kemper and Bauman,1998). In the inferior olive of the brain stem, neurons are reported as plentiful and abnormally enlarged in children with autism but small, pale, and reduced in number in autistic adults. BDNF, a serotonin growth factor, followed a similar pattern. In blood samples taken at postnatal day 1, from babies later diagnosed with autism, it was significantly elevated (Nelson et al.,2001). When measured in young adults who were autistic, BDNF levels in the blood were very low (Hashimoto et al.,2006).

Reduced numbers of small neurons are a frequently observed neuropathological feature in autism. In the cerebellum, Purkinje cells are smaller than normal and decreased in number by 41% (Wegiel,2010). The hippocampus (Bauman and Kemper et al.,2005; Courchesne et al.,2005), fusiform gyrus (Van Kooten et al.,2008) of the temporal lobe, and the amygdala all show smaller and fewer neurons (Schumann et al.,2004; Schumann and Amaral,2006). Another trend consistently observed in autism in conjunction with decreased cell size is increased cell-packing density. Both features are observed at all ages in the amygdala, entorhinal cortex, hippocampus, subiculum, mammillary bodies, anterior cingulate gyrus, and medial septal nucleus (Bauman and Kemper,1994; Palmen et al.,2004; Bauman and Kemper,2005; Pickett and London,2005). In a postmortem study of two infantile autistic children, pyramidal neurons of the hippocampus were found to be smaller relative to controls and in Cornu Ammonus areas 1 and 4 (CA1 and CA4), decreased complexity and diminished dendritic branching were observed (Raymond et al.,1989). In summary, fewer and less mature neurons (i.e., smaller in size and with fewer dendrites) occur in many brain regions of children and adults with autism.

Increased cellular GFAP immunoreactivity has also been reported in the brains of patients with autism (Vargas et al.,2005), whereas a separate brain postmortem study of seven patients with autism revealed a marked activation of microglia and astroglia (López-Hurtado et al.,2002). Cerebro-spinal fluid (CSF) from six living patients with autism showed a proinflammatory profile of cytokines, including a marked increase in macrophage chemoattractant proteins (MCP-1). Taken together, these results suggest a neuroinflammatory process in the cerebral cortex and white matter of patients with autism (Pickett and London,2005).

Neuropathology and Dystrophy

Neurodegenerative disorders such as Parkinson's disease (PD) tend to progress from neuropathology to dystrophy and then finally to neurodegeneration. Despite the evidence of neuropathology in autism, however, no studies have so far indicated that autism is neurodegenerative and only one study (discussed below) has shown evidence of dystrophy in autism—which we define as an intermediate state preceding neurodegeneration in which normal neuronal function is hindered due to axons and/or dendrites sustaining damage, but not to a fatal degree.

As previously mentioned, PD is a progressive, irreversible, and neurodegenerative disorder. Its primary neuropathological effect is that it thins the substantia nigra and gives it a mottled appearance (Hutchinson and Raff,1999) and its dystrophic effects include the accumulation of Lewy bodies (eosinophilic α-synuclein immunoreactive neuronal inclusions) in degenerating neurons (Dickson et al.,2009) and the region-specific loss of dopaminergic neurons from the substantia nigra pars compacta (Lees et al.,2009). Alzheimer's disease (AD) similarly elicits a neuropathology of decreased cortical size and enlarged ventricles along with amyloid plaque accumulation in the brain (Perl,2010). Its dystrophic effects include the formation of neurofibrillary tangles and Lewy Bodies (Kotzbauer et al.,2001). Huntington's disease (HD) is a degenerative disorder that is associated with the neuropathological reduction of volume in the left striatum, bilateral insula, dorsal midbrain, and bilateral intra parietal sulcus relative to controls (Thieben et al.,2002; Peinemann et al.,2005). Its dystrophic effects are thought to include progressive white matter atrophy (Peinemann et al.,2005).

A number of developmental disorders can also be viewed as neurodegenerative diseases. Down syndrome (DS) is a developmental disorder that usually progresses to AD neruopatholoogy. They exhibit early overdevelopment of immature dendritic spines followed by abnormal dendritic arborization and atrophy associated with plaques and tangles (Purpura,1975; Marin-Padilla,1976; Becker et al.,1986; Kanold,2004; Bauman and Kemper,2005; Iqbal et al.,2009). These children exhibit an overall phenotype of delayed brain maturation (Oliver and Holland, 1986). Their brainstems and cerebellums are abnormally small, and the myelination of fibers in their frontotemporal lobes is delayed. In terms of neuronal connectivity, they have a higher synaptic density (generally only seen early in development) relative to age-matched controls (Wisniewski,1990).

Patients with the developmental (and neurodegenerative) disorder of untreated phenylketonuria exhibit progressive white matter degeneration, hypomyelination, and developmental delays of the cerebral cortex characterized by smaller than normal pyramidal neurons and an increased cell packing density, to an extent normally only seen between the ages of 6 months and 2 years (Huttenlocher,2000). Schizophrenia is a developmental brain disorder associated with a neuropathology of neuroinflammation (Arion et al.,2007), structural abnormalities in white matter, continuous and progressive decreases in gray matter, and increases in lateral ventricle volume (Lin et al.,1998; Davis et al.,2003; Kumra et al.,2005). The dystrophic effects of schizophrenia stem from its prolonged activation of large numbers of microglia, as found in the postmortem analysis of anterior cingulate cortex and mediodorsal thalamus of schizophrenic patients who committed suicide (Block and Hong,2005; Nurun et al.,2006; Steiner et al.,2008). Schizophrenia also causes sublethal apoptotic activity, leading to dystrophic neurons with limited apoptosis in their terminal neurites and the loss of individual synapses (Jarskog et al.,2005; Glantz et al.,2006; Narayan et al.,2008). Thus, it is clear that most neurodegenerative disorders, developmental or otherwise, exhibit neuropathology, dystrophy, and ultimately neurodegeneration.

Dystrophic Serotonin Axons

Serotonergic fibers originate in the midbrain and project throughout forebrain (global neuronal system). Their distributive architecture imbues them with a powerful ability to influence global brain activity but also makes them sensitive to environmental changes in any region of the brain. Consequently, these fibers and their fine, highly branched morphology may function as a key homeostatic regulator of brain health (Azmitia,1999). Hyper-innervation, hypo-innervation (Ueda and Kawata,1994), pericellular aggregates (Leger et al.,2001), and dystrophic fibers (see Azmitia and Nixon,2008) have been widely reported as this global system develops and matures. Serotonergic axonal dystrophy has been shown in a number of other neurodevelopmental and neurodegenerative disorders, which commonly show depressive symptoms and neurochemical evidence for reduced 5-HT levels.

Serotonergic axons in postmortem human brains can be selectively labeled with an antibody against the 5-HTT (Austin et al.,2002). 5-HTT Immunoreactive axons were studied in postmortem brain tissue from individuals with PD, diffuse lewy body dementia (DLBD), and frontal lobe dementia (FLD) (Azmitia and Nixon,2008). In control brains and the three aforementioned neurodegenerative diseases, 5-HTT-immunoreactive (IR) axons were found in the midbrain, temporal lobe, and prefrontal cortex (PFC). Unlike controls, however, dystrophic 5-HT axons were found in the PFC and temporal cortical areas in postmortem brains from individuals with FLD, DLBD, or PD. The four major abnormalities observed were: (1) enlarged, twisted, and swollen varicosities, (2) fine fibers forming tight clusters, (3) isolated and splayed fibers with an irregular shape, and (4) densely labeled aggregates with degenerating profiles (Azmitia and Nixon,2008).

Several other postmortem studies show evidence of dystrophic 5-HT neurons. Lewy bodies were seen in the pontine raphe neurons (Ohama and Ikuta,1976), fewer 5-HT neurons in the median raphe relative to controls (MR; Halliday et al.,1990), and lower 5-HT concentration in the PFC (Scatton et al.,1983). PD brains develop a neocortical decrease in the density of the 5-HT transporter (SERT; Chinaglia et al.,1993; Haapaniemi et al.,2001; Guttman et al.,2007), particularly in the frontal cortices (Haapaniemi et al.,2001). Finally, in FLD, dystrophic 5-HT axons with swollen varicosities were found clustered in significant numbers in the deep layers of the PFC. The axons appeared to be splayed and degenerating.

Given these findings, is there evidence that autism causes dystrophy in 5-HT fibers, as the neurodegenerative disorders presented above do? That autism leads to neuropathology is well established. Here, we now present preliminary evidence that dystrophic 5-HT axons can be found in the superior temporal, piriform, and entorhinal cortices, and the amygdala.

Cortical Fibers

Serotonin axons were observed in all the cortical regions that were examined. The fibers appeared to enter from the underlying white matter and to ascend from Layer VI to Layer II. Tangentially oriented serotonin fibers innervated Layer I. Examination of composites of the piriform cortex reveals detailed regional innervation of 5-HT axons (Fig. 2). The 5-HTT-IR axons in the fusiform cortex are dense in both younger (2.75 years) and older (29 years) autistic donors. They are significantly increased relative to the fibers seen in undiagnosed control brains (25 years). Figure 2 indicates the composites of the 5-HTT-IR axons extending from Layer I down to Layer VI and the white matter. In the cortical sections from a 25-year-old control donor, 5-HTT-IR fibers are seen in all layers with an apparently higher density in the upper layers. A similar distribution of serotonergic fibers was reported in primates (Morrison and Magistretti,1983; Austin et al.,2002).

In the cortical section from the 29-year-old autism donor, the dense distribution of 5-HTT-IR axons is seen in all layers. Many fibers appear to be entering from the ventral white matter. Several apparently dystrophic fibers are seen in Layer III (white arrows). The 5-HTT-IR innervation zone appears to be smaller in the cortical section in the 2.8-year-old autism donor than in the 29-year-old donor. The deeper areas of cortex and white matter are sparsely innervated but laterally sweeping fibers are seen throughout the upper layers. No evidence of dystrophic axons has been found at this time.

An interesting observation concerns the morphology of the serotonin axons. In tissue from typical control brains, the 5-HTT-IR axons are fine, varicose, and highly branched (Fig. 3). However, a number of dystrophic fibers appear in autism donors starting as early as 12 years of age. Dystrophic fibers are seen in the amygdala, hippocampus, and in temporal lobe cortices in autism patients. Several types of dystrophic fibers are presented suggesting that local factors, such as glial abnormalities, rather than genetic deficits may be responsible for these changes.

In the terminal regions of the brains from young (teenagers) autistic doors, thick heavily 5-HTT stained axons were seen (Fig. 4). In the cortex, these abnormal fibers were observed traveling with a parallel trajectory in the white matter and in Layer II, and also extending perpendicular from white matter toward Layer II. The thick axons can be relatively long and straight or short and curved. They contain irregularly spaced varicosities, which can be large or small, circular or elliptical in shape. Occasionally, very fine processes can be seen emanating from the larger branch. These axons appear to have intensely 5-HTT immunoreactive variable endings: bulbous, tapered or “cork-screw” in shape (see Autism 14 yr. in Fig. 3). They can be considered to be degenerating. The thick-serotonin fibers in cortex observed in normal rodents (“M” type axons) have large, spherical varicosities with very fine intervaricose segments (Kosofsky and Mollivar,1987; Molliver et al.,1990) whereas those observed after neonatal 5-HTT injections are straight and smooth and suggested be either a developmentally arrested or a degenerating phenotype (Weaver et al., in press). Thus, the autism related fibers described here most probably represent a degenerating morphology because of there varied phenotype and unusual endings.