We demonstrate in this report that TH-ir, presumably noradrenergic LC neurons are vulnerable during the onset of dementia as evidenced by their loss in aMCI and AD and the association of this loss with multiple measures of cognitive deterioration and neuropathological accumulation. We further characterized the molecular underpinnings of LC neuronal vulnerability by single population microarray analysis. Results indicated selective changes in genes regulating mitochondrial function and neuritic/structural plasticity, which also correlated with antemortem cognitive status and postmortem plaque and tangle burden. Hence, the early loss of forebrain monoaminergic signaling mediated by the LC appears to be central to AD pathophysiology.
Rodent and non-human primate investigations have revealed the importance of the ascending noradrenergic LC projection system for cognitive function [11, 16, 85]. Based on a series of seminal experiments, Aston-Jones and Cohen proposed a theory of ‘adaptive gain’ whereby NE integrates sensory, attentional and memory processing by positively modulating signal gain in neurons to facilitate the processing of salient events by driving LC phasic activity in response to relevant stimuli [11]. LC bursts and pulsatile NE efflux in coeruleus target fields (e.g., hippocampus and prefrontal cortex) increases the activity of excitatory inputs and decreases the action of inhibitory inputs, which optimizes task performance [11, 16]. Despite the more caudal location of the LC deep within the brainstem it might be envisioned as the brain’s “watchtower,” scanning incoming sensory information for important events that require immediate attention [33]. Moreover, NE-dependent modulation of long-term alterations in synaptic strength and gene transcription, particularly within the hippocampus and prefrontal cortex, influences memory formation and experience-dependent alterations in neural function and behavior and plays a critical role in the ability of the LC to optimize performance [105]. Finally, evidence from transgenic mouse models of AD and neuronal cell culture studies show that NE exerts a wide array of neuroprotective effects including anti-inflammatory and pro-neurotrophic mechanisms [32, 43, 44, 58, 64, 72, 74, 117]. Taken together, these data suggest that the central NE projection system is essential for cognitive function and, in turn, that LC neuronal degeneration contributes to cognitive dysfunction.
Recently, two reports have shown that LC neurodegeneration coincides with both mounting Braak stage pathology [110] and cognitive impairment [7]. Theofilas and colleagues reported that LC volume decreases ~8.4% with each successive Braak stage, resulting in a significant ~25% loss of LC volume between control cases neuropathologically diagnosed postmortem as Braak stage III compared to stage 0 [110]. This volume loss mirrored a similar rate of change in total pigmented and non-pigmented LC neuron numbers, as measured by unbiased stereology [110]. The authors also noted that the topography of LC volume and cell loss appears to follow a rostrocaudal gradient similar to that reported in frank AD cases [49, 110]. Here, we report that the loss of TH-ir, presumably noradrenergic LC neurons also correlates with Braak stage. In addition, we found that TH-ir neuron loss correlated with increasing neuropathologic burden based on NIA-Reagan and CERAD diagnostic criteria. As reductions in LC neuron number have been associated with increased cortical amyloid plaque and NFT loads in cases of frank AD [18, 108], these observations in early stage cases indicate a strong relationship between LC projection system degeneration and the pathologic sequela of AD.
Arendt and colleagues used unbiased stereology to demonstrate a significant ~13% loss of neuromelanin-positive LC neurons in subjects classified as MCI/prodromal AD (CDR 0.5 who also displayed “low” to “intermediate” amyloid-Braak-CERAD (ABC) diagnostic scores [61]) compared to those classified as controls (CDR 0 and “not” ABC score) [7]. Subjects classified as mild/moderate AD exhibited ~30–45% LC cell loss compared to controls. This study revealed that LC cell loss, which is prominent in cases of frank AD [19, 24, 36, 75, 86, 119], appears to occur early in the clinical progression of AD, concurrent with cell loss in the nucleus basalis and entorhinal cortex [7].
The present morphometric analysis revealed a ~30% loss of TH-ir LC number during the transition from NCI to aMCI in cases classified independent of neuropathological diagnosis. Notably, our estimate of total TH-ir LC cell number in NCI/healthy controls (19,495?±?2,891) was similar to other studies [23, 50, 93, 112] and, more specifically, to Arendt and colleagues’ unbiased estimate of total neuromelanin-positive neurons (17, 487?±?2,736) for this group. The discrepancy in the observed loss of LC neurons between control and MCI/prodromal AD between these two studies (~30% compared to ~13%) could be related to cohort group classifications or possibly to a relatively greater loss of TH-expressing neurons compared to neuromelanin-positive neurons early in the disease process. In this regard, it would be interesting to determine whether a subset of neuromelanin-positive LC neurons undergoes a phenotypic downregulation of TH during the disease. Regardless, the present findings suggest that a disconnection of transmitter-identified LC projections to the forebrain contributes to the presentation of clinical disease.
With respect to the relationship between LC degeneration and cognitive status across the diagnostic groups, we found that reduced numbers of TH-ir LC neurons were associated not only with episodic memory deficits that define aMCI subtypes [63], but also with semantic memory, working memory, perceptual speed, and visuospatial ability. Since the LC innervates components of the dorsal memory network (e.g., dorsolateral and dorsomedial prefrontal cortex) [10, 84] and the amygdala [46, 99, 115], which are both dysfunctional early in the onset of AD [89], disruption of NE afferents to these structures may play a role in the deficits seen in these various cognitive domains [14]. In fact, we found that quantitative decreases in LC neuron number were associated with poorer performance on measures of global cognitive function, as well. Hence, we demonstrate a link between loss of noradrenergic tone, neuropathological criteria, and cognitive decline. However, regression analysis revealed that global cognitive decline, but not neuropathological status, was a predictor for LC cell number. These findings lend support to a previous autopsy study, which showed that higher LC neuronal density was associated with slower rates of antemortem cognitive decline, suggesting a role for the LC in cognitive reserve [120]. Altogether, our data supports an emerging theory that LC neuron degeneration is an early pathological event that contributes to cognitive dysfunction and the onset of AD, and is likely a site for therapeutic intervention.
The cellular and molecular mechanisms underlying LC selective vulnerability in the early stages of AD are unclear. To begin to address this question, we profiled gene expression patterns of transmitter-identified LC neurons microdissected from NCI, aMCI, and AD cases. We noted two major patterns of expression changes related to functional classes of genes regulating mitochondrial function and neuritic plasticity. With respect to the transcripts involved in mitochondrial dysfunction, Nrf1 and Cytc1 mRNA levels were both significantly downregulated in LC neurons in aMCI and mild AD relative to NCI. Nrf1 is a transcription factor that directs the expression of several functional classes of genes involved in mitochondrial function, including those regulating redox homeostasis, mitochondrial biogenesis, calcium homeostasis, and cytochrome oxidase activity [17, 40, 41]. Interestingly, Cytc1 is the heme-containing component in the cytochrome b-c1 complex III of the respiratory chain, accepting electrons from Rieske protein and transferring it to cytochrome c, which couples to cytochrome oxidase [47, 113]. By contrast, transcript levels of Pfkl and Pfkp isozymes were significantly upregulated in LC neurons in AD. These enzyme subunits catalyze the conversion of D-fructose 6-phosphate to D-fructose 1,6-bisphosphate, resulting in the first committing step of glycolysis [60]. Taken together, these results indicate that LC neurons are under considerable respiratory stress during the transition from NCI to prodromal and frank AD. In addition, these neurons displayed significant decreases in transcripts encoding the antioxidant enzymes mitochondrial Sod2 and Gpx1 in aMCI. Given the role of these two enzymes in detoxifying superoxide and hydrogen peroxide, respectively, LC neurons are also likely under considerable oxidative stress prior to the onset of aMCI [109]. These observations support tissue-based assessments demonstrating regional mitochondrial deficits during the progression of AD [12, 15, 59, 71, 101, 102, 116, 125], but importantly, identify these changes in neurons, as opposed to glial cells or admixed populations of cells, thus attributing these markers for mitochondrial dysfunction to LC neurons.
A second common thread of transcript dysregulation in aMCI centered on genes encoding proteins involved in axonal function and neuronal morphological plasticity. For instance, Map1b is critical for microtubule stabilization during axonal growth [114] and Ntn1 plays a role in axonal guidance [81], whereas Nfh is essential in maintaining axonal caliber and is dysregulated in human AD and animal models of the disease [42, 83]. Significant downregulation of these genes suggest that LC neurons may be undergoing an axonal degenerative process during prodromal stages, which progress to frank AD as evidenced by a significant increase in the cytoskeletal proteases Capn1 and Capn2 expression in AD neurons [92, 104]. Moreover, we found a global decrease in the ratio of 3R/4R tau isoforms in LC neurons in aMCI and AD, consistent with previous single cell gene expression studies of cholinergic nucleus basalis and CA1 pyramidal neurons isolated from MCI and AD cases [54]. The alteration in 3R tau gene dosage relative to 4R tau suggests a shift toward tau isoforms associated with slower axonal transport kinetics as well as NFT formation [37, 69, 107].
Finally, we observed a concomitant downregulation of genes encoding the postsynaptic receptor clustering protein Utrn and the dendritic spine marker Synpo in aMCI and AD neurons compared to NCI. In particular, Synpo is an actin-associated protein that may play a role in modulating actin-based shape and motility of dendritic spines and seems to be essential for the formation of spine apparatuses involved in synaptic plasticity [39, 68]. Dysregulation of postsynaptic gene expression in vulnerable LC neurons may signal perturbations prior to the onset of aMCI that are indicative of a failure in neuroplastic and/or remodeling programs [38, 39, 68, 89]. While we cannot rule out the possibility that these molecular changes are epiphenomenal to AD pathogenesis, the timing of these changes to stages prior to the onset of aMCI suggest that they represent pathways critical to disease progression.
Perhaps the most striking feature of the LC is the immensity and divergence of its noradrenergic forebrain efferents [45], extensive complex dendritic arborization pattern, and the long distance these perikarya project to reach their forebrain innervation sites [6, 8, 85]. In this regard, LC neurons are similar to other selectively vulnerable long forebrain projection neuron systems (e.g., cholinergic basal forebrain neurons, substantia nigra pars compacta neurons, and dorsal raphe neurons, among others [5, 16, 78, 80, 105]), that are heavily reliant on energy metabolism and cytoskeletal integrity to modulate synaptic input. These long projection systems are likely more prone to cellular stress given their position in the brain’s organization. Taken together, our findings present evidence for the molecular dysregulation of mitochondrial function and neuritic/structural plasticity coinciding with the loss of LC neurons prior to the transition from NCI to prodromal AD. Moreover, our observation that the dysregulation of these genes correlates with poorer global cognition and greater neuropathological burden suggests that these molecular pathways represent pathogenic mechanisms underlying the selective vulnerability of LC neurons which may be common to other vulnerable neurons. Notably, there is evidence from similar single neuron expression analysis studies showing alterations in mitochondrial and structural plasticity pathways in AD. For instance, we have demonstrated that cholinergic nucleus basalis neurons exhibit alterations in Sod2, Pfkl, and Capn1 in AD [29]. Other studies have also demonstrated an AD-related dysregulation of specific respiratory chain genes (e.g., cytochrome oxidase 5b) in medial temporal and posterior cingulate pyramidal neurons [70, 71]. However, the extent to which dysregulation of these pathways are evident in other selectively vulnerable cell groups during the onset of cognitive decline warrants further comparative assessments of these neuron groups as well as relatively unaffected populations in postmortem brain tissues. Moreover, whether dysregulation of these pathways is related to NFT or Lewy body pathology in these neurons is an additional question to be addressed in future studies [51, 55, 62, 111, 112]. The current findings also suggest the need for additional quantitative biochemical and immunohistochemical assessments of these mitochondrial and structural markers in LC neurons in aMCI.