Background
Alzheimer’s disease (AD) is the most common neurodegenerative disease and is characterized by cognitive impairment, amyloid-β (Aβ) deposition, and neurofibrillary tangle formation [
1‐
5]. Increasing evidence suggests that immune mechanisms contribute to the pathogenesis of AD including reactive microgliosis in postmortem samples, increased microglial activation marker, translocator protein (TSPO), binding on positron emission tomography (PET), [
6] and increased pro-inflammatory cytokines such as interleukin (IL)-6, IL-1β, TNF, and IFN-γ in the cerebrospinal fluid and serum [
7‐
11]. Thus, neuroinflammation is potentially a target for immunomodulatory therapies [
12,
13].
Genome-wide association studies implicate immune system dysfunction, particularly in myeloid-derived cells, as immune-related genes coding for triggering receptor expressed on myeloid cells 2 (TREM2) and CD33 confer increased risk for AD susceptibility [
14‐
18]. Current studies focus on microglia, resident myeloid cells of the brain, throughout AD pathogenesis, but recent literature in neurodegenerative diseases suggest extensive neuro-immune cross-talk between the brain and peripheral immune system [
19‐
22]. This cross-talk may derive either directly or indirectly from peripheral immune cells in the presence of a compromised blood brain barrier (BBB) such as in neurodegenerative disease [
23‐
25]. Additionally during inflammatory insult and microglial depletion, peripheral macrophage engraftment into the CNS was observed with these cells retaining a distinct and lasting transcriptional and functional identity [
26]. Thus peripheral immune myeloid cells could modulate disease progression and outcomes in the CNS. The accessibility of these peripheral myeloid cells, and the lack of accessibility to CNS microglia, prompted a detailed examination of blood monocyte populations during the pathoprogression of AD.
Immature and mature monocytes, here-after denoted as “peripheral myeloid cells,” originate from hematopoietic stem cells and mature into peripheral monocytes with the capability of differentiating into macrophages once they enter tissue parenchyma [
27‐
31]. Changes and shifts in peripheral myeloid populations are indicators of disease onset and progression for a multitude of diseases; the pro-inflammatory phenotypes have direct effects on their specific disease [
32‐
40]. A detailed analysis of peripheral monocyte population and phenotype changes have not been documented thoroughly in AD progression.
Myeloid-derived suppressor cells (MDSCs) are immature myeloid cells that exhibit robust suppressive function on T cell proliferation and mature myeloid cell function which has made them a target of multiple immunomodulatory therapies [
41‐
44]. Chronic AD immune signaling provides a practical mechanism where MDSC activation and recruitment potentially promotes an immunosuppressive environment [
45]. However, the actual role of MDSCs in AD pathology and progression remains unexplored. A detailed investigation into the role and function of mature and immature peripheral myeloid cells could provide insight into mechanisms of their direct involvement in AD pathogenesis as well as provide clues as to the changes in CNS myeloid cells through disease progression.
In this study, we investigated peripheral myeloid populations and their pro-inflammatory gene expression throughout the course of AD compared with age-matched controls. We also investigated changes in a novel and previously unrecognized cell type in AD, MDSCs, to determine their phenotypes during course of disease. Understanding the role of peripheral myeloid cells in AD-related inflammation and mechanisms of their immune suppression may lead to novel therapeutic interventions for the successful treatment of AD.
Discussion
The analyses of peripheral myeloid cells in AD patients provided three important findings. First, pro-inflammatory gene-expressing phenotypes of peripheral monocytes are increased in CDR1,2,3 patients. Second, pro-inflammatory gene-expressing phenotypes of peripheral monocytes are decreased in CDR0.5 patients. Third, MDSCs are increased in number and suppressive function in CDR0.5 and decreased in CDR2/3. MDSC from CDR0.5 suppressed Tresp proliferation and iPSC-derived pro-inflammatory myeloid cells in vitro more effectively than control MDSCs. MDSCs from CDR2/3 patients had significantly decreased suppressive function for both Tresp proliferation and pro-inflammatory myeloid cells.
Activation of myeloid cells has been extensively studied in humans and animal models of neurodegeneration, including AD, with much of the focus on brain microglia which are difficult to study directly during the course of disease. Peripheral myeloid cells have similar functions to brain microglia including chemotaxis, phagocytosis, and cytokine expression but constitute a cell population that can be collected and studied through simple, minimally-invasive methods [
28,
49‐
51]. Activation of immune cells can be assessed using flow cytometry by analyzing HLA-DR, a MHC class II cell surface receptor that is upregulated upon antigen presenting cell activation. HLA-DR expression was increased on mature monocytes from CDR1 and CDR2/3 patients when compared with controls and CDR0.5 which matches the pro-inflammatory gene expression seen in these patients. CDR0.5 monocytes did not exhibit pro-inflammatory gene expression but displayed increased inhibitory CD33 cell surface expression compared with controls and this expression diminished in CDR1 patients while decreasing below control values in CDR2/3 patients. While some studies implicate differential activation and polarization of immune cells due to the age and gender of the patient, we did not observe any significant effects contributed by these factors in our control or AD datasets. Advanced studies into gender and age-specific contributions to neurodegenerative disease are warranted.
The peripheral pattern of immune cell deactivation and reactivation is in accord with the biphasic immune marker profile reported in a study of CSF immune makers in AD [
52]; decreases in immune signaling were noted early in the course of disease and a strong increase in immune signaling was noted in later stages of disease. Additionally, a longitudinal imaging study of AD patients found initial microglial activation upon diagnosis followed by a reduction of activation as patients went through mild cognitive impairment stages. As patients progressed into later stages of AD, microglial activation increased again suggesting that there might be two activation stages; the first being a protective activation state while the later peak being a more pro-inflammatory and destructive process [
53].
Many studies utilize changes and shifts in monocyte populations to identify and predict inflammatory disease and progression in a multitude of inflammation-based disorders [
38‐
40]. Previous AD studies suggest monocyte involvement in AD although our analysis showed no difference of total monocyte number between controls and through the progression of AD [
54‐
57]. With no studies focused on AD monocyte population shifts, our results present a progressive decline of the classical monocyte population (CD14
+CD16
−) with a subsequent increase in both the intermediate (CD14
+CD16
+) and non-classical (CD14
lowCD16
+) subsets as disease progresses. Shifting of populations from the classical population, a patrolling subset, into intermediate and non-classical populations reflects monocyte maturation and pro-inflammatory activation [
58]. Identification of mechanisms whereby patient monocytes are shifting to their subsets and how they are inhibited early while being activated with increasing disease burden is needed.
MDSCs are immature myeloid cells that expand under pathological conditions such as cancer and immune disorders but have yet to be fully characterized in patients or animal models of AD. These cells suppress pro-inflammatory immune cells including mature myeloid cells and CD4
+/ CD8
+ T cells. MDSCs are roughly divided into early-stage, granulocytic, and monocytic MDSCs with the latter being most suppressive and characterized as CD14
+/HLA-DR
−/CD11b
+/CD33
+ cells [
41,
59]. MDSC numbers were increased in CDR0.5 and CDR1 patients, and decreased in CDR2/3 patients. While monocytic MDSCs are undocumented in AD, a previous study documented an increase in granulocytic MDSCs in amnestic mild cognitive impaired patients compared with mild AD patients. However, this previous study did not provide functional data, thus it is not known whether granulocytic MDSCs have differential functions as the course of AD progresses or if AD granulocytic MDSCs have different properties compared to those of healthy controls [
60].
MDSCs and mature pro-inflammatory monocytes have opposing roles in inflammation. The former provides strong immunosuppression while the latter contributes to pro-inflammatory signaling. Analyses of these phenotypes and signaling mechanisms in advancing AD would provide important information on the state of the immune system during the progression of AD. Additionally, these cells may directly engraft into the CNS given the evidence of a compromised BBB based upon the leakage of blood-derived constituents in postmortem AD brains [
61‐
63], the increased CSF/serum or CSF/plasma ratios of albumin [
64‐
66], and imaging showing an age-dependent increase in BBB permeability in the hippocampus, dentate gyrus, and CA1 region [
67]. Additionally, dysregulation of the neurovascular unit, responsible for maintaining BBB integrity, is known to accompany AD pathology [
68]. Peripheral immune cell involvement and engraftment into the CNS are postulated to play important roles in AD [
69,
70]. With that said, roles of infiltrating cells during CNS disease pathogenesis is controversial, contextual, and disease/stage-dependent [
22,
56,
71‐
73]. Whether and how peripheral myeloid cells dictate progression of AD necessitates further investigation, especially the question of peripheral immune cell migration into the CNS and other mechanisms of neuro-immune cross-talk.
Microglia and monocytes, through genetics and activation status in AD, play a critical role in AD pathogenesis [
6,
14‐
18]. Their responsibilities include phagocytosis, cytokine secretion, and restoration of homeostatic conditions with impairments to these functions being deleterious to sensitive neuronal environments in AD. We have developed CD14 myeloid cells from control iPS cells (unpublished Zhao et al.) and polarized them to be pro-inflammatory to examine differential effects of control/AD MDSCs on activated myeloid cells. Pro-inflammatory polarization of iPS cells provides a consistent, activated cell type to compare suppression between patients without the confounding variable of primed or differentially activated mature monocytes from patients.
MDSCs from CDR0.5 patients were able to suppress pro-inflammatory IL-6 expression and protein from activated cells more than controls while CDR1 and CDR2/3 markedly lost their suppressive function. IL-6 transcript and protein analysis were used due to MDSCs producing no IL-6 while IL-6 is a predominant pro-inflammatory output from activated myeloid cells [
74]. The IL-10 data from the RNA studies provided a mechanistic target for MDSC suppression of the myeloid cells. Adding IL-10ab to co-cultures reduced MDSC suppression of myeloid IL-6 transcripts that was most pronounced with CDR0.5 co-cultures but also observed in the other groups. IL-10ab only partially blocked IL-6 suppression leaving other additional mechanisms contributing to suppression, either through soluble signaling or direct cell contact. An extensive study into the intrinsic differences between MDSCs from prodromal and AD dementia is warranted to comprehensively parse out the predominating mechanisms of their suppressive functions and which of these are lost through AD progression.
Co-culture of MDSCs with Tresps showed increased CDR0.5 and CDR1 MDSC suppression compared with controls. The mechanism for MDSC suppression of T cell proliferation is through metabolism of arginine via increased Arg1 transcript in MDSCs [
75,
76]. Arg1 gene expression increased in CDR0.5 MDSCs and then diminished as AD advances. Analysis of CDR1 MDSC suppression shows enhanced T cell proliferation but decreased M1 suppression which may suggest a suppressive function transition occurring at this stage.
Limitations of these studies involve lack of direct evidence that peripheral myeloid cells enter the CNS parenchyma and influence disease progression in AD. Nevertheless, the definitive evidence for alterations in the BBB in AD, the comparability of CSF findings and peripheral myeloid cell phenotype alterations, and the PET scan evidence of increasing neuroinflammation during the course of disease are all in accord with extensive neuro-immune cross-talk between the brain and peripheral immune system. Advances in neuroimaging with specialized ligands may provide tools for analyzing central vs peripheral identities as well as the anti vs pro-inflammatory phenotypes contributing to AD pathogenesis. The identification of distinct transcriptional and functional phenotypes could assist in developing these technologies [
26]. Additionally, complex networks and signaling cascades involving additional cells outside our examination might assist in dictating inflammatory environments and cell phenotypes. Examination of these cells, both individually and in concert with each other, will help establish a more definitive immune environment during AD and through its course.