BMC Med: cellular transcriptional changes in peripheral blood in Alzheimer's patients

Alzheimer's disease (AD) is a progressive neurodegenerative disease that is the most common cause
of dementia worldwide.
Cumulative data indicate the role of
the peripheral immune system in the pathogenesis of AD.
However, there is a lack of comprehensive understanding
of the molecular characteristics of peripheral immune cells in AD patients.
To explore changes in cell composition in peripheral blood and changes in gene expression in single cell types, the researchers used algorithms in a large-scale bulk blood expression cohort to perform calculations and identified differentially expressed genes
in individual cell types by adjusting the cell ratio.
The researchers detected a significant increase and decrease in the proportion of neutrophils and B lymphocytes in the AD blood, respectively, which was reproducible
in the other three AD cohorts.
Differentially expressed genes in AD neutrophils are enriched into some AD-related pathways such as ATP metabolic processes and mitochondria
.

The researchers also found that in neutrophils of AD risk genes, including CD33 and IL1B, cell interaction activation of leukocytes and protein interaction network modules of mitochondrial and cytokine-mediated signaling pathways were significantly enriched
.
Changes in cell composition and levels of specific gene expression are significantly associated
with clinical and pathological changes.
Similar patterns
of disturbance in neutrophil cell proportions and gene expression levels can also be observed in patients with mild cognitive impairment (MCI).
In addition, the researchers noted an increase
in the abundance of neutrophils in the AD brain.
In summary, this study reveals a molecular disturbance map
at the AD cell level.
These changes highlight the role of neutrophils in the pathobiology of
AD.

Paper ID

Cellular transcriptional alterations of peripheral blood in Alzheimer's disease

Translation: Cellular transcriptional changes in peripheral blood in Alzheimer's patients

Journal: BMC Medicine

IF：11.
15

Published: August 2022

Corresponding author: Zhao Xingming

Corresponding author: Institute of Brain-like Intelligence Science and Technology, Fudan University

DOI number: 10.
1186/s12916-022-02472-4

Experimental design changes in the proportion of immune cells in AD patients

The researchers obtained bloodborne gene expression data from individuals in the ADNI database, including 116 cases of AD, 382 cases of MCI, and 246 cases of elderly NC.

The researchers first applied the EPIC algorithm to estimate the proportion of immune cells in these samples using gene expression profiles of reference immune cell types (Figure 1A).

The researchers found that the proportion of B cells in AD patients was significantly lower than that of NC patients and MCI, while the proportion of AD neutrophils was significantly higher than that of NC.

The researchers then used the neutrophil-to-lymphocyte ratio (NLR), the ratio between neutrophils and lymphocyte counts in peripheral blood, to assess neutrophil-mediated systemic inflammation
in these samples.
The researchers found a significant increase in NLR in AD patients compared to the control group (Figure 1B).

To further validate this observation, the researchers also estimated the proportion of
immune cells in three other separate blood transcriptome datasets (i.
e.
, ANM1, ANM2, and ZIB) using three computational tools, namely quanTIseq, CIBERSORT, and MCP-counter.
AS EXPECTED, THE RESEARCHERS FOUND THAT WITH THE EXCEPTION OF THE ANM2 DATASET UNDER CIBERSORT, THE NL R OF AD IN MOST DATASETS WAS SIGNIFICANTLY HIGHER THAN NC (FIGURE 1B) WHEN USING DIFFERENT COMPUTATIONAL TOOLS (FIGURE 1B).

Although the difference in NLR between MCI and NC was not statistically significant in most tests, the researchers could still observe the same trend
between AD and NC.
Overall, these findings suggest that AD and MCI have a higher
inflammatory response than NC.

The researchers then performed principal component analysis (PCA)
on the proportion of immune cells in these samples.
The first two principal components (PC) accounted for 59.
9%
of the total variance-covariance of the proportion of immune cells in the sample.
Neutrophils (negative) and B cells (positive) contribute the most to PC1 (35% of total variance-covariance) (Figure 1C
).
PC2 (24.
9% of the total variance-covariance) has the strongest
weight on CD8+ and CD4+ T cells.
The PCA scatterplot shows that the immune cell characteristics of most AD patients are indistinguishable from those of the control group (Figure 1C), but a subset of these patients have a lower PC1 positive score, i.
e.
, a significantly lower proportion of B cells and/or an increased proportion of neutrophils, indicating lower adaptive activity and/or higher innate and pro-inflammatory activity in these samples
.

The researchers then assessed the correlation between immune cell ratio, cognition, disease diagnosis, early detection of AD in brain structural features (derived from MRI data), and demographic variables (Figure 1D)
in NC and AD participants.
Consistent with the above results, NLR and neutrophil ratio were positively correlated with disease diagnosis, while negatively correlated with hippocampal and left entorhinal cortex volume and cognitive function (Figure 1D).

Similar patterns
of correlation can be observed when assessed using all AD, MCI, and NC individuals.
In addition, the researchers noted that cell ratio and NLR may be affected by age and sex (Figure 1D).

Considering that age and sex differences between different diagnostic groups can be important factors, the researchers performed logistic regression
between disease state and cell ratio by adjusting for two factors, namely age and sex.
After adjusting for factors, the proportion of neutrophils and B cells was significantly positively and negatively correlated
with AD status in the ADNI dataset, respectively, despite a decrease in the level of statistical significance.
The researchers also confirmed these results
in the ANM1 and ANM2 datasets.
The researchers observed the same direction between the ratio of neutrophils and B cells and MCI status, but no significant correlation
.
In addition, after adjusting for age and sex factors, there was no statistically significant
difference in the proportion of CD4+ T cells between MCI and AD.
Consistent with previous studies, these results suggest that, in addition to age and sex factors, the occurrence of AD drives to some extent an increase
in neutrophil proportion and NLR.
Figure 1.
Proportion of
peripheral immune cells in AD patients.
Cell ratio distribution
of six peripheral blood immune cells in patients with A, MCI, AD, and control groups.
B.
Comparison
of AD and NC neutrophil-to-lymphocyte ratio (NLR) in different datasets under different calculation tools.
Principal component analysis (PCA)
of cell proportions of C, AD and NC participants.
D.
Correlation
between immune cell ratio, diagnosis, cognitive measures, brain area, and demographic variables.

Differentially expressed genes for cell types in AD

By adjusting the co-construction factors, linear regression identified the differentially expressed gene (DEG) of AD in each dataset (Figure 2A), where there was significant overlap and strong agreement with the gene identified using Wilcoxon's test, confirming the reliability
of the results.
Consistent with changes in the proportion of immune cells and previous studies, the researchers found that a large number of up-regulated and down-regulated genes in AD were significantly enriched in three-quarters of the dataset in neutrophil activation/degranulation and B/T cell receptor signaling pathways, respectively (Figure 2B, C
).
The researchers also observed downregulation of MCI neutrophil activation/degranulation and upregulation
of RNA catabolic processes in the ANM1 and ANM2 datasets.

To further examine the intrinsic expression changes of individual immune cell types, the researchers determined the DEG in each cell type by adjusting the cell ratio and co-construction factor in the ADNI dataset (Figure 2D).

The researchers compared the overall DEG with the cell type of DEG and found that only a small fraction of the cell type overlapped
with the overall DEG.
The researchers noted that four of the six cell types (i.
e.
, neutrophils, monocytes, NK cells, and CD4+ T cells) had significant overlap with large amounts of DEG; The overlap is strongest in neutrophils, which may be due to
the high abundance (55-70%) of this cell type in the peripheral blood.

Next, the researchers explored the sharing of disturbed biological pathways in peripheral immune cell types in AD patients from the ADNI dataset
.
As shown in Figure 2E, dysregulated genes in all or most cell types are associated with immune response regulation, defense response regulation, cellular response to cytokine stimuli, and inflammatory response regulation, which is consistent with
dysregulation signals commonly found in AD peripheral blood mononuclear cells revealed by single-cell sequencing analysis.
In addition, the researchers observed cell type-specific disruptions in some biological processes, such as Ras protein signaling in B cells, positive regulation of cell migration in NK cells, regulation of biomineralization in CD4+ T cells, antigen processing, and presentation of monocytes
.
Interestingly, the researchers found that DEG in neutrophils was significantly enriched during ncRNA metabolism, peptide and ATP metabolism, ribosome subbiogenesis, and mitochondrial tissue, which were also downregulated at the overall level, suggesting that neutrophils may be a major factor
in peripheral transcriptomic alterations.
Mitochondrial dysfunction and low energy metabolism have been reported to be among the most common and earliest abnormalities in AD and MCI and may be associated with the production of Aβ, highlighting the potential role
of neutrophils in AD pathobiology.
Figure 2.
AD differentially expressed genes (DEGs)
at the population and cell type levels.
The Wayne diagram shows the overlap
of a large number of DEGs in four blood transcriptome datasets.
B, C, the highest enriched biological processes
in AD that upregulate (B) and downregulate (C) genes compared to NCs in each dataset.
The circular plot shows the overlap
between DEG genes for each cell type in ADNI.
E.
Enrichment network
for each cell type in ADNI.

Enrichment of AD risk genes in neutrophils

Protein interaction (PPI) networks can help elucidate the biochemical complexes or signal transduction components
that control signal output.
Given the function of neutrophils in the pathogenesis of AD, the researchers used Metascape to reconstruct the PPI network
of DEG in neutrophils.
The researchers identified densely linked modules (as small as three proteins) that represent specific molecular complexes and functions (Figure 3A).

For example, the most important module #1 is biogenesis rich in ribonucleoprotein complexes, including ribosomal protein families such as RPL35, RPL7A, and RPL6
.
Modules #2 and #3 are rich in cytokine-mediated signaling, including interleukin (IL) and tumor necrosis factor (TNF).

The researchers also observed biologically significant functions of other modules #3, #4, and #5, including ATP metabolic processes, mitochondrial transport, and the outer mitochondrial membrane
associated with mitochondrial tissue.

To confirm that the modules are associated with the pathogenesis of AD, the researchers assessed gene enrichment in the AD genetic risk gene set or the AD-associated gene set module from different sources (Figure 3B).

The researchers found that modules #2 and #3 were significantly enriched in multiple AD-associated gene sets
.
Module #2 exhibits enrichment mainly in the regulation of leukocyte cell adhesion and activation
.
Interestingly, the researchers found that two central genes, IL1B (a pro-inflammatory member of the interleukin-1 cytokine family) and FAS (encoding a TNF receptor superfamily), in module #2, were upregulated in neutrophils of AD (Figure 3B).

The researchers also noted that module #3 (mitochondrial tissue and cytokine-mediated signaling pathway correlation module) significantly enriched AD genetic risk genes
in meta-analysis of GWAS and large LOAD combined datasets.
For example, CD33 expression levels are positively correlated with hippocampal and entorhinal cortex
volume.
In previous studies, it has been reported that CD33 may play an important role in microglia-mediated Aβ clearance and neuroinflammatory pathways in the brain, with structural variants and SNPs associated
with a high risk of AD.
In addition, the researchers wanted to note that CD33 is one of
the central genes with a high clustering coefficient in Module #3.
These results highlight the function of neutrophil module #3 in the pathogenesis of
AD.

The researchers then performed a bioprocess enrichment analysis
of DEG in neutrophils.
The researchers found that mitochondrial tissue in AD and NC was inhibited, and cytokine-mediated signaling pathways and inflammatory responses were activated
.
To validate disruption of the neutrophil signaling pathway, the researchers further performed pathway enrichment analysis using DEGs determined from different datasets (Figure 3C-D).

While DEGs from different datasets show limited overlap (Figure 3C), the researchers want to highlight convergence of the enrichment pathway categories in at least three cohorts, including metabolic processes, immune system processes, and responses to stimuli (Figure 3D).

In addition, the researchers observed similar enrichment pathway patterns in MCI in different cohorts (Figure 3E).

Figure 3.
Protein-protein interaction network
in neutrophils.
A.
Protein-protein interaction network and MCODE module
of DEG in AD neutrophils.
B.
Enrichment of each module in the list of AD-related genes
.
The Wayne diagram shows the overlap
of DEG in AD and NC neutrophils in four sets of blood transcriptome data.
Heat map
of the enrichment pathway of DEG in AD(D) and MCI(E) in four datasets.

AD brain neutrophils are elevated

Evidence suggests that dysfunction and deterioration of the blood-brain barrier (BBB) may play a key role in the pathogenesis of AD through feedback loops of Aβ accumulation and accelerate the onset
of cognitive impairment and dementia.
Blood-derived subsets of white blood cells, such as lymphocytes and monocytes, have also been found in the brains of AD patients and animal models
.

To examine whether neutrophils have penetrated into the AD brain, the researchers estimated the proportion and abundance of
neutrophils in the AD brain using a transcriptome dataset of 97 AD patients and 105 controls.
Consistent with extravasation migration of neutrophils to the brain, the researchers observed a significant increase in absolute abundance of cerebellar (CBE) and temporal cortex (TCX) neutrophils in the AD brain (Figure 4A
).
The researchers also found that neutrophil abundance was significantly positively correlated
with neural fiber staging scores (Figure 4B) and amyloid phase (Figure 4C) in two brain regions.
However, the researchers did not observe a significant difference in the proportion of neutrophils between AD and NC, which may be due to large elevations
in brain immune cells such as microglia and astrocytes.
In addition, the researchers quantified the abundance of neutrophils (i.
e.
, 3, 6, 12, and 18 months) in the spatial transcriptome dataset of AD mouse models
.
As expected, the researchers found significant increases in the abundance of both hippocampal (HP) and cortical (CX) neutrophils in the late stages (i.
e.
, 12 and 18 months) rather than in the early stages (3 and 6 months) (Figure 4D-E).

Figure 4.
AD abundance of neutrophils in the brain
.
A.
Comparison
of absolute abundance of cerebellar (CBE) and temporal cortex (TCX) neutrophils in AD patients and control groups.
B.
Correlation
between temporal cortical neutrophil abundance scores and neural fiber staging scores.
C.
Correlation
between neutrophil abundance score and amyloid score.
D, E, comparison of absolute abundance of hippocampal (D) and cerebral cortex (E) neutrophils in AD and NC model mice of different ages
.

discuss

In this study, the researchers first analyzed
a large-scale bulk peripheral blood transcriptome cohort of AD, MCI, and NC.
The researchers then compared the cellular proportions of immune cells in these groups and further validated the results
using three separate blood transcriptome datasets and other computational tools.
Next, the researchers estimated the cellular intrinsic DEG
of the immune cell type.
The researchers identified cell type-specific and shared signaling pathways
between these immune cell types.
The researchers also reconstructed the protein-protein interaction network for DEG in neutrophils and identified several modules
associated with the pathogenesis of AD.
Finally, the researchers found that immune cells are more permeable in the AD brain
.

The researchers found that the proportion of neutrophils and B lymphocytes in AD patients increased and decreased, respectively, compared to NC, which was consistent with
elevated NLR in AD patients.
The researchers observed that neutrophil ratio and NLR were negatively correlated with AD magnetic resonance imaging (MRI) biomarkers and cognitive measures, and AD status even after adjusting for age and sex factors
.
Neutrophils are the most abundant type of white blood cell in human peripheral blood and are considered a key element
of innate immunity against infectious pathogens.
NLR in peripheral blood reflects the balance
between systemic inflammation and adaptive immunity.
Thus, elevated neutrophils and NLR in AD indicate a potential role
for systemic inflammatory responses in the pathogenesis of AD.
Interestingly, the researchers noted that the NLR of patients with MCI was higher than that of patients with NC, suggesting that this biomedical measurement can serve as a strong indicator
for early diagnosis of AD.
In addition, previous studies in mouse models of AD have shown that neutrophils in peripheral blood can adhere to cerebral capillaries and block cortical blood flow, penetrating into the brain parenchyma through integrin LFA-1, ultimately promoting BBB damage
.
Therefore, the researchers noted that in the early and late stages of AD, the abundance of neutrophils increased in multiple brain regions, and neutrophil abundance was positively correlated
with AD pathological changes.

Given the effects of mitochondrial dysfunction and decreased metabolism on cognitive decline and Aβ deposition, the specific downregulation of neutrophil metabolic processes and mitochondrial tissue highlights the role of
neutrophils in AD progression.
In addition, the researchers noted that mitochondrial tissue, cytokine-mediated signaling pathways, and leukocyte-cell adhesion network modules in neutrophils were significantly enriched
in AD risk or related genes.
Specifically, the expression of certain pro-inflammatory cytokines and receptors (e.
g.
, IL1B, IL1R1, IL21R, and TNFAIP6) in neutrophils in AD patients was significantly upregulated
compared to the control group.
These cytokines can target microglia and brain endothelial cells, resulting in increased microglia activation and altered
BBB permeability and transport, respectively.
In addition, CD33 is a biomarker of immature neutrophils that is downregulated in AD neutrophils, consistent with
elevated degranulation of AD neutrophils.
In addition, the expression of neutrophil oxidative phosphorylated subunit genes such as NDUFC2, NDUFA7, and NDUFV1 is reduced, reflecting mitochondrial dysfunction
in the AD brain.
These results suggest the potential role
of neutrophil molecular changes in the pathogenesis of AD.

There is a consistency between neutrophil degranulation enrichment and an increase in the proportion of neutrophils in a
large number of upregulated genes.
In addition, for most immune cell types, especially neutrophils, there is significant overlap
between large amounts of DEG and intrinsic DEG.
This suggests that a large number of transcriptome alterations are driven
by a combination of cell ratio and abundance and changes in cell expression.
Despite the significant overlap, most DEGs for each cell type were not detected at the overall level, highlighting the importance of
identifying cell-intrinsic DEGs for the development of diagnostic biomarkers.
In addition, the researchers note that DEG shows limited overlap between different cohorts (Figure 3C) and also with the overlap identified from single-cell sequencing datasets, which may be due to differences in sample preparation, sequencing methods, and analysis, as well as heterogeneity
of participants.
However, there is convergence in dysregulated signaling pathways, such as dysregulation of defense responses, dysregulated cellular responses to cytokine stimuli, and dysregulated inflammatory responses in most cell types, suggesting that pathways and network-based markers may be more robust
.

Although biological insights have been gained through changes in cell proportions and cellular molecules, there are some limitations
.
First, when performing differential expression analysis and logistic regression analysis, the researchers adjusted for covariates
for age and sex.
However, immune cell count and peripheral blood expression profiles are susceptible to hidden factors such as comorbidities, clinical, and lifestyle factors
.
Second, AD is a heterogeneous disease
with different pathophysiological mechanisms.
Thus, brain transcriptome data reveal molecular subtypes corresponding to different combinations of multiple dysregulated pathways, such as sensitivity
to tau-mediated neurodegeneration, Aβ neuroinflammation, synaptic signaling, immune activity, and mitochondrial tissue.
Therefore, the changes found in this study may reflect the average signal
of heterogeneous samples.
Finally, there is currently no single-cell sequencing data
on AD neutrophils.
Although changes in cell ratio and enriched signaling pathways can be reproduced in multiple independent datasets, comparisons of single-cell analysis of the same batch of data are lacking to illustrate the performance of the
method.

In conclusion, this study explores changes in the proportion of different immune cell types in the peripheral blood of AD patients and changes
in intrinsic expression of individual cell types.
The researchers observed a sustained increase
in neutrophil ratio and NLR in independent cohorts of AD and MCI participants compared to controls.
The researchers identified cell type specificity and shared signaling pathways
based on DEG in immune cell types.
The researchers then reconstructed the PPI network for DEG in each immune cell type and identified several modules from neutrophils associated with the onset of AD whose molecular functions are enriched in leukocyte interaction activation, mitochondrial tissue, and cytokine-mediated signaling pathways
.
Changes in neutrophil abundance and central gene expression levels of related modules in neutrophils were significantly correlated
with pathological changes in AD.
This work highlights the clinical value of the cellular abundance and expression profile of neutrophils, and in the future, the comprehensive analysis of these data in peripheral blood will help monitor the pathogenesis of AD/MCI and may provide new insights
for the development of therapeutic targets for AD.