Dysregulation of antimicrobial peptide expression distinguishes Alzheimer’s disease from normal aging

Transcriptomic data mining revealed a general upregulation of antimicrobial peptide (AMP) expression in the head of normal D. melanogaster during aging

To explore the expressive regularity of the innate immune system from a broad range of data, a total of 18 eligible experiments from GEO database comprising data of 52 young, 57 middle-aged, and 75 old healthy Drosophila head samples were obtained through data retrieval and selection of Drosophila head samples (Table 1). All expression profiles of innate immune genes were converted into occurrence of high expression. RD and OR values, which represented the difference in expression of innate immune genes, were calculated subsequently.

Among all categories of innate immunity relevant genes (Figure 1A–1C), only AMP genes exhibited a marked increase in gene expression with age. However, the degree of expression of other inducible and constitutive effector molecules in the host defense system of Drosophila, such as C-type lectins or lysozymes; pattern recognition receptors (PRRs), including peptidoglycan recognition proteins (PGRPs) and gram-negative binding proteins (GNBPs); major molecules involved in Toll, immune deficiency (IMD), and Janus kinase - signal transducer and activator of transcription (JAK-STAT) pathways, essentially remained unchanged. Meta-analysis (Figure 1D–1F, Supplementary Figure 1) further demonstrated that with the exception of drosomycin (Drs), among 14 major AMP genes, including attacin (Att) A, B, C and D; cecropin (Cec) A1, A2, B and C; diptericin (Dpt) A and B; Drs; defensin (Def); drosocin (Dro); and metchnikowin (Mtk); there was overexpression of all other genes between at least one younger group with that of at least one older group. The data mining results suggested that the upregulation of AMPs in the head of healthy aging Drosophila could be one of the most important changes in the innate immune system with age.

Figure 1. Comparison of innate immunity gene expression variances in WT flies with age among GEO datasets. The occurrence of high expression was introduced as a statistical parameter to compare the relative expression quantity of immune-related genes across age groups. The upper figures exhibit the rate differences among different classes of genes associated with innate immunity between the middle-aged and young (A), old and middle-aged (B), and old and young (C) groups, clearly demonstrating that AMPs had the most significant differential expression. The expression levels of other gene clusters were generally unchanged. Transcriptional differences in AMP genes were subsequently compared using meta-analysis. The odds ratio between the middle-aged and young (D), old and middle-aged (E), and old and young (F) groups, are shown in the lower graphs, indicating an increased expression of AMPs in the head of WT Drosophila during aging.

Age-associated overexpression of AMP genes in normal Drosophila was observed in RNA-Seq

To validate the gene expression variation of the innate immune system with age, whole transcriptome profiles of normal Drosophila heads at 3, 10, 20, and 30 days post eclosion were analyzed by high throughput RNA sequencing. The raw data had been deposited to the GEO database with an accession number of GSE109489. The relative quantity of gene expression and the significance value between groups were calculated by DESeq2 software. The results revealed that 12 out of 14 AMP genes, including AttA, AttB, AttC, AttD, CecA2, CecC, DptA, DptB, Drs, Def, Dro, and Mtk, had upregulated expression in at least one older group compared to that of at least one younger group (Figure 2B). Despite of slight variations on individual genes, the expression levels of these AMP genes generally presented a gradual tendency to increase with age, consistent with results from the big data analysis (Figure 2A). Lysozyme (Lys) S, and three PGRP genes, including -LC, -SA, and -SD, were overexpressed in at least one pair of age groups. However, there was no significantly different expression among other AMP-related genes.

Figure 2. RNA-seq analysis of differentially expressed genes associated with the innate immune system in control and AD Drosophila model. There were significant differences in transcriptional levels of AMPs and LysS between the two groups. The line chart (A) illustrates different expression patterns of differentially expressed immune-related genes between WT (red, round dots) and AD (blue, square dots) groups during aging. The box plot (B) presents differences in expression therein between the normal (red, left) and disease (blue, right) groups at each time point (3-, 10-, 20-, 30-days post eclosion).

High throughput sequencing indicated differential AMP expression pattern in normal and AD flies

Transcriptomic profiling of the Drosophila model of AD at the time points corresponding to those of the WT group was performed using RNA-Seq. Raw transcriptomic data was deposited to the same database with an identical accession number. Time series analysis illustrated that the expression of most AMPs declined to the lowest on day 10, elevated sharply on day 20, and continuously increased or remained unchanged on day 30. The expression levels of AMP genes displayed a rough trend of initial decrease followed by an increase, which was distinct to that of healthy controls (Figure 2A). The most significant differences in expression between WT and AD groups appeared on day 10, when 13 out of 14 AMPs, including AttA, AttB, AttC, AttD, CecA1, CecA2, CecB, CecC, DptA, DptB, Drs, Def, Dro, and Mtk, showed a remarkable decline in the AD group, with log2 fold changes of -1.816, -1.182, -2.011, -4.250, -2.065, -1.455, -2.735, -1.813, -3.011, -2.184, -3.525, -2.173, and -1.975, respectively (Figure 2B). mRNA levels of LysS, PGRP-LC, -SA, and -SD genes presented a rising trend with age in the AD group, but there were no significant differences among PGRP genes when compared with those of the WT group. However, the expression of LysS exhibited a comparable increase in Drosophila with AD. The log2-fold change values of LysS between the WT and AD groups at day 3, 10, 20, and 30 were 2.891, 3.929, 3.412, and 4.576, respectively (Figure 2).

Validation of AMP gene expression profiles by quantitative real-time PCR (qPCR)

Transcript levels of key differentially expressed innate immunity genes were investigated using RNA-Seq, including 14 AMP genes and LysS, and were validated using qPCR assays. The qPCR result generally recapitulated RNA-Seq data. Expression trends of the aforementioned genes in different age groups were generally consistent with those revealed by transcriptomic analysis (Figure 3A). Quantitative analysis confirmed that expression levels of nearly all AMP genes declined to some extent in the AD group at day 3 and 10. Among these, six AMPs, including AttC, CecB, CecC, DptB, Drs, and Dro, had the most significant differential expression, with log2-fold changes of -1.344, -2.097, -2.750, -2.398, -1.973, and -1.791, respectively. After day 10, AMP expression levels in the AD group rapidly increased. The expression of most AMP genes remained at similar levels to those of the control group, of which CecA and AttA were found elevated on day 20 and 30 with log2-fold changes of 1.250 and 1.255, respectively (Figure 3B). There was a continuously increasing trend in the mRNA level of LysS from day 3 to 30, which was markedly increased relative to control levels, with log2-fold change values of 1.413, 3.636, 3.555, and 3.331, respectively (Figure 3).

Figure 3. Quantitative PCR validation of differentially expressed immune-related genes in control and Aβ transgenic flies. The results confirmed age-associated alterations in expression trends and transcriptional regulatory levels among the AMP and LysS genes between healthy control and disease model flies. The line chart (A) displays the time series (3-, 10-, 20-, and 30-days post eclosion) gene expression in the head tissue of WT (red, round dots) and AD (blue, square dots) flies. The box plot (B) exhibits the comparison of mRNA levels between normal (red, left) and disease (blue, right) model flies among the age groups.

ELISA suggested that dysregulation of AMP expression was positively correlated with Aβ₄₂ concentration but not neuronal apoptosis in AD flies

The levels of Aβ₄₂ and neuronal apoptosis were determined as described previously using respective ELISA kits. ELISA revealed significantly increased Aβ₄₂ concentration and apoptotic DNA fragmentation within the disease group, indicating abnormal Aβ aggregation and neuronal apoptosis in Drosophila with AD (Figure 4A and 4B). Time series analysis showed that the degree of Aβ burden and apoptosis in the head of WT flies remained at low levels over time, whereas that of the AD group presented an obvious upward trend from day 3 to30. The degree of Aβ and apoptosis were most significantly increased on day 20 (Figure 4C and 4D).

Figure 4. Quantitative determination of Aβ₄₂ and apoptosis levels in brain tissue of control and AD model flies. Aβ₄₂ concentration and apoptotic DNA fragmentation representing the extent of apoptosis was examined with ELISA. Compared to the control group (left), which had low levels of Aβ₄₂ concentration (A, left) and cell apoptosis (B, left), the AD group had an increased concentration of Aβ (A, right) and apoptotic DNA fragments (B, right). The trends of Aβ₄₂ production (C) and neuronal apoptosis (D) in the WT (round dots) and AD (square dots) groups are displayed in the lower figures. Subsequent correlation analysis further indicated a potential relationship between AMP expression, Aβ₄₂ production, and neuronal apoptosis.

Pearson's correlation coefficient test (Figure 4E) revealed a significant positive correlation between the generation of human Aβ₄₂ peptide and programmed cell death in the head of the Drosophila model. No significant correlation was detected between AMP genes or LysS expression and apoptosis. Nevertheless, significant correlations between abnormal Aβ concentration and aberrant expression of AttB, AttC, CecA, Drs, Mtk, and LysS genes were detected.

Analysis of innate immune gene expression in the head of normal D. melanogaster during aging based on data mining strategies

To analyze the gene expression of innate immunity in Drosophila with age, transcriptome profiles, including microarray and high-throughput sequencing data, were retrieved from the GEO database by organism-specific keyword searching using terms of aging, age, and “time course.” Eligible experiments had to include head tissue of normal D. melanogaster, which were reared under routine culture conditions without any stimulation and contain at least two age groups (young: < 10 days; middle-aged: 10-29 days; old: ≥ 30 days) in a single experiment. The gene expression value was extracted from the original published datasets. The log2-fold change of gene expression relative to control was calculated. Values greater than 1 (2 folds) were considered significant. We compared the statistical parameters of differential expression obtained from individual datasets instead of comparing gene expression signals between experiments to avoid the difficulties in comparing data from different conditions. Each sample from an individual dataset was compared with other age groups. If the log2-fold change exceeded the cutoff threshold, it was marked as one occurrence of high expression. The number of occurrences (k) and number of samples in a single experiment (n) were counted separately.

For comparison of a certain class of genes, the frequency of high expression genes (response rate, P) was calculated (P = k / n), and the overall expression differences were presented as rate difference (RD) = response rate of the older group (Po) - response rate of the younger group (Py). For comparison of the expression of an individual gene among different age groups, the occurrence of high expression genes and total events were counted separately. The analysis was performed using Review Manager Version 5.0. The odds ratios (OR) were the principal measurements of the effects and were presented with 95% confidence intervals (CI). Differences with p < 0.05 were considered statistically significant.

Drosophila stocks

The Drosophila model of AD that expresses human Aβ₄₂ peptide in the brain was constructed using a cross of the w¹¹¹⁸ genetic background UAS-Aβ₄₂ flies driven by elav-GAL₄^c155 line, while WT (w¹¹¹⁸) flies which also outcrossed with the elav-GAL₄^c155 line were used as controls [39]. All flies were reared at 23°C and 42% relative humidity, and fed with standard corn meal food under 12 h/12 h light/dark cycles. Head tissues from 3-, 10-, 20, and 30-day-old male WT and AD Drosophila were collected for further experimentation.

RNA isolation and gene expression analysis

An approximate 200 fly heads for each of triplicates for each age group were collected, and total RNAs were extracted with Trizol reagent (Thermo Fisher, USA), from which mRNAs were purified by poly-T oligo-attached magnetic beads. Libraries were constructed for sequencing using a Superscript Double-Stranded cDNA Synthesis kit (Thermo Fisher, USA) according to manufacturer's specifications.

The prepared library was sequenced on the HiSeq X Ten platform (Illumina, USA) based on a 2 × 150 bp paired-end (PE150) sequencing protocol. The raw sequencing data were processed with SOAPnuke 1.5.2 with parameters of -l 15 -q 0.2 -n 0.05 -i, to remove adaptor sequences, poly-N reads, and low quality reads [40]. The obtained clean data were mapped to Drosophila melanogaster whole genome dmel_r6.11 using HISAT 2.0.4 with parameters of --phred64 --sensitive --no-discordant --no-mixed -I 1 -X 1000 [41, 42]. The relative quantities of gene expression were calculated using RSEM software.

The levels of differentially expressed innate immunity genes were further elucidated through standard qPCR experiment subsequently. Specific primers used in qPCR analysis are listed as Supplementary Table 1. The assays were performed in triplicate on ABI ViiA 7 Real-time PCR system (Applied Biosystems, USA) using QuantiTect SYBR Green PCR kit (Qiagen, Germany) as the fluorescent reporter. Relative expression was estimated with ribosomal protein L32 (RpL32) as the reference gene using the 2^−∆∆Ct method.

Protein extraction and ELISA assays

Aβ expression in the brains of Drosophila was examined using a High Sensitivity Human Amyloid β42 ELISA kit (Merck Millipore, USA) according to the manufacturer’s instruction. Briefly, 20 fly heads of each of three replicates for each age group were homogenized then diluted in Standard and Sample Diluent followed by centrifugal purification. The supernatant was transferred into a capture-antibody-coated ELISA plate for overnight incubation at 4°C. After careful rinsing with washing solution, the plate was incubated with the biotinylated detection antibody to form an antibody-amyloid-antibody-complex, which was visualized using the streptavidin-HRP method and measured at 450 nm with a microplate reader (Pulang, China).

Neuronal apoptosis in Drosophila was determined with a Cell Death Detection ELISA Plus kit (Roche, Switzerland). In brief, homogenate samples were prepared using the same method described above. The supernatant containing cytoplasmic histone-associated DNA fragments was added to a streptavidin-coated microplate and incubated with a mixture of anti-histone (biotin-labeled) and anti-DNA (peroxidase-conjugated) antibodies that included in the ELISA kit. The DNA-histone-complex was colored with the ABTS Substrate, then measured at 405 nm with a microplate reader (Pulang, China).

Statistical analysis

Transcriptomic analysis for in silico data was conducted using the DESeq2 package of R software [43]. Transcripts with absolute values of log2-fold change greater than 1 and false discovery rates less than 5% were considered to be significantly differently expressed. Statistical analyses for ELISA and qPCR studies were performed using SPSS Statistics 19.0 software and presented as mean ± SD. Dynamic changes in gene expression were statistically analyzed using one-way analysis of variance (ANOVA), with the Bonferroni post-hoc multiple comparison tests. Significance of qPCR validation among age groups was examined using the Mann-Whitney U test. Correlations between differential gene expression, Aβ₄₂ concentration, and neuronal apoptosis were examined using Pearson's correlation analysis. Results were plotted using the corrplot package of R software. A p-value <0.05 was considered statistically significant.