Fig 1.
Up-regulation of miRNA-34a and down-regulation of TREM2 in AMD whole retina and macular region versus age-matched controls; (A) color-coded cluster diagram; miRNA-34a and miRNA-155 showed the greatest up-regulation to 3.3- and 1.8-fold over their respective controls in whole retina and 8.8- and 1.6-fold over their respective controls in the macular region; a significant up-regulation was not observed for either miRNA-183 or 5S RNA controls in either (B) whole retina or (C) the macular region; one preliminary study involving miRNA-34a up-regulation and a miRNA-34a-mediated TREM2 down-regulation has previously been reported for the hippocampal CA1 region of AD brain [10,12]; miRNA-34a is part of an inducible pro-inflammatory miRNA quintet consisting of miRNA-9, miRNA-34a, miRNA-125b, miRNA-146a, miRNA-155 involved in degeneration in the human CNS; there were no significant differences between age for control or AMD tissues; all AMD cases were for moderate-to-advanced stages of disease; all post-mortem intervals were 2.5 h or less [20–22]; there were no significant differences in RNA quality (all RNA integrity numbers—RIN values—were 8.1–9.1) or yield between the control (N = 9) or AMD (N = 12) groups (p >0.05, ANOVA); (D-F) Western blot of TREM2 protein in the same AMD (A) and control (C) tissues in (D) whole retina and (E) the macular region; in whole retina (N = 10) TREM2 protein levels were reduced to 0.54-fold of control levels and in the macular region (N = 5) TREM2 levels were reduced to 0.22-fold of control; (E) note that TREM2 Western blot analysis on ≤10% TGSDS gels show multiple bands due to the variable post-translational glycosylation of ~25kDa core TREM2 protein (unpublished); TREM2 protein levels are shown as the mean plus one standard deviation (SD) are bar-graphed in (F); see text; *p<0.05; **p<0.001 (ANOVA).
Fig 2.
(A) Schematic complementarity map for a micro RNA-34a-TREM2 mRNA-3’-untranslated region (hsa-miRNA-34a-TREM2-mRNA-3’UTR) interaction between primary gene products on chromosome 1 and 6 [12]; human sequences shown; not drawn to scale; (B) hsa-miRNA-34a, encoded chr1p36.15 contains 3 canonical NF-kB sites (N) in the upstream promoter (http://www.genecards.org/cgi-bin/carddisp.pl?gene=MIR34A) [8,12,56];; E1 = exon 1; E2 = exon 2 of the miRNA-34a gene; miRNA-34a expression is known to be NF-kB-sensitive in human brain cells [12,58]; (C) miRNA-34a precursor is processed into a mature 22 nucleotide hsa-miRNA-34a sequence; the free energy of association (EA) between hsa-miRNA-34a and the TREM2 mRNA-3’UTR sequence is ~16.2 kcal/mol; the miRNA-34a seed sequence 3’-UGUGACGG-5’ is overlaid in yellow; the complementary TREM2-3’-UTR recognition (DNA) sequence 5’-ACACTGCT-3’ is overlaid in red; an ‘|’ indicates a full hydrogen bond between miRNA-34a and the TREM2-mRNA-3’UTR and a ‘:’ indicates a partial hydrogen bond; (D) the hsa-miRNA-34a recognition feature within the TREM2-3’UTR is located about midway in the 299 nucleotide (nt) TREM2-3’UTR; several other brain-enriched miRNAs located within the TREM2-3’UTR and may also affect TREM2 mRNA stability and regulate its expression; sequence structures in (B) and (D) are not drawn to scale; (E) TREM2 is encoded as a single copy gene at human chr6p21.1; the primary transcript is a 2.7k nt TREM2 mRNA (http://www.genecards.org/cgi-bin/carddisp.pl?gene=TREM2) with a half-life of about 20 hr [60]; it is noteworthy that the TREM2 gene has no strong NF-kB binding site within at least 11 kb of its transcription start site and NF-kB activation has no strong effects on TREM2 transcription (unpublished); single stranded ribonucleotide sequences and alignment derived using miRBASE algorithms (European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton UK; srv/microcosm/cgi-bin/targets/v5/ detail_view.pl? transcript_id = ENST00000 373113) [3–6,8,12,20,56,58].
Fig 3.
TREM2 and DAPI nuclear staining of C8B4 murine microglial (MG) cells (A): (a) control MG cells cultured 3 days, magnification 20x; (b) treated with TNFα; (c) treated with 50 nM miRNA-34a-sc (24 hr) or (d) treated with 50 nM miRNA-34a (24 hr); note significantly reduced TREM2 protein signals in stressed MG cells (b and d) compared to control (a) or miRNA-34a-sc-treated MG cells (C); Westerns blots were performed for TREM2 using an antibody directed against the 277 amino acid murine TREM2 (M227;sc-48765) or TYROBP (DAP12; C-20; sc-7853); SCBT, Santa Cruz, California, USA); nuclei stained with DAPI as in (Fig 6); (B) (upper panel) representative Western blot and (B) (lower panel) bar graph analysis of TREM2 and DAP12 protein levels in control, TNFα-, miRNA-NC or miRNA-34a-stressed MG cells; in this sample set TREM2 protein levels were found to be significantly reduced in TNFα- or miRNA-34a-treated MG cells compared to age matched controls; there were no significant differences in the abundance of the TYROBP/DAP12 adaptor protein amongst control, TNFα, miRNA-NC or miRNA-34a treated cells (Fig 7); N = 8; *p<0.05, **p<0.01 (ANOVA).
Fig 4.
Functional validation of a miRNA-34a-TREM2–3’UTR interaction.
(A) ribonucleotide sequence of the 299 nt TREM2-mRNA-3’-UTR is shown in the 5’-3’ direction; the 22 nt miRNA-34a-TREM2 3’UTR complementarity-interaction region is indicated by a black underline and the 8 nt TREM2-mRNA-3’-UTR seed sequence 5’-ACACUGCU-3’ is overlaid in yellow; a single arrowhead indicates the 5’ end of a poly A+ tail in the TREM2 mRNA (22 ‘A’ nt shown; the length of this poly A+ tail is variable); TREM2 mRNA sequence derived from NM_018965; TREM2 transcript is the major X1 variant (see also http://switchdb.switchgeargenomics.com/productinfo/id_801321/) (Fig 3); (B) TREM2-mRNA-3’UTR expression vector luciferase reporter assay (pLightSwitch-3’UTR; Cat#S801178; Switchgear Genomics, Palo Alto CA); in this vector, the entire 299 nucleotide TREM2 3’UTR was ligated into the unique Nhe1-Xho1 site; (C) control C8B4 murine microglial cells, 1 week in culture; phase contrast bright field microscopy 20x; C8B4 cells transfected with the TREM2-mRNA-3’UTR expression vector luciferase reporter were treated exogenously with miRNA-34a, a scrambled control miRNA-34a (miRNA-sc) or control miRNA-183; see references and text for further details [18,19]; (D) compared to control, C8B4 cells transfected with a scrambled (sc) control pLightSwitch-3’UTR vector, the TREM2-mRNA-3‘UTR vector exhibited decreased luciferase signal to a mean of 0.16-fold of controls in the presence of miRNA-34a; this same vector exhibited no change in the presence of the control miRNA-34a-sc or miRNA-183; for each experiment (using different batches of MG cells) a control luciferase signal was generated and included separate appropriate controls with each analysis; in addition a control vector β-actin-3’UTR showed no significant effects on the relative luciferase signal yield after treatment with either miRNA-183 or miRNA-34a (data not shown); dashed horizontal line set to 1.0 for ease of comparison; N = 5; *p<0.001 (ANOVA). The results suggest a physiologically relevant miRNA-34a- TREM2-mRNA-3‘UTR interaction and a miRNA-34a-mediated down-regulation of TREM2 expression in stressed MG cells. This pathogenic ineraction may be related to the down-regulation of other immune system genes by up-regulated pro-inflammatory miRNAs in the CNS [12,19,22] and/or an impairment in cellular phagocytosis or related phagocytic signaling [5–7,20,21].
Fig 5.
Reactive oxygen species (ROS)-, IL-1β- or TNFα-stressed MG cells—involvement of NF-kB and miRNA-34a and the effects of NF-kB inhibitors or anti-miRNAs (AMs)–(A) representative Western blot of TREM2 protein levels in variably stressed MG cells both in the presence and absence of NF-kB and AMs; miRNA-34a levels were determined using microfluidic miRNA array analysis in the same sample (Fig 1); (B) note ROS-, IL-1β- or TNFα-induced increases in miRNA-34a and TREM2 protein decreases in the same sample; when present the antioxidants and/or NF-kB inhibitors PBN, CAY10512, CAPE or curcumin quenched this induction; see text for further details; similarly anti-miRNA-34a (AM-34a) but not 4 other AM species: AM-183 (or AM-9, AM-125b or AM-146a; data not shown) selectively lowered miRNA-34a levels while increasing TREM2 to 0.92 of control levels; N = 6; *p<0.05, **p<0.01 (ANOVA); NS = not significant.
Fig 6.
Control C8B4 murine MG cells efficiently phagocytose Aβ42 peptides; miRNA-34a treated MG cells do not; (A) C8B4 MG cells (ATCC CRL-2540) were cultured for 3 days (control a-d, top row); cells were treated with 5 μM of Aβ42 for 24 hr before staining; Aβ42 peptide (American Peptide Company, Sunnyvale, CA, cat # 62-0-80A); Aβ42 peptide was prepared as previously described [33]; briefly, Aβ42 peptides were initially solubilized in hexafluoroisopropanol (HFIP; Fluka Chemical, cat# 52512; Sigma-Aldrich, St. Louis MO), aliquoted, and stored at −20°C as an HFIP film. After vacuum evaporation of HFIP, aliquoted peptide was re-suspended with DMSO to 5 mM and diluted to 5 μM into the cell culture media; (B) cells were treated with a scrambled miRNA-34a sequence (miRNA-34a-sc; 30 nM;, control, a-d, middle row); or (C) with an LNA-stabilized miRNA-34a (30 nM; miRNA-34a stressed; a-d; bottom row); treatments were for 24 hr before incubation with 5 uM of Aβ42 (made up as in [33]) for another 24 hr before assay; MG cells were subsequently stained using a murine amyloid beta MABN10 (red fluorescence λmax~650 nm; anti-Aβ antibody, clone W0-2; Millipore, Bellerica MA), a TREM-2 antibody (M-227): sc-48765 (green fluorescence; λmax~510 nm; Santa Cruz, Santa Cruz CA) or DAPI nuclear stain (blue fluorescence; λmax~470 nm) as in Fig 3; arrows indicate Aβ42 uptake into MG cells; note decreased presence of TREM2 in miRNA-34a treated MG cells (bottom row, panel B) and decrease in ingested Aβ42 peptide within C8B4 cells (C; bottom row, panel d). Taken together these results support a miRNA-34a-mediated impairment of sufficient TREM2 to phagocytose Aβ42 peptide from the extracellular space; note self-aggregation of Aβ42 peptide after 24 hrs and Aβ42 peptide affinity for TREM2 containing cells (leftmost panels) and internalization (rightmost panel; yellow merge; λmax~580 nm); magnification 20x; Aβ peptide quantification was performed using SlideBook 5.0 (Intelligent Imaging Innovations) and ImageJ (NIH) software; under these conditions about 42% of externalized Aβ42 was cleared; additional relevant methods have been described [10,44].
Fig 7.
Schematic of the structure and function of an NF-kB-regulated, miRNA-34a-mediated TREM2 sensor-receptor circuit down-regulated in AMD and in stressed MG cells; TREM2 is a variably glycosylated, single pass, integrated transmembrane sensor-receptor (green oval; deglycosylated MW ~26 kDa) embedded in the MG plasma membrane [37–39,46–53,85]; TREM2-mediated signaling via the tripartite TYROBP (DAP12) accessory receptor (brown oval; MW ~12 kDa) results in Aβ42 peptide engulfment, phagocytosis and ultimately, clearance of Aβ42 peptides (red spheres) from the extracellular space; TREM2 appears to be able to deal effectively with Aβ42 peptide monomers, however encounter difficulty ingesting Aβ42 peptide dimers, oligomers or higher order structures (Fig 6) (and unpublished observations); insufficient TREM2 may be in part responsible for the inability to adequately phagocytose Aβ42 peptide monomers resulting in their self-aggregation in the extracellular space; neurotoxic metals (such as aluminum) may contribute to the aggregation of external Aβ42 peptide monomers into higher order structures while also up-regulating additional miRNA-34a via NF-kB activation [21,42,43]; importantly, TYROBP (DAP12) protein levels were found to be unchanged in sporadic AMD or in stressed MG cells (unpublished observations); see (Fig 3); TREM2 mutations may affect MG cell’s ability to phagocytose [6–14;85]. Inset: the NF-kB-induced, pro-inflammatory miRNA-34a is found to be significantly increased in AMD retina and in stressed MG cells; miRNA-34a targeting of the TREM2 mRNA 3’UTR appears to be in part responsible for the down-regulation of TREM2 expression (see Figs 2 and 3); because miRNA-34a is an NF-kB-regulated transcript inducible by ROS and pro-inflammatory cytokine stressors from outside of the cell, free radical scavenging (PBN), anti-NF-kB (CAPE, CAY10512, curcumin) and/or anti-miRNA-34a (AM-34a) strategies (elongated black ovals) or combinatorial strategies may be clinically useful in the restoration of TREM2 and ‘homeostatic’ phagocytosis.