1. Introduction

As the aged population increases, Alzheimer’s Disease (AD) and AD-related dementias (ADRDs) will become one of the world’s largest medical and economic crises due to the lack of early diagnostic methods and durable therapies^{1, 2}. Age is the highest risk factor for AD, as disease prevalence increases exponentially as a function of age, and close to 40% of older adults over the age of 80 experience AD or ADRDs^{3, 4}. The most important genetic risk factor for developing late onset AD (LOAD) is the Apolipoprotein E (APOE) genotype, at least in Caucasians, where carriers of one or two copies of the APOE4 allele have a substantial increase in the risk for developing LOAD when compared to carriers of the APOE2 or 3 alleles^{4, 5}. Beyond APOE, recent meta-analyses combining genome-wide association studies (GWAS) have expanded the number of AD-risk loci⁶, but most of the disease-associated variants reside in non-protein coding regions of the genome, making it difficult to elucidate how they affect AD susceptibility and pathogenesis.

The exact mechanisms behind these two predominant AD risk factors remains a mystery but both are widely associated with increased abnormal protein aggregation, which is the most abundantly observed pathological hallmark of AD pathogenesis. AD and ADRDs all have an abnormal protein aggregation profile that includes: 1) extracellular amyloid senile plaque deposition occurring at least 5-10 years before mild cognitive impairments (MCI) are observed and, 2) aggregates of phosphorylated Tau (p-Tau) forming intracellular neurofibrillary tangles (NFTs) near MCI onset^{7, 8}. Novel biomarker studies have shown that hyperphosphorylated Tau species exist in plasma prior to onset of dementia symptoms, with specific phosphorylation sites seeming to affect the affinity of binding of the Tau protein to microtubules ⁹. With advancing age, protein aggregation is accelerated by the accumulation of reactive oxygen species (ROS) as well as the onset of neuroinflammation in the brain^10–12. Having APOE4 was shown to significantly increase accumulation of amyloid and NFTs in Caucasians, and also increases inflammation and reactive oxygen species in the human brain^{5, 13}. The effects of these AD risk factors have been studied mostly at the protein level in the brain, but not yet at RNA level, where structure and function have a myriad of important roles in numerous biological processes.

RNA higher order structures of particular interest are RNA guanine-rich G-quadruplexes (rG4s)¹⁴. G-quadruplexes are secondary structures that can fold under physiological conditions by guanine-rich DNA and RNA molecules and occur in the genomes and transcriptomes of many species, including humans¹⁵. rG4s are four-stranded helices formed by guanine tetrads through Hoogsteen base-pairing, with a monovalent cation channel running down the middle of the tetrads (Fig. 1a)¹⁶. G4s are potentially involved in the regulation of DNA and RNA processes, such as replication, transcription, translation, RNA localization, and degradation^17–19. Recently, rG4s were shown to form as a function of stress in eukaryotic cells²⁰. Under non-stress conditions, rG4s were largely unfolded, but upon introduction of stress (including ROS stress) the rG4s preferentially folded and remained folded until the cessation of stress²¹. This observation raises the possibility that rG4s could preferentially fold under the increased and chronic stress conditions of aging and AD in the human brain, leading to abnormally high levels of rG4 formation in older adults or in adults with protein misfolding diseases²². RNA has been known to be included in AD-related aggregates of both amyloid and NFTs in patients with AD for decades²³, and a recent sequencing study of nucleic acids embedded in AD aggregates showed an enrichment in potential rG4-forming sequences²⁴.

Fig 1.

It has recently been shown that rG4s can greatly enhance protein oligomerization^{25, 26}. Moreover, rG4s can bind to and greatly enhance the phase transition of tau²⁷. Similarly, Zwierzchowski-Zarate et al explored the effects of different RNA sequences on tau aggregation in a cellular model²⁸. In their study, several sequences caused an increase in tau aggregation, but the highest level of tau aggregation was achieved with a GGGC repeat sequence²⁸. However, the structure of the RNAs, including possible rG4 formation, was not explored or speculated upon. We therefore performed CD spectroscopy to determine if the sequence associated with the highest levels of tau fibrillation might be forming rG4 secondary structures. The CD spectra show that this RNA forms parallel rG4 structures in vitro (Supplementary Data Fig. 1).

Based on these previous in vitro and cellular studies, an increased presence of rG4s in the brain could contribute to AD-associated aggregation. However, the presence of rG4s in the human brain or other human tissues has not been shown previously, nor how it relates to aging or AD pathology. This was the focus of the current study.

2. Methods

Human brain tissue

This work was approved by the University of Colorado Institutional Review Board (IRB) and the study was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. However, the use of postmortem brain tissue for research is exempt according to federal IRB regulations and does not require IRB review and approval. All efforts were made to consider diversity, equity and inclusion (DEI) in the selection of brain tissues for this study. The tissues included in this study were obtained from the Medical University of South Carolina (MUSC) brain bank (Dr. Eric Hamlett). Postmortem consent was obtained from the next of kin, and the brain was rapidly removed. Fig. 1b shows the demographics for the 21 cases used for this study. The tissue was sliced in 1 cm coronal slices and subjected to free-floating fixation in 4% paraformaldehyde in phosphate buffered saline (PBS) for 72 hours. The fixed tissue was then transferred to a 30% sucrose in phosphate buffer (PB), followed by transfer to a cryoprotection solution in PBS (30% Glycerol, 30% Ethylene Glycol and 40% PO4 Buffer) and stored in −20° C until dissection. Anatomical blocks from 19 brain regions were dissected from the 1 cm slabs, embedded in paraffin and cut at a thickness of 5µm on a Microm Microtome (Thermo Fisher Scientific Inc., Waltham, MA) and stained for routine neuropathological staging according to the NIA-AA Revised Criteria for Diagnosis and Staging of Alzheimer’s Disease ²⁹.

Immunofluorescence

Paraffin-embedded sections were deparaffinized in xylene (RPI, CAT#111056) and rehydrated using wash steps at: 100%, 95%, 70%, 50% ethanol followed by distilled water. Heat-induced antigen retrieval was performed using 0.05% citraconic acid at pH 7.2 for 20 minutes in a steamer. Slides were then washed with a saline solution buffered with tris(hydroxymethyl) aminomethane (Tris, TBS) and blocked with 0.25% triton-X and 5% normal serum for 1 hour. After blocking, slides were rinsed again with TBS and incubated with 1X True Black Lipofuscin Autofluorescence Quencher (Biotium CAT#23007) diluted in 70% ethanol. Slides were washed 3 times with TBS then incubated with primary antibodies listed in Supplementary Data Table 1 overnight, followed by Goat anti-mouse AlexaFluor 555 (Thermo Scientific #A32727) or Donkey anti-rabbit Alexafluor 488 (Invitrogen #A-21206) at 1:500 dilution for 1 hour. Slides were then washed 3 times with TBS and cover-slipped using mounting media containing DAPI (Thermo Scientific, CAT#P36931). Staining controls included omission of primary or secondary antibodies (Supplementary Data Fig. 2). In Fig. 3b, we performed sequential double labeling, first with BG4, and then with rRNA antibody at 1:250 (Abcam #ab17119).

In addition, we examined whether Braak stage correlated with aging itself (Supplementary Data Fig. 3). Although not significant at the 0.05 level, there was a weak correlation between Braak and Age. Braak staging is a paradigm used to describe density and regional spread of neurofibrillary tangles (NFTs) in the Alzheimer brain and includes levels 0-VI⁷. CERAD (Consortium to Establish a Registry for Alzheimer’s Disease) is a staging paradigm for Alzheimer’s disease that takes into account amyloid plaque formation³⁰.

Imaging and Statistical methods

Slides were imaged on a confocal microscope at 10X, 20X and 60X magnification at wavelengths 355, 488, and 555. Images used for quantification were taken at constant microscope and intensity settings. All image quantification was performed using Fiji ImageJ. Quantification was performed by a researcher who was blinded to the neuropathological diagnosis of individual cases and with no knowledge of Braak Stage or Age for the case being quantified. Supplemental Table 2 presents side by side comparisons of quantification from 2 researchers who were both blinded to group diagnosis and individual cases. Data used in the paper are highlighted. One data point with too much variation between quantifications was removed to increase rigor. Background was subtracted using a rolling ball radius. Images were then divided into 3 regions of interests (ROI); OML, DG and CA4. (Supplementary Data Fig. 4) Images were submitted to an auto-threshold and particle’s analyzed to get a percent area measurement of BG4 staining. These were graphed and linearly fit using Kaleidagraph using Pearson fit. Spearman and Pearson R and p-values are reported for 18 cases for age correlation and 18 cases for Braak correlation. An unpaired T-test with unequal variance was performed for the two groups E2/E3 and E3/E3 vs E3/E4 and E4/E4. When the singular African American case was removed from the APOE quantification, the t-probability remained highly significant at 0.0045 vs 0.002. All quantification was performed on 10X images obtained from an Olympus microscope.

Circular Dichroism

CD spectra were obtained using a Jasco J-1100 circular dichroism at 25°C. Sequence was resuspended in 10mM potassium phosphate pH 7.5 buffer and diluted to 12.5 µM (per strand) RNA. The CD measurement was taken from 300 nm to 210 nm at 1nm intervals using a 50 nm/min scanning speed. The sequence used was 5’- CGGGCGGCGGGGGGGCCCGGGCGGCGGGGGGGCCCGGGCG-3’ ²⁸. Parallel structure was determined using previously described quantification methods for quadruplex topology³¹.

3. Results

3.1 Formation of rG4s in aging

To determine if rG4s form in the human brain, we used the well-characterized single-chain BG4 anti-quadruplex antibody (Sigma-Aldrich MABE917, antibody null controls in Supplementary Data Fig. 2) to stain human hippocampal postmortem tissue of different ages under conditions that strongly favor cytoplasmic rG4 identification using triton-X fixation in cells ^{32, 33}. The tissues were obtained in collaboration with the brain bank at the Medical University of South Carolina (MUSC). The study included 21 cases, male and female, of different ages and with or without a clinical or neuropathological diagnosis of AD (see Fig. 1b). All cases had received a neuropathological examination including assessment of Braak and CERAD stages according to the updated NIA criteria except in those cases in which no pathology was noted ^{7, 34}.

First, the rG4 staining profiles were compared between different ages, with an age range of 30-92 years of age (Fig. 1c). We discovered a significant age-related increase in rG4 immunostaining, with the densest staining observed in older individuals (Fig. 1c,d). The largest difference in staining between young and older cases was observed in the outer molecular layer (OML) and in the hilar region, CA4 (see Fig. 1c, legend arrows). The strongest overall staining for BG4 was found in the dentate gyrus granular cell layer (DG), but with less observable age-related increases. Staining density measurements confirmed these findings and demonstrated a highly significant increase in BG4 staining in the oldest cases studied within the OML (Fig. 1d) and the CA4 regions of cornu ammonis (Supplementary data Fig. 5a), confirming increased intracellular BG4 accumulation as a function of age. Plotting BG4 percent area versus age confirmed the observed findings and showed correlation (Spearman R= 0.49) with high significance (p= 0.038 in the OML region (Fig. 1d). Positive and significant Pearson but not Spearman correlations were observed between age and BG4 area in the CA4 region, Supplementary data Fig. 5a), and non-significant correlations were observed in the dentate granule cell layer. No significant correlation was found with BG4 stain and postmortem interval (PMI) (Spearman R = −0.18, p = 0.49 see Supplementary Data Fig. 6). In total, these data show that rG4s increased as a function of age in the human brain. We observed multiple outliers in both older and younger individuals due to Braak stage. For example, an 85-year-old individual with Braak stage 0, or 59-year-old individual with Braak stage 5.5 were outliers as a function of age. This led us to analyze whether rG4 formation increased with Braak stage, a measure of AD severity.

3.2 Formation of rG4s in AD

Next, we examined whether the formation of rG4s was associated with AD neuropathology. Braak staging is a neuropathological staging method that takes into account the density of neurofibrillary tangles (NFTs) both within each brain region and also from region to region³⁰, with higher numerical values representing greater AD severity⁷. Like in the case of aging, increased BG4 staining was readily apparent in AD cases, especially in those cases associated with a higher Braak stage (Fig. 2a). In the OML, plotting BG4 percent area versus Braak stage demonstrated a strong correlation (Spearman R= 0.72) with highly significantly increased BG4 staining with higher Braak stages (p = 0.00086) (Fig. 2b). Positive significant correlations (R = 0.52, p = 0.028) between Braak stage severity and BG4 staining were also observed within the CA4 region (Supplementary data fig. 4). Interestingly, the OML region is highly involved in AD pathology, and is a hippocampal region where AD pathology is often observed,³⁵ suggesting that the presence of rG4s in this particular brain region might have significance for amyloid as well as NFT formation during the AD disease process. In addition, a prominent loss of synapses in AD is found in the OML, suggesting a potential close connection with layer 2 of the entorhinal cortex³⁶.

Fig 2.

If rG4 formation is more prevalent in AD than in non-AD cases and directly involved in AD pathology, we would also predict that rG4 formation would be more prevalent in APOE genotypes that are predisposed for higher LOAD risk, i.e. patients with one or two copies of the APOE4 genotype. Indeed, there was a significant (t-probability 0.002) increase in BG4 staining levels in the OML in postmortem cases with one or more APOE4 alleles compared to those with only APOE2 or APOE3 alleles (Fig. 2c). This result, although correlative, is consistent with a previous study performed in vitro in which the RNA sequences found in aggregates in different APOE genotype cell lines were different²⁴.

3.3 Patterns of rG4 localization

Images obtained at high magnification using a confocal microscope (Olympus) showed that a large amount of the BG4 immunostaining was localized in the cytoplasm near the nucleus in a punctate pattern, reminiscent of known rG4 formation in vitro and in cell lines in rRNA ribosomal extensions³³. We therefore also stained with an anti-rRNA antibody to determine whether this was the case in human tissue as well. As expected, the stains overlapped (Fig. 3b, Fig. S7), showing that at least a large percentage of the observed BG4 signal arose from rG4s and not nuclear DNA G4s. This result also confirms that previously observed rG4 formation in ribosomes³³ also occurs in human brain tissue.

Fig 3.

To further explore the formation of rG4s in the human hippocampus, we examined the staining pattern in different cell types based on specific antibody markers. Co-staining with markers for the most common brain cell types, rG4s were clearly abundant in neurons, oligodendrocytes, and astrocytes (Fig. 3c). rG4s exhibited no apparent colocalization with microglia (Fig. 3c). This pattern is especially noticeable in the double labeling with SOX10 (Fig. 3c), which clearly shows the nuclear oligodendrocyte marker SOX10³⁷ staining in proximity to the cytoplasmic BG4 immunostaining.

We co-stained BG4 with different p-tau epitope antibodies to determine whether pathological aggregation of tau proteins co-occurred with BG4 immunostaining (Fig. 4). Indeed, we found that BG4 immunostaining co-localized with hyper-phosphorylated tau immunostaining in neurons located in the CA1-4 regions of the hippocampus (Fig. 4). We used two different antibodies directed against the serine 396 site (S396) and the threonine 231 (T231) to investigate co-localization with different p-Tau epitopes. Quantifying the percentage of cells co-stained with BG4 and pTau revealed that 85% of the S396 p-tau positive cells also had positive BG4 staining, and 82% of the T231 p-tau positive cells also had positive BG4 staining.

Fig 4.

4. Discussion

In this study, we demonstrated that rG4s form preferentially as a function of age and AD pathology in neurons and glial cells the human postmortem brain. While the granule cell layer in the dentate gyrus exhibited the most BG4 immunostaining overall, this was not an area most changed with aging and AD, as considerably greater changes were observed in the OML and the CA4 region of the hippocampus. rG4s were primarily found in the cytoplasm near the nucleus of neurons, astrocytes, and oligodendrocytes and were also found in neurons that contained regions of positive staining against two different epitopes of p-tau. The BG4 staining levels were also impacted by APOE status, with the presence of one or two APOE4 alleles associated with higher BG4 levels in the OML region. The OML is an interesting region of the hippocampus from the pathology standpoint, since this is the region of the hippocampus exhibiting the most prominent synaptic loss early in the disease process^{36, 38}. We have recently demonstrated a significant loss of presynaptic components of the OML in AD³⁸, strengthening the findings observed herein and suggesting a particular vulnerability of this region of the hippocampus to AD progressive pathology.

Combining these observations with those previously made on the reliance of rG4 structure on cellular stress and the roles of rG4s in protein oligomerization and tau aggregation in vitro, we can propose a model of how rG4s form and could perpetuate the effects of AD (Fig. 4c). As ROS and other protein stresses accumulate with age and according to APOE genotype, rG4s form. Although rG4s can rescue protein folding and could release proteins when unfolded at stress cessation³⁹, under persistent stress proteins RNA would not be released from the G-quadruplexes, leading to increased protein oligomerization and aggregation²⁶. This function has previously been suggested for RNA in biomolecular condensates more generally⁴⁰. The acceleration of protein oligomerization and aggregation by the rG4s could then contribute to a vicious cycle of increased aggregation, leading to synaptic loss as described recently for the OML³⁸. Recent in vitro work confirms an interaction between rG4 formation and p-Tau aggregation specifically⁴¹. In vitro experiments focused on the specific mechanisms for the rG4 influence on Tau phosphorylation and/or aggregation will be a future focus following these preliminary observations. Of note, the previous studies showing that the fixation conditions used here strongly favor rG4 identification instead of DNA G4 identification were performed in cells, and so it is possible that in tissue some DNA staining could also occur, however this is not suggested by the co-stain with rRNA (Fig. 3b).

Of note, the accumulation or rG4s in neurodegeneration could have roles in addition to those of protein aggregation. RNA transport dysregulation has also been a common factor identified in neurodegeneration⁴², and it could be possible that the rG4 build-up observed here could be in part a product of this process or could lead to it. It has also been proposed that the RNA transport dysregulation and protein aggregation processes could be linked ⁴³. This remains an area of needed future investigation.

This work raises the possibility of using rG4s as potential drug targets in AD, or given the high correlation with AD severity, as early AD biomarkers. Future work can focus on ameliorating AD symptoms through rG4 binding molecules or detecting brain-derived rG4s as disease biomarkers for diagnosis. Many different rG4-binding molecules have already been created and shown to bind in human cells⁴⁴, and provide a set of promising lead compounds for potential AD treatment.

Although here we have examined only AD, there are many other aging-related diseases featuring protein aggregation, including Parkinson’s Disease, ALS, Fragile X syndrome, and Huntington’s Disease ⁴⁵. Notably, all of these diseases now have significant evidence showing rG4s as interactors with the aggregating proteins ^46–58, and in the case of Fragile X, a recent study demonstrated that rG4s could contribute to its aggregation and neurotoxicity⁵⁹. In addition, a recent manuscript by Raguseo et al., showed an increased prevalence of G4-structures in C9orf72 mutant human motor neurons in ALS/FTD when compared to healthy motor neurons, and direct involvement of the rG4 complex in the aggregation process⁶⁰. These recent findings illustrate the need to explore rG4 complexes and their role in neurodegeneration further. The results here combined with these previous studies suggest that the formation of rG4s could be a common mechanism of neurodegeneration.

Acknowledgements

Funding for this project was provided by NIH R35GM142442 to S.H., and NIH 5R01AG071228-02, R01AG070153, and R01AG061566 as well as a grant from the BrightFocus Foundation (CA2018010) to A.C.G.

Additional information

Author Contributions

Idea generated by S.H., work conceptualized by S.H. and A.C.G. Experiments designed by S.H. and A.C.G, with assistance from all authors. Tissues collected and catalogued by E.D.H. Experiments performed by L.K., with assistance from H.S. and A.G. Primary writing performed by S.H., A.G.G., and L.K., with editing provided by all authors.

Authors declare no competing interests.

Additional files

Supplemental Materials