Super-resolution structure of DNA significantly differs in buccal cells of controls and Alzheimer's patients.

The advent of super-resolution microscopy allowed for new insights into cellular and physiological processes of normal and diseased cells. In this study, we report for the first time on the super-resolved DNA structure of buccal cells from patients with Alzheimer's disease (AD) versus age- and gender-matched healthy, non-caregiver controls. In this super-resolution study cohort of 74 participants, buccal cells were collected and their spatial DNA organization in the nucleus examined by 3D Structured Illumination Microscopy (3D-SIM). Quantitation of the super-resolution DNA structure revealed that the nuclear super-resolution DNA structure of individuals with AD significantly differs from that of their controls (p < 0.05) with an overall increase in the measured DNA-free/poor spaces. This represents a significant increase in the interchromatin compartment. We also find that the DNA structure of AD significantly differs in mild, moderate, and severe disease with respect to the DNA-containing and DNA-free/poor spaces. We conclude that whole genome remodeling is a feature of buccal cells in AD.


| Spatial organization of the nucleus
The concept of nuclear architecture linked to proper cellular function is established since the early work of Boveri (1914Boveri ( , 1929. Sophisticated experimental and analytical tools have become available since then. These technological advances have enabled today's researchers to add detail to Boveri's concept. We have learned that chromosome territories are the structure of chromosomes in interphase nuclei (Cremer & Cremer, 2001;Cremer et al., 2006;Fritz et al., 2016;Sehgal et al., 2016a) and were able to depict all of them together using successive hybridizations (Bolzer et al., 2005). We learned that chromosome positions are conserved during evolution (Tanabe et al., 2002) and that gene-poor and gene-dense chromosomes have their specific nuclear location in human cells (Tanabe et al., 2002).
Chromosome positioning is not a constant feature of nuclei as positional changes were recorded during keratinocyte differentiation (Sehgal et al., 2016b), during adipocyte differentiation (Charó, of chromosome territories differs in the rod nuclei of retinas in diurnal and nocturnal animals: In diurnal animals across the phylogenetic tree, most heterochromatin is observed at the nuclear periphery and euchromatin in the nuclear interior (Solovei et al., 2009). The rod nuclei of nocturnal animals; however, showed an inverted pattern with heterochromatin found in the nuclear center and euchromatin at the nuclear border. The authors concluded that this inverted organization of chromatin in rod cell nuclei acts as collecting lenses in nocturnal animals.

| Super-resolution microscopy
The 2014 Nobel Prize in Chemistry was awarded to Drs. Eric Betzig, William E. Moerner, and Stefan Hell for "the development of superresolved fluorescence microscopy," a transformative breakthrough that allows imaging of objects far smaller than the Abbe diffractionlimited spatial resolution (200 nm) and enables the study of individual molecules at ≤20 nm. While others have used this new approach to study open questions in immunology (Hu, Cang, & Lillemeier, 2016), cell biology and DNA repair (Lopez Perez et al., 2016), nuclear envelope and nuclear pores (Schermelleh et al., 2008;Xie, Horn, & Wright, 2016), in cells undergoing ischemic stress (Kirmes et al., 2015), and plant biology (Urban, Barclay, Sivaguru, Punyasena, 2016), we were the first to apply this technology to the study of the cancer cell genome (Righolt et al., 2014;Righolt, Knecht, & Mai, 2016;Sathitruangsak et al., 2015).
We examined nuclear DNA organization of normal, premalignant, and malignant cells. Nuclei of cancer cells showed more DNA-free and DNA-poor spaces compared to normal cells (Righolt et al., 2014Sathitruangsak et al., 2015). We developed a program to quantify the organization of nuclear DNA, using a measurement called granulometry, and applied it to the acquired super resolution images (Righolt et al., 2014Sathitruangsak et al., 2015). In these images, the granulometry of the image ("light granulometry") quantifies the DNA structure, and granulometry of the negative ("dark granulometry") quantifies the spaces void of DNA structure (Righolt et al., 2014). An increase in dark granulometry shows that DNA-poor space is increased in frequency, size, or both (Righolt et al., 2014). 1.3 | Alzheimer's and nuclear structural markers of Alzheimer's disease Alzheimer's disease (AD) is an increasing challenge for society.
According to the Alzheimer's Society, an estimated 5.4 million Americans of all ages live with AD in 2016. This number includes an estimated 5.2 million people age 65 and older (Alzheimer's Association, 2016;Hebert, Weuve, Scherr, & Evans, 2013), and approximately 200,000 individuals under the age of 65 who have younger-onset AD (Alzheimer's Association, 2016;Hebert et al., 2013). World-wide, nearly 44 million people suffer from AD and related dementias (http://www.alzheimers.net/resources/alzheimers-statistics/). The costs to the health care systems are immense: Currently, the costs amount to $236 billion in the United States. In 2050, it is expected to reach $1 trillion (http://www.alz.org/facts/). No treatment for the disease that slows or reverses it has been identified to date.
In a previous study, we have focused on the spatial organization of telomeres in buccal cells of AD (Mathur et al., 2014). Buccal cells are of neuroectodermal origin and therefore considered a suitable model for studies into changes observed in the brain (Forero et al., 2016;Mathur et al., 2014;Thomas, O'Callaghan, & Fenech, 2008). Our previous study (Mathur et al., 2014) was a blinded study with 41 AD patients and 41 age-and gender-matched caregivers. We measured the 3D nuclear telomere organization using TeloView  and showed that the 3D telomeric organization in buccal cells of the 82 study participants was significantly different in AD and non-AD.
Moreover, the 3D telomeric profiles were significantly different between mild, moderate, and severe AD patients. The 3D profiles described alteration in the number of telomeric signals detected, the telomere length measured and the aggregation of telomeric signals (Mathur et al., 2014). A recent meta analysis that focuses on telomere length alone has confirmed our findings of shorter telomeres in AD as a general feature of AD (Forero et al., 2016).
In the current study, we have investigated the nuclear organization of DNA in AD and non-AD using quantitative 3D-SIM super resolution microscopy. Our non-AD control participants were genderand age-matched non-caregivers as caregivers were shown to exhibit telomere shortening as a result of constant and long-lasting daily stress (Damjanovic et al., 2007;O'Donovan et al., 2012). Our data show for the first time evidence for AD-specific nuclear DNA organization. were classified as severe AD (Table 1) (Folstein et al., 1975;McDowell et al., 1997;Nasreddine et al., 1995). All AD patients were treated with cholinesterase inhibitors according to the Canadian guidelines for treatment of dementia (Gauthier et al., 2012;Herrmann & Gauthier, 2008
A coverslip specially suited for super-resolution microscopy was added (No. 1 ½, Schott, Mainz, Germany), sealed with nail polish and imaged immediately.

| Image processing
The image processing and measurement steps were performed in Matlab (MathWorks, Natick, MA) with the toolbox DIPimage (Luengo Hendriks et al., 1999). A central z-plane was manually selected for processing (Righolt et al., 2014). The nucleus was automatically detected by isodata thresholding (Ridler & Calvard, 1978) the widefield DAPI image and filling the holes in the binary image. The greyscale DAPI images were errorfunction clipped between the 10th and 90th percentile of the intensity over the detected cell (Verbeek & van Vliet, 1993

| Nuclear aspect ratio measurements
Nuclear aspect ratios of all 2,220 nuclei were manually measured as shown in Supplemental Fig. 1. Line measurements of the "major" and "minor" lengths were done through each nucleus and the differences between the two calculated.

| Statistical analyses
The coefficient of variation (the standard deviation divided by the mean) and skewness of the intensity histogram over the detected region was computed as well. To assess the significance of the measured difference, we used two-sided, two-sample Kolmogorov-Smirnov (KS) tests. A linear classification line based on the Fisher linear discriminant assuming equal priors was performed for the histogram features using the PRTools toolbox for Matlab (Duin, 2007). The structure of the DNA and DNA-poor space were compared between cases and controls by controlling for both granular size and matching GARCIA ET AL.
| 2389 using a general linear models (GLM). Abnormal/normal cell proportions were analyzed by Mantel-Haenszel stratified analysis. The nuclear aspect ratio data were analyzed using the GLM procedure.

| Buccal cell analysis of Alzheimer's disease (AD) and matched controls
We examined buccal cells of 74 study participants to define the DNA super-resolution structure of their nuclei based on 3D Structured Illumination Microscopy (3D-SIM), a super resolution imaging method.  (Folstein et al., 1975;McDowell et al., 1997;Nasreddine et al., 1995). Scoring of the AD study participants was carried out by a clinician (AG).

| Super-resolution DNA structure of buccal cells of AD and non-AD study participants
Buccal cells of all participants were imaged using 3D-SIM and subjected to granulometry analyses (Righolt et al., 2014). Light granulometry represents DNA structure as detected by 3D-SIM of 4′,6-diamidino-2-phenylindole (DAPI)-stained nuclei. In contrast, dark granulometry refers to spaces void of DAPI signals and thus spaces representing DNA-free/poor spaces detected in nuclei or the interchromatin compartment. Figure 1 illustrates the DNA structure of representative buccal cell nuclei in AD and non-AD (dark vs. light granulometries). The organization of the DNA differs significantly between normal ageand gender-matched individuals and the AD patients that scored with mild, moderate, or severe disease: Granulometry measurements were carried out (Table 2, Fig. 2). Figure 2 shows representative examples of light and dark granulometries representing the DNA-free spaces for mild, moderate, and severe AD versus non-AD controls, respectively ( Fig. 2 Panel A and B). Statistical analyses were performed on 2,220 nuclei from 74 study participants (i.e., on 30 nuclei per person) that had been imaged with 3D-SIM and analyzed by light and dark granulometries (Table 2). For all AD severities combined and compared to the non-AD controls, we found significant alterations in the DNA-poor spaces where dark granulometry increased (p < 0.0001). In contrast, the amount of DNA measured (light granulometry) was not significantly different (p = 0.6859). These data indicate that, as expected, no DNA is lost or gained between non-AD controls and AD subjects. However, the spatial organization of the DNA in the 3D nucleus is significantly remodeled in AD leaving these nuclei with more DNA-free/poor space and therefore a larger interchromatin compartment.
When individual comparisons were performed, distinctive features for each group became apparent. Buccal cells of mild AD manifest significant organizational differences in both light and dark granulometries (p < 0.0001 and p = 0.0009, respectively). Buccal cells of moderate AD display light granulometries distinct from those of non-AD (p < 0.0001) but not dark granulometries (p = 0.5704). Buccal FIGURE 1 Representative 3D-SIM images of buccal cell nuclei collected from Alzheimer's (AD) and controls and their respective light and dark granulometry images. Buccal cells were collected, placed onto slides and stained as described (Section 2). Buccal cell nuclei from mild, moderate and severe AD were compared to their age-and gender-matched non-AD controls. Top images: reconstructed 3D SIM images, middle images: light granulometry images; bottom images: dark granulometry images. For statistical analyses, see Table 2. Scale bar: 1 μm cells of severe AD exhibit significant differences in dark granulometries (p < 0.0001), while light granulometries are similar to those of the controls (p = 0.1427). These data indicate dynamic changes to the 3D nuclear organization of DNA during the onset and progression of the disease (Table 2).

| Nuclear aspect ratios of buccal cells from AD and non-AD study participants
While performing the above studies, we noted changes in the nuclear shapes of buccal cells at 3D-SIM resolution. We examined the nuclear dimensions (minor, major, and difference) for the above 2,220 nuclei as a simple evaluation method for nuclear morphology (Section 2 and Supplemental Fig. 1). While it is known that buccal cells have several sub-types (François et al., 2016), we have not selected for any subtype but rather collected data from all types present in a study participants' collected sample. Combining all buccal of all AD groups compared to all buccal cells of the controls yielded significant differences in the nuclear aspect ratios (p = 0.0296) (Supplemental Table S 1). This result indicates that the overall shape of the nuclei is more ellipsoid in buccal cells of AD than in non-AD. When this measurement is broken down into the AD subgroups of mild, moderate, and severe, it is not a strong discriminator between the AD subgroups (Supplemental Table S 1).

| DISCUSSION
The data presented in this study highlight novel features of the nuclear architecture of buccal cells of patients with AD compared to genderand age-matched controls. Our data demonstrate a significant increase in DNA-free/poor space in AD accompanied by some changes in nuclear morphology.

| 3D nuclear space and genomic remodeling in disease
This work illustrates that the 3D nuclear organization of the genome is  D'Esposito, & Bickmore, 2007;Mehta et al., 2010). It is in this general context that Meaburn (2016), in a recent review, poses the question whether the spatial organization of the genome may serve as a potential biomarker of disease. She writes, "Since spatial reorganization of the genome has been identified in multiple human diseases, it is likely that spatial genome positioning patterns as a diagnostic biomarker may be applied to many diseases." With respect to AD, the super-resolved organization of the genomic DNA structure may be stage-specific as our study indicates.
This conclusion complements our previous study that demonstrated the nuclear remodeling of telomeres in the 3D space of AD versus non-AD as a stage-specific event for mild, moderate, and severe AD versus controls (Mathur et al., 2014). In addition to the remodeling, telomeres were also significantly shorter, and this was confirmed in a recent meta-analysis (Forero et al., 2016).
There may be dynamic changes in the organization of the genome structure and its associated function that enable the remodeling of the genome in AD.
While not yet reported for AD, other studies have reported that nuclear re-organization occurs and may be linked to neuronal differentiation and disease. In a recent review, Wilczynski    Table 2 2014, 2016; Sathitruangsak et al., 2015). The data of the current study on the super resolution structure of AD versus non-AD indicate that the IC compartment increases significantly in AD.
Together, all available data suggest that the previously discussed models of the interchromatin domain and interchromosomal network arrangement of the genome (reviewed in Branco & Pombo, 2006) can be tentatively answered in favor of the interchromosomal model that assumes spaces in between chromosomes. In contrast, the interchromosomal network model assumes that chromosome territories mix and form contacts with each other in the absence of interchromosomal space. This latter model is not supported by the experimental evidence revealed by super-resolution studies. In fact, all data obtained by super resolution suggest that the IC is a dynamic feature of the genome and its overall plasticity. This holds true for AD as well as shown in the current study. In conclusion, buccal cells of AD patients show a significant increase in interchromosomal space.