POLK subcellular expression changes with increasing age, across multiple cortical regions.

A1) Validation of anti-POLK antibody (sc-166667) by immunofluorescence on mouse primary cortical neuronal cultures treated with Polk siRNA (top row) showing a marked reduction in nuclear POLK levels after 48hrs, compared to scrambled siRNA control (bottom row). A2) Western blot of mouse primary cortical neuronal culture treated with siPOLK whole cell lysate immunostained with sc-166667 HRP anti-POLK (HRP conjugated) antibody showing ∼20% reduction in POLK protein levels compared to siControl. B) Quantitation of 120kDa and 99kDa POLK bands from western blot using sc-166667 HRP antibody shows a decreasing trend of nuclear POLK levels with progressive age from unsorted cells of the whole mouse cortex. Full blot shown in Figure S1E. C) Experimental design to longitudinally compare POLK cellular localization in mice brains. D) Immunofluorescence (IF) followed by imaging and analysis using Cell Profiler of mouse brain sections. Visualization of POLK speckles (green) expression using immunohistochemistry in cells labeled by fluorescent-nissl (red) from wild-type mouse brain aged 1 month, and 18 months. Subcellular compartments were segmented and POLK was detected and measured separately inside the nucleus and in the cytoplasm. E) Representative low (20x) and high magnification (63x) images of Polk expression using IF from Cg1, M1, M2, and S1 cortical regions in ages 1 month (N=3), 10 months (N=2), and 18 months (N=3). POLK (green) and Nissl depicting all cells (purple) (scale bar = 10um). F) Boxplot showing nuclear and cytoplasmic POLK counts per unit area for each brain region in 1C grouped by age. n denotes the numbers of cells measured per time points under each boxplot. G) Means plots with 95% confidence intervals showing the nuclear POLK speckle count and size decreasing with increasing age in Cg1, M1, M2, and S1 cortical areas. H) Means plots with 95% confidence intervals show the cytoplasmic POLK granule count decreasing with a concomitant increase in size with age in Cg1, M1, M2, and S1 cortical areas.

Absolute counts of nuclear Polk speckles

Size (area) of nuclear Polk speckles

Absolute counts of cytoplasmic Polk granules

Size (area) of cytoplasmic Polk granules

Colocalization and coexpression of POLK with DNA damage marker proteins in POLK nuclear speckles

A) Experimental design to longitudinally compare POLK colocalization with various DNA damage markers in three age group mice brains. Asterix denotes markers that were studied in three age groups, rest were evaluated in middle age only. B) Schematic of cell class gating logic, registration, sub-cellular segmentation, POLK speckle detection, measurement of POLK counts, and POLK and DNA damage markers/repair protein intensities. C1) Representative images showing colocalization of gH2AX, and channel separation of POLK, gH2AX, and POLK speckle overlay on POLK and gH2AX. POLK speckles detected inside the nucleus are outlined in yellow. IN class cells are outlined in red, PN class in green. C2) Scatterplots of gH2AX intensity measured inside POLK nuclear speckle (y-axis) were plotted against POLK intensities (x-axis) shown for INs (red) and PNs (green). The correlation coefficient and p-values are indicated for each. C3) Dot mean plots with 95% confidence error bars of gH2AX intensities in the y-axis in INs (red) and PNs (green) plotted as a function of age (x-axis). D1) Representative images showing colocalization of 53BP1, and channel separation of POLK, 53BP1, and POLK speckle overlay on POLK and 53BP1. POLK speckles detected inside the nucleus are outlined in yellow. IN class cells are outlined in red, and PN class in green. D2) Scatterplots of 53BP1 intensity measured inside POLK nuclear speckle (y-axis) were plotted against POLK intensities (x-axis) shown for INs (red) and PNs (green). The correlation coefficient and p-values are indicated for each. D3) Dot mean plots with 95% confidence error bars of 53BP1 intensities in the y-axis in INs (red) and PNs (green) plotted as a function of age (x-axis). E1) Representative images showing colocalization of 8oxo-dG, and channel separation of POLK, 8oxo-dG, and POLK speckle overlay on POLK and 8oxo-dG. POLK speckles detected inside the nucleus are outlined in yellow. IN class cells are outlined in red, and PN class in green. E2) Scatterplots of 8oxo-dG intensity measured inside POLK nuclear speckle (y-axis) were plotted against POLK intensities (x-axis) shown for INs (red) and PNs (green). The correlation coefficients and p-values are indicated for each. E3) Dot mean plots with 95% confidence error bars of 8oxo-dG intensities in the y-axis in INs (red) and PNs (green) plotted as a function of age (x-axis). F1-F3) Left column, merged 63X representative images of IN and PN, with red arrowheads showing line scan area where intensities of POLK and PRKDC (F1), POLK and APE1 (F2), and POLK and LIG3 (F3) were measured. The right columns in F1-F3 shows corresponding IN and PN-derived scatterplots where y-axis shows PRKDC intensities (F1), APE1 (F2), LIG3 (F3), plotted against POLK intensities (x-axis). Separate plots shown for INs (red) and PNs (green). The correlation coefficients and p-values are indicated for each.

Cytoplasmic POLK expression co-localizing with stress granules and lysosomal proteins

A) Schematic of cytoplasmic compartments and condensates and respective markers in parenthesis that were assayed to colocalize with cytoplasmic POLK. B_ Cytoplasmic POLK (green) expression co-localizing with G3BP1 and LAMP1 (red) in fluorescent-nissl stained cells (blue) of mouse brain tissue in 18 months old brain. Line scan in the boxed region shows POLK co-localizing with LAMP1 and G3BP1. C – F) Cytoplasmic POLK (green), EEA1, CTSB, CTSD and GBA1 (red) in fluorescent-nissl stained cells (blue) of mouse brain tissue in 18 months old brain. Line scan in the boxed region shows level of POLK co-localization with the proteins. Highest colocalization with CTSD, partial with EEA1 and CTSB, and minimal GBA1.

Nuclear POLK is differentially expressed based on cell-type in old brains.

A) Representative image of the detection of inhibitory interneurons (IN), excitatory pyramidal neurons (PN) and non-neuronal (NN) cell bodies using automated image analysis pipeline from a four-channel image of cortical areas from 19 month old brain. B) Magnified view of the dotted box from subpanel A showing overlay of the nuclear, cytoplasmic segmentation outlines and detection of nuclear POLK speckles and cytoplasmic POLK granules in IN, PN, and NN cells from wild-type mice cortical areas M1 and S1. C) Boxplots from brain areas show the nuclear POLK count is higher in INs than PNs and NNs across cortical areas in both young and old age groups.

Cell class nuclear Polk counts per area

Microglia associated with INs and PNs show significantly higher levels of cytoplasmic POLK expression.

A) Schematic of the experimental design showing age groups, cell class gating logic, scoring spatial interaction, subcellular segmentation, detection, and measurement of POLK in the cytoplasmic compartment. B) A representative single 20x imaging field shows outlines of detected cell bodies of IN (red), PN (green). Iba1+ microglia (MG) that are attached or wrapped to neurons (MG-tied, filled light pink and arrows) and microglia that are within one body length of a neuron (blue fill with an asterisk) as well as unassociated microglia (blue fill). Channel separation shown for the four-channel super-resolution confocal image, fluorescent-nissl, NeuN, and Gad67 = IN, fluorescent-nissl, and NeuN = PN, and fluorescent-nissl, and Iba1 = MG. C) Representative images from M1 and S1 cortical areas shows an overall increase in Iba1-positive cells in early old age compared to young. Yellow arrowheads point to MG-tied neurons. D) Boxplot showing the difference in cytoplasmic POLK granule median intensity between MG-free and MG-tied INs and PNs across middle, early old, and late old time age groups. T-test p-values and effect sizes are shown for each comparison.

Kainic acid-induced neuronal activity causes POLK subcellular localization in young ex-vivo brain.

A) Schematic of the experimental design showing age groups, ex-vivo treatment of kainic acid, and control groups, followed by, immunostaining, imaging, and the measurement of nuclear and cytoplasmic POLK. B) Boxplot of c-FOS levels between ACSF control and kainic acid-treated groups at 80min and 160min post-exposure intervals between young and old brains. Due to large number of observations artificially influencing significance, data was randomly subsampled 200 times, each time with 100 cells per group for Wilcoxon rank-sum (Mann–Whitney U) two-sided test. Distribution of subsampled p values and Cohen’s are shown in Figure S5A. Median p value and Cohen’s d (grey columns in Figure S5A) are reported on the plot. C) Boxplot of nuclear POLK speckle counts per unit between ACSF control and kainic acid-treated groups at 80min and 160min post-exposure intervals between young and old brains. Wilcoxon rank-sum (Mann–Whitney U) two-sided test showed a significant increase in nuclear POLK at 160min in the young brains. D) Boxplot of cytoplasmic POLK granule counts per unit area between ACSF control and kainic acid-treated groups at 80min and 160min post-exposure intervals between young and old brains. Wilcoxon rank-sum (Mann–Whitney U) two-sided test shows there is a significant decrease with small effect size in cytoplasmic granule counts upon kainic acid treatment after 160min in young brains. E) Boxplot of cytoplasmic POLK granule size measured by area contained, between ACSF control and kainic acid-treated groups at 80min and 160min post-exposure intervals between young and old brains. Due to large number of observations artificially influencing significance, data was randomly subsampled 200 times, each time with 100 cells per group for Wilcoxon rank-sum (Mann–Whitney U) two-sided test. Distribution of subsampled p values and Cohen’s are shown in Figure S5B. Median p value and Cohen’s d (grey columns in Figure S5B) are reported on the plot.

Subcellular expression of POLK is a learnable feature predictive of organismal age and IN, PN, and NN cell class from mouse brain tissue.

A1-2) Random Forest (A1) and Gradient Boosting (A2) classifier can distinguish between 1-month, 10-month, and 18-month-old age groups upon training on the subcellular nuclear and cytoplasmic distribution of POLK counts using data from Figure 1F and 1G. Top bar shows the number of cells used for training, validation, and test samples of the total. Area Under Receiver Operator Curve (AUROC) values show the performance of the classifier for each age group when tested on the holdout groups. The bar plot shows the relative contribution of the image parameters and image metadata indicating cytoplasmic POLK counts as the major driver. A3-4) Random Forest (A3) and Gradient Boosting (A4) classifier can distinguish between middle to late-old ages 10-month, 18-month, and 24-27-month age groups upon training on the subcellular nuclear and cytoplasmic distribution of POLK counts, POLK intensities in nuclear speckles and cytoplasmic granules, MG association and other metadata from Figure 5. Top bar shows the data split for training, validation and test groups. Major driver of performance is cytoplasmic POLK intensity and counts. B1-2) Random Forest (B1) and Gradient Boosting (B2) classifier can distinguish between broad IN, PN, and NN cell class with high AUROC values. Nuclear POLK per unit area and nuclear area are the major feature drivers of performance.

S1A) Protein sequence alignment of mouse and human POLK, with the 133-310 amino acid epitope for sc-166667 anti-POLK antibody showing complete homology between species. S1B-E is validation of sc-166667 POLK antibody using various mouse cells. S1B) qPCR of Polk transcript shows a 35% reduction upon siRNA against Polk but not scrambled control siRNA. Polh mRNA levels were not affected by siRNA against Polk. S1C) Western blot immunostained with SC-166667 antiPOLK-HRP antibody of whole cell lysates of two biological replicates of mouse Neuro-2A cells treated with either siPOLK or shPOLK lentivirus, showed a decrease in 99kDa POLK band. S1D) Western blot immunostained with SC-166667 antiPOLK-HRP antibody of whole cell lysates of three biological replicates of mouse 4T1 cells treated with siPOLK showed a decrease in 99kDa and 120kDa POLK bands. S1E) Western blot immunostained with SC-166667 antiPOLK-HRP antibody of whole cell lysates of two biological replicates of mouse primary cortical neurons treated with 4 individual shPOLK lentivirus, shows a decrease in 99kDa and 120kDa POLK bands upon treatment with shPOLK#C. S1F) Full western blot showing all six replicates nucleus and cytoplasmic fractions from 7 and 22 month old whole brain unsorted cells. Quantitation of the 99kDa POLK band did not show a increase, possibly due to inability to extract POLK associated with lysosomal and stress granules (as observed in Figure 3) S1G) Immunofluorescence staining of wild-type 18-month-old mouse, from brain cortical area S1 shows a similar pattern and distribution of POLK nuclear speckles (arrowheads) and cytoplasmic granules (arrows) using sc-166667 and A12052. The bottom row is a crop of the boxed area in the corresponding top image. S1H) Representative low (20x) and high magnification (63x) images of REV1 and POLI expression using IF from S1 and M1 cortical regions in ages 1 month, 10 months, and 18 months. REV1 (green) and Nissl depicting all cells (purple) (scale bar = 10um in 20x image). Arrowheads point nuclear speckles and arrows indicate cytoplasmic granules. Wild-type mouse brain cortical areas S1 and M1 shows REV1 and POLI punctate nuclear expression resembling speckles and progressive cytoplasmic accumulation with age at 10 and 18 months.

Scatter plots showing intensity-intensity plots from INs (red) and PNs (green), where y-axis is KU70, XRCC1, and XRCC4, intensity levels plotted against POLK intensities in the x-axis.

The correlation coefficients and p-values are indicated for each.

S3A) Cytoplasmic POLK (green) expression co-localizing with G3BP1 (blue) in fluorescent-nissl stained cells (purple) of mouse brain tissue from M1 and S1 cortical regions in 18months old brain but not in young 1 month. S3B) Cytoplasmic POLK (green) expression co-localizing with LAMP1 (blue) in fluorescent-nissl stained cells (purple) of mouse brain tissue from M1 and S1 cortical regions of 18-month brain. S3C) Additional representative image with channel separation for immunofluorescence staining of wild-type mouse brain cortical areas S1, showing cytoplasmic POLK is colocalized with stress granule marker G3BP1 and endo/lysosomal marker LAMP1. Arrows indicate few representative sites of colocalization in both images.

Boxplots comparing the nuclear POLK median intensity and cytoplasmic POLK granule count between MG-free and MG-tied INs and PNs across middle, early-old, and late-old time age groups.

T-test p-values are shown for each comparison.

Boxplots showing the distribution of p-values and Cohen’s d values after performing nonparametric test by 200 repeated random subsampling at 100 cells per group from Figure 6B and 6E.

Shaded columns shows the values reported in Figure 6B and 6E.