Nuclear protein accumulation in cellular senescence and organismal aging revealed with a novel single-cell resolution fluorescence microscopy assay

Total protein content of early passage and senescent fibroblasts in culture

We first confirmed previous reports that senescent cells contain more protein using the strain of normal human diploid fibroblasts (HDF) commonly used in our lab (LF1, see Methods). Total protein was extracted from early passage, senescent, or early passage quiescent cells, and assayed using the EZQ Protein Quantitation Kit (Invitrogen, Supplemental Methods). Additionally, we fractionated whole cells into cytoplasmic and nuclear extracts, and determined their relative total protein concentrations (Supplemental Figure 1A). The values were normalized to either cell number, DNA content, or histone H2B content (Supplement Figure 1B). Over a large number of experiments we saw consistent increases in total, cytoplasmic and nuclear protein content. These data are in good agreement with previous reports [16, 28, 29].

A fluorescence microscopy method for total protein staining and measurement

We next stained cells grown on glass cover slips using the NanoOrange dye, which was originally developed for highly sensitive biochemical detection of proteins in solution by Molecular Probes [30]. We systematically explored the parameters affecting the performance of the assay (Supplemental Figure 2), and developed a protocol that provides highly consistent staining of whole cells (Methods). Notably, our staining method is stable over a wide range of paraformaldehyde fixation times, affording excellent preservation of cellular morphological features. As expected from the biochemical fractionation experiments, stained preparations of senescent HDF observed by confocal fluorescence microscopy displayed a clear increase of both cytoplasmic and nuclear signals (Figure 1A). To extend the generality of these findings, we also investigated senescent mouse fibroblasts and observed very similar increases of cytoplasmic and nuclear protein content (Figure 1D). We used cultures of mouse tail fibroblasts (MTF) prepared freshly from adult biopsy material (Methods), which in contrast to mouse embryo fibroblasts (MEF) undergo senescence even under physiological (2.5%) oxygen levels [31], and display characteristic increases of SA-b-Gal and senescence regulators such as p16^Ink4a (Cdkn2a [32] and our unpublished observations).

We subsequently focused on quantitative image analyses of the nucleus. Senescence-associated changes in protein content of this compartment have not been widely studied. More importantly, the development of methods for the examination of single cells in complex tissues depends on the ability to unambiguously assign the regions of interest, which in the case of the nucleus can be done accurately by DAPI staining. To enable the high throughput necessary for statistical significance, we applied the open source software CellProfiler and CellProfiler Analyst, which allows the analysis of thousands of images through automated pipelines (Methods).

We documented a highly significant senescence-associated increase in total nuclear protein in both HDF (Figure 1B, C) and MTF (Figure 1E, F), while as expected, DNA content did not change significantly. Interestingly, the mean protein intensities increased 1.93-fold for HDF and 1.64-fold for MTF (Table 1), suggesting that the nuclei had become more dense in protein. Since the DNA signals did not change significantly, the ratio of protein to DNA intensity also increased, 1.84-fold in HDF and 1.82-fold in MTF, further confirming the increase in protein density. Hence, the senescence-associated change in nuclear protein content can be conveniently observed and documented as an increase of DAPI-normalized mean NanoOrange intensity.

Multiparameter immunofluorescence and NanoOrange protein staining

The utility of the NanoOrange staining of total protein would be greatly enhanced if it could be combined with immunological detection of specific proteins. Preliminary experiments were performed to test whether the NanoOrange dye would interfere with antibody detection, or whether the extensive washing of immunofluorescence protocols would decrease the dye signal. Repeated trials where the same specimens were processed in triplicates (NanoOrange alone, immunofluorescence alone, NanoOrange and immunofluorescence combined) and analyzed for all parameters with CellProfiler did not reveal any significant interference (data not shown).

We subsequently investigated using these multiparameter assays major structural nuclear proteins (lamin A/C, LMNA), histones (histone H3), and proteins that have been documented to increase in senescent cells (the heterochromatin component macro H2A, mH2A). LMNA levels did not differ between early passage and senescent cells, either HDF (Fig. 2A-C) or MTF (Fig. 2D-F). In the same specimens, total protein was increased as observed previously. Thus, the increase in total protein is not a consequence of a general increase of all nuclear proteins, and conversely, the major structural nuclear component LMNA cannot account for the increase of total nuclear protein.

The same pattern was observed in a mutiparameter assay scoring histone H3 (H3) by immunofluorescence and total protein by NanoOrange staining: H3 did not increase, while total protein did (Figure 3A, B). The lack of H3 increase would be expected from the constant DNA signal, and indicates that the increase in total protein is unlikely to be accounted for by major changes in the DNA/histone ratio of chromatin. In support, the levels of histone H2B were not observed to increase in senescent cells by immunoblotting (Supplemental Figure 1B). It should be noted that the absence of increase in LMNA or H3 was not due to effects of NanoOrange staining, since (as noted above) single parameter assays for either LMNA or H3 gave the same results.

In contrast, a multiparameter assay scoring mH2A by immunofluorescence and total protein by NanoOrange staining showed the typical increase in total protein (Figure 4A-C), as well as the previously documented [11] increase in mH2A (Figure 4D-F) in the same specimens. This further underscores the multiplexing capability of NanoOrange staining with immunofluorescence. Since mH2A is a relatively minor component of chromatin, being typically found at levels some 2-orders of magnitude below canonical histones [33], its increase is unlikely to account for a significant fraction of the total nuclear protein increase.

Application of NanoOrange staining to tissue specimens

The eventual goal of the work presented here was to develop a novel single-cell bioassay to monitor age associated changes in mammalian tissues. We therefore examined liver sections obtained from healthy young (5 months) and old (36 months) male mice (Figure 5A-C). Liver tissue contains very abundant cytoplasmic proteins resulting in a bright NanoOrange signal. We focused our analysis on the nuclei, which were defined on the basis of DAPI staining. Compared to cells grown on cover slips, tissues such as liver contain nuclei oriented in 3 dimensions. Since the CellProfiler software currently cannot process z-axis stacks of images, we took a single optical section through the center of the tissue slice, and set software filters to retain only the largest objects in the DAPI channel, thus focusing the analysis on nuclei that were bisected close to their centers. This eliminates nuclei that were bisected close to one edge, and minimizes out of focus NanoOrange signals from over (or under) lying cytoplasm. All data were internally normalized to DNA content, and are expressed as ratios of mean NanoOrange to mean DAPI intensities acquired from the same optical section.

The mean protein intensity of the old nuclei was some 25% greater than that of the young nuclei (0.36 and 0.29, respectively). Notably, the two histogram distributions had different shapes: the old distribution contained a significant fraction of high intensity nuclei and commensurately fewer low intensity nuclei (Figure 5C). Given the large number of nuclei (>4000 for each condition) scored by the CellProfiler pipeline, the two histogram distributions are statistically highly significantly different. In addition, since the software keeps track of each individual nucleus, the high intensity nuclei can be extracted from the dataset and scored for additional parameters.

We also noted that at high magnification young liver nuclei displayed relatively uniform NanoOrange staining, while almost all the old nuclei contained prominent bright foci (Figure 5A,B, lower panels). These foci largely colocalized with bright foci in the DAPI channel, which in the mouse are known to demarcate regions of centromeric and pericentromeric heterochromatin. It is interesting that while the DAPI foci did not change appreciably between young and old specimens, only in the old tissues NanoOrange staining increased in these regions. This suggests that the protein density of centromeric heterochromatin increases with age.

Cell lines, culture conditions, and fixation of specimens

LF1 is a normal human diploid fibroblast (HDF) cell strain derived from first trimester female embryonic lung tissue [42]. HDF were cultured as described [11] in an atmosphere of 2.5% O₂, 5% CO₂, and 92.5% N₂. Primary adult mouse tail fibroblasts (MTF) were prepared from tail biopsies of 8-12 week old C57 mice as described [31]. MTF were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS under the same atmospheric conditions as HDF. Both HDF and MTF were routinely subcultured at 1:4 dilution upon reaching 80% confluence. Under our conditions HDF reach senescence at passage 50-52, and MTF at passage 11-13. Early passage HDF used in this study were at passage 15 and MTF at passage 3. HDF were cultured until proliferation ceased (monitored by daily microscopic observation) and incubated for 2 more weeks prior to harvest. MTF were monitored at each passage using the SA-b-Gal assay [9], and harvested at passage 13 when >80% of the cells were positive. For fluorescence microscopy cells were seeded onto 22 x 22 mm glass cover slips and culture was continued for at least 24 hr. Cover slips were always harvested when cells were sub-confluent, fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, and permeabilized with 0.2% Triton X-100 as described [11].

Antibodies

The rabbit polyclonal antibody to mH2A was directly labeled with Alexa Fluor 647 using a monoclonal antibody labeling kit (Invitrogen, A-20186) as described [11]. The mouse monoclonal antibody to human LMNA was from Millipore (MAB3211), and the goat polyclonal antibody to mouse LMNA was from Santa Cruz (sc-6215). Both antibodies target a region in common between the lamin A and C proteins (amino acids 464-572). Mouse monoclonal antibody to histone H3 was from Active Motif (mAb 39763). When required, appropriate Alexa Fluor 647 conjugated secondary antibodies were obtained from Invitrogen.

Staining of cells with NanoOrange

Staining was performed in 6-well cell culture clusters, following the procedure of Jones et al. [30], which was developed for the staining of proteins in solution, with some modifications. During the development of our protocol we varied several parameters (temperature and time of staining, dye concentration, effect of chaotropic agents) to investigate the robustness of the assay (see Supplemental Figure 2). We also varied the time of paraformaldehyde fixation from 5 min to 60 min and observed that this did not have a significant effect on the extent of staining (data not shown). All steps were performed at room temperature unless otherwise stated, and the specimens were protected from direct light at all times. One cover slip was placed in each well, fixed as above, and washed once with PBS containing 0.2% Triton X-100 for 2-5 min. The NanoOrange Protein Quantitation kit was purchased from Invitrogen (Cat. No. N-6666). NanoOrange protein quantitation reagent was diluted to 6.25 μl/ml with 1 x NanoOrange protein quantitation diluent and 3 ml was added to each well. The plate was wrapped in Saran Wrap and then in aluminum foil (to prevent desiccation and protect from light), and incubated at 95°C for 20 min in a hybridization oven. The plate was then allowed to cool and equilibrate to room temperature for 40 min, the staining solution was removed, the cover slips were washed 3 times for 5 min in PBS, nuclei were counterstained with 2 mg/ml DAPI in PBS for 20 min, and mounted for microscopy as described [11].

Combination of NanoOrange staining with immunofluorescence

The NanoOrange staining protocol (above) was performed up to the DAPI counterstaining step. Following the 3 PBS washes, immunofluorescence staining was initiated at the nonspecific binding blocking step, and continued essentially as described [11]. Three different blocking buffers were used, depending on the primary antibody: human LMNA, 3% donkey serum in PBS; mouse LMNA, 10% rabbit serum in PBS; mH2A and histone H3, 4% bovine serum albumin, 2% donkey serum, 2% rabbit serum in PBS. Primary antibodies were used at saturating concentrations, and incubated for 2 hr at room temperature (human LMNA, mouse LMNA) or overnight at 4 °C (mH2A, histone H3). As with the NanoOrange staining, all steps were performed at room temperature unless otherwise stated, and the specimens were protected from direct light at all times.

Processing of mouse liver tissue

Dissection, embedding and cryosectioning were performed as described [11]. Following the fixation and permeabilization step, the specimens were stained with NanoOrange as indicated above.

Quantitative imaging

Images were acquired using a Zeiss LSM 710 Confocal Laser Scanning Microscope. All microscope settings were set to collect images below saturation and were kept constant for all images taken in one experiment, as previously described [11]. All images were collected at 16-bit resolution in order to maximize the dynamic range of the detected intensities. Images were analyzed with the open source software CellProfiler [40] and CellProfiler Analyst [41]. ImageJ from the NIH (http://rsbweb.nih.gov/ij/) was used to convert the ZEN format files generated by the Zeiss software into TIFF files recognized by the CellProfiler software. The analysis was automated by the development of Cell Profiles pipelines (available on request), allowing the processing of large numbers of images and recording all values in databases. For each database values from at least 500 cells were compiled. Nuclei were defined using the DAPI channel (435-485 nm emission), and the total (sum of all pixels) and mean (total/number of pixels) intensities within this region were recorded for the other fluorophores: NanoOrange (560-620 nm emission), Alexa 647 (665 nm long pass emission). NanoOrange has an excitation maximum in the range of 470-490 nm, and can thus be efficiently excited by the 488 nm laser. However, both the excitation and emission peaks are rather broad (400-550 nm and 500-650 nm), making spectral overlap a significant problem in combination with dyes in the FITC/Cy2/Alexa 488 or Rhodamine/Cy3/Alexa 555 channels. However, NanoOrange can be very effectively used in combination with red-shifted dyes such as Cy5/Alexa 647. Fluorescence values were expressed as mean values in arbitrary fluorescence units (a.u.). In experiments with mouse liver tissue the mean intensities of NanoOrange were normalized to mean DAPI intensities for each nucleus. Data were displayed as histograms with nuclei distributed into bins based on their intensity levels, and plotted with the value of each bin shown as % of total nuclei. Statistical significance (p values) were assessed using the two-tailed Student's test (t-test), unless otherwise indicated. Error bars, where shown, designate standard deviations.