Advances in mass spectrometry have allowed the high throughput analysis of protein expression and modification in a host of tissues. Combined with subcellular fractionation and disease models, quantitative mass spectrometry and bioinformatics can reveal new properties in biological systems. The method described herein analyzes chromatin-associated proteins in the setting of heart disease and is readily applicable to other in vivo models of human disease.
In the nucleus reside the proteomes whose functions are most intimately linked with gene regulation. Adult mammalian cardiomyocyte nuclei are unique due to the high percentage of binucleated cells,1 the predominantly heterochromatic state of the DNA, and the non-dividing nature of the cardiomyocyte which renders adult nuclei in a permanent state of interphase.2 Transcriptional regulation during development and disease have been well studied in this organ,3-5 but what remains relatively unexplored is the role played by the nuclear proteins responsible for DNA packaging and expression, and how these proteins control changes in transcriptional programs that occur during disease.6 In the developed world, heart disease is the number one cause of mortality for both men and women.7 Insight on how nuclear proteins cooperate to regulate the progression of this disease is critical for advancing the current treatment options.
Mass spectrometry is the ideal tool for addressing these questions as it allows for an unbiased annotation of the nuclear proteome and relative quantification for how the abundance of these proteins changes with disease. While there have been several proteomic studies for mammalian nuclear protein complexes,8-13 until recently14 there has been only one study examining the cardiac nuclear proteome, and it considered the entire nucleus, rather than exploring the proteome at the level of nuclear sub compartments.15 In large part, this shortage of work is due to the difficulty of isolating cardiac nuclei. Cardiac nuclei occur within a rigid and dense actin-myosin apparatus to which they are connected via multiple extensions from the endoplasmic reticulum, to the extent that myocyte contraction alters their overall shape.16 Additionally, cardiomyocytes are 40% mitochondria by volume17 which necessitates enrichment of the nucleus apart from the other organelles. Here we describe a protocol for cardiac nuclear enrichment and further fractionation into biologically-relevant compartments. Furthermore, we detail methods for label-free quantitative mass spectrometric dissection of these fractions-techniques amenable to in vivo experimentation in various animal models and organ systems where metabolic labeling is not feasible.
The experimental workflow contains seven major steps (Figure 1). For any work involving samples that will be run on the mass spectrometer, the experimenter should wear a lab coat, gloves and hair net and take care to avoid contamination from dust and personal shedding of keratin.
1. Heart Homogenization and Nuclear Isolation
Mouse hearts are homogenized and an intact nuclei pellet is isolated (Figure 2).
2. Nucleoplasm and Detergent-extracted Chromatin Fractionation
The crude nuclei pellet is separated into a nucleoplasm and detergent-extracted chromatin fraction, containing proteins loosely associated with the DNA.
3. Acid-extraction Fractionation – DNA-bound Protein Enrichment
A separate fractionation enriches for proteins tightly bound to the DNA, including histones.
4. Purity Check
Western blots and electron microscopy confirm successful enrichment of nuclei and depletion of other organelles. To see typical results for all of the following quality control assays, please see our previous publication.18
Important considerations for enriching versus purifying nuclei. This protocol was developed to enable proteomic analyses of cardiac chromatin for the purpose of studying global gene regulation processes during disease. A major component of designing whole-proteome mass spectrometry experiments is the issue of dynamic range and being able to identify and characterize low-abundance proteins. In addition to the primary goal of providing information on nuclear and chromatin-specific biology, enriching for the nuclear proteome, increases the likeliness of detecting these low-abundance proteins, as does further subfractionation of the nuclei. However, as discussed, the adult cardiomyocyte has a dense cytoskeletal component connected to the nuclear membrane. We do not believe that we have completely purified the nuclei from these other cellular components. However, by enriching only for the nuclei, we have obtained an acceptable threshold to enable the types of analyses described.
Specifically, we have used EM to assess the purity of our crude nuclei pellet and found 60-80% of the fraction to be intact nuclei, roughly 10% mitochondria, and the rest debris. Additionally, we know from our previous study on the intact nuclei population, that while many myofilaments are not enriched by this protocol (including tubulin and actin as measured by Western) certain proteins (calsarcin-1) are enriched, suggesting a true biological population of these proteins in the cardiac nuclei.19
Additionally, we compared the proteins we found to be present in the acid-extracted chromatin fraction, to the predicted roles of these proteins by gene ontology. Importantly, this gene ontology analysis does not take into account the relative abundance of the different proteins (as does our Western blot data) but rather counts all proteins identified (enriched or not) as equal when identifying common pathways and cellular compartments. Analysis of these pathways and cellular compartments can be found in our previous publications.18,20 Most importantly, the success of the enrichment in the acid-extracted chromatin fraction allowed us to identify the presence of 54 histone variants in the adult mouse myocyte,18 many of which relied on one unique peptide for identification, and thus would likely not have been detectable without this enrichment protocol given the enormous complexity of the total cardiomyocyte proteome. Of the 1,048 proteins we identified from the cardiac nuclei, 56 of them (5.3%) were annotated by GO analysis to be part of the nucleosome (one component of the nucleus of interest). Another study looking at the whole heart, identified 6,180 proteins, of which only 11 proteins (0.18%) were annotated to be part of the nucleosome21. This further illustrates the strength of our protocol to meaningfully enrich for nuclear proteins.
5. Protein Gel and Enzymatic Protein Digest
Proteins are separated by one dimensional SDS-PAGE and trypsin digested to be run on the mass spectrometer.
6. Mass Spectrometry and Data Analysis
Samples are separated on an LC and analyzed by MS/MS. The spectra are searched against a protein database for protein identification.
7. Label-free Quantitation
Determine the relative abundance of proteins using label-free quantitation (Figure 4).
Important considerations for label-free quantitation. When performing label-free quantitation, specific attention must be given to ensure consistent sample preparation, digestion time and LC-MS/MS conditions as each sample must be processed and analyzed separately. In contrast to metabolic labeling approaches, comparisons in label-free experiments are made on data from distinct mass spectrometry runs (because the lack of labeling obviates the possibility of running them together), thereby necessitating high reproducibility in all aspects of sample analysis (i.e. of sample prep, LC and MS steps) and use of high mass accuracy mass spectrometers.
To account for slight changes in sample preparation and analysis a known standard may be added to each sample to assist in normalization of data. Additionally, most software programs allow normalization of signal intensities (e.g. by adjusting to background noise or a known abundant analyte) after data acquisition to account for differences in injection, ionization and fragmentation. Alignment algorithms exist in most of the above-referenced software programs that assist in correcting differences in peptide elution profiles. The use of biological and technical replicates is essential in this type of study, as it allows statistical analyses to confirm the reproducibility and consistency of any observed changes in protein abundance.
Figure 4 highlights the utility of this form of relative quantification. Shown in the left panel are the individual monoisotopic peptide peaks (overlaid from different mice), which have been designated as belonging to the protein HMGB1 (identified via database search). Each peak, essentially an extracted ion chromatograph for the given peptide, comes from a different mouse. The groups represent three different physiological states: basal, cardiac hypertrophy, and heart failure, with three biological replicates for each group. By integrating the area under the curve, a relative abundance can be calculated. In the middle panel the average of these abundances is shown. It is clear from this type of analysis that the abundance of the HMGB1 protein is changing during the course of the disease. In contrast, the bottom panels show the example of histone H4 that does not change with disease. In both cases, peptides coming from animals of the same phenotypic state show similar elution profiles and relative abundance, demonstrating the reproducibility of the sample preparation and LC/MS/MS.
The reproducibility of this technique is further confirmed using principal components analysis (PCA). As shown in Figure 5, application of ANOVA to the total proteome results of each animal analyzed reveals a distinct clustering by cardiac phenotype (biological replicates) with the tightest association between technical replicates (different mass spec runs of the same sample) from the same animal. This provides confidence that the abundance changes observed for individual proteins of interest are indeed attributable to the difference in the physiological state.
Of equal importance as the quantification and reproducibility of the results is the increased coverage of the nuclear proteome enabled by nuclear subfractionation. Figure 6 illustrates that analysis of the whole nucleus alone fails to uncover even a majority of the nuclear proteins identified when combining individual analyses from the separate compartments (1,048 unique proteins identified in total across the fractions versus 428 identified from the intact nuclei). Furthermore, the moderate overlap between the different fractions demonstrates the biological relevance of considering these compartments individually for added insight into nuclear function and gene regulation. Finally, the ability to specifically fractionate acid-extracted or detergent-extracted chromatin enriches for key chromatin structural proteins, such as histones, thereby enabling more focused mass spectrometric analyses (such as those that target post-translational modifications) on a group of proteins without requiring specific antibodies for each variant and isoform.
Figure 1. Flow chart for characterization of proteins in cardiac nuclear sub compartments. Whole mouse hearts are homogenized and the lysate enriched for nuclei. Nuclei are fractionated into nucleoplasm, detergent-extracted chromatin, and acid-extracted chromatin fractions. Proteins from each are isolated and separated by one-dimensional SDS-PAGE, and the bands trypsin digested and run by LC/MS/MS. After database searching, label-free quantification is used to identify abundance trends for proteins in the different fractions.
Figure 2. Nuclear fractionation schematic. Nuclei are enriched from whole heart homogenate via a sucrose gradient. Each nuclei sample can either be separated into nucleoplasm and detergent-extracted fractions or into an acid-extracted fraction that enriches for histones.
Figure 3. One-dimensional gel of the different fractions. 20 μg of whole heart lysate (WHL), intact nuclei (Nuc), nucleoplasm (NuP), detergent-extracted chromatin (DE), and acid-extracted chromatin (AE) fraction from mouse heart were run on a one-dimensional SDS-PAGE gel (12% acrylamide) along with acid-extracted chromatin fraction from HeLa cells and stained with Oriole. The different patterning of abundance of total protein at the various molecular weights demonstrate the existence of unique proteomes for the different sub compartments, including enrichment for the lower-molecular weight histones in the detergent-extracted chromatin which are further enriched in the acid-extracted fractions. Also of note is the substantially less-complex nature of the AE fraction from HeLa cells as compared to that from the heart.
Figure 4. Relative abundance quantification. Mice from three different cardiac states: basal (blue) hypertrophy (red) and failure (green) were collected in triplicate and their protein samples ran in two technical replicates by LC/MS/MS. Shown on the left are the monoisotopic peaks for a specific peptide (whose spectra is shown on the right). Relative abundance was calculated by integration and averaged between the replicates of each phenotypic state (middle panel) to compare the relative abundance of the peptide (and the protein from which it came) in the different states of heart failure. The top panel shows a peptide from HMGB1, whose abundance is altered in disease, while the bottom panel (Histone H4) does not change. Click here to view larger figure.
Figure 5. Principal component analysis reveals reproducibility. ANOVA analysis for each mouse (3 biological replicates/cardiac state x 2 technical replicates) plotted for peptide intensity (axes) shows clustering by disease state (basal in blue, hypertrophy in red, failure in green), which indicates successful reproducibility of the mass spec data and significant physiological differences between the different disease states.
Figure 6. Global characterization of distinct nuclear sub compartments. We identified 1,048 total unique proteins across all 4 fractions. 146 proteins were shared by all 4 fractions, compared to 749 proteins identified in the nucleoplasm, 380 in the detergent-extracted chromatin fraction, 426 in the acid-extracted chromatin fraction, and 428 in the intact nuclei. The Venn diagram depicts the moderate overlap in identified proteins between the fractions, highlighting the existence of distinct regions in vivo and the ability of this fractionation to isolate them. Additionally, the markedly incomplete characterization of the total nuclear proteome achievable by only analyzing unfractionated nuclei (orange) justifies further fractionation.
Two main methods for nuclear isolation have been reviewed previously:27 one is the Behrens technique of homogenizing lyophilized tissue in a non-aqueous solvent and the second, a modification of which we use here, of homogenizing tissue in an aqueous sucrose/salt solution followed by differential or density-gradient centrifugation.
Subfractionation of the nuclei by acid extraction on tissue samples is an important tool for studying chromatin which has been used since 1960,28 with the original goal being to analyze histones. Acid-extraction protocols were developed for this goal of purifying core nucleosomal histones components.29 A limitation of these previous studies is an absence of information on the non-histone proteins constituting, and modifying, chromatin; we have addressed this challenge with the current protocol for characterizing distinct biological subfractions of the nucleus and chromatin. This approach, which uses detergent extraction or acid extraction of chromatin in parallel, reveals the diversity of chromatin regulatory molecules and distinguishes between levels of endogenous regulation, thereby serving as a robust approach for hypothesis generation.
Applying the established acid-extraction methodology to the adult myocyte poses certain obstacles including the dense cytoskeleton and large mitochondrial population. While versions of acid-extraction have been used on whole heart tissue30 the experiments were conducted in a targeted manner to study known proteins; the current protocol is designed and has been executed to enable unbiased proteome analysis. Additionally, we found it beneficial to examine the detergent-extracted and acid-extracted chromatin separately (only 37% overlap was observed between the proteomes of these two fractions) to better understand the dynamics through which these populations of chromatin proteins interact with DNA. Other protocols using urea-based buffers have also been described,31,32 which, performed prior to acid extraction, enrich proteins less tightly bound to the genome. In addition to chromatin fractions, others have subfractionated the nucleus to isolate the nucleolus,33 but subfractionation for mass spectrometry analysis of myocardial nuclei isolated from whole hearts has not been reported.
Shotgun proteomic studies on intact nuclei-that is, unfractionated organelles-have been carried out on myocardial tissue using a MudPIT approach.15 This study identified 1,044 nuclear cardiac proteins, on the same scale as our protocol, however they relied on using an average of 8.5 replicates, with each run lasting 20 hr to detect low-abundance proteins. In contrast, we employed subfractionation, to both increase the number of identifiable proteins (without needing lengthy chromatography-our LC runs are 85 min) while importantly also offering additional biological information on the sub-nuclear localization and relative abundance of the proteins.
From a whole mouse heart (approximately 0.14 g) we recover approximately 1.2 mg (0.85% of total) of protein in the intact nuclei. From the intact nuclei we achieve the following recovery for each of the subfractions (Note: the total values exceed 100% as acid-extraction and detergent-extraction are done on separate hearts, not in series): Nucleoplasm (15%), detergent-extracted chromatin (85%), and acid-extracted chromatin (54%). By comparison, Thermo Scientific has a kit (NE-PER Nuclear and Cytoplasmic Reagents) which their data sheet says yields 1.6mg cytosolic protein and 0.6 mg nuclear protein (1.5%) from 40 mg of mouse heart with a 10% cross-contamination between the two fractions. Sigma has a kit (NXTRACT, CellLytic NuCLEAR Extraction Kit) which they have tested on rabbit muscle, but not heart, and recovered from 100 mg tissue 0.46 mg of nuclear protein (0.46%) in what they term their “crude nuclear pellet” with a “few percent” of cytoplasmic contamination. Other companies, such as Epigentek also offer kits, however we were unable to find specific details on the enrichment and purity on their websites for comparison.
We have described a novel technique to addresses the current need for successful enrichment of cardiac nuclei for proteomic studies. We further describe methodology for sub compartment fractionation, which both increases the ability to detect less abundant proteins as well as allows for higher resolution, quantification and determination of protein localization. To this end, the protocol for mass spectrometry analysis offers recommended guidelines, however case-by-case considerations will most likely be necessary. For example, we found that the large number of cardiac histone variant isoforms with sequence similarity required manual inspection of the spectra to assign major peaks and confirm parent ion mass before being able to confidently assign the spectra.14 Search parameters can also be customized for post-translational modification (PTM) identification and excellent reviews exist detailing methods and concerns with data interpretation for PTM analysis (including PTM enrichment,34 software for PTMs,35 analysis of combinatorial PTMs,36 alternative fragmentation strategies37 and alternative digestion and processing38,39).
The enrichment and fractionation described is from the whole heart (which is 50-70% cardiomyocyte by mass). While this sacrifices aspects of cell specificity which may be especially important for certain families of nuclear proteins, it allows for a much more rapid suspension of the proteins into proper buffers, which is not possible with adult myocyte cell isolation protocols. In this way, this methodology better preserves the integrity of the sample against degradation. However, also described are instructions for applying this protocol to isolated neonatal rat ventricular myocytes, since both isolated-cell and whole organism models are necessary tools for studying heart failure. In principle this method could also be applied to adult myocytes extracted by enzyme digestion from isolated hearts-with the caveat that changes in protein binding that occur during the isolation (which can last over an hour) will confound interpretation of the proteomic observations.
We developed this methodology to understand the changes in the nuclear proteome and chromatin makeup that contribute to disease.14 However other applications include studying stimulus-specific changes in proteome composition in normal physiology. Additionally, sub-fractionation offers the ability for a new level of characterization of how the proteins operate and localize across distinct functional regions of the nucleus.
The authors have nothing to disclose.
The Vondriska lab is supported by grants from the National Heart, Lung and Blood Institute of the NIH and the Laubisch Endowment at UCLA. EM is recipient of the Jennifer S. Buchwald Graduate Fellowship in Physiology at UCLA; HC is the recipient of an American Heart Association Pre-doctoral Fellowship; MP is the recipient of an NIH Ruth Kirschstein Post-doctoral Fellowship; and SF is the recipient of an NIH K99 Award.
Name of the reagent | Company | Catalogue number |
Dulbeco Modified Eagle Medium | Invitrogen | 11965 |
Protease pellet | Roche | 04 693 159 001 |
100 μm strainer | BD Falcon | 352360 |
Ultracut ultramicrotome | Reichert | |
100CX Transmission Electron Microscope |
JEOL USA, Inc. | |
Oriole | BioRad | 161-0496 |
Histone H2A antibody | Santa Cruz | sc-8648 |
Nucleoporin p62 antibody | BD Biosciences | 610498 |
Adenine nucleotide transporter antibody | Santa Cruz | sc-9299 |
BiP antibody | Santa Cruz | sc-1050 |
Tubulin antibody | Sigma | T1568 |
Histone H3 antibody | Abcam | ab1791 |
Fibrillarin antibody | Cell Signaling | C12C3 |
SNRP70 antibody | Abcam | ab51266 |
E2F-1 antibody | Thermo Fisher | MS-879 |
Retinoblastoma antibody | BD Biosciences | 554136 |
Hypoxia inducible factor-1 antibody | Novus Biologicals | NB100-469 |
BCA protein assay | Thermo Scientific | 23227 |
Reverse phase column | New Objective | PFC7515-B14-10 |
BioWorks Browser | Thermo Scientific |