Comparison of Nuclear Matrix and Mitotic Chromosome Scaffold Proteins in Drosophila S2 Cells—Transmission of Hallmarks of Nuclear Organization Through Mitosis*

Chromatin condenses several folds to form mitotic chromosomes during cell division and decondenses post-mitotically to reoccupy their nuclear territory and regain their specific transcriptional profile in a precisely lineage specific manner. This necessitates that the features of nuclear architecture and DNA topology persist through mitosis. We compared the proteome of nuclease and high salt resistant fraction of interphase nucleus known as nuclear matrix (NuMat) and an equivalent biochemical fraction in the mitotic chromosome known as mitotic chromosome scaffold (MiCS). Our study elucidates that as much as 67% of the NuMat proteins are retained in the MiCS indicating that the features of nuclear architecture in interphase nucleus are retained on the mitotic chromosomes. Proteins of the NuMat/MiCS have large dynamic range of MS signal and were detected in sub-femtomolar amounts. Chromatin/RNA binding proteins with hydrolase and helicase activity are highly enriched in NuMat as well as MiCS. Although several transcription factors involved in functioning of interphase nucleus are present exclusively in NuMat, protein components responsible for assembly of membrane-less nuclear bodies are uniquely retained in MiCS. Our study clearly indicates that the features of nuclear architecture, in the structural context of NuMat, are retained in MiCS and possibly play an important role in maintenance of cell lineage specific transcriptional status during cell division and thereby, serve as components of cellular memory.

Chromatin condenses several folds to form mitotic chromosomes during cell division and decondenses post-mitotically to reoccupy their nuclear territory and regain their specific transcriptional profile in a precisely lineage specific manner. This necessitates that the features of nuclear architecture and DNA topology persist through mitosis. We compared the proteome of nuclease and high salt resistant fraction of interphase nucleus known as nuclear matrix (NuMat) and an equivalent biochemical fraction in the mitotic chromosome known as mitotic chromosome scaffold (MiCS). Our study elucidates that as much as 67% Genetic information that is encoded in a linear DNA sequence, needs to be organized in three-dimensional space of the nucleus in the form of chromatin for appropriate expression of genes (1)(2)(3). During interphase, interior of the nucleus is organized on a ribonucleo-protein network known as nuclear matrix (NuMat) 1 (4). Specific attachment of chromatin fibers to NuMat create a higher order structure made up of discrete topological loop domains (5,6). NuMat, along with facilitating the packaging of DNA, organizes the nuclear microenvironment in a way that transcription, replication, splicing and repair are executed in a coordinated manner over the highly compacted DNA. However, during open mitosis the three-dimensional organization of the nucleus breaks down as the chromosomes condense to enable faithful segregation of duplicated chromatids into daughter cells. Most genes are thought to be silenced during cell division but are precisely re-activated in daughter cells to achieve gene expression pattern that is indistinguishable from the mother cell (7,8). One of the mechanisms by which this is accomplished is by bookmarking of genes mediated by locus specific retention of transcription factors and deposition of heritable epigenetic marks (9 -11). Several studies suggest that along with this, the information for 3D organization of chromatin in the nucleus is also stably inherited in daughter cells. For example, chromosome positions are transmitted through mitosis and nonrandom higher order chromatin arrangements can be stably propagated over many cell cycles (12,13). Early evidence for transmission of information for higher order chromatin organization through mitosis came from the pioneering work by Laemmli and co-workers, who showed that in histone depleted mitotic chromosomes, DNA loops are attached to a proteinaceous scaffold. This arrangement of DNA loops in mitotic chromosomes is similar to the DNA loops that are attached to NuMat in interphase nucleus (14,15).
Considering that NuMat and mitotic chromosome scaffold (MiCS) are comparable biochemical structures, it is reasonable to expect that at least certain components of the NuMat are retained in MiCS (16,17). However, our knowledge of the precise relationship between composition of NuMat and MiCS remains limited. Hence, it is important to identify the key structural and functional components of NuMat that are rep-resented in MiCS. In this study, we have carried out a detailed proteomic analysis of NuMat and MiCS in Drosophila S2 cells.
Our results indicate that the bulk of MiCS proteins are a subset of NuMat proteome whereas only a small fraction of MiCS proteins is unique. These findings suggest that NuMat proteins are packaged into MiCS as chromosomes condense. We hypothesize that during mitotic exit, as the chromosomes decondense, these proteins provide the information to organize each chromosome into their territories and functional domains in the daughter nuclei. Thus, the NuMat and MiCS share much of the components responsible for hierarchical organization of chromatin that are likely to be, at least partially, specific to a cell type.

EXPERIMENTAL PROCEDURES
Cell Culture-S2 cells were cultured in Schneider's Media (GIBCO, Gaithersburg, MD) with 10% heat inactivated FBS at 25°C. Cells were arrested at metaphase with colcemid (Sigma, St. Louis, MO) at a concentration of 20 g/ml for 20-24 h.
Transfection of S2 Cells and Preparation of Stable Cell Lines-S2 cell transfections were done using effectene transfection reagent (Qiagen, Netherlands, 301425) according to manufacturer instructions. Briefly, 10 6 cells/ml in S2 cell culture medium were plated in each well of the 6-well plate 24 h before transfection. For each well, 1 g/2 l of plasmid DNA was mixed with 94.8 l of enhancer buffer and 3.2 l of enhancer in a microfuge tube. The solution was mixed by pipetting and incubated at room temperature for 5 mins. 4 l of effectene reagent was added to the tube, mixed by pipetting, and incubated at room temperature for 10 mins. 500 l of fresh S2 medium was added to each reaction mixture. In the meanwhile, 500 l of the medium was aspirated out from each well. The reaction mixture thus prepared was added dropwise into each well. The 6-well plates were incubated in 25°C incubator for 48 h. 100 g/ml hygromycin was added to each well and the plates were further incubated in 25°C incubator for 48 h. Following this the media was changed every 48 h with fresh media containing 100 g/ml hygromycin. After 3 weeks when cell death was negligible the cells were transferred to T25 flasks and maintained in media containing 100 g/ml hygromycin. The expression of proteins was induced by adding 500 M CuSO 4 in the media and incubating the cells in 25°C incubator for 24 h. All the steps were carried out in laminar flow hood.
Nuclei Purification-Pelleted cells were suspended in nuclear isolation buffer (NIB) with 0.25 M sucrose (NIB -15 mM Tris, pH 7.4, 40 mM KCl, 1 mM EDTA, 0.1 mM EGTA, 0.1 mM PMSF, 0.25 mM spermidine, 0.1 mM spermine). Cell suspension was passed 3 times through a 22-gauge needle to rupture cell membrane. Homogenate thus obtained was spun at 600 ϫ g for 2 mins to remove unbroken cells and debris. Supernatant was spun at 3000 ϫ g to collect crude nuclear pellet. The crude nuclear pellet was suspended in 1 M sucrose in NIB, and centrifuged at 6000 ϫ g to get pure nuclear pellet. The pellet was washed twice before being used for NuMat preparation.
Chromosome Purification-Chromosomes were purified according to Mg-Hex buffer method (modified from Wray and Stubblefield (18) by Lewis and Laemmli (14)). Briefly, arrested S2 cells were pelleted, washed and resuspended in RSB (10 mM Tris pH 7.4, 10 mM NaCl, 5 mM Mgcl 2 ). After incubation for 15 mins at room temperature, cells were pelleted down at 1400 ϫ g and resuspended for swelling in Mg-Hexelene buffer (1 M Hexelene glycol, 0.1 mM PIPES, 1 mM MgCl 2 , 0.1 mM and 1% thiodiglycol). Swollen cells were pelleted and resuspended in Mg-Hex buffer with 0.5% Na-DOC and homogenized with 20-25 strokes at 400 rpm with Teflon pestle in a tight-fitting glass tube. Homogenate was spun at 600 ϫ g for 5 mins to remove nuclei and unbroken cells. Supernatant was collected and spun at 1400 ϫ g for 30 mins to collect chromosomes.
An aliquot of nuclei/chromosomes was lysed in 0.5% SDS and absorbance was measured at 260 nm to find out the DNA content. Based on the DNA content, nuclear and chromosomal yield was calculated in OD units.
The NuMat and MiCS pellets were used for 1D and 2D PAGE, Western blotting and LC-MS/MS. SDS-PAGE, Western Blotting, and DIGE-NuMat/MiCS protein samples were resolved on 12% SDS-PAGE gel and were either silver stained for 1D profile or transferred to PVDF for western hybridization with Lamin Dm0, topoisomerase 2, histone or GFP antibody.
For DIGE, 150 g of protein was labeled with Cy Dyes from Ettan DIGE (GE Healthcare, Chicago, IL) and applied to 11 cm (3-10) strips from BioRad in rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 1% Pharmalyte). Focusing was done at 4000 V with linear ramp for 40,000 V Hrs. After focusing, the strips were equilibrated in buffer (Tris-Cl 50 mM pH 8.8, 6 M urea, 30% v/v glycerol, 2% SDS) containing 0.01g/ml DTT and 0.025 g/ml iodoacetamide. The focused proteins were resolved for second dimension on 12% SDS-PAGE. Gels were scanned using Typhoon scanner, documented and analyzed for spot pattern through Image Master 2D Platinum v7.05 DIGE software.
In-Gel Digestion and LC-MS/MS-For tryptic digestion, 2-4 g of protein were resolved on 12% SDS-PAGE. Gel slices were cut and dehydrated twice using 50% ACN followed by 100% ACN and the protein in the gel slices were digested with 100 l-200 l of Trypsin solution (10 g/ml, Promega, Madison, WI) in 25 mM ammonium bicarbonate. Digestion was carried out at 37°C overnight. Digested peptides were eluted with 50% ACN and 5% TFA, vacuum dried and stored at Ϫ70°C till loaded on to the mass spectrometer.
Tryptic peptides were loaded on a 50 cm long column, packed with 1.8 m C 18 beads. Column had been heated at 48°C in a homemade oven. The peptides were separated using linear gradient from 5% to 35% of buffer B (95% acetonitrile and 0.5% acetic acid) with flow rate of 75 nl/min followed by a wash up to 95% of buffer B. The length of the gradient was adjusted to 480 mins. The Proxeon LC system was directly connected with Thermo fisher scientific LTQ Orbitrap Velos instrument using Proxeon nanoelectrospray source. The nano source was operated at 2.2 kV and the ion transfer tube at 200°C without sheath gas. The mass spectrometer was programmed to acquire in a data dependent mode. The scans were acquired with resolution 60,000 at m/z 400 in orbitrap mass analyzer with lock mass option enabled for the 445.120024 ion. The 25 most intense peaks containing double or higher charge state were selected for sequencing and fragmentation in the ion trap by using collision induced dissociation with normalized collision energy of 40%, activation q ϭ 0.25, activation time of 10 ms and one micro scan. Dynamic exclusion was activated for all sequencing events to minimize repeated sequencing. Peaks selected for fragmentation more than once within 30 s were excluded from selection for next 90 s and the maximum number of excluded peak was 500.
Data Processing and Analysis-The raw spectra obtained were processed with Andromeda search engine (MaxQuant software version 1.1.0.39). Search was performed against D. melanogaster protein database (FlyBase assembly version Dmel_5.37 that contains 23605 entries in the forward database). Trypsin/P was specified as the cleavage enzyme and up to two missed cleavages were allowed. The initial precursor mass tolerance was set at 10 ppm and fragment mass deviation was set at 0.25 Da. The search included cysteine carbamidomethylation as fixed and oxidation of methionine as variable modification. The identification has been done keeping 1% false discovery rate at the peptide and protein level. Proteins identified with multiple peptides as well as one unique one unique peptide with high confidence were listed and analyzed further.
Experimental Design and Statistical Rationale-Three biological replicates from independent biological preparations were analyzed for each of NuMat and MiCS. Only proteins identified in all three biological replicates with two or more peptides, were considered as true NuMat and MiCS components and were used for further analysis. NuMat and MiCS thus identified were compared and the common and unique proteins were listed (supplementary Table S3). For estimation of sensitivity of single run analysis, total summed peptide intensities of each protein group in NuMat and MiCS were taken as proxies to estimate the protein amount. Summed peptide intensities were normalized to the molecular weight of proteins. Spearman Rank correlation was used to judge the reliability of quantification measurements between biological replicates. Further process was followed from the earlier report (19).
Bioinformatics Analysis-Annotation of the proteins was derived using clusterProfiler, an R package. Enrichment of protein domains in NuMat and MiCS proteins were analyzed based on InterPro database where analysis was done using DAVID annotation tool (v6.8 (http://david.abcc.ncifcrf.gov)).
In Situ NuMat/MiCS Preparation in Drosophila Embryo and Immunostaining-Drosophila embryos (0 -2 h old) were collected, de-chorionated and washed thoroughly with running water to remove sodium hypochlorite. Embryos (0.1 g) were then rinsed in distilled water and pretreated in 4:1 heptane:PBS for 2 min. Fixation was carried out in 0.8 ml heptane, 0.1 ml 10ϫ PBS with 4% formaldehyde for 20 min at room temperature with continuous mixing. After fixation, heptane and aqueous phase was removed. A mixture of 1:1 (v/v) ice cold methanol/heptane was added to devitellinize the embryos. The tube was shaken vigorously, and the embryos were allowed to settle to the bottom. This step is repeated several times. These fixed embryos were brought to aqueous medium by several washes in PBS ϩ 0.1% Triton-X-100 (PBT) and used directly for immuno-staining as control embryos. To prepare NuMat, the fixed embryos were brought to aqueous medium by several washes in DB (composition same as in MuMat/MiCS preparation) without DNase I. Digestion was carried out with DNase I (200 g/ml) in DB for 30 min at room temperature.
Embryos were then washed two times in extraction buffer without NaCl by suspending and then allowing to settle down in the tube because of gravity. Extraction was carried out with 2 M NaCl in extraction buffer or 20 min. The embryos were finally washed several times in PBS with 0.5% Triton-X-100 (PBT). These embryos with in situ mitotic chromosome scaffolds were used for immuno-staining.
For antibody detection, anti-Lamin Dm0 (mouse), anti-fibrillarin (rabbit) and an in house developed anti-NuMat (rat) antibody were used alone or in combination. The primary antibodies were used at a dilution of 1:100 in PBT. Incubation was carried out for 1 h at 37°C or at 4°C overnight. Primary antibody was followed by secondary antibody treatment (anti-mouse/rabbit/rat Cy3/FITC at 1:300 dilution) for another hour. Embryos were then washed and mounted in Vectasheild with DAPI. Confocal laser scanning was carried out on a Zeiss LSM 510 META (Carl Zeiss Inc., Germany) with excitation at 488 nm, 543 nm, 633 nm (Ar-ion and HeNe lasers) and 760 nm at a pinhole of 1 AU. The scanning was done in the multi-track mode using a 10ϫ, 63ϫ and 100ϫ 1.4NA objective. The emission of FITC was acquired using a 500 -530 BP filter, that of Cy3 with a 565-615 BP filter and that of Cy5 with 650 -710 BP filter set. Optical sections were taken 0.35 m intervals. Individual optical sections were projected to give information in 3D using the Zeiss LSM software version 3.2 SP2. Later the images were assembled using Adobe Photoshop 6.0.
Plasmids-The vectors carrying the cDNA sequences of Orc2 (GH13824), Orc5(RE16687), Mcm2 (LD47441) and Mcm5(RE67590) proteins were obtained from DGRC (Drosophila Genome Research Center). eGFP cDNA was amplified from pEGFP-C1 plasmid using appropriate primers with kpn1 and spe1 restriction enzyme sites introduced in the forward and reverse primers respectively. The Orc2, Orc5, Mcm2 and Mcm5 full length cDNAs were amplified from the DGRC clones using appropriate primers harboring BamH1(for Orc2 and Orc5) or EcoR1(for Mcm2) or Xho1(for Mcm5) restriction enzyme sites in the forward primers and Kpn1 restriction enzyme site in the reverse primers. The amplified cDNA sequences of Orc2, Orc5, Mcm2, and Mcm5 were digested with Kpn1 and ligated with Kpn1 digested eGFP amplicon. The resulting product was further amplified using the respective forward primers for each of the genes and the reverse primer for GFP. The products were digested with appropriate restriction enzymes and cloned into pMk33-cTAP(SG) vector, between BamH1 (for Orc2 and Orc5) or EcoR1 (for Mcm2) or Xho1 (for Mcm5) and Spe1 sites, such that the proteins are C-terminally tagged with GFP. For empty vector control GFP was amplified from pEGFP-C1 plasmid using a different forward primer harboring a BamH1 restriction enzyme site instead of Kpn1. This was subsequently digested and cloned between BamH1 and Spe1 sites into pMk33-cTAP(SG) vector. The pMk33-cTAP(SG) plasmid contains an inducible metallothionine promoter(MT) and a hygromycin selection marker.

NuMat and MiCS Preparation and Quality Control-NuMat
is the salt and nuclease resistant fraction of the interphase nucleus, whereas MiCS is an equivalent biochemical fraction from mitotic chromosomes. S2 cell nuclei were purified according to published protocol and were found to be free of cytosolic contaminants when observed under fluorescence microscope (Fig. 1A, a). The enlarged pictures in inset shows that the nuclei were free of adhering membranes. The isolated nuclei were also assessed by western hybridization with antibodies against proteins known to localize to other subcellular compartments (Fig. 1B). Aliquots were analyzed at different steps of nuclei isolation and analyzed by Western blot analysis. Western blot analysis showed that endoplasmic reticulum, golgi and cytosolic contaminants were efficiently removed from the isolated nuclei. S2 cells were arrested in metaphase with microtubule depolymerizing drug colcemid and highly purified mitotic chromosomes were obtained using Mg-Hexelene buffer method, adapted from Laemmli et al. (14). Fluorescence microscopy after DAPI staining showed that the chromosomal fraction was homogenous, and no contaminating nuclei were detected (Fig. 1A, b). Nuclei and chromosomes were digested with nucleases and subjected to high salt extraction to yield NuMat and MiCS respectively. The representative silver stained SDS-PAGE pictures showed that NuMat/MiCS were enriched in high molecular weight proteins (Fig. 2, Lanes 2 and 6). Digestion with nucleases removes bulk of chromatin and associated proteins (Fig. 2, Lanes 3 and 7) whereas stepwise salt and detergent extraction further re-moves the tightly associated proteins like histones (Lanes 4 and 8). Quality of the preparations was assessed by Western blot analysis with known NuMat and MiCS protein (Fig. 2, Lower panels). A known NuMat protein, lamin Dm0, was used as positive control for NuMat preparations and topoisomerase II that localizes in MiCS was used as a positive control for chromosome scaffold preparations. These proteins, as expected, were enriched in the respective insoluble fractions of NuMat/MiCS. During preparation, histones were eluted out by salt extraction and thus western with antibody against H3 served as a negative control to assure that the preparation is of good quality. Interestingly, lamin Dm0 was also seen in MiCS and topoisomerase II was also present in NuMat. Because ribonucleoproteins(RNPs) and RNA is known to be an integral part of NuMat, we estimated the amount of RNA that is retained in NuMat and MiCS and expressed it as percentage of that in nuclei and mitotic chromosomes respectively (supplemental Fig. S2). We find that ϳ11% RNA is retained in both the structures.

DIGE Analysis of NuMat and MiCS
Proteome-To get an estimate of total number of protein in NuMat and MiCS and to get an idea of overlap among the two proteomes, DIGE analysis was carried out (supplemental Fig. S1A). NuMat proteins were labeled with Cy5 (supplemental Fig. S1A, a), MiCS proteins were labeled with Cy2 (supplemental Fig. S1A, b) and an internal control comprising of pooled proteins from both the fractions was labeled with Cy3 (supplemental Fig. S1A, d). The spots were detected and analyzed by Image Master 2D Platinum Version 7.0, DIGE analysis software (GE Health Care) keeping threshold values at default level. Spot volume was chosen after manual assessment and exclusion of artifacts. The software identified a total of 1479 proteins spots of which 793 were common to NuMat and MiCS (supplemental Fig.  S1B). The experiment was performed twice and each time the numbers of unique and overlapping spot was nearly the same. This gave us a rough estimate that more than half of NuMat and MiCS proteome overlap with each other.

LC-MS/MS Profiling of NuMat and MiCS
Proteome-LC-MS/MS analysis was performed for NuMat and MiCS from three biological replicates, using 2 g of proteins per run. The FT-MS raw spectra were analyzed using MaxQuant with integrated Andromeda search engine. A total of 2892 and 2300 proteins were identified with a minimum of one unique peptide in NuMat and MiCS respectively (supplemental Tables S1, S2, and S5). Although, proteins identified by single unique peptide may also be NuMat/MiCS components, to reduce the possibility of false positive identifications, we used very stringent criteria for data analysis and considered only the proteins, that occur in all three biological replicates and were identified with two or more unique peptides, as true NuMat and MiCS components. Hence, 1953 proteins in NuMat and 1511 proteins in MiCS were considered for further analysis. Comparing NuMat and MiCS proteome we find that 67% (1390) of proteins were common to both, 27% (563) proteins were unique to NuMat and 6% (121) were unique to MiCS ( Fig. 3A and supplemental Table S3).
The proteins were assigned an abundance value that was calculated using MS signal of peptides of each protein, adjusted for protein length and compared with total MS signal. These quantifications give an approximate idea of abundance of individual proteins in a sample. Quantification was carried out based on added peptide intensities for each protein obtained for each of three biological replicate single runs. It has been reported earlier that there is a good correlation between added peptide intensity of each protein and total amount of proteins (19,(21)(22)(23). Spearmen rank correlation of protein quantification between biological replicates was greater than 0.78 for NuMat (p Ͻ 0.0001) and greater than 0.77 for MiCS (p Ͻ 0.0001) indicating that the quantification measurements were reliable among biological replicates. Protein abundance ranged over five orders of magnitude, with sub-femtomole amounts. The dynamic range of NuMat proteome was higher than that of the MiCS proteome, with about 300 additional proteins detected at sub-femtomole level (Fig. 3B). Although the experiment does not provide accurate copy numbers for each of the proteins measured, as this would require isotope labeled standards, the method has been shown to grossly estimate protein abundance in a sample (19). Even though the quantification may not be very accurate for individual proteins there is no deviation in global ranks from these values. However, at very low abundance level of proteins, their abundance measurement may be unreliable because the relationship between absolute amount of protein and MS signal may be less accurate. Interestingly, the 1390 proteins common to NuMat and MiCS show similar abundance in the respective proteomes with a wide range of dynamics of MS signal and sensitivity. (Fig. 3C).
General Features of the NuMat and MiCS Proteome-To elucidate the biochemical nature of common proteins of Nu-Mat and MiCS we plotted the distribution of pI values of these proteins over the range of 2-13 with a class interval of 0.5 (Fig.  4A). The plot shows that the proteins common to NuMat and MiCS have a bimodal distribution as compared with the whole Drosophila proteome, which shows a trimodal distribution. Although the proteins common to NuMat and MiCS and the background Drosophila proteome show an abundance of proteins of pI value of ϳ6, the former is enriched in proteins with pI value of ϳ8.5 as opposed to the latter which shows slight abundance of proteins with pI values of ϳ7.5 and ϳ9.
Gene ontology classification was done using clusterProfiler, an R package, to identify the molecular function of the proteins present in NuMat and MiCS (supplemental Table S4) (24). Based on the molecular function ontology, proteins that hydrolyze nucleoside triphosphates to release energy are highly enriched in common NuMat/MiCS proteome. Further analysis of the enrichment Table indicates that DNA/RNA binding proteins that have hydrolase as well as helicase activity are found abundantly in common proteome of NuMat and MiCS. Many chromatin binding transcription factors were uniquely enriched in NuMat whereas several proteins that bind to and hydrolyze RNA were uniquely associated with MiCS.
Depending on their molecular function in NuMat or MiCS, we observe that certain classes of proteins like RNA binding proteins, cytoskeletal proteins and unfolded protein binding chaperones were the least dynamic class, with almost all such proteins being common to NuMat/MiCS. Proteins with acetyltransferase, ligase and isomerase activity were also commonly associated with NuMat as well as MiCS. On the other hand, structural constituents of ribosome, nucleotide/nucleoside binding proteins and proteins with pyrophosphatase activity were moderately dynamic as almost 75% of them were present in NuMat as well as MiCS whereas the rest of them were present only in the NuMat. Proteins involved in DNA/ RNA polymerization, nucleocytoplasmic transport and transcription factors were highly dynamic with half of them (ϳ50%) common to NuMat as well as MiCS and the other half present only in NuMat. Interestingly, enzymes with methyltransferase/demethylase and phosphatase activity were exclusively present in NuMat. Similarly, ribonucleases were only present in MiCS (Fig. 4B).
The tool provides an enrichment score using the whole Drosophila proteome as background. This score is a modified Fisher Exact p value score. The results obtained were in parallel to the enrichment results obtained based on molecular function gene ontology classification. Based on the enrichment score for different categories, we find that proteins that carry a helicase domain are enriched in both NuMat and MiCS proteomes. This was also evident in the molecular function categorization where pyrophosphatases with helicase activity were found to be significantly enriched in common NuMat/ MiCS proteome. Proteins with WD40 repeat, SET domain, C 2 H 2 zinc finger and PHD domain that are mostly found in DNA binding transcription factors, were uniquely enriched in NuMat. Four proteins with WD40 repeats were found uniquely associated with MiCS.
Validation of NuMat/MiCS Proteins by Western Blotting-We observed that many proteins involved in specific functions in interphase nucleus such as DNA replication were also present in MiCS. As DNA replication occurs in association with the NuMat during S-phase of cell cycle, presence of these proteins in NuMat is understandable, but the relevance of these proteins in MiCS seemed confounding. We thus chose four proteins, namely ORC2, ORC5, MCM2 and MCM5, involved in replication licensing for validation. We tagged these proteins with eGFP and expressed them in S2 cells. Stably transfected cell lines were obtained by growing the transfected cells in the presence of antibiotic (hygromycin) for several passages. NuMat and MiCS were prepared from these cells and Western blotting was carried out using anti GFP antibody (Fig. 5). The results confirmed that the queried proteins were present in NuMat as well as MiCS, thus validating the result obtained by LC-MS/MS.
Visualization of NuMat/MiCS Proteins by Immunostaining-To directly visualize the reorganization of NuMat proteins during mitosis we used an antibody that was raised in-house, using the whole NuMat proteins from S2 cell nuclei as antigen. The antibody was used for immuno-staining Drosophila early (60 -90 min) embryo that is an excellent model to visualize dynamics of mitotic events because the nuclear divisions are synchronous and it is relatively easy to spot an embryo with a mitotic wave sweeping through. In intact nuclei (control), the NuMat antibody gives a punctate staining confined to the nuclear space defined by the lamin rim (Fig. 6A, nuclei panel). After in vivo NuMat/MiCS preparation, when bulk of chromatin and soluble proteins had been removed, the antibody continues to stain the NuMat proteins that undergo a striking reorganization into mitotic chromosomes and spindle like structure, which colocalizes with Lamin Dm0, during mitosis (Fig.  6A, NuMat, MiCS panels). Quality of NuMat/MiCS preparation was judged by the absence of DAPI signal that indicates complete digestion and removal of chromatin. Similar results were obtained when we carried out the immuno-staining with a known NuMat protein, fibrillarin. In intact nuclei (control), the protein remains associated with the nucleolus, remains distinctly associated with nucleolar remnant in NuMat preparation and were seen to distribute in the chromosomal space during mitosis (Fig. 6B). For this experiment, slightly older embryos (100 -140 min) were used, because nucleolus forms only after the thirteenth division when the nuclei have been separated into separate cells because of cellularization.
Distribution of Proteins Involved in Specific Functions in NuMat/MiCS-We then analyzed the distribution of few proteins with known cellular functions in NuMat/MiCS. We observe that many proteins show interesting and differential distribution in the two structures (Summarized in Table I).
Whereas epigenetic modifiers such as histone methyltransferases and demethylases were mostly present in NuMat, histone acetyl transferases and deacetylases were found to associate with both NuMat and MiCS. Mediator complex proteins and most Polycomb group (PcG) of proteins were present only in NuMat, whereas RISC complex proteins, Trithorax group (trxG) of proteins and insulator binding proteins were present in NuMat as well as MiCS. Cohesin and Condensin complex proteins were not only present on MiCS but were also associated with NuMat. Further, we investigated the role of NuMat/MiCS in organizing the various nuclear sub-compartments, which lack a delimiting membrane and are known to be assembled by the hierarchical recruitment of component proteins from the surrounding nucleoplasm. The protein components responsible for biogenesis of several nuclear sub-compartments were found in NuMat as well as MiCS (Table II). DISCUSSION In an earlier study, NuMat and MiCS proteomes were shown to be largely similar in HeLa cells, however the exact protein constituents were not known till now (17). This extensive proteomics approach not only uncovers the fact that the MiCS proteome is a subset of NuMat proteome but also gives a glimpse into the complexity of both the structures indicated by the wide variety of proteins. We for the first time report an in depth quantitative proteomics of Drosophila NuMat and  MiCS exemplified by detection of proteins present in as low as sub-femtomole amount. Although the common proteins of NuMat and MiCS were detected with comparable intensities, the higher number of unique proteins in NuMat also included a significant proportion of low abundance proteins compared with MiCS. This is expected because the interphase nucleus is functionally more complex than mitotic chromosomes. Proteome analysis studies have elucidated that bacterial and archaeal proteomes exhibit a bimodal distribution of pI values, whereas eukaryotic proteomes show a trimodal distribution (27,28). Nuclear proteins are almost evenly distributed throughout the pI range of 4.5-10 in contrast to the bimodal distribution around pI values ϳ6 and ϳ8.5 observed in our study. This observation is interesting as proteins are generally least soluble at their isoelectric point (29). As intracellular pH is close to neutrality, the pI values of greater than or less than pH 7 of NuMat/MiCS proteins suggests their inherent insoluble nature. This also points to potential role of NuMat/MiCS proteins as structural units in nondiffusible nuclear architecture.
We observed many heat shock proteins/chaperones that are mostly common between NuMat/MiCS were also one of the most abundant proteins as observed based on protein intensities. Presence of molecular chaperones and protein folding catalysts such as heat shock proteins are known to associate with nuclear multiprotein complexes and their presence in NuMat is well documented (30 -32). Because, chaperones protect nonspecific aggregation of their substrates in a crowded micro-environment, perhaps they assist in the proper folding and ordered assembly of the proteinaceous network in NuMat and MiCS and maintain them under changing conditions of the cell cycle.
Several studies in past have shown that RNA is an essential component of NuMat. Considering the similarity between the NuMat and MiCS proteome, including number of RNA binding proteins, RNA components in the later one is also likely to be important. Although we maintain strict RNase free conditions during our preparation process, that we used both MNase and DNase I for digestion of metaphase chromosomes for preparation of MiCS, whereas only DNase I was used for digestion of nuclear chromatin. However, it is known that the specific activity of MNase for DNA is ϳ3-fold greater than that for RNA (33). This and the short digestion time for MNase used in this study, is unlikely to significantly affect the RNP content in MiCS. This was also confirmed by the fact that both NuMat and MiCS retain ϳ11% of nuclear and chromosomal RNA respectively. Therefore, the effect of MNase digestion on the structural integrity and protein composition of MiCS is expected to be minor, if any. This is further supported by the observation that most of the RNPs, proteins which are expected to be associated in an RNA dependent manner, detected in this study are common between both NuMat and MiCS.
One of the limitations of the NuMat field has been the suspicion that because of high salt concentration (2 M NaCl) various components may acquire interactions that are irrelevant to their normal context (34,35). These concerns are at least partially addressed by stepwise elution (0.4 M followed by 2 M Nacl), used in our study, thereby reducing the complexity by removing bulk of extracted proteins in the first step. As discussed above, high degree of consistency among biological replicates further supports the relevant and non-random nature of components identified in NuMat. Finally, remarkable similarity in the constituents of NuMat and MiCS also indicates the reliability of these constituents and argues against any significant artifactual/random aggregation.
One of the interesting findings of this study is the fact that cytoskeletal proteins like cheerio and zipper (fly homologs of filamin and myosin II, respectively) were present in both Nu-Mat and MiCS, whereas abundant proteins of the cytoskeleton were found only in NuMat (e.g. actin) or were present in very low amounts (e.g. tubulin) in NuMat/MiCS, also strengthens the idea that cytoskeletal proteins in NuMat/MiCS (filamin and myosin) are associated with each of these structures as important structural/functional components. It has been shown earlier that actin is associated with regulation of gene expression in the nucleus, by interacting with transcription factors and chromatin remodeling complexes, which are associated only with interphase nucleus, and hence, is presence only in NuMat (36). Cheerio has been reported to be involved Pc Disassembly/Reassembly (67) in repression of rRNA transcription whereas zipper is known to help attach chromatin to the nucleoskeleton, functions which are essential in both nucleus and mitotic chromosomes (37,38). The role of several cytosolic proteins in nuclear architecture is also supported by our finding that ϳ22% of the NuMat/MiCS proteins are known to localize to different cytosolic organelles like mitochondria, golgi, endoplasmic reticulum, etc. Association of such proteins with the NuMat as well as MiCS raises the, possibility of these proteins having a role in maintaining cellular memory through mitosis to re-establish similar cellular state in daughter cells. In recent years, many proteins have been found to behave as moonlighting proteins (39,40). The role of MiCS in mitotic cellular memory would also suggest several nuclear function proteins might be retained at the appropriate loci in mitotic chromosomes. For example, the replication licensing proteins (Orc2, Orc5, Mcm2 and Mcm5) that we chose for validation of our results, are also known to show chromosome condensation defects indicating that these proteins may have functions other than replication (41,42). The replication machinery itself is correlated with differential organization of the genome in the interphase nuclei in the form of early and late replication associated domains (RADs), which overlap with actively transcribing and inactive regions, respectively (43). Replicating units are also known to be differentially distributed between euchromatin and heterochromatin. Their presence in both NuMat and MiCS suggests that replication licensing proteins might be the information carriers of RADs from mother to daughter cell during mitosis. On the other hand, several proteins known to have a defined function in mitotic chromosomes are also found in NuMat. For example, cohesins involved in cohesion and segregation of sister chromatids are found in NuMat. Recently, they have been shown to be involved in mediating enhancer promoter interactions and in organizing chromatin loops during replication (44,45). Interestingly, these proteins have also been implicated in re-establishment of transcription foci after cell division suggesting their role in transmission of transcriptional memory to daughter nuclei (46 -48).
One of the known mechanisms by which cellular memory is transmitted from mother to daughter cells during mitosis is by bookmarking of genes in the form of locus specific histone modifications and the recruitment of complexes which maintain open/closed chromatin state thus enabling/repressing transcription (49). Our results provide a glimpse into the distribution/localization of histone modifiers, mediator complex proteins, RISC complex proteins, insulator body proteins and proteins of trxG and PcG proteins in NuMat and MiCS, all of which have plausible roles in cellular memory. We observe that mediator complex proteins that regulate the process of active transcription are present only in NuMat, but trxG proteins that mark the active genes, are present in NuMat as well as MiCS (50,51). On the other hand, PcG proteins that maintain silenced chromatin are uniquely present in NuMat, indi-cating that the marks identifying active genes are transmitted through mitosis whereas those identifying repressed genes are probably deposited and maintained during interphase. Our data also suggests that though active transcription complexes are assembled in interphase, some seed components of these complexes persist through mitosis by associating with mitotic chromosomes. These complexes might also be responsible for the low levels of transcription that has recently been shown to occur during mitosis (52).
At the syncytial blastoderm stage in Drosophila, nuclear division is characterized by semi-closed mitosis where the nuclear membrane ruptures only partially near the spindle poles (53)(54)(55). The localization of Lamin Dm0 (Figure 6A), where it forms a spindle like structure between the spindle poles but is discontinuous at the spindle poles, also reflects the partially closed mitosis in early embryos. On the other hand, in post syncytial stage embryos ( Figure 6B), when mitosis is completely open, Lamin Dm0 localizes to the cell periphery. Interestingly, Lamin Dm0, which is one of the most abundant proteins in NuMat, is also present in MiCS, which is intriguing as nuclear membrane and lamina disintegrate during mitosis. Interaction between chromatin and nuclear lamina is essential for higher order organization of chromatin because lamin associated domains comprise 30 -40% of whole genome in mammals (56). Several nuclear pore proteins, in addition to their role as transport channel have been implicated in chromatin organization and gene regulation and are shown to bind at distinct genomic loci that are not nuclear envelope contact sites (57,58). Retention of lamina components and nuclear pore complex proteins in MiCS hints toward the structural basis for cellular memory during cell division. The relevance of such protein in MiCS needs to be studied further.
Nuclear bodies that lack a delimiting lipid membrane represent yet another aspect of nuclear organization because they are the sites of many important nuclear functions. Most nuclear bodies, just like the rest of interphase nucleus, disassemble during mitosis but are efficiently and rapidly reassembled after mitosis from components present in the surrounding nucleoplasm (59). We observed that protein components responsible for the formation of several nuclear bodies were present in NuMat as well as MiCS indicating that the information for facilitating the formation of these nuclear bodies is carried over from the mother to daughter nuclei in association with MiCS. For example, Drosophila histone locus bodies form when transcription of histone genes commences in interphase by hierarchical recruitment of components, first of which is Mxc (60). Mxc persists as a MiCS protein probably to aid the re-assembly of histone locus body after mitosis. Similarly, coillin, which is considered as the marker protein for Cajal bodies that are the site for snRNP biogenesis, is abundant in both NuMat and MiCS (61). Thus, the presence of key components of nuclear bodies in MiCS supports the model in which nuclear body formation is driven by initial seeding pres-ent in the mitotic chromosome followed by self-assembly as cell enters interphase (62). Taken together, based on these observations, we suggest that mitotic chromosome may represent a miniaturized version of interphase chromosome territory, which retains key protein components as seed elements to re-establish functionally equivalent organization of the nucleus in daughter cells. MiCS constituents, therefore, play a role in mitotic cellular memory and provide a structural basis for the same.
In conclusion, the retention of majority of NuMat proteins in MiCS might be instrumental in transmitting structural and functional information of nuclear architecture to daughter cells. NuMat proteins retained in MiCS may also provide structural basis, to re-establish the nuclear architecture and gene expression pattern specific to the cell type. This study gives a flavor of the rich complexity involved in the multifaceted process of maintaining cellular memory and serves as an useful resource for future candidate-based studies. Functional studies on the constituents of nuclear architecture will be needed to establish the precise molecular basis of cellular memory.