Functional architecture of the cell nucleus: Towards comprehensive toponome reference maps of apoptosis

Biochimica et Biophysica Acta 1783 (2008) 2080–2088

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b b a m c r

Functional architecture of the cell nucleus: Towards comprehensive toponome reference maps of apoptosis Walter Schubert a,b,⁎, Manuela Friedenberger a, Marcus Bode a, Andreas Krusche a, Reyk Hillert a a b

Molecular Pattern Recognition Research Group, Institue of Medical Neurobiology, Otto-von-Guericke-University Magdeburg, Leiziger Street 44, 39120 Magdeburg, Germany CAS-MPG Partner Institute for Computational Biology, International Faculty, Shanghai, China

a r t i c l e

i n f o

Article history: Received 20 May 2008 Accepted 20 July 2008 Available online 31 July 2008 Keywords: Apoptosis Toponome TIS Cell nucleus Fluorescence microscopy Liver cell

a b s t r a c t We have recently described the MELC/TIS fluorescence robot technology that is capable of colocalizing at least a hundred different molecular cell components in one cell. The technology reveals new hierarchical properties of protein network organisation, referred to as the toponome, in which topologically confined protein clusters are interlocked within the structural framework of the cell. In this study we have applied MELC/TIS to construct a three-dimensional toponome map of the cell nucleus of a single human hepatocyte undergoing apoptosis. The map reveals six different spatially separated toponome domains in the nuclear interior of one apoptotic cell. In the drive to decipher the apoptosis-specific molecular network on the single cell level, the present toponome map is a first milestone towards the construction of much larger maps addressing hundreds of molecular cell components across the stages of apoptosis. © 2008 Elsevier B.V. All rights reserved.

1. Introduction The cell nucleus is the largest single organelle in the cell. It contains the overall functional plan for myriads of different functionalities, such as regulated transcription of single genes, coordination of large gene programmes during cell differentiation and DNA replication, maturation of ribosomes in the nucleolus, and transfer of factors across nuclear pores. Emerging views indicate that the genome and its functional domains are dynamically linked to the overall nuclear architecture [1]. For example, specific factors are organized in 20–40 speckled structures [2], and DNA is replicated only once in each cell cycle. Considering the concept that almost every function in the cell relies on the coordinated action of molecular machines [3] at highly defined subcellular locations [4,5], the precise detection and quantitative as well as functional analysis of these machines within the cell is one of the biggest challenges in the post-genome era [6]. For example, which clusters of transcription factors are formed at the promotor region of specific genes? Would we be able to decipher the overall functional plan of the nuclear DNA by studying transcription of hundreds of specific genes in one single cell nucleus across the whole genome? How is this machinery disturbed in individual cells undergoing apoptosis? Although apoptosis-specific molecular machines, such as the apoptosome, have been identified [7,8], significant gaps remain in our knowledge of this process. There must be an apoptosis-

⁎ Corresponding author. Tel.: +49 391 6117175; fax: +49 391 6117176. E-mail address: [email protected] (W. Schubert). 0167-4889/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.bbamcr.2008.07.019

specific molecular network with clear cut topological constraints that finally leads to this particular form of cell death (in contrast to necrosis that induces inflammatory reactions of the surrounding tissue). How can this network be quantitatively described in individual apoptotic cells? These and many similar problems addressing the functional architecture of the cell nucleus place significant methodological constraints on molecular imaging technologies able to identify the topological properties of molecular networks in situ, by colocalizing a very large number of molecular cell components in one cell nucleus and cytoplasm [4]. We have recently described the microscopic fluorescence robot technology MELC/TIS that is capable of imaging at least 100 different molecular cell components (MCC) in one cell [4,5]. MELC/TIS overcomes the spectral limitation of traditional fluorescence microscopy by using large dye-conjugated tag libraries and automatically bleaching a dye after imaging and re-labelling the same or another MCC in the identical sample with the same dye coupled to a tag having the same or another specificity, and repeat similar cycles with other tags for multiple times revealing multidimensional colocation patterns [4,5,9–11]. By this approach we have addressed for the time as what we term the toponome (the functional protein networks, or, the biological code of the cell) [11–14]. After the first description of the technology in 1990 [9], the importance of this approach to cell function has been increasingly recognized [5,11,15– 19]. The technology has proven to solve key problems in biology and therapy research: it has (i) uncovered a new cellular transdifferentiation mechanism of vascular cells giving rise to myogenic cells in situ/in vivo [20], a finding that has led to efficient cell therapy

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

models of muscle disorders [21–24]; (ii) discovered a new target protein in sporadic amyotrophic lateral sclerosis by hierarchical protein network analysis [25], a finding that has been confirmed by a mouse knock out model [26], (iii) uncovered a lead target protein in tumour cells that controls cell polarization as a mechanism that is fundamental for migration and metastasis formation [43] and (iv) found new functional territories in the CNS defined by highdimensional synaptic protein clusters [27,28]. The MELC/TIS technology is unparalleled when it comes to the colocalization of a very large number of proteins using photonic microscopy. The ability to colocalize proteins on a large scale is regarded as a bottleneck in the drive to understand how proteomes are organized in individual cells and how interlocked clusters of proteins exert control over cellular functions. In the present study we have chosen a primary human hepatocyte undergoing apoptosis to construct a toponome reference map. It allows the investigator to study simultaneously in one cell changes in the cell nucleus and in the cytoplasm. We regard this mapping procedure as a new principal towards the analysis of molecular and structural changes during apoptosis. The map does not focus on specific molecular clusters in the nuclear interior but rather on the feasibility to localize simultaneously 15 different MCC as specific markers of cellular organelles, substructures of the cell nucleus and molecular events taking place in apoptosis. Together we consider this map as a reference map (a pan-cellular molecular ‘landscape’) that can be used in future studies to construct toponome cybermaps of the cell nucleus, in which hundreds of additional MCC can be colocalized relative to the apoptosis-toponome structures presented in this study. 2. Material and methods 2.1. Tag library The tags used to label proteins and ligands, assembled as a toponome mapping library, are summarized in Table 1. The molecules selected for the toponome imaging procedures in this study are all reference markers of subcellular organelles or molecular systems in the cell undergoing substantial changes during apoptosis or liver disease [7,8,29–43].

Table 1 Summary of molecules labelled by MELC/TIS and corresponding locus link annotations Molecule/moiety recognized

Official symbol

Name

c-fos

[FOS]

CD49f CD138 ConA lig. Cytochrome C Cytokeratin

[ITGA6] [sdc1] – [CYC1] [KRT8], [KRT18] –

v-fos FBJ murine osteosarcoma viral oncogene homolog [Homo sapiens] Integrin, alpha 6 [Homo sapiens] Syndecan[Homo sapiens] Alpha-man, alpha-glc Cytochrome c-1 [Homo sapiens] Keratin 8 [Homo sapiens] and keratin 18 [Homo sapiens] Golgi complex-specific antigen

Golgi complex Ab-1 Hoechst 3328 lig. Lamp1

– [LAMP1]

Membrin

[GOSR2]

Mitochondria Ab-2 p53

–

p170

[ITGAM]

Prop lig. WGA lig.

– –

[TP53]

2081

2.2. Cells Primary human hepatocytes were cultured on glass cover slips as described [5]. Before use for toponome mapping procedures cells were washed with 10% DMSO in PBS, fixed in fresh 4% paraformaldehyde, and washed in PBS. Cells were then air dried and snap frozen in liquid nitrogen via pre-cooled isopentane. Before use samples were rehydrated in PBS at 20 °C, incubated with normal goat serum for 30 min, and washed again in PBS, as described [5]. 2.3. MELC/TIS technology for toponomics Multi-Epitope-Ligand-Cartography (MELC) [4,9,14], recently further developed as Toponome Imaging System (TIS) [5,44], and now termed MELC/TIS technology, and its use in biology and medicine has been described previously in detail [4,5]. It is a multidimensional microscopic robot technology which runs cycles of fluorescence tagging, imaging and bleaching in situ. This technology combines three advances: a technique capable of mapping hundreds of different proteins at light microscopic resolution in one tissue section or cell sample; a method for selecting the most prominent combinatorial molecular patterns by representing the resulting data as binary vectors; and a system for imaging the distribution of these protein groupings in a ‘toponome map’. We have shown that this approach reveals new hierarchical properties of protein network organisation [4], in which the frequency distribution of different protein groupings obeys Zipf's law [4,45]. To map a large number of proteins/ligands simultaneously in a sample, a slide with the sample is placed on the stage of an inverted wide-field fluorescence microscope equipped with fluorescence filters for FITC [5]. Fluorochrome-labelled tags and wash solutions are added and removed robotically under temperature control, avoiding any displacement of the sample and objective. In each cycle, a tag is added; phase contrast and fluorescence images are acquired by a highsensitivity cooled CCD camera; the sample is washed with PBS and bleached at the excitation wavelengths; and post-bleaching phase contrast and fluorescence images are acquired. Usually (but not exclusively) data are acquired by using a 63x oil objective (1.4 aperture) yielding a pixel dimension of 216 × 216 nm (0.0467 μm2) as outlined in detail in our earlier study [5]. Data acquisition is fully automated using home made software. By this approach, the technology overcomes the spectral limitation of traditional fluorescence microscopy [46]. 2.4. Construction of 2D toponome maps

Nucleic acids Lysosomal-associated membrane protein 1 [Homo sapiens] Golgi SNAP receptor complex membrane 2 [Homo sapiens] Mitochondria specific 60 kDA nonglycosylated protein Tumor protein p53 (Li-Fraumeni syndrome) [Homo sapiens] Integrin, alpha M (complement component receptor 3, alpha; also known as CD11b (p170), macrophage antigen alpha polypeptide) [Homo sapiens] Nucleic acids NeuAc, (glcNac)2, (glcNac)3

Fluorescence images produced by each tag are aligned pixel-wise using the phase contrast images, with an in- register accuracy of ±1 pixel. Background and illumination faults are then removed by flatfield correction. Fluorescent pixels are then parsed by regarding the list of fluorescence intensities I1, I2, I3…In for proteins 1, 2, 3…n in any particular pixel or voxel as the values of an n-dimensional vector associated with that pixel. This vector can then be binarized by selecting thresholds T1, T2, T3…Tn for proteins 1, 2, 3…n, and setting the vector values for any protein m to zero if Im b Tm and to 1 if not, using thresholds manually set by human experts from within an automatically generated range. The binarized images are then combined to form a list of combinatorial molecular phenotypes (CMPs) representing the proteins/molecules expressed in each pixel, or groups of CMPs, representing regions of interest. Given CMPs (protein clusters) or groups of CMPs are visualized in false color at their location. This imaging procedure can be performed in 2D or 3D [4]. A rapid method to map a large number of different CMPs in one sample is to introduce a threshold grey value for each fluorescence signal (representing the location of a labeled molecule). Each signal

2082

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

Table 2 List of the most prominent combinatorial molecular phenotypes (CMPs) detected in this study (out of N7.000) in one cell shown in Figs. 1, 2, and 4 at different magnifications and spatial aspects

of a sample; deconvoluting these images using a standard algorithm (XCOSM) [47] working with a specific point spread function; setting thresholds for each tag signal from each optical plane as before; overlaying all binarized images to construct large-scale protein/ligand colocalization maps using home made software; and constructing 3D toponome maps in the same way as was done for two dimensions. The latter two visualization steps were performed by using algorithms provided by IMARIS software packages. Calibration of the procedures was performed as described by us in detail [4]. 3. Results 3.1. Rationale of the toponome imaging procedures

Note that 15 colocalized molecules/moieties (decoding list on the right), are differentially combined in the cell to reveal 39 different CMPs that are mutually exclusive by their intracellular location. The CMP colours correspond to the colours displayed in the toponome maps of Figs. 1, 2 and 4.

within a CMP is mapped as tag present (=1) or tag absent (= 0), depending on whether the value for the fluorescence signal is above or below this threshold, respectively (= 1 bit information per protein/ligand at a pixel/voxel) (Table 2; Fig. 1). CMPs assembled as a group will eventually have unique features as defined by the assembly's of lead proteins (L = 1), absent proteins (A = anticolocated, 0) and wild card proteins (W = variable occurrence of a protein 0 and 1 in given CMPs of a CMP group). We define such CMP groups as a CMP motif), denoting a given functional region of a cell or tissue [4]. By using the three symbol code (LAW, above) [11], differences of cell states and cell types can be readily identified in studies comparing large experiments, or, diseases with normal conditions. Note that in the present study CMP motifs were not analyzed, because the detected CMPs were sufficient to define functional territories in the cell. 2.5. Three dimensional toponome imaging 3D imaging of MELC/TIS runs was performed by generating and visualizing z-stack raw images for each tag signal from top to bottom

The design of this study was based on the known facts of apoptosis, in which cytochrome c (Cytc) forming a complex with Apaf-1 triggers a cascade of events that lead to fragmentation of multiple organelles (Golgi, ER, etc) and cellular structures, such as the cell nucleus, and the cytoskeleton [7]. The study was focussed on cultured primary human hepatocytes undergoing (in part) spontaneous apoptosis. Apoptotic cells were readily identified by the fragmentation of the cell nucleus and condensation of nuclear DNA (Fig. 1d), while non-apoptotic cells, present in the same cell culture, showed nuclei with homogeneous staining for nuclear DNA (Fig. 1k). To illustrate the working of a toponome imaging system (TIS) for the analysis of apoptosis, we have chosen 15 different MCC (Table 1) to be colocalized in a human primary hepatocyte undergoing apoptosis. Every single chosen MCC can be regarded as a functional reference marker for given cellular compartments or domains so that their colocalization in a TIS experiment will reveal as what we term a toponome reference map (TRM). As an overall strategy to construct functional toponome maps for an extremely large number of MCC (N100) [4], the map presented here is regarded as multidimensional hepatocyte specific subcellular MCC-‘landscape’ that can be used as a basis for future in-depth analyses of much larger toponome maps. As described in the method section we have generated 3D images for each of the corresponding MCC, resulting from 10 different deconvoluted z-stack images per MCC, each z-stack image having a thickness of approx. 200 nm (Fig. 1, a–j). After setting a threshold for each MCC signal by an approach described earlier [4], we have measured the total number of CMPs displayed by the cell of interest. We found more than 7.000 different CMPs in the one cell, some of which were confined to the cell nucleus. Table 2 illustrates the 39 most frequent CMPs, each of which occupies a unique space within the cell as illustrated in ten optical sections from top to bottom (Fig. 1, a–j). Principally CMPs are mutually exclusive functional sites within any cell [4]. Thereby CMPs, contained in a toponome map, visualize many different functional domains within a cell simultaneously. Hence the CMPs displayed in Fig. 1 and decoded in Table 2 reveal an overall TRM both of cytoplasmic and nuclear structures in liver cell apoptosis. Below we focus on the nuclear CMPs. 3.2. Detection of nuclear domains To illustrate the high selectivity of nuclear CMPs, we have assembled a gallery of two representative optical sections across the cell nucleus (Fig. 2 part I and part II). Visualization of the selective and mutually exclusive location of different CMPs was performed as follows. First, to give a cellular overview of the most frequent 39 CMPs (summarized in Table 2), these CMPs were displayed inside and around the region of the cell nucleus, simultaneously in different colours, by aligning horizontally two out of ten optical sections (optical sections Z5 and Z6: Fig. 2a, b, part I, respectively). These images show that the intranuclear CMPs are separated from the cytoplasmic CMPs by a rim of cytokeratin structures (CMP9) (compare with the CMP decoding list shown in Table 2).

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

2083

Fig. 1. Illustration of a toponome map across a human hepatocyte undergoing apoptosis (a–j) and illustration of a non-apoptotic cell at 3D from the same cell culture (k). Apoptosis is evident by fragmentation of the cell nucleus (d, 1) and of cytoplasmic organelles (d, arrows 2). Together ten different optical sections across the cell are shown from top to bottom (a-j, corresponding to optical sections Z1–Z10). Each optical section has a thickness of approx. 200 nm. Note that 15 different cellular marker molecules have been colocalized in different colours to reveal most prominent 39 CMPs (molecules and CMPs are listed in Table 2). These CMPs are mutually exclusive, thereby singling out nuclear and cytoplasmic substructures of the cell. (k) Note homogenous nuclear staining in the non-apoptotic cell by propidium iodide (arrow 1) and Hoechst 3328 (arrow 2). Bar: 10 μm.

As expected, the cytoplasmic structures of the Golgi organelle show a pronounced fragmentation (Fig. 2 b, CMP6: note many single turquoise structures around the red cytokeratin rim), which is in keep

with the action of caspases on cytoplasmic organelles during apoptosis [7]. Out of the CMPs listed in Table 2, we have selected seven CMPs that are located inside the nuclear region (CMPs 1; 3; 5; 8;

2084

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

Fig. 2. Part I. Illustration of two optical sections (level Z6 and Z5) across the cell nucleus of the identical apoptotic cell shown as overview in Fig. 1. Note that the cell nucleus is separated from the cytoplasmic structures by a cytokeratin rim (a-h: CMP9, red). (a, b) Simultaneous display of the 39 most frequent CMPs in different colours (decoded in Table 2): Note slight differences in level 5 and 6 (a, b, respectively). (c–h) Three CMPs (CMPs 1, 3, and 5) are shown that specifically single out distinct intranuclear domains (D1, D2, D3sub [D3 subunit]): For direct visual comparison of any given CMP, the corresponding images from the two optical sections (levels Z5 and Z6) are aligned horizontally (CMP1: c, d; CMP3: e, f; CMP5: g, h). Bar: 5 μm. Part II. In addition to the three CMPs of Fig. 2 part I, this figure illustrates another four distinct intranuclear CMPs of the identical toponome data set (i, j: CMP 8; k, l: CMP12, m, n: CMP25; o, p: CMP26). Note that these CMPs denote the spatially separated intranuclear domains D3, D4, D5, and D6, in addition to the three domains illustrated in Fig. 2 part I. Bar: 5 μm.

12; 25; 26). We displayed these CMPs together with the cytokeratin rim (CMP9, red) for better orientation (Fig. 2, part I, Fig. 2 part II). Note that the separation of Fig. 2 into part I and part II was chosen to clearly

visualize the small intranuclear structures at an appropriate magnification across the two optical sections (Fig. 2 part I and part II: optical sections of level Z5 and Z6, vertical collection of seven different CMPs).

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

2085

Fig. 2 (continued).

As illustrated in Fig. 2 (part I) and Fig. 2 (part II), the seven different CMPs detected in the nuclear interior are all present as punctuate structures, each with a unique intranuclear distribution pattern in the x/y and the z-direction of the image stack. When the single intranuclear distribution patterns are compared, it becomes evident, that there are essentially six intranuclear, spatially separated domains singled out by CMP1, CMP5, CMP8, and CMP12. Domains 1 and 2 (Fig. 2 part I, D1, D2) are spatially separated

territories characterized by clusters of Cytc signals (CMP5) (Fig. 2 part I, g, h). These domains also contain signals displayed as CMP1 (CD49f) (Fig. 2 part I, c). Domain 3 is characterized by DNA structures displaying CMP8 (Hoechst ligand) (Fig. 2 part II, i, j), CMP25 (propidium iodide-ligand) (Fig. 2 part II, m, n), and CMP26 (coexistence of Hoechst- and propidium-ligands) (Fig. 2 part II, o, p). This domain also contains signals of CMP3 (c-fos) as a subdomain (Fig. 2 part I, e, D3sub). Domains 4 to 6 are singled out by CMP12 (Lamp1).

2086

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

Fig. 3. Schematic illustration of the intranuclear domains (D1–D6) detected in Fig. 2. Note that the outer line symbolizes the nuclear envelope. CMPs below refer to the CMP coding list in Table 2. D1: CMP5 (Cytc) + CMP1 (CD49f); D2: CMP5 (Cytc) + CMP1 (CD49f); D3: CMP8 (Hoechst ligand) + CMP25 (Hoechst ligand + Propidium ligand) representing the region of condensed DNA. This domain is crossed by a subdomain of cfos structures (CMP3) (punctuate line) linking D1 and D2; D4–D6: CMP12 (Lamp1).

These domains are clearly separated from the former domains, but appear to be closely apposed to the border of domain 3 (compare Fig. 2 part II, k, with Fig. 2 part II, m). Fig. 3 gives a schematic illustration of these detected domains. These findings are in keep with earlier results having shown that Hoechst 3328 and propidium iodide differentially bind to nuclear DNA structures [29]. Moreover, the data also coincide with reports having shown that cytoplasmic factors, such as Lamp1 [41] and mitochondria-associated molecules, such as the apoptosis-inducing factor (AIF) [42] are recruited to the cell nucleus during apoptosis. Recruitment of mitochondrial Cytc may be in line with nuclear translocation of mitochondrial AIF, but to our knowledge nuclear Cytc recruitment, as suggested by the present study, has not been clearly shown yet. Cytc released from mitochondria can initiate apoptosis by triggering a caspase-activating complex (the apoptosome) [7,8]. The apoptosome, a cytosolic signalling platform that activates a cascade of proteolytic caspases, is a wheel-like molecular machine composed of Cytc and the cytosolic protein Apaf-1 [8]. In our present study we did not colocalize Apaf-1. This leaves open whether the Cytc (CMP5) signals detected in the nuclear interior singles out intranuclear apoptosomes or simply reflects nuclear Cytc translocation. Localization of c-fos (CMP3) in the region of nuclear DNA (CMPs 8, 25, and 26) is conceivable because c-fos is a transcription factor for early response genes, which is upregulated in liver cell apoptosis [40]. To our knowledge, presence of CD49f (CMP1) inside the cell nucleus during apoptosis is a new finding. It is in keep with upregulation of CD49f in liver diseases [35].

As illustrated in the SRTC map (Fig. 4a) the surface-rendered intranuclear signals of Cytc (corresponding to domain 1 of Fig. 2, part I) are seen (Fig. 4a, asterisk). These signals are located in close proximity to the condensed nuclear DNA (green colour in Fig. 4a). The latter corresponds to domain 3 (Fig. 2 part II). It represent a major hallmark of apoptosis [7]. Interestingly and in addition to the 2D optical sections from Fig. 2, the 3D image of Fig. 4a uncovers that Cytc forms small spherical structures which are directly facing the surface of the condensed DNA (Fig. 4a, arrow 5). This suggests that these sites may correspond to specific interaction structures of Cytc and nuclear DNA. When all of the 39 most frequent CMPs (Table 2) are simultaneously displayed as a CMPT map at 3D (Fig. 4b), intranuclear CMP28 is detected in addition to the CMPs which have already been described in Fig. 2. It shows a clear cut intranuclear location (Fig. 4b, arrow 28). It contains signals of both CD138 (syndecan-1) and of the glycotopes NeuAc/(GlcNac)2/(GlcNac)3 visualized by binding of the wheat germ agglutinin WGA. To our knowledge these molecular assemblies have not been shown yet at intranuclear sites in apoptosis. WGA binding to nuclear pore proteins has been shown earlier [31] and regulation of syndecan-1 can be implicated in apoptosis [33]. Whether the association of these molecules as CMP28 reflects a specific function in apoptosis is not known. The close proximity of CMP28 to the nuclear envelope (denoted by red cytokeratin rim) may suggest that

3.3. Synopsis at 3D The total number of CMPs that we have detected in the apoptotic liver cell significantly exceeds those illustrated in Fig. 2. The intranuclear domains shown in Fig. 2 give the most prominent features of liver cell apoptosis detected by toponomics so far. These domains were derived from two out of ten optical sections across the cell. Therefore we attempted to provide a more synoptic view of this cell nucleus by rotating it to an oblique position (Fig. 4). Two different visualization procedures were chosen to show how the detected intranuclear domains are spatially interrelated in x/y and z across all of the ten optical sections of the cell: a surface-rendered toponome colocalization (SRTC) map and a CMP-toponome (CMPT) map were generated for direct visual comparison (Fig. 4a and Fig. 4b, respectively).

Fig. 4. 3D toponome images (rotated to an oblique position) are shown to illustrate a detail of the hepatocyte cell nucleus (identical cell from Figs. 1, 2 and 3). (a) Surfacerendered toponome colocalization (SRTC) map; (b) corresponding CMP-toponome map (CMPT map). Note that the numbers of arrows indicate concrete CMPs (decoded in Table 2) that define mutually exclusive sites and domains in the nuclear interior. The 3D aspect uncovers the spatial interrelationship of these CMPs. Bar: 5 μm.

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088

this association is brought about by the action of caspases on proteins which are physiologically located at or near the perinuclear structures. 4. Discussion In the present study we have applied TIS to address apoptosis in an attempt to examine the feasibility to construct large-scale toponome reference maps of the cell nucleus on the single cell level. The demonstrated ability to colocalize 15 different MCC in one apoptotic cell at 3D is a milestone towards much larger maps focussing on the complete apoptosis-specific molecular network in one cell, referred to as the apoptosis toponome. We consider that this large-scale imaging approach will be feasible and essential to unravel the molecular organization of apoptosis, for the following reasons. First, we have shown that at least 100 different MCC can be colocalized to reveal high-dimensional toponome maps of one biological sample [4]. This suggests that large tag libraries focussing on at least 100 molecules relevant for apoptosis can be constructed in the future for apoptosis-toponome mapping procedures. Powerful software and mathematical methods for toponome analyses have been developed [48–51], that can be readily applied to apoptosistoponome research. Second, progress in understanding apoptosis has followed separate lines of investigation which can be grouped in biochemical and imaging studies. These studies gave rise to ideas on how the many identified components of apoptosis might be interconnected as an apoptosis-specific molecular network [7,8]. One critical mechanism in apoptosis is the formation of the apoptosome, a molecular complex that is formed between cytochrome c and Apaf-1 that activates a cascade of caspases finally leading to a controlled demolition of the cell [7,8]. It is known from proteomics studies that caspases, as the agents of this demolition process, act on hundreds of proteins [52], and nearly every cell organelle as well as many cytoskeletal proteins are targets of pronounced fragmentation. Proteolysis of the cytoskeleton probably contributes to the rounding and retraction of the cell seen in the early stages of apoptosis [7]. Despite extensive data assembled around these processes, significant gaps remain in our knowledge of these mechanisms. For example, which protein clusters are formed between the 10 known caspases and cellular target proteins as substrates in a cell undergoing apoptosis; where are these clusters located in apoptotic cells; how are these clusters interrelated as a network, and how do they change during apoptosis? Moreover, accumulating evidence appears to suggest that caspases participate in other cellular events, such as differentiation [7]. This suggests that the identical caspases may differentially associate with various MCC at different subcellular locations to generate different molecular networks with topologically and functionally distinct properties. Furthermore, toponomics will allow in the future to colocalize all the molecular compontens of the apoptosome to address the question when and where these components associate during the apoptosis. Important clues to the identification and understanding of these networks may come from large-scale toponome mapping studies. Third, the importance of large-scale toponome mapping becomes evident from our studies showing that two different functional states of a cell (spherical vs. explore state) can be easily distinguished by the protein clusters formed, rather than by averaging the abundance of the corresponding proteins in these cells [4]. The explore state was characterized by specific protein clusters interlocked as arrays along the cell surface. These clusters had in common one protein (APN) while other proteins (different cell surface integrins) were variably associated with the former protein. We termed this protein the lead protein. We could show that inhibition of this lead protein led to disassembly of the protein clusters and to severe loss of function (ability to polarize), substantiating the lead protein hypothesis [14] and providing a rationale for the identification both of cellular protein

2087

networks and the corresponding control protein(s) in situ/in vivo. We consider that this new paradigm may be applied to apoptosis: Once lead proteins have been identified in apoptosis-specific toponome maps (vs. control cells), they can be selectively inhibited or deleted, and the functional consequences observed. The construction of apoptosis-specific toponome maps as an approach to analyse the functional organization of apoptosis-specific molecular networks, is still in its infancy. The map of the apoptotic cell nucleus presented in this study is a first ‘draft’ of a toponome reference map in apoptosis. It gives multiple molecular ‘landmarks’ in the nuclear interior during apoptosis of a human liver cell. We consider the six nuclear domains identified by simultaneous CMP imaging as essential territories of the intranuclear processes during apoptosis. A systematic TIS-based approach using the identified apoptosis-specific domains as landmarks together with a proteomewide set of affinity ligands might decipher the complete functional plan of apoptosis using lead proteins as anchor points. We believe that the toponome-based identification of the protein hierarchies in apoptosis, as revealed by lead proteins within their topologically defined CMP motives [4], will become essential for a comprehensive cellular understanding of apoptosis. A corresponding mapping platform is in progress [53]. Acknowledgements The study was funded by BMBF through NGFN2, subproject FKZ01Gr0446, NGFNplus, and NBL3, the Deutsche Forschungsgemeinschaft through Schu672/10-1, the Tschira foundation and LOM of the medical faculty, University of Magdeburg. References [1] X. Wei, S. Somanathan, J. Samanabandu, R. Berezney, Three-dimensional visualization of transcription sites and their association with splicing-rich nuclear speckles, J. Cell Biol. 146 (1999) 543–558. [2] D. Spector, Macromolecular domains within the cell nucleus, Annu. Rev. Cell Biol. 9 (1993) 265–315. [3] B. Alberts, The cell as a collection of protein machines: preparing the next generation of molecular biologists, Cell 6 (1998) 291–294. [4] W. Schubert, B. Bonnekoh, A.J. Pommer, L. Philipsen, R. Boeckelmann, J. Maliykh, H. Gollnick, M. Friedenberger, M. Bode, A. Dress, Analyzing proteome topology and function by automated multidimensional fluorescence microscopy, Nat. Biotechnol. 24 (2006) 1270–1278. [5] M. Friedenberger, M. Bode, A. Krusche, W. Schubert, Fluorescence detection of protein clusters in individual cells and tissue sections by using toponome imaging system: sample preparation and measuring procedures, Nat. Protoc. 2 (2007) 2285–2294. [6] A. Abbott, Mapping togetherness (Research highlight), Nature 443 (2006) 609 (referring to ref. 4). [7] R.C. Taylor, S.P. Cullen, S.J. Martin, Apoptosis: controlled demolition at the cellular level, Nat. Rev. Mol. Cell Biol. 9 (2008) 231–241. [8] S.J. Riedl, G.S. Salvesen, The apoptosome: signalling platform of cell death, Nat. Rev. Mol. Cell Biol. 8 (2007) 405–413. [9] W. Schubert, Multiple antigen — mapping microscopy of human tissue, in: G. Burger, M. Oberholzer, G.P. Vooijs (Eds.), Excerpta medica, Elsevier, Amsterdam, 1990, pp. 97–98, Adv. Analyt. Cell. Pathol. [10] W. Schubert, Exploring molecular networks directly in the cell, Cytometry A. 69 (2006) 109–112. [11] W. Schubert, A three symbol code for organized proteomes based on cyclical imaging of protein locations, Cytometry A. 71 (2007) 352–360. [12] W. Schubert, Cytomics in characterizing toponomes: towards the biological code of the cell, Cytometry A. 69 (2006) 209–211. [13] W. Schubert, Breaking the biological code, Cytometry A. 71 (2007) 771–772. [14] W. Schubert, Topological proteomics, toponomics, MELK technology, Adv. Biochem. Eng. Biotechnol. 83 (2003) 189–209. [15] R.C. Ecker, B. Rogojanu, M. Streit, K. Osterreicher, G.E. Steiner, An improved method for discrimination of cell populations in tissue sections using microscopybased multicolor tissue cytometry, Cytometry A. 69 (2006) 119–123. [16] R.C. Ecker, A. Tarnok, Cytomics goes 3D: toward tissomics, Cytometry A. 65 (2005) 1–3. [17] A. Mittag, D. Lenz, A.O. Gerstner, A. Tarnok, Hyperchromatic cytometry principles for cytomics using slide based cytometry, Cytometry A. 69 (2006) 691–703. [18] W. Laffers, A. Mittag, D. Lenz, A. Tarnok, A.O. Gerstner, Iterative restaining as a privotal tool for n-color immunophenotyping by slide-based cytometry, Cytometry A. 69 (2006) 127–130. [19] E. Bedner, L. Du, F. Traganos, Z. Darzynkiewicz, Caffeine dissociates complexes between DNA and intercalating dyes: application for bleaching fluorochrome-stained

2088

[20]

[21]

[22]

[23]

[24] [25]

[26]

[27]

[28]

[29] [30]

[31]

[32]

[33]

[34]

W. Schubert et al. / Biochimica et Biophysica Acta 1783 (2008) 2080–2088 cells for their subsequent restaining and analysis by laser scanning cytometry, Cytometry 43 (2001) 38. W. Schubert, Antigenic determinants of T Iymphocyte α/β receptor and other leukocyte surface proteins as differential markers of skeletal muscle regeneration: detection of spatially and timely restricted patterns by MAM microscopy, Eur. J. Cell Biol. 58 (1992) 395–410. L. De Angelis, L. Berghella, M. Coletta, L. Lattanzi, M. Zanchi, M.G. Cusella De Angelis, C. Ponzetto, G. Cossu, Skeletal myogenic progenitors originating from embryonic dorsal aorta coexpress endothelial and myogenic markers and contribute to postnatal muscle growth and regeneration, J. Cell Biol. 147 (1999) 869–878. M. Sampaolesi, S. Blot, G. D'Antona, N. Granger, R. Tonlorenzi, A. Innocenzi, P. Mognol, J.L. Thibaud, B.G. Galvez, I. Barthélémy, L. Perani, S. Mantero, M. Guttinger, O. Pansarasa, C. Rinaldi, M.G. Cusella De-Angelis, Y. Torrente, C. Bordignon, R. Bottinelli, G. Cossu, Mesangioblast stem cells ameliorate muscle function in dystrophic dogs, Nature 444 (2006) 552–553. B. Zheng, B. Cao, M. Crisan, B. Sun, G. Li, A. Logar, S. Yap, J.B. Pollett, L. Drowley, T. Cassino, B. Gharaibeh, B.M. Deasy, J. Huard, B. Peault, Prospective identification of myogenic endothelial cells in human skeletal muscle, Nat. Biotechnol. 25 (2007) 1025–1034. Y. Torrente, et al., Autologous transplantation of muscle-derived CD133+ stem cells in Duchenne patients, Cell Transplant. 16 (2007) 563–577. W. Schubert, Method of blocking cytotoxic activity in patients with amyotrophic lateral sclerosis using antibodies to FcgammaRIII, 1999 US patent No. 6,638,506 (first published as international patent application WO 99/29731, 1999). H.A. Mohamed, D.R. Mosier, L.L. Zou, L. Siklós, M.E. Alexianu, J.L. Engelhardt, D.R. Beers, W.D. Le, S.H. Appel, Immunoglobulin Fcgamma receptor promotes immunoglobulin uptake, immunoglobulin-mediated calcium release, and neurotransmitter release in motor neurons, J. Neurosci. Res. 69 (2002) 110–116. M. Bode, M. Irmler, M. Friedenberger, C. May, K. Jung, C. Stephan, H.E. Meyer, C. Lach, R. Hillert, A. Krusche, J. Beckers, K. Marcus, W. Schubert, Interlocking transcriptomics, proteomics and toponomics technologies for brain tissue analysis in murine hippocampus, Proteomics 8 (2008) 87–96. W. Schubert, M. Bode, R. Hillert, A. Krusche, M. Friedenberger, Toponomics and neurotoponomics: a new way to medical systems biology, Expert Rev. Proteomics 5 (2008) 169–361. X. Shi, R.B. Macgregor Jr, Volume and hydration changes of DNA-ligand interactions, Biopohysical Chemistry 125 (2007) 471–482. R. Haars, A. Schneider, M. Bode, W. Schubert, Secretion and differential localization of the proteolytic cleavage products Abeta40 and Abeta42 of the Alzheimer amyloid precursor protein in human fetal myogenic cells, Eur. J. Cell Biol. 79 (2000) 400–406. D. Vautier, P. Chesné, C. Cunha, A. Calado, J.P. Renard, M. Carmo-Fonseca, Transcription-dependent nucleocytoplasmic distribution of hnRNP A1 protein in early mouse embryos, J. Cell Sci. 114 (2001) 1521–1531. P.N. Bishop, R.E. Bonshek, C.L. Jones, A.E. Ridgway, R.W. Stoddard, Lectin binding sites in normal, scarred and lattice dystrophic corneas, Brit. J. Ophthalmol. 75 (1991) 22–27. H. Sun, I.M. Berquin, R.T. Owens, J.T. O'Flaherty, I.J. Edwards, Peroxisome proliferator-activated gamma-mediated up-regulation of syndecan-1 by n-3 fatty acids promotes apoptosis of human breast cancer cells, Cancer Res. 68 (2008) 2912–2919. V. Cortes, L. Amigo, K. Donoso, I. Valencia, V. Quinonses, S. Zanlungo, E. Brandan, A. Rigotti, Adenovirus-mediated hepatic syndecan-1 overexpression induces

[35]

[36]

[37] [38] [39]

[40]

[41]

[42]

[43]

[44] [45] [46] [47]

[48]

[49]

[50] [51]

[52] [53]

hepatocyte proliferation and hyperlipidaemia in mice, Liver Int. 27 (2007) 569–581. J. Laurson, C. Selden, M. Clements, D. Mavri-Damelin, S. Coward, M. Lowdell, H.J. Hodgson, Putative human liver progenitor cells in explanted liver, Cells Tissues Organs 186 (2007) 180–191. N.O. Ku, P. Strnad, B.H. Zhong, G.Z. Tao, M.B. Omary, Keratins let liver live: Mutations predispose to liver desease and crosslinking generates Mallory-Denk bodies, Hepatology 46 (2007) 1639–1649. J.F.R. Kerr, A.H. Wyllie, A.R. Currie, Apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics, Br. J. Cancer 24 (1972) 239–275. A.H. Wyllie, J.F. Kerr, A.R. Currie, Cell death: the significance of apoptosis, Int. Rev. Cytol. 68 (1980) 251–305. A.M. Robertson, C.C. Bird, A.W. Waddell, A.R. Currie, Morphological aspects of glucocorticoid-induced cell death in human lymphoblastoid cells, J. Pathol. 126 (1978) 181–187. C. Bernt, T. Vennegeerts, U. Beuers, Ch. Rust, The human transcription factor AP-1 is a mediator of bile acid-induced liver cell apoptosis, Biochem. Biophys. Res. Comm. 340 (2006) 800–806. M. Nakahara, A. Nagasaka, M. Koike, K. Uchida, K. Kawane, Y. Uchiyama, S. Nagata, Degradation of nuclear DNA by Dnase II-like acid Dnase in cortical fiber cells of mouse eye lens, FEBS J. 274 (2007) 3055–3064. X. Zhang, J. Chen, S.H. Graham, L. Du, P.M. Kochanek, R. Droviam, F. Guo, P.D. Nathaniel, C. Szabó, S.C. Watkins, R.S. Clark, Intranuclear localization of apoptosisinducing factor (AIF) and large scale DNA fragmentation after traumatic brain injury in rats and neuronal cultures exposed to peroxinitrite, J. Neurochem. 82 (2002) 181–191. T. Sakai, L. Lui, X. Teng, R. Mukai-Sakai, H. Shimada, R. Kaji, T. Mitani, M. Matsumoto, K. Toida, K. Ishimura, Y. Shishido, T.W. Mak, K. Fukui, Nucling recruits Apaf-1/pro-caspase-9 complex for the induction of stress-induced apoptosis, J. Biol. Chem. 279 (2004) 41131–41140. M. Bode, A. Krusche, Toponome Imaging System (TIS): imaging the proteome with functional resolution. Application note, Nat. Methods 4 (2007) iii–iv. G.K. Zipf, Human behavior and the principle of least effort, Addison-Wesley, Cambridge MS, 1949. R.F. Murphy, Putting proteins on a map, Nat. Biotechnol. 24 (2006) 1223–1224. J.A. Conchello, J.G. McNally, Fast regularization technique for expectation maximization algorithm for computational optical sectioning microscopy: Image acquisition and processing, in: C.J. Cogswell, G.S. Kino, T. Wilson (Eds.), Proc. SPIE, 2655, 1996, pp. 199–208. T. Nattkemper, H. Ritter, W. Schubert, A neural classifyer enabling highthroughput topological analysis of lymphocytes in tissue sections, IEEE Trans. Inf. Techn. in Biomed. 5 (2001) 138–149. T.M. Nattkemper, T. Twellmann, W. Schubert, H. Ritter, Human vs. machine: Evaluation of fluorescence micrographs, Computers in Biology and Medicine 33 (2003) 31–43. A. Dress, T. Lokot, L.D. Pustyl`nikov, W. Schubert, Poisson numbers and poisson distributions in subset surprisology, Ann. Combinatorics 8 (2004) 473–485. W. Schubert, M. Friedenberger, R. Haars, T. Nattkemper, H. Ritter, Automatic recognition of muscle invasive T lymphocytes expressing dipeptidyl-peptidase IV (CD26), and analysis of the associated cell surface phenotypes, J. Theoret. Med. 4 (2002) 67–74. A.U. Lüthi, S.J. Martin, The CASBAH: a searchable database of caspase substrates, Cell Death Differ. 14 (2007) 641–650. K. Cottingham, Human toponome project, J. Proteome Res. 7 (2008) 1806.