CONTRASTE DE POSIBILIDADES DE LO PLANTEADO EN EL PGIRS/2G Y

The results obtained by induced formation of HCC add to the understanding of nuclear architecture and enable to critically revisit the postulations of present model views. Which model views are compatible with observations of interconnected CTs, a reproducibly expandable interchromatin space and the lack of observing giant chromatin loops?

7.6.1 Giant chromatin loops

Are giant chromatin loops major elements of nuclear architecture as proposed by the random walk-giant loop (RW-GL) computer model (Sachs et al., 1995) or by experimental observations on the localization of HSA11p15.5 (Mahy et al., 2002a)? The relevance of the RW-GL model was already discussed in the Introduction (see 4.3.4). Experimental data correlating with the theoretical RW-GL calculations was also fully compatible with the multi- loop subcompartment (MLS) model where no giant loops are implemented.

The presented experiments of this study do also not support any claims for the existence of giant 30 nm fiber chromatin loops. No extended fields of out-looped DNA fibers (Fig. 1D) were observed around CTs in fixed as well as in living cell nuclei. In fact, investigations focused on the highly expressed and protruded regions of HSA 11p15.5 and HSA 6p21.3 revealed that even these are built from higher order chromatin with compaction ratios much beyond that of sheer 30 nm fibers (Fig. 31). These results rule out the claim of extended chromatin ‘out-looping’, which was indirectly concluded from incomplete visualizations of only parts of these regions (Chubb and Bickmore, 2003; Mahy et al., 2002a). The presented results, however, do match postulations of the MLS-model (Fig.1 C) for the existence of small scale chromatin loops building up sub-compartments. These sub-compartments could correspond to the observed ~1 Mb chromatin domains as well as to the chromomeres observed by Wanner and Formanek, 2000. A chain of such domains could accordingly extend from the respective CT building up protruded regions like observed for the investigated highly expressed loci (see also subsequent chapter, 7.7). An organization of chromatin in small scale sub-compartments would additionally demand much less orchestration during the conformational changes over a NCC-HCC-NCC cycle compared to a model claiming entangled giant loops.

Applying the ‘Occam's razor’ principle the simpler explanation of an observation should be preferred. Since no tenable evidence for extended out-looped 30 nm chromatin fibers could be provided, this postulation is actually not favored.

7.6.2 'Nuclear matrix'

The 'nuclear matrix', a postulated scaffold organizing chromatin in interphase nuclei is another nuclear component, whose existence is questioned (Hancock, 2000; Pederson, 1998; Pederson, 2000). Again, if its proposed function could be adopted by more simple and already demonstrated principles, postulations for its existence would be challenged (Cremer et al., 1995; Hancock, 2000; Hancock, 2004b; Pederson, 2000). In that line of argument it was discussed that the structural organization of a cell nucleus could largely be triggered by macromolecular crowding effects with no need for an additional scaffold (Hancock, 2004b). Kreth et al demonstrated that the existence of distinct CTs could be fully explained with a computer model just based on sphere-like ~1 Mb chromatin domains interconnected by short flexible chromatin linkers (Cremer and Cremer, 2001; Kreth et al., 2004b).

Factors proposed to be components of the 'nuclear matrix' (e.g. SAF-A) preferentially localize in the interchromatin space and at the surface of HCC bundles as demonstrated in this study (Fig. 25). Since these seem to be the regions subjected to small scale movements during HCC formation (Fig. 15), a potential 'nuclear matrix' would necessarily be considerably flexible. However, if the 'nuclear matrix' would be responsible for the stable organization of

chromatin, how could it fulfill this function, granting the strikingly similar patterns of HCC, although it itself gets massively subjected to global topological changes whilst HCC formation? These considerations would rather suggest a 'nuclear matrix' to be located in the interior of HCC bundles, potentially at the bases of multi-loop sub-compartments (Munkel et al., 1999; Sadoni et al., 2004) and acting as a backbone for the emerging HCC bundles at their central axis. On the other hand, since the 'nuclear matrix' is also postulated to be involved in the stable anchoring of active small scale chromatin loops (Cook, 1999), a localization at sites of functional activity would be expected. However, as demonstrated in this study, these sites were localized at the HCC bundles surfaces.

In summary, the results presented are contradictory to an organizing 'nuclear matrix' and go better with a view that 'nuclear matrix' preparations lead to the precipitation of interchromatin proteins, which in part are involved in nuclear functions (Cremer et al., 1995; Pederson, 2000). This matches with the localization of 'nuclear matrix' proteins at the periphery of HCC bundles and is additionally in accordance with the rapid recovery of all tested nuclear functions, when the massive conformational changes of chromatin were reversed.

Still, the question whether and how small scale loops of active chromatin are tethered remains open. In the view of Peter Cook, not the polymerases are the mobile entities while transcription or replication is processed, but the DNA template is pulled through a fixed factory of polymerases and additional factors (Cook, 1999). This claim derives from reflections of the entwining and replication/transcription processes, which would necessarily lead to extensive entanglement of the replicates/transcripts with their respective templates in case of an immobile DNA template. The question on how these factories are fixed remains to be addressed.

7.6.3 Interchromatin space

The observed widening of interchromatin spaces in a non-random reproducible pattern supports the claim for the existence of a channel-like compartment as postulated by the chromosome territory-interchromatin compartment (CT-IC) model (Cremer and Cremer, 2001, Fig. 1H). According to the CT-IC model, the observed 3D interchromatin network would represent a widened interchromatin compartment. Since functional processes are postulated to be restricted to the surface of chromatin domains, a preferential localization of these processes at the surface of the HCC bundles would have been expected according to the CT-IC model. Indeed, nascent RNA, nascent DNA and RNA polymerases located around the surface of chromatin bundles (Fig. 22 and 23).

The older interchromatin domain model postulating interchromatin spaces to completely surround every CT (Zirbel et al., 1993, Fig. 1G), could be ruled out by the presented results. Formation of HCC revealed that CTs are not organized as single entities embedded in

interchromatin space, but are interconnected in a 3D chromatin network aligned by a 3D interchromatin channel system.

Preferential localizations of nuclear functions as postulated by the CT-IC model and as found in this study are hard to interpret in the light of the interchromatin network (ICN) model (Branco and Pombo, 2006; Branco and Pombo, 2007). The extensive chromatin intermingling postulated to lead to the ICN (Fig. 1E) would necessarily be expected to either [1] maintain the extent of intermingling during the HCC state leading to extensively intermingled fibers in the HCC bundles with no higher order of active versus inactive sites or lead to [2] de-mingling of chromatin fibers, which would be unlikely to yield the same chromatin pattern in repeated treatments. Such a reproducibility of HCC pattern formation would demand a complex and organized orchestration of intermingling chromatin fibers to assure their correct de-mingling when retracted and correct and organized re-intermingling when recovering the NCC state. Branco et al. deduced their postulation of wide fields of intermingling fibers between neighboring CTs from the observed overlap of fluorescent signals in thin (150 nm) cryosections. Although their approach allowed a much better resolution compared to confocal microscopy of intact nuclei, their images still suffered from out-of-focus light, were still limited by an axial resolution of ~200-300 nm and their evaluations were largely depending on the user set gray value thresholds. Considering these limitations, (Albiez et al., 2006) came to the conclusion, that the overlap of neighboring CTs was very much overestimated by (Branco and Pombo, 2006). In fact, an updated CT-IC model given below (see 7.7), is fully compatible with a partial overlap of ~1 Mb chromatin domains of neighboring CTs in 150 nm thick cryosections without the claim for any intermingling fibers (see supplemental in Albiez et al., 2006).

7.6.4 Chromatin conformation

The ICN model was supported by observations at the resolution level of EM revealing chromatin structures in the order of 30 nm fiber (Dehghani et al., 2005). The deduced lattice- model describes the entire chromatin to form a network of chromatin fibers with no order in respect to the space in between (Dehghani et al., 2005) Fig. 1F). Indeed, a huge channel system of interchromatin spaces cannot be distinguished with the electron spectroscopic imaging (ESI) method, which delineates and distinguishes the distribution of phosphorus and nitrogen in ultrathin EM sections. However, the conclusion, that the observed distribution of the two elements indeed reflects the distribution of only DNA and proteins respectively has to be critically scrutinized. First of all, RNA contains phosphorus as well, hindering a distinguishing between DNA and RNA (Cremer and Cremer, 2006b). As soon as RNA transcription is terminated, the transcript may get released to the interchromatin channel system and thereby blurring the topography of DNA as delineated by the ESI technique.

Furthermore heavily phosphorylated proteins could hamper a proper separation of the signals (Cremer and Cremer, 2006b). Applying the osmium ammine staining technique, which is supposed to solely stain DNA by a Feulgen-type reaction (Cogliati and Gautier, 1973), lacunas and channels in the order as described in this and previous studies are detected (Cremer and Cremer, 2001; Verschure et al., 1999; Visser et al., 2000).

Nevertheless, also the osmium ammine approach delivers images which suggest a structural organization of at least some nuclear regions in the order of 30 nm fibers (e.g. Fig 24). Where are the ultrastructural counterparts of the ~1 Mb chromatin domains observed at the light microscopic level? Dheghani et al. claim that ~1 Mb chromatin domains are misperceptions of local concentrations of 30 nm fibers (Dehghani et al., 2005). However their statement may as well be a misinterpretation of local chromatin arrangements, missing to view larger regions as organized entities. A comprehensive view on these considerations is given at the end of this discussion (see 7.8).

7.7 Updated chromosome territory-interchromatin