Co-localisation with nuclear markers

For the following immunolocalisations the antibody Grosveld 2 was used on Hela cells unless stated otherwise. First, co-localisations with markers staining different nuclear subcompartments were carried out (Fig. 19). DEK- and DNA-staining by propidium iodide overlap after RNAse treatment to reduce background staining of RNA by the dye. However, neither a preference for transcriptionally active

euchromatin nor for heterochromatin as visualised by tetra-acetylated histone H4 and HP1α staining, respectively, could be seen. Comparison with fibrillarin, a component of nucleoli confirmed that DEK is almost absent in this compartment.

The 'nuclear matrix' or 'nuclear scaffold' is the structure that remains after extensive nuclease treatment of nuclei. It consists mainly of proteins, but also to a considerable amount of nuclease-protected RNA. Proteins that are localised at different regions of the nuclear matrix were used as markers: Lamin A/C, two proteins made from the same transcript by alternative splicing are a main component of the nuclear lamina sitting on the inside of the inner nuclear membrane. Some Lamin A/C is also found throughout the nucleus. The other marker is SAF-A, the scaffold attachment factor A that is a component of ribonuclear particles and binds to A/T-rich scaffold/matrix attachment regions on DNA (SAR/MAR DNA). Comparison with the two markers shows that the intensity of DEK staining is reduced towards the periphery of the nucleus. This contrasts to SAF-A, whose concentration decreases towards the center of the nucleus.

Fig. 19 DEK versus markers for subnuclear domains. The DEK antibody Grosveld 2 was used with antibodies against the proteins indicated on the left for co-immunolocalisations on Hela-cells.

Exceptions are indicated with indices: ¹GFP-DEK; ²monoclonal DEK. Scale-bar: 5µm. DEK does not co-localise with a marker protein for subnuclear domains, although there is some overlap with DNA and euchromatin. DEK is excluded from the nucleoli and does not stain the periphery of nuclei (compared to Lamin A/C and SAF-A).

Functional markers for four nuclear processes were examined next: replication (α-PCNA, α-BrdU), transcription (α-BrU), DNA repair (γH2AX) and, as a summary of the previous chapter, splicing (α-SR-proteins). With markers for replication it is also possible to roughly determine the progress of S-phase when cells are at this stage of the cell-cycle. Because heterochromatin is replicated late in S-phase, stained islets of replication activity is interpreted as heterochromatin and the cell is thus in late S-phase. Cells with distinct heterochromatin are mouse 3T3 cells, therefore the

replication-coimmunolocalisations are carried out in this cell-line. Replication markers used here are PCNA, the sliding clamp that stains replication forks and also DNA-repair sites. The other marker is BrdU, which was added to the medium 30 minutes before the experiment. Thus in contrast to PCNA, BrdU marks the DNA that has been replicated within the last half hour. The results of Fig. 20 demonstrate that DEK is neither accumulated nor reduced at sites of DNA-replication. Neither is the DEK-level changed in newly synthesised DNA in comparison to yet unreplicated DNA or DNA that was replicated longer than 30minutes ago (BrdU staining). BrdU treatment was also carried for 2h or with a 30min incubation 'pulse' followed by removal of BrdU and an 18h 'chase' period without changing the result (data not shown). BrdU

staining of 3T3 cells also seems to confirm that DEK is not accumulated in heterochromatin. However, the density of DNA (DNA/volume) is higher in

heterochromatin compared to euchromatin, so equal staining may mean a lower DEK/DNA ratio. The experimental design applied here is not precise enough to make such statements, i.e. heterochromatin may be less accessible for antibodies.

The same result - neither accumulation nor reduction of DEK at sites of the marker-protein - is found for BrU staining which marks sites of transcription and of nascent transcripts when incubated for short times (here: 10 minutes). As already

demonstrated in 3.2.1, DEK does also not co-localise with splicing speckles. Finally, DEK does not migrate to sites of DNA-damage 1h after gamma-irradiation with 2Gy.

...

Fig. 20 DEK vs. functional markers. Co-localisation of DEK (Grosveld 2 antibody) with marker proteins (see text on the left) in Hela cells. ¹3T3-cells taken instead of Hela-cells to visualise heterochromatin. Scale-bar: 5µm. DEK does not co-stain with the used markers for DNA-repair and splicing. Some overlap between BrU + early S-phase-staining (PCNA / BrdU) and DEK is due to the extensive staining of the whole nuclei intrinsic to these markers.

DEK binds preferentially to four-way junction DNA (Waldmann et al. 2002), a property shared by architectural and nucleosome-binding proteins. Therefore DEK was compared to GFP-H1 (gift of M. Bustin, NIH, Bethesda, Maryland; Misteli et al.

2000) and GFP-HMGN1 (the former HMG-14/-17; gift of T. Misteli, NIH, Bethesda, Maryland; Phair and Misteli 2000; Catez et al. 2002), two core-nucleosome binding proteins. As seen in Fig. 21, the DEK stain overlaps with GFP-H1 fluorescence.

However, DEK's distribution appears more granular and is not as evenly spread over chromatin. Thus both proteins have a similar but probably not identical localisation.

This question is important in respect to the finding of Alexiadis et al. (Alexiadis et al.

2000) that DEK binds to histones H2A and H2B. In chapter 3.4.2 the binding between histones and DEK is re-examined by far Western blotting.

GFP-HMGN1, the other nucleosome binder, showed extensive staining of nucleoli.

Thus its localisation is inconsistent with that of cellular HMGN1, which is not present in nuclei. The GFP-construct worked well in a publication of Phair and Misteli (Phair and Misteli 2000), where GFP-HMGN1 was transiently transfected in BHK (Baby Hamster Kidney) cells.

Fig. 21 DEK vs. core histone binders. GFP-fusion constructs of two core-nucleosome -binding proteins, histone H1 and HMGN1 (= HMG-14/-17) compared to DEK (monoclonal DEK antibody).

Scale-bar: 5µm. Some overlap between DEK and histone H1 is seen, although the staining pattern of GFP-H1 is finer and less grainy compared to DEK. GFP-HMGN1 is incorrectly localised in nucleoli.