The HL cell lines U-HO1 had a doubling time of about 3-4 days, and U-HO1-mock about 7 days, whereas U-HO1-PTPN1 grew much slower with a doubling time of about 14 days. In steady state culture, the number of at least bi-nucleated RS-cells was about 4%-5% in both, U-HO1 and U-HO1-mock, but significantly higher (18-22%; p < 0.0001) in U-HO1-PTPN1 (Figure 1A, B). Both cell lines, U-HO1 and U-HO1-PTPN1 had equally high telomerase activity and in U-HO1-PTPN1 stable expression of the specific protein-tyrosin-phosphatase was confirmed by Western blotting (Figure 2A, B). Stable expression of PTPN1 was associated with a high apoptosis rate and had a significant impact on presence of phosphorylated STAT5, which was high in U-HO1 but nearly absent in U-HO1-PTPN1 (Figure 3A, B).

Mononuclear H-cells of U-HO1 and U-HO1-PTPN1 showed largely identical 3D telomere characteristics (Figure 4 and Table 1). The only difference was a small increase (p = 0.0407) of very short telomeres including so-called "t-stumps" (Xu and Blackburn, 2007), occurring in U-HO1 H-cells (Figure 4 and Table 1).

Reed-Sternberg cells of both, U-HO1 and U-HO1-PTPN1, showed several common characteristics as previously identified in RS-cells of the HL cell lines HDLM-2, L-428, L-1236 and patient biopsies 4 6 . As expected, U-HO1-RS-cells and U-HO1-PTPN1 RS-cells significantly differed from their mononuclear H-cell precursors by their nuclear volume (p < 0.0001), number of telomeres (p < 0.0001) and increase of telomere aggregates (p < 0.003). Strikingly, RS-cells of U-HO1 differed from U-HO1-PTPN1-RS cells by a highly significant increase in very short telomeres including so-called "t-stumps" (p < 0.0001) (Table 1 Figure 5). RS-cells of U-HO1 often showed numerous very short and short telomeres without loss of mid-sized telomeres and stable number of telomeres/1000 μm³ of nuclear volume, consistent with the hypothesis that these RS-cells were still able to divide further through endoreplication, resulting in giant RS-cells (Figure 6). Indeed, such telomere rich U-HO1-RS cells were regularly identified and telomere-poor or -free "ghost" nuclei were rare. Contrary to the U-HO1 RS-cells, the PTPN1 expressing RS-cells did not show any increase of very short telomeres (Figure 5 and 7) resulting in multinuclear RS-cells with a low number of telomeres/1000 μm³ of nuclear volume when compared to their mononuclear H-precursors (p = 0.0175). In such U-HO1-PTPN1 expressing RS-cells formation of telomere-poor "ghost" nuclei was frequently observed.

The cell lines U-HO1, U-HO1-PTPN1 and U-HO1-mock were grown in suspension in 25 cm² culture flasks in Iscove/RPMI-1640 medium (4:1) supplemented with 10% fetal calf serum, L-glutamine (2 mM), penicillin (100 U/ml) and streptomycin (100 U/ml) in 5% CO₂ in a humidified atmosphere at 37^o C. U-HO1 and U-HO1-mock were passed into new flasks containing fresh culture media twice weekly, U-HO1-PTPN1 once weekly, due to much slower growth.

Plasmids were transfected into U-HO1 cells using the Nucleofector device (Amaxa, Cologne, Germany). 2 × 10⁶ cells and 4 μg expression vector were used for each plasmid nucleofection (Nucleofector Kit V, program X-01). pcDNA4/TO vector was obtained from Invitrogen (Invitrogen, San Diego, CA) and used as a control mock-vector. The PTPN1 coding sequence (NM 003745, nucleotide 155 to 790) was amplified by PCR on cDNA from human tonsil and inserted into pcDNA4/TO vector under control of CMV promoter. The plasmids were purified using Endofree Plasmid Purification MaxiKit (Qiagen, Hilden, Germany). PTPN1 sequence was checked by sequencing (MWG Biotech, Ebersberg, Germany). Cells were harvested 24 h and 48 h post transfection or after at least 4 weeks under zeocin selection for stable transfection. Success of stable transfection was proved by immunoblotting against PTPN1.

TRAP (Telomeric Repeat Amplification Protocol) was performed using the TRAPeze Telomerase detection Kit (S7700, Chemicon, Temecula, CA, USA). 100 ng of protein extract were subject to the TRAPeze procdure with and without heat inactivation of telomerase (ΔT). Pictures were shown as the negative of Ethidium bromide coloration of a 10% PAGE (TBE 0.5X).

Apoptosis was assessed by flow cytometry according to the standard procedure of Nicoletti et al. 27 .

For Western blotting 20 μg of protein were charged per lane. The following primary antibodies were used: PTPN1 (murine clone AE4-2J) (Merck, Darmstadt, Germany), β-actin (murine clone AC-74), (Sigma, Deisenhofen, Germany). For detection peroxidase conjugated goat anti-rabbit IgG or goat anti-mouse IgG, (Peribo Science, Bonn, Germany) was used. Signals were visualized using the SuperSignal West Dura extended-duration substrate (Perbio Science, Aalst, Belgium).

Cytocentrifuge preparations of U-HO1 and and U-HO1-PTPN1 cells were fixed in acetone (5 min), air dried, and stained with MayGrünwald/Giemsa (MGG) according to standard conditions. For the detection of phospho-STAT5, fresh cells of U-HO1 and U-HO1-PTPN1 were collected, centrifuged, fixed in ethanol and embed in paraffin. The specimen were cut in 5 μm thick sectiones and further treated as usual. The primary monoclonal antibody against phosphorylated STAT5 (Cell Signaling Technology, Beverly, MA) was diluted of 1:50 and used according to manufactorers instructions. For detection the EnVision detection system (Dako, Copenhagen, Denmark) was used and 3-Amino-9-ethylcarbazole served as the substrate.

U-HO1 and U-HO1-PTPN1 cells were collected (200 × g for 10 min) and resuspended in PBS containing 3.7% formaldehyde (Fluka) and incubated for 20 min. Thereafter, the telomere Q-FISH protocol was performed 5 28 by using Cy3-labelled PNA probes (DAKO). Imaging of interphases after telomere FISH was performed by using Zeiss AxioImager Z1 with a cooled AxioCam HR B & W, DAPI, Cy3 filters in combination with a Planapo 63 ×/1.4 oil objective lens. Images were acquired by using AXIOVISION 4.6 (Zeiss) in multichannel mode followed by constraint iterative deconvolution as specified below.

At least 30 H-cell interphase nuclei and at least 30 RS-cell interphase polykaria were analyzed for both cell lines. AXIOVISION 4.6 with deconvolution module and rendering module were used. For every fluorochrome, the 3D image consists of a stack of 80 images with a sampling distance of 200 nm along the z and 107 nm in the xy direction. The constrained iterative algorithm option was used for deconvolution 29 .

Telomere measurements were done with TeloView TM 30 . By choosing a simple threshold for the telomeres, a binary image is found. Based on that, the center of gravity of intensities is calculated for every object resulting in a set of coordinates (x, y, z) denoted by crosses on the screen. The integrated intensity of each telomere is calculated because it is proportional to the telomere length 31 .

Telomere aggregates are defined as clusters of telomeres that are found in close association and cannot be further resolved as separate entities at an optical resolution limit of 200 nm 32 .

Telomeres with a relative fluorescent intensity (y-axis) ranging from 0-5000 units are classified as very short, with an intensity ranging from 5,000-15,000 units as short, with an intensity ranging from 15,000-30,000 units as mid-sized, and with an intensity >30,000 units as large. The cut-off of 5,000 units is arbitrary and based on our 5 years' experience. Telomeres with an intensity ≪ 5'000 are still easily identifiable on the Teloview-Program, but barely detectable on 3D reconstitution images.

Total telomere volume is the sum of all very short, short, mid-sized and large telomeres and aggregates within one H- or RS-cell.

The Nuclear volume is calculated according to the 3D nuclear 4',6-diamidino-2-phenylindoline staining as described earlier 33 .

For both cell lines, normally distributed parameters are compared between the two types of cells using nested or two-way analysis of variance. Multiple comparisons using the least-square means tests followed in which interaction effects between two factors were found to be significant. Other parameters that were not normally distributed were compared using a non-parametric Wilcoxon rank sum test. Significance levels were set at P = 0.05. Analyses were perfirmed using SAS v9.1 programs.