8.2.1 Experimental conditions
Two main aspects had to be considered for selecting an appropriate experimental setup: First, nothing is known about the half-life of MLL-AF9 protein, which might be significantly different to MLL protein. Second, MLL-AF9 leads to changes in the expression of target genes via epigenetic mechanisms (Mohan, et al. 2010; Slany 2009) whose reversal might take considerably more time than e.g. direct interaction by transcription factors (Ehrenhofer-Murray 2004; Cheng, et al. 2008). Thus a prolonged knockdown of MLL-AF9 over several days was envisaged. To simplify the experimental procedure, we aimed at a stable transfection of THP1 cells with shRNA plasmids (two plasmids containing distinct shRNA sequences against MLL-AF9 and one with a control shRNA sequence). Surprisingly, though over 94% of cells remained positive for the surface selection marker LNGFR, only 1% of cells stayed positive for the GFP marker and no MLL-AF9 knockdown could be observed via qRT-PCR. Three main reasons could account for this failure to downregulate MLL-AF9: (1) the H1 promoter of the shRNA could be inactivated like obviously the CMV promoter of GFP was, (2) shRNA levels could have been too low to efficiently affect MLL-AF9 transcript levels or (3) the shRNA was not processed correctly. Because of the time-consuming generation of stable transfected THP1 cells, we decided to proceed with the study employing the tested, very efficient and gentle transient transfection of siRNAs into THP1 via Dreamfect. To nevertheless ensure a functionally efficient MLL-AF9 knockdown, experiments were performed over eight days with repeated siRNA transfections on day 0, 3 and 6. Prior to each transfection event, cell densities were determined by Cellscreen microscopy and cells were reseeded at 0.05×106 cells per ml. Fetal bovine serum (FBS), containing growth factors, has previously been reported to affect miRNA expression, leading to a mitogenic miRNA profile (Paroo, et al. 2009). These effects might mask changes which result from the knockdown treatment. Thus fetal bovine serum levels were reduced to 0.5% from day 6 on. This concentration was determined to correspond to culture conditions in which THP1 cells remain proliferating over at least 6 days, while showing a clearly reduced proliferation level. Analyses were usually performed on day 8 of experiments which corresponds to 48 hours after serum reduction and last transfection.
8.2.2 Validation of MLL-AF9 knockdown
Knockdown was efficient and reduced MLL-AF9 transcript levels on day 8 of experiments to 22 ±6 % residual expression (figure 11 a), while MLL and AF9 wildtype levels were not significantly altered at any time point (day 8 levels are shown in figure 12).
Results
40
Figure 11: Confirmation of MLL-AF9 knockdown in THP1 cells over the experimental time course. Employed were two siRNA against MLL-AF9 (siR_3 and siR_4) and two non-targeting control siRNAs (n-tg_1 and n-tg_2). (A) Relative MLL-AF9 transcript levels measured by qRT- PCR. (B) Relative HOXA9 transcript levels which served as surrogate marker of MLL-AF9 protein reduction. MLL-AF9 has previously been shown to directly upregulate HOXA9 transcription. Graphs represent data from five independent experiments. Bars indicate standard error of the mean.
Figure 12: MLL and AF9 wildtype transcript levels in presence or absence of MLL-AF9
knockdown in THP1 cells. Relative transcript levels on day 8 of experiments are shown. Employed siRNA were: siR_3 and siR_4 against MLL-AF9 and non-targeting control siRNAs n-tg_1 and n-tg_2. Graph represents data from five independent experiments. Bars indicate standard error of the mean.
MLL-AF9 protein is hard to detect, which is already implied by the fact that other groups working with knockdown strategies to reduce endogenous MLL-AF9 were not able to show reduction on protein level (Kawagoe, et al. 2001; Pession, et al. 2003). Here, MLL-AF9 from THP1 cells was detectable on Western blot by antibodies detecting MLL N-terminus or AF9 C-terminus in protein isolated from the organic phase of Qiazol (similar to Trizol) and redissolved in equal volumes of 2% SDS and 2 × Laemmli sample buffer (figure 13 a) or in equal volumes of 9.5M Urea / 2%CHAPS and 2 × Laemmli sample buffer. Several lines of evidence indicate that the observed band is specific for MLL-AF9: (1) it shows the predicted size (approximately 170 kDa), (2) it is detected by both, MLL (N-terminal) and AF9 (C-terminal) antibodies and (3) it is somewhat larger in MonoMac6 (MM6) cell line than in THP1 cells which is expected because MM6 cells harbor a further 3’-terminal breakpoint in MLL. However, due to the high amount of protein needed and because complete dissolution of intermediate protein precipitates can hardly be achieved following this isolation procedure, these samples were not employed for semiquantitative assessment of MLL-AF9 in knockdown experiments. Cell lysates prepared with standard lysis buffers (e.g. Laemmli sample buffer or RIPA buffer) were tested but did not lead to the MLL-AF9 specific band (figure 13 b). To overcome insufficient dissolution of MLL-AF9 protein or protein bound to DNA, sonication of cell lysates was tested. Surprisingly, MLL and AF9 wildtype could be readily detected in control lysates of MCF-7 or HELA cells, while no MLL or MLL-AF9 or AF9 specific band was detectable in THP1 lysates, even on the same Western blots (figure 13 c). This phenomenon may point to a higher tendency of MLL- AF9 to form protein aggregates as compared to wildtype MLL protein.
Results
42
Figure 13: Detection of MLL-AF9 by Western blot (A) Identification of MLL-AF9 protein band in THP1 cells: As expected, MLL-AF9 band is larger (~11 kDa) in MonoMac6 (MM6) cell line as compared to THP1 cells, because of a further 3’-terminal breakpoint in MLL. MLL-AF9 band is stained by MLL (N-terminal) as well as by AF9 (C-terminal) antibody. Protein was extracted from Qiazol organic phase as described in material and methods. (B) MLL-AF9 band is not detectable by neither MLL nor AF9 antibody in THP1 cell lysates prepared with Laemmli sample buffer (or RIPA buffer) but is detectable in THP1 protein extracted from Qiazol organic phase. (C) In lysates prepared with RIPA buffer, neither MLL nor MLL-AF9 or AF9 bands are detectable in THP1 lysates, but wildtype MLL and AF9 bands are detectable in HELA control lysates (HELA cells express MLL and AF9 but do not carry MLL-AF9 translocation). Shown is the image of a single blot from which two lanes were excised for clarity reasons. Similar results were obtained with MCF7 cell line lysates. Band above MLL wildtype band in HELA might be unprocessed MLL protein (430 kDa). MLL and AF9 wildtype bands were assigned due to the expected sizes of the proteins (320 kDa and 63 kDa respectively).
An alternative, new technique for relative quantitation of protein, TaqMan® Protein Assay (Applied Biosystems, Life Technologies, Carlsbad, CA, USA), was additionally tested for MLL- AF9 quantitation (figure 6 in material and methods, page 26). Unfortunately, despite promising quality control results for the assay probes, specific signal of MLL-AF9 was not high enough above background (∆CT>3) to yield reliable quantitative data. Disadvantageous binding sites for
the antibodies (i.e. located far from each other on the large MLL-AF9 protein) or again, non- sufficient dissolution of MLL-AF9 protein, might explain this result.
To nevertheless ensure that MLL-AF9 knockdown is effective on protein level, we analyzed HOXA9 transcript level as a surrogate marker. HOXA9 transcription is raised by MLL-AF9 through direct interaction between MLL-AF9 protein complex and HOXA9 promoter (Erfurth, et al. 2008; Cierpicki, et al. 2010). A MLL-AF9 knockdown specific reduction of HOXA9 transcript to 57 ±8 % residual expression on day 8 was detected by qRT-PCR (figure 11 b). This finding clearly indicates a functionally sufficient reduction of MLL-AF9 protein.