Comparison of Splicing Factor 3b Inhibitors in Human Cells

Pre-mRNA splicing is mediated by the spliceosome, a macromolecular machinery comprised of the U1, U2, U4/6, and U5 small nuclear ribonucleoproteins.1 The U2 small nuclear ribonucleoproteins contain splicing factor 3b (SF3b) complex that is critical for the accurate selection of 3′-splice sites.2, 3 In 2007, the SF3b complex was found to be directly inhibited by FR901464,4 a natural product isolated from the culture broth of a bacterium, Pseudomonas sp. No.2663 (Scheme 1).5–7 Since then, some of the readily available FR901464 analogues— spliceostatin A8 and meayamycin9 among others10–12—have been widely used to elucidate the mechanisms of pre-mRNA splicing and SF3b-related diseases.2, 4, 13–25 In 2011, herboxidiene (GEX1A) was also found to be an inhibitor of SF3b,26–31 and our group reported meayamycin B as a new FR901464 analogue (Scheme 1).20, 32, 33 Meayamycin B was more synthetically accessible than meayamycin and the most potent antiproliferative agent among the known FR901464 analogues. However, it has not yet been studied as a splicing inhibitor. Meanwhile, despite their widespread use, FR901464 analogues have not been directly compared within the same splicing system. With the addition of new SF3b inhibitors such as meayamycin B and herboxidiene, a direct comparison of these two compounds, FR901464, spliceostatin A, and meayamycin in a splicing assay would enable rational selection of an SF3b inhibitor for specific experimental designs.

Scheme 1. Structures of FR901464, its analogues, and herboxidiene/GEX1A.

It should be noted that there appears to be confusion in the literature about the identity of FR901464 analogues. For example, in a recent review article, FR901464, spliceostatin A, and meayamycin were listed as if they were unrelated compounds.34 Both spliceostatin A and meayamycin are synthetic analogues of FR9014648, 9 and should be treated as such22, 35, 36 and not as natural products.18

In this study, we used a constitutive splicing reporter developed by the Moore group (Figure 1 A).22 Briefly, the construct consists of an exon 6–intron 6–exon 7 cassette from the human triose phosphate isomerase (TPI) minigene upstream of a firefly luciferase open reading frame. To ensure that the expressed luciferase activity is inversely correlated with the splicing efficiency of the minigene, the constitutive in-frame stop codon in the intron was removed and a G was inserted in TPI exon 7 so that the downstream luciferase gene is in-frame when the intron is retained. A negative-control construct contains the same minigene but with a mutated 5′-splice site; therefore, it constantly expresses high levels of luciferase. These constructs were stably transfected into human embryonic kidney 293 (HEK293) cells, and the resulting cell lines were called HEK293-II and HEK293-III, respectively.22 The Z scores of a splicing assay using HEK293-II paired with HEK293-III were reported.22 However, the Z score of an assay using the HEK293-II cell line by itself and a potent SF3b inhibitor as a positive control was not reported.

Figure 1. Optimization of a cell-based splicing inhibition assay in a 96-well format. A) TPI minigene construct for screening splicing inhibitor. B) Time-dependent increase of luminescence intensity in HEK293-II cells treated with meayamycin B (10 nM) or DMSO (0.5 % v/v) for various lengths of time. C) Determination of the optimal cell density. HEK293-II cells were seeded at various densities in triplicates and treated with meayamycin B (10 nM) for 16 h. Coefficient of variation values were obtained using GraphPad Prism. D) Evaluation of the Z score of the cell-based assay. Half of a 96-well plate confluent with the HEK293-II cells (1.6×10⁴ cells per well) were treated with meayamycin B (10 nM), whereas the other half received DMSO (0.5 % v/v). The treatment lasted for 16 h before luminescence was assayed to calculate the Z score. Dotted lines for each set of data indicate 3× standard deviations. Data represent three separate experiments.

To configure the cellular assay for compound screening using a small-molecule control, we set out to further optimize and validate the HEK293-II-based splicing assay to meet the accepted high-throughput criteria in a 96-well format.37 To this end, meayamycin B was chosen as a pre-mRNA splicing inhibitor because it is the most potent among known FR901464 analogues against various human cancer cell lines and is structurally similar to the known splicing modulators meayamycin, spliceostatin A (see the Supporting Information for the preparation and purity of spliceostatin A), and FR901464.9, 19, 32 To find a convenient end-point of the splicing assay, the inhibitory activity of meayamycin B was monitored over time. This study revealed that meayamycin B caused a linear increase in luciferase activity that plateaued around 30 h. Sixteen hours was the shortest period of exposure that was in the linear portion of the curve, and showed a reasonably large signal-to-background ratio (Figure 1 B); hence, 16 h was chosen as a time point for further assay development. We then determined the optimal cell density at the 16 h time point. In the presence of meayamycin B (10 nM, ED₉₅), luciferase activity showed a linear correlation with cell number over a wide range of cell densities. We chose 1.6×10⁴ cells per well as judged by the lowest standard deviation of luciferase expression (Figure 1 C, left) and a calculated coefficient of variation of less than 7 % (Figure 1 C, right). This optimal cell density was in agreement with previously reported data in a 384-well format.22

Meayamycin B is also an antiproliferative agent; therefore, we assessed the growth inhibition of the HEK293-II cells by meayamycin B to ensure that the luciferase data would not be skewed by cell growth inhibition, and therefore would truly reflect the splicing inhibition. To this end, we employed a total protein assay using bicinchoninic acid (BCA) and a cell proliferation assay using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) after a 16 h incubation of HEK293-II cells with meayamycin B (10 nM). The BCA assay was used to assess cell proliferation because in our experiments, cell number correlated with total protein concentrations (Figure S1 in the Supporting Information). Meayamycin B (10 nM), induced a 15 % inhibition of growth in HEK293-II cells (data not shown). Hence, further luciferase assays were normalized with total protein concentrations.

With the optimized conditions in hand, we proceeded to determine the Z score of the splicing inhibition assay. The HEK293-II cells (1.6×10⁴ cells per well) in 96 wells were treated with meayamycin B (10 nM) or DMSO for 16 h. As shown in Figure 1 D, a half-plate minimum/half-plate maximum signal design produced a Z score of 0.56±0.04 (n=3), validating high-throughput compatibility.37, 38 It is worth noting that we did not use the HEK293-III (negative control) cell line in our assay, but a dose-dependence study showed that a 16 h exposure elicited a negligible increase of luciferase activity, indicating the absence of other influence such as transcription and translation (Figure S2).

With this validated method, we investigated the potencies of FR901464, spliceostatin A, meayamycin, meayamycin B, and synthetic39 herboxidiene. As shown in Figure 2 A, 16 h treatments with these compounds resulted in a dose-dependent splicing inhibition with ED₅₀ values of 3.5±0.75, 4.7±1.7, 1.6±0.4, 0.23±0.12, and 257±11.1 nM, respectively. The splicing inhibition was further verified by a semiquantitative RT-PCR experiment that showed the relative amounts of both spliced and unspliced TPI mRNA (Figure 2 B). In both assays, meayamycin B was found to be the most potent splicing inhibitor among the compounds tested.

Figure 2. Constitutive splicing inhibition of the triose phosphate isomerase (TPI) minigene by SF3b inhibitors in HEK293-II cells. A) Splicing inhibition of TPI minigene by SF3b inhibitors for 16 h as expressed with luciferase activity. B) Inhibition of splicing of TPI minigene by SF3b inhibitors after 16 h as examined by semiquantitative RT-PCR. C) Time-dependence of splicing inhibition by SF3b inhibitors. The relative luminescence unit (RLU) is defined as the ratio of luminescence intensities from compound-treated and DMSO-treated normalized by the corresponding cell density. Data points represent average values of triplicate experiments±standard deviation. Each graph represents the results from three separate experiments. D) Correlation between splicing inhibition and antiproliferative activity of meayamycin B (left panel) and herboxidiene (right panel). Data points represent average values of triplicate experiments±standard deviation. Each graph represents the results from three separate experiments.

We next studied the time-dependence of splicing inhibition using the ED₉₅ concentrations calculated from the 16 h assays. Because these compounds are potent antiproliferative agents, cell viability (see Figure S3) under each treatment was measured and factored into the evaluation of splicing inhibition over time. As shown in Figure 2 C, the luciferase activity (i.e., splicing inhibitory activity) increased to a similar extent (relative luminescence unit, RLU=five to tenfold increase over vehicle control) for the first 24–48 h in the presence of FR901464, spliceostatin A, and herboxidiene before it dropped back to near baseline in 72 h. The magnitude of luciferase activity was much larger in the presence of meayamycin and meayamycin B (RLU=20–30-fold increase over vehicle increase) and remained as such for 72 h. We postulated that the prolonged effect of the meayamycins was partly due to their superior stability in cell culture medium.9, 19

Thus far, we have discussed FR901464, its analogues, and herboxidiene as SF3b inhibitors. However, these compounds are also antiproliferative agents.5–7, 11, 19, 20, 26, 30, 31, 40 Surprisingly, the relationship between these two activities has not been studied using the Pearson correlation. To determine how well splicing inhibition and cell growth inhibition caused by meayamycin B and herboxidiene correlate with each other, HEK293-II cells were exposed to meayamycin B and herboxidiene for 16 h, and the luciferase activity and cell proliferation were measured. The results are shown in Figure 2 D. The Pearson correlation coefficients (r) for meayamycin B and herboxidiene were determined to be −0.944±0.01 (p<0.0001) and −0.603±0.17 (p<0.0008), respectively. Therefore, the splicing inhibition and antiproliferative activities for both compounds are correlated in these transformed HEK293 cells, with a stronger correlation for meayamycin B.

Many of the natural products that inhibit pre-mRNA splicing (FR901464,4 pladienolides,41 herboxidiene,26 and isoginkgetin22) are endowed with either an epoxide or a Michael acceptor. However, a complete understanding of these important electrophilic functional groups remains elusive.42 Thus, we screened a panel of epoxides and Michael acceptors (Scheme 2) in the HEK293-II-based assay; none of these compounds inhibited splicing (1, 10, and 50 μM) as determined by the luciferase assays and RT-PCR of the treated HEK293-II cells (Figure S4). These results indicate that an electrophilic functional group by itself is not sufficient to inhibit pre-mRNA splicing.

Scheme 2. Epoxides and Michael acceptors. Chiral compounds were tested as racemates.

In conclusion, we implemented a previously reported cell-based splicing assay system in a 96-well format using meayamycin B as a positive control. The validated assay permitted a direct, quantitative, and comparative evaluation of herboxidiene and the currently available analogues of FR901464 for inhibition of constitutive splicing activity. The assay delivered graded responses and ED₅₀ values for all agents; meayamycin B was the most potent, while meayamycin, spliceostatin A, and FR901464 were equipotent, and herboxidiene was less potent. Meayamycins inhibited pre-mRNA splicing for longer periods than spliceostatin A, FR901464, and herboxidiene. We found that simple electrophilic functional groups did not suffice for splicing inhibition in HEK293-II cells. Antiproliferative activity and splicing inhibition by meayamycin B were strongly correlated. Our data validate meayamycin B as the most potent spliceosome inhibitor to date whose cytotoxic effects might be a direct result of splicing inhibition. We posit that meayamycin B is a useful chemical probe with which to further validate the spliceosome as an anticancer target, and a starting point for development of novel antineoplastic agents.

Experimental Section

Complete experimental details for the preparation of test compounds, optimization of the cell-based assay, luciferase assay, cell proliferation assay, RT-PCR and correlation assay experiments are provided in the Supporting Information.

Acknowledgements

This work is in part supported by the US National Cancer Institute (R01 CA 120792 and 3P50 CA 097190 0751). We thank Sami Osman in our group for providing meayamycin B, Professor Melissa Moore (University of Massachusetts) for providing the HEK293 cell lines.