BaxΔ2 Family Alternative Splicing Salvages Bax Microsatellite-Frameshift Mutations

Abstract

Mutation or aberrant splicing can interrupt gene expression. Tumor suppressor Bax is one of the susceptible genes prone to microsatellite frameshifting mutations in coding regions. As a result, tumors exhibiting microsatellite instability (MSI) often present a “Bax-negative” phenotype. We previously reported that some Bax-negative cells in fact contain a functional Bax isoform (BaxΔ2), generated when unique alternative splicing “salvages” the shifted reading frame introduced by a microsatellite mutation. Here we compared Bax alternative splicing profiles in a range of cell lines and primary tumors with and without Bax microsatellite mutations. We found that MSI tumors exhibit a high Bax alternative splicing frequency, especially in exon 2, and produce a family of alternatively spliced isoforms that retain many important Bax functional domains. Surprisingly, these BaxΔ2 family isoforms can rescue Bax from all common microsatellite frameshift mutations. Production of BaxΔ2 requires specific cis mutations, while trans components are not cell-type specific. Furthermore, all BaxΔ2 family isoforms are more potent cell death inducers than the parental Bax without directly targeting mitochondria. These results indicate that the BaxΔ2 family can potentially salvage Bax tumor suppressor expression otherwise lost to mutation.

Introduction

Microsatellite instability (MSI) results from a deficiency of mismatch repair (MMR) proteins.^1-3 Microsatellite tracts are particularly prone to replication error-induced insertions and deletions because of DNA polymerase slippage along these sequences.⁴ When MMR proteins are mutated and unable to repair microsatellite insertions and deletions, the tumor develops MSI.^5,6 Single and double base pair insertions and deletions produce frameshifting mutations, typically culminating in nonfunctional, truncated proteins or nonsense-mediated decay of transcripts due to a premature termination codon (PTC) introduced by the frameshift.⁷ Many genes are prone to developing MSI mutations because they encode microsatellites in their exons, such as the MMR genes MSH3, MSH6, and MLH3⁸; tumor suppressors TGF-β RII, IGFIIR, and PTEN^9,10; DNA damage sensors ATM and RAD50^11,12; and the pro-death caspase-5.¹³

Bax was one of the first affected genes identified in MSI tumors, in which a coding 8 guanine (G8) microsatellite tract in exon 3 was found to be mutated in more than 50% of a set of MSI colon cancer tumors.¹⁴ Such tumors are typically considered “Bax-negative.” Paradoxically, many studies correlating patient prognosis with Bax tumor expression have found Bax statistically predictive for both poor and positive outcomes.^15-20 Low Bax expression in tumors often correlates significantly with a poor patient prognosis or lack of response to particular treatment.^16,18-20 In contrast, some studies have found MSI tumors predict a better patient outcome, particularly when Bax is mutated.^15,21 Whether these seemingly contradictory results have an underlying molecular explanation remains a question.

Alternative splicing contributes significantly to the diversity of proteins in a cell.²² However, in the case of tumor suppressors, alternative splicing in cancer often generates loss-of-function isoforms.^23,24 Both cis mutations or trans changes can result in exon skipping, intron retention, and usage of cryptic 5′ and 3′ splice sites in the final transcript.^25,26 This can produce mRNA transcripts missing coding sequences, causing frameshifts with subsequent premature termination PTC. Transcripts with PTCs are typically degraded by nonsense-mediated decay mechanisms.²⁷ However, aberrant transcripts that are not degraded may translate to truncated proteins, which often lack important functional domains.^28-30

We previously showed that Bax microsatellite mutated cells are not necessarily “Bax-negative,” and instead can produce an alternatively spliced functional Bax isoform, BaxΔ2.³¹ Here we show that a family of BaxΔ2 can also be generated through various combinations of Bax-MSI mutations and alternative splicing in both Bax MSI cell lines and primary tumors. Interestingly, all common Bax-MSI-mediated frameshift mutations can be salvaged by various splicing in Bax exon 2 to produce viable BaxΔ2 family isoforms. This is the first study of Bax functional isoforms generated from Bax-mutated DNA.

Results

MSI cell lines and tumors have a high frequency of alternative splicing at Bax exon 2

Bax microsatellite exon 3 mutations cause a reading frameshift and subsequent premature termination of the Bax transcript when constitutively spliced (Fig. 1A). Previously we identified a unique, functional Bax-MSI isoform, BaxΔ2, in which alternative splicing of Bax exon 2 rescued the reading frameshift introduced by the microsatellite mutation.³¹ Because unique alternative splicing was required for the salvage process, we questioned whether splicing patterns and frequency were Bax-MSI specific. To determine whether there was a relationship between Bax alternative splicing and the Bax MSI status, we screened a panel of 12 cell lines, which represented Bax microsatellite stable (MSS, G8) and microsatellite unstable cell lines (MSI, G7 or G9) (Table 1). Bax transcripts from the cell lines were amplified by RT-PCR from total mRNA using primers corresponding to the 5′ and 3′ UTR of Bax. The amplified pool of Bax cDNA was then cloned into a vector and sequenced. We found a high frequency of alternative splicing activities centered on exons 1 to 3 (Fig. 1B). The overall splicing patterns were similar between Bax-MSS and Bax-MSI cell lines (Fig. 1B). However, it is readily apparent that Bax alternative splicing events are significantly more prevalent in Bax-MSI cell lines than in Bax-MSS cell lines (P < 0.01) (Fig. 1B). Notably, exon 2 alternative splicing represented the greatest percentage of all splicing events in Bax-MSI cell lines and was significantly higher than in Bax-MSS cell lines (P < 0.05) (Fig. 1B). These results indicate that there is a higher frequency of alternative splicing, particularly around exon 2, in Bax MSI cells as compared with Bax-MSS cells.

Figure 1.

(A) Schematic diagram of Bax gene (the sizes of exons and intron are not scaled) and location of Bax microsatellite region. Ex = exon; G = guanidine; PTC = premature termination codon. (B) Alternative splicing events in corresponding to Bax exon and intron positions in Bax microsatellite stable (MSS) versus instable (MSI) cell lines. Total AS events = total number of alternative splicing events. A total of 89 cell line transcripts were analyzed, including 32 Bax-MSS transcripts and 57 Bax-MSI transcripts.

Table 1.

BaxΔ2 Family Transcripts in Cancer Cell Lines.

MSI status	Source	Organ	Bax MSI	Baxα	BaxΔ2	BaxΔ2ω	BaxΔ2G9	BaxΔ2G9ω
Bax-MSS	PC3	Prostate	G8	+	−	−	−	−
	LNCaP	Prostate	G8	+	−	−	−	−
	SW1116	Colon	G8	+	−	−	−	−
	HepG2	Liver	G8	+	−	−	−	−
Bax-MSI	DU145	Prostate	G9	−	−	−	−	−
	104-S	Prostate	G7	−	−	−	−	−
	104-R	Prostate	G7	−	+	−	−	−
	104-IS	Prostate	G7	−	+	−	−	−
	MCF-7	Breast	G7, G8	+	+	−	−	−
	LS174T	Colon	G7	−	+	+	−	−
	LoVo	Colon	G7, G9	−	+	+	+	+
	HCT116	Colon	G7,G8	+	+	−	−	−

Note: Bax-MS = Bax microsatellite sequence detected; + = the specific isoform is detected; − = not detected.

MSI cell lines and tumors express a family of BaxΔ2 isoforms

We previously showed that unusual Bax exon 2 splicing could rescue the shifted reading frame in transcripts that would otherwise prematurely truncate due to a coding microsatellite mutation. As a result, a new Bax isoform, BaxΔ2 was generated through a combination of alternative splicing of exon 2 and a microsatellite mononucleotide deletion from Bax G8 to G7.³¹ To further investigate whether alternative splicing can rescue other Bax microsatellite mutations, such as G8 to G9, we sequenced and analyzed more than 300 transcripts from 12 cell lines and 20 primary tumors. Three additional BaxΔ2 isoforms were identified in addition to parental BaxΔ2 (Fig. 2). All isoforms were generated in a manner similar to BaxΔ2 or in combination with other alternative splicing events (Fig. 2A). For example, BaxΔ2 was generated from exon 2 splicing using an alternative 3′ site within exon 2, while BaxΔ2-G9 resulted from alternative splicing of exon 2 at a typical constitutive intron-exon boundary. Baxω is a previously identified isoform that retains part of intron 5;³² here, we refer to BaxΔ2 in combination with intron 5 retention as BaxΔ2ω (Fig. 2A). Notably, all newly identified isoforms have unique 10 amino acid sequences at the beginning of exon 3 due to the reading frameshift caused by exon 2 alternative splicing (Fig. 2B, underlined). The reading frameshift is restored at the point of microsatellite mutation (G7) or (G9), thus “rescuing” the transcript from premature termination. The presence of BaxΔ2 isoform transcripts in cell lines and primary tumors are summarized in Tables 1 and 2, respectively. It is important to note that all BaxΔ2 isoforms were only detected in Bax microsatellite mutated cells. However, not all the microsatellite mutated cells had BaxΔ2 isoforms. Within the 12 cell lines examined, BaxΔ2 was detected in 5 cell lines, BaxΔ2ω in 2 cell lines, and BaxΔ2-G9 and BaxΔ2ω-G9 in 1 cell line (LoVo) (Table 1). The primary tumor RNA contained both tumor and surrounding normal tissue, and therefore we expected that Baxα isoforms would predominate and that alternative isoforms would be found at a low frequency. Therefore, we sampled 6 to 19 transcripts per tumor RNA sample to increase the likelihood of detecting alternative Bax isoforms. Out of 10 prostate cancer tumor samples, 87/107 Baxα isoforms were detected; 1/107 BaxΔ2 and 1/107 BaxΔ2G9 transcript were detected. Out of 10 colon cancer tumor samples, 99/120 Baxα isoforms were detected and 1/120 BaxΔ2 transcripts (Table 2). Additional alternatively splicing events were detected in both prostate and colon tumors, mostly concentrated around the first 3 exons (Table 2), which is consistent with cell line results (Fig. 1B). These data indicate that BaxΔ2 family members are Bax microsatellite mutation-specific isoforms; however, transcriptional production of BaxΔ2 isoform is tightly regulated.

Figure 2.

BaxΔ2 family isoforms. (A) Schematic comparison of BaxΔ2 isoforms. BH = Bcl-2 homology domain; TM = transmembrane domain. Hatched box, sequence reading frameshit. (B) Amino acid sequence alignment of BaxΔ2 isoforms with corresponding exons labeled on the top.

Table 2.

BaxΔ2 Family Transcripts in Primary Human Tumors.

Tumors	Origin	Grade	Bax MSIa	Baxα	BaxΔ2	BaxΔ2G9	BaxΔ2 splicingb
1	Colon	II	G7, G8	16	1	—	—
2	Colon	II	G8	6	—	—	—
3	Colon	II	G8	7	—	—	—
4	Colon	III	G8	14	—	—	—
5	Colon	III	G8	11	—	—	—
6	Colon	III	G8	7	—	—	—
7	Colon	III	G8, G9	7	—	—	—
8	Colon	IV	G8	14	—	—	—
9	Colon	IV	G8	8	—	—	—
10	Colon	IV	G8	9	—	—	—
11	Prostate	III	G7, G8	10	1	—	1
12	Prostate	III	G8	12	—	—	—
13	Prostate	III	G7, G8	11	—	—	1
14	Prostate	III	G8	10	—	—	1
15	Prostate	III	G8	6	—	—	—
16	Prostate	III	G8, G9	2	—	4	—
17	Prostate	III	G8	10	—	—	—
18	Prostate	IV	G8	8	—	—	—
19	Prostate	IV	G7, G8	12c	—	—	—
20	Prostate	IV	G8	7	—	—	—

^aBax MS summarizes Bax microsatellite sequences detected in mRNA transcripts for each tumor sample.

^bThese sequences have Bax G8, but with BaxΔ2 alternative splicing.

^cThe transcripts include 2 sequences which have Bax G7, but Baxα constitutive splicing.

Bax exon 2 splicing variants can salvage all Bax microsatellite frameshift mutations

The common microsatellite mutation in Bax is a mononucleotide deletion or insertion in Bax exon 3 from G8 to G7, or from G8 to G9, leading to loss of Bax expression. By analyzing transcripts from both cell lines and primary tumors, we realized that both Bax G7 and G9 mutants could be salvaged as new isoforms but by different exon 2 splicing mechanisms. For a Bax G7 mutation, if exon 2 is spliced using an alternative 3′ acceptor site within exon 2, the reading frame is restored at the G7 microsatellite mutation site as the BaxΔ2 isoform. For a Bax G9 mutation, if exon 2 is spliced using a constitutive splicing mechanism, the reading frame is restored at the G9 microsatellite mutation site as the BaxΔ2G9 isoform (Fig. 3A and B). Of note, splicing exon 2 by either a constitutive or alternative splicing mechanism by itself causes a reading frame shift due to the non-multiple of 3 nucleotides in exon 2. For a Bax microsatellite multi-nucleotide deletion from G8 to G6, or insertion from G8 to G10, the result will be same as the G9 or G7 mutations, respectively, due to the multiple of 3 nucleotides. This means that all Bax shifted reading frames due to MSI deletion or insertion can be rescued by Bax exon 2 splicing variants to produce a viable Bax isoforms (Fig. 3A).

Figure 3.

Generation of viable transcripts by different exon 2 splicing mechanisms in Bax microsatellite mutants. (A) Bax with G8 wt microsatellite produces Baxα by a constitutive splicing; Bax with G7 mutation could produce BaxΔ2 by alternative splicing using alternative 3′ splicing site (Ex2 alt 3′ sp); Bax with G9 mutation could produce BaxΔ2G9 by alternative splicing whole exon 2 (Ex2 alt sp). PTC = premature termination codon. (B) Alternative splicing hypothesis. Numbers in italics indicate predicted splice site strength. (C) BaxG7 minigene (MG) wild type (wt) or exon 2 alternative 3′ splicing acceptor site mutant (1643A>T) were transfected into HCT116 cells. The expressions of green fluorescent proteins and BaxΔ2 from immunostaining with anti-BaxΔ2 antibody were observed under a fluorescence microscope. (D) Immunoblotting the transfected cells from (C) with anti-BaxΔ2 antibody.

Splicing Bax exon 2 to generate BaxΔ2 appeared to require the use of an unusual alternative 3′ acceptor site within exon 2. To examine this splicing hypothesis, we analyzed all potential splicing boundaries for Bax using Alternative Splice Site Predictor (ASSP) (http://www.es.embnet.org/~mwang/assp.html).³³ ASSP predicted an unusual alternative 3′ acceptor site located in exon 2 with a relatively high score (Fig. 3B). To further confirm this unusual 3′ acceptor site was indeed used to generate BaxΔ2, we generated a single point mutant (1643A>T) for the predicted acceptor site using site-directed mutagenesis. The 1643A>T mutant was harbored in a BaxG7 minigene, in which a bax genomic construct from exon 1 to the 5′ end of exon 4 was cloned in-frame with a 5′ GFP sequence.³¹ The sequence translated in-frame for GFP expression with BaxΔ2 splicing, but with constitutive splicing, the GFP sequence would be translated out-of-frame. Figure 3C shows that GFP-expressing cells could be detected in both Bax G7 wt and 1643A>T mutant minigene transfections, although the GFP expression in 1643A>T mutant is weaker than that from wt (Fig. 3C). However, expression of BaxΔ2-GFP fusion protein could only be detected in Bax G7 wt, and not 1643A>T mutants, by immunoblotting with anti-BaxΔ2 antibody (Fig. 3D). This indicates that the GFP expression from the 1643A>T mutant resulted from a non-BaxΔ2 splicing event since it did not react with anti-BaxΔ2 antibody (Fig. 3D). Furthermore, this result indicates that the cryptic alternative 3′ acceptor site in the exon 2 coding region is critical for producing the BaxΔ2 isoform from Bax G7 mutated genes.

BaxΔ2 family products requires a specific cis mutation but not cell-line specific trans factors

Generation of the BaxΔ2 family has 2 requirements: (a) a frameshift mutation in the Bax microsatellite region as a cis element and (b) a microenvironment for alternative splicing of exon 2. It is conceivable that the production of BaxΔ2 is a cis and/or trans-regulated phenomenon. To examine these requirements, we cloned and sequenced Bax genomic DNA from several cell lines and sequenced the 5′ UTR through exon 4 and exon/intron boundaries. At least 4 sequences were analyzed for each cell line and compared to the Bax genomic sequence (GenBank AY217036.1). The predicted Bax microsatellite mutation and known polymorphisms were found, but no mutations were found at splice site boundaries. Furthermore, many cell lines produce BaxΔ2 splicing, but no point mutations were shared across cell lines. Taken together, it is likely that this Bax microsatellite cis mutation is the only cis element required for the BaxΔ2 exon 2 alternative splicing events.

To test the trans hypothesis, we screened a panel of cells differing in their Bax status, including BaxG7/G7 (LS174T colon) or G8/G8 (PC3 prostate) cancer cells, immortalized non-cancer cells (293 human embryonic kidney fibroblast), and murine fibroblasts (NIH3T3). Cells were transfected with the BaxG7 minigene construct, and the production of BaxΔ2-GFP fusion proteins was observed under a fluorescence microscope and analyzed by immunoblotting with anti-BaxΔ2 specific antibody. As shown in Figure 4A, all cells were able to splice the BaxG7 minigene to produce BaxΔ2-GPF fusion proteins regardless of the host cell Bax microsatellite and transformation status. This result suggests that the splicing trans components are not cell-line specific.

Figure 4.

The splicing efficiency of BaxΔ2 minigenes is dependent on cellular context. (A) The BaxΔ2 minigene productivity is analyzed by immunoblotting of different cells indicated transfected with BaxG7 minigene or GFP control with anti-BaxΔ2 antibody. (B) Schematic illustration of Bax minigenes constructs for BaxG7, BaxG8, and BaxG9. (C) Minigene assay. PC3 and LoVo cells were transfected with the minigene constructs listed in (B), and expression of GFP fusion proteins were measured by a fluorescence plate reader and quantitated using the BaxG8 minigene product for internal normalization.

We next examined whether a single cell type was capable of processing all 3 minigenes, BaxG7, G8, and G9 (Fig. 4B). PC3 (G8/G8) and LoVo (G7/G9) cells were transfected with 3 different minigenes individually (Fig. 4B). The minigene splicing products were observed under a fluorescence microscope and quantitated using normalized GFP. As shown in Figure 4C, all transfected cells were capable of the splicing necessary to produce Bax-GFP fusion products. However, the efficiency of the production varied. PC3 cells favored G8 constitutive splicing as compared to LoVo cells (Fig. 4C), while BaxΔ2 G7 splicing was significantly higher in LoVo cells than PC3 cells (Fig. 4C). These data suggest that although the trans requirements are common to a variety of cell lines, BaxΔ2 isoform expression is controlled by the cellular context.

BaxΔ2 family isoforms are potent cell death inducers but do not target mitochondria

Like the parental Baxα, BaxΔ2 is able to heterodimerize with Bcl-2 or homodimerize with Baxα.³⁴ Unlike Baxα, which stays in the cytosol and targets mitochondria only on a death stimulus, BaxΔ2 protein forms aggregates in cytosol but does not directly target mitochondria.³¹ As all BaxΔ2 family isoforms share the same BH domains but have distinct N- and/or C-termini, we expected that they might have a similar pro-death function but different cellular localizations. To examine this, Bax^−/− MEFs cells were transfected with GFP-tagged BaxΔ2, BaxΔ2ω, BaxΔ2(G9), and BaxΔ2(G9)ω isoforms, using GFP alone as a control. The cellular localizations of BaxΔ2 family proteins were determined by green fluorescence co-stained with a mitochondrial marker (MitoTracker, red). Figure 5A shows that all BaxΔ2 protein isoforms were localized in the cytosol as aggregates and not co-localized with mitochondria. Importantly, all BaxΔ2 isoforms were capable to induce cell death in the Bax-negative prostate cancer DU145 cells and colon cancer LS174T cells (Fig. 5B), and all at a higher efficiency than parental Baxα (Fig. 5B). These data suggest that BaxΔ2 family isoforms are a group of potent pro-death molecules, which can potentiate cell death in Bax-negative cancer cells without directly targeting mitochondria.

Figure 5.

BaxΔ2 family proteins induce cell death but not target mitochondria. (A) The GFP-tagged BaxΔ2 family construct as indicated (GFP alone used as a control) were transfected into Bax^−/− MEFs cells for 16 hours, then the cells were then stained with MitoTracker for 30 minutes before fixed with 4% paraformaldehyde. The cells were imaged under a fluorescence microscope. (B) Bax-negative prostate cancer DU145 and colon LS147T cells were transfected with GFP-tagged BaxΔ2 family isoform constructs as indicated (Baxα as a control) for 16 hours. Percentage of cell death was calculated using floating green cells versus total green transfected cells.

Discussion

In this study, we identified a unique family of Bax isoforms that were found only in Bax MSI mutated tumors. Either frameshift mutations or alternative splicing could result in a Bax-negative phenotype. However, with various combinations of Bax microsatellite mutations and high activity of alternative splicing events in tumor, a family of potential functional Bax isoforms (BaxΔ2) could be generated (Fig. 2). BaxΔ2 transcripts were only detected in Bax microsatellite mutated cancer cell lines and primary tumors (Tables 1 and 2). The site-mutagenesis data show that the 3′ alternative splicing site in exon 2 is critical for generation of BaxΔ2. The BaxΔ2 splicing trans requirements do not seem cell line specific but are tightly controlled by cellular context (Fig. 4). Interestingly, all Bax microsatellite mutation, G6, G7, G9, and higher, can be salvaged by variations of BaxΔ2 splicing events. Furthermore, all BaxΔ2 family isoforms are more potent cell death inducers than the parental Bax without directly targeting mitochondria (Fig. 5). These data suggest that some “Bax-negative” cells actually express viable Bax alternatives, which should be taken into the consideration during analysis of “Bax-negative” data.

The initial prototypical Bax N-term 11 amino acid sequence from exon 1 is shared among all 4 Bax isoforms we have identified (Fig. 2B). This sequence is followed by a frameshifted region generated by the alternatively splicing that is unique to each isoform and diverges from the constitutive Baxα sequence (Fig. 2B, underlined). At the microsatellite site, where the reading frame shifts according to the mutation status, all sequences realign to prototypical Baxα for the remainder of the coding sequence. As a result, all isoforms retain Bax functional BH domains. We have shown that, like Baxα, BaxΔ2 is a potent inducer of cell death. However, BaxΔ2 does not seem to target mitochondria but forms aggregates in the cytosol.³¹ It would be interesting to investigate whether the unique oligopeptides in the BaxΔ2 family play an important role in their cellular targeting.

Whereas N-terminus alternative splicing tends to impact Bax targeting, C-terminus alternative splicing appears to impact mitochondrial membrane insertion.³⁵ Lack of the transmembrane domain, as in the Baxβ isoform, causes constitutive targeting and insertion into the mitochondria.³⁶ Bax lacking a constitutive C-terminus is rapidly proteasomally degraded, even without an apoptotic stimuli.³⁶ In the study here, several Bax-MSI isoforms included partial intron 5 retention, known as Baxω.³² Little is known about the ability of Baxω to translocate and insert into the mitochondria, although its initial characterization showed that it could potentiate cell death.³² Interestingly, our data show that all BaxΔ2 family proteins have a similar cellular localization pattern (Fig. 5A), indicating that the non-mitochondrial targeting may be independent of their N- or C-terminal variations.

It has been shown that Bax may have several functional isoforms generated by alternative splicing or degradation of parental Baxα. Some Bax isoforms play an important role not only in cellular apoptotic dynamic balance but also in the prognosis and outcome of the tumor development and treatment. For example, Baxψ, an isoform identified in glioblastoma multiforme tumors, was correlated with longer patient survival relative to tumors expressing Baxα alone.³⁷ We have previously shown that BaxΔ2 positive cancer cells are more selectively sensitive to subgroup of chemotherapeutic drugs, such as Adriamycin.³¹ With the advantage that BaxΔ2 is specific in Bax mutated cells, this suggests that cancer patients with BaxΔ2 expressing tumors may have selective response to certain chemotherapeutic drugs.

Materials and Methods

Materials and cell lines

The BaxΔ2 monoclonal antibody was generated using the short unique peptide (GFHPGSSRAN) by Precision Antibody (Columbia, MD). Antibody against β-actin was purchased from Millipore (Billerica, MA). The transfection reagent Lipofectamine 2000 was purchased from Invitrogen (Grand Island, NY). Prostate cancer cell lines DU145 and PC3 and colon cancer cell lines LS174T, LoVo, HCT116, and SW1116 were originally obtained from the American Type Culture Collection (Rockville, MD). LNCaP sublines 104-S, 104-R, and 104-IS were generous gifts from Dr. John Kokontis at the University of Chicago. Androgen-dependent 104-S cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; Mediatech, Inc., Herndon, VA) supplemented with 10% fetal bovine serum (FBS) and 1 nM synthetic androgen R1881. PC-3, DU145, 104-R, and 104-IS cells were cultured in DMEM supplemented with 10% charcoal stripped FBS. LoVo cells were cultured in Ham’s/F12 supplemented with 10% FBS. Other cells were cultured in DMEM supplemented with 10% FBS. All cultures were maintained at 37°C and 5% CO₂.

Primary tumor RNA

Purified primary tumor RNA was purchased from Bioserve Biotechnologies Ltd. (Beltsville, MD). Bioserve collects and processes their samples with full adherence to proper informed consent, as well as strict institutional review board and Health Insurance Portability and Accountability Act compliance. There is no link to personal identification.

Cloning and DNA sequencing

Total RNA from different cell lines was purified using PureLink RNA Mini Kit from Invitrogen, according to the manufacturer’s instructions. The RNA quantity and concentration was measured by NanoDrop Spectrophotometer from Alpha Innotech. cDNA transcripts of various cell lines and primary tumors were reverse transcribed using total RNA by ThermoScript RT-PCR System. The PCR product was amplified with forward primer 5′-gctctagagagcggcggtgatggacgggt-3′ and reverse primer 5′-ggaattccagctgggggcctcagcccat-3′ and inserted into PCR-Script vectors. Random clones were selected for sequencing and analyzed using Geneious software (Biomatters Ltd, Auckland, New Zealand).

Bax DNA sequencing and Bax-GFP minigene constructs

Bax genomic DNA was extracted using the AquaPure Genomic DNA Kit from Bio-Rad (Hercules, CA). Briefly, fresh cells collected by gently scraping and pelleting, or frozen pellets, were lysed in a detergent buffer with RNase. DNA was isolated from solution using a salt precipitation followed by ethanol precipitation, then rehydrated. Isolated DNA was amplified in a PCR reaction using primers located in the 5′ UTR of Bax containing a Bgl II restriction site (5′-ctcagatctCCACGTGAAGGACGCACGTTCAGC-3′) and the 3′ end of Bax exon 4 followed by an EcoR1 restriction site (5′-aacgaattcTGGCAAAGTAGAAAAGGGCGACAACC-3′). Amplified product was gel-extracted and ligated into pEGFP-N1 mammalian expression vector. At least 4 clones from each cell line were sequenced. Bax-GFP minigene constructs were selected from these clones. Primers were designed so that Bax genomic sequences would produce transcripts in-frame with GFP on the appropriate microsatellite status and splicing event combination.

Site-directed mutagenesis of Bax minigene

A single point mutation was performed using QuikChange II Site-Direct Mutagenesis Kit (Stratagene) according to the manufacturer’s instruction. The potential BaxΔ2 acceptor site was mutated from A to T at Bax nucleotide position 1643 (1643A>T, in exon 2, Gen Bank AY217036.1) using the Bax G7 minigene construct as a template with forward primer 5′-gcccttttgcttctggggtgagtttgaggtc-3′ and a reverse primer 5′-gacctcaaactcaccccagaagcaaaagggc-3′. Colonies were selected and the desired mutation was validated by DNA sequencing.

Transient transfection and Immunoblotting

Transient transfection was performed using Lipofectamine 2000 transfection reagent according to manufacturer’s instruction. Cells were allowed to grow to 70% to 80% confluences in 6-well plates before transfection. After 24 hours posttransfection, the expression of the minigene GFP fusion product was examined under a fluorescence microscope. The cells were then harvested and lysed in NP-40 lysis buffer with a cocktail of protease inhibitors. Protein concentration was determined using the Bio-Rad protein assay kit, and samples containing an equal amount of protein were separated on SDS-polyacrylamide gel and transferred to a PVDF membrane. Following transfer, the membrane was incubated in 3% non-fat milk for one hour, then anti-BaxΔ2 or anti-actin antibody at 4°C overnight. The expression of these proteins was visualized by Pierce’s ECL chemiluminescence detection reagent.

Cellular localization and cell death assay

Bax^−/− MEF cells were transfected with BaxΔ2-GFP, BaxΔ2ω-GFP, BaxΔ2(G9)-GFP, BaxΔ2(G9)ω-GFP for 16 hours. The cells were then incubated with MitoTracker for 30 minutes at 37°C and fixed with 4% paraformaldehyde. The cellular localization of GFP-tagged proteins and mitochondria were imaged under a fluorescence microscopy. For cell death, BaxΔ2 negative prostate cancer DU145 and colon cancer LS174T cells were transfected with BaxΔ2 family constructs as above for 16 hours. Only GFP-positive cells were counted for the death assay. The percent cell death was calculated as green floating cells versus total GFP positive cells.

Analytical tools and statistics

Nucleotide sequences were analyzed for alignment to known Bax sequences using Geneious Bioinformatics Software (Biomatters). Splice site sequence predictions and their relative strength were made using Alternative Splice Site Predictor (ASSP) (http://www.es.embnet.org/~mwang/assp.html). Protein secondary structure predictions were made using PROFsec within PredictProtein (https://www.predictprotein.org/). P-values for splice site comparisons were made using a chi-squared test.