Cardioprotective SNPs in SLC28A3 and lncRNA SLC28A3-AS1 result in transcriptional changes and alternative splicing to reduce doxorubicin cytotoxicity

Abstract

Cardiotoxicity is a therapeutic challenge for anthracycline-based treatments for solid tumors and leukemia. Genome-wide association studies have revealed that single nucleotide polymorphisms in the SLC28A3 locus (encoding Concentrative Nucleoside Transporter 3, CNT3) are significantly associated with reduced doxorubicin-induced cardiotoxicity. However, the mechanistic understanding of the functional effects of these genomic variants is lacking. We designed studies focused on clinically associated SNPs within SLC28A3 using minigenes, site-directed mutagenesis, splicing assays, modulation of SLC28A3 and its antisense long noncoding RNA (lncRNA, AS1), and doxorubicin transport and cytotoxicity measurements to gain more insight. We demonstrated that the cardioprotective synonymous SNP rs7853758 in the Ex14 coding region of SLC28A3 and the variant rs11140490 in Ex1 of its antisense lncRNA (SLC28A3-AS1) have functional consequences in regulating CNT3 transcript and protein expression using alterations in RNA levels and alternative splicing. Additionally, the deep intronic region of Int13, which harbors the SNP rs7030019, is critical for the splicing of CNT3 precursor mRNA at Ex13–14. Furthermore, we identified alternatively spliced variants of the AS1 lncRNA that differentially regulate CNT3 gene expression, doxorubicin transport, and cytotoxicity. Together, these findings suggest that antisense and splicing mechanisms may be exploited to modulate CNT3 function to reduce doxorubicin cytotoxicity, enabling the development of predictive biomarkers and chemotherapeutic management of anthracycline toxicities.

Graphical Abstract

Graphical Abstract.

Introduction

Anthracyclines, which are widely used to treat solid tumors, including those in breast cancer patients and leukemia and lymphoma patients, cause life-threatening cardiotoxicity, in addition to toxicity to other organs, such as liver, kidneys, adipose tissue and brain [1–10]. Anthracycline-induced cardiotoxicity can manifest either as asymptomatic echocardiographic abnormalities or can progress to clinical heart failure requiring emergency interventions [11, 12]. In ~ 9% of patients, it most often manifests as congestive heart failure with reduced ventricular function, in addition to other general symptoms including myocarditis, vascular toxicity, arterial hypertension, cardiac arrhythmias, and pulmonary hypertension, with many patients developing a spectrum of dysfunctions through various mechanisms [13–17]. Typically, one or more symptoms are reported in more than 50% of anthracycline-treated patients, limiting the overall antitumoral benefits they derive from anthracycline-based chemotherapy [18]. Younger age at diagnosis, female sex, concomitant radiation therapy, comorbidities such as diabetes and hypertension, and concurrent use of other anticancer drugs (e.g. trastuzumab, paclitaxel) are also recognized as additional risk factors that trigger anthracycline-induced cardiotoxicity [10, 19, 20].

Anthracyclines such as daunorubicin, doxorubicin, and mitoxantrone, which generally cause dose-dependent cardiotoxicity, inhibit topoisomerase IIβ and intercalate into DNA, resulting in DNA damage, mitochondrial injury, and reactive oxygen species generation, all of which lead to programmed cell death [10, 21]. Conceptually, the therapeutic effect of such DNA-damaging agents on cancer cells, unintentionally results in adverse effects in healthy cell types, such as cardiomyocytes, resulting in cardiomyopathy, congestive heart failure, and accelerated aging, among others. Notably, despite an established dose–dependent association with cardiotoxicity, there is extensive interpatient variability in the risk of cardiac dysfunction for any given anthracycline dose, with clinical variables alone failing to adequately predict anthracycline-induced cardiotoxicities [22]. Thus, there is an unmet clinical need to develop cardioprotective therapies that can be used in conjunction with anthracycline therapies; however, this field has met with challenges [23–25].

Intriguingly, pharmacogenomic studies and genome-wide approaches in large patient cohorts in the past decade have begun to illuminate the role of single nucleotide polymorphisms (SNPs) associated with anthracycline-induced cardiac dysfunction [11, 18, 26–37]. These SNPs have been identified in several anthracycline biotransformation genes such as those encoding glutathione transferases, drug transporters, and DNA repair enzymes [38]. Among all the predicted risk genes with genetic polymorphisms, an evaluation of 2977 SNPs in 220 key drug biotransformation genes in a cohort of 156 anthracycline-treated children, with added replication in a second cohort of 188 children, revealed a significant association of a synonymous coding variant, rs7853758 (L461L), within the SLC28A3 gene (solute carrier family 28 member 3) with anthracycline-related cardiomyopathy (OR: 0.35; P = 1.8 × 10⁻⁵ for all cohorts combined) [11, 39, 40].

SLC28A3 encodes for a membrane-bound CNT3 transporter protein, which facilitates the unidirectional passage of nucleosides and nucleoside analogs across biological membranes, with implications for the uptake of anthracyclines such as doxorubicin [41, 42]. CNT3 belongs to the concentrative nucleoside transporter (CNT) family, which encompasses three subfamilies of sodium-dependent transporters designated CNT1, CNT2, and CNT3 (encoded by SLC28A1, SLC28A2, and SLC28A3, respectively) [43–45]. Earlier genetic analysis and functional characterization of CNT3 suggest that this transporter does not tolerate the nonsynonymous changes important for human fitness [46]. To date, several independent pharmacogenomic studies have recapitulated the synonymous genomic variant rs7853758 (G > A, L461L) and the intronic variant rs885004 in SLC28A3 as statistically associated with a lower incidence of anthracycline-induced cardiotoxicity. More recently, these SNPs were suggested to be in linkage disequilibrium with SLC28A3 gene locus in 3 cardio-protected patients and in a small European cohort of 99 individuals [27]. Within these SNPs, base editing of a variant in its antisense long noncoding RNA (lncRNA, AS1) rs11140490 in hiPSCs showed a causal relationship with cardiotoxicity by altering doxorubicin transport and cytotoxicity. Despite significant progress in this area, no risk prediction tool that could inform personalized decisions regarding anthracycline-based treatment and post-treatment surveillance has been developed to date, predominantly because of a lack of further information on the mechanism of action of the SNPs in the risk alleles on anthracycline-induced cardiotoxicity and methods for toxicity intervention. Therefore, we hypothesized that these genomic variants and AS1 may play contributing roles via alternative RNA processing or gene expression mechanisms, leading to altered CNT3 expression and function and eventually modulating anthracycline transport, uptake and cytotoxic activities. To test these possibilities, we investigated the specific role of SNPs rs7853758 and rs11140490 and of the region encompassing SNPs rs7030019, rs7853758, rs7853066, rs4877831, rs4877833, and rs10868135 within the SLC28A3 and in the intronic region of its lncRNA AS1, to evaluate the functional consequences on CNT3 gene expression and doxorubicin cellular transport and toxicity.

Results

Cardioprotective synonymous SNP rs7853758 decreases the spliced CNT3 transcript level: The intronic region encompassing SNP rs7030019 is essential for optimal CNT3 splicing

The synonymous SNP rs7853758 (Ex14: G > A, L461L) and the intronic SNP rs7030019 on SLC28A3 gene have been previously reported to be coinherited in cardioprotected patients [27, 47]. These SNPs are displayed in a schematic of SLC28A3 gene organization in relation to exons and introns (Fig. 1A). To understand the mechanism of action of these SNPs, we generated 3 distinct minigenes and designed studies utilizing these constructs to study their roles in CNT3 RNA biology, specifically their effects on CNT3 mRNA splicing and transcript levels (Supplementary Fig. S1A–C). All the constructs were confirmed for sequence integrity. A minigene construct containing entire Ex14-Int14-Ex15 and 300 bp spanning the upstream and downstream ends of the introns was able to efficiently generate Ex14–15 spliced RNA at the indicated time points upon transfection in HEK293 cells (Fig. 1B  i, ii). A larger construct containing the entire Ex13-Int13-Ex14-Int14-Ex15 sequence and 300 bp spanning the upstream and downstream ends of the introns was also able to generate Ex13-Ex14–15 spliced RNA at the indicated time points upon transfection in HEK293 cells (Fig. 1C  i, ii). Interestingly, a smaller product was obtained, which was more prominently observed on a higher % agarose gel and on subsequent polyacrylamide gel electrophoresis (PAGE) analysis with further optimized PCR conditions (Supplementary Fig. S2A). Sequencing of this smaller product revealed the presence of the Ex14 ∆ variant in this splicing reaction (Supplementary Fig. S2B). Interestingly, the WT construct lacking the region of location of an additional SNP (rs7030019) in Int13 (Ex13–14-15 truncated minigene) was inefficient in generating spliced products corresponding to Ex13–14-15 (Fig. 1D  i, ii). This was also evident in the real-time qRT–PCR analysis, which revealed a 99.9% reduction in the Ex13–14-15 spliced product with the ‘Truncated’ minigene compared with the ‘Full’ minigene (Fig. 1D  iii). This finding indicates that Int-13 (1936 nts) contains splicing regulatory regions deep within the intron (encompassing the SNP rs7030019; 958 nt downstream of Ex13, 977 nt upstream of Ex14), beyond the flanking 300 bases that are usually sufficient for splicing of the adjoining exons.

Figure 1.

(A) Schematic overview of CNT3 (SLC28A3), long noncoding RNA AS1 regions, and the locations of cardioprotective SNPs in the genome. (B) i) the ‘Ex 14-15 minigene construct’ contains full-length exons 14 and 15, full intron 14 and 300 intronic bases on either side of exons in a pcDNA3.1 vector containing the T7 promoter to drive transcription. ii) upon transfection in HEK293 cells, the minigene successfully generated spliced ex 14–15 products. (C) i) the ‘Ex 13-14-15 full minigene construct’ contains full-length exons 13, 14 and 15; full-length introns 13 and 14; and 300 intronic bases on either side of exons in a pcDNA3.1 vector. ii) upon transfection in HEK293 cells, the minigene successfully generates spliced ex 13–14-15 products. (D) i) the ‘Ex 13-14-15 truncated construct’ contains full-length exons 13, 14 and 15; full-length intron 14; a partial intron 13 (600 bases) lacking the intronic SNP (rs7030019)-containing region; and 300 intronic bases on either side of exons in a pcDNA3.1 vector. ii) In HEK293 cells, the minigene was ineffective at generating spliced ex 13–14-15 products. GAPDH was used as a housekeeping control, and NEOMYCIN was used as a transfection control. iii) qRT–PCR analysis of the spliced ‘13–14-15’ product generated using ‘full’ vs. ‘truncated’ ex 13–14-15 minigene constructs reveals loss of splicing with intron 13 deletion. (E) SNP rs7853758 was introduced into the ‘Ex 13-14-15 Full Minigene’ construct using site-directed mutagenesis, and the transcript levels and splicing efficiency of the resulting SNP-containing minigene were compared with those of the WT minigene. i) PAGE analysis using primers for Ex13 and Ex15, ii) qRT–PCR analysis of spliced ‘13–14-15’ products using primers for Ex13–14 and Ex15, iii) PAGE analysis using primers for Ex13–15 and Ex15, and iv) semiquantitative analysis using primers for Ex13–15 and Ex15. Assays and gels were normalized for transfection and cell number using the NEOMYCIN control. Tx = transfection. The gels represent at least 3 replicates. The primer pairs are depicted with small arrows. Twenty-four SNPs are depicted with blue bars for the SLC28A3 gene and black bars for AS1, the Int13 SNP (rs7030019) is depicted with a red bar, and the Ex14 synonymous SNP (rs7853758) and the AS1 Ex1 SNP (rs11140490) are depicted with red asterisks. ^*P < 0.05; ^****P < 0.0001.

Next, we introduced the SNP rs7853758 in the Ex14–15 and Ex13–14-15 full minigenes by SDM to assess its effect on splicing or transcript levels (Supplementary Fig. S2C–F). While we did not observe a change in the splicing pattern, a reduction in the transcript levels was observed in the presence of this SNP in the Ex13–14-15 minigene construct when primers that detect both variants were used (Fig. 1E  i). Next, we designed primers and performed PCR to specifically capture the Ex13–14-15 (Fig. 1E  ii) and Ex13–15 (Ex14∆) variants (Fig. 1E  iii, iv). The Ex13–14-15 and Ex13–15 (Ex14∆) variants presented 95% and 66% reductions, respectively. These findings indicate that the deep intronic Int13 region of the SLC28A3 gene (encompassing the SNP rs7030019) contains splicing regulatory elements, an Ex14∆ alternative spliced variant was generated for CNT3, and the SNP rs7853758 decreased the CNT3 transcript levels compared with those of the WT.

Cardioprotective SNP rs11140490 leads to alternative splicing in lncRNA AS1

There is limited knowledge on the endogenous production and role of the lncRNA AS1 in regulating CNT3 and anthracycline cardiotoxicity; however, recently, the SNP rs11140490 was shown to play an important causal cardioprotective role [27]. This SNP is located within the Int18 region of the CNT3 gene; thus, it also overlaps with Ex1 of its predicted AS1 (Fig. 2A). Since this variant is at the 2^nd nt position of the 5′ splice donor site in Ex1, we hypothesized that it might play a role in alternative splicing of AS1. To answer this question, we designed an AS1 minigene construct containing Ex1-truncated Int1-Ex2 300 bp upstream and downstream of Ex1 and 2 and confirmed its sequence identity (Fig. 2B  i, Supplementary Fig. S3). Notably, the truncated Int1 region (total Int1 ~ 11.6 kb) harbors six other SNPs that have been reported within the 7SNPs in the lncRNA AS1 [27], namely, rs7030019, 7 853 758, rs7853066, rs4877833 and rs4877831, and rs10868135 (Fig. 2A and B). The AS1 minigene construct efficiently generated the Ex1–2 spliced product at the indicated time points upon transfection in HEK293 cells (Fig. 2B  ii). Thus, these intronic SNPs (1–6, Fig. 2A) are unlikely to have an independent causal effect on the AS1 splicing pattern, as the truncated Int used in the minigene construct was sufficient to generate the correct Ex1–2 spliced product. Interestingly, we observed a faint band that migrated slower than the full-length (FL) AS1 spliced product, which was more prominent on the higher % gel and subsequent PAGE gels, which, upon sequencing, revealed the presence of alternative spliced products corresponding to the deleted (∆) and ∆ + trimmed AS1 variants (Fig. 2B  ii, Supplementary Fig. S4). Next, we introduced the SNP rs11140490 in the minigene construct to evaluate and compare the splicing pattern with that of the WT (Fig. 2C  i, Supplementary Fig. S3). A clear shift in the splicing pattern was observed with the SNP minigene compared with the WT (Fig. 2C  ii). The full-length spliced AS1 (FL) and the alternatively spliced variants ((∆) and ∆ + trimmed) were cloned and inserted into a mammalian expression vector for subsequent downstream analysis of their effects on CNT3 RNA, protein and function (Fig. 2D). RNA structure prediction using mfold identified stem loop structures for the AS1 FL WT, which were altered with the change in a single nucleotide at the SNP (Fig. 2E i, ii). To quantify the different spliced forms, we designed variant-specific primers and performed PCRs to capture the FL and deleted variants specifically (Table 1). While the FL spliced form was reduced (50%) with SNP introduction (Fig. 2F  i), the number of deleted variants (∆) and the overall ratio of ∆/FL increased 205 and 320%, respectively (Figs. 2F  ii, iii).

Figure 2.

(A) Schematic depiction of the lncRNA AS1 genomic region, highlighting its position relative to the SLC28A3 gene on chromosome 9. The AS1 gene, which is predicted to consist of two exons shown as peach rectangles, is interrupted by an intron shown in gray. The transcriptional direction is indicated by a gray arrow. The SNP rs11140490 is marked by a red asterisk^* at the splice junction of Ex1, with additional intronic SNPs ($) represented as gray bars. (B) i) the ‘AS1 Ex 1-2 minigene construct’ contains full-length exons 1 and 2, partial intron 1 lacking the intronic SNPs, and 300 bases included from either side of the exons cloned and inserted into a pcDNA3.1 vector. ii) post-transfection in HEK293 cells, this minigene effectively produces spliced ex 1–2 AS1 products at the indicated time points. (C) i) schematic of the ‘AS1 Ex1-2 minigene construct with SNP rs11140490 (A>G)’, highlighting the introduction of the SNP in Ex1 using site-directed mutagenesis. ii) post-transfection in HEK293 cells, this minigene generated the full-length Ex1–2 spliced product, as well as ‘variants’ of this spliced product through alternative splicing. (D) Endogenously spliced products from ‘AS1 Ex 1-2 minigene WT and SNP constructs’ were cloned and inserted into the pcDNA3.1 expression vector to generate expression constructs. (E) RNA structure prediction using mfold for AS1-FL i) WT (dG = −96.52) and ii) SNP (dG = −97.76). The SNP position is marked with a red circle. (F) i) qRT–PCR analysis of ‘full’ length (FL)-spliced ‘ex 1–2’ products generated using ‘WT’ or ‘SNP’ ‘ex 1–2’ minigene constructs, normalized for transfection, and the number of cells with the NEOMYCIN control indicates a decrease in AS1 transcript levels. ii) qRT–PCR analysis of ‘deleted’ (∆) spliced ‘ex 1–2’ products generated using ‘WT’ or ‘SNP’ ‘ex 1–2’ minigene constructs, normalized for transfection, and the number of cells with the NEOMYCIN control indicates an increase in the level of alternatively spliced AS1 transcripts. iii) the ratio of ∆/FL AS1 variants also increased with SNP introduction. Tx = transfection. The gels represent at least 3 replicates. The primer pairs are depicted with small arrows. ^**P < 0.01; ^***P < 0.001; ^****P < 0.0001.

Table 1.

Primers used in this study.

CNT3 Minigene Construct Primers
Code	Sequence	Binding Position	F/R
SA423	atatatggtaccggtaccAAGTTTCCACAACACAGG	300 bp upstream of Ex13 with R. Enz.	F
SA424	atatatctcgaggcggccgcATTGATTGCAGTCCCAGC	300 bp downstream Ex15 with R. Enz.	R
SA425	atatatggcgcgccgctagcAGAGGCTGAGGTGGGAGGATAAC	300 bp downstream Ex13 with R. Enz.	R
SA426	atatatggcgcgccgctagcCTGAGGAAGGAGGATCACTTGG	300 bp upstream Ex14 with R. Enz.	F
SA427	atatatggtaccggtaccCTGAGGAAGGAGGATCACTTGG	300 bp upstream Ex14 with R. Enz.	F
SA440	TACTACAGGCATGCACCACC	downstream of Ex13	F
SA441	ATTGGAGCCTCCAACTCCTG	upstream of Ex14	R
SA449	CCTTCCTGGCCCTGtTGTCTTTTATG	Ex14 (SDM)	F
SA450	CAATCAGATTCACAGCGATGTTGGC	Ex14 (SDM)	R
CNT3 Transcript Detection Primers
SA445	GTTCCATCCTCCCACTTGTTA	Ex13	F
SA446	TGATTCAGGGAATCTTCTAG	Ex14	F
SA447	CTCAAAACTCAGCTGTGGGTA	Ex14	R
SA448	TGATATATATTGCTGCACACCG	Ex15	R
SA549	GCCATGAAAATGGAAAGTGGTG	Ex13–14 only	F
SA546	GCCATGAAAATGGAAAGTGGCT	Ex13–15 only	F
SA544	CAGCACCTGCGTCATTGGC	Ex13	F
SA545	GCTGTCCTGCCATTCCACTC	Ex15	R
AS1 Minigene Construct Primers
SA428	atatatggtaccggtaccGTGGGTGAGTCAGAGCCCAG	300 bp upstream of Ex1 with R. Enz.	F
SA429	atatatctcgaggcggccgcTTCTAACGAGTTTCCAGG	300 bp downstream of Ex2 with R. Enz.	R
SA430	atatatggcgcgccgctagcGGTGAACAAGAAACTACC	300 bp downstream of Ex1 with R. Enz.	R
SA431	atatatggcgcgccgctagcCCAGGGGTCAGAAAAGTTTTTC	300 bp upstream of Exon 2 with R. Enz.	F
SA451	TGGCAGAAACAGTgGGTGAGCAG	Ex1 (SDM)	F
SA452	TCTATGGCTCATGTAACTGTTTC	Ex1 (SDM)	R
AS1 Expression Construct Primers
SA432	atatatggtaccggtaccAAAATTCAGCATCTGTACTTC	Ex1 with R. Enz.	F
SA433	atatatctcgaggcggccgcTTATGAAAACACCAACCTC	Ex2 with R. Enz.	R
AS1 Transcript Detection Primers
SA454	GCCCTTCAAAGAATACAGACTGTGG	Ex1	F
SA457	GTTGGTCACAGTTGCTTTGGC	Ex2 deleted region	R
SA532	TTCTTTTCCATTGCTGCT	Ex2 on deleted junction	R
Generic Primers
SA436	GGGAAGGTGAAGGTCGGAGT	GAPDH	F
SA437	GAGGGATCTCGCTCCTGG	GAPDH	R
SA455	GGCTGCTATTGGGCGAAGTG	NEOMYCIN	F
SA456	CCTGATGCTCTTCGTCCAGATC	NEOMYCIN	R
SA443	TTAATACGACTCACTATAGGG	T7 Promoter	F
SA444	CATTTAGGTGACACTATAGA	SP6 Promoter	R
CNT3 Short and Long PCR Primers
hSLC28A3_long	F1 – GGAGCACACAAACACCAAAC R1- TGAGTCACCCTGGACTACTT	Exon 3 (F)- Exon 19 (R)	2328
hSLC28A3_short	F2- CCAAGGTGATAGCTTGTTGCC R2- GAGAAGTGGCTGACCTCAAA	Exon 18 (F)-Exon 19 (R)	179

Note: SDM = Site Directed Mutagenesis; R. Enz. = Restriction Enzyme sequences; F = forward; R = reverse

SLC28A3-AS1 lncRNA regulates CNT3 expression at RNA and protein levels: Differential activity of AS1 lncRNA variants

The SNP rs7853758 induced a reduction in spliced SLC28A3 transcript levels, and the SNP rs11140490, associated with alterations in the levels of SLC28A3-AS1 lncRNA variants, which has overlapping sequences with SLC28A3 mRNA (Ex19 of SLC28A3 mRNA and Ex1 of SLC28A3-AS1), may have functional consequences for doxorubicin-induced cardiotoxicity (Figs 2A and 3A). Previous studies have also demonstrated that doxorubicin uptake is facilitated by a Na⁺-dependent influx mechanism^{34, 35} and that the loss of CNT3 reduces the doxorubicin IC₅₀ in iPSC-induced cardiomyocytes.²³ Here, we examined the transcript and protein levels of CNT3 in HEK293 cells in the presence or absence of exogenous SLC28A3-AS1 using AS1 expression constructs encoding AS1-FL, AS1-∆ and AS1-trimmed variants of AS1 described above (Fig. 2D). Notably, AS1 (427 bp) has an overlapping complementary sequence to potentially bind to 427 nt in the SLC28A3 DNA or precursor mRNA sequence or to a 178 nt sequence in the mature CNT3 mRNA transcript (3 nt in Ex18, 127 nt in Ex19 and 48 nt in the 3’UTR), with a 249 nt overhang at the 3′ end of AS1 (Fig. 3A). Since HEK293 cells express minimal amounts of endogenous CNT3 (C_t values of ≥38), we decided to express CNT3 exogenously using transient transfection of a full-length CNT3 expression plasmid to investigate its effects on mature CNT3 transcripts. We observed high induction of CNT3 transcripts (~50 000-fold) upon transfection with the pCMV3-CNT3-FLAG plasmid harboring full-length CNT3 with a FLAG epitope at the C-terminus (Supplementary Fig. S5). Transfection of the AS1 variant-expressing constructs (pcDNA3.1-AS1-FL, AS1-Δ and AS1-trimmed) at increasing plasmid concentrations resulted in a dose-dependent increase in the corresponding AS1 expression (Fig. 3B  i-iv). A concordant reduction in CNT3 transcript levels was observed upon cotransfection of AS1 variants with CNT3-FLAG plasmids, as measured using long-range PCR spanning Ex3-Ex19 (including the AS1 overlapping region) (Fig. 3C  i-ii). Moreover, this reduction in the CNT3 mRNA transcript level was also evident in the short 151-bp amplicon qRT–qPCR spanning Ex18–19 (Fig. 3D  i-iii) and PCR products run on agarose gels (Fig. 3E  i) and semi-quantified (Fig. 3E  ii). Furthermore, the AS1-mediated reduction in CNT3 transcript levels was also evident with decreased CNT3 protein levels, as determined by FLAG immunoblotting, with increased potency observed with AS1-Δ and AS1-trimmed than with AS1-FL (Fig. 3F  i-ii).

Figure 3.

(A) Schematic representation of the CNT3 mRNA transcript showing the AS1 complementary region and primers used for amplification of short and long amplicons of CNT3. (B) i-iii) agarose gels showing AS1-specific RTPCR products from HEK293 cells exogenously expressing CNT3-c-FLAG co-transfected with increasing concentrations (0.1 ng-10 000 ng) of AS1 expression constructs (AS1-FL, AS1-Δ, or AS1-trimmed). GAPDH was used as a housekeeping control. ii) respective densitometry plots of AS1 PCR products (iv). (C) i-ii) long amplicon (2.3 kb) RT–PCR products of CNT3 obtained using the primer pair F1-R1 (Ex3-Ex19) from HEK293 cells exogenously expressing CNT3-c-FLAG and AS1 variants. GAPDH was used as a housekeeping control. (ii) respective densitometry plots of CNT3 transcript levels. (D) i-iii) short amplicon (179 bp) quantitative RT-PCR of CNT3 using the primer pair F2-R2 (Ex18-Ex19) for HEK293 cells with the exogenous CNT3-C-FLAG and AS1 constructs at a concentration of 1 μg and 10 μg. Ct values of CNT3 and GAPDH (Ct_CNT3/GAPDH) are listed above the bars. (E) i) short amplicon (179bp) RT-PCR of CNT3 using the primer pair F2-R2 with GAPDH used as a housekeeping control, and the corresponding relative mRNA expression analysis is shown in (E) ii. (F) i-ii) representative immunoblots of HEK293 cells co-expressing CNT3 with C-terminal FLAG and AS1 constructs probed with an anti-FLAG antibody (F7425, sigma Aldrich). Β-actin served as a housekeeping control. The data are presented as the means±SEMs (n = 3; P < 0.05) and were analyzed with one-way ANOVA using GraphPad prism 10.0.

SLC28A3-AS1 lncRNA reduces CNT3-mediated doxorubicin transport and cytotoxicity

We first assessed the transport activity of HEK-293 cells transfected with a CNT3-expressing plasmid by measuring doxorubicin uptake using a fluorescence-based assay. CNT3-overexpressing cells demonstrated significantly greater doxorubicin accumulation than did the vector control, with the difference becoming more pronounced at concentrations exceeding 320 μg/mL (Fig. 4A  i). Furthermore, CNT3-mediated uptake of doxorubicin, as measured by doxorubicin fluorescence, was reduced when the cells were cotransfected with AS1 expression constructs (AS1-FL, AS1-Δ, or AS1-trimmed) (Fig. 4A  ii). Moreover, these reductions caused by AS1 variants were found to be in the order of potency: AS1-trimmed > AS1-Δ > AS1-FL. The results were further confirmed by confocal fluorescence imaging of doxorubicin in CNT3-overexpressing HEK293 cells in the presence of AS1 expression constructs. While reductions in doxorubicin nuclear staining were evident with all the AS1 constructs, the AS1-trimmed and AS1-Δ constructs showed maximal inhibition of doxorubicin fluorescence (Fig. 4B). Furthermore, qPCR analysis revealed no changes in CNT3 expression across different doxorubicin concentrations alone, further supporting these findings that the observed reduction in CNT3 expression is attributable to the AS1 expression and not to doxorubicin (Supplementary Fig. S6). Overall, these results demonstrate that CNT3 overexpression enhances doxorubicin transport, whereas AS1 and its variants can impair CNT3-mediated doxorubicin transport, with the deleted and trimmed AS1 variants resulting in a greater reduction in doxorubicin accumulation.

Figure 4.

(A) CNT3-mediated doxorubicin accumulation in HEK293 cells was quantified as relative fluorescence units (RFU) normalized to total protein. i) CNT3-mediated transport of doxorubicin at increasing drug concentrations (40, 80, 160, 320, and 640 μg/mL) in HEK293 cells in the absence and presence of exogenously expressed CNT3, and ii) HEK293 cells expressing CNT3, were co-transfected with AS1-FL, AS1-∆, or AS1-trimmed constructs and treated with a single doxorubicin concentration (320 μg/mL). (B) Doxorubicin (treated at 320 μg/mL) uptake in HEK293 cells expressing CNT3 was examined using immunofluorescence confocal microscopy 16 h post-transfection with various AS1-expressing constructs. Nuclei stained with DAPI are in blue. CNT3-FLAG (green), doxorubicin (red). Original magnification, 40×. Bars represent 20 μM. (C) Cytotoxicity assay IC₅₀ estimations in HEK293 cells transfected with CNT3-c-FLAG and various AS1-expressing constructs and treated with 100 μM phloridzin (a CNT3 inhibitor) after 72hours. The bars represent the means ± SEMs (n = 3; ^*P < 0.05, ^***P < 0.001 ^****P < 0.0001).

To measure the functional effects of AS1 on CNT3-mediated doxorubicin transport, we treated HEK293 cells expressing CNT3 and/or AS1 isoforms with doxorubicin at concentrations ranging from 0.312–10 μM for 72h, followed by cell viability measurements. Phlorizin dihydrate was used as a chemical inhibitor of CNT3, as it has been shown to inhibit sodium-dependent transport previously implicated in enhanced doxorubicin-induced cellular influx [41, 42]. The calculated IC₅₀ values (in μM) as a measure of cell cytotoxicity demonstrated that the presence of exogenous CNT3 (IC₅₀ 0.213 ± 0.006) enhances the drug sensitivity of HEK293 cells compared with that of control cells with endogenous CNT3 levels (IC₅₀ 0.372± 0.027) (Fig. 4C). However, CNT transport inhibition using phlorizin dihydrate (100 μM; IC₅₀ 0.837 ± 0.0725) conferred resistance to CNT3-mediated drug susceptibility (Fig. 4C). Similar protective effects were observed in HEK293 cells with ectopic AS1-FL and AS1-Δ and AS1-trimmed variants, with significantly high IC₅₀ values (AS1-FL IC₅₀ 0.730 ± 0.160; AS1-Δ IC₅₀ 0.816 ± 0.006; AS1-trimmed IC₅₀ 0.744 ± 0.019), confirming that AS1 and its variants can regulate CNT3 and decrease the sensitivity of these cells to doxorubicin, indicating protection against drug-induced cytotoxicity (Fig. 4C).

Discussion

The anthracycline class of drugs continues to constitute a cornerstone of modern chemotherapy regimens for various types of cancers, such as breast cancer, lung cancer, sarcomas, lymphomas and leukemias. However, they pose a significant clinical challenge because of the accompanying cardiotoxicity observed in many patients. In the present study, we applied the principles of pharmacogenomics to understand the mechanistic role of genetic makeup, i.e. multiple SNPs in the SLC28A3 gene, encoding the CNT3 transporter, which were found to be tightly associated with cardioprotection in patients receiving chemotherapy [27]. We demonstrated that within the 24 SNPs of SLC28A3 suggested to be cardioprotective and in linkage disequilibrium, the synonymous SNP rs7853758 in the Ex14 coding region of SLC28A3 and the variant rs11140490 in Ex1 of its antisense lncRNA (SLC28A3-AS1) have functional consequences in regulating CNT3 transcript and protein expression using alterations in RNA levels and alternative splicing (Fig. 5). Furthermore, we demonstrate that AS1 lncRNA and its variants regulate CNT3 levels and function, providing new insights into the role of CNT3 (and its antisense lncRNA) in anthracycline transport, which is relevant to the clinically observed cardiotoxicities. Moreover, the deep intronic region of Int13, which harbors the SNP rs7030019, is critical for the splicing of CNT3 precursor mRNA at Ex13–14, whereas 6 SNPs within the AS1 intronic region do not seem to be critical for AS1 mRNA splicing or expression. Furthermore, our studies demonstrated the presence of alternatively spliced variant forms of AS1, which differ in their ability to regulate CNT3 mRNA and protein levels, and its functional effects in mediating doxorubicin uptake and toxicity. Our findings suggest that antisense and splicing mechanisms may be exploited to modulate CNT3 function to reduce doxorubicin-induced cardiotoxicity.

Figure 5.

Schematic illustrating that among the 24 SNPs (listed in Supplementary Table S1, and shown in Figs 1 and 2) in SLC28A3—Identified as cardioprotective and in linkage disequilibrium—The synonymous SNP rs7853758 in the exon 14 coding region of SLC28A3, and the variant rs11140490 in exon 1 of its antisense lncRNA (SLC28A3-AS1), exert functional effects on the regulation of CNT3 transcript and protein expression. These effects are mediated through changes in RNA transcript levels and alternative splicing. Furthermore, SLC28A3-AS1 lncRNA and its variants modulate CNT3 expression and function, including doxorubicin transport and cytotoxicity, which are relevant to clinically observed cardiotoxicities.

Long noncoding RNAs (lncRNAs) are regulatory RNAs that constitute a large fraction of the human genome. While there are only ~ 20 000 protein-coding genes, there are ~ 36 000 lncRNA genes and ~ 191 000 lncRNA loci transcripts (GENCODE, version 47). We demonstrated that the AS1 sequence (including FL, Δ, and trimmed) is complementary to the intronic and exonic sequences of SLC28A3 and that the overexpression of AS1 variants regulates the expression and function of CNT3 to varying degrees; however, the mechanism of action and regulation of these variants remain to be investigated. For example, it is not clear how AS1 transcription is regulated, as lncRNAs can be transcribed by RNA polymerase I (Pol I), Pol II or Pol III; they can be highly cell type-specific and expressed only under certain conditions [48]. LncRNAs can regulate many aspects of cell differentiation and development and other physiological processes, and their role in heart health, development, aging and disease is constantly evolving [49]. AS1-FL is 427 bases long and shares 178 nt complementarity with the SLC28A3 mature mRNA transcript in Ex18, Ex19 and the 3’UTR. The most likely mechanistic explanation for AS1 regulation of CNT3 function would be its binding to SLC28A3 mRNA at this region and triggering nuclease-mediated mRNA degradation. However, other methods of CNT3 regulation, such as microRNA sponging, mRNA transcription, and post-transcriptional and translational alterations, cannot be excluded. Moreover, there is a good possibility that AS1 functions independently of SLC28A3, since we observed modest, yet significant, differences in CNT3 levels and functions with respect to doxorubicin uptake and toxicity in the presence of AS1 overexpression. The independent effects may be caused by associations with chromatin-modifying complexes or other RNA, protein or peptide interactors; altering the transcription of other genes; binding to other genomic regions; and regulating transcription, translation, metabolism and signaling, as has been demonstrated for other lncRNAs [48, 50]. Interestingly, despite their name, some lncRNAs have been found to associate with ribosomes and encode proteins or peptides that could have physiological or pathological roles [51]. While there is a possibility that AS1 FL could encode short peptides, an in silico open reading frame translational analysis of AS1 neither supports nor nullifies this theory, thus further experimental evidence will be definitive. Our study also raises the question of the mechanism and implications of the differential effects of AS1-FL, AS1-Δ and AS1-trimmed variants on CNT3 levels and activity, including doxorubicin uptake in cells (higher potency was observed with AS1-Δ and AS1-trimmed variants). Although all three variants share the same 178-nt complementary sequence to the Ex19–3’UTR of the SLC28A3 gene and mRNA, they may have different binding capacities due to differences in their RNA structures, interaction partners, stability, localization, etc. Moreover, the introduction of the SNP rs11140490 resulted in increased expression of the AS1-Δ and AS1-trimmed variants and decreased expression of the AS1-FL variant, which partly explains the causal role of this SNP in regulating CNT3-dependent doxorubicin uptake, a limiting step for xenobiotics causing tissue-specific toxicity. Moreover, as evident from different mfold predicted RNA structures for WT and SNP-AS1 (Fig. 3E), it is possible that SNPs containing AS1 have differential binding capabilities to SLC28A3 mRNA compared with WT AS1, resulting in differential CNT3 expression and function.

The role of alternative splicing in health and disease is a well-documented phenomenon, including our own studies on alternative splicing of a tumor suppressor [52–56]. Interestingly, of the 24 SNPs reported previously, 5 SNPs (rs4877272, rs3750406, rs10868133, rs11140489, and rs7858075) are present in the 3’UTR, and 2 SNPs (rs11140488 and rs12003403) extend further downstream (Fig. 1A). These SNPs may alter the promoter activity of AS1 and thus regulate AS1 expression or the stability of CNT3 mRNA by altering the 3′ end processing steps, such as cleavage and polyadenylation. Additionally, our findings of CNT3 transcript level reduction with the introduction of the SNPs rs7853758 in Ex14 and rs11140490 in Ex1 of AS1 corroborate the eQTL results for these variants reported earlier [27]. Of particular interest, we observed almost complete elimination of the splicing of Ex13–14, with the major intronic portion removed in the truncated minigene, despite possessing the canonical regulatory ~ 300 bases on either end. To confirm that this methodology works, our data demonstrate the correct splicing of AS1 Ex1–2 using a similar approach. The deep intronic region of Int13 deleted in the truncated minigene harbors the SNP rs7030019, suggesting that this SNP may be significant. Such examples of deep intronic mutations that play a regulatory role in human diseases, especially in cancer, are well documented [57, 58]. Another interesting finding in our study is the presence of the Ex14Δ variant in both the WT and the SNP-containing CNT3 minigene (Fig. 1E  i). Although we did not observe a change in the ratio of this variant to the WT with the SNP, Ex14 exclusion may lead to the production of a truncated nonfunctional CNT3 or CNT3 with different functional consequences (Ex 14 is 169 bp; thus, its exclusion would result in a frame shift), as demonstrated for another splice variant of CNT3, involving part of the out-of-frame insertion of intronic sequence [51].

Future studies to explore the cardiomyocyte-specific regulation of CNT3 and SLC28A3-AS1 are warranted. While this study primarily used HEK293 cells, it provides proof-of-concept functional effects of these cardioprotective SNPs. Although HEK293 cells may not fully capture the complexity of cardiac expression of CNT3, the findings might have broader implications for understanding how cardiotoxicity, or doxorubicin-induced toxicity in general, may occur and provide therapeutic avenues for circumventing drug toxicity. Given that anthracyclines affect multiple organs, with primary effects on cardiac tissues, future studies could examine whether similar mechanisms are at play in cardiomyocytes or heart-derived cell lines, which will allow for an increased personalized dosage regimen for cancer patients harboring these SNPs. Furthermore, determining whether these effects can be retained in cancer cells without increasing the dosage of anthracyclines is critical. Additionally, further data analysis on these cardioprotective SNPs that were previously suggested to exist in linkage disequilibrium within the SLC28A3 [27], revealed that this haplotype structure does not seem to be conserved amongst different ancestries, as the African and African American population deviates from the SNP distribution of Alternate allele vs Reference allele compared to the European and Asian population, which seem to be more similar (Supplementary Fig. S7, Supplementary Table S1). Even within the entire population, one of the synonymous SNPs, rs7867504, occurs at a higher frequency compared to the other 23 SNPs within SLC28A3. These observations should be considered when designing genetic aware trials, although more studies will need to be undertaken to guide such practices in clinic. However, regardless of SNPs, our study suggests that AS1 FL, Δ, or trimmed mimics, wholly or partially resembling (e.g. antisense oligonucleotides) the 178-nt region complementary to the SLC28A3 mature mRNA, can be explored as potential future drugs as cotherapeutics with cancer-treating anthracyclines.

Our study focused on the genomic variants and role of CNT3 in doxorubicin uptake; however, Genome Wide Association Studies (GWAS) have suggested that other SLC transporters harbor SNPs that are significantly associated with the chemotherapeutic outcomes of anthracyclines [26]. The CNT family encompasses three subfamilies of sodium-dependent transporters, designated CNT1, CNT2, and CNT3, which differ from each other and their equilibrative counterparts (called ENTs). The expression of CNTs is more restricted than that of ENTs, which are more widely expressed, which affects their functional roles in the body. For example, CNT1, which is predominantly expressed in tissues such as the kidney and intestine, impacts endogenous nucleoside absorption and reabsorption [49], whereas CNT3, which is expressed in the pancreas and heart [59], may play additional roles, such as regulating adenosine receptor signaling, among others. Additional associations (P < 0.01) with risk and protective variants in CNT1 have also been linked to anthracycline-induced cardiotoxicity. Moreover, in addition to the uptake of doxorubicin, high CNT3 expression levels are associated with favorable outcomes in acute myeloid leukemia patients treated with nucleoside analogs such as cladribine and fludarabine. Furthermore, our current study revealed that CNT3 plays a more significant role at higher concentrations of doxorubicin, and, not surprisingly, the SLC28A3 protective effect reached statistical significance only at anthracycline doses > 250 mg/m² [OR 0.43 (95% CI 0.22–0.78), P = 0.0093] in the clinic.

In summary, our study highlights the mechanistic roles of cardioprotective SNPs in the SLC28A3 gene and its lncRNA AS1, using mechanisms of transcription and post-transcriptional RNA processing. Additionally, our findings demonstrate that CNT3 overexpression significantly enhances doxorubicin transport under basal conditions. Furthermore, AS1 and its variants can impair CNT3-mediated doxorubicin transport, with the deleted and trimmed AS1 variants resulting in a greater reduction in doxorubicin transport. Overall, these studies provide significant progress for the development of predictive biomarkers (e.g. SNP variants) and cardioprotective therapeutic strategies (e.g. antisense approaches) to eventually enable a more personalized clinical approach to chemotherapeutic regimens targeted by anthracyclines.

Materials and methods

Materials

High-glucose Dulbecco’s minimal essential medium (DMEM; D115), MTT (thiazolyl blue tetrazolium bromide, catalog no-M5655), doxorubicin (catalog no-D1515), and phloridzin dihydrate (catalog no-274313) were purchased from Sigma Aldrich (St. Louis, MO, USA). Cell culture dishes (Corning^® 100 mm TC-treated) were procured from Corning and Nunc Thermo Scientific (Waltham, MA). A plasmid isolation kit (catalog no. D6942–02) was obtained from Macherey-Nagel and OMEGA (USA). Trypsin (catalog no. 25200–056) was purchased from Thermo Fisher Scientific (NY, USA). Low-EEO agarose (BP160–100) was purchased from Thermo Fisher Scientific (Florence, KY, USA).

Cells, reagents, antibodies, and plasmids

The HEK293 cells used in the present study were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) and were used within 20 passages or within a period of 2–3 months of revival. Fetal bovine serum (FBS) was purchased from Sigma–Aldrich (St. Louis, MO). The bicinchoninic acid (BCA) protein assay reagent and West Pico Chemiluminescent substrate were obtained from Pierce Chemical (Rockford, IL). All cell culture media were purchased from Cellgro Corning (Manassas, VA). Rabbit polyclonal anti-FLAG^® (#F7425) and mouse monoclonal β-actin (#A1978) antibodies were purchased from Sigma–Aldrich (St. Louis, MO). Rabbit polyclonal CNT3 H-44 (#sc-134529) antibody was procured from Cell Signaling Technology (Danvers, MA). The hSLC28A3 ORF mammalian expression plasmid with a C-Flag tag (designated pCMV3-CNT3-FLAG) and the control vector (pCMV3-FLAG) were procured from Sino Biological, Inc. (HG18648-CF).

Cloning

Minigene constructs were engineered with a focus on exons 13, 14, and 15 of SLC28A3 and exons 1 and 2 of AS1, including the SNP-containing regions and critical intronic segments.

Minigene constructs for CNT3

1. Ex 14–15 minigene containing entire Ex14-Int14-Ex15 and 300 bp spanning the upstream and downstream ends of the introns:

1. 2) Ex 13–14-15 truncated minigene containing the entire Ex13-truncated Int13-Ex14-Int14-Ex15 and 300 bp spanning the upstream and downstream ends of the introns:

1. 3) The Ex-13-14-15 full minigene construct contains the entire Ex13-Int13-Ex14-Int14-Ex15 and 300 bp spanning the upstream and downstream ends of the introns:

Minigene constructs for AS1

1. 4) Minigene containing the Ex1-truncated Int1-Ex2 and 300 bp upstream and downstream of Ex1 and 2

Targeted amplification of the relevant genomic regions was performed using human genomic DNA (Zyagen, San Diego, CA), Phusion DNA polymerase (Life Science Technologies), and primers containing restriction enzyme sites for Acc65I, KpnI, NheI, XhoI, AscI and NotI (New England Biolabs) (Table 1). Following amplification, the resulting PCR products were enzymatically digested with KpnI, NheI, or XhoI and ligated into the predigested pcDNA3.1 vector with compatible KpnI and XhoI sites via T4 DNA ligase (New England Biolabs). Ligated products were transformed into TOP10 competent cells (Life Science Technologies) and cultured on ampicillin-containing agar plates at 37°C overnight, and positive colonies determined by colony PCR were expanded in LB media. The plasmids were isolated using the Endo-free Plasmid Mini Kit II (Macherey-Nagel). To confirm the sequence integrity and the presence of the desired inserts, the plasmids were subjected to sequencing at the Genomic Core Facility at Stony Brook University using specific inserts and vector primers (Table 1).

Site-directed mutagenesis (SDM)

SNPs were introduced using the SDM primers listed in Table 1 to incorporate the SNP rs7853758 (G > A, L461L) on CNT3 and the SNP rs11140490 on AS1. High-fidelity DNA polymerase was utilized to amplify the entire plasmid (minigene constructs) to incorporate the desired mutation, and the parental DNA template was digested using the DpnI enzyme (New England Biolabs) to selectively degrade the methylated, nonmutated DNA strand. The remaining mutated strand was ligated to form a circular plasmid and transformed into competent cells for propagation. Plasmid DNA was isolated from the resulting colonies, followed by sequencing to confirm sequence integrity and introduction of the desired SNP.

Expression constructs

Spliced and amplified products were cloned and inserted into the pcDNA3.1 vector using primer pairs with appropriate restriction enzyme sites and Phusion DNA polymerase. The sequences were confirmed by sequencing the positive clones.

Minigene splicing assays

Splicing assays were performed by transfecting the minigene constructs into HEK293 cells and performing PCR using specific primer pairs (Table 1).

Cell culture and transfection

HEK293 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS and 1x penicillin–streptomycin (15140122) at 37°C in a CO₂ incubator. Cell transfections were carried out using either a CalPhos™ Mammalian Transfection Kit (#631312) from Takara Bio, Inc. (San Jose, CA, USA) or Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) in Opti-MEM (Gibco). Briefly, HEK293 cells were cultured under the indicated conditions and grown to ~ 60% confluency before being transfected with the respective plasmids at 1 μg. HEK293 cells were cotransfected with pCMV-CNT3-Flag and AS1 original, deleted, and trimmed variant plasmids at concentrations ranging from 0.1 ng to 10 μg.

RNA extraction, reverse transcription (RT), PCR, sequencing, and real-time qRT–PCR

Total RNA was extracted at the indicated time points or at 24 h post-transfection using the TRIzol reagent and the RNA easy kit (Life Science Technologies, Carlsbad, CA) or RNeasy Plus Mini Kit (74134) from Qiagen (Germantown, MD, USA). Five hundred nanograms of RNA were treated with DNase (Life Technologies Corporation) to remove any contaminating genomic DNA, followed by deactivation of the DNase with 25 mM EDTA (Life Technologies Corporation) at 65°C for 10 minutes and reverse transcription into cDNA using the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA) or RevertAid First Strand cDNA Synthesis Kit (#K1621) from Thermo Fisher Scientific (Florence, KY, USA), following the manufacturer's protocol. PCR analysis was performed using Taq Master mix (Qiagen) for gel analysis or Phusion DNA polymerase (Life Technologies Corporation) for downstream cloning using specific forward and reverse primers (Table 1). RT–PCR was performed on an Eppendorf thermocycler using standard Taq DNA Polymerase (M0273L, New England Biolabs, Ipswich, MA). The PCR conditions for Taq DNA polymerase, unless otherwise specified, were as follows: 95°C for 5′, 38 X (95°C for 30 s, 55°C for 30 s, 72°C for 30 s), and 72°C for 10′. The PCR conditions for Phusion DNA polymerase were as follows: 98°C for 30 s, 38 X (98°C for 5 s, 60°C for 10 s, 72°C for 45–50 s), and 72°C for 10′. Amplified products were analyzed by electrophoresis using 1–2% agarose gels or 5% nondenaturing polyacrylamide (PAGE). For sequencing or cloning, products were resolved on a 1% agarose gel, bands of interest were extracted using a gel extraction kit (Qiagen), and the resulting DNA was sequenced at the Genomic Core Facility at Stony Brook University via vector-specific or amplicon-specific primers. ImageJ was used for semiquantitative gel densitometry, where specified. Real-time qRT–PCR analysis was performed using PowerUp™ SYBR™ Green Master Mix (Life Technologies Corporation) under the following PCR conditions: 95°C for 5′, 40 X (95°C for 15 s, 55°C for 1′), followed by melt curve analysis. The data were analyzed using the ∆∆CT method [60] and normalized to either the housekeeping gene GAPDH or the transfection vector control gene NEOMYCIN. The PCR conditions for the CNT3 short and long amplicons are given in Tables 1 and 2.

Table 2.

Thermal cycler settings for CNT3 and AS1 PCR.

Type	Stage	Temperature	Time	Cycles
Short PCR	Initial denaturation	95°C	30 sec	1x
	Denaturation	95°C	30 sec	30x
	Annealing	60 °C	30 sec
	Extension	72°C	30 sec
	Final extension	72°C	30 sec	1x
	Hold	4 °C	∞
Long PCR	Initial denaturation	95°C	30 sec	1x
	Denaturation	95°C	30 sec	30x
	Annealing	60 °C	30 sec
	Extension	72°C	3 min
	Final extension	72°C	5 min	1x
	Hold	4 °C	∞

SDS–PAGE and western blotting

Western blotting analysis was carried out as previously reported (Nayak et al. 2022). Briefly, whole-cell lysates were prepared in ice-cold lysis buffer containing 150 mM NaCl, 1% Triton X-100, 0.1% sodium dodecyl sulfate, 50 mM Tris HCl (pH 8.0), 1 mM EDTA, 2 mM PMSF, 1 mM NaF, Pierce 1x protease inhibitor (A32955) and 1x phosphatase inhibitor (A32957). The protein amount was measured using a BCA protein assay kit, and 30 μg of protein was loaded onto 12% polyacrylamide sodium dodecyl sulfate (SDS–PAGE) gels and transferred (100 V, 2 h) to a PVDF membrane (Bio-Rad Laboratories, Hercules, CA). The membrane was then blocked with 5% BSA for 2 h at RT, incubated overnight at 4°C with the appropriate primary antibody (1:1000), and washed 3x with TBS-T for 5 min each. The membrane was incubated with HRP-conjugated secondary antibodies (Bethyl Laboratories Inc.) at a 1:10000 dilution in 5% skim milk for 1 h, and the bands were visualized using the Supersignal West Pico Chemiluminescent substrate (34 580 from Pierce, Rockford, IL) in a ChemiDoc Touch Imaging system (Bio-Rad Laboratories, Hercules, CA).

Doxorubicin transport measurement

HEK293T cells at 2 × 10⁵ cells/well were seeded onto 24-well plates and incubated in growth media overnight. Sodium-containing transport buffer [20 mM Tris–HCl (pH 7.4), 130 mM NaCl, 3 mM K₂HPO4, 1 mM MgCl₂.6H₂0, 5 mM glucose, and 2 mM CaCl₂] was used to conduct the transport assay. Briefly, HEK293 cells were transfected with pCMV3-FLAG (control) or pCMV3-CNT3-FLAG for 24 h using Lipofectamine, and the cells were reseeded for transient transfection with various AS1 plasmid constructs using the calcium phosphate transfection method. The transfected cells were treated with the indicated concentrations (40, 80, 160, 320, and 640 μg/ml) of doxorubicin hydrochloride (Chemscience LLC, NJ, USA) for 30 minutes at 37°C to evaluate transporter functionality. The reaction was stopped by washing the cells three times with ice-cold sodium buffer. After washing, the cells were lysed with 500 μl of 1x SDS buffer per well. The plate was incubated on ice for 15 min, and the lysates were collected and measured using a fluorescence plate reader with an excitation wavelength of 480 nm and an emission wavelength of 590 nm. A standard curve was prepared using known concentrations of doxorubicin diluted in sodium buffer to quantify intracellular doxorubicin concentrations. A BCA protein assay was performed on cell lysates to normalize doxorubicin uptake to protein content.

Doxorubicin and CNT3 visualization

After transfection and doxorubicin treatment (320 μM), the cells were fixed with 4% paraformaldehyde (PFA) for 10 min at room temperature. Following fixation, the cells were permeabilized with 0.5% Triton X-100 in PBS for 30 min. After permeabilization, the cells were blocked with 5% goat serum in PBS for 1 h to prevent nonspecific antibody binding. Primary antibody incubation was performed for one hour at room temperature with the anti-FLAG antibody (1:1000 dilution). The cells were subsequently washed three times with 1X PBST and incubated with an Alexa Fluor 488-conjugated anti-rabbit secondary antibody (Invitrogen, Waltham, MA, USA) for 1 h at room temperature. For visualization of the nuclei, the cells were counterstained with DAPI (1 μg/ml in distilled water) for 15 min. The slides were mounted with ProLong Gold Antifade Reagent (Thermo Fisher Scientific). Confocal microscopy was performed using an Olympus confocal microscope with a 60× oil immersion objective lens. Doxorubicin fluorescence was excited at 480 nm, and emission was recorded at 600 nm.

Cellular proliferation and cytotoxic IC₅₀ estimation

Cell viability estimation was carried out using the MTT assay (Riss et al. 2013). Briefly, HEK293 cells with or without AS1 or CNT3 transfection were counted, and 5 × 10³ cells were seeded per well of a 96-well plate, incubated overnight, and treated with or without phloridzin (100 μM). Phenol red-free medium containing dissolved MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; 5 mg/ml) was added to each well to a final concentration of 1 mg/ml. Following incubation for 4 h, the media in each well were replaced with 100 μl of dimethyl sulfoxide for solubilization. The plate was then covered with tinfoil and agitated on an orbital shaker for 15 min. The absorbance at 570 nm was recorded with a filter reference at 620 nm using a BioTek Synergy H1 plate reader (Agilent Technologies, CA, USA). The viability rate of the treated cells was calculated as follows:

IC₅₀ estimations were performed using nonlinear regression with four parameters, and dose–response curves were plotted. The IC₅₀ values from 3 biological replicates were averaged and plotted as bar graphs. The calculations were performed with GraphPad Prism 10.0.

Statistical analyses

Comparisons between values were performed using Student’s t test (two-tailed) or one-way or two-way analysis of variance (ANOVA) with Tukey’s post hoc analysis. For all the statistical analyses, P < 0.05 was considered statistically significant unless otherwise indicated. The results are expressed as the means ± SEM, unless otherwise mentioned, using Graph Pad Prism 10 software (Graph Pad Software, San Diego, CA, USA).

Declaration of generative AI and AI-assisted technologies in the writing process

None.

Web resources

ClinVar, https://www.ncbi.nlm.nih.gov/clinvar/.

Author contributions

Shipra Agrawal (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing), Monoj K Das (Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft), Tejinder Kaur (Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft), Sithumini Lokupathirage (Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft), Christian Reilly (Investigation, Methodology), Mahika Yarram (Investigation, Methodology), Rajgopal Govindarajan (Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing).

Conflict of interest statement: The authors declare that they have no competing financial or nonfinancial interests.

Funding

NIH R01DK133440 to SA and NIH R01HL164443 to RG.

Data availability

None.