The transporter ABCB7 is a mediator of the phenotype of acquired refractory anemia with ring sideroblasts

Abstract

Refractory anemia with ring sideroblasts (RARS) is characterized by mitochondrial ferritin (FTMT) accumulation and markedly suppressed expression of the iron transporter ABCB7. To test the hypothesis that ABCB7 is a key mediator of ineffective erythropoiesis of RARS, we modulated its expression in hematopoietic cells. ABCB7 up and down-regulation did not influence growth and survival of K562 cells. In normal bone marrow, ABCB7 down-regulation reduced erythroid differentiation, growth, and colony formation, and resulted in a gene expression pattern similar to that observed in intermediate RARS erythroblasts, and in the accumulation of FTMT. Importantly, forced ABCB7 expression restored erythroid colony growth and decreased FTMT expression level in RARS CD34+ marrow cells. Mutations in the SF3B1 gene, a core component of the RNA splicing machinery, were recently identified in a high proportion of patients with RARS and eleven of the thirteen RARS patients in this study carried this mutation. Interestingly, ABCB7 exon usage differed between NBM and RARS, as well as within the RARS cohort. In addition, SF3B1 silencing resulted in down-regulation of ABCB7 in K562 cells undergoing erythroid differentiation. Our findings support that ABCB7 is implicated in the phenotype of acquired RARS and suggest a relation between SF3B1 mutations and ABCB7 down-regulation.

Introduction

Myelodysplastic syndromes (MDS) with ring sideroblasts are characterized by accumulation of aberrant mitochondrial ferritin (FTMT) in the erythroblasts (1, 2). Three MDS subgroups have this specific feature; refractory anemia with ring sideroblasts (RARS) and isolated erythroid dysplasia, refractory cytopenia with multilineage dysplasia and ring sideroblasts (RCMD-RS), and RARS with marked thrombocytosis, RARS-T. Mutations involving TET2 and JAK2 reported in these MDS subtypes have not been associated with survival and risk for leukemic transformation (3, 4). Recently, we reported frequent mutations of the splicing factor gene SF3B1 in a majority of patients with MDS and ring sideroblasts(5). In an extended report, mutations of SF3B1 independently predicted for the presence of ring sideroblasts as well as for a favorable prognosis (6). Controversially, Patnaik et al., (7) and Thol et al., (8), argue that SF3B1 does not hold independent prognostic value and has no impact on patient outcome. Although altered RNA splicing has been suggested as a mechanism underlying the observed phenotypic changes in RARS (9, 10), the mechanism by which these mutations drive ring sideroblast formation and contribute to clonal advantage remains to be elucidated.

Patients with RARS show both erythroid hyperplasia and apoptosis in the bone marrow, indicating that apoptosis is executed during the later stages of erythroid differentiation (11, 12). However, mitochondrial release of cytochrome c and accumulation of FTMT is evident in early RARS erythroblasts, almost 100% of RARS granulocytes are clonal, and in addition myeloid colony growth is markedly reduced (13-15). Taken together, this indicates that the propensity for sideroblast formation is present at the hematopoietic stem cell level.

The iron transporter ABCB7 is mutated in X-linked sideroblastic anemia with ataxia (16, 17). Silencing of ABCB7 in HeLa cells leads to an iron deficient phenotype with mitochondrial iron accumulation (18). In mice, ABCB7 is essential for hematopoiesis in general (19). Furthermore, Mx1-Cre–mediated ABCB7 gene deletion and ABCB7^E433K mutation in mice results in siderocytosis without marrow ring sideroblast formation (20). We have previously shown that ABCB7 is down-regulated in RARS CD34+ marrow cells and that the expression level correlates with the percentage of marrow ring sideroblasts. The expression level decreases during differentiation, in contrast to a continuous increase during normal erythroid maturation (21-23).

To test the hypothesis that ABCB7 is a mediator of aberrant iron accumulation and reduced erythroid growth in acquired RARS, we modulated its expression in primary human hematopoietic cells. Down-regulation of ABCB7 led to markedly reduced erythroid growth and accumulation of FTMT. Normalization of ABCB7 expression in RARS erythroblasts rescued aspects of the RARS phenotype. Moreover, our data suggests a link between SF3B1 mutation and alternate exon usage of ABCB7 in primary RARS progenitors.

Materials and Methods

K562 cells

Human K562 cells were cultured in RPMI1640 (Sigma, USA) containing 10% fetal bovine serum (FBS), at 37°C and 5% CO2 in air. Cultures were split every 4 days to maintain an exponential growth before experiments. Live cell numbers were assessed by trypan blue staining and counted with a haemocytometer. To induce erythroid differentiation, K562 cells were treated by Hemin 50 μM in NaOH 0.1 N for 72 h. Differentiation of K562 cells was studied by analyzing the expression level of gamma-globin and glycophorin A (GPA) using quantitative RT-PCR and flow cytometry, respectively.

Human CD34+ cell separation and erythroblast culture

Aspirated bone marrow from patients with MDS with <5% myeloblasts and ≥15% ring sideroblasts (the term RARS is used hereafter both for RARS and RCMD-RS) and from healthy volunteers was subjected to mononuclear cell isolation by Lymphoprep (Axis-Shield, Oslo, Norway) density gradient within 1 h from aspiration, and CD34+ cells were positively selected using a magnetic-activated cell sorting (MACS) labeling system (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturers’ protocols. The purity of CD34+ cells isolated with this system was assessed at regular intervals and shown to be above 95% at all occasions. Following separation, CD34+ cells were cultured (1 × 10⁵/ml) for 14 d in Iscove’s medium supplemented with 15% BIT9500 serum substitute and recombinant human interleukin (rh-IL)-3 (10 ng/ml), rh-IL-6 (10 ng/ml), rh-stem cell factor (rh-SCF; 25 ng/ml), and 1% L-glutamine. Erythropoetin (Epo) was added to the medium at day 7 to a final concentration of 2 IU/ml, and fresh medium supplemented as above (plus Epo) was added at day 9 and 11. The erythroid maturation pattern obtained from these cultures has previously been described in detail (14, 15). Informed consent was obtained from both patients and controls, and the study followed the guidelines of the ethical research committee at Karolinska Institute.

Lentiviral vector construction and lentivirus preparation

pLKO.1-GFP-shABCB7

shRNA constructs against ABCB7 (shown in supplementary table 1) were obtained from Open biosystems. We evaluated five different constructs for ABCB7 silencing. We also constructed shRNA vectors with a GFP reporter by cloning NdeI/SpeI fragments from pLKO.1-shABCB7-D12 and pLKO.1-shABCB7-E1 into the pLKO.1-EGFP vector backbone (a kind gift from Jonas Larsson), resulting in pLKO.1-EGFP-shABCB7-D12 and pLKO.1-EGFP-shABCB7-E1.

pRRL-TRE-ABCB7-IRES-GFP-PGK-rtTA

IRES-GFP was amplified from MSCV-IRES-GFP²¹ with primers (supplementary table 2); IRESGFPMluIAgeIUpper and MSCVGFPNheLower and inserted into pSIN-TREmSEAP-hPGKrtTA2S (a kind gift from Maria Antonietta Zanta Boussif) opened with NheI and MluI (removing mSEAP) resulting in pRRL-TRE-IRES-GFP-PGK-rtTA. Human ABCB7 (Accession BC006323, Mammalian Gene Collection) was amplified by PCR using the primers MluIABCB7-fwd and MluI-ABCB7-rev and cloned into pRRL-TRE-IRES-GFP-PGK-rtTA opened with MluI resulting in the plasmid (pRRL-TRE-ABCB7-IRES-GFP-PGK-rtTA)

pRRLSIN.cPPT.PGK.IRES.YFP.WPRE and pRRLSIN-ABCB7-IRES.YFP.WPRE

The region containing IRES-YFP was amplified from MSCV-IRES-YFP (24) by PCR using the primers BamHI-MluI-AgeI-IRES-fwd and SalI-YFP-rev (supplementary fig 2). This fragment was ligated into pRRLSIN.cPPT.PGK.GFP.WPRE (from Didier Trono, Addgene plasmid 12252) from which the GFP was removed using BamHI and SalI resulting in the vector pRRLSIN.cPPT.PGK.IRES.YFP.WPRE. Human ABCB7 (Accession BC006323, Mammalian Gene Collection) was amplified by PCR using the primers MluI-ABCB7-fwd and AgeI-ABCB7-rev and cloned into pRRLSIN.cPPT.PGK.IRES.YFP.WPRE opened with MluI and AgeI (supplementary table 2).

Lentivirus production by 293FT cells

293FT cells (Invitrogen) were plated in 10cm Petri dishes in 7 ml Dulbecco’s modified Eagle’s Medium (DMEM) supplemented with Glutamine and 10% FBS. Cells were transfected using Lipofectamine™2000 (Invitrogen) with the packaging plasmid pΔR8.9 and envelope pMDG together with the lentiviral vector containing the gene of interest or a shRNA against it. After48-72 h, viral supernatant was harvested, filtered and frozen at −80°C.

Lentiviral transduction of BM CD34+ cells

Bone marrow (BM) CD34+ cells were cultured in Iscove’s medium supplemented with 15% BIT 9500 serum substitute and recombinant human interleukin (rh-IL)-3 (10 ng/ml), rh-IL-6 (10 ng/ml), rh-stem cell factor (rh-SCF; 25 ng/ml), and 1% L-glutamine, as previously reported (14). After 16 hours of prestimulation, normal bone marrow (NBM) CD34+ cells were added to wells that were coated with the fibronectin fragment CH-296 (Retronectin; Takara Shuzo, Otsu, Japan) and pre-loaded with viral supernatant. Cells were incubated for 24 hours, washed and used for subsequent experiments.

Colony-forming unit cell (CFU-C) assay

Lentiviral transduced BM CD34+ cells on day 3 (10 000 cells for RARS BM CD34+ cells, 1000 cells for normal BM CD34+ cells) were plated in triplicate in 35-mm dishes in H4230 methylcellulose (Stem Cell Technologies). The cells were cultured at 37°C in a humidified atmosphere with 5% CO2. The number of erythroid colonies (CFU-E and BFU-E), myeloid colonies (granulocytic colonies, CFU-G, monocytoid colonies, CFU-M, and mixed granulocyte/monocyte colonies, CFU-GM), YFP⁺ /GFP⁺ erythroid, and YFP⁺ /GFP⁺ myeloid colonies were counted on day 14 after transduction. Colonies were counted in a blinded fashion to avoid unconscious bias in evaluating the effect of the experimental vector.

Quantitative real-time RT-PCR

Total RNA was isolated and cDNA was reversely transcribed using Superscript II (Gibco BRL) according to manufacturer’s instructions. Real-time quantitative polymerase chain reaction (qRT-PCR) was performed for selected genes. Based on a recent publication on gene expression profiling of RARS erythroblasts(21) we tested genes with potential role in RARS pathogenesis including ABCB7, ALAS2, FTMT, MFN2, GATA1, GATA2, HSPA1B (HSP70), HSPA9B, FANCC, FOXO3A, gamma-Globin, MAP3K7. The expression level of the housekeeping gene GAPDH was used to normalize for differences in input cDNA. qRT-PCR was carried out using TaqMan gene expression pre-synthesized reagents and master mix (Applied Biosystems, Foster City, CA, USA) in 7500 Real-Time PCR system (Applied Biosystems). The expression ratio was calculated using the ΔΔCT method (25).

ABCB7 exon usage

Primers covering different regions of ABCB7 were designed based on the full ABCB7 mRNA sequence (NM_004299, NCBI) (Supplementary table 3). Real-time qPCR and data analysis were performed in a total volume of 20 μl using barcoded 96-well well plates (Applied BioSystems, Foster City, CA). Five microliters of cDNA, 10 μl SYBR green PCR Master Mix (Applied BioSystems) and both forward and reverse primers at 10 pmol/ul concentration were added to each well. To reach a total volume of 20 μl per well, DNase-RNase-free distilled water (Sigma) was added. The reaction was run at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Specificity of the PCR reaction was validated by melting curve analysis. All PCRs were performed in triplicate. Comparison between NBM and RARS exon usage was performed using the paired Student t-test.

Immunohistochemical staining for mitochondrial ferritin

K562 or erythroblasts cells transduced were fixed by 4% formaldehyde and analyzed for FTMT with a polyclonal antibody raised in rabbits and specific against FTMT (kindly provided by Sonia Levi, Italy). Bound antibody was detected by an immuno-alkaline phosphatase method (NovoLink Min Polymer Detection System). Negative controls were performed by replacing the primary antibody with PBS. Finally, slides were counterstained with Hematoxylin. Slides were studied and analysed on a Olympus BH-2 light microscope. In each case the percentage of positive cells was recorded.

Flow Cytometry

CD34+ cells from normal BM or RARS were cultured as described above. At defined time points (day 7, 11 and 14) aliquots were used for flow cytometric analysis. Since transduced cells (signifying cells with down-regulated ABCB7 or control shRNA) expressed GFP, antibodies were chosen to accommodate other colors. Cells were incubated with antibodies against APC-conjugated anti-Glycophorin-A, APCCy7-conjugated anti-CD36, PECy5-conjugated C-kit and PECy7-conjugated CD34. All antibodies were obtained from Biolegend or eBioscience. All analyses were performed on a LSRII-Fortessa.

ABCB7 protein detection in K562 cells by Immunohistochemistry

K562 cells transduced by either ABCB7-IRES-GFP or GFP-control vector were fixed by 4% formaldehyde and cyto-centrifuged on slides. ABCB7 protein expression was analyzed with a polyclonal IgG antibody raised in rabbit to human ABCB7 protein (ab65149, Abcam, UK), followed by a biotinylated goat-anti-rabbit IgG (Multilink HK 340-5K, BioGenex, USA) secondary antibody. Immunoreactivity was visualized by using Fast Red chromogen (Fast Red Substrate Pack, HK182-5K, BioGenex, USA), with alkaline phosphates (HK 331-5K, BioGenex, USA). Negative controls were produced by replacing the primary antibody with PBS. Finally, slides were counterstained with Hematoxylin. Protein detection was analysed by using an Olympus BH-2 light microscope.

Mutational analysis of SF3B1

The coding exons of SF3B1 were screened using massively parallel pyrosequencing of DNA pools from selected samples using the genome sequencer FLX system (Roche, Branford, CT, USA). A total of 27 oligonucleotide primer pairs were designed using Primer 3 v.0.4.01, targeting all protein coding exons for transcript CCDS 33356, as annotated by Ensembl genome browser. A total of n=13 samples were individually amplified and indexed using oligonucleotide 8-mer tags in 15μl PCR reactions. DNA pools for massively parallel sequencing were prepared and high throughput sequencing of pooled products was performed. Individual sample sequencing information was deconstructed as previously described (26), mapping of individual sample FLX 454 reads to the human genome (Build 37) was done using BWA and to facilitate variant calling pileup files were constructed with Samtools (27). All novel sequence variants were verified using conventional PCR based Sanger sequencing.

RNAi knockdown of SF3B1 in K562 cells

K562 cells were cultured at 37 °C in a 5% CO₂ humidified atmosphere, in RPMI1640 medium supplemented with 10% fetal bovine serum. For transfection of siRNA, 2 × 10⁶ K562 cells were electroporated in an Amaxa Nucleofector I, using the Amaxa cell optimization kit V (Amaxa, Gaithersburg, MD) according to the manufacturer’s recommendations. For each transfection, 30 pmol of siRNA were used. The transfection efficiency was assessed after 24 h by checking the percentage of GFP-positive cells obtained using the pmaxGFP fluorescent expression plasmid, confirming >80% of successfully transfected cells. Three non-overlapping siRNAs targeting SF3B1 and two different scramble sequences with GC content similar to the siRNA sequences (Stealth Select RNAi™, Invitrogen) were used in two independent experiments.

RNA was extracted using TRIzol Reagent (Invitrogen), following manufacturer’s instructions, and 2μg were retrotranscribed using RETROscript kit (Ambion). Expression levels of SF3B1 and ABCB7 were assesed in triplicate by real-time quantitative PCR on a LightCycler 480 instrument (Roche Diagnostics) using TaqMan gene expression assays (Applied Biosystems). Ct values were normalized to those of β2-microglobulin.

Results

Down-regulation of ABCB7 in normal progenitors reduces erythroid survival and colony growth and leads to accumulation of mitochondrial ferritin

We previously showed that ABCB7 expression is progressively down-regulated during erythroid differentiation of RARS progenitors (21). In order to optimize silencing of ABCB7 in human CD34+ progenitors, we assessed transduction efficiency of five different vectors and corresponding ABCB7 expression in K562 cells, selected because of their capacity to differentiate into recognizable progenitors of the erythrocytic lineage (Supplementary Figure 1). ABCB7 silencing did not influence growth and survival of K562 cells (data not shown). In initial experiments in normal bone marrow (NBM) CD34+ progenitors using the most efficient of the five silencing vectors, i.e., shRD12, we observed almost complete abrogation of colony growth, while the mock-transduced cells yielded comparable numbers of colony-forming units (CFUs) compared to untransduced cells (data not shown). Using the vector shRE1, for subsequent NBM experiments, we observed a clear inhibitory but not detrimental effect on erythroid colony formation (Figure 1A). Silencing of ABCB7 significantly reduced erythroid colony growth in three consecutive experiments. Myeloid colony formation was also reduced but not to the same extent (15-44%).

Figure 1. Downregulation of ABCB7 in normal CD34+ marrow cells reduces viability and inhibits erythroid maturation.

Bone marrow CD34+ cells from healthy individuals were transduced by lentiviral vectors expressing shRNAs either specific for ABCB7 or scrambled control. After successful transduction, aliquots were plated in CFU-assays and erythroblast cultures. A. Analysis of erythroid and myeloid colony growth from normal CD34+ cells (n=3), either untransduced (control), or transduced by scrambled-control or shABCB7, respectively. B. Expression of shABCB7-GFP (dashed black lines) or scrambled-control-GFP (grey lines) over the course of 14 days in erythroblast cultures. The percentage of GFP+ cells was analyzed by flow cytometry. Shown are two normal donors. C and D. Flow cytometric analysis of erythroid maturation during erythroblast culture on days 5, 7, 11 and 14. Shown is the degree of erythroid maturation analyzed by expression of CD36 and glycophorin A in the GFP+ fraction. Depicted are two representative examples from the same experiment as shown in A. EPO was added from day 7. E. The expression of FTMT, ALAS2, FOXO3A, and MAP3K7 was analyzed by qRT-PCR during the early phase of erythroblast culture (days 3 and 7). NBM CD34+ cells were transduced with pLKO.1-GFP-shABCB7 or pLKO.1-GFP and sorted on day 3 post transduction. Subsequently, GFP+ cells were cultured until day 7. qRT-PCR analysis was performed on cells sorted at the time of FACS (= day 3) and at day 7 of erythroblast culture. The black bar represents data of pLKO.1-GFP-shABCB7 positive cells, which was normalized to that of pLKO.1-GFP positive cells (grey bar).

Next, we studied ABCB7-silenced CD34+ cells using a well-established erythroblast culture model that results in >80% GPA+ erythroblasts at day 14, using normal bone marrow (13-15) and measured the size of the ABCB7-silenced or scrambled GFP+ population over the course of 14 days (Figure 1B). While no significant effect on relative survival was observed during the first 7 days of culture, the ABCB7 silenced cells decreased during the latter part of differentiation and fell below 1% at day 14. We employed flow cytometric measurement of CD36 and GPA expression as surrogate parameters for early and later stages of erythroid differentiation, respectively. Silencing of ABCB7 demonstrated no change in the acquisition of CD36 (Figure 1C-D). Analysis of GPA expression, however, demonstrated a decrease compared to progenitor cells transduced with scrambled control vector. These sets of experiments suggest that silencing of ABCB7 mainly affects the later stages of erythroid maturation. To address the question whether ABCB7 is equally expressed in myeloid cells we analyzed BM non-cultured erythroid (GPA+) and non-erythroid (GPA-) from the bone marrow of 3 RARS and 3 NBM. ABCB7 was less expressed (p = 0.0054) in non-erythroid (mainly myeloid) cells compared to erythroid cells (supplementary Figure 3).

Changes in gene expression in response to silencing of ABCB7

To further explore the gene expression pattern during differentiation of CD34+ cells with down-regulated ABCB7, we flow-sorted shRE1-GFP+ cells from transduced NBM CD34+ cells two days after transduction and analyzed expression levels in the pure population of cells after 3 and 7 days. Genes were chosen from a previous publication in which we defined aberrantly expressed genes in cultured RARS vs normal erythroblasts (21). Even a moderate down-regulation of ABCB7 at the time of sorting, i.e., 3 days of in vitro culture, caused a two-fold increase of FTMT expression (Figure 1E). As expected from previous data, ALAS2 expression did not change significantly after 3 days of culture and decreased dramatically at day 7, indicating a markedly reduced heme synthesis at this time point. qRT-PCR after 7 days also showed marked reduction of FOXO3A, involved in protection from oxidative stress, and MAP3K7, a negative regulator of apoptosis.

Over-expression of ABCB7 in RARS restores erythropoiesis and reverts gene expression towards the normal range

Next we addressed the question of whether forced expression of ABCB7 in RARS CD34+ cells could rescue erythroid maturation and restore colony growth of these cells. CD34+ cells from four RARS patients over-expressing ABCB7 clearly demonstrated enhanced erythroid colony growth compared with cultures transduced with mock vector (Figure 2A). The increase consisted mainly of YFP-expressing BFU-E colonies (Figure 2B, p = 0.014). Myeloid colonies showed a moderate increase (1.4-2 fold) not reaching statistical significance. The increased expression of ABCB7, varying from 28-462%, was tested by qRT-PCR and compared to un-transduced controls (Figure 2C). We also confirmed increased ABCB7 protein expression following transduction in transduced K562 cells (Supplementary Figure 2). Given the substantial increase in ABCB7 expression, we confirmed the functional effect of ABCB7 over-expression on erythroid colony-forming capacity. Importantly, ABCB7 forced expression decreased the expression level of FTMT in two out of three transduced cases suggesting that ABCB7 inhibited the mitochondrial iron accumulation (Figure 2D).

Figure 2. Up-regulation of ABCB7 restores erythroid function in RARS progenitors.

CD34+ cells from RARS patients were transduced with lentiviral vectors driving expression of ABCB7-YFP or a control vector (mock-YFP). Transduced cells were plated in CFU-assays and erythroblast cultures. A. Colony growth in RARS CD34+ cells (n=4), transduced by either mock or ABCB7-vector. Results are normalized to untransduced cells and error bars depict mean ± SE. B. Number of YFP+ colonies. Shown is the mean ± SE of YFP+ CFU-Cs. (p<0.05). C-D. Transduced cells were cultured in erythroblast cultures for 14 days and expression of ABCB7 and FTMT in RARS erythroblasts was analyzed at day 10.

Mutations of SF3B1 in the RARS patient cohort

Given the recent findings of mutations in the splicing factor SF3B1 in MDS with ring sideroblasts, we studied the mutational status in twelve of thirteen patients included in this paper in order to explore a potential link between SF3B1 and ABCB7. In one case, DNA material was not sufficient for analysis. SF3B1 was mutated in eleven of the twelve patients studied. The mutational status disclosed the K700E mutation in five patients, while the rest of the patients exhibited H662Q, E662D, K666M and N626D mutations (Table 1).

Table 1.

Patient No	Age	Sex	WHO Subtype /IPSS	Karyotype	SF3B1 mutation	Protein change	Analysis
1	76	F	RARS/Low	46,XX	Mutated c.2098 A>G	K700E	ABCB7 Over-expression
2	73	M	RCMD-RS/INT-1	46,XY	Mutated c.1986 C>G	H662Q	ABCB7 Over-expression Erythroid differentiation
3	71	M	RARS/Low	46,XY	Mutated c.2098 A>G	K700E	ABCB7 Over-expression
4	71	M	RCMD-RS/Low	46,XY	Mutated c.2098 A>G	K700E	ABCB7 Over-expression Exon usage
5	86	F	RARS/Low	46,XX	Wild type	-	ABCB7 Over-expression
6	86	F	RARS	46,XX	ND	-	Erythroid differentiation
7	72	M	RCMD-RS	46,XY	Mutated c.2098 A>G	K700E	Erythroid differentiation
8	63	M	RARS-T (JAK2-V617F mut)	46,XY	Mutated c.1866 G>C	E622D	Erythroid differentiation
9	49	M	RARS-T (JAK2 wt)	46,XY	Mutated c.2098 A>G	K700E	Erythroid differentiation
10	73	F	RCMD-RS	46,XX		H662Q	Exon usage
11	73	F	RARS	46,XX		K666M	Exon usage
12	82	M	RCMD-RS	46,XY		N626D	Exon usage
13	67	M	RARS	46,XY		E622D	Exon usage

Altered ABCB7 exon usage in primary RARS cells

To assess ABCB7 exon usage patterns in SF3B1-mutated patients, we performed qRT-PCR in 5 SF3B1 mutated RARS patients (Table 1) and 3 healthy controls. RNA samples were obtained at day 7 (Figure 3A) and day 14 (Figure 3B) of erythroid differentiation. Primers were designed to cover different regions of ABCB7 mRNA (Supplementary table 3). Figure 3A and B show the average expression in the control group and in the individual RARS cases. Profound down-regulation was observed for all regions, but the level varied substantially between patients and, interestingly, also during erythroid differentiation. Figure 3 C-F display delta Ct values for all analyzed samples and show how exon usage is more heterogeneous within the RARS population compared to NBM. Moreover, the average down-regulation compared to NBM for the exon regions differed; exon 2-4 62%±17.9 (p=0.001), exon 6-8 57.7%±32.3 (p=0.025), exon 7J8-9 38.8%±24.2 (p=0.04), and exon 11-13 34%±14.3 (p=0.007).

Figure 3. ABCB7 exon usage in RARS and NBM progenitors.

ABCB7 exons expression was quantified by q-PCR at day 7 (Figure 3A) and day 14 (Figure 3B) in RARS and NBM controls. Graph shows marked difference exon usage within RARS samples at both time points. C-F. ABCB7 Exon usage analysis at day 7. C: Exon 2-4, D: Exon 6-8; E: 7J8-9; F: Exon 11-13.

SF3B1 silencing in K562 cells reduces ABCB7 expression

For this experiment we used siRNA to knock-down SF3B1 expression in K562 cells. Three non-overlapping siRNAs targeting SF3B1 were used alongside two scramble control siRNAs. Expression levels of SF3B1 were monitored using real-time quantitative PCR over 10 days in culture. Efficient knock-down of SF3B1 (approximately 90%) was observed at days 1 and 2 post-transfection compared to cells transfected with scramble siRNA sequences (Figure 4A). The SF3B1 silencing was transient, and its expression levels increased gradually reaching similar levels to cells transfected with scramble siRNA sequences by day 10 post-transfection. Interestingly, silencing of SF3B1 led to a decrease in ABCB7 expression (Figure 4B). When the same experiment was performed in the context of hemin-induced erythroid differentiation, down-regulation of ABCB7 was more pronounced (Figure 4C and D) and correlated with γ-globin expression, analyzed as a surrogate parameter for erythroid differentiation (Figure E).

Figure 4. Relative expression levels of SF3B1 and ABCB7 post SF3B1 silencing.

K562 cells were transfected by using three non-overlapping siRNAs (#13-15) targeting SF3B1 (A and B). Identical experiments were also performed using hemin to induce erythroid differentiation in K562 cells (C, D and E). Depicted is the relative gene expression of SF3B1, ABCB7 and γ-globin. Percentages are calculated with respect to expression levels of both genes in cells transfected with two scramble sequences with GC content similar to the one of siRNA sequences.

Discussion

The formation of ring sideroblasts is an intriguing biological phenomenon characterized by the accumulation of aberrant FTMT. The recent finding of mutations in a core component of the spliceosome, SF3B1, in a majority of patients with RARS, and the association between this mutation and a potentially favorable prognosis raises several new questions (5-8, 28). However, SF3B1 mutations per se do not invariably lead to sideroblast formation, as exactly the same mutations have been reported in poor prognosis chronic lymphocytic leukemia patients refractory to fludarabine (29). The recent work by Yoshida et al., and others indicates that splice factor mutations are common in myelodysplasia in general, while being uncommon in de novo acute myeloid leukemia and myeloproliferative disorders (5, 30). Hence, SF3B1 mutations may be linked to ring sideroblast formation involving a pathway that is specific for a certain cell type or line of differentiation.

ABCB7 expression is underexpressed in CD34+ cells and differentiating erythroblasts from RARS patients (21, 31). However, several other genes are also underexpressed in RARS, and a functional relation between ABCB7 and RS formation has not been established in acquired RARS. To answer the question if ABCB7 deficiency contributes to the phenotype of RARS, we modulated its expression in primary human hematopoietic cells. Our selection of appropriate readout systems was based on previous work showing impaired erythroid cell survival and reduced clonogenic growth (11), expression of FTMT during erythroid differentiation (1, 14, 15), and aberrant gene expression in CD34+ cells and intermediate erythroblasts (21, 31). Pondarre et al., previously demonstrated, using a murine model, the essential nature of ABCB7 in the development of many tissues and how it is essential for hematopoiesis (20). Heme biosynthesis was directly or indirectly inhibited by the partial loss of function mutations in ABCB7. The present study shows that ABCB7 down-regulation, is functionally involved in the defective erythropoiesis of RARS, and that forced ABCB7 expression can restore erythroid function in this subtype of MDS.

Silencing of ABCB7 in the NBM cells inhibited erythropoiesis more than myelopoiesis, indicating that the ABCB7 protein is more important for differentiation of the erythroid lineage than the myeloid lineage. In particular, ABCB7 down-regulation impaired late erythroid differentiation and survival, while earlier and intermediate stages were unaffected. This pattern corresponds well with RARS morphology, which is dominated by intermediate and late dysplastic erythroblasts. Hence, it is possible that reduced ABCB7 levels results in marked cellular phenotypic changes first when iron turnover or heme synthesis increases above a critical threshold.

Next, we investigated whether over-expression of the gene could rescue erythroid growth of RARS CD34+ cells and found that the number of erythroid colonies increased by 77-115% in four subsequent patient samples. In addition, myeloid colony growth was moderately increased. The increase was specifically seen in the YFP-expressing erythroid colonies, indicating a direct function of the over-expressed gene. These data are the first to show that modification of a single gene could rescue erythropoiesis in RARS. In a series of clinical and biological studies we have shown that in vivo as well as in vitro treatment of RARS CD34+ cells and erythroblasts with G-CSF can improve erythroid survival and allow for terminal differentiation (14, 15, 32). However, G-CSF inhibits mitochondria-mediated apoptosis via up-regulation of a number of anti-apoptotic genes including heat shock proteins and MFN2 but does not influence the expression of the ABCB7 gene or mitochondrial ferritin protein expression (14, 15, 21).

The erythroblast culture model does not produce ring sideroblasts in a mode that allows for statistical comparison. However, we previously showed that FTMT, the key protein of ring sideroblasts, accumulates during early erythroid differentiation of RARS progenitors, and we used this as a marker for aberrant iron accumulation (14, 15). Down-regulation of ABCB7 followed by sorting of GFP-positive cells and subsequent erythroblast culturing induced a gene expression pattern similar to that we previously observed in RARS erythroblasts (21). Specifically, FTMT gene and protein expression increased after ABCB7 silencing in normal progenitors during erythroid differentiation. ABCB7 down-regulation and FTMT over-expression both lead to increased RNA-binding activity of iron regulatory proteins (IRPs) and consequently an increase in cellular iron uptake from transferrin (18, 33), which mainly incorporates into FTMT. The avidity of FTMT for iron is stronger than that of its cytosolic counterpart (33). Hence, FTMT over-expression may lead to the functional iron deficiency in RARS erythrocytes, evident by the finding of hypochromic erythrocytes in RARS (34). Moreover, FTMT expression decreased in RARS samples with forced expression of ABCB7, in spite of relatively moderate transduction frequencies. Expression of MAP3K7, a negative regular of apoptosis, was also reverted towards the normal range. Hence, forced ABCB7 expression improved erythroid survival and decreased aberrant mitochondrial iron accumulation, the two key features of RARS.

SF3B1 mutations are frequent in RARS, occurring in 64%-90% of all patients studied (5, 6, 9, 10, 28, 35, 36), and it is important to understand if there is an association between SF3B1 gene mutation and ABCB7 down-regulation in RARS. Eleven of the thirteen RARS patients in the present study had SF3B1 mutations and ABCB7 exon usage differed both compared to normal bone marrow and during erythroid differentiation, indicating that ABCB7 transcription indeed may be altered in mutated cases. The variation between RARS samples may either be a function of the different SF3B1 mutations, or reflect additional mutations in the MDS clone of these cases (5, 28). We also report that transient siRNA-mediated down-regulation of SF3B1 during erythroid differentiation results in reduced expression of ABCB7, strongly suggesting a link between the two genes. Whether this relationship forms part of the pathophysiology of RARS remains to be determined.

ABCB7 is neither mutated nor methylated in acquired RARS (21), but in this study we present evidence that the RARS phenotype is associated with down-regulation of this gene, and that forced expression of ABCB7 can restore erythroid growth and survival of RARS progenitors while decreasing the expression of aberrant mitochondrial ferritin. Random usage of different ABCB7 exons might explain the reduced ABCB7 expression in RARS disease, which may be secondary to altered SF3B1 function due to mutation. We suggest ABCB7 down-regulation as a common pathway leading to the RARS phenotype.

Supplementary Material

Supplementary Figure 1

Supplementary Figure 1. Efficiency of ABCB7 shRNA silencing and ABCB7 upregulation. A. Transduced K562 cells with shRNAs; level of ABCB7 mRNA was quantified by qRT-PCR. B. Transduced NBM CD34+ cells with pLKO.1-EGFP-shABCB7-E1 or with pLKO.1-EGFP as control. On day 3, ABCB7 mRNA levels were quantified by qRTPCR. C. Transduced K562 cells with an inducible lentiviral vector. D. CD34+ cells from three patients were transduced with lentiviral vectors either expressing YFP only (control) or ABCB7 and YFP (ABCB7 over-expression) mock).

Supplementary Figure 2

Supplementary Figure 2. The expression of ABCB7 protein by Immunohistochemistry. A.K562 cells stained by ABCB7 antibody (ab65149) at concentration of 5 μg/ml. B. ABCB7 staining of K562 cells over-expressing ABCB7 using the same concentration of antibody, black arrows shows a positive cells.

Supplementary Figure 3

Supplementary Figure 3. Comparison between ABCB7 expression in erythroid and non-erythroid RARS and NBM. Erythroid (GPA+) and non-erythroid (GPA−) from bone marrow of 3 RARS and 3 NBM were analyzed. ABCB7 is less expressed (p = 0.0054) in nonerythroid (mainly myeloid) cells compared to erythroid cells.