Induction of fetal hemoglobin: Lentiviral shRNA knockdown of HBS1L in β0-thalassemia/HbE erythroid cells

Imbalanced globin chain output contributes to thalassemia pathophysiology. Hence, induction of fetal hemoglobin in β-thalassemia and other β-hemoglobinopathies are of continuing interest for therapeutic approaches. Genome-wide association studies have identified three common genetic loci: namely β-globin (HBB), an intergenic region between MYB and HBS1L, and BCL11A underlying quantitative fetal hemoglobin production. Here, we report that knockdown of HBS1L (all known variants) using shRNA in early erythroblast obtained from β0-thalassemia/HbE patients triggers an upregulation of γ-globin mRNA 1.69 folds. There is modest perturbation of red cell differentiation assessed by flow cytometry and morphology studies. The levels of α- and β-globin mRNAs are relatively unaltered. Knockdown of HBS1L also increases the percentage of fetal hemoglobin around 16.7 folds when compared to non-targeting shRNA. Targeting HBS1L is attractive because of the potent induction of fetal hemoglobin and the modest effect on cell differentiation.


Introduction
The most common hereditable anemia in Thailand is thalassemia. There are two major types of thalassemia, α-and β-. The disease forms of β-thalassemia in Southeast Asia occur by a combination of β 0 with β + , and the most common form of β + is hemoglobin E [1]. G to A substitution at codon 26 of β-globin transcript leads to glutamate replaced by lysine residues. This mutation also generates an abnormal splice site resulting in an aberrant RNA processing competing with the normal splice site of β E -globin mRNA culminating in a reduction in β E -globin mRNA production and makes hemoglobin E a mild form of β + -thalassemia [2].
The symptoms of β-thalassemia/HbE display a complex trait [3,4]. There are numerous modifiers, for example, coinheritance with α-thalassemia [5], the number of α-globin genes, types of β-globin mutations [6], and modifiers of HbF levels [7]. The gold standard treatment for β-thalassemia is an allogenic bone marrow transplantation. With several constraints such as applicable to the relatively young age patients and suitable HLA-matched donors make this curative approach unavailable for most β-thalassemia patients [8]. Alternatively, genetherapy approaches to elicit the induction of β-or γ-globins, can diminish the load of unpaired α-globin chains. This will reclaim the balance of α/β-like globin ratios and thereby amend the ineffective erythropoiesis and shorten RBC lifespan [9]. Transduction of stem cells with lentivirus carrying short hairpin sequence (shRNA) targets to BCL11A and SOX6 has been shown to effectively induce γ-globin mRNA expression [10,11].
A genome-wide association study (GWAS) from a European twin cohort with significantly different F-cell values revealed that 10% of the variability is attributable to the HBB region, 19% in HBS1L-MYB intergenic region (HMIR), and 15% in BCL11A on chromosome 2 [12]. GWAS study in Thai β 0 -thalassemia/HbE patients exhibits an association of the same three regions with the disease severity and HbF levels [13].
In humans, HBS1L is identified as one of 5 protein-coding genes (AHI1, MYB, ALDH8A1, HBS1L, and PDE7B) in around 1.5 Mb region associated with HbF levels on the chromosome 6q23 [14]. HBS1L functions in combination with PELOTA in identifying the stalled ribosome complexes on truncated mRNAs and causing them to be separated into subunits [15]. Additionally, this incident starts mRNA degradation pathways.
We previously indicated that an SNP in exon 1 of HBS1L was associated with fetal hemoglobin and disease severity in Thai Chinese β 0 -thalassemia/HbE patients [16]. Expression of HBS1L is significantly diminished in erythroblasts throughout Phase II culture obtained from individuals with hereditary persistent fetal hemoglobin (HPFH) [17]. Here we extended that suppression of HBS1L expression by shRNA triggered considerable induction of γ-globin expression in both mRNA and protein levels while did not affect both α-and β-globin expression. Moreover, unlike any HbF inducer known so far, HbF's induction slightly perturbed erythroid differentiation characterized by flow cytometry.

Subjects
This study was permitted by the Ethical Committee, Siriraj Institutional Review Board (Si 290/ 2015). All participants aged greater than 18 years old. The 5 β 0 -thalassemia/HbE patients were recruited from the Thalassemia clinic, Nakhon Pathom Hospital, Thailand, and the 5 healthy volunteers were recruited at the Institute of Molecular Bioscience, Mahidol University, Thailand (S1 Table). The β 0 -thalassemia/HbE patients did not receive blood transfusion for at least 1 month before participation. The healthy donors did not carry any known types of thalassemia or hemoglobin variants. All participants provided their written informed consent to be a part of the study.

Sample collection and CD34 + cell differentiation
Mononuclear cells (PBMCs) were isolated from the peripheral blood by gradient centrifugation. First, plasma was removed from the peripheral blood samples after centrifugation at 600g for 10 min at room temperature. The packed cells were then diluted with 1X Dulbecco's phosphate-buffered saline (DPBS) containing 2 mM EDTA. The diluted cell solution at the volume of 25 mL was layered on 15 mL of lymphoprep™ density medium (density 1.077 g/mL) (Axisshield PoC AS, Oslo, Norway), followed by centrifugation at 600g for 20 min at room temperature without break. The PBMCs layer located between plasma layer and lymphoprep™ layer was carefully harvested and washed twice with the 1X DPBS buffer containing 2 mM EDTA and once with 1X DPBS containing 2 mM EDTA and 0.5% fetal bovine srum (FBS). Primary CD34 + hematopoietic progenitor cells were positively selected from the PBMCs using MACS 1 CD34 MicroBead kit (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) according to manufacturer's protocol.

ShRNA-carrying lentivirus vector development, selection, and transduction
For knockdown tests, lentivirus with HBS1L shRNAs was created. HBS1L shRNA1 (TRCN0000353597) and HBS1L shRNA2 (TRCN0000353653), two distinct target sense sequences of HBS1L shRNA, were chosen for this study because they were perfectly matched to HBS1L transcripts. While shRNA2 targeted all variants of HBS1L, shRNA1 only targeted the spliced variant transcripts 1 and 2. Table 1 lists the target sequences for shRNA1, shRNA2, and shNTC. A diagram of the HBS1L variants shows the shRNA-target locations (shRNA1 and shRNA2) (S1 Fig).  To transform bacteria, HBS1L shRNAs were ligated into the pLL-Puro, a modified version of the pLL3.7 lentiviral plasmid. In this plasmid, the EGFP gene was replaced by a puromycinresistant gene [18]. By digesting them with the XhoI and XbaI restriction enzymes, five clones of each of the HBS1L shRNA1 and HBS1L shRNA2 were chosen and characterized for insertion. Direct sequencing was used to verify the insert sequences.
The lentiviral vector plasmids previously chosen were co-transfected into HEK293T cells along with the three packaging plasmids pMDLg/pRRE, pRSV-Rev, and pMD2.G using the X-tremeGENE HP transfection reagent (Roche Mannheim Germany). Puromycin (Invitrogen, Carlsbad, CA, USA) selection was used to titrate lentiviral particles from the cultured supernatant at 48 and 72 hours after transfection in order to calculate the infection multiplicity (MOI). On day 4 of erythroblast culture, lentivirus containing shRNA was transduced with a MOI of 20 in 500 μL of Phase II media supplemented with 8 μg/mL of polybrene (Sigma-Aldrich 1 ) for 24 hours before being further cultivated in new Phase II medium for an additional 24 hours. Cells were then under selection in the presence of 1 μg/mL puromycin for 48 h, replaced with fresh Phase II medium without puromycin and continued culture until day 14.
At day 8 of culture, untransduced cells and normal erythroblast cells transduced with HBS1L shRNA1 and shRNA2, and shNTC (non-targeted shRNA) were extracted for RNA and protein extractions. According to the results of qPCR and Western blot, the knockdown of HBS1L by shRNA2 was more effective than that by shRNA1.
The synthesized cDNAs were quantified with specific primers for HBS1L transcripts using SYBR master mix (Applied Biosystems) according to the manufacturer's recommended conditions. Expression of α-, β-, and γ-globin was measured by SYBR green-based qPCR using primer sequences [19]. Quantitative PCR was performed on CFX96™ Real-Time system (Bio-Rad). The expression of α-, β-, and γ-globin mRNA in shNTC and shHBS1L transduced cells were calculated by 2 -ΔΔCt methods relative to untransduced (UNT) control as described below.
In comparison to untransduced cells, the abundance of the mRNAs for the following erythroid-related transcription factors, namely BCL11A, ZBTB7A, KLF1, GATA1, GATA2, MYB, and ATF4, was measured and displayed as a fold change [20]. The primer sequences utilized in this study are listed in S2 Table.

Wright-Giemsa staining
Cell morphology was evaluated from 5X10 4 of erythroid cells. The cells were spun by cytospin (StatSpin CytoFuge 2, Beckman Culter, Inc., CA, USA), stained with modified Wright-Giemsa dye (Sigma-Aldrich, Inc., MO, USA) for 5 minutes, and washed twice with water for two minutes. The slides were dried and cleaned with 70% ethanol before cell morphology observation under an Olympus CX31 light microscope.

Hemoglobin measurement by high-performance liquid chromatography (HPLC)
The remaining erythroid cells on day 14 were all collected and washed once with 1X DPBS. The cell pellet was lyzed for hemoglobin measurement by HPLC using the VARIANT™ II βthalassemia Short Program (Biorad, Hercules, CA, USA). This cation exchange HPLC technique uses an increasing sodium phosphate gradient buffer to separate different hemoglobin variants. Hemoglobin was measured using absorbance at 415 nm, and correction for turbidity was done by the absorbance at 690 nm [21]. Each 6.5-minute assay detects the most prevalent aberrant hemoglobin variations while also providing quantitative findings for the percentages of HbA2/E and HbF.

Statistical analysis
The analysis was conducted in R version 4.03 [22]. The Shapiro-Wilk test was performed to check for normal distribution. Differences between suitable groups were checked with paired t-test. We used one-tail t-test for comparison of globin mRNAs and two-tail t-test for the rest of comparisons. The significant levels of less than 0.05 were employed.

HBS1L expression during erythroid differentiation
HBS1L expressed at a low level but detectable in the early to intermediate stage of erythroid differentiation (day 6-10) and declined in the late stage (day12-14). This similar expression was observed in both healthy donors and β 0 -thalassemia/HbE patients in Fig 2. There were none that attained statistical significance at any time.

Evaluation of HBS1L shRNA effectiveness
Erythroblasts on day 4 were transduced with lentivirus carrying HBS1L shRNA1, shRNA2, and shNTC. Healthy cells were collected at day 8 for RNA and protein studies. Fig 3A, the expression of HBS1L mRNA reduced in cells transduced with HBS1L shRNA2 down to 0.25 (p-value = 0.0004, N = 3) compared with HBS1L shRNA1 down to 0.87. These results also corresponded with the reduction of HBS1L protein localized to both cytoplasmic and nuclear portions (Fig 3B). In our hands, we could not separate isoforms of HBS1L in SDS-PAGE. Accordingly, we managed to transduce erythroblasts with HBS1L shRNA2 or shNTC on day 4 for the subsequent experiments.
The flow cytometry revealed a slight increase in the percentages of R2 populations upon either shRNA2 or shNTC knockdown in β 0 -thalassemia/HbE patients (Fig 4D, N = 3). The R2 populations reflect the basophilic erythroblasts. This increment indicated a slight delay of erythroid cell differentiation after HBS1L knockdown in β 0 -thalassemia/HbE erythroblasts compared with untransduced but not shNTC transduced cells. It was noteworthy that regardless of treatment, cells were morphologically indistinguishable in Fig 4E and 4F. Fig 4G displayed the percentage of enucleation for each group. However, in the cases of shNTC and shRNA2 knockdown, the percentage of enucleated cells tended to decrease (N = 2, each). In the cells of the patients, the effects of HBS1L knockdown on enucleation appeared to be less.

Effect of HBS1L knockdown on hemoglobin production
To evaluate the impact of HBS1L knockdown on α-, β-and γ-globin expression, erythroblasts were harvested for RNA extraction on day 8 in healthy donors and day 10 in β 0 -thalassemia/ HbE patients, due to the low number of cells. The results indicated that HBS1L knockdown effectively induced γ-globin gene expression up to 1.69-fold at mRNA level in patients (compared with shNTC, p-value = 0.0117) and 1.44 folds in healthy donors (compared with shNTC, p-value = 0.05048), but there was no significant effect on α-or β-globin mRNA levels (Fig 5A and 5B). On average, we found a reduction of HBS1L mRNA expression of 71% in β 0thalassemia/HbE patients and 84% in healthy donors.
HPLC analysis of hemoglobin types revealed markedly elevated fetal hemoglobin levels in all cases after transduction. Fetal hemoglobin was indeed elevated regardless of baseline levels (untransduced). The base line fetal hemoglobin levels were inferred by the untransduced conditions (healthy donors and patients). The fetal hemoglobin was increased in additional 16.7% (p-value = 0.0085) in patients and 7.3% (p-value = 0.0516) in healthy donors after HBS1L knockdown (compared with shNTC control) (Fig 5C, left, N = 3) on day 14. Notably, fetal hemoglobin levels were not significantly different in shNTC transduction compared to the untreated. The percentage of HbA has not changed under any circumstances (Fig 5C, middle). The level of HbA2/E was significantly decreased reciprocally (Fig 5C, right). Typical HPLC analysis of both healthy and patient were represented in Fig 5D and 5E.
The erythroid-specific transcriptional factors BCL11A, GATA1, GATA2, KLF1, ZBTB7A, MYB, and ATF4 were then examined in untransduced, shNTC, and shRNA2 transduced erythroblasts (Fig 6). Also demonstrated was the quantity of HBS1L transcript that persisted in shNTC and shRNA2. The expression of GATA1, GATA2, KLF1, and MYB was relatively unchanged in all conditions. Notably, the levels of ATF4 transcripts were remarkably increased, 3-4 fold, in both healthy donors and β 0 -thalassemia/HbE patients. Both healthy and β 0 -thalassemia/HbE erythroblasts transduced with shRNA2 showed a modest rise in ZBTB7A. The level of BCL11A transcript, on the other hand, slightly decreased as a result of transduction of shRNA2 only in β 0 -thalassemia/HbE erythroblast.

Discussion
There are 3 isoforms of HBS1L proteins, V1, V2, and V3, while V3 is the short isoform. Sankaran and colleagues [25] reported nonsense and splice site mutations of isoforms V1, V2 and V4, rendering null expression while preserving the expression of isoform V3 (S1 Fig). The authors further reported 2 patients with relatively normal hematological pictures as well as fetal hemoglobin level. These patients also exhibited various growth and skeletal abnormalities.
To our knowledge, this was the first study to demonstrate the expression of HBS1L during erythroid differentiation. Knocking down of all variants of HBS1L selectively upregulated mRNA levels 2-3 folds (when compared to the untransduced conditions) of γ-globin expressions in both β 0 -thalassemia/HbE and healthy groups. When compared to cells that had not been transduced, the knockdown of BCL11A was originally reported to result in a 5-fold activation of γ-globin per total β-like globin mRNAs [26].  HBS1L together with PELO are specialized ribosome release factors involved in no-go and nonstop mRNA degradation. Endonucleolytic cleavage of the matching mRNA is initiated when the HBS1L/PELO complex binds a stalled ribosome [27].
ATF4 (Activating Transcription Factor 4) is a genuine direct target of NMD (nonsense mediated decay) since upregulation of its mRNA is caused by Rent1/Upf1 knockdown [28]. UPF1 serves as the focal point for bringing together other NMD factors and enlisting RNA decay factors [27]. In addition to the downstream reading frame that codes for the ATF4 protein, the ATF4 mRNA possesses two additional upstream open reading frames (uORFs). Only uORFs are translated when there is no stress, which causes the ribosome to pause at a termination codon upstream of exon junction complex. This could explain why ATF4 transcripts are vulnerable to NMD [28]. For Upf1 to mediate the connection with the RNA exosome complex and hence mediate NMD, Ski7 is required as an interacting factor in yeast [29]. Intriguingly, it has been discovered that the short form of HBS1L (HBS1LV3) serves as a human Ski7 homolog [30].
We suggest that HBS1LV3, like Ski7 in the Ski complex, associates with exosomes to boost UPF1 and cause the degradation of ATF4 mRNA. According to our experiments, the abundance ATF4 was increased after the knockdown of all HBS1L isoforms. As demonstrated by the induction of HbF by the protein phosphatase 1 inhibitor salubrinal in sickle erythroid progenitor cells, ATF4 is implicated in the elevation of γ-globin mRNA and an increase in HbF levels in erythroid progenitors via the eIF2α-ATF4 stress signaling pathway [31].
Studies utilizing the chromatin immunoprecipitation (ChIP) approach in erythroid precursors reveal a high GATA-1 binding signal in the middle of the intergenic HBS1L-MYB region on chromosome 6q23 [32]. It has been demonstrated that the presence of the rs66650371 minor allele in the intergenic region strongly correlates with MYB expression and reduced GATA-1 binding at this site in primary erythroid cells [33]. HBS1L expression, however, was not investigated. Additionally, RNAi knockdown of the transcription factors identified as residing in this region, LDB1, TAL1, and KLF1, suppresses MYB expression rather than HBS1L [32,33].
We used lentiviral vectors carrying shRNA targeted at all major HBS1L transcript variants. The expressed shRNAs entered the RISC complex to cleave complementary mRNAs after being transformed into siRNA by Dicer. It was anticipated that the GATA-1 transcript would continue to express itself. Indeed, we demonstrated that after HBS1L knockdown, the expression of GATA-1 and MYB remained unaltered. The CRISPR-Cas9 method was also reported to induce Hb F without affecting GATA-1 level when ZNF410 was knocked down [34]. Together, our findings imply that GATA-1 and the HBS1L-MYB intergenic region might not be involved in the increase of HbF following HBS1L knockdown.
Reduction of HBS1L expression had a marginal increase in the percentages of basophilic erythroblasts assessed by flow cytometry. Our finding was attractive because most of fetal hemoglobin interventions so far have unfavorable effects on markedly delayed erythroid differentiation. The upregulation was also seen at the protein levels. Hemoglobin analysis showed that, on average, in healthy donors and patients with HBS1L knockdown, the percentages of fetal hemoglobin ranged from 2 and 18 to 10 and 40, respectively. Fetal hemoglobin levels prominently rose to 40% over basal when BCL11A was knocked down in healthy cells [26]. Alternatively, HBS1L knockdown provided a potential target for treating β-thalassemia as well as other hemoglobinopathies such as sickle cell disease.