HMGA2 Moderately Increases Fetal Hemoglobin Expression in Human Adult Erythroblasts

Induction of fetal hemoglobin (HbF) has therapeutic importance for patients with beta-hemoglobin disorders. Previous studies showed that let-7 microRNAs (miRNAs) are highly regulated in erythroid cells during the fetal-to-adult developmental transition, and that targeting let-7 mediated the up-regulation of HbF to greater than 30% of the total globin levels in human adult cultured erythroblasts. HMGA2 is a member of the high-mobility group A family of proteins and a validated target of the let-7 family of miRNAs. Here we investigate whether expression of HMGA2 directly regulates fetal hemoglobin in adult erythroblasts. Let-7 resistant HMGA2 expression was studied after lentiviral transduction of CD34(+) cells. The transgene was regulated by the erythroid-specific gene promoter region of the human SPTA1 gene (HMGA2-OE). HMGA2-OE caused significant increases in gamma-globin mRNA expression and HbF to around 16% of the total hemoglobin levels compared to matched control transductions. Interestingly, no significant changes in KLF1, SOX6, GATA1, ZBTB7A and BCL11A mRNA levels were observed. Overall, our data suggest that expression of HMGA2, a downstream target of let-7 miRNAs, causes moderately increased gamma-globin gene and protein expression in adult human erythroblasts.


Introduction
Sickle cell disease (SCD) and the beta-thalassemias are among the most common genetic disorders worldwide [1,2]. These diseases are caused by the occurrence of mutations in the adult globin gene [2][3][4][5]. Although many clinical trials are ongoing [6], the only drug currently approved by the Food and Drug Administration for the treatment of SCD is hydroxyurea, which is known to increase levels of fetal hemoglobin (HbF) [7]. HbF has a sparing effect in sickle hemoglobin polymerization [8], and HbF levels of >20% are sufficient to ameliorate the pathogenesis of the disease [9]. In recent years, many advances have been made in the investigation of factors that regulate HbF levels. Genome-Wide Association Studies (GWAS) identified several relevant polymorphisms to the variation of human HbF levels, including the BCL11A locus [10,11]. Additional studies included the correction of the SCD phenotype by the knockdown of BCL11A in adult mice [12], and the discovery of an erythroid-specific enhancer that regulates the genetic variation of BCL11A levels [13,14]. More recent studies identified the transcription factor LRF/ZBTB7A as causing robust increases in the levels of HbF in human cultured erythroblasts [15]. Altogether, these studies provide new insights into candidate-therapeutic targets for the treatment of beta-hemoglobin disorders; however a clinically effective approach for the reactivation of HbF in adult erythroblasts is still missing.
In erythroid cells, the let-7 family of microRNAs (miRNAs) is highly regulated during the fetal-to-adult developmental transition, where let-7 family members are consistently up-regulated in adult blood compared to cord blood reticulocytes [16]. Additional studies demonstrated that the let-7 miRNAs, as well as LIN28A and LIN28B (proteins that bind and inhibit let-7 miRNAs) regulate HbF expression levels to >30% of the total hemoglobin produced in cultured erythroblasts from adult humans [17]. Furthermore, over-expression of LIN28A was also shown to ameliorate the hypoxia-related sickling of cultured mature erythrocytes from pediatric patients with sickle cell disease [18]. More recently, high LIN28B expression was associated with elevated levels of HbF in patients with juvenile myelomonocytic leukemia, thus identifying expression of LIN28B as the hallmark of a novel fetal-like molecular subgroup in this disease [19].
High Mobility Group AT-hook 2 (HMGA2) is a member of the HMGA non-histone chromatin protein family known as 'architectural transcription factors' and are downstream targets of the let-7 miRNAs. HMGA2 mRNA is negatively-regulated by the binding of let-7 miRNAs to the 3'-UTR region [20,21]. Interestingly, HMGA2 mRNAs are subjected to alternative splicing modifications and form four different isoforms that vary concerning their 3' structure, resulting in novel 3'-coding and -untranslated regions.
HMGA2 is known to both positively and negatively regulate the transcription of several genes by means of directly binding the DNA AT-hook domains (encoded by HMGA2 exons 1 to 3) to the minor groove of AT-rich regulatory elements, altering the DNA conformation or the chromatin structure [22][23][24]. During development, HMGA2 has been shown to have a role in the regulation of growth and adipogenesis in mammalian animal models [24,25]. Furthermore, HMGA2 is involved in a wide variety of biological mechanisms, such as cell-cycle progression, cell proliferation, cell differentiation and senescence [22,26]. Of note, overexpression of HMGA2 transcripts, possibly due to disruption of the let-7 miRNAs binding sites at the HMGA2 locus, accompanied an increased expression of endogenous HbF in a patient receiving lentiviral beta-globin gene therapy for the treatment of severe transfusion-dependent beta-thalassemia [27].
Here we investigated HMGA2 as a regulator of fetal hemoglobin. We observed moderate increases in gamma-globin gene expression and HbF levels in human adult cultured erythroblasts. Our data support that the LIN28-let-7 heterochronic pathway increases HbF, in part, by the targeting of erythroblast HMGA2 transcripts.

Ethics Statement
All related studies were performed after human subject review and Intramural National Institutes of Health Institutional Review Board approval. These studies were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all research subjects prior to participation in this study. Peripheral blood CD34(+) cells from healthy adult human volunteers were mobilized and isolated using a Clinimax device as previously described [28]. All cell processing, isolation, and cryopreservation were performed by personnel in the Cell Processing Section of the National Institutes of Health Department of Transfusion Medicine.

Cell culture
Cryopreserved healthy adult human CD34(+) cells were cultured ex vivo in a serum-free system for 21 days consisting of three phases: phase I from days 0 to 7, phase II from days 7 to 14 and phase III from days 14 to 21, as previously described [17]. Cell counts were performed throughout the culture period in a Z1 Coulter Particle Counter (Beckman Coulter, Indianapolis, IN) following manufacturer's instructions.

HMGA2 virus production and lentiviral transduction
Virus production was performed as previously described [29]. Briefly, the plasmid mixture was prepared for co-transfection in HEK293T cells (Thermo Scientific, Waltham, MA) following the manufacturer's instructions for the Calcium Phosphate Transfection Kit (Life Technologies, Grand Island, NY). The co-transfection mixture consisted of the vector plasmid (empty vector control or HMGA2-OE vector) with CAG kGP1.1R, CAG4 RTR2 and CAGGS vsv-g packaging helper virus plasmids [30,31]. The lentivirus-containing supernatant was concentrated following the Lenti-X Concentrator (Clontech) manufacturer's protocol and resuspended in 1/100 of the original supernatant volume in phase I culture medium. The presence of lentiviral particles was confirmed using the Lenti-X GoStix (Clontech) following the manufacturer's instructions. Cryopreserved CD34(+) cells were thawed and seeded at a density of 250,000 cells/ml in phase I culture medium. On culture day 3, cells were resuspended at a density of 2,000 cells/μl in phase I culture medium and transduced with lentiviral particles (an estimated MOI of 5 was calculated for the viral transductions). On the day after transduction, cells were resuspended in 4.0 ml phase I culture medium containing puromycin and transferred on day 7 into phase II culture medium without puromycin. For each condition, a puromycin selected control of mock-transduced cells was cultured in matched conditions until the end of phase II and analyzed to confirm puromycin selection by flow cytometry.

Flow cytometry analyses
Erythroid differentiation was assessed with transferrin receptor (CD71) and glycophorin A (GPA) antibodies (Invitrogen, Carlsbad, CA) on culture days 14 and 21 using the BD FAC-SAria I flow cytometer (BD Biosciences, San Jose, CA) as previously described [32]. Enucleation was quantitated by thiazole orange staining (Sigma) on culture day 21. To assess fetal hemoglobin, cells were stained with antibody directed against HbF (Invitrogen).

Quantitative PCRs
Total RNA was isolated using miRNeasy mini kit with QIAzol following manufacturer's instructions (Qiagen) and quantitated by reading absorbance at 260 nm in the spectrophotometer. In all samples, complementary DNA (cDNA) was synthesized using SuperScript III reverse transcriptase (Thermo Fisher) from the same amount of total RNA following manufacturer's instructions and as previously described [28]. Quantitative real-time PCR amplification was carried out in a 7700 Sequence Detection System using TaqMan Universal PCR Master Mix (Thermo Fisher) used following manufacturer's instructions. Q-RT-PCR assays and conditions were performed as previously described [17,33,34], with the addition of ZBTB7A (Hs00792219_m1) Assay-on-Demand Gene Expression Product (Thermo Fisher/Applied Biosystems, Grand Island, NY) for gene expression analyses. For the let-7 family of miRNAs, cDNA synthesis and Q-RT-PCR assays and conditions were performed as previously described [16] for let-7a, let-7b, let-7c, let-7d, let-7e, let-7f, let-7g, let-7i and miR-98. In all Q-RT-PCR reactions, absolute quantification for each transcript was determined by comparison with a standard curve that was run in parallel with biological samples as previously described [28]. For each let-7 family member, standard curves were constructed on the basis of the synthetic targeted mature miRNA oligonucleotide of known concentration (1:10 serial dilutions, n = 6). Reactions were performed in triplicate. A file containing each of the standard curves statistics is included as S1 Appendix.

Western blot analysis
Nuclear and cytoplasmic extracts were prepared using the NE-PER Nuclear and Cytoplasmic Extraction kit (Pierce Biotechnology, Rockford, IL) from culture day 14 erythroblasts as previously described [17]. Equal amounts of protein were separated following the Western blot protocol as previously described [17]. Blots were probed with antibody against HMGA2 (Abcam, Cambridge, MA). Lamin B1 was used as loading control (Abcam, Cambridge, MA).

Confocal analysis
For confocal microscopy, slides containing 70,000 sorted live cells on culture day 14 were fixed with 3.7% paraformaldehyde in PBS (Electron Microscopy Sciences, Hatfield, PA) and permeabilized with 2% BSA/PBS containing 0.1% Triton X (Thermo Fisher Scientific, Grand Island, NY). After cell permeabilization, slides were probed with antibody against human HMGA2 (1:500, ab184616, Abcam) overnight at 4˚C in 2% BSA/PBS solution. Slides were incubated with secondary antibody against rabbit IgG (H+L) conjugated with CF-555 (1:500, Sigma) in 2% BSA/PBS solution for one hour at room temperature. After secondary antibody incubation, slides were washed three times in PBS and once in distilled water then mounted with ProLong Diamond Antifade Mountant with DAPI (Thermo Fisher Scientific). Laser scanned confocal images were obtained from LSM 5 Live Duoscan (Carl Zeiss, Oberkochen, Germany) and analyzed with Zen2007 software.
HPLC for adult and fetal hemoglobins HPLC samples were prepared and analyzed as previously described [28,35].

Statistical analysis
Replicates are expressed as mean ± SD values and significance was calculated by one-or twotailed paired Student's t-test.

HMGA2 over-expression has minor effects in erythroblast in vitro differentiation and enucleation
HbF regulation in humans remains an active area of investigation. Recently, the heterochronic LIN28-let-7 pathway that has been associated with developmental timing events in C. elegans has been reported to be involved in the modulation of HbF levels [17,19]. Since the LIN28 RNA-binding proteins negatively regulate the let-7 family of miRNAs [36] and the let-7 miR-NAs are negative regulators of HMGA2 [21], we investigated the role of HMGA2 in the regulation of HbF. For this purpose, erythroid-specific over-expression of HMGA2 was performed and compared to empty vector controls. To prevent degradation in the adult CD34(+) cells, the 3'-untranslated region (UTR) of the gene where the let-7 miRNAs binding sites are located was excluded, and only the open reading frame from the HMGA2 variant 1 was used in the over-expression lentivirus transgene, keeping intact the region encoding the DNA-binding AT-hook domains. Transductions were performed in CD34(+) cells from adult healthy volunteers cultivated ex vivo in erythropoietin-supplemented serum-free media for 21 days. Expression of transgenic HMGA2 (HMGA2-OE) was confirmed by Q-RT-PCR (empty vector control: below detection limits; HMGA2-OE: 2.51E+04 ± 3.44E+04 copies/ng) and Western analyses at culture day 14 (Fig 1). Pancellular HMGA2 expression was demonstrated by confocal analysis of HMGA2-OE compared to control transduction (S1 Fig).
Cell proliferation and erythroblast differentiation were compared between empty vector control samples and HMGA2-OE. Cell counts at culture day 14 showed no significant differences between HMGA2-OE and empty vector control transductions (S2 Fig). Furthermore, erythroblast maturation characterized by expression and gradual loss of transferrin receptor (CD71) and expression of glycophorin A (GPA) on the cell surface was analyzed in the empty vector control and HMGA2-OE samples at day 14 and at the end of the culture period (Fig  2A), as well as enucleation was assessed by thiazole orange (TO) staining in the empty vector control and HMGA2-OE samples at the end of the culture period (Fig 2B). Maturation was equivalent among HMGA2-OE and empty vector control cells at culture day 14. However, less terminal maturation and enucleation were detected by culture day 21 [Average % CD71(+) and GPA(+): empty vector control: 54.4 ± 9.7, HMGA2-OE: 71.6 ± 11.1, p = 0.01; Average % CD71(-) and GPA(+): empty vector control: 45.6 ± 9.7, HMGA2-OE: 28.3 ± 11.1, p = 0.01]. Despite the decreased maturation, HMGA2-OE cells maintained the ability to enucleate to an average of 35.2 ± 17.0% in the ex vivo culture conditions.

HMGA2 over-expression moderately regulates HbF levels in human adult erythroblasts
To investigate the effects of HMGA2-OE in the expression levels of the globin mRNAs in cultured adult erythroblasts, Q-RT-PCR analysis of the alpha-and beta-globin genes was performed at culture day 14. No major differences were observed in mu-, theta-, zeta-, delta-and epsilon-globin transcripts in HMGA2-OE samples compared to empty vector control transductions (Fig 3A and 3B). Average increases of 23% and 27% in mRNA levels of alpha-and betaglobin were detected, respectively, but those increases did not reach statistical significance due to donor variability (S3A and S3B Fig). In contrast, gamma-globin mRNA was consistently increased in each donor resulting in an average increase of 189% observed in HMGA2-OE compared to controls (empty vector control: 5.02E+05 ± 8.62E+04 copies/ng; HMGA2-OE: 1.45E+06 ± 7.31E+05 copies/ng; p = 0.037; Fig 3B and S3C Fig). HPLC analyses at culture day 21 demonstrated a modest but significant increase in HbF levels in HMGA2-OE compared to controls (empty vector control: 5.41 ± 2.15%; HMGA2-OE: 16.53 ± 4.43%; p = 0.003; Fig 3C-3E and S3D Fig), which is consistent with the increase in gamma-globin mRNA levels. In addition, HbF staining by flow cytometry at culture day 21 demonstrated that HMGA2-OE has a pancellular expression of HbF compared to the control transductions ( Fig 3F). Overall, the data showed here supports the notion that the LIN28-let-7 pathway regulates HbF, at least in part, by targeting of the erythroblast HMGA2 transcripts.

HMGA2 over-expression has minor effects in erythroid-and HbF-related transcription factors
To determine if HMGA2-OE acts via known regulators of gamma-globin, Q-RT-PCR analyses were performed for expression of the LIN28 genes, the let-7 family of miRNAs as well as several erythroid-related transcription factors. LIN28A and LIN28B were not affected by HMGA2-OE  remaining below the detection limits. Likewise, no significant changes were detected in the levels of the let-7 family of miRNAs in HMGA2-OE compared to empty vector control samples (Fig 4). These results demonstrate that upon high-level erythroblast expression of HMGA2 there is no feedback loop that regulates LIN28 transcripts or the let-7 miRNAs. In addition, the expression pattern of the transcription factors KLF1, SOX6, GATA1, ZBTB7A and BCL11A were investigated and, despite a down-regulation trend in the expression level of these transcription factors, there were no statistically significant changes detected in HMGA2-OE compared to empty vector control transductions (Fig 5A-5E). Overall, HMGA2-OE caused mild changes in several transcription factors that are known regulators of HbF, but none of the changes reached statistical significance in this study.

Discussion
Understanding the molecular processes controlling the fetal-to-adult hemoglobin transition is crucial to determine the drivers and effectors of HbF, which have the potential to be used in the treatment of beta-hemoglobin disorders. Here we focus on recent discoveries that highlight the interplay between the heterochronic LIN28-let-7 pathway and the regulation of HbF. Our results demonstrate that HMGA2, a downstream target of the let-7 miRNAs [37], plays a moderate role in the modulation of gamma-globin and HbF levels. Importantly, HMGA2 had only modest effects in the activation of HbF, so these results are unable to fully explain the more robust HbF effects of let-7 miRNA suppression by LIN28 in the same culture model [17].
HMGA2 is physiologically expressed during embryonic and fetal developmental stages, and it is down-regulated to very low or absent levels in normal adult tissues [38]. In addition, Hmga2 transcripts are differentially expressed between fetal and adult mice hematopoietic stem cells [39]. Here we expressed a let-7-resistant HMGA2 variant 1 driven by the erythroid specific gene promoter region of the human SPTA1 gene. Of note, HMGA2 proteins do not have a direct transcriptional activation function; however, they can control gene expression by interaction with DNA and transcription factors (TFs) to form complexes that support the TFs with binding to the DNA [40], or exclusively interacting with the TFs and, consequently, increasing their affinity to bind to its target DNA [41].
It has been shown that transgenic mice carrying a 3'UTR-truncated Hmga2 cDNA showed proliferative hematopoiesis, erythropoietin-independent erythroid colony formation, as well as hypercellular bone marrow and splenomegaly [42]. Conversely, similar hyperproliferative HMGA2 Increases Fetal Hemoglobin Levels changes were not associated with increased expression of a truncated HMGA2 protein, accompanied by an increase in expression of endogenous HbF, in a patient receiving lentiviral betaglobin gene therapy for the treatment of beta-thalassemia [27]. We report here that erythroidspecific-let-7-resistant HMGA2 over-expression had only minor effects in ex vivo erythropoiesis and enucleation. While mRNA levels of alpha-and beta-globin genes were both increased upon HMGA2 over-expression, these increases were most likely the result of donor variability ( S3 Fig). However, a moderate and consistent activation of gamma-globin expression was observed upon HMGA2 over-expression in all donors, as well as increased and pancellular HbF levels to an average of 16% of the total hemoglobin produced. Despite the high HMGA2 protein levels detected in each donor's cells, HbF was increased to levels lower than those observed in previous studies directly targeting LIN28 or let-7 miRNAs performed under similar culture conditions [17]. Due to a modest effect on HbF with HMGA2-OE, additional factors are being sought that may act in combination with HMGA2 for the activation of HbF in adult erythroblasts, such as other downstream mRNA targets of the let-7 family of miRNAs.
Over-expression of HMGA2 did not significantly affect the levels of LIN28 or any of the let-7 family members, suggesting that the effects were not mediated by feedback regulation within the let-7 regulatory circuit. To explore potential downstream effects of HMGA2 over-expression in adult erythroblasts, the expression levels of transcription factor regulators of HbF were investigated. KLF1, SOX6, GATA1, ZBTB7A and BCL11A mRNA levels were not significantly reduced with the over-expression of HMGA2. We interpreted these results as an indication that HMGA2 effects are not primarily mediated by a singular transcription factor in this group. Alternatively, the subtle differences in the erythroblast differentiation (Fig 2) mediated by HMGA2-OE may influence the effects of transcription factors upon globin gene regulation as well as HbF levels.
Finally, our studies on HMGA2 add to recent efforts aimed toward developing an experimental bridge that connects the ancient and well-conserved LIN28-let-7 developmental timing circuit with the abundant knowledge of globin gene regulation during human ontogeny. Importantly, since direct targeting of the let-7 miRNAs in adult erythroblasts causes more robust effects in the reactivation of HbF levels, the results shown here suggest that HMGA2 expression is not sufficient to fully explain the downstream effects of the LIN28-let-7 circuit in the regulation of HbF levels.