Human erythroid cells

The HUDEP human erythroid cell lines have been described elsewhere.¹⁹ Cultures of primary human erythroid progenitors^{20, 23-25} were established from anonymized buffy coats (Dutch blood bank Sanquin, project NVT0146) or ~10 mL of peripheral blood derived from SCD patients (Erasmus MC MEC-2018-1422).

Cell culture

Cells were cultured at 5% CO₂, 20% O₂, and 95% ± 5% humidity at 37°C. Cell density was kept between 0.3 and 1.5 × 10⁶ cells/mL refreshing the medium every second or third day. The cells were cultured in fully defined serum-free Cellquin medium developed at Sanquin Research.^{20, 23, 24} Cellquin medium was prepared on site from Iscove's modified Dulbecco's medium (IMDM) (Cat. #P04-20 250; PAN Biotech); the full list of ingredients is provided in Table 1. A readily prepared version of Cellquin is available (Cat. #P04-20251K; PAN Biotech). Doxycycline, used in the proliferation medium, promotes the expression of the HPV16-E6/E7 transgene that was introduced to immortalize the HUDEP cells¹⁹ and was added to the cultures every other day. In the differentiation medium, doxycycline was removed to allow the cells to exit the cell cycle, which is required for terminal differentiation. Cell density and diameter distribution of HUDEP cultures were determined with a CASY Model TTT electrical current exclusion cell counter (Omni Life Science). For storage in liquid nitrogen, HUDEP cells were frozen in fetal calf serum (FCS, Cat. #FBS-12A; Capricorn Scientific) containing 10% dimethyl sulfoxide (DMSO), typically storing 5–20 million cells in 1 mL suspension.

HEK293T cells were cultured at 5% CO₂, 20% O₂, and 95% ± 5% humidity at 37°C in DMEM supplemented with 50,000 U/L penicillin, 50 mg/L streptomycin (Cat. #P0781; Sigma-Aldrich), and 10% FCS (Capricorn Scientific). Cell density was kept below 90% confluence.

Sanger sequencing

Polymerase chain reaction (PCR)-amplified DNA fragments (Supporting Information S1: Table S1) were subjected to Sanger sequencing and analyzed using CLC Main Workbench 8 (Qiagen). To estimate Cas9 efficiency, Sanger sequence trace deconvolution was performed with TIDE.²⁶

Guide RNA (gRNA) design

CRISPOR²⁷ was used to design gRNAs in the 3′-UTR of HBG1. Two HBG1-specific gRNAs were identified (Supporting Information S1: Table S1). Template DNA for homology-directed repair was derived from plasmid DNA or synthesized as a single-stranded oligonucleotide (Supporting Information S1: Table S1; IDT).

Nucleofection

HUDEP2 cells (2 × 10⁶) were nucleofected (EW-113 Program 1) on a 4D-Nucleofector system (Lonza) using pLentiCRISPR v2 (Addgene #52 961)²⁸ plus double-stranded template DNA or Cas9-sgRNA ribonucleoprotein (RNP) complex plus single-stranded template DNA (Alt-R CRISPR-Cas9 system; IDT). To isolate clones, cells were single-cell sorted (FACSARIA III; Becton Dickinson) 2 days postnucleofection.

Aγ-HiBiT detection

The Nano-Glo HiBiT lytic assay (#N3030; Promega) was used for Aγ-HiBiT luminescence detection with a GloMax plate reader (Promega). Western blots were probed using the Nano-Glo HiBiT blotting assay (#N2410; Promega).

SNP arrays

HUDEP2 DNA was used for the Global Screening Array (GSAv3); data were analyzed with GenomeStudio (Illumina).

Next generation sequencing

RNAseq and ATACseq were performed as described.²⁹ Reads were mapped against the GRCh38 human reference using HiSat2 (version 2.1.0).³⁰ RNA expression values were called using htseq-count.³¹

Flow cytometry

CD253a (#561775; Becton Dickinson) and CD117 (#562435) were used for flow cytometry (LSR-FORTESSA; Becton Dickinson); data were analyzed with Flowjo v.10.6 (Becton Dickinson).

Quantitative PCR

Globin expression was determined by qPCR using PSMD1 as a reference (Supporting Information S1: Table S1).^{29, 32, 33} The $$ method was used to calculate expression levels.³⁴

Protein analysis

High-performance liquid chromatography (HPLC) analysis of HbF/HbA and western blots were performed as described.²⁹ Primary antibodies recognized γ-globin (sc-21756; Santa Cruz Biotechnology), β-globin (sc-21757), and NPM1 (ab10530; Abcam). Appropriate secondary antibodies enabled detection with an Odyssey CLx Imaging System (LI-COR Biosciences).

HTS

Assay optimization for the use of Aγ-HiBiT cells in HTS applications and the pilot screen of 5632 known bioactive drugs were performed at the EMBL Chemical Biology Core facility.

Statistical tests

Statistical analyses were performed using two-tailed t-tests (GraphPad Software). *p < 0.05; **p < 0.01; ***p < 0.001.

Sequence analysis for HBG1-specific CRISPR-Cas9 guide RNA designs

The fetal γ-globin chains are encoded by two nearly identical and closely spaced genes, HBG1 and HBG2. The high level of sequence homology between the two genes poses the risk of simultaneous Cas9-induced double-strand breaks at both genes resulting in large deletions or inversions.³⁵ CRISPR-Cas9-mediated tagging of the endogenous fetal γ-globin chains therefore requires a gene-specific gRNA design that leaves the other β-like globin genes in the HBB locus unaffected (Figure 1A,B). A comparison of the two 3′-UTR regions in the human reference genome GRCh38 identified a unique nucleotide at position +17 (G_HBG2 vs. T_HBG1). Also, HBG1 contains an extra A at position +55 that is not present in HBG2 (Figure 1B). In HUDEP2 cells, gene-specific PCR amplicons of HBG1 and HBG2 were sequenced, identifying four unique nucleotides in the 3′-UTR between positions +3 and +6 (TCAC>CTCT, a polymorphism known with variant ID: rs386750130) (Figure 1C). Of note, this 4-nucleotide variant was also found in the HBG1 3′-UTR of HUDEP1 and HEK293T cells. Furthermore, in HUDEP2 cells, the extra A at position +55 was allele-specific rather than gene-specific. Sanger sequencing of cloned 5 kb amplicons that contained the 3′-UTRs of HBG1 and HBG2 showed that on one chromosome, both genes contain the extra A (A+ allele), while on the other chromosome, both genes lack the +55A (A- allele) (Figure 1C). This allele-specific variant allows differentiation between sequence data from either allele when evaluating the knockin of a reporter gene. These +55A variants are listed in dbSNP as rs3841756 (HBG1) and rs34879481 (HBG2). With this sequence information, two gRNAs specifically targeting the HBG1 3′-UTR were designed. gRNA #1 maps to an HBG1-specific PAM (protospacer adjacent motif, 5′-NGG-3′) and contains 3 gene-specific nucleotides in the spacer sequence, while gRNA #2 contains 4 gene-specific nucleotides and directs Cas9 to cut at the Aγ-globin stop codon (Figure 1D).

Tagging the endogenous HBG1 gene in HUDEP cells

The HBG1-specific gRNAs were cloned into an expression plasmid containing Cas9 (p.LentiCRISPRv2 Addgene #52961²⁸) and transfected into HEK293T cells to evaluate their DNA cutting efficiency. Sequence trace deconvolution with the TIDE algorithm²⁶ suggested gRNA #2 was most efficient (1 bp insertion in 12.9% of traces vs. 5.7% for gRNA #1). HBG1 was then targeted in the fetal-like HUDEP1 cells¹⁹ to test the feasibility of CRISPR-Cas9-mediated tagging of HBG1. The Cas9_gRNA #2 plasmid was combined with a double-strand DNA template containing eGFP flanked by 800 bp homology arms (HA, Figure 2A). Although the knockin efficiency was low, indicated by less than 1% GFP+ cells in flow cytometry (Figure 2B,C), the correctly edited cells could be enriched by fluorescence-activated cell sorting (FACS) (Figure 2D,E). DNA sequence analysis of the sorted cells confirmed that the tag landed correctly in the endogenous HBG1 gene (data not shown). For a useful reporter cell that could identify novel inducers of HbF expression, HBG1 must be tagged in HUDEP2 cells that do not express HbF. Without expression of the tagged gene, FACS enrichment of the correctly edited cells would be impossible underscoring the need for a more efficient knock-in strategy. Also, GFP was not considered the optimal reporter gene. Detection of fluorescence using microscopy or flow cytometry would pose significant limitations in terms of sensitivity and sample analysis time, and drug screens would be hampered by the inherent auto-fluorescence of some of the pharmacological compounds, for example, pomalidomide (Figure 2F). In contrast, luciferase-based assays are amenable to fast, specific, and sensitive robotized analysis and are not affected by auto-fluorescence (Figure 2G). We opted for a split NanoLuciferase system³⁶ in which an 11-amino acid tag (HiBiT, High-affinity small BiT) is fused in-frame to HBG1. Upon lysis of the cells, its counterpart Large BiT (LgBiT) is added to form catalytically active luciferase. The luminescent signal intensity provides a direct and sensitive measure for the amount of Aγ-HiBiT protein in the lysate. To optimize editing efficiency, we used a HUDEP2 population in which expression of the HBG1/2 genes had been spontaneously reactivated (TV and SP, unpublished results). First, Cas9 cutting efficiency was improved by delivering Cas9 and the gRNA as a ribonucleic protein (RNP) complex rather than on an expression plasmid (Figure 2H). Second, a single-strand oligodeoxynucleotide (ssODN) template was used to introduce the HiBiT tag in-frame at the C-terminus of Aγ-globin (Figure 2H). Combined, the knockin efficiency increased to over 20% (Figure 2I). Sequence analysis demonstrated correct knockin of the HiBiT-tag at the HBG1 gene. This was confirmed by western blot analysis of Clone #10 in which both HBG1 genes contain the HiBiT tag. Compared to the untagged Gγ-chains, the Aγ-HiBiT chains were detected as a slower migrating band by the γ-globin antibody (Figure 2J). Importantly, when the HiBiT luminescence assay was used to probe the western blots, only the slower migrating Aγ-HiBiT band was detected (Figure 2K). To generate the reporter cell line, this procedure was repeated in HUDEP2 cells that did not express HbF. Here, we report the clonal isolation, molecular characterization, and HTS validation of the HbF reporter cell line.

The Aγ-HiBiT reporter cell line resembles HUDEP2 cells in proliferation and differentiation

Sequence analysis confirmed that the clonal Aγ-HiBiT reporter cell line contained an in-frame knockin of the HiBiT-tag at the C-terminus of HBG1 on the A− allele. On the A+ allele, an insA was observed in the stop codon, leaving the TGA stop codon intact (Figure 3A). This insertion probably prevented further cutting by Cas9 resulting in heterozygous rather than homozygous knockin of the HiBiT-tag. Since the sequence of the HBG2 genes was unaffected (data not shown), 1 out of 4 HBG genes now carried the HiBiT-tag. After genotyping this correctly tagged cell line, its phenotype was characterized and compared to untreated HUDEP2 cells. Untreated HUDEP2 cells are known to have multiple chromosomal gains. Trisomies have been described for chromosomes 6, 8, 17, 18, 19, and 21.^{37, 38} In line with these reports, SNP array analysis of our untreated HUDEP2 cells revealed trisomy for chromosomes 6, 8, 18, 19, and 21. The reporter clone contained an additional copy of chromosome 7, but no chromosomes were lost (Figure 3B,C). A comparison of open chromatin in the HBB locus of HUDEP1, HUDEP2, and the reporter cells with ATACseq showed that the ATAC sites in the reporter cells matched the pattern observed in HUDEP2 cells (Figure 3D). The absence of ATACseq peaks at the HBE, HBG1, and HBG2 genes indicated that these genes remained effectively silenced. RNA sequence traces confirmed that transcriptional activity in the reporter cells is limited to the adult HBD and HBB genes (Figure 3D,E). Expression levels of 52 previously reported regulators of HbF were compared between the reporter cells and control HUDEP2 cells (Supporting Information S1: Table S2). The expression profile of these HbF-regulating factors, including the two key repressors of HbF, BCL11A³⁹ and ZBTB7A,⁴⁰ correlated closely between the two cell lines (Figure 3F). The EIF2AK1 gene, encoding a kinase and known repressor of HbF,⁴¹ is located on chromosome 7. The reporter cell line, with three copies of chromosome 7, showed ~2-fold increased expression of this gene. Two other reported HbF regulators located on chromosome 7, JAZF1^{42, 43} and EZH2,^{44, 45} were not upregulated in the reporter cell line. The expression pattern of hemoglobin subunits in the reporter cell line closely resembles that observed in HUDEP2 cells. Both lines express very low levels of HbF and high levels of HbA. Hemoglobin expression depends on the differentiation state of the culture at the time of RNA isolation, which might explain the minor differences in the absolute number of transcripts picked up by RNAseq (Figure 3E). Upon induction of differentiation, HUDEP2 cell diameter decreases, while the expression of globin chains gradually increases. The parental HUDEP2 and reporter cells responded similarly after the medium was changed to differentiation conditions. Upon differentiation, the red color of the cell pellets increased indicating higher levels of globin expression, while cell diameter decreased (Figure 3G,H). Flow cytometry showed that during differentiation the reporter cells changed expression of cell surface markers similar to differentiating HUDEP2 cells, that is, loss of CD117 and gain of CD235a (Figure 3I). Together these results confirmed that the Aγ-HiBiT reporter cell line closely resembles the adult erythroid progenitor phenotype of HUDEP2 cells.

Application of the Aγ-HiBiT reporter cell line to evaluate genetic HbF-inducing strategies

In genetic experiments, the Aγ-HiBiT reporter cell line was evaluated as a tool to quickly evaluate different genetic HbF-inducing strategies. This requires that the reporter cells survive gene editing experiments. Also, any increase in detected Aγ-HiBiT signal should correlate quantitatively with the extent of HbF induction as measured by standard methods, such as western blotting or HPLC. Using Cas9 guided to the promoters of HBG1 and HBG2 by a gRNA previously published by Traxler et al.,⁴⁶ a 13 bp deletion is preferentially introduced. This deletion is a known HPFH mutation.⁴⁷ In the reporter cells, the HPFH mutation was successfully introduced in ~27% (estimate based on sequence trace deconvolution) of the HBG1/2 promoters (Figure 4A). After editing the HBG1/2 promoters in HUDEP2 and the reporter cells, γ-globin protein levels increased in both cell types (Figure 4B). Western blots on lysates from the reporter cells demonstrated induction of HiBiT-tagged and untagged γ-globin chains. The lower intensity of the slower migrating Aγ-HiBiT chain fits with the notion that the HiBiT tag was introduced in one of the four HBG genes (Figures 3A and 4B). Untreated reporter cells contained 2.3% HbF, which increased to 23.1% in the population after gene editing, as was quantified by HPLC (Figure 4C,D). Similarly, in the HUDEP2 population, the CRISPR experiment increased the HbF levels to 19.6% (Figure 4E). As an alternative genetic approach, the reporter cells were subjected to editing outside the HBB locus with the disruption of an erythroid-specific enhancer of BCL11A, a key repressor of HbF.⁴⁸ Also, for this orthogonal strategy, sequence disruption and HbF induction were observed (Figure 4F,G). Next, we tested if the induction of HbF was reflected by the luminescent reporter signal. The HiBiT Lytic Assay showed a sharp increase in luminescence signal from the reporter cells after gene editing of the HBG promoters (gRNA-1, Figure 4H) and the BCL11A enhancer (gRNA-1617, Figure 4H).

Pomalidomide as a positive control compound for HbF induction

To use the reporter cell line for the discovery of γ-globin-inducing compounds, a compound that can serve as a positive control is required. We used pomalidomide, a thalidomide derivative that was previously described to have HbF-inducing properties.^49-52 Primary erythroid progenitors are known to express relatively high levels of HbF when cultured.²⁰ Indeed, we observed a baseline level of ~2% HBG1/2 as a fraction of HBG1/2+HBB messenger RNA (mRNA) in primary erythroid progenitors obtained from a healthy donor (Figure 5A). Under proliferation conditions, this increased to 12%–15% at Day 5 of treatment with 10 µM pomalidomide, and this ratio was maintained when pomalidomide treatment was continued under differentiation conditions (Figure 5A). Western blot analysis confirmed these observations on γ-globin expression upon pomalidomide treatment (Figure 5B). HPLC analysis of cells harvested at the end of the experiment (Day 10) showed that the HbF level of ~3% in untreated cells (DMSO solvent control) had increased to ~12% upon pomalidomide treatment. In primary erythroid cells obtained from an SCD patient, we observed a baseline level of ~10% HBG1/2 as a fraction of HBG1/2+HBB mRNA (Figure 5C). This increased to 15%–20% at Day 5 of treatment with 10 µM pomalidomide under proliferation conditions, further increasing to 30%–40% when pomalidomide treatment was continued under differentiation conditions (Figure 5C). Western blot analysis confirmed these observations on γ-globin expression upon pomalidomide treatment (Figure 5D). In HUDEP2 cells, baseline expression of HbF is low.^{19, 29} Following the schedule used for the primary cells (Figure 5A–D), we treated HUDEP2 and the reporter cells with 10 µM pomalidomide and analyzed globin expression by quantitative reverse-transcription polymerase chain reaction (RT-qPCR) (Figure 5E). Increased γ-globin expression was detectable at Day 3 and Day 5, and a further increase was observed when pomalidomide treatment was continued under differentiation conditions, reaching a maximum of 1%–5% at Day 10 (Figure 5E). HPLC analysis of cells harvested at the end of the experiment (Day 10) showed an HbF level of ~1%–2% in untreated cells and of ~3%–5% in pomalidomide-treated cells (Figure 5F). Detection of Aγ-HiBiT with the HiBiT Lytic Assay in the reporter cell line showed a significantly increased signal at Day 3 (3.5-fold, Figure 5G), which increased further until Day 10 (5.8-fold, Figure 5G). Western blot analysis of the parental HUDEP2 cells showed increased expression of β-globin during culture under differentiation conditions, while γ-globin was barely detectable even at Day 10 of pomalidomide treatment (Figure 5H). Similarly, increased expression of β-globin during culture under differentiation conditions was observed for the reporter cells (Figure 5I), while γ-globin expression could not be detected with the γ-globin antibody (data not shown). When the blot was probed with the HiBiT luminescence assay, increased expression of Aγ-HiBiT was already detectable at Day 3 of pomalidomide treatment and further increased at Days 5–10 (Figure 5I). Collectively, we conclude that HUDEP2 cells and the reporter cell line display very similar patterns of pomalidomide-mediated induction of γ-globin. These results also strongly support the specificity and sensitivity of Aγ-HiBiT detection by the lytic assay (Figure 5G) and on western blots (Figure 5I). Next, we treated reporter cells with a range of pomalidomide concentrations for 3 days in proliferation conditions. ATP quantification as a measurement of cell viability indicated no severe toxicity up to a final concentration of 50 μM pomalidomide (Figure 5J), and we observed a maximum 6.9-fold increase of Aγ-HiBiT signal with an EC₅₀ around 0.9 µM pomalidomide (Figure 5K). To determine if pomalidomide activates HbF specifically or if it is promoting differentiation with an indirect effect on all globin expression levels, we next tested pomalidomide treatment under differentiation conditions. As expected, the Aγ-HiBiT signals increased upon induction of differentiation for 3 days. Compared to DMSO-treated cells, the addition of 50 μM pomalidomide resulted in a further 7.7-fold increase of the Aγ-HiBiT signal (Figure 5L). Thus, pomalidomide also increased γ-globin expression under differentiation conditions. Importantly, treatment with an unrelated compound that displayed high toxicity (CC₅₀ = 1.7 µM, Figure 5M) did not increase the Aγ-HiBiT signal (Figure 5N). This indicated that cytotoxic stress did not reactivate HbF in the reporter cells, an important prerequisite for use in HTS applications. Finally, we tested the sensitivity of the reporter cells to the solvent DMSO. To our surprise, DMSO concentrations above 0.2% affected cell viability (CC₅₀ = 0.55%, Figure 5O). For compound screening, final DMSO concentrations should therefore be kept below 0.1%.

The reporter cell line is HTS compatible

Having established that the Aγ-HiBiT reporter cells resemble the parental HUDEP2 cells and that pomalidomide can serve as a robust positive control for HbF induction, we aimed to validate the assay for high-throughput drug screening. HTS libraries include up to hundreds of thousands of compounds. To make a large screen feasible and cost-effective, these compounds are pre-plated in white opaque 384-well plates that are specially designed for robotized luminescence assays. The reporter cells would be required to survive plating by a liquid handling robot and a predefined incubation period in the individual wells. Subsequent robotic addition of the HiBiT lytic buffer, which also contains LgBiT and the luciferase substrate, followed by automated luminescence read-out would enable HTS. The pomalidomide experiments suggested that 3–4 days of incubation are required before measuring HbF induction. Longer incubation times require a medium change, which is undesirable in the HTS setting as this introduces a source of uncontrollable variability. To develop the HTS application, daily viability assays were performed after plating the reporter cells to 384-well plates in concentrations ranging from 0 to 50,000 cells in 50 µL medium per well. This showed that cultures starting with 10,000–15,000 cells per well displayed a linear expansion for up to 3 days (Figure 6A). Therefore, 12,500 cells per well in 50 µL medium were incubated for 3 days in 384-well plates pre-plated with pomalidomide dissolved in DMSO at concentrations ranging from 20 µM to 1 nM (10 concentrations, threefold serial dilutions) (Figure 6B). These initial tests in the 384-well format showed that further optimization was required. First, the maximum Aγ-HiBiT induction by pomalidomide, deduced from fully automated luminescence read-out after the addition of 50 µL lytic buffer per well, was lower than the ~6–7-fold observed in 96-well plates. Second, with 0.1% DMSO, some outlier wells occurred (6/180, Figure 6B). Third, with 100 µL total volume per well after the addition of lytic buffer, the airflow created by the automated plate reader caused the transfer of liquid between wells. In addition, high-speed injection of the lytic buffer created air bubbles. These issues were addressed by adding a reduced volume (25 µL) of lytic buffer at a lower speed and incubating the mixture for 10 min before read-out by the automated plate reader. The rapid expansion of the reporter cells is another prerequisite for HTS applications. In the fully defined serum-free proliferation culture medium, the reporter cells expand ~5-fold every 3 days. We prepared stocks of 20 million cells per vial stored in liquid nitrogen. For large experiments, we typically thaw 60 million cells and plate them at a density of 1 million cells/mL. The cells recover within 24 h after which they can be plated at 0.3 million cells/mL and expanded. Starting with 60 million cells, over 0.5 billion cells can be grown within a week, a number sufficient to perform ~40,000 assays in the 384-well format.

A pilot screen of an annotated drug library identifies a known HbF inducer

As a pilot screen, a collection of 5632 known bioactive/medicinal compounds was screened at 10 µM final concentration using the Aγ-HiBiT reporter cell line (Figure 6C). Pomalidomide was used as a positive control and to normalize for variation between the plates. Compared to the DMSO controls, the majority of the tested compounds resulted in a lower luminescence signal; this was interpreted as a sign of toxicity. The Aγ-HiBiT signal was detectable in the untreated reporter cells and a decrease in signal, therefore, indicated that the cells failed to survive the 3-day incubation period. Five compounds (0.089%) showed a >20-fold induction of the Aγ-HiBiT signal, while none of the DMSO-only wells exceeded this threshold (Figure 6C). In addition to the positive control wells, pomalidomide was also present in the screened collection, where it showed the expected sixfold induction. Furthermore, the library contained compounds targeting known HbF regulators. Three inhibitors of histone methyltransferases EHMT1/EHMT2,^53-56 UNC0224, UNC0321, and A-366, displayed 4.9-, 4.7-, and 3.8-fold Aγ-HiBiT induction, respectively. The histone deacetylase (see Migliaccio et al.⁵⁷ for review) inhibitor Bufexamac showed a 2.7-fold induction of the Aγ-HiBiT signal. The modest response to hydroxyurea (1.6-fold induction) is in agreement with previously reported data.⁵⁸ We then tested serial dilutions of the top hits ML 113 HCl, 5-chloro-2′-deoxycytidine, 5-bromo-2′-deoxyuridine (5-BRDU), 5-aza-2′-deoxycytidine (Decitabine), and Nociceptin (Fosfestrol was not available for testing). ML 113 HCl did not induce Aγ-HiBiT expression and affected cell viability (CC₅₀ = 8.9 µM, Figure 6D). Strong Aγ-HiBiT induction was observed upon treatment with 5-chloro-2′-deoxycytidine (EC₅₀ = 0.21 µM) and 5-BRDU (EC₅₀ = 0.14 µM); 5-BRDU displayed a modest impact on cell viability over the concentration range tested (Figure 6D). The addition of Nociceptin, a 17-amino acid neuropeptide, led to sharply increased signals in the HiBiT Lytic Assay at the high end of the concentration range without affecting cell viability (Figure 6E). Since we found that Nociceptin alone activates the HiBiT Lytic Assay (Figure 6E), we conclude that this is a false-positive hit. The top hit, with an Aγ-HiBiT induction over 40-fold, was decitabine (Figure 6C), a well-known HbF inducer that has been tested in clinical trials.^{50, 59, 60} Finding a known activator is a promising outcome of the pilot screen that contributes to the HTS validation of the reporter cell line. Much like typical HTS projects, in this pilot screen, all compounds were tested once at a fixed concentration of 10 μM. In larger screens identifying clusters of active compounds that share similarities in their chemical structures can help prioritize hits that are most likely true positives. In our pilot screen, a cluster of nucleotide analogs including decitabine all showed an induction of Aγ-HiBiT signal outperforming pomalidomide (Figure 6C,D). Finding such a cluster is promising and helps prioritize hit compounds selected for validation experiments. Treatment of the Aγ-HiBiT reporter cells with serially diluted decitabine in a 96-well format confirmed decitabine as a strong HbF-inducing agent (EC₅₀ = 0.09 µM, Figure 6F). Decitabine displayed a modest impact on cell viability over the concentration range tested (Figure 6G). We conclude that the reporter cell line is capable of identifying clusters of HbF-inducing compounds. Most importantly, the pilot screen validates the Aγ-HiBiT reporter cell line as a robust tool for the unbiased identification of novel HbF-inducing compounds using the HTS approach.