Probing the requirement for CD38 in retinoic acid-induced HL-60 cell differentiation with a small molecule dimerizer and genetic knockout

CD38 is an ectoenzyme and receptor with key physiological roles. It metabolizes NAD+ to adenosine diphosphate ribose (ADPR) and cyclic ADPR, regulating several processes including calcium signalling. CD38 is both a positive and negative prognostic indicator in leukaemia. In all-trans retinoic acid (RA)-induced differentiation of acute promyelocytic leukaemia and HL-60 cells, CD38 is one of the earliest and most prominently upregulated proteins known. CD38 overexpression enhances differentiation, while morpholino- and siRNA-induced knockdown diminishes it. CD38, via Src family kinases and adapters, interacts with a MAPK signalling axis that propels differentiation. Motivated by evidence suggesting the importance of CD38, we sought to determine whether it functions via dimerization. We created a linker based on the suicide substrate arabinosyl-2′-fluoro-2′-deoxy NAD+ (F-araNAD+), dimeric F-araNAD+, to induce homodimerization. CD38 homodimerization did not affect RA-induced differentiation. Probing the importance of CD38 further, we created HL-60 cell lines with CRISPR/Cas9-mediated CD38 truncations. Deletion of its enzymatic domain did not affect differentiation. Apart from increased RA-induced CD11b expression, ablation of all but the first six amino acids of CD38 affected neither RA-induced differentiation nor associated signalling. Although we cannot discount the importance of this peptide, our study indicates that CD38 is not necessary for RA-induced differentiation.

We then used FRET 36 to detect CD38 dimerization in HL-60 cells. We directly conjugated monoclonal antibodies against CD38 with either Alexa Fluor 488 or 594 and, after determining that treatment with either F-araNAD + or dF-araNAD + did not alter CD38 expression (Fig. 2B), we performed the FRET measurement via flow cytometry (Fig. 2C). Untreated samples displayed background levels of fluorescence in the FRET channel. Samples with no CD38 antibodies are included to show that F-araNAD + and dF-araNAD + do not have background fluorescence. RA treatment induced CD38 expression and the acceptor fluorescence mean increased. When F-araNAD + was added, there was no effect on the acceptor fluorescence. When dF-araNAD + was added, however, the acceptor fluorescence mean increased. Therefore, in the presence of dF-araNAD + , the donor and acceptor fluors were closer, supporting that dF-araNAD + increased the dimerization of CD38. Thus, while some CD38 homodimerization may be present after RA treatment alone, treatment with dF-araNAD + increases the . HL-60 cells were cultured for 24 h with 1 μM RA as indicated. At 23 h, cells were treated with 1 μM F-araNAD + (+) or dF-araNAD + (D) as indicated. 1 × 10 6 cells were put in PBS with Alexa Fluor 594-conjugated CD38 antibody as indicated and analysed by flow cytometry. Error bars indicate standard error of the mean (SEM). (C) FRET means (n = 3). HL-60 cells were cultured for 24 h with 1 μM RA as indicated. At 23 h, cells were treated with 1 μM F-araNAD + (+) or dF-araNAD + (D) as indicated. 1 × 10 6 cells were put in PBS with Alexa Fluor 488-and 594-conjugated CD38 antibodies as indicated and analysed by flow cytometry. Error bars indicate SEM. ***p < 0.001; ****p < 0.0001. Two-way analysis of variance (ANOVA) using Tukey's multiple comparisons test was used to determine significance. (D) Western blot showing covalent dimerization of CD38 when treated with dF-araNAD + . HL-60 cells were cultured with RA and treated with F-araNAD + or dF-araNAD + at 8 or 23 h as indicated and lysate was collected at 24 h. 25 μg of lysate per lane was run. Membrane images for each protein are cropped to show only the bands of interest. degree of CD38 homodimerization. This indicates that not all available CD38 was homodimerized before addition of dF-araNAD + .
Since dF-araNAD + functions via covalent interactions, we also assessed and confirmed CD38 dimerization in HL-60 cells via western blot 32,33,35 . HL-60 cells that had been treated with RA to induce CD38 expression were treated with dF-araNAD + at 8 h or 23 h and harvested at 24 h. In both cases, a band corresponding to twice the molecular weight of CD38 was present in the samples treated with RA and dF-araNAD + ( Fig. 2A and D). These results were consistent with those of the purified CD38 and FRET experiments, indicating that dF-araNAD + induced dimerization of CD38. dF-araNAD + does not affect RA-induced differentiation. In order to assess the effects of CD38 dimerization on RA-induced differentiation, we analysed several myeloid lineage differentiation markers over the course of a 72 h treatment. Since a previous report demonstrated that CD38 catalytic activity has no influence on RA-induced differentiation, we were not concerned about blocking the active site 30 . We measured CD38 surface expression at 24 h and cell density (i.e., population growth), CD11b surface expression, and G 1 /G 0 cell cycle arrest every 24 h. HL-60 cell cultures were initiated with or without RA at 0 h, and F-araNAD + or dF-araNAD + was added to flasks as indicated at 8 h or 23 h when CD38 expression was significantly induced. This represented testing for both early and late potential dimerization effects.
Expression levels of the differentiation markers varied as a function of RA, but not F-araNAD + or dF-araNAD + , treatment (Fig. 3). As expected, with RA treatment, CD11b and CD38 expression increased, the proportion of cells in G 1 /G 0 increased, and cell population growth was retarded compared to control cells not treated with RA. Addition of F-araNAD + or dF-araNAD + at either time point, however, had no effect on the induced expression of any of these markers.
Characterization of CRISPR/Cas9-mediated CD38 disruption. Using the CRISPR/Cas9 system, we generated three CD38-targeted cell lines, CRISPR 1, 2, and 3, to test whether CD38 is required for RA-induced differentiation. Using the E-CRISP design tools, we targeted sites as early as possible along the gene to perform knockouts 37 .
To ensure successful disruption, CD38 levels were measured by both flow cytometry and by western blot following 48 h treatment with RA. We detected no CD38 expression in any of the pooled cell lines (Figs 4B and 5A). We further characterized CD38 expression in the CRISPR-derived cell lines using RT-PCR (Fig. 4A). Comparing wild-type CD38 with the truncated forms, CRISPR 1 and 3 both lack nearly the entire extracellular domain, but retain the cytoplasmic and transmembrane domains (Fig. 4A). CRISPR 1 retains the N-terminal 49 amino acids before a substitution at position 50 and a premature stop codon. CRISPR 3 has a CD38 product of 62 amino acids before a premature stop codon, with the first 45 amino acids of wild-type CD38 conserved (Fig. 4A.). CRISPR 2 is a mixed population, as we saw two distinct bands after performing RT-PCR. Sequencing both bands, however, revealed that the CD38 truncations were similar -both lacked the vast majority of the protein, with only the first six or seven amino acids of CD38 left (Fig. 4A.). Both had premature stop codons after 15 amino acids (Fig. 4A).
The extracellular domain of CD38 contains its enzymatically active domain, and since we largely deleted this, we anticipated a loss of RA-induced enzymatic activity with the CRISPR mutants. We confirmed this using the nicotinamide guanine dinucleotide (NGD + ) assay (Fig. 4C), which detects ADP-ribosyl cyclase activity. While CD38 is a major source of this activity, CD157 also possesses this catalytic activity 38 . HL-60 cells normally express both CD38 and CD157 at very low levels, but they are upregulated following RA treatment. We calculated the fold increase in cyclic GDP ribose (cGDPR) production from RA-treated to untreated cells for each mutant line and normalized to wild-type HL-60 levels. In each of the CRISPR-derived lines, RA-induced production of cGDPR was significantly reduced, further confirming the disruption of CD38. Wildtype and CRISPR cell lines were cultured for 48 h with 1 μM RA as indicated to stimulate CD38 expression and evaluated for cGDPR production via an NGD + assay (n = 4). The difference in fold increase between RA treated and untreated samples was taken for each cell line and normalized to a control value of 1 in wild-type cells. Error bars indicate SEM. *p < 0.05; **p < 0.01. One-way ANOVA using Tukey's multiple comparisons test was used to determine significance. The raw data NGD + catabolism curves are shown in Supplementary Information.
SCIentIfIC RepoRtS | 7: 17406 | DOI:10.1038/s41598-017-17720-4 CD38 disruption does not affect RA-induced differentiation. We evaluated the effects of CD38 disruption on RA-induced differentiation as we did with induced CD38 dimerization. We measured cell density, CD11b, CD38, and G 1 /G 0 cell cycle arrest over the course of a 72 h RA treatment. For both wild-type and CRISPR-derived cell lines, RA treatment induced growth retardation, upregulation of CD38 and CD11b expression, and G 1 /G 0 accumulation (Fig. 5). The responses of the CRISPR-derived knockouts were not significantly different from wild-type cells except for the CRISPR 2 cell line, which displayed increased levels of CD11b at 48 and 72 h after RA treatment (Fig. 5B).
CD38 disruption does not affect prominent signalling proteins associated with RA-induced differentiation. Since an ensemble of signalling molecules propels differentiation, we determined whether any RA-response associated signalling proteins were affected by loss of the majority of CD38. c-Cbl, Lyn, Slp-76, and Vav1, which have been shown to interact with CD38, were of particular interest [19][20][21] . We treated wild-type HL-60s, as well as the CRISPR-derived cell lines, with or without RA for 48 h, collected lysates, and analysed them by western blot (Fig. 6). For all the proteins we analysed, we observed that the RA-induced effects were similar for wild-type and CRISPR-derived cells (Fig. 6). This corroborates the previously described results for phenotypic shift of the cells. Levels of the signalling proteins Lyn, c-Cbl, Slp-76, Vav1, and phospho-MEK increased similarly with RA treatment for wild-type and CRISPR-derived lines (Fig. 6). RA treatment induced expression of a component of the NAD + phosphatase oxidase complex, p47 phox, in all cell lines as well (Fig. 6). MEK and ERK levels remained stable (Fig. 6).

Discussion
While a previous report indicated that CD38 is found as homodimers on the cell surface of murine B lymphocytes, our results suggest that dF-araNAD + is able to induce an additional degree of CD38 homodimerization in HL-60 cells 31 . This increased CD38 homodimerization did not affect RA-induced differentiation. We must note, however, that there are several alternative explanations for why dF-araNAD + did not affect RA-induced differentiation. It is possible that the induced CD38 dimer is not in the correct orientation to initiate downstream signalling or that high-order oligomerization other than dimerization is needed.
Additionally, we do not know whether a particular degree of dimerization is required for signalling. One of the best studied transmembrane receptors, epidermal growth factor receptor (EGFR), is typically found in a 3:7 ratio of dimer to monomer under resting conditions and a 1:1 ratio when its cognate ligand, EGF, is bound 39 . The western blot of relative abundance of CD38 monomer and dimers, although not quantitative, appears consistent with this range of receptor dimerization. In vitro dimerization of EGFR with EGF and activation of EGFR tyrosine kinase activity are coincident events, but transmembrane receptor dimer formation and downstream signalling are not always coupled [40][41][42] . It is experimentally intractable to test every possibility such as the dimer to monomer ratios, relation between dimer formation and downstream signalling, required orientation, and potential high-order oligomerization for CD38. Thus, we took a simpler alternative approach to study the role of CD38 in RA-induced differentiation: We genetically disrupted endogenous CD38 using the CRISPR/Cas9 system to probe for effects of losing the protein.
Results from the CRISPR/Cas9-mediated CD38 truncations indicate that the extracellular domain of CD38 is not necessary for RA-induced differentiation. CRISPR 1 and CRISPR 3 possess only seven and three amino acids, respectively, of the extracellular domain of CD38, while CRISPR 2 has none, yet the phenotypic response to RA treatment is largely unaltered. Because the enzymatic activity of CD38 is dependent on the extracellular domain, specifically E226, we also conclude that this enzymatic activity is not necessary for RA-induced differentiation, consistent with previously reported results 30 .
CRISPR 2 is a mixed population expressing two different short peptides of CD38, both of which are 15 amino acids long, with the first six or seven amino acids conserved. Phenotypic markers for RA-induced differentiation in the CRISPR 2 line were equal to or, in the case of CD11b, better than wild-type HL-60 cells. The transmembrane portion and much of the cytoplasmic domain are thus also not required for RA-induced differentiation. We cannot rule out the importance of the remaining short peptide; however, the weight of the evidence suggests that CD38 is not necessary for RA-induced myeloid differentiation. This conclusion can be integrated with that of an earlier report showing that siRNA-and morpholino-mediated knockdown of CD38 compromised RA-induced differentiation of HL-60 cells 18 . That report tested differentiation using only one assay, nitroblue tetrazolium reduction, a marker for inducible reactive oxidative metabolism. Occurrence of mature myeloid cells is commonly measured by either the CD11b cell surface marker or by functional differentiation evidenced by 12-O-tetradecanoylphorbol-13-acetate-inducible superoxide production. It is also betrayed by occurrence of G 1 / G 0 arrest. While Munshi et al. measured superoxide production, which is a functional marker for mature cells, we measured CD11b, which is a widely accepted surface marker for mature cells, as well as G 0 arrest and expression of p47 phox, a signature component of the superoxide production machinery. Evidenced by the cell surface and cell cycle indicators, there was cell maturation and cell cycle arrest following treatment of the CRISPR-derived cell lines with RA. Induced expression of p47 phox was also not compromised. Synthesizing our data and theirs, it is possible that while some aspect of the NADPH machinery generating superoxide is CD38 dependent, the expression of the mature cell surface marker is not, nor is cell cycle arrest and nor is the induced expression of p47 phox. It is potentially also noteworthy that crippling CD38 may affect NAD + metabolism so that it adversely affects the function of NADPH machine and ergo superoxide production. Without being able to rule this out, superoxide production may not be a reliable marker of differentiation when NAD + metabolism is subject to manipulation. Hence, unless a 6-amino acid fragment of the cytoplasmic tail can account for any differentiation-driving action of the entire CD38 receptor, it appears that CD38 is not necessary for the signalling or differentiation attributed to RA.
In summary, our data show that despite the evidence suggesting that CD38 is a driver of RA-induced differentiation, neither induction of dimerization nor deletion of the vast majority of the protein affected the differentiation process. It appears that CD38 is ergo not necessary for RA to induce differentiation, even though enhancing its expression can enhance signalling and differentiation. A full understanding of its function remains to be determined. Given that CD38 is upregulated so prominently and so early in response to RA treatment, it is possible that it may play an important role in the function of differentiated cells. HL-60 cells differentiate into neutrophil-like cells after RA treatment. Consistent with this, CD38 knockout mice are known to have immune defects 43 . However, exactly how CD38 affects the function of immune cells remains to be elucidated.

Direct conjugation of Alexa Fluors to primary antibodies. Conjugation of Alexa Fluor succinim-
idyl esters to CD38 primary antibodies was performed using the manufacturer's protocol. Briefly, 30 µL 0.75 M sodium bicarbonate, pH 8.3, was added to 200 µL CD38 antibody. 1 mg of Alexa Fluor was dissolved in 100 µL DMSO, and 20 µL Alexa Fluor 488 or 594 was added to one vial of primary antibody while vortexing. The tubes were then shaken at room temperature for 1 h at 200 RPM. 20 µL 1.5 M hydroxylamine, pH 8.5, was then added to each tube and they were shaken for an additional 1 h. Finally, the conjugated antibodies were dialyzed in 2 L of PBS in the dark at room temperature for 2 h. Western blot analysis. Cells were pelleted, washed twice with PBS, and lysed with ice cold M-PER Mammalian Protein Extraction Reagent (Pierce, Rockford, IL) with protease and phosphatase inhibitor cocktails (Sigma, St. Louis, MO). Samples were incubated overnight at −80 °C and debris was pelleted. Protein concentration was determined using the Pierce BCA Protein Assay according to the manufacturer's protocol. Lysate was subjected to standard SDS-PAGE, using 25 µg of lysate per lane under denaturing conditions. Membranes were blocked with 5% dry nonfat milk in PBS and probed with antibodies described above. Images were captured on a Bio-Rad VersaDoc MP 5000 and analysed with Quantity One software (Hercules, CA).
Lentiviral particles were produced using 2.5 µg pMD2.g (Addgene #12259), 7.5 µg psPAX2 (Addgene #12260), and 10 µg pLentiCRISPR v2 with CRISPR 1, 2, or 3 inserted. These were co-transfected into HEK 293 T cells at roughly 50% confluence in 10 cm cell culture dishes with DMEM and 10% FBS using TransIT-LT1 transfection reagent (Mirus, Madison, WI) according to the manufacturer's protocol in a 3:2 ratio of reagent (µL) to plasmid (µg). After 48 h, media containing viral particles was collected and 5 mL of additional media was added to the dishes for 24 h until final collection. The 15 mL total viral media was concentrated using Amicon Ultra (Millipore, Billerica, MA) centrifugal filters with 30 kDa MWCO. Concentrated viral media was stored at −80 °C until use.
Transduction of HL-60 cells with the lentiviral particles was performed in 6-well plates. 100 µL concentrated viral particles was added to 5 × 10 4 cells in 1 mL RPMI 1640 with 5% heat-inactivated FBS. After 72 h, transduced cells were put into 10 mL flasks and cultured in RPMI 1640 with 5% FBS and selected for 3 weeks in 400 ng/ mL puromycin. The cell lines were evaluated for CD38 expression following 24 h treatment with 1 µM RA. In all cases, while the majority did not upregulate CD38 in response to RA treatment, there was a population of 15-30% that expressed CD38 levels comparable to wild-type HL-60 cells. Thus, a single cell sort into a 96-well plate was performed on each cell line using a Becton Dickinson FACS Aria III SORP. Upon confluence, each well was transferred to a 24-well plate and RA-induced CD38 expression was assessed as before. Several wells per cell line which did not express CD38 were pooled and designated CRISPR 1, CRISPR 2, and CRISPR 3 according to the sgRNA used. mRNA isolation and sequencing. Total RNA was extracted from CRISPR 1, CRISPR 2, and CRISPR 3 cell lines using a Qiagen RNeasy Plus Mini kit following the manufacturer's protocol (Hilden, Germany). RT-PCR was performed using Invitrogen SuperScript III One Step RT-PCR with Platinum Taq (Invitrogen, Carlsbad, CA) on a PTC-100 thermocycler from MJ Research (Waltham, MA). Forward (5′-AGAAGGGGAGGTGCAGTTTC-3′) and reverse (5′-TGTTGCAAGGTACGGTCTGA-3′) primers for CD38 were purchased from IDT (Coralville, IA). Products were separated on a 1.5% acrylamide gel and appropriate bands were excised and purified with a Qiagen QIAquick Gel Extraction Kit according to the manufacturer's protocol. Because the lines were established from several populations, there were two bands corresponding to slightly different CD38 mutants in the CRISPR 2 line, denoted upper and lower. DNA samples were sequenced at Cornell University's Life Sciences Core Laboratory. DNA sequences were translated to amino acid sequences using ExPASY Translate and compared to the sequence of wild-type CD38.
Detection of CD38 enzymatic activity. ADP-ribosyl cyclase activity was detected by fluorometric analysis of the NGD + metabolic product cyclic GDP ribose (cGDPR) using a spectrofluorometer (Horiba Jobin Yvon FluoroMax3, NJ) as previously described 30 . Cells were cultured for 48 h treated with or without 1 µM RA to induce CD38 expression. 1 × 10 6 cells were pelleted, washed twice with PBS, resuspended in 0.7 mL PBS, and treated with 100 µM NGD + for 1 h with constant rotation at 37 °C. Cells were then pelleted and the supernatant was analysed for cGDPR. Readings at 300 nm excitation/410 nm emission were used to measure cGDPR content.
Statistics. Statistics were analysed using GraphPad Prism version 7.02. One-way and two-way analysis of variance, as appropriate, with Tukey's multiple comparisons test was used to determine significance. Error bars indicate mean ± standard error of the mean (SEM). Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.