Inhibition of UBE2L6 attenuates ISGylation and impedes ATRA-induced differentiation of leukemic cells.

Ubiquitin-conjugating enzyme E2L6 (UBE2L6) is a critical enzyme in ISGylation, a post-translational protein modification that conjugates the ubiquitin-like modifier, interferon-stimulated gene 15 (ISG15), to target substrates. Previous gene expression studies in acute promyelocytic leukemia (APL) cells showed that all-trans-retinoic acid (ATRA) altered the expression of many genes, including UBE2L6 (200-fold) and other members of the ISGylation pathway. Through gene expression analyses in a cohort of 98 acute myeloid leukemia (AML) patient samples and in primary neutrophils from healthy donors, we found that UBE2L6 gene expression is reduced in primary AML cells compared to normal mature granulocytes. To assess if UBE2L6 expression is important for leukemic cell differentiation - two cell line models were employed. The human APL cell line NB4 and its ATRA-resistant NB4R counterpart, as well as the ATRA sensitive human AML HL60 cells along with their ATRA-resistant subclone - HL60R. ATRA strongly induced UBE2L6 in NB4 APL cells and in ATRA-sensitive HL60 AML cells, but not in the ATRA-resistant NB4R and HL60R cells. Furthermore, short hairpin (sh)RNA-mediated UBE2L6 depletion in NB4 cells impeded ATRA-mediated differentiation, suggesting a functional role for UBE2L6 in leukemic cell differentiation. In addition, ATRA induced ISG15 gene expression in NB4 APL cells, leading to increased levels of both free ISG15 protein and ISG15 conjugates. UBE2L6 depletion attenuated ATRA-induced ISG15 conjugation. Knockdown of ISG15 in NB4 APL cells inhibited ISGylation and also attenuated ATRA-induced differentiation. In summary, we demonstrate the functional importance of UBE2L6 in ATRA induced neutrophil differentiation of APL cells and propose that this may be mediated by its catalytic role in ISGylation.

Ubiquitin/ISG15-conjugating E2L6 (UBE2L6) is a critical enzyme in ISGylation, a post-translational protein modification that conjugates the ubiquitin-like modifier, interferon-stimulated gene 15 (ISG15), to target substrates. Previous gene expression studies in acute promyelocytic leukemia (APL) cells showed that all-trans-retinoic acid (ATRA) altered the expression of many genes, including UBE2L6 (200-fold) and other members of the ISGylation pathway. Through gene expression analyses in a cohort of 98 acute myeloid leukemia (AML) patient samples and in primary neutrophils from healthy donors, we found that UBE2L6 gene expression is reduced in primary AML cells compared with normal mature granulocytes. To assess whether UBE2L6 expression is important for leukemic cell differentiation-two cell line models were employed: the human APL cell line NB4 and its ATRA-resistant NB4R counterpart, as well as the ATRA-sensitive human AML HL60 cells along with their ATRA-resistant subclone-HL60R. ATRA strongly induced UBE2L6 in NB4 APL cells and in ATRA-sensitive HL60 AML cells, but not in the ATRA-resistant NB4R and HL60R cells. Furthermore, short hairpin (sh)RNA-mediated UBE2L6 depletion in NB4 cells impeded ATRA-mediated differentiation, suggesting a functional role for UBE2L6 in leukemic cell differentiation. In addition, ATRA induced ISG15 gene expression in NB4 APL cells, leading to increased levels of both free ISG15 protein and ISG15 conjugates. UBE2L6 depletion attenuated ATRA-induced ISG15 conjugation. Knockdown of ISG15 in NB4 APL cells inhibited ISGylation and also attenuated ATRA-induced differentiation. In summary, we demonstrate the functional importance of UBE2L6 in ATRA-induced neutrophil differentiation of APL cells and propose that this may be mediated by its catalytic role in ISGylation.

Introduction
Acute myeloid leukemia (AML) is a clonal disorder characterized by the accumulation of immature hematopoietic precursors in the bone marrow and peripheral circulation (Caceres-Cortes, 2013). Overall survival is poor, particularly for patients over 60 years of age, who have an overall 5-year survival of~10% (Kantarjian and O'Brien, 2010;Thein et al., 2013). Improved therapeutic strategies with tolerable toxicity profiles are needed.
Acute promyelocytic leukemia (APL) is a clinically, pathologically, and molecularly distinct subtype of AML (Tallman and Altman, 2009). It is distinguished in 95% of cases by a translocation of chromosomes 15 and 17, which leads to the expression of a fusion oncoprotein PML-retinoic acid receptor alpha (RARa) (Tallman and Altman, 2009). This protein disrupts functional retinoid signaling in APL cells, repressing gene transcription and halting myeloid maturation at the promyelocyte stage (Tang and Gudas, 2011). Therapeutic doses of all-trans-retinoic acid (ATRA) reactivate gene transcription and overcome this differentiation block allowing clinical remission (Tang and Gudas, 2011). As we have previously reviewed, ATRA also encourages the degradation of the PML-RARa protein through cooperating pathways including proteasomal degradation and autophagy (Orfali et al., 2014).
Cellular protein activity and stability is regulated by post-translational modification (Krishna and Wold, 1993). One such modification is 'ubiquitination', the reversible addition of ubiquitin protein (8.5kDa) to the lysine residues of target substrates (Komander and Rape, 2012). This is catalyzed by a series of enzymes: (a) Ubiquitin-activating enzymes (E1) use ATP to convert ubiquitin to a high-energy thioester; (b) ubiquitinconjugating enzymes (E2) bind active ubiquitin on their cysteine residues; and (c) ubiquitin-ligase enzymes (E3) interact with E2 enzymes and catalyze the formation of a covalent bond between ubiquitin and its target substrate. E3 ligases regulate substrate specificity. Ubiquitin can also be removed from target proteins through the action of deubiquitinases (Friend et al., 2014). Ubiquitin-like modifiers (UBLs) are proteins that share significant structural and some sequence homology with ubiquitin and modify substrates in a similar enzyme-controlled fashion (Hochstrasser, 2009). While protein ubiquitination is known to alter protein activity or target substrates for degradation by the 26S proteasome, the functional roles of UBL modifications are less well-defined and remain under investigation (Hochstrasser, 2009). We now recognize at least ten UBLs, including small ubiquitin-like modifier, autophagy-related proteins 8 and 12 (autophagy-related protein 8, autophagy-related protein 12), and interferon-stimulated gene 15 (ISG15)-which is an important modification in the present study (Hochstrasser, 2009).
ISG15 expression and ISGylation are induced during erythropoiesis, and primary erythroblasts harvested from ISG15 À/À knockout mice show impaired differentiation in ex vivo culture (Maragno et al., 2011). Transcriptional profiling of human granulopoiesis has shown that ISG15 expression is similarly induced during terminal neutrophil differentiation and a PU.1 binding site has been identified within the ISG15 promoter region (Meraro et al., 2002;Theilgaard-Monch et al., 2005). To date, however, a functional role for ISGylation in granulopoiesis has not been proven.
Our work has found that UBE2L6, the gene encoding the E2 enzyme of ISGylation, is strongly upregulated following ATRA treatment of APL cells. Through a series of short hairpin (sh)RNA knockdown experiments, we have investigated for the first time the functional importance of this enzyme in the ATRA-mediated granulocytic differentiation of APL cells. We report that inhibiting UBE2L6 expression results in reduced ISGylation and impaired APL cell differentiation. Interference with ISG15 expression similarly impedes differentiation. Through improving our understanding of ISGylation and protein PTMs involved in ATRA-mediated differentiation of APL cells, we hope to identify ways of promoting differentiation therapy in other AML subtypes.

Patient study
A cohort of 98 AML patient samples, collected through the HOVON/SAKK (Dutch-Belgian Hematology-Oncology/Swiss Group for Clinical Cancer Research Cooperative Group) protocols 04, 04A, 29, and 42 between 1987 and 2006, were provided by P. Valk and B. Lowenberg. Patient characteristics have been previously outlined (Schlafli et al., 2012). Primary neutrophils from healthy donors were isolated using Polymorphprep (Axis-Shield, Dundee, Scotland). All patients provided written informed consent in accordance with the Declaration of Helsinki.
1lg of RNA was reverse-transcribed using qScript (Quanta Biosciences #95047, Beverly, MA, USA) as per product protocol at a final reaction volume of 20 lL, and the resulting cDNA was diluted 1 : 10 in H 2 0. Subsequent qRT-PCRs were carried out using 2 lL of template together with 19 SYBR Green Supermix (Quanta Biosciences #84091), forward and reverse primers at 0.25 µM and 2.5 lL H 2 0 in a final reaction volume of 15 lL. Reactions were run on a Bio-Rad MyiQ TM (Irvine, CA, USA) Single Color Real-time PCR detection system with each cycle including a 94°C 9 20 s denaturation step, 60°C 9 20 s annealing step, and a 72°C x 30 s extension step. Primer pairs were designed to span distinct exon software to avoid genomic DNA signaling, and gene expression amplicons were validated with sequencing at the Genomics Resources Core Facility, WCMC. Sequences of specific primers were as follows: UBE2L6_F CTGGAAGCCTTGCACCAAGA, UBE2L6_R GAAC ATGAGTTAGGAGGGCCG, ISG15_F GGTGGA CAAATGCGACGAAC, and ISG15_R TCGAAGGT-CAGCCAGAACAG. The transcript levels in biological replicates (n = 3) were normalized to hPRT transcript levels, and relative differences were calculated using the Pfaffl method. Graphical displays and measurements of statistical significance were performed on GRAPHPAD PRISM software (San Diego, CA, USA).

Morphology examination
Cells were cytospun onto glass slides and stained with Rapi-Diff (Braidwood Laboratories 22007, 22008, 22009, London, UK) according to product guidelines. Morphology was examined using an Olympus DP70 digital microscope at 400X magnification (Mason Technology, Cork, Ireland).

UBE2L6 is induced during the neutrophil differentiation of leukemic cells
We have previously examined the gene expression changes induced by ATRA in APL cells by sequencing RNA extracted from NB4 cells treated with 1 lM ATRA for 72 h alongside untreated controls (Orfali et al., 2019). These data showed that ATRA induced a 200-fold increase in UBE2L6 expression. Other members of the ISGylation pathway were also found to be coregulated (RNAseq data reproduced in Table 1). As NB4 cells respond to ATRA by differentiating toward mature neutrophils, this prompted us to question whether UBE2L6 expression is important for leukemic cell differentiation.
We first examined UBE2L6 mRNA expression in 98 primary AML patient samples (M0-M4), six samples of normal CD34 + (HSC) cells, and 24 donated mature granulocyte samples using a TaqMan low-density array. Relative UBE2L6 mRNA levels are shown as differences in Ct values as compared to mRNA levels for the housekeeping genes HMBS and ABL1. Expression was significantly lower in AML patient samples and HSC cells than in granulocytes, suggesting that increased expression may be important for the mature granulocyte phenotype (Mann-Whitney U-test, ****P ≤ 0.0001) (Fig. 1A).
To test this hypothesis further, we treated NB4 cells with ATRA along with their ATRA-resistant counterparts NB4R cells. We measured UBE2L6 expression by quantitative real-time (q)PCR at 72 h, assessing Ct values relative to the housekeeping gene hPRT. Validating our earlier RNA sequencing observations, we detected a 180-fold increase in UBE2L6 expression in differentiating NB4 cells (****P ≤ 0.0001), but only a 0.23-fold difference in NB4R cells (**P = 0.0021) (Fig. 1B). The HL60 cell line (human M2 AML), although it does not carry the PML-RARa oncoprotein, also differentiates down a granulocytic lineage in response to ATRA therapy and can be used as a second model of leukemic cell differentiation. Following 96 h of ATRA treatment, we found a 189-fold increase in UBE2L6 expression in HL60 cells (***P = 0.0003). ATRA-resistant HL60R cells, however, failed to induce UBE2L6 (Fig. 1C).
These results indicate that UBE2L6 is prominently upregulated during leukemic cell differentiation rather than solely on ATRA treatment and that this effect is not restricted to APL cells carrying the PML-RARa fusion oncoprotein. In order to investigate whether UBE2L6 has a functional role in leukemic cell differentiation, we generated UBE2L6 knockdown NB4 cells using a lentiviral delivery system to deliver target-specific shRNA. NB4 cells transduced with a nontargeting shRNA were used as a control (SHC). Functional knockdown was confirmed by detecting reduced UBE2L6 protein levels following ATRA treatment in two knockdown clones: shU-BE2L6_499 and shUBE2L6_1082. Superior knockdown efficiency is evident in shUBE2L6_1082 ( Fig. 2A). ATRA-induced neutrophil differentiation was reduced in UBE2L6 knockdown NB4 clones compared with control cells. At a transcript level, we detected reduced expression of granulocyte colony-stimulating factor receptor (GCSFR), a marker of neutrophil differentiation, by qPCR at 72 h (*P = 0.0123, ***P = 0.0004) (Fig. 2B). At a protein level, surface CD11b expression was reduced in ATRA-treated knockdown cells when analyzed by flow cytometry at 72 h. A direct correlation was observed between UBE2L6 knockdown efficiency and detectable CD11b levels, with both shUBE2L6_499 and shUBE2L6_1082 showing a significant reduction in CD11b (**P = 0.0017 and P = 0.0067, respectively) (Fig. 2C). Morphologically, ATRA-treated control cells displayed characteristic features of granulocytic differentiation with increased cytoplasmic volume and visible nuclear indentation (Fig. 2D lower left panel, arrows). This phenotype was stunted in knockdown clones (Fig. 2D lower middle and right panels). Finally, we assessed functional differentiation using the nitro blue tetrazolium assay, which tests the reducing power of the neutrophil enzyme alkaline phosphatase. We observed and quantified a decreased nitro blue tetrazolium reduction in UBE2L6 knockdown clones (**P = 0.0069, *P = 0.0354), with a Fig. 1. UBE2L6 expression is increased during leukemic cell differentiation. (A) UBE2L6 mRNA levels of primary AML patient samples, normal CD34 + (HSC) cells, and mature granulocytes from healthy donors were quantified using qPCR. The relative DCt expression was calculated by the difference in UBE2L6 expression to the housekeeping genes HMBS and ABL (Mann-Whitney U-test ****P ≤ 0.0001). (B) NB4 and NB4R cells were seeded at 0.2 9 10 5 cells per mL and treated with 1 lM ATRA for 72 h. Successful differentiation was confirmed in NB4 cells by flow cytometric analysis of CD11b expression. ATRA-resistant NB4R cells did not differentiate (data not shown). Total RNA was extracted, and UBE2L6 mRNA expression was quantified by qPCR. Values are given as n-fold induction compared with untreated cells and normalized to housekeeping gene hPRT (n = 3) (t-test ****P ≤ 0.0001, **P ≤ 0.01). (C) HL60 and HL60R cells were treated with 1 lM ATRA for 96 h. Successful HL60 differentiation was confirmed by qPCR measurement of GCSFR expression. ATRA-resistant HL60R cells failed to differentiate (data not shown). Total RNA was extracted, and UBE2L6 expression was quantified by qPCR. Values are given as nfold induction compared with untreated cells and normalized to housekeeping gene HMBS (n = 3) (t-test ***P ≤ 0.001).  direct correlation again seen between knockdown efficiency and functional differentiation (Fig. 2E (i) lower panels and Fig. 2E (ii)).
Together, these results demonstrate that UBE2L6 depletion impedes ATRA-mediated granulocytic differentiation of APL cells and prompted us to question the potential mechanism involved in this differentiation block.

UBE2L6 mediates protein ISGylation in ATRA-treated APL cells
As described earlier, UBE2L6 is an E2 ligase critical in the conjugation of ISG15 to target proteins during ISGylation, a process with an unknown role in leukemic cell differentiation. A search of UBE2L6 human protein interactions on the publicly available STRING database (Search Tool for the Retrieval of Interacting Genes/Proteins) (www.string-db.org) (Franceschini et al., 2013) depicts a high confidence of interaction between UBE2L6 and ISG15, as well as interactions with the E1 and all E3 ligases of the ISGylation pathway (Fig. 3A).
We tested levels of both free and conjugated ISG15 proteins in UBE2L6 knockdown NB4 cells using western blot analysis. We found a prominent induction of free ISG15 protein following ATRA treatment in control cells and in knockdown clones. Notably, conjugated ISG15 was markedly increased after 72 h of ATRA treatment in control cells but was impaired in UBE2L6 knockdown cells, with the most efficient knockdown clone shUBE2L6_1082 having the least amount of visible conjugates (Fig. 3B (i,ii)).

ISG15 is induced during the neutrophil differentiation of leukemic cells
Our RNA sequencing data showed a 17.54-fold increase in ISG15 gene expression in NB4 cells with ATRA treatment (Orfali et al., 2019) (Table 1). We validated this finding by qPCR measurement of ISG15 expression in ATRA-treated NB4 cells at 72 h, which showed a 23-fold increase in expression in differentiating cells (****P ≤ 0.0001). No significant change in expression was found in ATRA-treated NB4R cells (Fig. 4A). HL60 cells undergoing differentiation with ATRA treatment showed a fourfold induction of ISG15 at 96 h (**P = 0.0011), whereas no induction was seen in HL60R cells (Fig. 4B). These findings associate ISG15 induction with leukemic cell differentiation rather than solely with ATRA treatment.

Knockdown of ISG15 inhibits ATRA-induced neutrophil differentiation of NB4 APL cells
To investigate whether the induction of ISG15 during ATRA-mediated leukemic cell differentiation had functional significance, we knocked down ISG15 in NB4 APL cells. We confirmed efficient ISG15 protein knockdown in two clones, shISG15_319 and shISG15_352, detecting reduced basal levels of free ISG15 in knockdown cells compared with controls and by detecting reduced induction of free ISG15 by 72 h of ATRA treatment. ISG15 conjugates were not seen before or after ATRA treatment in knockdown cells consistent with the knockdown blocking ATRA-induced ISGylation (Fig. 5A).
ATRA-induced neutrophil differentiation was reduced in both ISG15 knockdown clones compared with control cells, analogous to UBE2L6-depleted cells. At the transcript level, we found reduced GCSFR mRNA expression in ATRA-treated knockdown clones compared with ATRA-treated control cells at 72 h (****P ≤ 0.0001, **P = 0.0002, respectively) (Fig. 5B). At the protein level, surface CD11b expression was reduced in both ATRA-treated knockdown clones (*P = 0.0434 in shISG15_319 clone), but did not achieve significance in the shISG15_352 knockdown clone (Fig. 5C). ISG15 knockdown cells failed to morphologically differentiate into mature myeloid forms after 72 h of ATRA treatment (Fig. 5D lower middle and right panels), and their ability to reduce nitro blue tetrazolium was also diminished as is shown in Fig. 5E (i) lower middle and right panels. Decreased nitro blue tetrazolium reduction was quantified and is presented in Fig. 5E (ii) (*P = 0.0312, **P = 0031).
Our findings suggest a functional role for ISGylation in the ATRA-mediated neutrophil differentiation of APL cells. In the context of our earlier results, Fig. 4. ISG15 expression is induced during leukemic cell differentiation. (A) NB4 and NB4R cells were seeded at 0.2 9 10 5 cells per mL and treated with 1 lM ATRA for 72 h. Successful differentiation was confirmed in NB4 cells by flow cytometric analysis of CD11b expression. ATRA-resistant NB4R cells did not differentiate (data not shown). Total RNA was extracted, and ISG15 mRNA expression was quantified by qPCR. Values are given as n-fold induction compared with untreated cells and normalized to housekeeping gene hPRT (n = 3) (t-test ****P ≤ 0.0001). (B) HL60 and HL60R cells were treated with 1 lM ATRA for 96 h. Successful HL60 differentiation was confirmed by qPCR measurement of GCSFR expression. ATRA-resistant HL60R cells failed to differentiate (data not shown). Total RNA was extracted, and ISG15 expression was quantified by qPCR. Values are given as n-fold induction compared with untreated cells and normalized to housekeeping gene HMBS (n = 3) (t-test **P ≤ 0.01).   ISGylation may be the prominent mechanism by which UBE2L6 regulates differentiation.

Discussion
Our results demonstrate that UBE2L6 is underexpressed in AML cells compared with their mature myeloid counterparts. Using two cell line models of AML cell differentiation, we show that cells undergoing ATRA-mediated neutrophil differentiation strongly induce UBE2L6. We show that shRNA depletion of UBE2L6 in leukemic cells impedes their ability to differentiate, reporting for the first time a functional importance for this enzyme in ATRA-mediated leukemic cell differentiation.
While UBE2L6 was first identified as an E2-conjugating enzyme in ubiquitination, it is now thought to preferentially function as an E2 enzyme for ISG15 conjugation (Jeon et al., 2010;Kim et al., 2004;Zhao et al., 2004). We demonstrate that UBE2L6 modulates ISGylation in leukemic cells and further show that genetic inhibition of ISG15 strongly interferes with the neutrophil differentiation of ATRA-treated APL cells. We hence propose that the effects of UBE2L6 on leukemic cell differentiation are likely to involve its activity in ISGylation.
In addition to ISG15 and other elements of the cellular ISGylation machinery, UBE2L6 is induced by type 1 interferon signaling and contains an interferonstimulated response element in its promoter region (Kim et al., 2004). ATRA has previously been shown to upregulate ISGylation machinery in NB4 cells and to stimulate ISGylation (Dao et al., 2006). This effect is thought to be due to ATRA stimulating the secretion of type I interferon as antibody-mediated blockade of the interferon relative receptor complex suppresses this ISGylation (Dao et al., 2006). Experiments by Pitha-Rowe and colleagues identified that UBE1L, the E1-activating enzyme of ISGylation, is induced by ATRA treatment in ATRA-sensitive but not ATRA-resistant APL cells (Kitareewan et al., 2002). Subsequently, this group identified RARa binding in a domain of the UBE1L promoter, which was repressed by the PML-RARa oncoprotein (Kitareewan et al., 2002). While our observation that UBE2L6 is induced only in ATRA-sensitive NB4 APL cells could be explained by direct repression by PML-RARa, we observe equivalent UBE2L6 expression levels in ATRA-sensitive and ATRA-resistant HL60 cells which do not carry the PML-RARa oncoprotein. We thus speculate that UBE2L6 is activated during the leukemic cell differentiation program with a functional purpose regardless of the presence of fusion oncoproteins.
Equally, ISG15 induction has been reported following ATRA treatment only in differentiating NB4 APL cells and not in ATRA-resistant NB4R cells (Guo et al., 2010;Pitha-Rowe et al., 2004). Up until now, a similar finding in HL60 and HL60R cells has not been reported. Our findings again suggest that ISG15 is activated during the cellular differentiation program rather than being regulated by the PML-RARa oncoprotein. The ISG15 promoter contains a PU.1 binding site in addition to two ISREs, and it is possible that it is activated by this transcription factor during leukemic cell differentiation (Meraro et al., 2002).
The precise cellular functions of ISGylation remain under speculation. Proteomic studies in a range of cell lines have now collectively identified over 300 putative ISGylation targets, with no functional or compartmental class over-represented among them (Giannakopoulos et al., 2005;Malakhov et al., 2003;Zhao et al., 2005). HERC5, the primary E3 ligase of ISGylation, associates with ribosomes and broadly captures newly synthesized proteins of both endogenous origin and exogenous origin for ISGylation (Durfee et al., 2010). Induced by both type I interferon and lipopolysaccharide, ISGylation is thought to play a role in our defense against viral pathogens, but the exact mechanisms at play remain under investigation . While ubiquitination can modulate protein function or promote proteasomal degradation (depending on specific linkages) (Ebner and Versteeg, 2017), it is unclear whether ISGylation has analogous effects on target proteins and whether these effects may be contextual . ISGylation can stabilize proteins, competitively preventing their ubiquitination and subsequent degradation, as is seen with interferon regulatory factor 3 (Shi et al., 2010). It may also modify enzymes involved in ubiquitination impeding their function and thus negatively regulating proteasomal degradation of ubiquitin substrates (Takeuchi and Yokosawa, 2005;Zou et al., 2005). ISGylation might also inactivate or destabilize proteins through proteasomal channels. UBE1L-mediated ISGylation of cyclin D1 in lung cancer cells reduces detectable protein levels with an antiproliferative effect. The reduction in cyclin D1 is reversed with the overexpression of the deISGylating enzyme USP18 . Further study is warranted into the context-dependent and possibly tumor-suppressive actions of ISGylation in physiologic and pathologic settings.
An inhibitory effect on ATRA-mediated neutrophil differentiation of NB4 APL cells as a direct result of shRNA depletion of either UBE2L6 or ISG15 has not been previously reported. Our findings propose a functional role for these genes in differentiation. We have previously reviewed the effects of ATRA on the PML-RARa protein in APL cells (Orfali et al., 2014). Briefly, in addition to derepressing transcription, ATRA induces the degradation of the PML-RARa oncoprotein through caspase-3-mediated cleavage, ubiquitin/ proteasome-mediated proteolysis, and lysosomal-mediated autophagy (Orfali et al., 2014). Investigating a temporal correlation between the reduction in PML-RARa and an induction of UBE1L in ATRA-treated NB4 cells, Shah et al. (2008) reported ISGylation of the PML domain of PML-RARa with subsequent repression. This repression was opposed by the overexpression of USP18. Subsequent work showed that knockdown of USP18 destabilized PML-RARa and promoted apoptosis in NB4 APL cells but did not have an effect on differentiation (Guo et al., 2010). A similar destabilizing effect of ISGylation on a leukemic oncoprotein is suggested for the BCR-ABL kinase that drives chronic myeloid leukemia (CML). The expression of BCR-ABL in mouse bone marrow cells results in splenomegaly and an abnormal myeloproliferation resembling CML. Bone marrow cells harvested from USP18 À/À mice and transfected with BCR-ABL prior to transplantation into wild-type recipient mice developed a CML-like state in only 40% of cases, whereas all mice transplanted with USP18 +/+ BCR-ABL expressing cells developed disease (Yan et al., 2007). Degradation of PML-RARa may be one mechanism through which UBE2L6 and ISGylation contribute to ATRA-mediated APL cell differentiation as we have observed. However, we speculate that this pathway may have broader functions in this process. This is supported by a proposed role for ISGylation machinery in the differentiation of other hemopoietic cells. High levels of USP18 in murine hematopoietic cells block the cytokine-induced terminal differentiation of monocytic cells (Liu et al., 1999). ISG15 along with UBE1L and UBCM8 (the murine orthologue of UBE2L6) is induced during erythroid development in mice, and erythroblasts cultured ex vivo from ISG15 À/À mice show impaired differentiation (Maragno et al., 2011). Future work will examine the effects of modulating ISGylation in non-APL models of leukemic differentiation.
Our previous work has proposed that promoting autophagy may enhance the differentiating effects of ATRA on leukemic cells (Orfali et al., 2015). With further study, promoting ISGylation may prove to have similar benefits. Inhibiting USP18, the negative regulator of ISGylation, has been shown in both cell line and in vivo systems to enhance ISG15 conjugation (Ketscher et al., 2015). A small-molecule inhibitor of this isopeptidase is awaited and will greatly advance study in this arena (Basters et al., 2012).

Conclusions
We have identified a novel function of UBE2L6 in the granulocytic differentiation of APL cells, mediated by ISGylation. Our work contributes to the growing area of study of post-translational protein alteration by ubiquitin-like modifiers. A greater understanding of the protein handling that occurs during leukemic cell differentiation might allow us to modulate these processes and broaden the application of differentiation therapy for the improved treatment of AML.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. ISGylation-in the first step of ISGylation, ISG15 is activated by the E1 enzyme UBE1L in an ATP-dependent reaction.