Skip to main page content
U.S. flag
See More

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2014 Jul 3;511(7507):90-3.
doi: 10.1038/nature13283. Epub 2014 May 28.

The sonic hedgehog factor GLI1 imparts drug resistance through inducible glucuronidation

Hiba Ahmad Zahreddine  1 , Biljana Culjkovic-Kraljacic  1 , Sarit Assouline  2 , Patrick Gendron  1 , Andrea A Romeo  3 , Stephen J Morris  3 , Gregory Cormack  1 , James B Jaquith  4 , Leandro Cerchietti  5 , Eftihia Cocolakis  2 , Abdellatif Amri  1 , Julie Bergeron  6 , Brian Leber  7 , Michael W Becker  8 , Shanshan Pei  9 , Craig T Jordan  9 , Wilson H Miller  2 , Katherine L B Borden  1
Affiliations

The sonic hedgehog factor GLI1 imparts drug resistance through inducible glucuronidation

Hiba Ahmad Zahreddine et al. Nature. .

Abstract

Drug resistance is a major hurdle in oncology. Responses of acute myeloid leukaemia (AML) patients to cytarabine (Ara-C)-based therapies are often short lived with a median overall survival of months. Therapies are under development to improve outcomes and include targeting the eukaryotic translation initiation factor (eIF4E) with its inhibitor ribavirin. In a Phase II clinical trial in poor prognosis AML, ribavirin monotherapy yielded promising responses including remissions; however, all patients relapsed. Here we identify a novel form of drug resistance to ribavirin and Ara-C. We observe that the sonic hedgehog transcription factor glioma-associated protein 1 (GLI1) and the UDP glucuronosyltransferase (UGT1A) family of enzymes are elevated in resistant cells. UGT1As add glucuronic acid to many drugs, modifying their activity in diverse tissues. GLI1 alone is sufficient to drive UGT1A-dependent glucuronidation of ribavirin and Ara-C, and thus drug resistance. Resistance is overcome by genetic or pharmacological inhibition of GLI1, revealing a potential strategy to overcome drug resistance in some patients.

PubMed Disclaimer

Conflict of interest statement

Conflict of interest

KLBB, SA, WHM, HAZ, BCK, BL, JB, GC, PG, EC, LC, SP, CT, MBJ have no competing financial interests. SJM, JBJ and AAR are/were employees of Pharmascience.

Figures

Figure 1
Figure 1. Ribavirin resistance in patients and cell lines
A. Bone marrow biopsies for patient 11 (Wright-Giemsa staining, 50X magnification). EOT, end-of-treatment. B. Cell growth in untreated (UN) or 20 uM ribavirin (Rib), which is clinically achievable. C. Westerns of eIF4E targets Mcl-1 and c-Myc. D. 3H ribavirin uptake.. E. eIF4E co-immunoprecipitation with 3H-ribavirin relative to IgG. Loading is in Extended Data Figure 6. F. Westerns of Gli1 and UGT1A. Panels B, D and E: Means +/− SD (error bars). *** p-value<0.001 (Student’s t-test). All panels: Experiments were completed in triplicate, at least three independent times. HSP90 and tubulin provide loading controls.
Figure 2
Figure 2. Gli1 status underpins resistance
A. Gli1 mRNA levels before treatment(BT), at best response(Resp) and EOT for ribavirin-treated patients/healthy volunteers. B. Gli1 mRNA (right) and UGT1A protein (left) for patients treated with Ara-C- therapies at relapse(R)/diagnosis(D). C–D. Effects of Gli1-overexpression on drug sensitivity: ribavirin(20 uM), Ara-C(200 nM, clinically achievable). E. siRNA to Gli1 restores drug sensitivity. siLuc, non-specific control RNA. A–B. RNA results were normalized to ubiquitin. All panels: Means+/−SD (error bars). Experiments were completed in triplicate at least three independent times. ** p-value<0.01, *** p-value<0.001 (Student’s t-test). Figure 4A–C shows corresponding westerns. Actin, tubulin and GADPH provide loading controls.
Figure 3
Figure 3. Targeting Gli1 activity
A. FRII cell growth +/− GDC-0449 (GDC) and ribavirin or Ara-C. B. The ribavirin-eIF4E complex is measured as a function of Gli1 status using 3H-ribavirin and eIF4E immunoprecipitation relative to IgG controls: FaDu-Gli1 cells, siGli1, GDC-0449 treatment. See Extended Data Figure 6 for immunoprecipitation controls. C. Methyl-cellulose colony growth assays in specimens from M4/M5 AML patients or healthy volunteers (bone marrow). All panels: Mean +/− SD (error bars). Results are representative of at least three independent experiments in triplicate (A–D) or at least two (E) in replicates of 5. p-value <0.01 **, and p-value<0.001 *** (Student’s t-test).
Figure 4
Figure 4. Link between Gli1, UGT1A and drug glucuronidation
A–C. Westerns as a function of Gli1. HSP90 provides a loading control. D–I. MS/MS extracted ion chromatographs are shown. Rib-Glu Ribavirin Glucuronide; RDP* (ribavirin diphosphate minus H2O). Fragmentation of Rib-Glu peak confirms its identity (Extended Data Figure 7). J. Ribavirin, the best-supported glucuronidation site indicated by red arrow. K. Glucuronic acid L. RDP M. 3H ribavirin-eIF4E-GST complexes competed by equimolar amounts of compounds. 3H ribavirin did not bind beads alone. See Extended Data Figure 7D for loading. Averages +/− SD (error bars). Experiments were carried out in triplicate, at least three independent times.
Figure 1
Figure 1
Ribavirin resistance in some ribavirin monotherapy clinical trial patients. BM biopsies for responding (patient 8 achieved a PR; patient 11 a CR is shown in Figure 1A) and non-responding patients (patient 9 was a PD and patient 13, SD). Note abundance of blasts before treatment and at the end of treatment (EOT) and reduction in blasts and restoration of hematopoiesis during response. For Wright-Giemsa stained samples magnification was 50× with oil (patients 8 and 13), and for Lysozyme staining 20X (Patient 9).
Figure 2
Figure 2
Characterization of FaDu and THP-1 derived resistant cell lines. A. Detailed growth curves for FRI and FRII resistant cells. All cell lines have indistinguishable doubling times in the absence of ribavirin (far right panel). B. THP-1 resistant (TR) cell lines are not sensitive to treatment with ribavirin at the doses and times used. Ribavirin no longer targets eIF4E activity (i.e. Mcl-1) in resistant cells (far right panel). There were no changes in eIF4E levels between resistant and parental cell lines (and Figure 1C). Actin provides a loading control. C. Resistance is retained after 6 months growth in the absence of ribavirin. D. Incubation of live cells with 3H-ribavirin indicates that THP-1 resistant cells have impaired uptake of ribavirin similar to FRI cells. E. eIF4E cap binding and eIF4G binding activity are retained in FRII cells. F. FRII cells are sensitive to eIF4E knockdown measured by cell growth. G. Effects of RNAi mediated knockdown of Gli1 or eIF4E on UGT1A levels. Western blots were probed as indicated. RNAi mediated knockdown of Gli1 led to reduced levels of UGT1A whereas knockdown of eIF4E did not. For UGT1A, a pan-UGT1A antibody was used. Antibody controls for UGT1A and Gli1 are shown in Extended Data Figure 8. Results are representative of at least three independent experiments. Average values are reported and error bars indicate standard deviations. Experiments were carried out in triplicate, three independent times. Western blots are representative of at least three independent experiments.
Figure 3
Figure 3
Pro-drug metabolism is impaired in Type I resistance. A–B. Western analysis reveals that ADK levels were reduced in FRI cells (panel A) and THP1 resistant cells (panel B). Treatments were 48 hours at 20μM ribavirin. C–D. Knockdown of ADK leads to ribavirin resistance as shown by cell growth. Western blot confirms knockdown of ADK. Hsp90 provides a loading control. E–F. ENT1 and ADK mRNA levels for patients’ specimens. Patient 11 (CR) was responding clinically at (and prior to) 84 and 112 days and relapsed around day 252, when both ADK and ENT1 mRNA levels decreased. Analysis of RNA samples isolated before and at the end of the first 28 day cycle for patient 9 (who did not respond to ribavirin) compared to a healthy volunteer. Averaged values for ADK and ENT1 RNAs were normalized to glucose 6-phosphate dehydrogenase (G6PDH). Error bars indicate SD and centre values are averages. All experiments were performed in triplicate at least three independent times. ** indicates p-value<0.01 and *** p-value<0.001 as determined by two-tailed Student’s t-test. Results are representative of at least three independent experiments.
Figure 4
Figure 4
Confocal micrographs of leukemic blasts isolated from bone marrows of responding and non-responding patients at before treatment (BT), Response or End of Treatment (EOT). Immunostaining for Gli1 and UGT1A are shown. DAPI is in blue. Note nuclear accumulation of Gli1 in non-responding patients, indicating elevated Gli1 activity. High levels of Gli1 and UGT1A suggest primary resistance. All confocal settings were identical between specimens and thus lower signal is indicative of less protein. A 100X objective with no digital zoom was used for patients 9, 11, 13, 17 and 19. The same objective but a digital zoom of 2X was used for patients 8, A, B and C. Note Patient C was still in remission at EOT (see main text). For each patient, staining was carried out three independent experiments. Controls for specificity of Gli1 and UGT1A antibodies are provided in Extended Data Figure 8.
Figure 5
Figure 5
Higher Gli1 expression is found in poor cytogenetic risk group and predicts a trend of worse survival outcome in AML. To study the prognostic value of Gli1 gene expression in AML, we mined the publicly available AML data set published by The Cancer Genome Atlas Research Network (A) Gli1 gene expression in 176 de novo AML patients grouped by cytogenetic risks. The expression level is represented by RPKM value which stands for Reads Per Kilobase of transcript per Million mapped reads in RNA-seq. Each patient is represented by a symbol. Error bars represent median ± IQR (Inter-Quartile Range) of each group. Nonparametric Mann-Whitney test was used to analyze the differences between groups. A total of 176 de novo AML patients with complete mRNA-seq and cytogenetic risk classification data are included in this analysis. (B, C) Kaplan-Meier plots of events-free survival (EFS) and overall survival (OS) of 168 de novo AML patients segregated by median Gli1 expression (RPKM=0.8596) (B) or high Gli1 expression (RPKM greater than or equal to 2) (C). Each tick on the survival curve represents a censored event because the patient is still alive at the end of the TCGA study. A total of 168 de novo AML patients with complete mRNA-seq, and reliable EFS, and OS data are included in this analysis (Patient information details are described in the Supplemental Table 1 of the NEJM study. Mantel-Cox test was performed to calculate log-rank p values. We also observed that abnormally low levels of Gli1 were also correlated with poor outcome (data not shown), suggesting that Gli1 levels must be in a “Goldilocks” zone.
Figure 6
Figure 6
Effects of modulation of Gli1 levels on UGT1A A. Effects of the direct Gli1 inhibitor GANT-61 on restoring ribavirin sensitivity (20 uM) in FRII cells. Effects are dependent on GANT-61 dose. B–D. Controls for eIF4E-ribavirin immunoprecipitations (IP) shown in Figures 1E and 3B. Inputs, supernatants (Sn) and IP controls for 3H ribavirin anti-eIF4E IPs are shown for GDC-0449 treated cells (B), FRII and FRI cells (C), and RNAi mediated knockdown of Gli1 (D). E–G. qPCR analysis of Gli1 (E) and UGT1A (F) using a pan-UGT1A primer or primers for specific UGT1As (G). mRNA levels were normalized to RNA Polymerase IIa. These findings are consistent with Extended Data Table 1 which indicates lower levels of UGT1A mRNA levels. Further, UGT1A3 and UGT1A8 decreased similarly (data not shown). Experiments were carried out in triplicate, at least three independent times. Average values are reported and error bars indicate standard deviations. These findings, that Gli1 elevation leads to reduce mRNA levels but increased protein levels are counter-intuitive. We hypothesize that Gli1 elevation increases protein stability of UGT1As (see below) and this leads to some sort of feedback mechanism leading to reduced UGT1As. Other scenarios are possible but the main point that Gli1 elevation leads to increased UGT1A protein production is clear. G. Gli1 increases UGT1A protein stability as shown by studies with the proteasomal inhibitor MG132 and a pan-UGT1A antibody. Here, MG132 addition stabilizes levels of UGT1A in parental cells, but in FRII cells where levels are already increased, there is no further increase with MG132. This indicates that UGT1A proteins are already stabilized in the FRII cells. All results are representative of three independent experiments. H. Western blot analysis with a pan-UGT2B antibody indicates that UGT2B levels are unchanged in FRII relative to FaDu cells suggesting the glucuronidation effects are mediated mostly through the UGT1A family. Tubulin provides a loading control.
Figure 7
Figure 7
Mass Spectrometry Analysis of Ribavirin and Ara-C glucuronidation. A. MS/MS collision induced fragmentation analysis indicate that a breakdown product of the ribavirin glucuronide missing the ribose ring (exact mass 288.07) was further fragmented into a fragment of this glucuronide (exact mass 244.08, red asterisk) and to the triazole ring, the key moiety of ribavirin (exact mass 112.04). No ribose-glucuronide or ribose fragment was detected in our experiments suggesting this is not a major glucuronidation site in these cells. However, we cannot rule out that this exists and could not be detected. B. Microsomes expressing UGT1A1, UGT1A4, UGT1A6 and UGT1A9 were treated with RTP, underwent HILIC chromatography and the resulting EIC is shown. The Rib-Glu peak is clearly present and fragmentation analysis as in part A confirms that this is glucuronidated ribavirin. We note that microsomes only expressing UGT1A1 do not glucuronidate RTP; and that RTP, but not ribavirin, is glucuronidated in microsomes. These studies suggest that UGT1A4, UGT1A6 and/or UGT1A9 are required for glucuronidation as is some phosphorylation event prior to glucuronidation. C. Using HILIC chromatography, we isolated the fraction containing the Rib-Glu peak in part B. A portion of this was re-assessed by MS/MS to be sure that the correct peak was isolated. This material was used in the 3H ribavirin competition assay in Figure 4M. Material was quantified using a standard curve of ribavirin (see Materials and Methods). D. Western blot demonstrating equal loading of eIF4E-GST in the 3H ribavirin pulldown assay shown in Figure 4M. All results are representative of at least three independent experiments. E. AraC is glucuronidated (AraC-Glu) in FRII cells but not parental FaDu cells where AraC-TP (triphosphate) is observed. AraC-TP is also observed in FRII cells, but at much lower levels than AraC-Glu. Treatment of FRII cells with GDC-0449 results in the loss of the AraC-Glu peak and causes no alteration to the parental FaDu cells. Fragmentation strongly suggests that the cytosine is the major site of glucuronidation (data not shown). We did not observe masses consistent with an arabinose breakdown product or an arabinose-glucuronide but cannot rule out that they are present at low levels or that our isolation procedure precluded their detection. F. Structures of AraC and AraC-TP are shown. The red arrow indicating the most likely glucuronidation site, as per our mass spectrometry data. Note that no glucuronides were observed when reactions were incubated in the absence of UDP-glucuronic acid (data not shown).
Figure 8
Figure 8
Investigations into why Gli1 levels are elevated in FRII cells. A. Analysis of expression of a subset of hedgehog signalling pathway proteins. Western blots are probed as indicated and are representative of three independent experiments. Tubulin and Hsp90 provided loading controls. Patched 1 (PTCH1) was the most significant change. PTCH1 is 210 kDa, with an often observed degradation product at 170 kDa. IHH indicates Indian Hedgehog; SHH sonic hedgehog, Smo smoothened. B. GLI1 DNA methylation. CpG methylation was interrogated on bisulfite converted DNA from GLI1 promoter region and first exon. The amplicon covered 29 CpG, 25 of them located within a CpG island. DNA CpG methylation is shown as percent methylation for FaDu (top) and FRII (bottom) cells. There was no difference observed between the cell lines. C–D. Antibody controls. Analysis of Gli1 (C) and pan-UGT1A (D) antibodies as a function of RNAi mediated knockdown of these proteins as indicated. Note that UGT1A family members have approximately the same molecular weight. Results are representative of at least three independent experiments.
See this image and copyright information in PMC

Comment in

References

    1. Deschler B, de Witte T, Mertelsmann R, Lubbert M. Treatment decision-making for older patients with high-risk myelodysplastic syndrome or acute myeloid leukemia: problems and approaches. Haematologica. 2006;91:1513–1522. - PubMed
    1. Deschler B, Lubbert M. Acute myeloid leukemia: epidemiology and etiology. Cancer. 2006;107:2099–2107. doi: 10.1002/cncr.22233. - DOI - PubMed
    1. Pulsoni A, et al. Survival of elderly patients with acute myeloid leukemia. Haematologica. 2004;89:296–302. - PubMed
    1. Hiddemann W, et al. Management of acute myeloid leukemia in elderly patients. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 1999;17:3569–3576. - PubMed
    1. Assouline S, et al. Molecular targeting of the oncogene eIF4E in acute myeloid leukemia (AML): a proof-of-principle clinical trial with ribavirin. Blood. 2009;114:257–260. doi: 10.1182/blood-2009-02-205153. blood-2009-02-205153 [pii] - DOI - PubMed

Publication types

MeSH terms