Multiple cancer pathways regulate telomere protection

Abstract

Telomeres are considered as universal anti-cancer targets, as telomere maintenance is essential to sustain indefinite cancer growth. Mutations in telomerase, the enzyme that maintains telomeres, are among the most frequently found in cancer. In addition, mutations in components of the telomere protective complex, or shelterin, are also found in familial and sporadic cancers. Most efforts to target telomeres have focused in telomerase inhibition; however, recent studies suggest that direct targeting of the shelterin complex could represent a more effective strategy. In particular, we recently showed that genetic deletion of the TRF1 essential shelterin protein impairs tumor growth in aggressive lung cancer and glioblastoma (GBM) mouse models by direct induction of telomere damage independently of telomere length. Here, we screen for TRF1 inhibitory drugs using a collection of FDA-approved drugs and drugs in clinical trials, which cover the majority of pathways included in the Reactome database. Among other targets, we find that inhibition of several kinases of the Ras pathway, including ERK and MEK, recapitulates the effects of Trf1 genetic deletion, including induction of telomeric DNA damage, telomere fragility, and inhibition of cancer stemness. We further show that both bRAF and ERK2 kinases phosphorylate TRF1 in vitro and that these modifications are essential for TRF1 location to telomeres in vivo. Finally, we use these new TRF1 regulatory pathways as the basis to discover novel drug combinations based on TRF1 inhibition, with the goal of effectively blocking potential resistance to individual drugs in patient-derived glioblastoma xenograft models.

Keywords: ERK kinase; TRF1 inhibitors; drug resistance; glioblastoma; telomeres.

Figure EV1. Validation of the novel pathways of TRF1 regulation

1. A

   Representative image of the pathways included in Reactome database.

2. B

   Quantification of TRF1 nuclear fluorescence in CHA9‐3 lung cancer cells treated with DMSO or structurally different MEK inhibitors for 24 h at 1 μM. Data are representative of n = 2 biological replicates

3. C

   Quantification of TRF1 nuclear fluorescence in CHA9‐3 lung cancer cells treated with DMSO or structurally different ERK inhibitors for 24 h at 1 μM. Data are representative of n = 2 biological replicates

4. D

   Western blot images (up) and p‐ERK protein levels (down) of CHA9‐3 lung cancer cells treated with the indicated compounds for 24 h at 1 μM. Data are representative of n = 2 biological replicates

5. E

   Quantification of TRF1 nuclear fluorescence in CHA9‐3 lung cancer cells treated with DMSO or structurally different HSP90 inhibitors for 24 h at 1 μM. Data are representative of n = 2 biological replicates

6. F

   Quantification of TRF1 nuclear fluorescence in CHA9‐3 lung cancer cells treated with DMSO or different tubulin agents for 24 h at 1 μM. Data are representative of n = 2 biological replicates.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01, ***P < 0.001.Source data are available online for this figure.

Figure 1. Identification of novel compounds with the ability to downregulate TRF1 protein levels

1. A

   Experimental procedure: 114 compounds approved by the FDA or in clinical trials are assessed by the Opera High Screening system for their ability to reduce TRF1 protein levels in CHA9.3 lung cancer mouse cells.

2. B

   Representative images (top) and quantification (bottom) of TRF1 nuclear fluorescence of patient‐derived h676 GSCs cells treated with the indicated compounds for 24 h at 1 μM. Scale bars, 5 μm. Data are representative of n = 3 biological replicates.

3. C

   Western blot images (top) and TRF1 protein levels (bottom) of patient‐derived h676 GSCs cells treated with the indicated compounds for 24 h at 1 μM. Data are representative of n = 3 (PLKi, HSP90i, and RTKi) and n = 4 (Aurorai, mTOR, CDKi, docetaxel, gemcitabine, ERKi, MEKi) biological replicates.

4. D

   Schematic representation of the novel TRF1 regulatory pathways. Asterisk indicates targets of TRF1 inhibitory compounds found in the screening.

Data information: Data are represented as mean ± SEM. n represents biological replicates. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01, ***P < 0.001.Source data are available online for this figure.

Figure 2. New TRF1 chemical inhibitors induce DNA damage in lung cancer and glioblastoma cells

1. A

   Representative images (left) and percentage (right) of cells presenting 1 or more γH2AX and RAP1 colocalizing foci (TIFs) upon treatment of CHA9‐3 lung cancer cells with the indicated compounds. White arrowheads point to colocalization of γH2AX and RAP1. Scale bars, 5 μm. Data are representative of n = 6 (DMSO) and n = 3 (mTORi, PI3Ki, RTKi, MEKi, ERKi, HSPO90i, CDKi, docetaxel) biological replicates.

2. B

   Representative images (left) and percentage (right) of γH2AX‐positive cells per field in DMSO or compound‐treated patient‐derived h676 GSCs. Scale bars, 50 μm. Data are representative of 6 (DMSO) and 3 (mTORi, docetaxel, ERKi, MEKi, RTKi, HSP90i, gemcitabine, CDKi) biological replicates.

3. C

   Quantification of multitelomeric signals (MTS) in patient‐derived h676 GSC metaphases upon treatment with the indicated compounds. Representative images of the qFISH in the metaphases (left). Multitelomeric signals are indicated by arrowheads. Scale bars, 1 μm. Data are representative of n = 31 (DMSO), n = 18 (mTORi), n = 11 (MEKi), and n = 24 (RTKi) biological replicates.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3. New TRF1 inhibitory compounds reduce stemness in patient‐derived glioma stem cells

1. A–H

   Dose–response curves of h543 and h676 patient‐derived GSCs treated with the indicated compounds at several concentrations. Data are representative of n = 2 biological replicates. Data are represented as mean ± SEM normalized to DMSO.

Figure 4. ERK2, bRaf, and mTOR kinases phosphorylate TRF1 in vitro

1. A–D

   1 or 2 μM of GST or GST‐TRF1 was incubated with the indicated concentrations of mouse ERK2 kinase (A), human BRaf kinase (WT or V600E) (B, C), or mouse MEK1 kinase (D) in the presence of 5 μCi [γ‐³²P]ATP. The mixture was resolved by SDS–PAGE followed by autoradiography.

2. E

   1 μM of GST‐TRF1 and 0.2 μM of mouse ERK2 kinase were incubated in the presence of ERK and MEK inhibitors.

3. F

   2 μM of GST‐TRF1 and 0.1 μM of human BRaf kinase were incubated in the presence of the bRaf inhibitors dabrafenib and vemurafenib.

4. G, H

   1 or 2 μM of GST or GST‐TRF1 was incubated with the indicated concentrations of human mTOR kinase (G) in the presence of the mTOR inhibitors rapamycin and Ku0063794 (H).

5. I–K

   Phosphopeptide peak intensity normalized to total TRF1 signal in samples containing only TRF1 or TRF1 plus ERK2 (I), TRF1 plus bRAF^WT or bRAF^V600E (J), and TRF1 plus mTOR (K); data are representative of n = 2 independent experiments.

6. L

   Schematic representation of TRF1 protein with the phosphorylation sites by ERK2, bRAF, mTOR, and AKT.

7. M–O

   Representative image (down) and quantification (up) of in vitro phosphorylation assays with the indicated GST‐TRF1 wild‐type or mutated forms in the presence of mouse ERK2 kinase. Data are representative of n = 4 independent experiments.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.Source data are available online for this figure.

Figure 5. Phosphorylation of TRF1 residue T330 stabilizes TRF1 telomeric foci in vivo

1. A

   Trf1 ^lox/lox p53 ^−/− MEFs were transduced with eGFP‐tagged Trf1 WT or mutant alleles to overexpressed TRF1 depicted variants. Endogenous TRF1 was deleted by transduction with Cre recombinase. Opera High Content Screening (HCS) system was used to quantify the GFP spot intensity per cell.

2. B

   Western blot images of Trf1 ^lox/lox MEFs with or without overexpression of eGFP‐Trf1 WT or mutant alleles followed by Cre recombinase transduction.

3. C

   Quantification of eGFP‐TRF1 inhibition in Trf1 ^Δ/Δ MEFs transduced with eGFP‐Trf1 WT or mutant alleles as indicated. Data are representative of n = 15 biological replicates

4. D

   Growth curves of Trf1 ^Δ/Δ MEFs transduced with eGFP‐Trf1 WT or mutant alleles as indicated. Data are representative of n = 5 biological replicates

5. E

   Western blot images of Trf1 ^lox/lox MEFs with or without overexpression of eGFP‐Trf1 WT or mutant alleles followed by Cre recombinase transduction.

6. F

   Quantification of eGFP‐TRF1 inhibition in Trf1 ^Δ/Δ MEFs transduced with eGFP‐Trf1 WT or mutant alleles as indicated. Data are representative of n = 5 biological replicates

7. G

   Western blot images of p53 ^−/− MEFs with or without overexpression of eGFP‐Trf1 WT or mutant alleles followed treatment with ERKi. Data are representative of n = 2 biological replicates

8. H

   Representative images (above) and percentage (bottom) of telomeric and 53BP1 colocalizing foci (TIFs) per cells of p53 ^−/− MEFs with or without overexpression of eGFP‐Trf1 WT and the indicated mutants upon treatment with the ERKi. White arrowheads: colocalization of telomeric and 53BP1. Scale bars, 10 μm. Data are representative of n = 2 independent experiments.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, ***P < 0.001.Source data are available online for this figure.

Figure EV2. GFP‐tagged TRF1 telomeric foci in MEFs expressing either WT or mutants in ERK‐dependent phosphosites of GFP‐TRF1

Representative images of Trf1 ^Δ/Δ MEFs transduced with eGFP‐Trf1 WT or mutant alleles as indicated. Scale bars, 5 μm.

Figure 6. Genetic model of the telomeric role of ERK1/2‐mediated TRF1 phosphorylation

1. A

   Western blot image (above) and quantification (bottom) of TRF1 protein levels upon genetic depletion of ERK1/2 in p53 ^−/− MEF line. Data are representative of n = 3 independent experiments.

2. B

   Representative images (above) and quantification (bottom) of TRF1 telomeric foci in ERK1/2 RNA interfered p53 ^−/− MEFs. Scale bars, 5 μm. Data are representative of n = 2 independent experiments.

3. C

   p53 ^−/− MEFs were sequentially transduced with lentiparticles encoding short hairpins against ERK1/2 and retroparticles for eGFP‐tagged Trf1 WT or mutant alleles to overexpress TRF1‐depicted variants in the absence of ERK1/2.

4. D

   Western blot image of eGFP‐tagged and endogenous TRF1 protein levels upon genetic depletion of ERK1/2 in p53 ^−/− MEF line.

5. E

   Representative images (above) and percentage (bottom) of telomeric and 53BP1 colocalizing foci (TIFs) per cell in p53 ^−/− MEF with or without overexpression of eGFP‐Trf1 WT and indicated mutants upon genetic depletion of ERK1/2. White arrowheads: colocalization of telomeric and 53BP1. Scale bars, 10 μm. Data are representative of n = 2 biological replicates.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01.Source data are available online for this figure.

Figure EV3. Patient‐derived xenografts become resistant to PI3Ki

1. A

   Longitudinal tumor growth follow‐up in ETP‐47037 or vehicle‐treated mice injected with patient‐derived h676 GSCs. Data are representative of n = 4 independent tumors.

2. B

   Longitudinal tumor growth follow‐up in ETP‐47037 or vehicle‐treated mice injected with patient derived h543 GSCs. Data are representative of n = 4 independent tumors.

3. C

   Western blot images (left) and p‐AKT/AKT or pS6/S6 protein levels (right) in vehicle‐ or ETP‐47037‐treated tumors. Data are representative of n = 4 independent tumors.

4. D

   TRF1 nuclear fluorescence in ETP‐47037 or vehicle‐treated tumors. Scale bars, 10 μm. Data are representative of n = 4 independent tumors

5. E

   Western blot images (left) and TRF1 protein levels (right) in vehicle‐ or ETP‐47037‐treated tumors. Data are representative of n = 4 independent tumors.

Data information: Data are represented as mean ± SEM.Source data are available online for this figure.

Figure EV4. In vitro combinatorial studies with the new TRF1 inhibitory compounds

1. A, B

   Number of spheres formed by patient‐derived h676 GSCs 7 days after treatment with the indicated compounds as single agents or in combination. Data are representative of n = 2 biological replicates

2. C–H

   Diameter of spheres formed by patient‐derived h676 GSCs 7 days after treatment with the indicated compounds as single agents or in combination. n represents biological replicates: in (C) DMSO n = 42, PI3Ki n = 49, RTKi n = 56, Comb n = 82; in (D) DMSO n = 48, PI3Ki n = 57; ERKi n = 48, Comb n = 71; in (E) DMSO n = 50; PI3Ki n = 49, MEKi n = 45, Comb n = 47; in (F) DMSO n = 39, PI3Ki n = 38, HSP90i n = 34, Comb n = 31; in (G) DMSO n = 42, PI3Ki n = 35, Gem n = 32, Comb n = 28; in (H) DMSO n = 37, PI3Ki n = 35, Doc n = 34, Comb n = 30.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by ***P < 0.001.

Figure 7. In vitro combinatorial studies of PI3Ki with novel TRF1 inhibitory compounds

1. A–F

   Representative images (left) and quantification (right) of number of spheres formed by patient‐derived h676 GSCs 7 days after treatment with the indicated compounds as single agents or in combination. Scale bars, 100 μm. Data are representative of n = 6 biological replicates.

2. G–L

   Western blot images (left) and TRF1 protein levels (right) measured in patient‐derived h676 GSCs 24 h after treatment with the indicated compounds as single agents or in combination. Data are representative of n = 3 (combination in K), n = 4 (combination in G, L), n = 5 (RTKi in G, combination in H–J, docetaxel in L), n = 6 (DMSO in G–L, and ERKi in H), n = 11 (PI3Ki in G–J, L) biological replicates.

Data information: Data are represented as mean ± SEM. Significant differences using unpaired t‐test are indicated by *P < 0.05, **P < 0.01, ***P < 0.001.Source data are available online for this figure.

Figure 8. In vivo combinatorial studies of PI3Ki with novel TRF1 inhibitory compounds in patient‐derived GBM xenograft models

1. A–D

   Longitudinal tumor growth follow‐up in mice injected with patient‐derived h676 GSCs and treated with the indicated compounds in single agents or combination. n represents number of tumors, in (A): vehicle n = 16, PI3Ki n = 8, ERKi n = 16, combination n = 8; in (B): vehicle n = 16, PI3Ki n = 16, MEKi n = 8, combination n = 8; in (C): vehicle n = 8, PI3Ki n = 16, docetaxel n = 16, combination n = 8; in (D): vehicle n = 16, PI3K n = 16, gemcitabine n = 4, combination n = 4. P‐values represent the mean of all the time points.

2. E–H

   Representative images (top) and quantification (bottom) of TRF1 nuclear fluorescence in tumors treated with the indicated compounds as single agents or in combination. Scale bars, 10 μm. Data are represented as mean ± SEM. n represents number of tumors: in (E): vehicle n = 15, PI3Ki n = 10, ERKi n = 8, combination n = 2; in (F): vehicle n = 15, PI3Ki n = 10, MEKi n = 8, combination n = 6; in (G): vehicle n = 15, PI3Ki n = 10, docetaxel n = 8, combination n = 4; in (H): vehicle n = 15, PI3K n = 10, gemcitabine n = 8, combination n = 6.

Data information: Significant differences using unpaired t‐test with Welch's correction are indicated by *P < 0.05, **P < 0.01, ***P < 0.001.