Actinomycin D enhances killing of cancer cells byimmunotoxin RG7787 through activation of theextrinsic pathway of apoptosisXiu Fen Liua, Laiman Xianga,1, Qi Zhoua, Jean-Philippe Carralotb, Marco Prunottob, Gerhard Niederfellnerc,and Ira Pastana,2

aLaboratory of Molecular Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892; bOffice ofInnovation, Hoffman-La Roche AG, 4070 Basel, Switzerland; and cRoche Pharmaceutical Research & Early Development, Discovery Oncology, InnovationCenter Penzberg, Roche Diagnostics GmbH, 82377 Penzberg, Germany

Contributed by Ira Pastan, July 27, 2016 (sent for review June 16, 2016; reviewed by Michael G. Rosenblum and Daniel A. Vallera)

RG7787 is a mesothelin-targeted immunotoxin designed to havelow-immunogenicity, high-cytotoxic activity and fewer side effects.RG7787 kills many types of mesothelin-expressing cancer cells linesand causes tumor regressions in mice. Safety and immunogenicityof RG7787 is now being assessed in a phase I trial. To enhance theantitumor activity of RG7787, we screened for clinically used drugsthat can synergize with RG7787. Actinomycin D is a potent transcrip-tion inhibitor that is used for treating several cancers. We reporthere that actinomycin D and RG7787 act synergistically to kill manymesothelin-positive cancer cell lines and produce major regressionsof pancreatic and stomach cancer xenografts. Analyses of RNA ex-pression show that RG7787 or actinomycin D alone and togetherincrease levels of TNF/TNFR family members and NF-κB–regulatedgenes. Western blots revealed the combination changed apoptoticprotein levels and enhanced cleavage of Caspases and PARP.

cancer therapy | apoptosis | immunotherapy | pancreatic cancer |mesothelioma

Recombinant immunotoxins (RITs) are chimeric proteins thatcontain an antibody fragment directed against a tumor-

selective surface antigen attached to a protein toxin. We haveconstructed immunotoxins by attaching a 38 kDa fragment ofPseudomonoas exotoxin A (PE38) to the Fv portion of mAb thatbinds to cancer cells but not to essential tissues (1, 2). RITs killcells by ADP-ribosylating and inactivating elongation factor(EF)-2, leading to protein synthesis arrest, a fall in MCL-1 levels,and induction of apoptosis (3, 4). SS1P is a RIT that targetsmesothelin, a protein highly expressed on mesothelioma, pan-creatic, ovarian, lung, and stomach cancers. Because SS1P con-tains a bacterial toxin, it is immunogenic and can only be givenfor one treatment cycle to most patients. However, when com-bined with pentostatin and cyclophosphamide to suppress anti-body formation, SS1P has produced major and prolonged tumorregressions in some patients with advanced chemo-refractorymesothelioma (5–7).RG7787 (now named LMB-100) is in clinical trials for re-

fractory pancreatic cancer (NCT02810418) and mesothelioma(NCT02798536). It is a derivative of SS1P containing mutationsthat make it less immunogenic, more active in killing target cells,and better tolerated by patients (7). The targeting moiety ofRG7787 is a humanized antimesothelin Fab; its effector moiety is a24-kDa ADP ribosylation domain of PE fused via a furin cleavablelinker to the Fab. The domain III variant used in RG7787 containsmutations that silence many human B-cell epitopes and someT-cell epitopes. RG7787 is cytotoxic to many mesothelin-expressingcell lines and when combined with pacilitaxel produces completeremissions in pancreatic cancer-bearing mice (7).The mechanism by which immunotoxins kill cells is not completely

understood. After binding to specific receptors, immunotoxins entercells by endocytosis, and in the endocytic compartment, the furinseparates the Fv from the toxin. Then the toxin is transferred in a

retrograde fashion through the Golgi and endoplasmic reticuluminto the cytosol. There the toxin catalyzes the ADP ribosylation ofEF-2, leading to protein synthesis arrest and apoptosis (4).Actinomycin D (Act D) is a polypeptide antibiotic isolated from

the genus Streptomyces. Act D intercalates into DNA, preventingthe progression of RNA polymerases (8, 9). It is widely used as atranscription inhibitor. RNA polymerase I, catalyzing ribosomalRNA transcription, is most sensitive to Act D (IC50, 0.05 μg/mL);RNAP II (0.5 μg/mL) and RNAP III (about 5 μg/mL) are lesssensitive (9). Nanomolar concentrations of Act D block tran-scription of RNA polymerase I and induce nucleolar stress byinterfering with ribosome biogenesis (10–12). Act D is the firstantibiotic used for treating cancer; these cancers include gesta-tional trophoblastic neoplasia, Wilms tumor, rhabdomyosarcoma,Ewing’s sarcoma, and NPM1-mutated acute myeloid leukemia (13).The mechanism by which Act D causes tumor cell death isnot established.Many anticancer drugs exert their effects by triggering apo-

ptosis (14, 15). Apoptosis can be triggered by the extrinsic (re-ceptor) pathway or at the mitochondrial level by the intrinsicpathway. Activation of members of the TNFR superfamily, suchas FAS, TRAILR1 and R2, TNFR, and CD137, results in re-cruitment of death domain proteins and caspase-8 activation,which can result in the cleavage of Bid and activation of theintrinsic apoptotic pathway (16). TNFR superfamily activationleads to activation of NF-κB, a key regulatory protein that caninitiate cell death (17–19).

Significance

Because effective cancer therapy usually requires a combina-tion of drugs, we searched for clinically used anticancer agentsthat would enhance the activity of immunotoxin RG7787 sothey could be combined in humans. We show here that acti-nomycin D activates the extrinsic pathway of apoptosis andacts synergistically with RG7787 to kill a variety of cancer celllines and cause striking tumor regression in mice. These dataindicate that combining immunotoxins like RG7787 that killcells by inhibiting protein synthesis with actinomycin D is auseful strategy to enhance their antitumor activity in humans.

Author contributions: X.F.L., M.P., G.N., and I.P. designed research; X.F.L., L.X., Q.Z., andJ.-P.C. performed research; X.F.L., J.-P.C., and I.P. analyzed data; and X.F.L., J.-P.C., M.P.,G.N., and I.P. wrote the paper.

Reviewers: M.G.R., University of Texas MD Anderson Cancer Center; and D.A.V., Universityof Minnesota Cancer Center.

The authors declare no conflict of interest.1Retired.2To whom correspondence should be addressed. Email: pastani@mail.nih.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1611481113/-/DCSupplemental.

10666–10671 | PNAS | September 20, 2016 | vol. 113 | no. 38 www.pnas.org/cgi/doi/10.1073/pnas.1611481113

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0

To improve the antitumor activity of RG7787, we have screenednine DNA-damaging agents in clinical use that enhance RG7787action. We report here that Act D and RG7787 act synergisticallyto increase killing of many mesothelin-expressing cancers andcause major tumor regressions in tumor-bearing mice. Thesedata support the combined use of these agents in ongoing trialswith RG7787.

ResultsAct D Enhances Killing of KLM1 Cells by RG7787. We screened sevenDNA-damaging drugs and found that Act D dramatically syner-gized with RG7787 to kill 25 cancer cell lines (lung, pancreas,ovary) (Fig. S1 and Table S1). To determine the cytotoxic mech-anism, we stained the treated cells with annexin V and 7AAD andused flow cytometry to measure cell apoptotic death. Fig. 1Ashows that 9% of cells treated for 24 h with RG7787 at 100 ng/mLhad died, Act D alone at 10 ng/mL did not cause cell death, butthe combination was very effective, killing about 20% of the cells.To examine the effect of lower concentrations of these agents, weextended the treatment time to 72 h (Fig. 1B) or 96 h (Fig. 1C).Under both conditions, the combination was more than additive,and 60–75% of the cells underwent apoptosis. Fig. 1D shows

photomicrographs of KLM1 cells after 4 d of treatment withRG7787 (10 ng/mL) or Act D (10 ng/mL) or both. Cells treatedwith Act D alone appeared larger and thinner, and there werefewer cells, indicating inhibition of cell growth. With RG7787many cells died and small clusters of cells survived. In the com-bination group, only a few nonviable rounded cells were presenton day 4, which did not grow out when the drugs were removed(Fig. 1D).

Act D Accelerates Killing of Tumor Cells. Because immunotoxinRG7787 has a relatively short half-life in the circulation, it isimportant that cells are killed after a short exposure to RG7787(20). To evaluate if Act D treatment shortens the time neededfor RG7787 to kill cells, KLM1 cells were treated with low dosesof RG7787 in combination with Act D for 6, 24, or 48 h, and thecells were transferred to drug-free medium and followed. Fig. 1Eshows that exposure to each agent alone for 6 h had little effecton the cells, but the combination decreased cell numbers. Treat-ment with either agent for 24 or 48 h slightly decreased cell num-bers, but there were very few cells after combination treatment for24 h and no cells after 48 h of treatment.

Act D Enhances RG7787 Killing of Many Cancer Cells. We next ex-amined the stomach cancer line MKN28 (Fig. S2A shows pho-tomicrographs of these cells). Because they die more slowly thanKLM1 cells, we treated for 3 d and grew them in drug-freemedium for 2 more days. After 5 d the MKN28 cells in thecontrol and the Act D group reached confluence. RG7787 at20 ng/mL killed some cells, but after 5 d, the surviving cellsstarted to regrow. However, the combination of Act D and RG7787eliminated almost all of the cells. Similar results were observedwith the pancreatic cancer line, AsPC1 pancreatic cells, andRH16 human mesothelioma cells when treated with RG7787and Act D (Fig. S2 B and C).To verify that the treated cells were dying by apoptosis and to

quantify the effect, we used flow cytometry and stained cells withannexin V and 7AAD. Table 1 shows data from eight cell lines ofdiverse cancer types: pancreas, breast, stomach, lung, cervix, andmesothelioma. With all these cancer types, we found that killingby the combination was more than additive.The combination of Act D and RG7787 also caused rapid

killing of pancreatic cancer AsPC-1 cells (Fig. S3A). In 6 h, thecombination killed equivalent numbers of cells as RG7787 singletreatment for 24 or 48 h. In addition, combined treatment for 24or 48 h completely killed almost all of the cells. Hay and MKN28cells were also efficiently killed by low doses of RG7787 whencombined with Act D (Fig. S3 B and C).

Mouse Experiments. To evaluate the effect of combination treat-ment, we used KLM1 pancreatic tumors, which we previouslyshowed were mainly growth-inhibited by RG7787 alone (21). Fig.2A shows that tumors had reached 100 mm3 on day 6 aftertreatment was started. The PBS control group continued to growand reached about 500 mm3 on day 15. Tumors in the RG7787group had a slight decrease in size after the first cycle of treat-ment but had grown significantly by day 22. Treatment with ActD slowed tumor growth but did not cause tumor shrinkage.However, tumors in the combination group started to shrinkfrom the second day of treatment. On day 30, five of eight micehad no measurable tumors, and three of eight had very smalltumors below 20 mm3. On day 41, two mice still had no mea-surable tumors. The other six tumors started to grow, but allwere below 50 mm3. No mice died from the therapy, althoughsome mice in the combination group lost up to 10% of theirweight during the first few days of treatment. They had fullyrecovered by the end of the second cycle of treatment. This ex-periment was repeated with similar results. We did not try threecycles of treatment.

Fig. 1. KLM1 cells are susceptible to Act D and RG7787 treatment. (A–C)KLM1 cells were treated with RG7787 with or without 10 nM Act D for 24 h(A, RG7787, 100 ng/mL), 72 h (B, RG7787, 10 ng/mL), or 96 h (C, RG7787, 5 ng/mL).The cells were stained with annexin V and 7AAD and analyzed by FACS.Percentages of dead cells in the graph are generated by subtracting deadcells from the untreated control group. The data were generated fromat least three separate experiments. (D) Microscopy examination of cellstreated with RG7787 (10 ng/mL) or Act D (10 ng/mL) for 4 d. (E) Act Daccelerated and resulted in complete killing of tumor cells by RG7787. KLM1cells were treated with Act D (10 ng/mL), RG7787 (10 ng/mL), or a combi-nation. After 6, 24, or 48 h, the cells were changed to fresh media to let thesurviving cells grow for 1 wk. The images were taken after cells were fixedand stained with crystal violet. Comb, combination of Act D and RG7787;Con, untreated control.

Liu et al. PNAS | September 20, 2016 | vol. 113 | no. 38 | 10667

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0

We also examined the effect of RG7787 and Act D on MKN28tumors (Fig. 2B). MKN28 tumors grew slower than KLM1 tu-mors and had reached 100 mm3 on day 9, when we started treat-ment. Tumors in the control group (n = 6) reached 500 mm3 on day20. Tumor growth slowed in both the RG7787 group (n = 6) andthe Act D group (n = 6), but tumor shrinkage was not observed.In contrast, with combination treatment (Combo, n = 6), thetumors decreased in size up until day 21 and then began to re-grow. The combination significantly inhibited the tumor growthmuch more than RG7787 or Act D single treatment (P < 0.01)from day 15 to day 28.

Mechanism Studies.One way to increase immunotoxin action is toincrease the rate and amount of immunotoxin taken into the cell.To measure uptake, cells were exposed to Alexa Fluor 647–RG7787, and the amount accumulated was measured in a flowcytometer. Fig. 3A shows that the uptake of Alexa–RG7787 isnot changed by exposure to Act D when measured from 20 minto 3 h. After internalization and furin cleavage, the toxin istransferred to the endoplasmic reticulum and then to the cytosol,where it ADP-ribosylates EF-2. To determine if the modificationand inactivation of EF-2 was stimulated by Act D treatment, cellswere treated for 2, 6, and 24 h, and cell lysates were prepared foranalysis of the state of EF-2. In extracts of untreated cells, EF-2was modified by NAD-biotin in the presence of toxin, and themodified EF-2–biotin was detected on a Western blot (Fig. 3B).In cells treated with immunotoxin, EF-2 becomes resistant tomodification with NAD-biotin, because the site is already mod-ified. Fig. 3B shows that at 2 and 6 h, the EF-2 in RG7787-treated cells is resistant to ADP ribosylation and that Act D hasno effect on this modification. Treatment for 24 h with Act D

slightly decreased ADP ribosylation, but the combination withRG7787 did not further decrease levels of ADP–EF-2 (24 hlonger exposure). This indicates that Act D does not affect thetoxin trafficking to the endoplasmic reticulum, transfer to thecytosol, or EF-2 modification and must act on some subsequentstep to arrest protein synthesis.

Act D and RG7787 Changed Apoptotic Proteins. Western blots wereperformed to determine which apoptotic proteins were changedby this treatment (Fig. 4A). The combination greatly increasedlevels of PARP, tBID, and Caspase-3, -8, and -9, all of which areproapoptotic proteins. Act D by itself had very little effect onthese proteins, except for elevating BIM, which is antiapoptotic,but BIM levels were decreased by RG7787. RG7787 alone low-ered levels of MCL-1, BIM, and BCLxl and had small effects onCaspase-8 and -9 and on PARP.

RNA Changes Induced by Act D and RG7787. Act D is an effectiveinhibitor of RNA polymerase 1, and this inhibition results in adecrease in the synthesis of ribosomal RNA (9). To determine ifthe low concentrations of Act D used in our experiments affectedlevels of RNA encoding proteins in the apoptotic pathway, weused a RNA array containing 84 different proapoptotic andantiapoptotic genes. KLM1 and RH16 cells were treated withAct D, RG7787, or both for 24 h and prepared RNA for analysis.The ratios of RNA levels compared with untreated cells areshown in Table S2. For both cell types, the levels of RNA for thecontrol house-keeping genes (ACTB, B2M, GAPDH, RPLP0, orHPRT1) were largely unaffected. This result indicates that at thelow concentration of Act D used in these experiments, there isnot an overall inhibition of RNA synthesis. Unexpectedly, we

Fig. 2. Act D stimulated RG7787 killing of tumor cells in xenografts in mice. (A) KLM1 cells were s.c. injected and treated with Act D and RG7787 as describedin Materials and Methods. There is a statistically significant difference between RG7787 and combination treatment beginning from day 11 (*P < 0.001, n = 8per treatment group). (B) MKN28 cells were similarly injected and treated as in A at the indicated days. There are statistically significant differences betweenRG7787 and combination (*P ≤ 0.01, n = 6) starting from day 15 until the end of the experiment.

Table 1. Synergistic killing of tumor cells with Act D and RG7787

Cell lines Tumor types Con Act D RG7787 Act D + RG7787

KLM-1 Pancreatic 4.5 ± 1.1 16 ± 3.1 23.6 ± 4.4 66.9 ± 4.0HCC70 Breast 8.1 ± 1.7 15.9 ± 5.2 23.4 ± 7.5 64.4 ± 1.6HAY Mesothelioma 6.8 ± 0.35 13.1 ± 1.1 21.7 ± 2.2 39.9 ± 3.5RH16 Mesothelioma 6.8 ± 0.7 15.1 ± 5.8 23.3 ± 0.1 75.5 ± 4.5MKN28 Stomach 6.1 ± 0.9 13.9 ± 0.7 8.9 ± 1.0 58.8 ± 16.7NUGC4 Stomach 6.6 ± 0.25 10.6 ± 1.8 14.5 ± 2.3 37.8 ± 2.5KB 31 Cervical 9.6 ± 0.3 15.4 ± 0.87 22.5 ± 1.6 68.0 ± 2.3L55 Lung 6.5 ± 0.35 7.2 ± 0.04 20.5 ± 1 33.5 ± 0.85

Tumor cells were cultured with Act D at either 2.5 ng/mL (Hay), 5 ng/mL (RH16), 10 ng (KLM1, KB, MKN28,HCC70, NUGC4), or 15 ng/mL (L55) with or without RG7787 at 5 ng/mL (KLM1, KB, Hay, MKN28, NUGC4),10 ng/mL (HCC70, RH16), or 50 ng/mL (L55) for 3 or 4 d (RH16). Cell death was determined by FACS analysis afterstaining of annexin V and 7AAD. The data were generated from two or three separate experiments.

10668 | www.pnas.org/cgi/doi/10.1073/pnas.1611481113 Liu et al.

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0

observed that the level of many RNAs is increased by treatmentwith Act D or RG7787 or both. RNAs that changed over threefoldare listed in Table 2. Act D produced dramatic increases in KLM-1cells in many members of the TNF and TNFR superfamily, in-cluding TNFα and TNFβ, DR2, TNFR2, CD137, and CD27. Othergenes whose expression is elevated by Act D are NF-κB target genes(BCL2A1, CIAP2, and GADD45α). In RH16 cells, there was anincrease in Fas and CD27 but not other TNFR family members.RG7787 increased the levels of many of the same RNAs, and theywere also elevated in cells treated with the combination. However,there were differences. For example, CD27 was elevated by Act Dand the combination but was decreased by RG7787 alone.To confirm the array findings, we chose several genes with

the largest changes and performed real time-PCR using KLM1RNA. As seen in Table S3, TNFα and TNFβ were elevated from10- to 98-fold by either Act D, RG7787, or the combination. TheTNFR family member CD137 was elevated 56-fold by RG7787alone, and DR5 increased two- to threefold. The other NF-κBtarget genes—BCL2A1, BIK, CIAP2, and GADD45α—were alsofound to be elevated. We also performed real-time PCR withRNA from RH16 cells and confirmed the increased expressionof the same genes as seen in the apoptotic RNA array experi-ments with RH16 cells (Table S2).

Changes in Proteins of the NF-κB Pathway. Because activation ofthe TNF/TNFR pathway increases the activity or expression ofNF-κB, we examined levels of proteins in the NF-κB pathwayusing KLM1 cells. Fig. 4B shows that P-MAPK (T202/Y204), thekinase that phosphorylates NF-κB, is undetectable in controlcells, slightly increased in Act D-treated cells, dramatically ele-vated by RG7787, and further elevated by the combination. Wealso observed that total MAPK is increased by treatment withAct D, RG7787, or the combination. P-NF-κB (S536), which isthe activated form of NF-κB, was also increased by Act D,RG7787, or combination treatment (1.8-, 2.7-, and 3.3-fold, re-spectively). Total NF-κB was only slightly increased (less thantwofold). To confirm that NF-κB was activated, we isolated the

nuclear fraction of cells, in which the activated form of NF-κBaccumulates. As shown in Fig. 4C, nuclear NF-κB levels wereincreased 3.1-fold by Act D treatment, and over 50-fold byRG7787, and by combination treatment.

Role of TNF Family Members. Because RNA and protein of mem-bers of the TNF/TNFR family were increased by Act D, we in-vestigated if TNFα family members could have a direct role inRG7787-mediated cell killing. As shown in Fig. 5A, cells treatedwith either FASL or Trail or TNFα did not undergo cell death.However, when combined with RG7787, these agents increasedcell death but not to the extent produced by Act D.

Effect of Act D with Other Inhibitors of Protein Synthesis. There are avariety of protein toxins that kill cells by inhibiting protein syn-thesis. Diphtheria toxins (DPs) like PE ADP-ribosylates EF-2,whereas Gelonin inactivates the 60S ribosomal subunit. Thesetoxins are also used to make immunotoxins (22, 23). To determineif there is also enhancement by Act D, KLM-1 cells were exposedto Act D and various toxins for 72 h and cell death was measuredby flow cytometry. Because we do not have immunotoxins con-taining these toxins, we compared their activities with native PE.We find that killing of cells by Act D combined with PE or DT orGelonin is more than additive (Fig. 5B). These data also show thatAct D-stimulated cell killing is not mesothelin-dependent.

Fig. 4. Western blot analysis of apoptotic proteins and signaling molecules.(A) Apoptotic proteins changed by Act D and RG7787. KLM1 cells weretreated with 10 ng/mL Act D with or without RG7787 for 24 h, and cell ly-sates were analyzed by Western blot using anti-MCL1, anti-BCLxl, anti-Bim,anti-Bid, anti-cleaved Caspase-3, anti-cleaved Caspase-8, anti-cleaved Cas-pase-9, and anti-PARP. Actin is the loading control. (B) Increased level of TNF/TNFR superfamily and activation of NF-κB. Total protein lysates from KLM1cells treated with 10 ng/mL Act D with or without 100 ng/mL RG7787 wereblotted with anti-DR5, anti-CIAP2, anti–P-MAPK (p38), anti-MAPK (p38),anti–NF-κB (p65), or P-NF-κB (p65). Actin is the normalization for the loading.(C) After KLM1 cells were treated as above, the nuclear fraction was isolatedusing a cyto/nuc fractionation kit (Pierce). The nuclear fraction was used forWestern blot with anti-total NF-κB and anti-HDAC2 (loading control). Therelative numbers were obtained from scanning the Western blot using NIHimage, and the ratios were calculated by setting the untreated control as 1.SE, short exposure.

Fig. 3. Act D did not increase cellular uptake of RG7787 and did not enhanceADP ribosylation. (A) KLM1 cells were treated with Alexa 647-labeled RG7787,and internalized RG7787 cells were determined by FACS analysis as describedin Materials and Methods. (B) KLM1 cells were treated with RG7787 with orwithout Act D for 2, 6, or 24 h, and ADP ribosylation was performed in vitro asdescribed in Materials and Methods. L.E., longer exposure.

Liu et al. PNAS | September 20, 2016 | vol. 113 | no. 38 | 10669

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0

DiscussionWe have found that Act D has remarkable synergy with immu-notoxins targeting mesothelin-expressing malignancies. RNAand protein analyses showed that cells treated with RG7787activated both the intrinsic and extrinsic apoptotic pathways. Thesynergy was found in various epithelial tumors with mesothelinexpression. We also found the stimulatory effect of Act D is notrestricted to PE-containing immunotoxins, because Act D alsoenhanced the killing of cells by other toxins that inhibit proteinsynthesis.Act D is an old drug developed before apoptosis and other

mechanisms of cell death were elucidated. It was developed as an

anticancer agent in the 1950s and approved for treatment ofhumans in the 1960s (8). It is known that Act D can bind to GC-rich regions in DNA duplexes and is especially effective at dis-ruption of ribosome RNA biogenesis (9, 10). It is also knownthat Act D is cytotoxic when used at high concentrations. In thisstudy, we used relatively low (5–10 ng/mL) concentrations of ActD. Analysis of RNA arrays showed that the levels of RNA for themajority of the 86 genes in the apoptotic array, including severalhouse-keeping genes, were not decreased in a major way com-pared with untreated cells. A decrease should have been ob-served if overall RNA synthesis had been inhibited. Instead, formany genes the levels of RNA were increased. The most affectedgenes are TNF/TNFR family members [TNFα, TNFβ, TNFR2(KLM1), CD27, CD70 (KLM1), CD137 (KLM1), and TRAILR2/DR5, FAS (RH16)], NF-κB–regulated genes (BCL2A1, CIAP2,BCL10, GADD45α), caspase family members (Caspase-1, -3, -5,-7, -9, and -10, which vary between the two cell lines), and others(BIM and RIPK2). Both BIM RNA and protein were elevated byAct D treatment but not by RG7787. BIM is located in mito-chondria and plays an important role in promoting apoptosis (24).It is noteworthy that GADD45α is a target of both NF-κB and

p53. It has been shown that Act D can stabilize p53 in the nu-cleolus and that targets of P53 are activated (10, 25). We foundthat Act D has very little effect on total NF-κB protein, but itenhances phosphorylation of NF-κB and its accumulation in thenucleus. It is likely the dramatic increase in GADD45α RNAcould be due to both p53 and NF-κB accumulation in the nucleus.We previously reported that immunotoxins targeting meso-

thelin-expressing cells sensitize the cells to killing by TRAIL(26), but we did not examine the mechanism of this effect. Ourfinding that RG7787 increases expression of TNFα, TNFβ, FAS,CD127, and DR5 and DR5 protein provides an explanation forthat result. We also found that RG7787 activated the NF-κBpathway. NF-κB RNA increased in the apoptotic arrays fromboth KLM1 and RH16 cells (Table S1). Also NF-κB p65 phos-phorylation and total NF-κB in the nuclear fraction were ele-vated. Expression of BCL10, CIAP2, and two NF-κB–regulatedgenes was also increased. It was recently reported that loss ofdiphthamide, which is an essential component for EF-2–ADPribosylation, activates NF-κB and renders cells hypersensitive toTNF-mediated apoptosis (27). It is also possible that the arrestof protein synthesis caused by RG7787 induces cell stress, acti-vates MAPK P38, and leads to phosphorylation of NF-κB, therebysensitizing cells to apoptosis.When we compared the RNAs that were increased by treat-

ment with Act D and RG7787, we found many of the sameRNAs were increased, suggesting that the two agents have over-lapping mechanisms of action. Genes whose expression wasincreased include many members of the BCL-2 family (BIM, BID,

Fig. 5. (A) TNF family members can stimulate RG7787 and Act D cell killing. KLM1 cells were incubated with 10 ng/mL of TNFα, FasL, or TRAIL with or without10 ng/mL RG7787 for 3 d. The dead cells were measured by FACS after labeling with annexin V and 7AAD. Dead cells (%) represent the effects of the drug-treated group by subtracting the dead cells from untreated control cells. Graphs are generated from an average of at least two separate experiments. (B) ActD stimulated other protein synthesis inhibitors’ killing of KLM1 cells. KLM-1 was incubated with 10 ng/mL Act D with or without 200 ng/mL CHX, 90 pg/mL DT,30 ng/mL PE, or 5 pg/mL HB21-FvPE40 (HB21), 1 μg/mL of gelonin (Gel) for 3 d. The dead cells were measured and graphed as described in A.

Table 2. Genes affected by Act D or RG7787 using apoptoticarray

KLM1 RH16

Symbol Act D RG7787 Comb Act D RG7787 Comb

TNFα 10 147.6 103.1 1.2 6.9 11.2TNFβ 1.2 7.5 13.4 4.8 2.9 3.5TNFSF7 (CD70) 2.7 4.1 3.3 1.6 1 2.8Fas 1.1 2.4 1 5.9 0.1 6.1TRAILR2/DR5 2.7 8.3 4.6 2 1 3.9TNFR2 10.5 15.5 8.1 0.5 0.6 0.4TNFRSF9 (CD137) 4.6 68.8 15.7 0.7 4.5 2.8TNFRSF7 (CD27) 3.4 0.2 0.2 2.4 0.5 1.8Caspase-1 0.7 0.3 0.4 15 1.3 13.5Caspase-10 2.1 1.9 3 3.6 1.5 5.4Caspase-3 2.9 1.4 3.8 1.8 1.6 2.4Caspase-5 2.8 0.7 5.6 2.8 1.3 3.5Caspase-7 3.4 1.1 2.2 1.2 0.9 1.8Caspase-9 1.9 2.4 3.2 1 3.4 2.4Bcl-10 2.2 4.3 3.2 2.3 1.9 3.9BCL2A1 3.4 7.4 21.6 2.1 19.2 22.6Bim 3.1 0.7 2.3 1.8 0.7 1.7Bik 1.8 6.1 3.6 2.5 1.1 4.1CIAP2 4.7 46.3 13.2 8.6 5.7 9.8GADD45A 4.1 63.1 30 4.5 4.1 21.2TP53 2.4 5.6 8.8 1.5 1.1 1.6TP53BP2 1.6 5.8 3.7 1.6 1.7 1.8RIPK2 2.3 2.9 2.2 3.1 2.4 4.8

KLM1 or RH16 cells were treated with Act D (KLM1, 10 ng/mL; RH16, 5 ng/mL)or RG7787 (KLM1, 100 ng/mL; RH16, 200 ng/mL) for 24 h. Then total RNA wasisolated and apoptotic array performed. The numbers are generated using web-based data analysis (Qiagen) by setting the untreated control as 1 and usingACTB, B2M, GAPDH, HPRT1, and RPLP0 as the internal control. The genes thatchanged over threefold were selected and presented in the table.

10670 | www.pnas.org/cgi/doi/10.1073/pnas.1611481113 Liu et al.

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0

BCL10, and BCL2A1), the Caspase family [Caspase-1 (RH16)],Caspase-10, Caspase-5, and Caspase-7 (KLM1)], and members ofthe TNF and TNFR family [TNFα and TNFβ (RH16), CD137,DR5, TNFR2 (KLM1), and Fas (RH16)]. These changes occurred inboth KLM-1 and RH16 cells, indicating the changes are not limitedto one type of cancer cell. In some but not all cases, the increase waslarger in doubly treated cells (also seen in Fig. S3). The importanceof the variable increase between singly and doubly treated cells isnot clear, because we only measured RNA levels in cells after a 24-htreatment and some cells may have been dying, which could alterRNA levels. Possibly the increase in RNA will be larger in doublytreated cells.The mechanism by which Act D and RG7787 act synergisti-

cally to kill target cells is complex. Most prominently, Act Ddramatically enhanced RG7787-induced increases of cleavedCaspase-3, -8, and -9 and PARP. The enhancement could comefrom activation of NF-κB–mediated signaling. Secondly, Act Dcan interfere with ribosomal biogenesis and causes stabilizationof P53 in the nucleolus (25). Increased P53 was shown to or-chestrate a transcriptional response to stress and cause cell-cyclearrest and cell death (12). We observed an increase of both p53and p53BP RNA in the KLM1 cells. We also found that the p53responsive gene GADD45α increased in KLM1 and RH16 cells,which can lead to growth arrest. Because the addition of TNFα,FASL, or Trail produced only a small increase in cell death whencombined with RG7787 whereas Act D addition is very potent

(Fig. 5A), we conclude that Act D enhancement of RG7787cytotoxic activity involves more genes and pathways than justdeath receptor-mediated cell death. Finally, our data stronglysupport the use of Act D to enhance immunotoxin actionin humans.

Materials and MethodsRG7787 was made at Roche. Act D, TNFα, FasL, and TRAIL were from Sigma.Antibodies to EF-2; anti-BAX; BCLxl; BIM; Caspase-3, -8, and -9; cleavedCaspase-3, -8, and -9; PAPP; MAPK; P-MAPK (Thr180/Tyr182); CIAP2; RelA/p65(NF-κB); and P-NF-κB (Ser-536) were from Cell Signaling; anti-DR5 was fromProSci. KLM1, AsPC1 (21), MKN28 (28), and KB31 (29) were all described. L55was from S. Albelda, University of Pennsylvania, Philadelphia. NCI-Meso16(RH16) was from R. Hassan, National Cancer Institute, National Institutes ofHealth, Bethesda, MD, and grown as described (29). Flow cytometry, ADPribosylation of eEF-2, and real-time PCR were as described (30). Primers arelisted in Table S4. Apoptotic array followed the manufacturer’s instructions(Qiagen). To make tumors, two million KLM1 or MKN28 cells were implanteds.c. into athymic nude mice. Mice received two cycles of treatment. Act D(0.6 mg/kg) was injected i.p. once a week for 2 wk; RG7787 was injected i.v.(2.5 mg/kg) three doses every other day for 2 wk. Tumor volumes weremeasured and calculated as described (21). The animal protocol was ap-proved by the National Cancer Institute Animal Care and Use Committee.

ACKNOWLEDGMENTS. This work was supported by the Intramural ResearchProgram of the NIH, National Cancer Institute (NCI), Center for Cancer Re-search, and a Cooperative Research and Development Agreement betweenNCI and Roche.

1. Pastan I, Hassan R, Fitzgerald DJ, Kreitman RJ (2006) Immunotoxin therapy of cancer.Nat Rev Cancer 6(7):559–565.

2. Wayne AS, Fitzgerald DJ, Kreitman RJ, Pastan I (2014) Immunotoxins for leukemia.Blood 123(16):2470–2477.

3. Du X, Youle RJ, FitzGerald DJ, Pastan I (2010) Pseudomonas exotoxin A-mediatedapoptosis is Bak dependent and preceded by the degradation of Mcl-1. Mol Cell Biol30(14):3444–3452.

4. Weldon JE, Pastan I (2011) A guide to taming a toxin–Recombinant immunotoxinsconstructed from Pseudomonas exotoxin A for the treatment of cancer. FEBS J278(23):4683–4700.

5. Hassan R, et al. (2007) Phase I study of SS1P, a recombinant anti-mesothelin im-munotoxin given as a bolus I.V. infusion to patients with mesothelin-expressingmesothelioma, ovarian, and pancreatic cancers. Clin Cancer Res 13(17):5144–5149.

6. Hassan R, Ho M (2008) Mesothelin targeted cancer immunotherapy. Eur J Cancer44(1):46–53.

7. Hassan R, Alewine C, Pastan I (2016) New life for immunotoxin cancer therapy. ClinCancer Res 22(5):1055–1058.

8. Farber S, D’Angio G, Evans A, Mitus A (1960) Clinical studies on actinomycin D withspecial reference to Wilms’ tumor in children. Ann N Y Acad Sci 89:421–425.

9. Bensaude O (2011) Inhibiting eukaryotic transcription: Which compound to choose?How to evaluate its activity? Transcription 2(3):103–108.

10. Burger K, et al. (2010) Chemotherapeutic drugs inhibit ribosome biogenesis at variouslevels. J Biol Chem 285(16):12416–12425.

11. Andersen JS, et al. (2005) Nucleolar proteome dynamics. Nature 433(7021):77–83.12. Rao B, Lain S, Thompson AM (2013) p53-based cyclotherapy: Exploiting the ‘guardian

of the genome’ to protect normal cells from cytotoxic therapy. Br J Cancer 109(12):2954–2958.

13. Falini B, Brunetti L, Martelli MP (2015) Dactinomycin in NPM1-mutated acute myeloidleukemia. N Engl J Med 373(12):1180–1182.

14. Fulda S, Debatin K-M (2006) Extrinsic versus intrinsic apoptosis pathways in anticancerchemotherapy. Oncogene 25(34):4798–4811.

15. Azijli K, Weyhenmeyer B, Peters GJ, de Jong S, Kruyt FA (2013) Non-canonical kinasesignaling by the death ligand TRAIL in cancer cells: Discord in the death receptorfamily. Cell Death Differ 20(7):858–868.

16. Wei MC, et al. (2000) tBID, a membrane-targeted death ligand, oligomerizes BAK torelease cytochrome c. Genes Dev 14(16):2060–2071.

17. Brenner D, Blaser H, Mak TW (2015) Regulation of tumour necrosis factor signalling:Live or let die. Nat Rev Immunol 15(6):362–374.

18. Oeckinghaus A, Hayden MS, Ghosh S (2011) Crosstalk in NF-κB signaling pathways.Nat Immunol 12(8):695–708.

19. Ravi R, et al. (2001) Regulation of death receptor expression and TRAIL/Apo2L-inducedapoptosis by NF-kappaB. Nat Cell Biol 3(4):409–416.

20. Mason-Osann E, Hollevoet K, Niederfellner G, Pastan I (2015) Quantification of re-combinant immunotoxin delivery to solid tumors allows for direct comparison ofin vivo and in vitro results. Sci Rep 5:10832.

21. Hollevoet K, et al. (2014) In vitro and in vivo activity of the low-immunogenic anti-mesothelin immunotoxin RG7787 in pancreatic cancer. Mol Cancer Ther 13(8):2040–2049.

22. Lyu MA, Cao YJ, Mohamedali KA, Rosenblum MG (2012) Cell-targeting fusion con-structs containing recombinant gelonin. Methods Enzymol 502:167–214.

23. Li YM, Vallera DA, Hall WA (2013) Diphtheria toxin-based targeted toxin therapy forbrain tumors. J Neurooncol 114(2):155–164.

24. Harada H, Grant S (2012) Targeting the regulatory machinery of BIM for cancertherapy. Crit Rev Eukaryot Gene Expr 22(2):117–129.

25. Rubbi CP, Milner J (2003) Disruption of the nucleolus mediates stabilization of p53 inresponse to DNA damage and other stresses. EMBO J 22(22):6068–6077.

26. Du X, Xiang L, Mackall C, Pastan I (2011) Killing of resistant cancer cells with low Bakby a combination of an antimesothelin immunotoxin and a TRAIL Receptor 2 agonistantibody. Clin Cancer Res 17(18):5926–5934.

27. Stahl S, et al. (2015) Loss of diphthamide pre-activates NF-κB and death receptorpathways and renders MCF7 cells hypersensitive to tumor necrosis factor. Proc NatlAcad Sci USA 112(34):10732–10737.

28. Alewine C, et al. (2014) Efficacy of RG7787, a next-generation mesothelin-targetedimmunotoxin, against triple-negative breast and gastric cancers. Mol Cancer Ther13(11):2653–2661.

29. Liu XF, FitzGerald DJ, Pastan I (2013) The insulin receptor negatively regulates theaction of Pseudomonas toxin-based immunotoxins and native Pseudomonas toxin.Cancer Res 73(7):2281–2288.

30. Liu X, Müller F, Wayne AS, Pastan I (2016) Protein kinase inhibitor H89 enhances theactivity of Pseudomonas exotoxin A-based immunotoxins. Mol Cancer Ther 15(5):1053–1062.

31. Lehár J, et al. (2009) Synergistic drug combinations tend to improve therapeuticallyrelevant selectivity. Nat Biotechnol 27(7):659–666.

Liu et al. PNAS | September 20, 2016 | vol. 113 | no. 38 | 10671

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Apr

il 6,

202

0