Top Banner
Peptidic degron for IMiD-induced degradation of heterologous proteins Vidyasagar Koduri a , Samuel K. McBrayer a , Ella Liberzon a , Adam C. Wang a , Kimberly J. Briggs a,1 , Hyejin Cho a,2 , and William G. Kaelin Jr. a,b,3 a Department of Medical Oncology, DanaFarber Cancer Institute and Brigham and Womens Hospital, Harvard Medical School, Boston, MA 02215; and b Howard Hughes Medical Institute, Chevy Chase, MD 02185 Contributed by William G. Kaelin Jr., December 9, 2018 (sent for review November 2, 2018; reviewed by Deepak Nijhawan and Charles L. Sawyers) Current systems for modulating the abundance of proteins of interest in living cells are powerful tools for studying protein function but differ in terms of their complexity and ease of use. Moreover, no one system is ideal for all applications, and the best system for a given protein of interest must often be determined empirically. The thalidomide-like molecules (collectively called the IMiDs) bind to the ubiquitously expressed cereblon ubiquitin ligase complex and alter its substrate specificity such that it targets the IKZF1 and IKZF3 lymphocyte transcription factors for destruction. Here, we mapped the minimal IMiD-responsive IKZF3 degron and show that this peptidic degron can be used to target heterologous proteins for destruction with IMiDs in a time- and dose-dependent manner in cultured cells grown ex vivo or in vivo. tunable proteins | protein stability | ubiquitylation | thalidomide | proteasome S ystems for regulating the abundance of a given protein of interest are useful for studying that proteins function and for modeling the consequences of pharmacologically inhibiting that protein. The systems available today can be divided into those that alter protein abundance through changes in mRNA abundance and those that act primarily at the level of protein stability (13). Some of these systems require three components, such as a suitably engineered promoter, a drug-responsive transcriptional effector, and a drug; others are two-component systems, such as systems incor- porating polypeptides that confer drug-induced stability or drug- induced instability when fused to heterologous proteins. Each of these approaches has inherent advantages and disadvantages. The ideal choice for any one protein will often depend upon a number of factors, including the normal turnover of that protein and its mRNA, tolerance of that protein to in-frame fusion events, and function of the protein. Moreover, these approaches differ with respect to the technical expertise and time, as well as the ease and cost of obtaining the reagents needed to implement them. Finally, the degree of control achieved with these approaches for any given protein can currently only be determined empirically. In light of these consid- erations, we reasoned it would be useful to have another system for controlling protein abundance, especially one that was technically simple and used readily available and inexpensive components. Thalidomide-like drugs, such as lenalidomide and pomalidomide [collectively referred to as immunomodulatory drugs (IMiDs)], bind directly to the cereblon ubiquitin ligase complex and alter its sub- strate specificity such that it targets the lymphoid-restricted tran- scription factors IKZF1 and IKZF3 for destruction (48). Loss of these two transcription factors is both necessary and sufficient to account for the antimyeloma activity of these compounds (4, 5). Many ubiquitin ligases can recognize and polyubiquitylate their substrates even if those substrates are fused to heterologous re- porters (5, 9). Consistent with this observation, cereblon targets fusion proteins consisting of firefly luciferase fused to either IKZF1 or IKZF3 for destruction in an IMiD-dependent manner (4, 5). Ebert and coworkers (4) showed that the IKZF3 degron recognized by cereblon in the presence of an IMiD is located in IKZF3 zinc finger 2 and contained within residues 136180 (Fig. 1A). We did fine mapping of this degron in hopes of identifying a short, peptidic, degron that could be used to target heterologous proteins for de- struction in an IMiD-dependent manner. Results Mapping and Functional Characterization of the Minimal IMiD-Responsive IKZF3 Degron. Modeled on our earlier work, we made a mammalian expression plasmid that encodes two bioluminescent proteins from a single mRNA: (i ) a protein of interest fused to the C terminus of firefly luciferase (FLuc) and (ii ) renilla luciferase (RLuc), with both protein ORFs preceded by internal ribosome entry site (IRES) el- ements. (Fig. 1B). Using this vector system, we expressed various fragments of IKZF3 fused to firefly luciferase in transiently trans- fected 293T cells treated with pomalidomide or vehicle. We then measured the ratio of firefly to renilla luciferase activity as a sur- rogate for destabilization of the fusion proteins by pomalidomide. As expected, firefly luciferase fusion proteins containing full- length IKZF3 (1509) or the IKZF3 degron identified by Ebert and coworkers [IKZF3 (136180)] were down-regulated by pomalidomide (Fig. 1C). Analysis of N-terminal and C-terminal deletions of IKZF3 (136180) showed that this degron could be narrowed further to IKZF3 (146168) (Fig. 1C). The C-terminal boundary of the degron is clearly at residue 168 because elimi- nating this residue inactivated the IKZF3 degron [Fig. 1C, see IKZF3 (144168) and IKF3 (144167)]. The N-terminal boundary is Significance Systems for degrading proteins at will are useful for a variety of biological experiments. Although a number of such systems have been described, they vary widely in terms of complexity, ease of obtaining the necessary reagents, and costs. Moreover, no one system seems to work for all proteins, and the ideal system often must be determined empirically. Thalidomide-like drugs (IMiDs) reprogram the ubiquitiously expressed cereblon ubiquitin ligase complex to degrade the lymphocyte transcription factors IKZF1 and IKZF3. Here, we show that an IKZF3-derived 25mer constitutes a modular degron that can be used to target heterologous proteins for destruction by IMiDs, which are widely available and cross the bloodbrain barrier, in cell culture and in mouse experiments. Author contributions: V.K., K.J.B., H.C., and W.G.K. designed research; V.K., S.K.M., E.L., A.C.W., K.J.B., and H.C. performed research; V.K., S.K.M., and E.L. contributed new re- agents/analytic tools; V.K., S.K.M., and W.G.K. analyzed data; and V.K. and W.G.K. wrote the paper. Reviewers: D.N., University of Texas Southwestern; and C.L.S., Memorial SloanKettering Cancer Center. The authors declare no conflict of interest. Published under the PNAS license. 1 Present address: Discovery Biology, Tango Therapeutics, Cambridge, MA 02141. 2 Present address: Department of Biology, Peloton Therapeutics, Dallas, TX 75235. 3 To whom correspondence should be addressed. Email: [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1818109116/-/DCSupplemental. Published online January 25, 2019. www.pnas.org/cgi/doi/10.1073/pnas.1818109116 PNAS | February 12, 2019 | vol. 116 | no. 7 | 25392544 BIOCHEMISTRY Downloaded by guest on June 14, 2021
6

Peptidic degron for IMiD-induced degradation of …2018/12/09  · Peptidic degron for IMiD-induced degradation of heterologous proteins Vidyasagar Koduria, Samuel K. McBrayera, Ella

Jan 31, 2021

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
  • Peptidic degron for IMiD-induced degradation ofheterologous proteinsVidyasagar Koduria, Samuel K. McBrayera, Ella Liberzona, Adam C. Wanga, Kimberly J. Briggsa,1, Hyejin Choa,2,and William G. Kaelin Jr.a,b,3

    aDepartment of Medical Oncology, Dana–Farber Cancer Institute and Brigham and Womens Hospital, Harvard Medical School, Boston, MA 02215;and bHoward Hughes Medical Institute, Chevy Chase, MD 02185

    Contributed by William G. Kaelin Jr., December 9, 2018 (sent for review November 2, 2018; reviewed by Deepak Nijhawan and Charles L. Sawyers)

    Current systems for modulating the abundance of proteins ofinterest in living cells are powerful tools for studying proteinfunction but differ in terms of their complexity and ease of use.Moreover, no one system is ideal for all applications, and the bestsystem for a given protein of interest must often be determinedempirically. The thalidomide-like molecules (collectively called theIMiDs) bind to the ubiquitously expressed cereblon ubiquitin ligasecomplex and alter its substrate specificity such that it targets theIKZF1 and IKZF3 lymphocyte transcription factors for destruction.Here, we mapped the minimal IMiD-responsive IKZF3 degron andshow that this peptidic degron can be used to target heterologousproteins for destruction with IMiDs in a time- and dose-dependentmanner in cultured cells grown ex vivo or in vivo.

    tunable proteins | protein stability | ubiquitylation | thalidomide |proteasome

    Systems for regulating the abundance of a given protein ofinterest are useful for studying that protein’s function and formodeling the consequences of pharmacologically inhibiting thatprotein. The systems available today can be divided into thosethat alter protein abundance through changes in mRNA abundanceand those that act primarily at the level of protein stability (1–3).Some of these systems require three components, such as a suitablyengineered promoter, a drug-responsive transcriptional effector, anda drug; others are two-component systems, such as systems incor-porating polypeptides that confer drug-induced stability or drug-induced instability when fused to heterologous proteins. Each ofthese approaches has inherent advantages and disadvantages. Theideal choice for any one protein will often depend upon a number offactors, including the normal turnover of that protein and its mRNA,tolerance of that protein to in-frame fusion events, and function ofthe protein. Moreover, these approaches differ with respect to thetechnical expertise and time, as well as the ease and cost of obtainingthe reagents needed to implement them. Finally, the degree ofcontrol achieved with these approaches for any given protein cancurrently only be determined empirically. In light of these consid-erations, we reasoned it would be useful to have another system forcontrolling protein abundance, especially one that was technicallysimple and used readily available and inexpensive components.Thalidomide-like drugs, such as lenalidomide and pomalidomide

    [collectively referred to as immunomodulatory drugs (IMiDs)], binddirectly to the cereblon ubiquitin ligase complex and alter its sub-strate specificity such that it targets the lymphoid-restricted tran-scription factors IKZF1 and IKZF3 for destruction (4–8). Loss ofthese two transcription factors is both necessary and sufficient toaccount for the antimyeloma activity of these compounds (4, 5).Many ubiquitin ligases can recognize and polyubiquitylate theirsubstrates even if those substrates are fused to heterologous re-porters (5, 9). Consistent with this observation, cereblon targetsfusion proteins consisting of firefly luciferase fused to either IKZF1or IKZF3 for destruction in an IMiD-dependent manner (4, 5).Ebert and coworkers (4) showed that the IKZF3 degron recognizedby cereblon in the presence of an IMiD is located in IKZF3 zincfinger 2 and contained within residues 136–180 (Fig. 1A). We did

    fine mapping of this degron in hopes of identifying a short, peptidic,degron that could be used to target heterologous proteins for de-struction in an IMiD-dependent manner.

    ResultsMapping and Functional Characterization of theMinimal IMiD-ResponsiveIKZF3 Degron.Modeled on our earlier work, we made a mammalianexpression plasmid that encodes two bioluminescent proteins from asingle mRNA: (i) a protein of interest fused to the C terminus offirefly luciferase (FLuc) and (ii) renilla luciferase (RLuc), with bothprotein ORFs preceded by internal ribosome entry site (IRES) el-ements. (Fig. 1B). Using this vector system, we expressed variousfragments of IKZF3 fused to firefly luciferase in transiently trans-fected 293T cells treated with pomalidomide or vehicle. We thenmeasured the ratio of firefly to renilla luciferase activity as a sur-rogate for destabilization of the fusion proteins by pomalidomide.As expected, firefly luciferase fusion proteins containing full-

    length IKZF3 (1–509) or the IKZF3 degron identified by Ebertand coworkers [IKZF3 (136–180)] were down-regulated bypomalidomide (Fig. 1C). Analysis of N-terminal and C-terminaldeletions of IKZF3 (136–180) showed that this degron could benarrowed further to IKZF3 (146–168) (Fig. 1C). The C-terminalboundary of the degron is clearly at residue 168 because elimi-nating this residue inactivated the IKZF3 degron [Fig. 1C, seeIKZF3 (144–168) and IKF3 (144–167)]. The N-terminal boundary is

    Significance

    Systems for degrading proteins at will are useful for a variety ofbiological experiments. Although a number of such systems havebeen described, they vary widely in terms of complexity, ease ofobtaining the necessary reagents, and costs. Moreover, no onesystem seems to work for all proteins, and the ideal system oftenmust be determined empirically. Thalidomide-like drugs (IMiDs)reprogram the ubiquitiously expressed cereblon ubiquitin ligasecomplex to degrade the lymphocyte transcription factors IKZF1 andIKZF3. Here, we show that an IKZF3-derived 25mer constitutes amodular degron that can be used to target heterologous proteinsfor destruction by IMiDs, which are widely available and cross theblood–brain barrier, in cell culture and in mouse experiments.

    Author contributions: V.K., K.J.B., H.C., and W.G.K. designed research; V.K., S.K.M., E.L.,A.C.W., K.J.B., and H.C. performed research; V.K., S.K.M., and E.L. contributed new re-agents/analytic tools; V.K., S.K.M., and W.G.K. analyzed data; and V.K. and W.G.K. wrotethe paper.

    Reviewers: D.N., University of Texas Southwestern; and C.L.S., Memorial Sloan–KetteringCancer Center.

    The authors declare no conflict of interest.

    Published under the PNAS license.1Present address: Discovery Biology, Tango Therapeutics, Cambridge, MA 02141.2Present address: Department of Biology, Peloton Therapeutics, Dallas, TX 75235.3To whom correspondence should be addressed. Email: [email protected].

    This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplemental.

    Published online January 25, 2019.

    www.pnas.org/cgi/doi/10.1073/pnas.1818109116 PNAS | February 12, 2019 | vol. 116 | no. 7 | 2539–2544

    BIOCH

    EMISTR

    Y

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

    http://crossmark.crossref.org/dialog/?doi=10.1073/pnas.1818109116&domain=pdfhttps://www.pnas.org/site/aboutpnas/licenses.xhtmlmailto:[email protected]://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/cgi/doi/10.1073/pnas.1818109116

  • approximate because some of the N-terminal IKZF3 residuesin IKZF3 (146–168) might be serving as spacers between the N-terminal firefly luciferase and the degron itself rather than directlycontacting cereblon. Unfortunately, the IKZF3 fragments we stud-ied, including both IKZF3 (136–180) and IKZF3 (146–168), weregrossly impaired as degrons relative to IKZF3 (1–509) when fusedto the N terminus of firefly luciferase (SI Appendix, Fig. S1), pre-venting us from further refining the N-terminal degron boundaryfree of neighboring sequences (see also below). In pomalidomidedose titration experiments, IKZF3 (144–168) slightly outperformedIKZF3 (146–168), which corresponds exactly to IKZF3 Zinc Finger2, with respect to pomalidomide-induced degradation of their cor-responding firefly luciferase fusion proteins (SI Appendix, Fig. S2 Aand B) and so this sequence (RPFQCNQCGASFTQKGNLLR-HIKLH) was used in the studies below.As expected, based on the behavior of full-length IKFZ3, the

    function of the IKZF3 (144–168) degron was crippled by intro-ducing a Q147H mutation (5) and was blocked in cells in whichcullin-dependent ligases were inactivated with the NEDD8-activating enzyme inhibitor MLN4924 or in which cereblonitself was deleted using CRISPR/Cas9 (Fig. 1 D and G). TheIKZF3 (144–168) degron was at least as capable as full-lengthIKZF3 at targeting firefly luciferase for destruction in cis in dose-titration experiments with three progressively more potent IMiDs(Fig. 1 E, F,H, and I). In time course experiments, degradation was

    apparent within the first 6 h and maintained for 24 h in cells con-tinuously exposed to an IMiD (Fig. 1 J–M). Finally, we confirmedthat varying the length of the linker between FLuc and IKZF1between 3 and 27 amino acid residues did not affect degradation bypomalidomide (SI Appendix, Fig. S2 C and D).

    IMiD-Dependent Degradation of Degron Fusion Proteins Occurs in Boththe Nucleus and Cytoplasm. The IKZF3 (144–168) degron alsotargeted green fluorescent protein (GFP) for pomalidomide-induced degradation when fused to the GFP C terminus andcould do so in both the cytosol and in the nucleus in the context ofGFP-IKZF3 (144–168) fusion proteins that were engineered tocontain strong nuclear localization or nuclear export signals (Fig. 2A and B) as determined by fluorescence activated cell sorting (Fig.2C) and immunoblot analysis (Fig. 2D). The IKZF3 (144–168)polypeptide did not function as a degron when fused to the Nterminus of GFP (SI Appendix, Fig. S3). Finally, this degron tar-geted exogenous HIF2α for pomalidomide-induced degradationwhen fused to the C terminus of HIF2α, leading to down-regulation of HIF2α-responsive mRNAs in cells in which endog-enous HIF2α was inactivated using CRISPR/Cas9 (Fig. 3).

    IMiD-Dependent Degradation of Oncogenic Fusion Protein InhibitsTransformation in Soft Agar Assay. Next, we sought to test whetherour IMiD-dependent degradation strategy could be used to modulate

    DMSOPomMLN4924MLN4942 + Pom

    0.0

    0.5

    1.0

    1.5

    DMSO10 nM100 nM

    Thal Len Pom0.0

    0.5

    1.0

    1.5IKZF3

    1 μM10 μM

    Thal Len Pom0.0

    0.5

    1.0

    1.5WT Degron

    IRES

    FLuc POI

    IRES

    RLuc

    FLuc

    DMSOPomMLN

    CRBN

    Vinculin

    DM

    SO

    Thal Len Pom

    FLuc

    Vinculin

    J K

    DMSO10 nM100 nM1 μM10 μM

    FLuc

    /RLu

    c (P

    pm/ D

    MS

    O)

    FLuc

    /RLu

    c (P

    om/D

    MS

    O)

    FLuc

    /Rlu

    c (P

    om/D

    MS

    O)

    0.0

    0.5

    1.0

    FLuc

    /Rlu

    c (P

    om/D

    MS

    O)

    DMSO 6h 12h 18h 24h0.0

    0.5

    1.0

    DMSO 6h 12h 18h 24h

    WT Degron

    FLuc

    Vinculin

    FLuc

    Vinculin

    A B

    C

    D

    G

    E

    H

    F

    I

    DMSO 6h 12h 18h 24h DMSO 6h 12h 18h 24h

    L M

    0.0 0.5 1.0

    144-166144-167148-168147-168146-168145-168144-168136-180

    1-509

    0.0 0.5 1.0

    136-160136-164136-168136-172136-176136-180

    1-509

    0.0 0.5 1.0

    156-180152-180148-180144-180140-180136-180

    1-509

    FLuc/RLuc (Pom/DMSO)

    DM

    SO

    Thal Len Pom

    FLuc

    Vinculin

    Pom

    Pom

    Pom

    Pom

    FLuc/RLuc (Pom/DMSO) FLuc/RLuc (Pom/DMSO)

    FLuc

    /RLu

    c (P

    om/D

    MS

    O)

    DegronCRBN

    WT WTQ147H+/+ +/+ -/-

    CRBN

    WT Q147H WT Degron:

    IKZF3

    Fig. 1. Mapping of IMiD-responsive IKZF3 degron. (A) IKZF3 schematic. Shaded boxes, zinc finger domains; dotted box, region containing the IMiD-responsivedegron. (B) Schematic for bicistronic reporter vector. POI, protein of interest. (C) Firefly luciferase (FLuc) values, normalized to Renilla luciferase values (RLuc), of293T cells transfected to express the indicated fragments of IKZF3 fused to the C terminus of FLuc using the vector in B and treated with 1 μM pomalidomide for24 h relative to cells treated with DMSO. (D) FLuc/Rluc values of 293T cells (CRBN+/+ or CRBN−/−) transfected to express IKZF3 144–168 (WT Degron) or IKZF3 144–168;Q147H (Q147H Degron) fused to FLuc and treated with 1 μM pomalidomide, 1 μM MLN4924, or both for 24 h relative to cells treated with DMSO. (E and F)FLuc/RLuc values of 293T cells transfected to express full-length IKZF3 (E) or IKZF3 144–168 (WT Degron) (F) fused to FLuc and treated with the indicated con-centrations of thalidomide, lenalidomide, or pomalidomide for 24 h relative to cells treated with DMSO. (G–I) Immunoblots of cells treated as in D–F, respectively.(J and L) FLuc/Rluc values of 293T cells transfected to express full-length IKZF3 (J) or IKZF3 144–168 (WT Degron) (L) fused to FLuc and treated with 1 μMpomalidomide for the indicated duration relative to cells treated with DMSO. (K and M) Immunoblots of cells treated as in J and L, respectively.

    2540 | www.pnas.org/cgi/doi/10.1073/pnas.1818109116 Koduri et al.

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

    https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/cgi/doi/10.1073/pnas.1818109116

  • oncoprotein stability and function. We lentivirally infected im-mortalized melanocytes, PmeL* cells (10), to express themicrophthalmia-associated transcription factor (MITF), which is

    a known melanoma oncoprotein capable of inducing anchorage-independent growth (11), fused to the WT degron, to theQ147H degron, or unfused. Pomalidomide suppressed the

    GFP GFPWT Degron

    GFP-NLSWT Degron

    GFP-NESWT Degron

    GFP

    H33352

    A

    C

    GFP

    GFP

    GFP

    GFP NLS

    NES

    Degron

    Degron

    Degron

    GGS

    GGS

    GGS

    B

    D

    ********

    ****

    GFP

    GFP-

    WT D

    egron

    GFP-

    NLS-

    WT D

    egron

    GFP-

    NES-

    WT D

    egron

    0.0

    0.5

    1.0

    Rel

    ative

    GFP

    Flu

    ores

    cenc

    eW

    ith P

    om T

    reat

    men

    t

    GFP

    Vinculin

    GFP

    GFP

    -WT

    Degr

    on

    GFP

    -NES

    -WT

    Degr

    on

    GFP

    -NLS

    -WT

    Degr

    on

    EV

    Pom

    Fig. 2. The core IMiD-responsive IKZF3 degron functions in different cellular compartments. (A) GFP fusion protein schematics. Degron, IKZF3 144–168; GGS,glycine-glycine-serine linker; NES, nuclear export signal; NLS, nuclear localization signal. (B) Fluorescence images of 293T cells stably infected to express theindicated GFP fusion proteins. H33352, nuclear dye. (Images captured with 60× objective lens.) (C and D) GFP fluorescence intensity, as determined by FACS(C), and immunoblot analysis (D) of 293T cells stably infected as in B and treated with 1 μM pomalidomide or DMSO for 24 h.

    A B

    C D

    Fig. 3. The core-IMiD responsive IKZF3 degron functions when fused to a heterologous transcription factor. (A) Immunoblot analysis of 786-O renal car-cinoma cells that underwent CRISPR/Cas9-mediated gene editing with a HIF2α sgRNA or control sgRNA, where indicated, before being stably infected with anlentivirus encoding sgRNA-resistant HIF2α-IKZF3(144–168) (HIF2α-WT Degron) or HIF2α-IKZF3(144–168;Q147H) (HIF2α-Q147H Degron) and then treated witheither DMSO or 1 μM pomalidomide for 24 h. (B–D) Abundance of the indicated mRNAs measured by qPCR in 786-O cells treated as in A after normalization toACTBmRNA and then to the values of the 786-O cells treated with DMSO and the control sgRNA. For all panels, data presented are means ± SD; **P < 0.01; ns,not significant. Two-tailed P values were determined by unpaired t test.

    Koduri et al. PNAS | February 12, 2019 | vol. 116 | no. 7 | 2541

    BIOCH

    EMISTR

    Y

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

  • anchorage-independent growth of PmeL* cells expressing the MITF-WT Degron fusion relative to cells expressing unfused MITF or theMITF-Q147H Degron fusion (SI Appendix, Fig. S4). These resultsindicate that our inducible strategy for heterologous protein degra-dation might be useful for studying oncoprotein function and de-pendence in preclinical cancer models.

    IKZF3 (144–168) Degron Functions in Multiple Orthotopic MouseModels. To ask if the IKZF3 (144–168) degron could functionin vivo, we stably infected 786-O renal carcinoma cells, MDA-MB-231 breast carcinoma cells, and HOG oligodendrogliomacells to produce Firefly Luciferase-IKZF3 (144–168) or FireflyLuciferase-IKZF3 (144–168; Q147H). Pomalidomide down-regulated the former and not the latter in all three cellular con-texts and did not have any effect on cellular viability at concen-trations up to 10 μM (Fig. 4 A–C and SI Appendix, Figs. S5 andS6A). Next these cells were grown orthotopically in the kidneys,mammary fat pads, and brains, respectively, of immunocompro-mised mice. Once tumors were established, as determined by se-rial bioluminescence imaging (BLI), mice were randomized toreceive 30 mg/kg pomalidomide or vehicle by oral gavage twicedaily for a total of four doses. As expected, pomalidomide causeda greater than 50% reduction in BLI signals in tumors containingthe Firefly Luciferase-IKZF3 (144–168) chimera, but not in tu-mors containing the Q147H chimera (Fig. 4 D–I). In drug washoutand rechallenge experiments, mice did not exhibit tolerance topomalidomide, with each rechallenge producing an ∼50% re-duction in Firefly Luciferase-IKZF3 (144–168) BLI signal (Fig. 4 Jand K). The tumors formed by the two different chimeras werecomparable in size at necropsy (SI Appendix, Fig. S6 B–D).

    DiscussionThere are a number of methods to regulate the transcription orstability of a protein of interest. Directly regulating protein sta-bility, however, creates an opportunity to more rapidly alter theabundance and, hence, function, of a protein of interest com-pared with methods that act at the transcriptional level. More-over, it will perhaps more faithfully mimic the effects of smallmolecule protein antagonists, especially those that act wholly orin part by destabilizing their targets. The approach designed herecomplements several ingenious approaches that have been de-scribed over the past decade for chemically stabilizing orchemically destabilizing proteins of interest.One system for chemically stabilizing a protein of interest in-

    volves fusing it to a variant of human FKBP12 (FKBP12*) that istargeted for degradation unless it is bound to an artificial ligandcalled Shield-1 (12). FKBP12* also has a point mutation (F36V)such that it binds to Shield-1 with 1,000-fold selectivity comparedwith wild-type FKBP12. The FKBP cassette is considerablylarger than the one described here (107 versus 25 amino acidresidues) and so it might be more prone to alter protein function.A modified version of this system allows the stabilization andrelease of an unfused protein of interest (traceless shield), but atthe expense of expressing two foreign proteins: an FRB (FKBP-Rapamycin-Binding) domain-UbN fusion and a FKBP12*-UbCprotein of interest fusion (13). In this embodiment, Shield-1stabilizes the protein of interest, which can then be released byrapamycin-induced reconstitution of the ubiquitin protease. Fi-nally, this technique has been further modified by Nabet et al(14), who showed that a heterobifunctional chemical ligandcomprised of AP1867 and an IMiD could be used to trigger thedegradation of proteins of interest fused to FKBP12*.A second method for chemically stabilizing proteins involves

    fusing the protein of interest to an unstable variant of Escherichiacoli dihydrofolate reductase (ecDHFR) that is stabilized in thepresence of trimethoprim (TMP) (15, 16). The biodistribution ofTMP has been better studied than that of Shield-1 and is knownto cross the blood–brain barrier. However, ecDHFR might prove

    to be immunogenic. Moreover, both the FKBP12*/Shield-1 andecDHFR/TMP systems require that Shield-1 and TMP, re-spectively, be continuously present until the moment when acuteprotein destabilization is desired. This could prove cumbersomeand costly, especially in animal models.To circumvent this problem, Wandless and coworkers (17)

    fused FKBP12 (F36V) to an additional 19 amino acids thatcreate a cryptic degron that is displayed only after Shield-1 isadded and showed that this chimera could be used to targetheterologous proteins for destruction with Shield-1. In a com-plementary approach, called “SMASh,” Lin and coworkers (18)fused a modular degron to proteins of interest with interveningsequences encoding the hepatitis C NS3 protease and an NS3protease cleavage site such that the degron is constitutively re-leased unless cells are treated with the protease inhibitorAsunaprevir. In this latter system, unlike the former, the stableversion of the protein is minimally altered, having only the short“stub” generated by protease cleavage. However, Asunaprevircan only act on newly synthesized proteins because mature formsof the protein will already have excised the artificial degron. Inaddition, Asunaprevir does not cross the blood–brain barrier.Another method for targeting heterologous proteins for destruc-

    tion exploits the naturally occurring plant hormones called auxins,which bind to the ubiquitin ligase SCFTIR1 and trigger the degrada-tion of members of the AUX/IAA family of transcriptional repressors(19, 20). Natsume and coworkers showed that the fusion of a 68-amino acid sequence derived from the IAA17 transcriptional re-pressor to a target heterologous protein rendered that protein sus-ceptible to rapid degradation in the presence of an auxin (19). Anadvantage of this system is that plant auxins do not have knowntargets in mammalian cells and are therefore unlikely to cause tox-icity. However, use of the auxin system in mammalian cells requiresthe enforced expression of the plant-derived TIR1 protein in additionto tagging the protein of interest with a 68-amino acid residue tag.The IKZF3 degron system described here has a number of po-

    tential advantages compared with previously described systems. First,it requires only a simple genetically encoded modification of a pro-tein of interest that can easily be introduced with a synthetic oligo-nucleotide and that leads to a modest peptidic addition analogous toappending an epitope tag. In fact, one can envision the IKZF3 25merserving as both a degron and an epitope tag if suitable antibodies canbe raised. Moreover, the IKZF3 degron system utilizes IMiDs, whichare well-studied, bioavailable, and easily obtainable. In particular,pomalidomide is bioavailable and active in the CNS. Finally, theIKZF3 degron system is understood in sufficient detail to allow forthe incorporation of additional specificity controls, including, asshown here, loss of function degron point mutants and pharmaco-logical and genetic inhibitors of the cereblon ubiquitin ligase.So far it appears the IKZF3 25mer degron functions best when

    fused to a target protein’s C terminus rather than its N terminus.As larger IKZF3 fragments do function as degrons when fused tothe N terminus of firefly luciferase, we presume this relates tosteric factors and protein folding. Determining whether this ap-parent bias is true will require testing additional fusion partnersin the future. For reasons we do not yet fully understand, theIKZF3 25mer degron does not target all proteins for destruction.For example, we have been unable to target the VHL andFOXP3 proteins for destruction with the IKZF3 25mer degron,whether fused at the N terminus or C terminus (data not shown).The empirically determined percentage of endogenous proteinsamenable to this approach after CRISPR/Cas9-mediated in-troduction of an IKZF1 (or 3)-derived degron is ∼30–40% (SIAppendix, Fig. S7). We hypothesize that the cereblon ubiquitinligase complex cannot bind to the IKZF degron in some fusionsdue to steric effects or, once bound, is not able to productivelyreach a surface lysine residue that is capable of being poly-ubiquitylated. Both of these issues might, theoretically, beaddressed by exploring different spacers between the degron and

    2542 | www.pnas.org/cgi/doi/10.1073/pnas.1818109116 Koduri et al.

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

    https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/cgi/doi/10.1073/pnas.1818109116

  • A

    D

    G

    B

    0

    2

    4

    6

    8

    10

    Biol

    umin

    esce

    nce

    (Pos

    ttrea

    tmen

    t : P

    retre

    atm

    ent) *

    MDA-MB-231

    0.0

    0.5

    1.0

    1.5

    2.0

    2.5

    Biol

    umin

    esce

    nce

    (Pos

    ttrea

    tmen

    t : P

    retre

    atm

    ent) *

    786-O

    E

    H

    J K

    0 3 6 9 12 15 18107

    108

    109

    1010

    Days

    Biol

    umin

    esce

    nce

    (pho

    tons

    /sec

    )

    MDA-MB-231

    WTQ147H

    0 3 6 9 12 15 18108

    109

    1010

    Days

    Biol

    umin

    esce

    nce

    (pho

    tons

    /sec

    )

    786-O

    WT

    Q14

    7H WT

    Q14

    7H

    Pre Post786-O

    WT

    Q14

    7H WT

    Q14

    7H

    Pre PostMDA-MB-231

    C

    F

    I

    Pre Post

    Pre Post

    WT

    Q147H

    HOG

    - + - +

    FLuc

    Pom

    Vinculin

    786-O

    - + - +

    FLuc

    Pom

    Vinculin

    MDA-MB-231

    - + - +

    FLuc

    Pom

    Vinculin

    0

    2

    4

    6

    8

    HOG

    **WTQ147H

    Biol

    umin

    esce

    nce

    (Pos

    ttrea

    tmen

    t : P

    retre

    atm

    ent)

    HOG

    N=12 N=16 N=8

    Degron WT Q147H Degron WT Q147H Degron WT Q147H

    Fig. 4. The core IKZF3 IMiD-responsive degron functions in vivo. (A–C) Immunoblot analysis of 786-O renal carcinoma cells (A), MDA-MB-231 breast carci-noma cells (B), and HOG oligodendroglioma cells (C) stably infected to express FLuc-IKZF3 144–168 (WT) or FLuc-IKZF3 144–168;Q147H (Q147H) and treatedwith 1 μM pomalidomide or DMSO for 24 h. (D–F) Representative BLI of orthotopic tumors formed by cells as in A–C, respectively, before (pre) and after (post)systemic administration of 30 mg/kg pomalidomide by oral gavage twice a day for 48 h (4 doses). (G–I) Quantification of BLI signals from mice treated as in D–F, respectively. (J and K). Quantification of BLI signals from representative mice as in G and H, respectively, over time. Yellow bars indicate treatment withpomalidomide as above. Note logarithmic scale. For all panels, data presented are means ± SD; **P < 0.05; **P < 0.01. Two-tailed P values were determinedby unpaired t test.

    Koduri et al. PNAS | February 12, 2019 | vol. 116 | no. 7 | 2543

    BIOCH

    EMISTR

    Y

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

  • the heterologous protein, changing the placement of the degronrelative to the heterologous protein, extending the length of thedegron, and/or incorporating lysine residues into the spacer (seealso SI Appendix, Fig. S7). Similarly, we have not been able totarget the VHL protein for degradation with a SMASh tag, de-spite drug-induced degron retention, suggesting that differentsystems may or may not work for any given protein. A strength orweakness of the IKZF3 degron system, depending upon theapplication, is that mouse cereblon does not engage IKZF3 whenbound to an IMiD. However, a mouse with a humanized cere-blon has been developed (21).Two recent studies showed that indisulam redirects an ubiq-

    uitin ligase containing the substrate adaptor DCAF15 to degradeRBM39, much as the IMiDs redirect the cereblon ligase to targetIKZF1 and IKZF3 for destruction (22, 23). Moreover, it is nowclear that some IMiDs can also direct the destruction of caseinkinase 1α, ERF3, and SALL4 (24–26). Mapping the relevantdegrons in RBM39, casein kinase 1α, and ERF3A, and comparingtheir behavior to the IKZF1/3 degron, might yield additionalportable degrons for creating tunable proteins. The IKZF3

    degron system might also be improved by random mutagenesis of theIKZF3 degron followed by positive selection for enhanced binding tocereblon in the presence of an IMiD, as well as by next generationIMiDs that are more potent than existing compounds.

    Materials and MethodsA detailed description of the cell lines, plasmids, and antibodies used in thispaper can be found in the SI Appendix. Detailed protocols for lentiviral in-fection, immunoblotting, luciferase assays, FACS analysis, fluorescence mi-croscopy, real-time qPCR, soft agar assays, and orthotopic xenograftexperiments can also be found in SI Appendix. All experimental proceduresrelated to orthotopic xenografts were approved by the Institutional AnimalCare and Use Committee of Dana–Farber Cancer Institute.

    ACKNOWLEDGMENTS. We thank Benjamin Ebert, Eric Fischer, Egon Ogris,and Gromoslaw Smolen for critical reading of the manuscript; members ofthe W.G.K. laboratory for useful discussions and comments on the manu-script; Matthew Oser for help with fluorescence microscopy; Eric Fischer forsharing unpublished data; and Gang Lu and Rizwan Haq for reagents. Thiswork was supported by NIH Grants (to W.G.K.), T32 NIH Training GrantCA009172 and American Society of Hematology Research Training Award(to V.K.). W.G.K. is a Howard Hughes Medical Institute Investigator.

    1. Lampson MA, Kapoor TM (2006) Targeting protein stability with a small molecule.Cell 126:827–829.

    2. Hannah J, Zhou P (2015) Methods: A small-molecule SMASh hit. Nat Chem Biol 11:637–638.

    3. Rakhit R, Navarro R, Wandless TJ (2014) Chemical biology strategies for post-translational control of protein function. Chem Biol 21:1238–1252.

    4. Krönke J, et al. (2014) Lenalidomide causes selective degradation of IKZF1 and IKZF3in multiple myeloma cells. Science 343:301–305.

    5. Lu G, et al. (2014) The myeloma drug lenalidomide promotes the cereblon-dependentdestruction of Ikaros proteins. Science 343:305–309.

    6. Gandhi AK, et al. (2014) Immunomodulatory agents lenalidomide and pomalidomideco-stimulate T cells by inducing degradation of T cell repressors Ikaros and Aiolos viamodulation of the E3 ubiquitin ligase complex CRL4(CRBN.). Br J Haematol 164:811–821.

    7. Chamberlain PP, et al. (2014) Structure of the human cereblon-DDB1-lenalidomidecomplex reveals basis for responsiveness to thalidomide analogs. Nat Struct Mol Biol21:803–809.

    8. Fischer ES, et al. (2014) Structure of the DDB1-CRBN E3 ubiquitin ligase in complexwith thalidomide. Nature 512:49–53.

    9. Yen HC, Xu Q, Chou DM, Zhao Z, Elledge SJ (2008) Global protein stability profiling inmammalian cells. Science 322:918–923.

    10. Garraway LA, et al. (2005) Integrative genomic analyses identify MITF as a lineagesurvival oncogene amplified in malignant melanoma. Nature 436:117–122.

    11. Haq R, et al. (2013) Oncogenic BRAF regulates oxidative metabolism via PGC1α andMITF. Cancer Cell 23:302–315.

    12. Banaszynski LA, Chen LC, Maynard-Smith LA, Ooi AG, Wandless TJ (2006) A rapid,reversible, and tunable method to regulate protein function in living cells usingsynthetic small molecules. Cell 126:995–1004.

    13. Lin YH, Pratt MR (2014) A dual small-molecule rheostat for precise control of proteinconcentration in mammalian cells. ChemBioChem 15:805–809.

    14. Nabet B, et al. (2018) The dTAG system for immediate and target-specific proteindegradation. Nat Chem Biol 14:431–441.

    15. Iwamoto M, Björklund T, Lundberg C, Kirik D, Wandless TJ (2010) A general chemicalmethod to regulate protein stability in the mammalian central nervous system. ChemBiol 17:981–988.

    16. Sheridan RM, Bentley DL (2016) Selectable one-step PCR-mediated integration of adegron for rapid depletion of endogenous human proteins. Biotechniques 60:69–74.

    17. Bonger KM, Chen LC, Liu CW, Wandless TJ (2011) Small-molecule displacement of acryptic degron causes conditional protein degradation. Nat Chem Biol 7:531–537.

    18. Chung HK, et al. (2015) Tunable and reversible drug control of protein production viaa self-excising degron. Nat Chem Biol 11:713–720.

    19. Natsume T, Kiyomitsu T, Saga Y, Kanemaki MT (2016) Rapid protein depletion inhuman cells by auxin-inducible degron tagging with short homology donors. CellReports 15:210–218.

    20. Nishimura K, Fukagawa T, Takisawa H, Kakimoto T, Kanemaki M (2009) An auxin-based degron system for the rapid depletion of proteins in nonplant cells. NatMethods 6:917–922.

    21. Fink EC, et al. (2018) CrbnI391V is sufficient to confer in vivo sensitivity to thalidomideand its derivatives in mice. Blood 132:1535–1544.

    22. Uehara T, et al. (2017) Selective degradation of splicing factor CAPERα by anticancersulfonamides. Nat Chem Biol 13:675–680.

    23. Han T, et al. (2017) Anticancer sulfonamides target splicing by inducing RBM39degradation via recruitment to DCAF15. Science 356:eaal3755, and erratum (2017),356:eaan7977.

    24. Matyskiela ME, et al. (2016) A novel cereblon modulator recruits GSPT1 to the CRL4(CRBN) ubiquitin ligase. Nature 535:252–257.

    25. Krönke J, et al. (2015) Lenalidomide induces ubiquitination and degradation of CK1αin del(5q) MDS. Nature 523:183–188.

    26. Donovan KA, et al. (2018) Thalidomide promotes degradation of SALL4, a transcrip-tion factor implicated in Duane radial ray syndrome. eLife 7:e38430.

    2544 | www.pnas.org/cgi/doi/10.1073/pnas.1818109116 Koduri et al.

    Dow

    nloa

    ded

    by g

    uest

    on

    June

    14,

    202

    1

    https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1818109116/-/DCSupplementalhttps://www.pnas.org/cgi/doi/10.1073/pnas.1818109116