Development of a First-in-Class RIPK1 Degrader to Enhance Antitumor Immunity

The scaffolding function of receptor interacting protein kinase 1 (RIPK1) confers intrinsic and extrinsic resistance to immune checkpoint blockades (ICBs) and has emerged as a promising target for improving cancer immunotherapies. To address the challenge posed by a poorly defined binding pocket within the intermediate domain, we harnessed proteolysis targeting chimera (PROTAC) technology to develop a first-in-class RIPK1 degrader, LD4172. LD4172 exhibited potent and selective RIPK1 degradation both in vitro and in vivo. Degradation of RIPK1 by LD4172 triggered immunogenic cell death (ICD) and enriched tumor-infiltrating lymphocytes and substantially sensitized the tumors to anti-PD1 therapy. This work reports the first RIPK1 degrader that serves as a chemical probe for investigating the scaffolding functions of RIPK1 and as a potential therapeutic agent to enhance tumor responses to immune checkpoint blockade therapy.


Introduction
Immune checkpoint blockades (ICBs) have transformed cancer therapy by disrupting inhibitory signals that typically weaken robust anti-tumor immune responses 1 .Despite the success of ICBs, a signi cant subset of patients remain unresponsive to ICBs owing to various immuno-resistances, which are often propagated by cancer cells 2 .The exploration of combinational therapies involving novel immunomodulatory agents with anti-PD-1/PD-L1 has emerged as a promising approach to overcome intrinsic or acquired resistance to ICBs 3 .
Receptor-interacting protein kinase 1 (RIPK1) regulates cell fate through its kinase-dependent andindependent functions and controls proin ammatory responses downstream of multiple innate immune pathways, including those initiated by tumor necrosis factor-α (TNF-α), toll-like receptor (TLR) ligands, and interferons (IFNs) 4 .Recent studies have shown that genetic knockout of RIPK1 in cancer cells signi cantly sensitizes tumors to anti-PD1, leading to drastic changes in the tumor microenvironment, including increased in ltration of effector T cells, reduction of immunosuppressive myeloid cells, and enhanced immunostimulatory cytokine secretion [5][6][7] .Notably, RIPK1-mediated ICB resistance requires ubiquitin scaffolding function through its intermediate domain instead of its kinase function.Genetic depletion of RIPK1, but not inactivation of its kinase domain, sensitizes B16F10 tumors to ICBs 5,6 .Hence, targeting RIPK1 scaffolding functions holds promise as a strategy to synergize with ICBs to promote antitumor immunity.While all RIPK1 inhibitors developed thus far have focused on inhibiting kinase function for the treatment of autoimmune, in ammatory, and neurodegenerative diseases 8 , the development of inhibitors speci cally targeting the intermediate domain of RIPK1 remains challenging due to the absence of a wellde ned binding pocket within this domain.A proteolysis-targeting chimera (PROTAC) is a heterobifunctional molecule that binds both a targeted protein and an E3 ubiquitin ligase to facilitate the formation of a ternary complex, leading to ubiquitination and ultimate degradation of the target protein 9 .
Using PROTAC technology, we developed LD4172, a rst-in-class highly potent and speci c RIPK1 degrader, to abolish the scaffolding functions of RIPK1.We showed that LD4172 potently induces RIPK1 degradation with high speci city and substantially sensitizes multiple preclinical cancer models to anti-PD1 therapy.

Development of RIPK1 Degrader LD4172
To develop RIPK1 PROTACs, we tested two types of RIPK1 binders: type II RIPK1 inhibitor 1 (also referred as T2I), which targets both the ATP-binding pocket and the allosteric hydrophobic back pocket 10 , and type III RIPK1 inhibitor 2, which only binds the hydrophobic back pocket of the kinase domain 11 .To identify the ideal attachment sites for PROTAC linkers, we performed molecular docking of 1 with RIPK1, which revealed a solvent-exposed ethyl group in the 7H-pyrrolo [2,3-d] pyrimidine ring (Fig. 1A).The cocrystal structure of RIPK1 in complex with 2 (PDB: 6R5F) showed that the oxadiazole moiety in 2 was solvent-exposed, providing an ideal exit vector for linker attachment (Fig. 1B).
To identify an appropriate E3 ligase pair for RIPK1 degradation, we synthesized a small library by conjugating RIPK1 binders 1 and 2 to ligands for different E3 ligases, including Cereblon (CRBN), von Hippel-Lindau tumor suppressor (VHL), murine double minute 2 (MDM2), and a hydrophobic adamantane tag (Fig. 1C).As shown in Fig. 1D-E, PROTACs formed by conjugating type II inhibitor 1 to a VHL ligand induced the most e cient degradation of RIPK1 in Jurkat cells.
We further optimized RIPK1 PROTACs through linker lengths ranging from two to 14 methylene groups (Fig. 1F).We found that PROTACs with linker lengths of more than six methylenes were able to effectively degrade > 90% of RIPK1 at 1 µM after 24 h incubation in Jurkat cells, showing a monotonic trend (Fig. 1F-G).Consistently, the PROTACs exhibited maximal degradation with an 8-to 10-methylene linker and signi cantly reduced potency with either shorter or longer linkers in B16F10 mouse melanoma cells (Fig. 1F-G).Considering the potency of both human and mouse cells, we chose a combination of a type II RIPK1 binder, a VHL ligand, and a 10-methylene linker as the lead RIPK1 degrader, designated as LD4172 (Fig. 2A).

LD4172 Induces Potent RIPK1 Degradation In Vitro
LD4172 induced potent RIPK1 degradation (concentration to induce 50% protein degradation DC 50 = 4 to 400 nM) in a panel of human and mouse cancer cell lines (Fig. 2B-C, S1).To investigate the kinetics of LD4172-induced RIPK1 degradation and resynthesis rates, Jurkat and B16F10 cells were treated with LD4172 for different time points, followed by washout after 24 h.With 1 µM LD4172 treatment, > 90% of RIPK1 was degraded within 2 and 4 h in Jurkat and B16F10 cells, respectively (Fig. 2D-E).Upon removal of LD4172, the re-synthesis half-life of RIPK1 was ~ 48 h and ~ 24 h in Jurkat and B16F10 cells, respectively (Fig. 2D-E).Collectively, these data demonstrated that LD4172 is a potent RIPK1 degrader with rapid and sustained effects in vitro.

LD4172 Engages RIPK1 and Forms a Ternary Complex
To elucidate the formation of a binary complex during RIPK1 degradation, we developed a competitive NanoBRET (Nano-Bioluminescence Resonance Energy transfer)-based target engagement (TE) assay to quantify the binding between RIPK1 and LD4172 in cells 12,13 .First, we developed a RIPK1 tracer by conjugating the LD4172 warhead T2I with a BODIPY-590 uorescent dye, dubbed T2-590 (refer to Supporting Information for details).The dissociation equilibrium constant (K d ) between the tracer and RIPK1 was determined to be 0.5 µM by titrating the tracer in HEK293T cells expressing a nLuc-RIPK1 fusion protein.Subsequently, with the tracer concentration at its K d value, LD4172 competed with the tracer with an IC 50 value of 3.7 µM (Fig. 2F).Based on the Cheng-Prusoff equation, the apparent K i between LD4172 and RIPK1 in the cells was 1.9 µM.Using a recombinant human RIPK1 protein, we measured the biochemical K i between LD4172 and human RIPK1 to be 4.8 nM (Fig. 2G), which is 395 folds smaller than the corresponding K i in cells.This is usually expected, considering that the large molecular weight of LD4172 may lead to poor cellular permeability.However, the fact that the DC 50 values of LD4172 are much smaller than its TE IC 50 values demonstrates the sub-stoichiometric degradation of RIPK1 induced by LD4172.Additionally, we synthesized an LD4172 negative control (LD4172-NC, also referred to as NC, Fig. 2A) using a VHL ligand diastereomer that does not bind to VHL.As expected, LD4172-NC showed TE similar to that of LD4172 (Fig. 2F).
To test whether LD4172 induces ternary complex formation with RIPK1 and VHL, we co-transfected HEK293 cells with nLuc-RIPK1 and VHL-Halo labeled with BODIPY-590 dye.The addition of LD4172, but not LD4172-NC, induced NanoBRET between RIPK1 and VHL, demonstrating the formation of a ternary complex among {RIPK1-LD4172-VHL} (Fig. 2H).LD4172 Degrades RIPK1 with High Speci city Through Ubiquitin-Proteasome System (UPS) The mechanistic action of PROTACs involves bringing the protein of interest (POI) into close proximity to E3 ligase, which ubiquitinates the POI for degradation by the proteasome.To con rm that LD4172 functions through this mechanism, we disrupted ternary complex formation by introducing an excess of RIPK1 or VHL ligands, which led to attenuation of RIPK1 degradation induced by LD4172.Moreover, blocking Cul2 E3 ligase with the neddylation inhibitor MLN4924 or inhibiting the proteasome with car lzomib reversed the potent degradation of RIPK1 by LD4172 in both Jurkat and B16F10 cells (Fig. 2I).These ndings indicate that LD4172 induces protein degradation through ternary complex formation and the UPS machinery.
The RIPK1 binder used in LD4172 is a typical type II kinase inhibitor bound to some off-target kinases, including TrkA, Flt1, Flt4, Ret, Met, Mer, Fak, FGFR1, and MLK1 10 .To evaluate the speci city of LD4172, we performed mass spectrometry (MS) analysis of the whole cellular proteome.Because Jurkat and B16F10 cells lack expression of all the aforementioned off-target kinases, MDA-MB-231 cells were chosen and treated with either LD4172 (200 nM) or LD4172-NC (200 nM) for 6 h.Among the > 10,000 proteins detected, RIPK1 was the only protein degraded by LD4172 (the red dot in Fig. 2J), and no degradation of off-target kinases was observed (blue dots in Fig. 2J, Supporting data values).This nding is consistent with previous studies showing that PROTACs with promiscuous target protein binders can achieve enhanced selectivity through protein-protein interactions with the E3 ligase involved 14 .

LD4172 Sensitizes B16F10 Cells to TNFα-Mediated Apoptosis
In contrast to situations where RIPK1 is kinase-dead, genetic deletion of RIPK1 has been found to trigger apoptosis both in vitro and in vivo 15 .To investigate apoptosis in the B16F10 mouse melanoma cell model and its correlation with the mechanism of RIPK1 downregulation rather than kinase inhibition, we employed various tool molecules, including LD4172, T2I, TNFα, and the pan-caspase inhibitor Z-VAD-FMK.Results demonstrated that signi cant cell death (Fig. 3A-C), particularly apoptosis, was induced by the combination of TNFα and LD4172, as evidenced by enhanced surface exposure of phosphatidylserine (Fig. 3A), along with increased expressions of cleaved caspase3/7 and PARP (Fig. 3B-C), which can be reversed with Z-VAD-FMK treatment (Fig. 3A-C).In contrast, inhibition of RIPK1 kinase activity by T2I did not trigger TNFα-mediated apoptosis (Fig. 3A-C).Additionally, apoptosis induced by LD4172 plus TNFα involves membrane ruptures, as indicated by the enhanced production of ATP in extracellular environments (Fig. 3D), loss of nuclear HMGB1 (High Mobility Group Box 1, Fig. 3D-E) and downregulated calreticulin (Fig. 3D).
LD4172 Exhibits Acceptable Pharmacokinetic Properties and Tissue-selective RIPK1 Degradation LD4172 has half-lives of 21.1 and 9.7 minutes in human and mouse liver S9 fractions, respectively, corresponding to intrinsic clearance (CL int ) of 32.8 and 71.6 µL•min − 1 •mg − 1 protein.In human primary hepatocytes, the half-life of LD4172 is 56.3 minutes, which corresponds to a predicted CL int of 15.6 mL•min − 1 •kg − 1 in human.It should be noted that the predicted intrinsic clearance in primary hepatocytes was > 2,000 times slower than that in liver S9 fractions, possibly due to the low membrane permeability of LD4172, which protects it from being metabolized (Table 1).
To investigate the pharmacodynamics of LD4172 in vivo, we administered LD4172 via the i.p. route and observed a 60% reduction in RIPK1 levels in tumors (20 mg/kg, b.i.d., i.p.) (Fig. 4B-C).In contrast, less than 50% RIPK1 degradation was observed in the spleen, and no signi cant RIPK1 degradation was observed in other organs, including the lymph nodes, PBMCs, lungs, and bone marrow (Fig. 4B-C).
The hERG channel inhibition assay is a commonly used safety assay to identify compounds that exhibit cardiotoxicity related to hERG inhibition in vivo.LD4172 exhibited no obvious inhibition of hERG, even at 30 µM (Table 1), indicating that LD4172 has a good safety margin for hERG inhibition.

LD4172 Sensitizes Tumors to Anti-PD1 Therapy
Utilizing CRISPR-Cas9 technology, we generated RIPK1-knockout (KO) B16F10 cells and implanted them into mice to examine their response to anti-PD1 treatment (Fig. 4D-E).Align with previous reports [5][6][7] , our ndings demonstrated that tumors lacking RIPK1 exhibit heightened sensitivity to anti-PD1 treatment (Fig. 4E).Subsequently, we explored whether pharmacological degradation of RIPK1 could replicate the effects observed in RIPK1-null B16F10 tumors.Consistent with the genetic study, mice treated with anti-PD1 or LD4172 alone showed tumor progression similar to that of the untreated mice.However, LD4172 sensitized B16F10 tumors to anti-PD1 therapy (Fig. 4F-I), with long-term administration of LD4172 showing no impact on mouse body weight (Fig. 4J).To test whether inhibition of RIPK1 kinase activity also enhances tumor responses to ICB therapy, we treated B16F10 xenograft tumors with the RIPK1 kinase inhibitor T2I, alone or in combination with anti-PD1.Unlike the RIPK1 degrader LD4172, the RIPK1 kinase inhibitor T2I failed to sensitize B16F10 tumors to anti-PD1 treatment (Fig. 4K).
We also tested a syngeneic MC38 colon cancer model, which exhibited a limited response to anti-PD1 treatment.Consistent with the B16F10 tumor model, LD4172 substantially sensitized MC38 tumors to anti-PD1 therapy (Fig. S2A-B).

LD4172 Triggers Immunogenic Cell Death in B16F10 Tumor
To understand the observed synergistic effects of LD4172 and anti-PD1, we administered vehicle, LD4172, anti-PD1, or a combination of LD4172 and anti-PD1 in C57BL/6J mice with B16F10 tumors for a short duration.A ve-day treatment with LD4172 was su cient to induce substantial degradation of RIPK1 in the tumor (Fig. 5A, 1st column).Consistent with the in vitro ndings, LD4172 also triggered signi cant cell death in the tumor (Fig. 5A, 2nd column).Importantly, a notable increase in cleaved caspase 3/7 levels was observed in the LD4172-treated tumors, indicating the occurrence of apoptosis (Fig. 5A, 3rd and 4th columns).While apoptotic cell death was traditionally considered non-immunogenic, accumulating experimental data have revealed its potential to drive immune cell in ltration and anticancer immunity [16][17][18][19] .Supporting the activation of immunogenic apoptosis, we observed a signi cant increase in plasma HMGB1 levels (Fig. 5B) and enhanced exposure of calreticulin on the surface of B16F10 tumor cells (Fig. 5A, 5th column).

LD4172 Enhances Anti-tumor Immunity
To elucidate how the combination of LD4172 plus anti-PD1 promotes anti-tumor immunity, multiparameter ow cytometry was employed to evaluate tumor-in ltrating lymphocytes (TILs) within the tumor microenvironment (TME) of mice receiving different treatments (Fig. S3).Initially, we con rmed the successful blockade of PD1 on T cells (CD8 + PD1+) with an anti-PD1 antibody (Fig. 5C).LD4172induced ICD led to a notable expansion of CD4 + T cells (Fig. 5A, 7th column, and 5D), conventional dendritic cells (cDC1, CD45 + CD11C + IAIE + XCR1+, Fig. 5E), and macrophages (CD45 + CD11b + F4/80+, Fig. 5A, 8th column, and 5F) within the TME, all of which contribute to antigen presentation and cytotoxic T cell priming and activation.In addition, combined therapy with LD4172 and anti-PD1 not only induced extensive TIL in ltration (Fig. 5D-H) but also signi cantly enhanced anti-PD1 positivity in immunologically cold B16F10 tumors, as demonstrated by increased in ltration of cytotoxic CD8 + T cells (CD8 + IFN-γ+, Fig. 5A, 6th column, and 5G-H) and decreased in ltration of FOXP3 + T regulatory cells (Fig. 5A, 7th column) within the TME.Additionally, to con rm the contribution of CD8 + T cells to the antitumor effect, we conducted a CD8 + T cell depletion experiment, revealing that the synergy between anti-PD1 and LD4172 was nulli ed in the absence of CD8 + T cells (Fig. 5I).Results from the cytokine array pro ling of plasma further supported synergistic effects of combined treatment, showing a signi cant enhancement in the production of immune cell proliferation cytokines, including IFN-γ and IL2 (Fig. 5J).

Discussion
RIPK1 is a critical regulator involved in cellular processes and proin ammatory signaling and exerts its effects through both kinase-dependent and kinase-independent mechanisms.In particular, its ubiquitin scaffolding function through K376 has been implicated in conferring intrinsic and extrinsic resistance to immune checkpoint blockade and is a potential target for cancer immunotherapy 5,6 .However, the development of inhibitors that speci cally target the intermediate scaffolding domain is challenging because of the absence of a well-de ned binding pocket.In this study, we used PROTAC technology to address this limitation and successfully developed a rst-in-class RIPK1 degrader, LD4172, with potent and speci c RIPK1 degradation both in vitro and in vivo.Notably, LD4172 also exhibited therapeutic e cacy in vivo, leading to RIPK1 degradation in tumors and demonstrating a synergistic effect in inhibiting tumor growth when combined with anti-PD1 treatment.These ndings highlight the potential of developing RIPK1 degraders as a promising therapeutic strategy to enhance the antitumor immunity of anti-PD1 blockades.
The suboptimal pharmacokinetic properties of LD4172, particularly its high in vivo clearance and low unbound plasma drug concentration, present challenges for achieving optimal RIPK1 degradation in vivo following intraperitoneal administration.Interestingly, intratumoral administration of LD4172 improved RIPK1 degradation in the tumors (Fig. S4), suggesting that the incomplete degradation of RIPK1 in tumors via i.p. injections may be attributed to poor penetration and/or accumulation of LD4172 in tumor tissues.Additionally, LD4172 had poor permeability in cells based on the divergence observed between biochemical and cellular assays for target engagement (Fig. 2G-H).To enhance the pharmacodynamic properties of LD4172, optimization of medicinal chemistry is necessary to further improve its physicochemical and pharmacokinetic characteristics.Previous studies have reported successful optimization strategies, including optimizing linker structures 20 , E3 ligands, and kinase warheads 21 , introducing intramolecular hydrogen bonds 22 , and converting the drug into a prodrug form 23 .Our future work will implement these optimization approaches to enhance membrane permeability and the overall pharmacokinetic pro le of LD4172, thereby improving its potency for RIPK1 degradation in vivo.
As a rst-in-class therapeutic modality, the on-target toxicity pro les of RIPK1 degraders remain unclear.Although mice are a convenient model system for exploring the functions of cellular signaling pathways, human genetics provides the best models of human diseases and guides the selection of new targets for drug discovery 24 .Unlike Ripk1 knockout mice, which die at 1-3 days of age due to their widespread roles in multiple tissues and organs 25 , homozygous loss-of-function RIPK1 mutations are well tolerated in humans 26 .Patients with complete loss of RIPK1 protein only showed symptoms con ned to the immune system, with primary immunode ciency and/or intestinal in ammation 26 .Additionally, the phenotypes of genetic knockout may be different from those of chemical-induced protein degradation, which is acute and transient and can be tissue-speci c 27 .Although the safety pro les of RIPK1 degraders remain to be tested in future clinical studies, human genetic data suggest that pharmacological RIPK1 degradation is potentially safe and tolerable, especially with transient intervention in well-controlled clinical settings.Furthermore, we found that decreasing RIPK1 dosing frequency by 50% resulted in a similar tumor inhibition effect when combined with anti-PD1 (Fig. S5), suggesting that the safety pro le of RIPK1 degraders can be further improved by optimizing the dosing regimen.Importantly, we primarily observed RIPK1 degradation in tumors to a lesser extent in the spleen and no degradation in other organs (Fig. 4B-C).In contrast to small-molecule inhibitors, protein degraders can achieve tissue-speci c target degradation by leveraging tissue-speci c E3 ligases.One notable example is a BCL-X L degrader developed by the groups of Zhou and Zheng, which spares platelets due to the low expression of VHL in platelets 28 .However, according to proteinatlas.org,VHL expression is ubiquitous in major organs, which does not explain the tumor-selective RIPK1 degradation induced by LD4172.Albumin accounts for ~ 60% of the total plasma protein and preferentially accumulates in tumors due to the high demand for amino acids and energy 29,30 .Considering that 98.6% of LD4172 is bound to plasma proteins (Table 1), it is possible that LD4172 piggybacks albumin accumulation in tumors to achieve tumor-selective RIPK1 degradation, further alleviating potential toxicity concerns associated with RIPK1 degradation in normal tissues.
In summary, we developed a rst-in-class RIPK1 degrader with high degradation speci city in cells and tumor selectivity in vivo.Our work not only provides a high-quality chemical probe to explore the effects of RIPK1 degradation in biology, but also the rst proof-of-concept study demonstrating that pharmacological degradation of RIPK1 synergizes with anti-PD1 to overcome resistance to ICBs by enhancing the in ltration of effector immune cells and promoting the secretion of immunostimulatory cytokines.Considering the predicted safety pro le of RIPK1 degradation based on human genetics, we envision that further optimized RIPK1 degraders have the potential to improve cancer immunotherapy.

Cell Lines
Human and mouse hematopoietic cell lines, namely Jurkat, Ramos, THP1, U937, TK1, and A20, and mouse melanoma B16F10 cell lines, were procured from ATCC.MC38 colon carcinoma and H2023 lung carcinoma cells were provided by Dr. Weiyi Peng.A375 melanoma cells were acquired from the Cell Core at the MD Anderson Cancer Center.Human breast cancer cells MDA-MB-231 and BT474 were obtained from Baylor College of Medicine Cell Core, whereas mouse breast carcinoma 4T1 cells were a gift from Dr. Xiang Zhang.

Induction of Bone Marrow-Derived Macrophages and Dendritic Cells
To isolate bone marrow cells, femur and tibia bones were dissected from 6-to 8-week-old C57BL/6J mice.

Western Blotting
Cells were seeded into six-well plates at a density of 5×10 5 cells/mL in 2 mL of complete culture medium.
Following an overnight adaptation period, cells were treated with serially diluted LD4172 compounds for 24 h.After treatment, whole-cell lysates were prepared using a lysis buffer (1×RIPA supplemented with protease and phosphatase inhibitor cocktail).Protein concentrations in the lysates were measured using the BCA protein assay.Subsequently, equal amounts of protein (20 µg) from each sample were loaded onto a sodium dodecyl sulfate-polyacrylamide gel and separated by electrophoresis (Bio-Rad) at 120 V for 1.5 hours.The separated proteins were then transferred to a polyvinylidene uoride (PVDF) membrane using a Transblot Turbo system (Bio-Rad).

Apoptosis detection using FITC-conjugated Annexin V/PI
Apoptosis quanti cation was conducted utilizing a FITC-conjugated Annexin V/PI assay kit (556547, BD Biosciences) and analyzed through ow cytometry.Brie y, 2×10 5 of B16F10 cells were seeded onto sixwell plates and treated as speci ed for 72 hours at 37°C.Treated and untreated cells were harvested, washed with PBS, and resuspended in 100 µl of binding buffer.Subsequently, cells were stained with PI (50 µg/ml) and FITC-conjugated Annexin V (10 mg/ml) for 15 minutes at room temperature in the dark.After adding another 400 µl of binding buffer, the cells were subjected to LSR II Flow cytometer (BD Biosciences) for analysis, and ow cytometry data were processed using the FlowJo software.

Extracellular ATP Assay
To detect ATP secretion after treating B16F10 cells with speci ed treatments, the RealTime-Glo™ Extracellular ATP Assay (GA5010, Promega) was conducted following the manufacturer's protocol.In brief, 1×10 4 B16F10 cells were plated into each well of an opaque 96-well plate, after 72 hours of treatment, 1X assay reagent was dispensed, and luminescence was recorded at regular intervals.

TR-FRET Biochemical Binding Assay
A time-resolved uorescence resonance energy transfer (TR-FRET) assay was performed to evaluate the binding of the indicated compounds and RIPK1 by competition with a BODIPY-FL labeled RIPK1 tracer (Supplementary Information, T2I-488).The assay was performed in 20 µL assay buffer (50 mM Tris, pH7.5, 0.1% Triton X-100, 0.01% BSA, and 1mM DTT) with 0.3 nM Tb-anti-GST (61GSTTLF, Cisbio), 2 nM GST-RIPK1 (R07-11G-10, SignalChem), 150 nM RIPK1 tracer, and serially diluted compounds (10,000 to 0.64 nM, 5-fold dilutions) in opaque 384-well plates.Unless speci ed otherwise, all assays were performed in triplicate.The assay mixtures were incubated at room temperature in the dark for 120 min, and the signals were collected using a BioTek Synergy H1 microplate reader to measure the uorescence emission ratio (I520 nm/I490 nm) of each well using a 340-nm excitation lter, a 100-µs delay, and a 200µs integration time.Raw data from the plate reader were used directly for the analysis.The curve-tting software GraphPad Prism 9 was used to generate graphs and curves and determine IC 50 values.

NanoBRET Live-cell Ternary Complex Assay
Human RIPK1 cDNA insert was cloned into pLenti6.2-ccdB-nLucplasmid (87075, Addgene, a kind gift from Taipale Lab) using gateway cloning kit (11791020, Thermo) and standard protocol to obtain pLenti6.2-RIPK1-nLucfusion vector.The day before transfection, 1 million HEK293T cells were plated in a 60 mm dish and allowed to grow overnight in DMEM/10% FBS.The next day, the cells were cotransfected overnight at 37 deg with 1ng/ml pLenti6.2-RIPK1-nLucfusion vector, 100ng/ml HaloTag®-VHL Fusion Vector (N273A, Promega) along with 1ug/ml carrier DNA vector (E4881, Promega) using the calcium phosphate method.After 18h, transfected cells were trypsinized and resuspended in Opti-MEM (11058-021, Gibco) supplied with 4% FBS and 100nM HaloTag® NanoBRET™ 618 Ligand (G9801, Promega) to a cell-density of 0.2M/ml (for background subtraction group, 618 ligand was omitted).Plate 100ul cells into each 96-well (136101, Thermo).The plate was further incubated at 37°C overnight to allow HaloTag-VHL to be labeled with 618 ligand.Next day, cells were further treated with 10uM MG132 for 0.5h followed by 1µM PROTAC or DMSO for 4h.Immediately before reading the plate, prepare 4x concentrated NanoGlo nLuc substrate (N157, Promega) was prepared by diluting the stock into Opti-MEM 1,000-fold, and the NanoGlo substrate was then added into each well to bring the nal concentration to 1x.Donor emission at 450 nM and acceptor emission at 610 nM were measured on a BioTek Synergy H1 plate reader equipped with lter cube set 450/80 and 610 LP.The corrected mBU was calculated as follows:

Proteomics Study
One million MDA-MB-231 cells were seeded in 6-well plates.The following day, cells were treated in triplicate with LD4172 (200 nM) or LD4172-NC (200 nM) for 6 h.The cells were washed thrice with icecold PBS.The cell pellets were lysed, reduced, alkylated, and digested using EasyPep™ MS Sample Prep Kits (A45733,ThermoFisher) according to the manufacturer's instructions.The same amount of peptide from each condition was labeled with a tandem mass tag (TMT) reagent (90113, ThermoFisher).The 10plex TMT reagent was incubated with each peptide sample at a ratio of 1:8 (peptide:TMT label).The 10plex labeling reactions were performed for 1 h at room temperature.The labeled peptide samples were quenched by adding 50 µL of 5% hydroxylamine and 20% formic acid solution for 5 min and then mixed.The mixed samples were desalted and fractionated o ine into 24 fractions on a 250×4.6 mm Zorbax 300 Extend-C18 column (Agilent) using an Agilent 1260 In nity HPLC system.

Em 450nm
The 24 fractions were dried in vacuo and resuspended in 5% acetonitrile in water (0.1% FA).Each sample was rst separated by nano LC through a 5-40% ACN gradient within 75 min and ionized by electrospray (2.4 kV), followed by MS/MS analysis using higher-energy collisional dissociation (HCD) at a xed 38.0 collision energy on an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scienti c) in datadependent mode with a 3 sec cycle-time.MS1 data were acquired using the FTMS analyzer in pro le mode at a resolution of 120,000 over a range of 400-1,600 m/z.Following HCD activation and quadrupole isolation with a window of 0.7 m/z, MS/MS data were acquired using an orbitrap at a resolution of 50,000 in centroid mode and normal mass range.
Proteome Discoverer 2.4 (Thermo Fisher Scienti c) was used.RAW le processing and controlling peptide-and protein-level false discovery rates, assembling proteins from peptides, and protein quanti cation from peptides.Searches were performed using full tryptic digestion against the SwissProt human database with up to two miscleavage sites.Oxidation (+ 15.9949 Da) of methionine and Deamidation on N and Q (0.984 Da) were set as variable modi cations, while carbamidomethylation (+ 57.0214 Da) of cysteine residues and TMT 10-plex labeling of peptide N-termini and lysine residues were set as xed modi cations (+ 229.163Da).Data were searched with mass tolerances of ± 10 ppm and 0.02 Da on the precursor and fragment ions (HCD), respectively.The results were ltered to include peptide spectrum matches (PSMs) with a high peptide con dence.PSMs with precursor isolation interference values > 50% and average TMT-reporter ion signal-to-noise values (S/N) < 10 were excluded from quantitation.Isotopic impurity correction and TMT channel normalization, based on the total peptide amount, were applied.Protein quanti cation uses both unique and random peptides.For statistical analysis and adjusted p-value calculation, an integrated analysis of variance (ANOVA) hypothesis test on individual proteins was used.TMT ratios with adjusted p-values below 0.01 were considered signi cant.
The mass spectrometry raw data les for quantitative multiplexed proteomics have been deposited in the MassIVE dataset under accession number MSV000092377.

Molecular Docking
Molecular docking studies were carried out using Schrödinger software.Schrödinger adopted the Glide algorithm to dock exible ligands into the protein-binding site.The crystal structure of the RIPK1 kinase domain in complex with the isoquinolin-1amine analog (PDB: 4NEU) was used as the receptor structure in molecular docking studies.

Figures
Figures

Figure 1 Design
Figure 1

Figure 5 LD4172
Figure 5 5B16F10 cells.The medium was replaced 24 h after electroporation.Single-cell clones were screened for protein expression by western blotting.Con rmed gene-deleted clones were pooled and cultured for two weeks in vitro before being implanted in vivo.Animal studiesAll animal experiments were conducted according to the protocol approved by the Institutional Animal Care and Use Committee of the Baylor College of Medicine.All tumor studies were performed in female mice.Female 6-week-old C57BL/6J mice were ordered from Jackson Labs, and experiments were carried out in age-matched animals.Mice were housed in the TMF Mouse Facility at the Baylor College of Medicine under SPF/climate-controlled conditions with 12-hour day or night cycles.They were continuously supplied with fresh chow and water from an autowater system.DAPI for 20 minutes (1:30,000, 422801, Biolegend).The slides were mounted with ProLong™ Diamond Antifade Mountant (Thermo Fisher, P36970) and imaged using a Zeiss LSM780 confocal microscope with a ×60 objective.Consistent image exposure times and threshold settings were applied for all groups.