Proteolysis Targeting Chimera Degraders of the METTL3–14 m6A-RNA Methyltransferase

Methylation of adenine N6 (m6A) is the most frequent RNA modification. On mRNA, it is catalyzed by the METTL3–14 heterodimer complex, which plays a key role in acute myeloid leukemia (AML) and other types of blood cancers and solid tumors. Here, we disclose the first proteolysis targeting chimeras (PROTACs) for an epitranscriptomics protein. For designing the PROTACs, we made use of the crystal structure of the complex of METTL3–14 with a potent and selective small-molecule inhibitor (called UZH2). The optimization of the linker started from a desfluoro precursor of UZH2 whose synthesis is more efficient than that of UZH2. The first nine PROTAC molecules featured PEG- or alkyl-based linkers, but only the latter showed cell penetration. With this information in hand, we synthesized 26 PROTACs based on UZH2 and alkyl linkers of different lengths and rigidity. The formation of the ternary complex was validated by a FRET-based biochemical assay and an in vitro ubiquitination assay. The PROTACs 14, 20, 22, 24, and 30, featuring different linker types and lengths, showed 50% or higher degradation of METTL3 and/or METTL14 measured by Western blot in MOLM-13 cells. They also showed substantial degradation on three other AML cell lines and prostate cancer cell line PC3.


■ INTRODUCTION
Post-transcriptional (epitranscriptomic) modifications of RNA have a key role in gene expression and cell homeostasis regulation. 1,2The N6-adenosine methylation (m 6 A) is the most abundant among over 150 reported modifications. 3It has been found on mRNA, tRNA, rRNA, and several noncoding RNAs. 4 The m 6 A is a dynamic and reversible modification deposited by proteins defined as "writers" and removed by "eraser" proteins.−9 In this way, the m 6 A modification can mediate the expression or silencing of specific genes. 2his epitranscriptomic machinery enables processes such as stem cell differentiation, 10 cell response to stress, 11 and regulation of the circadian cycle 12 under physiological conditions.Its dysregulation has been linked to a growing number of pathological conditions.−20 Methyltransferase-like 3 (METTL3) and METTL14 form the heterodimeric protein complex that catalyzes the deposition of the m 6 A modification (writer).METTL3 is the catalytic subunit that binds the cosubstrate S-adenosyl-L-methionine (SAM) while METTL14 facilitates RNA binding and stabilization of the complex. 21,22Many studies show that increased m 6 A levels can lead to enhanced cell proliferation, antiapoptotic effects, promotion of migration, and invasion. 23Moreover, METTL3 has been reported to promote other cancerogenic processes independently of its catalytic activity. 24Mouse knockout studies have revealed that the depletion of m 6 A modification leads to early embryonic lethality.Mouse embryonic stem cells (mESCs) could survive the METTL3 gene knockout and continue to proliferate but lost the ability to differentiate. 25 In human hematopoietic stem/progenitor cells (HSPCs), m 6 A modification controls myeloid differentiation.Short hairpin RNA (shRNA)-mediated silencing of METTL3 in HSPCs promotes cell differentiation and reduces cell proliferation. 26These two examples demonstrate the relevance of m 6 A modification in normal cell differentiation processes, but its effects are largely dependent on the cellular context.Importantly, METTL3 mRNA and protein are expressed at higher levels in AML cells than in healthy HSPCs, which can result in a therapeutic window to target the protein with inhibitors and degraders.
−30 The low

JACS Au
nanomolar inhibitors UZH2 and the compound published by Storm Therapeutics (STM2457) have shown antiproliferative effects in acute myeloid leukemia (AML) cell lines, strengthening the therapeutic potential of targeting the METTL3−14 complex. 27,29A small molecule inhibitor of METTL3 called STC-15 (SAM-competitive, developed by Storm Therapeutics) is currently in phase 1 clinical trials (https://clinicaltrials.gov/ study/NCT05584111).A total of 66 patients have been enrolled and dosed (since Nov. 15, 2022) to evaluate the safety, pharmacokinetic, pharmacodynamic, and clinical activity of STC-15 in subjects with advanced malignancies.Serious side effects after dosing STC-15 do not seem to have emerged as the clinical trials are ongoing for nearly 14 months.A recent in vivo study using the METTL3 inhibitor STM2457 (a predecessor of the compound STC-15 currently in clinical trials) reported milder, more nuanced, and manageable effects of pharmacological METTL3 inhibition on normal hematopoiesis than those observed in METTL3 knockout studies.The observed lineage bias in the earliest hematopoietic progenitors included an increase in neutrophils and a decrease in erythroids, indicating anemia as a potential side effect of catalytic METTL3 inhibition. 31Available data show that targeting METTL3 by knockout or inhibition affects normal cells, but the effect depends on the cellular and systemic context.However, the high The EC max value is the PROTAC concentration at the maximum of the signal in the ternary complex formation assay (TCFA).The IC 50 value is the PROTAC concentration required to inhibit 50% of the catalytic activity of METTL3-14, as measured by the binary FRET assay.a Degradation <10% or stabilization.b The compound was inactive in the TCFA.c Not determined.cellular concentration of SAM (60 to 160 μM as measured in rat liver) 32 can limit the scope of SAM-competitive inhibitors.
−36 PROTACs are heterobifunctional molecules bearing a protein of interest (POI) ligand covalently linked to an E3 ligase ligand.Upon binding to both targets, PROTACs promote ubiquitination of the POI and its subsequent degradation by the 26S-proteasome.This is a promising approach already applied to a variety of targets, in particular in the epigenetic field. 37Their catalytic-like mechanism of action results in the recycling/reuse of the PROTAC molecules upon protein degradation.Moreover, because of the degradation of the whole protein, PROTACs eliminate both its enzymatic and scaffolding functions, acting as a chemical knockout of the protein.
Here, we report a medicinal chemistry campaign aimed at the development of PROTAC molecules against METTL3−14, the human m 6 A-RNA writer complex.

Protein Structure-Based Design
The medicinal chemistry campaign builds upon our previous results obtained during the development of small molecule inhibitors for METTL3−14. 27Here, we start from two potent and selective METTL3−14 inhibitors, UZH2 (IC 50 = 5 nM, selectivity data in Table S1) and its desfluoro derivative AD22 (IC 50 = 89 nM, compound 10 in ref. 27 ) (Figure 1A).As E3 ubiquitin ligase, we selected Cereblon (CRBN), for which the most common ligands are 4-amino thalidomide (pomalidomide) and lenalidomide. 38,39The general structure of the synthesized PROTACs is represented in Scheme 1.The UZH2/ AD22 atom for the covalent bond with the linker was identified by crystallographic analysis.The binding poses of UZH2 (PDB 7O2F) and AD22 (PDB 7O0P) in the METTL3−14 complex (Figure 1B) provide an exit vector from the pyrimidine ring into the solvent-exposed area.Replacing the methylamino moiety with a propyl diamino motif (handle) allowed a convenient connection to the CRBN ligand via a linker.The amino group directly connected to the pyrimidine ring was intended to maintain a favorable hydrogen bond interaction with the side chain of Asp377 in METTL3 (Figure 1C).The terminal amino functionality allowed the final amide bond formation, thus connecting the POI ligand with pomalidomide through the linker (Scheme 1).The propyl diamino moiety is formally considered to be part of the linker.Nevertheless, it affects the affinity of the PROTACs for METTL3−14 as measured in our time-resolved FRET assay (hereafter referred to as binary assay). 40For this reason, this portion is called handle to clearly distinguish it from the rest of the linker.The structures of all the synthesized PROTACs (compounds 1−35) are reported in Table 1.The first set of PROTAC molecules (compounds 1−4) consists of AD22 as the POI ligand, propyl diamine handle, polyethylene glycol (PEG) linker, and pomalidomide as the E3 ligase binder (Scheme 1A).Their synthesis was achieved by a final amide bond formation, as described in further detail in Scheme 4 in the Synthesis section.
The rationale behind the use of PEG linkers in the first generation was the commercial availability of PEG chains with different numbers of PEG subunits. 41This allowed us to cover different distances between the PROTAC moieties for POI and CRBN.This is useful for investigating the optimal range for the formation of the ternary complex CRBN/PROTAC/ METTL3−14.−44 The degradation of METTL3 and METTL14 was measured individually by Western blot at various PROTAC concentrations (10, 5, 1, 0.1, 0.01 μM) at the 16 h time point in MOLM-13, which is an AML cell line.However, none of the four first-generation PROTACs (compounds 1−4) showed degradation activity (Table 1).Compounds 1, 2, 3, and 4 were also tested in a biochemical ternary complex formation assay (TCFA). 45elatively high effective concentrations at the peak of the Hook curve (EC max : 6.8, 2.8, 1.9, and 2.0 μM) reflect low affinity toward both CRBN and METTL3−14 (Table 1).In addition, the amplitude of the Hook curve (Table S2) is low compared to those of other compounds in this paper.This can be an indication of weak cooperativity and less stable ternary complex formation, which is needed for successful degradation activity. 46s PROTACs can suffer from low cellular permeability due to their high molecular weight and/or high hydrophilicity, 47 we decided to assess permeability and target engagement in cells using the cellular thermal shift assay (CETSA). 48The first set of PROTACs (compounds 1−4) exhibited low protein stabilization and only at the very high concentration of 100 μM.Considering the CETSA (Figure 2) and TCFA results, the lack of degradation might be a result of low cell permeability and/or inability to form a stable ternary complex.

Optimization of Cell Permeability and POI Affinity
To address the low permeability, we synthesized a second set of AD22-based PROTACs (compounds 5−9), aiming to increase the lipophilicity of the molecules.−51 The positive effect of the increased lipophilicity on cell membrane permeation was confirmed through CETSA experiments, as indicated by more pronounced stabilization effects (Figure 2).To improve the stability of the ternary complex, we decided to change the POI ligand by replacing AD22 with the ∼20-fold more potent inhibitor UZH2 (Scheme 1C).As expected, the resulting PROTACs (compounds 10−13) showed a much higher affinity for METTL3−14 in comparison to their AD22 analogues (Table 1).Furthermore, the EC max measured in TCFA was substantially improved.Compounds 10, 11, and 12 showed an EC max below 1 μM, with a 3-to 6-fold improvement compared to  S3).The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(c) Correlation of the degradation of METTL3 and METTL14 in MOLM-13 at PROTAC concentration of 2 μM; the dotted black line is a linear regression of all the compounds (corr.coeff.r = 0.95).The nine most active PROTACs are highlighted in turquoise.me-14 and me-24 (in red) are the methylated negative controls of PROTACs 14 and 24 (for structures, see Figure 4B).the nonfluorinated analogues (Table 1).Despite the increased cellular permeability and binding affinity to the target protein, none of the tested compounds showed degradation of METTL3−14, as measured by Western blot at 0.2, 2, and 20 μM PROTAC concentration at multiple time points (6, 16, 36  h).As the PROTACs synthesized so far featured the same handle motif (propyl diamine), we questioned its impact on ternary complex formation and protein degradation.From previous studies, we knew that the replacement of the methyl amino group in UZH2 with an aryl or aliphatic ring can provide favorable lipophilic interaction with the edge of the METTL3− 14 binding site. 30−54 With this in mind, we moved on to the synthesis of PROTACs bearing a more rigid handle/linker (Scheme 1D).

Optimization of Length and Rigidity of the Linker
The next set of seven PROTAC molecules contained a benzyl diamine handle instead of a propyl diamine.In addition, we varied the length of the linear portion of the linker, retaining the alkyl (compounds 14−17) and alkyl-triazole (compounds 18− 20) motifs from previous optimization steps.The presence of the aromatic ring significantly improved the affinity for METTL3−14 (measured by the FRET-based binary assay) as well as the values of EC max (Table 1).Substantial degradation of both METTL3 and METTL14 proteins after 24 h incubation was observed with 2 μM PROTAC concentration in MOLM-13 cells.It is important to note that for SAM-competitive PROTACs the cellular activity at low micromolar concentration is in line with the low micromolar activity measured previously for UZH2 in cellular assays 25,29 which is due to the aforementioned high concentration of SAM.Compounds 14, 19, and 20 reduced the level of both METTL3 (by 52, 33, and 42%, respectively) and METTL14 (by 52, 40, and 51%, respectively).The most promising derivative, PROTAC 14, contained a shorter linker in comparison with the other PROTACs of this set.Furthermore, we synthesized another set of 11 PROTACs that contained lipophilic and rigid handles such as piperidine (compounds 21−25), piperazine (compounds 29−32), and triazole (compounds 26 and 27) in combination with different alkyl linker lengths.After a round of protein degradation screening in cells and quantification by Western blot analysis (2 μM, 24 h MOLM-13) compounds 22, 23, 24, 29, 30, and 31, featuring piperidine or piperazine handle, showed a 50% or higher degradation of METTL3 and/or METTL14 (Table 1, Figure 3A,B). 55,56The correlated degradation of the two proteins of the heterodimeric complex METTL3−14 provides evidence that a PROTAC binding at the SAM-pocket of METTL3 can degrade both proteins (Figure 3C, Figure S1).PROTAC 30 displayed the most significant degradation activity, achieving a reduction of both METTL3 and METTL14 by about 60%.In contrast, the PROTACs with a triazole ring as a handle (compounds 26 and 27) performed worse, showing degradation efficacies of 13 and 12%, respectively.
The TCFA shows gradual EC max improvement when reducing the linker length, as we can see for the piperidine handle series PROTACs 25, 24, 23, 22, and 21 (Figure 3D).Among them, the shortest PROTAC, compound 21, has the best EC max (0.06 μM), but in terms of degradation, it is worse than its slightly longer analogues 22 and 24, with 21, 46, and 36% METTL3 reduction, respectively.A possible explanation for the discrepancy between the biochemical TCFA and the cellular degradation assay is that the TCFA employs truncated protein constructs.PROTAC 21 might be too short to form a stable ternary complex with full-length proteins, and some steric clashes can cause a nonoptimal conformation of the ternary complex.This EC max improvement tendency is not observed at even shorter linker lengths, as shown with PROTAC 32 (no linker), which is inactive in both TCFA and cell degradation assays.Thus, there seems to be an optimal range for the linker length (three to five methylene groups).
Figure 3E shows the distribution of the percentage of protein degradation in MOLM-13 as a function of the EC max measured in TCFA (Table 1).Even though the correlation is not strong (correlation coefficient r = 0.44), only PROTAC 28 shows degradation higher than 20% and EC max > 1 μM.The biochemical TCFA can be considered a useful screening technique for prioritizing the PROTACs for biological characterization.
To further increase the rigidity of the handle/linker part, we synthesized three PROTACs (compounds 33−35) with reduced linker flexibility.Among them, PROTAC 33 turned out to be the best, showing 44% METTL3 reduction, while compound 34 displayed a not significant protein reduction and 35 only 21% METTL3 reduction.Considering that compounds 34 and 35 contain a much shorter linker than that of PROTAC 33, these results further highlight the importance of linker length.
Once the most promising handles and ideal length to achieve protein degradation were defined, we tried to modify the connection to the CRBN ligand. 57,58PROTAC 23 is linked to thalidomide at position 5.This modification is well tolerated, and the compound causes 50% degradation of METTL3.Nonetheless, compared to its 4-substituted analogue (PROTAC 22), the difference in both degradation (50% vs 46%) and EC max (0.16 μM vs 0.19 μM) is negligible (Table 1, Figure 3D).
At this point, we decided to select a subset of PROTACs for further validation in multiple AML cell lines.We first focused on the eight PROTACs that showed submicromolar EC max and at least 50% degradation of METTL3 and/or METTL14 in MOLM-13 (14, 20, 22, 23, 24, 29, 30, 31, highlighted in Figure 3C).Note that PROTAC 33 (also highlighted in Figure 3C) had not yet been prepared when we decided to focus on a small subset of PROTACs.We further restricted the selection to only five PROTACs as PROTAC 23 is the meta-substituted equivalent of 22, and PROTACs 29−31 differ only in the number of methylene groups.Thus, the degradation activity of compounds 14, 20, 22, 24, and 30 was investigated in different AML cell lines (THP-1, NOMO-1, and KASUMI-1) (Figure 3B, Figures S1 and S3).While the degradation levels observed in THP-1 and NOMO-1 cell lines were comparable to those in MOLM-13, a higher degradation of both METTL3 and METTL14 was measured in KASUMI-1.The treatment of KASUMI-1 cells with 2 μM of PROTAC 30 for 24 h caused a 70% degradation of the POI.
To further validate the top PROTACs, we tested some of them against cell lines of solid tumors.We decided to focus on two prostate cancer cell lines (DU145 and PC3) because of recent evidence for the importance of METTL3 in prostate cancer. 59,60The selected PROTACs 14, 20, 22, 24, and 30 showed only a minor effect in the DU145 cell line (Figure 3B, Figure S3).In contrast, the METTL3 levels were reduced substantially in PC3 after 24 h, with PROTACs 20 and 22 showing the highest degradation values, 48 and 64% reduction of METTL3, respectively (Figure 3B, Figure S3).This result indicates that the degradation activity of our PROTAC molecules is not limited to leukemia cell lines, but they have a good potential also against prostate cancer.Moreover, the correlated degradation of METTL3 and METTL14 was observed not only in MOLM-13 (Figure 3C) but also in all of the other cell lines tested (Figure 3B, Figure S1).

Validation of PROTACs Cellular Activity
At this stage, we evaluated the METTL3−14 protein levels after PROTAC treatment in combination with high concentrations of the small-molecule inhibitor lenalidomide or UZH2 (Figure 4A).These controls consist of saturating the binding pockets of CRBN or METTL3−14, respectively, thus preventing the formation of the ternary complex.Cells were treated with PROTACs 20 and 24 under three different conditions: 2 μM PROTAC, 2 μM PROTAC + 10 μM lenalidomide, and 2 μM PROTAC + 10 μM UZH2.As expected, we did not observe any degradation when applying a high concentration of lenalidomide.Interestingly, the presence of a high UZH2 concentration led to elevated levels of METTL3−14 both in the control as well as in combination with the PROTAC.This observation indicates a possible cellular compensatory mechanism aimed at preserving the protein catalytic activity in the presence of an inhibitor. 61It also offers a potential explanation for the difficulties in reaching degradation levels of METTL3−14 above 50%.
To confirm that our compounds lead to protein degradation by hijacking the Ubiquitin−proteasome system (UPS), we synthesized negative controls based on a single methylation, which results in inactive lenalidomide/pomalidomide derivatives. 62−64 Thus, we prepared the methylated versions of PROTACs 14 and 24 (me-14 and me-24) (Figure 4B).These methylated compounds did not show any METTL3−14 degradation at standard testing conditions (2 μM, 24 h) (Figure 4C).
Both control PROTACs gave no signal in the biochemical ternary complex formation assay (Figure 4D).The methylated derivatives are inactive, indicating their inability to form a ternary complex and further confirming the specificity of our PROTACs.Compound me-14 was also tested in the CETSA on both METTL3 and CRBN to provide evidence that it is cell permeable and engages the POI and not CRBN (i.e., validation of the negative control).As expected, the compound showed stabilization of METTL3 but not of CRBN (Figure S4).Furthermore, we set up an in vitro ubiquitination assay to quantify the ubiquitination of METTL3 and METTL14 in the presence of different concentrations of PROTACs.Compounds 14 and me-14 were tested at 2, 8, and 32 μM (Figure 4E).While me-14 does not increase the ubiquitination of METTL3 compared to the control without compound, PROTAC 14 raises the ubiquitination level of METTL3 to ∼40% at 2 μM.The decrease of ubiquitination at higher concentrations of compound 14 is consistent with the Hook effect.The ubiquitination of METTL14 is not as pronounced (Figure 4F).This can indicate that the ubiquitination site(s) is (are) mainly on METTL3.Interestingly, Zeng et al. showed that METTL14 can get ubiquitinated by STUB1 in the METTL3− 14 interface and METTL3 therefore seems to prevent METTL14 ubiquitination. 65The simultaneous degradation of METTL3 and METTL14 caused by the PROTACs (Figure 3C) could be explained either by both proteins being subjected to the proteasome as a complex or by the reduced stability of METTL14 without METTL3.
Lastly, we analyzed the concentration dependence of degradation of METTL3 and METTL14 in the MOLM-13 cell line.We selected PROTACs 14 and 30, which have the highest METTL3 degradation in MOLM-13 (52 and 57%, respectively).The activity of PROTAC 14 was also compared to the one of its negative control me-14.We treated MOLM-13 cells with PROTACs 14, 30, and me-14 at five different concentrations (Figure 5).As expected, the concentration dependence shows the so-called Hook effect, 66,67 resulting from a saturation of CRBN and METTL3−14 at high PROTAC concentrations, where the PROTAC/CRBN and PROTAC/ METTL3−14 binary complexes prevent the ternary complex formation and thus degradation.This result confirms that our compounds behave according to the principle of the PROTAC mechanism of action.For PROTAC 30, the highest degradation was observed at 0.1 μM concentration, reaching up to 60% degradation, and for PROTAC 14, the highest effect was seen at 1 μM, with a METTL3 reduction of around 50%.Compound me-14 showed no degradation and even increased METTL3 levels at 10 μM, similar to that of the UZH2 inhibitor alone.
Overall, among the 35 PROTAC compounds synthesized in this study, nine (14, 20, 22, 23, 24, 29, 30, 31, and 33) caused at least 50% degradation of METTL3 and/or METTL14 in MOLM-13.Five of them (14, 20, 22, 24, and 30) were tested in different cell lines (AML and prostate cancer) and showed significant activity with METTL3−14 degradation up to 70% of the endogenous level (Figure 3B).The concentration-dependent degradation activity (Hook effect, Figure 5A), correlation in the degradation of the two proteins of the METTL3−14 heterodimeric complex (Figure 3C), methylated PROTAC controls (Figure 4), and the additional validation experiments with competitive small-molecule ligands (Figure 4A) provide strong evidence of cellular target engagement and selectivity of our PROTACs.
To assess whether degradation induced by our compounds translated into enhanced cell death, we performed cell viability assays on MOLM-13 with all of the UZH2-based PROTACs (Table S3).Compounds 20, 22, 24, and 30 were also tested in the other AML cell lines THP-1 and Kasumi-1 and in the prostate cancer cell lines PC3 and DU145 (Figure 6A).Significant effects on the cell viability of some PROTACs were observed only at the highest concentration tested (10 μM).This concentration is higher than the EC max values measured in the biochemical assay because, as mentioned before, the PROTACs (and UZH2) compete with the micromolar concentration of SAM in the cellular assays.Interestingly, on PC3 cell line only PROTACs (compounds 22, 24, and 30) and not UZH2 showed an antiproliferative effect.
To better understand cell viability results in the MOLM-13 cell line, we measured the changes in cellular m 6 A/A levels (by LC-MS quantification) after PROTAC treatment (Figure 6B).At 2 μM (concentration used for degradation screening), we did not observe significant effects on the levels of m 6 A/A (except for a slight reduction for compound 22).Measurable reduction of m 6 A/A was observed at a 10 μM concentration.Taking into consideration the concentration-dependent activity of PRO-TAC molecules (Hook effect), it is more likely that the observed reduction in m 6 A modification is due to inhibition of the catalytic activity of METTL3 by the UZH2-based warhead rather than protein degradation.Partial inhibition of the catalytic activity might also explain the observed cytotoxic effect at the highest tested concentrations of PROTACs.In conclusion, the modest cytotoxicity and reduction of m 6 A/A suggest that degradation levels higher than 50−70% are required to observe phenotypic effects specific to the PROTAC-induced METTL3−14 degradation.

Synthesis
Starting from spiro compounds 36 and 37, the preparation of POI ligands bearing the handle moiety was conducted following the general strategy reported in Scheme 2: After Boc deprotection followed by an S N Ar reaction with 4,6dichloro or difluoro pyrimidine, compounds 38 and 39 were obtained.A second S N Ar was needed to afford compounds 40, 41, and 42a−47a.Interestingly, due to the poor reactivity of chloro-pyrimidine 38 toward S N Ar, we were only able to prepare compounds 42a and 43a.Switching to its fluorinated analogue 39 was necessary to synthesize compounds 40, 41, and 44a−47a in good yield (from 50 to 70%).Compounds 42−47 were then obtained upon removal of the protecting group from precursors 42a−47a.
Compounds 1−20, 22−25, 29−31, and 33 were synthesized via an amide coupling reaction between compounds 42−46 and 48 and the corresponding pomalidomide/lenalidomide carboxylic acids.HATU coupling agent provided decent yields only for compounds 1−4, 8, 9,and 29−31.For the other molecules, COMU performed better and provided the desired products with yields of up to 60% (Scheme 4).The final amide coupling (Scheme 4) did not work for compounds 21 and 35.Therefore, the synthetic route was slightly modified.Both intermediates 49 and 51 were prepared starting from 45.An amide coupling between the latter and Bocglycine, followed by amino group deprotection, yielded compound 49.For the synthesis of compound 51, the acetylation of 43 using 2-chloroacetyl chloride resulted in the formation of 50.Afterward, we converted 50 into 51 through an S N 2 reaction with Boc-piperazine, followed by the removal of the protecting group (Scheme 5).
With intermediates 49 and 51 in our hands, we were finally able to obtain PROTACs 21 and 35 using S N Ar and 4-fluoro thalidomide.Using the same reaction as for the final step, we prepared PROTAC 34 from compound 47 (Scheme 6).
To synthesize 26 and 27, compounds 52 and 53 were first prepared starting with lenalidomide, which was reacted with the corresponding carboxylic acids. 69A following S N 2 reaction with NaN 3 allowed us to prepare intermediates 54 and 55. 70 PROTACs 26 and 27 were finally obtained through the click reaction (Scheme 7).
Compound 56 was obtained from a S N Ar reaction between 4fluoro thalidomide and propargyl amine. 71Similarly to 26 and 27 (Scheme 7), the click reaction between 56 and 41 was employed to synthesize PROTAC 28 (Scheme 8).
Following the same procedure used for 56 (Scheme 8), we prepared the protected version of intermediate 57.After Boc removal, PROTAC 32 was obtained via the S N Ar reaction between compounds 57 and 39 (Scheme 9).
For the synthesis of me-14 and me-24 (Scheme 10), 4-fluoro thalidomide was methylated using CH 3 I. 72 The two PROTACs were then synthesized from 58, following the synthetic route used for the nonmethylated analogues PROTACs 14 and 24 (Scheme 4).

■ CONCLUSIONS
Here, we report a medicinal chemistry design of PROTACs against m 6 A-RNA writer METTL3−14.We used as starting information the crystal structure of METTL3−14 in complex with the potent (IC 50 = 5 nM) and selective inhibitor UZH2.We efficiently optimized the PROTAC linker by first employing the desfluoro derivative of UZH2 as a moiety for METTL3−14.While PEG-and alkyl-based linkers were considered initially, only the PROTACs with alkyl-based linkers demonstrated cell penetration.Subsequently, we synthesized 26 PROTACs based on UZH2 with alkyl linkers of varying lengths.The formation of the ternary complex and ubiquitination of METTL3 were confirmed by a FRET-based assay and an in vitro ubiquitination assay.The cellular characterization of the PROTACs is still highly necessary, but biochemical TCFA emerges as a valuable instrument for efficiently screening PROTACs for further validation (Figure 3E).Notably, five PROTACs (14, 20, 22, 24, and 30) with distinct rigid extensions of UZH2 achieved substantial METTL3−14 degradation (50% or higher) in multiple AML cell lines and the prostate cancer cell line PC3, showcasing their potential as valuable tools in targeted protein degradation research.In comparison with the catalytic inhibitor UZH2, the PROTACs 22, 24, and 30 show higher antiproliferative activity on prostate cancer PC3 but not on AML cell lines.The elevated level of METTL3−14 in the presence of UZH2 and the lack of reduction of the m 6 A/A level of polyadenylated RNA (measured by LC/MS) suggest that a substantially higher degradation of METTL3−14 (probably above 90%) is required for a strong antiproliferative effect.Similarly, compounds with higher activity will be needed to characterize the potential side effects of METTL3 PROTACs in future in vivo studies.
In investigating the features of PROTACs for successful protein degradation, this study adds a valuable contribution to the understanding of the crucial role of the linker.The biochemical and cellular characterizations provide evidence that a minimum linker length is required to achieve the formation of the tripartite complex between PROTAC, E3 ligase CRBN, and POI(s) METTL3−14.In this regard, the linkers of PROTACs 32 and 34 are too short for the formation of a ternary complex that allows for degradation of METTL3−14.The PROTACs that promote a more significant degradation (14, 20,  22, 23, 24, 29, 30, 31, and 33) are characterized by a rigid "handle" (benzyl, piperidine, and piperazine) and a longer linker.The structure of these PROTACs presents only minimal differences compared to other molecules synthesized during this study (18, 19, 21, and 25), which shows how small differences in geometry, length, and/or rigidity of the linker can have a major impact on the formation and stabilization of the ternary complex, and the eventual protein degradation.

METTL3−14 Expression and Purification
For determining the half maximal inhibitory concentration (IC 50 ) with the full-length complex, the recombinant complex construct pFastBacDual-StrepII-GFP-TEV-METTL3-His-TEV-METTL14 was expressed using the baculovirus/Sf9 insect cell expression system and purified as described previously. 21or ternary complex formation assays, genes containing the methyltransferase domains (MTD) of METTL3 (residues 354−580) and METTL14 (residues 107−395) were cloned into the MacroBac vector 438-GST. 73A histidine tag (His) was inserted at the N-terminus of GST by site-directed mutagenesis to yield the construct 438-His-GST-METTL3 MTD -METTL14 MTD .Recombinant baculovirus to express the complex of His-GST-METTL3 MTD and METTL14 MTD was generated by using the Bac-to-Bac system.For protein expression, suspension cultures of Sf9 cells in Sf-90 II SFM medium (Thermo Fisher) were infected at a density of 2 × 10 6 mL −1 .Cells were harvested 72 h postinfection, resuspended in buffer A (50 mM Tris-HCl pH 8.0, 500 mM NaCl) supplemented with protease inhibitor cocktail (Roche Diagnostics GmbH, Germany), phenylmethylsulfonyl fluoride (PMSF), salt active nuclease (Merck), and lysed by sonication.The protein complex was purified by Ni-affinity chromatography on a 5 mL HisTrap HP column (Cytiva), equilibrated, washed with buffer A, and eluted with 250 mM imidazole.The complex was further purified by size exclusion chromatography using a Superdex 200 Increase 10/300 GL column (Cytiva) in 20 mM Tris-Cl, pH 8.0, and 200 mM KCl.The complex was aliquoted and flash-frozen in liquid nitrogen and stored at −80 °C until further use.

Reader-Based TR-FRET Assay
The inhibitory potencies of the PROTACs for METTL3 were quantified by a homogeneous time-resolved fluorescence (HTRF)based enzyme assay as previously described. 40Briefly, the level of m 6 A in an RNA substrate after the reaction catalyzed by METTL3−14 was quantified by measuring specific binding to the m 6 A reader domain of YTHDC1 (residues 345−509) by HTRF.PROTACs that inhibit METTL3 decrease the m 6 A level and, thus, reduce the HTRF signal.Dose−response curves of titrations with the PROTACs were plotted in OriginLab 2018 and fitted with nonlinear regression "log(inhibitor) vs normalized response with variable slope" from which IC 50 values were determined.Each PROTAC was measured in triplicates on a Corning 384 U bottom white polystyrene plate.

Biotinylated CRBNTBD Expression and Purification
His-TEV-CRBN TBD -Avi was created by inserting a tobacco etch virus (TEV) cleavage site at the N-terminus and an Avi-tag at the C-terminus of the CRBN thalidomide binding domain (TBD, CRBN residues 318−442) and the construct was cloned into the pETDuet-1 vector between SacI and HindIII restriction sites together with full-length enzyme BirA, which was cloned into the same vector between NdeI and XhoI restriction sites.
The construct was overexpressed in Rosetta (DE3) cells upon induction with 100 μM isopropyl thio-beta-D-galactoside (IPTG) for 20 h at 18 °C.The expression medium was supplemented with 50 μM D-biotin.Harvested cells were resuspended in lysis buffer containing 100 mM Tris-HCl, pH 8.0, 500 mM NaCl, 1 mM 1,4-dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mM ethylenediaminetetraacetic acid (EDTA) and lysed by sonication.The lysates were centrifuged at 18,000 rpm at 4 °C for 1 h in an SS-34 rotor.The soluble fraction was then loaded onto a HisTrap FF crude column (GE Healthcare) and washed with lysis buffer supplemented with 50 mM imidazole.The protein was eluted with a buffer containing 250 mM imidazole, 100 mM Tris-HCl, pH 8.0, and 500 mM NaCl.Recombinant TEV protease cleaved the His 6 tag during overnight dialysis at 4 °C against 100 mM Tris-HCl, pH 8.0, and 500 mM NaCl buffer.The dialyzed sample was passed through the HisTrap FF crude column to remove the His 6 -tagged TEV protease and uncleaved protein.The protein was further purified by size-exclusion chromatography using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare) in 50 mM HEPES, pH 7.5 and 150 mM NaCl buffer.The protein was aliquoted and stored at −80 °C until further use.Biotinylation was confirmed by an avidin shift assay, where a final CRBN-Avi of 5 μM was mixed with different amounts of NeutrAvidin (10, 20, 40 μM) (Thermo Fisher # 31000) and the proteins were analyzed by SDS-PAGE (data not shown).

Ternary Complex Formation Assay
Ternary complex formation between the PROTACs, METTL3, and CRBN was quantified by a homogeneous time-resolved fluorescence (HTRF)-based enzyme assay.The HTRF signal of a titration series with PROTACs at constant METTL3 and CRBN concentrations underlies the hook effect, leading to a characteristic bell-shaped curve where the concentration of the ternary complex decreases at high PROTAC concentrations.Curves of titrations with the PROTACs were plotted in GraphPad Prism 9.5.1 and fitted with a Gaussian function, if appropriate.His-GST-METTL3 MTD -METTL14 MTD was used at a final concentration of 15 nM.CRBN TBD -Avi(biotin) was used at a final concentration of 10 nM.XL665-conjugated streptavidin (Cisbio, 610SAXLB) was used at a final concentration of 1.25 nM.Anti-GST Eu 3+ -labeled antibody (Cisbio, 61GSTKLB) was used at a final concentration of 0.8 nM.The final reaction volume was 20 μL in 50 mM HEPES (pH 7.5), 150 mM NaCl, 0.1% BSA, 100 mM KF.The assays were carried out in triplicate on a Corning 384 U bottom white polystyrene plate (20 ul working volume).The reaction was incubated for at least 3 h at room temperature (RT) in the dark before the HTRF signal was measured using a Tecan Spark plate reader (Tecan).The plate reader recorded with a delay of 100 μs the emission at 620 and 665 nm after excitation of the HTRF donor with UV light at 320 nm.The ratio of the emissions = F acceptor donor 665nm 620nm was considered for further analysis.The maximal control contained compound 16; the blank contained no compounds−this was replaced by the appropriate buffer (with DMSO).The Hook curves were determined by normalization with the maximal control (compound 16) where the maximum of the Hook curve is determined as the fraction of the maximum of compound 16 for each PROTAC.The concentrations resulting in the maximum signal (EC max ) and the amplitudes of the Hook curves were determined from the parameters of the Gaussian fit (if appropriate) or as the coordinates of the data-point with the highest TR-FRET signal.
For the ternary complex formation assay (TCFA), one of the protein partners must have a GST-tag and the other a biotin-tag.We chose to have the biotin-tag on the CRBN-TBD as it worked well in a previous project (unpublished data).Hence, we chose to put the GST-tag on either METTL3 or METTL14.We cloned, expressed, purified, and tested four N-terminally GST-tagged METTL3−14 constructs for the TCFA: two full-length complexes with either METTL3 or METTL14 GST-tagged and two MTD-only complexes with either METTL3-MTD or METTL14-MTD GST-tagged.The full-length complexes were either prone to aggregation or gave low signal in the TCFA.We attribute this to either folding problems due to the GST-tag or due to too long distances for efficient TR-FRET when the GST-tag is attached to the long unstructured N-terminal tails of the proteins.Hence, we conducted the TCFA with the MTD constructs, which behaved well and gave good TR-FRET signals.

Cell Viability Assay
Cells were seeded in white clear-bottom 96-well plates at a density of 6−20 × 10 3 cells/well in 50 μL of the complete RPMI medium and treated with 50 μL of increasing concentrations of the indicated compounds dissolved in DMSO (final concentration of compounds 0.01−10 μM) or DMSO only as a negative control (0.01% (v/v)) and incubated for 72 h at 37 °C with 5% CO 2 .Cell viability was determined using a CellTiter-Glo luminescent cell viability assay (Promega) based on the detection of ATP according to the manufacturer's instructions.100 μL of the reagent was added to each well and incubated for 10 min at room temperature.The luminescence was recorded using a Tecan Infinite 3046 M1000 microplate reader from the top.Background luminescence value was obtained from wells containing the CellTiter-Glo reagent and medium without cells.The resulting data was analyzed in GraphPad Prism 9.

Cellular Thermal Shift Assay (CETSA)
One million of MOLM-13 cells were suspended in 100 μL of PBS (10010023, Thermo Fisher Scientific) containing 2× protease inhibitor cocktail (11697498001, Roche), for each condition tested.Cells were incubated with compounds or DMSO control (1% (v/v)) for 1 h at 37 °C.They were then heat treated at 54 °C in a thermoblock for 3 min, followed by cooling to room temperature (3 min).Next, samples were lysed by threefreeze−thaw cycles in liquid nitrogen and centrifuge at 16000 g for 30 min, 4 °C.Equal volumes of control and tested samples (12 μL) were analyzed by Western blot.The changes in the amount of METTL3 protein (after normalization for β-actin and/or GAPDH) were monitored by performing densitometry in Image Studio Lite software and analyzed in GraphPad Prism 9.

Quantification of m6A/A Ratio in Polyadenylated RNA by UPLC-MS/MS Analysis
UPLC-MS/MS was performed as previously described. 28riefly, MOLM-13 cells were seeded into 6 well plates at a density of 1 × 10 6 cells/mL in 2 mL of complete RPMI medium.Cells were treated with the indicated concentrations of compounds or DMSO control (final concentration 0.5% (v/v)) for 24 h.Following the incubation, cells were collected by centrifugation and washed once with PBS, and total RNA was extracted using 0.5 mL of GENEzol reagent according to the manufacturer's instructions.The final volume of 50 μL of total RNA eluate was subjected to two rounds of purification using 25 μL of Sera-Mag magnetic oligo(dT) particles (Cytiva) per sample.The polyadenylated RNA was eluted with nuclease-free water in a final volume of 25 μL, and its concentration was determined using NanoDrop.One hundred nanograms of mRNA were digested to nucleosides and dephosphorylated in a one-pot reaction using 0.5 μL of nucleoside digestion mix (M0649S, NEB) in 25 μL of total reaction volume for 4 h at 37 °C.The samples were used for UPLC-MS/MS analysis without further purification steps.The data were plotted by using GraphPad Prism 9.
■ ASSOCIATED CONTENT * sı Supporting Information

Figure 1 .
Figure 1.Protein structure-based design of PROTACs.(a) 2D structures of UZH2 and its desfluoro precursor AD22.(b) Overlap of the crystal structures of METTL3 bound to AD22 (carbon atoms in cyan, PDB 7O0P) and UZH2 (carbon atoms in orange, PDB 7O2F).(c) Zoom in on the hydrogen bond between the side chain of Asp377 and the methylamine of AD22 and UZH2.Scheme 1.General Structure of PROTACs and Optimization Strategy

Figure 2 .
Figure 2. Evaluation of AD22-based PROTACS 1-9 in the MOLM-13 (AML) cell line.The stabilization of METTL3 (top) and Cereblon (CRBN, bottom) was quantified by CETSA at 54 °C.The SAMcompetitive inhibitor AD22 was employed as a control for METTL3 (top, left), while lenalidomide (LEN) was used as a control for Cereblon (bottom, left).The dashed line represents the protein level of the DMSO control used for normalization (y = 1).

Figure 3 .
Figure 3. (a) Representative Western blots from a cellular degradation assay with PROTACs 14, 20, 22, 24, and 30 in the MOLM-13 cell line.Full membranes are shown in Figure S2.(b) METTL3 Western blot quantification by densitometry from a cellular degradation assay at PROTAC concentration of 2 μMin the AML cell lines MOLM-13, THP-1, NOMO-1, and KASUMI-1 and in the prostate cancer cell lines PC3 and DU145 (representative Western blots are shown in Figure S3).The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(c) Correlation of the degradation of METTL3 and METTL14 in MOLM-13 at PROTAC concentration of 2 μM; the dotted black line is a linear regression of all the compounds (corr.coeff.r = 0.95).The nine most active PROTACs are highlighted in turquoise.me-14 and me-24 (in red) are the methylated negative controls of PROTACs 14 and 24 (for structures, see Figure 4B).(d) Correlation between the number of −CH 2 − groups in the linker and the EC max value measured in the TCFA.The color of the data points reflects the degradation of METTL3 as measured by Western blot in MOLM-13 (legend on the right).While the EC max improves until the shortest length of the linker (one −CH 2 −), the highest degradation is observed at intermediate lengths, i.e., (−CH 2 −) 3 .(e) Scatter plot of the degradation of METTL3 in MOLM-13 as a function of the EC max values measured in the TCFA.The data points on the y-axis (red) did not show any detectable ternary complex formation in the TCFA.The linear regression for the compounds with both degradation and EC max values (blue) is shown (black continuous line).The horizonal dashed line marks a METTL3 degradation level of 20%, and the vertical dashed line marks EC max = 1 μM.
Figure 3. (a) Representative Western blots from a cellular degradation assay with PROTACs 14, 20, 22, 24, and 30 in the MOLM-13 cell line.Full membranes are shown in Figure S2.(b) METTL3 Western blot quantification by densitometry from a cellular degradation assay at PROTAC concentration of 2 μMin the AML cell lines MOLM-13, THP-1, NOMO-1, and KASUMI-1 and in the prostate cancer cell lines PC3 and DU145 (representative Western blots are shown in Figure S3).The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(c) Correlation of the degradation of METTL3 and METTL14 in MOLM-13 at PROTAC concentration of 2 μM; the dotted black line is a linear regression of all the compounds (corr.coeff.r = 0.95).The nine most active PROTACs are highlighted in turquoise.me-14 and me-24 (in red) are the methylated negative controls of PROTACs 14 and 24 (for structures, see Figure 4B).(d) Correlation between the number of −CH 2 − groups in the linker and the EC max value measured in the TCFA.The color of the data points reflects the degradation of METTL3 as measured by Western blot in MOLM-13 (legend on the right).While the EC max improves until the shortest length of the linker (one −CH 2 −), the highest degradation is observed at intermediate lengths, i.e., (−CH 2 −) 3 .(e) Scatter plot of the degradation of METTL3 in MOLM-13 as a function of the EC max values measured in the TCFA.The data points on the y-axis (red) did not show any detectable ternary complex formation in the TCFA.The linear regression for the compounds with both degradation and EC max values (blue) is shown (black continuous line).The horizonal dashed line marks a METTL3 degradation level of 20%, and the vertical dashed line marks EC max = 1 μM.

Figure 4 .
Figure 4. Control experiments, TCFA and in vitro ubiquitination.(a) METTL3 and METTL14 levels in the presence of lenalidomide or UZH2 in MOLM-13.The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(b) Chemical structures of PROTAC negative controls me-14 and me-24.(c) METTL3 and METTL14 degradation by PROTACs 14 and 24 in comparison with their negative controls me-14 and me-24 in MOLM-13.The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(d)BiochemicalFRET-based ternary complex formation assay (TCFA) with PROTACs 14 and 24 and their methylated negative controls me-14 and me-24.(e) In vitro ubiquitination assay results with compound 14 and its negative control me-14.All data originate from biological and biochemical duplicates or more.(f) Western blot analysis of the in vitro ubiquitination assay.The ubiquitination reaction mixture (E1, E2, CUL4A-RBX1, CRBN-DDB1, and METTL3-METTL14 in reaction buffer) was incubated with or without ATP and at different concentrations of compounds 14 and me-14 as indicated at 30 °C for 2 h.The proteins were separated by SDS-PAGE followed by Western blot analysis with α-METTL3, α-METTL14, and α-Ubiquitin antibodies.Shown here is one representative Western blot of three biological replicates of the experiment.For panel (e), densitometry was performed using the α-METTL3 blots of all three replicates.In the α-METTL14 blot, a weak band appears above the METTL14 band at 32, 8, and 2 μM of compound 14, presumably indicating monoubiquitination of METTL14.However, the ubiquitination of METTL3 is clearly more efficient.The band indicated with an asterisk (*) originates from the unspecific detection of METTL3 with the α-METTL14 antibody.

Figure 5 .
Figure 5. Cellular characterization.(a) Concentration dependence of METTL3 (red) and METTL14 (blue) degradation by PROTACs 30 and 14 and the methylated negative control of 14 (me-14) in MOLM-13 cells.The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(b) Representative Western blots.Full membranes are shown in Figure S5.

Figure 6 .
Figure 6.Cellular characterization.(a) Cell viability assay for PROTACs 20, 22, 24,and 30 and the METTL3 catalytic inhibitor UZH2 in AML cell lines (top) and prostate cancer cell lines (bottom).The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).(b) LC-MS quantification of m 6 A/A levels in polyadenylated RNA in MOLM-13.The dashed line represents the protein level of the DMSO control, used for normalization (y = 1).

Scheme 9 .
Scheme 9. Synthesis Route for Compound 32 a

Table 1 .
Synthesized PROTACs and Activity Data