Methotrexate-based PROTACs as DHFR-specific chemical probes

Summary

Methotrexate (MTX) is a tight-binding dihydrofolate reductase (DHFR) inhibitor, used as both an antineoplastic and immunosuppressant therapeutic. MTX, like folate undergoes folylpolyglutamate synthetase-mediated γ-glutamylation, which affects cellular retention and target specificity. Mechanisms of MTX resistance in cancers include a decrease in MTX poly-γ-glutamylation and an upregulation of DHFR. Here we report a series of potent MTX-based PROteolysis TArgeting Chimeras (PROTACs) to investigate DHFR degradation pharmacology and one-carbon biochemistry. These on-target, cell-active PROTACs show proteasome- and E3 ligase-dependent activity, and selective degradation of DHFR in multiple cancer cell lines. By comparison, treatment with MTX increases cellular DHFR protein expression. Importantly, these PROTACs produced distinct, less-lethal phenotypes compared to MTX. The chemical probe set described here should complement conventional DHFR inhibitors and serve as useful tools for studying one-carbon biochemistry and dissecting complex polypharmacology of MTX and related drugs. Such compounds may also serve as leads for potential autoimmune and anti-neoplastic therapeutics.

Graphical Abstract

eTOC blurb

Methotrexate is an essential cancer and disease-modifying anti-rheumatic drug. The work of Rana et al. creates methotrexate derivatives capable of selectively degrading dihydrofolate reductase, one of the drug’s major targets to potentially offer chemical probes to study the incompletely understood polypharmacological basis of the drug’s efficacy and toxicity.

Introduction

Dihydrofolate reductase (DHFR) is an essential enzyme in the folate metabolic pathway that catalyzes the NADPH-dependent reduction of dihydrofolic acid (DHF) to tetrahydrofolic acid (THF). Tetrahydrofolate is a cofactor in several one-carbon (“methyl”) transfer reactions which participate in amino acid and nucleobase synthesis. Methotrexate (MTX) has been used for decades as a potent anti-cancer agent (e.g., acute lymphoblastic leukemias), as well as to treat autoimmune diseases like rheumatoid arthritis, and the rare systemic disorder granulomatosis with polyangiitis.^1,2 MTX is a drug and prodrug with a complex pharmacology, leading to reduced intracellular levels of THF cofactors, which results in inhibition of thymidylate, purine, and DNA biosynthesis. Mechanistically, MTX undergoes poly-γ-glutamylation at its γ-carboxylic acid by folylpolyglutamate synthetase (FPGS) in cells, which leads to intracellular accumulation and inhibition of other folate-dependent enzymes such as thymidylate synthase (TYMS).^3,4 Reduced THF cofactor concentrations result in DNA replication arrest. Unfortunately, therapeutic resistance to MTX is a significant clinical barrier. Mechanisms of resistance include DHFR amplification, decreased intracellular transport of MTX, decreased retention of MTX (due to lack of poly-γ-glutamylation), increased export of MTX (due to overexpression of some members of the ATP binding cassette transporters), mutated DHFR (resulting in weak binding to MTX), increased levels of lysosomal γ-glutamyl hydrolase (leading to hydrolysis of MTX polyglutamates), and others.^5–8

Since the discovery of MTX⁹, numerous analogs have been designed and synthesized in attempts to improve MTX’s clinical efficacy, its side-effect profile, and to study its mechanism(s) of action.^10–13 For example, γ-fluoromethotrexate (FMTX) shows reduced cellular toxicity in preclinical models which are attributed to its poor polyglutamylation at the γ-carboxylic acid.¹⁴ Despite such efforts, MTX remains a mainstay of treatment for highly proliferative cancers and its pharmacology remains incompletely understood.^15–19

PROteolysis TArgeting Chimeras (PROTACs), which hijack human cellular quality control machinery to catalytically and selectively degrade proteins of interest (POI), offer an avenue to investigate MTX pharmacology.²⁰ This strategy involves heterobifunctional molecules connected by a linker wherein one end of the molecule binds to the POI and the other end of the molecule binds to an E3 ubiquitin ligase, leading to POI ubiquitinylation and proteasomal degradation, as opposed to conventional pharmacologic inhibition without protein degradation. This approach can degrade previously undruggable proteins including transcription factors²¹, kinases^22–25, and scaffold proteins²⁶, and in the case of MTX offers an alternative to drugging DHFR and characterizing the various targets of MTX. In terms of prior work related to DHFR degraders, Long and colleagues demonstrated that trimethoprim linked to the hydrophobic moiety (Boc)₃Arg degrades Escherichia coli DHFR in a ubiquitin-independent manner using a heterologous expression system in mammalian cells.²⁷

We designed and synthesized MTX-based PROTACs with the goal of greatly reducing the cellular concentration of human DHFR. Herein, we show these PROTACs selectively degrade cellular DHFR by the ubiquitin-proteasome system and display a folate-dependent cell viability phenotype distinct from the conventional DHFR enzymatic inhibitor MTX and other antifolates. The MTX-PROTAC analog with potent, selective DHFR degrading activity accompanied by low cellular toxicity has been designated versortrexate (VSTX). Such compounds and derivatives can serve as chemical tools for studying the complex regulation of DHFR and its role in one-carbon cellular metabolism, as well as the mechanism and pharmacology of antifolate therapeutics.²⁸

Results

Design and synthesis of MTX-based PROTAC chemical probes.

A series of MTX-based PROTACs (3–9) were designed by linking MTX to the E3 ligase small-molecule ligand (E3L), thalidomide of the cereblon (CRBN) substrate receptor subunit (Figure 1A). Several PROTACs containing either hydrophilic (polyethylene glycol, PEG) or hydrophobic (alkyl) linkers were synthesized (Figure 1A and S1). Linkers were attached at the γ-carboxylate position of MTX. Control analogs were also prepared by either methylating the CRBN glutarimide at the NH position (4 and 7)²⁹ or attaching a linker at the α-carboxylate group (5 and 8) of MTX which is critical for its binding to DHFR.³⁰ The effects of PROTACs 3–9 trifluoroacetate salts, which behave similarly to chloride salts in cell culture (Figure S2), were then characterized for recombinant DHFR binding, cellular toxicity and their ability to degrade cellular DHFR as compared to the effects of MTX (1) and the threo and erythro diastereomeric mixture of FMTX (2)¹⁵ on these parameters (Figure 1 and Table 1).

Figure 1. Methotrexate-based antifolate PROTACs.

(A) Structures of MTX, FMTX and MTX-PROTACs, with a polyglycol (3-5) or alkyl (6-9) linker tethering the CRBN E3 ligase ligand. IE3L analogs are N-methyl CRBN ligand derivatives lacking E3 ligase binding. (B) Relative DHFR binding affinities of MTX (1), FMTX (2), and PROTC analogs (3-9) determined using an MTX-Cy5 based FP assay (top panel). Cytotoxicity of the agents in HBL1 cells after 72 h as determined using CellTiter-Glo and normalized to a digitonin control (bottom panel). (C) DHFR levels measured in HBL-1 cell line from recombinantly expressed DHFR-HiBiT as described in Figure S3. Data are mean ±SD from three independent experiments.

Table 1.

Activity summary of antifolates and MTX-based DHFR PROTACs

Compound	Relative binding (pIC50)a	Cytotoxicity (pEC50)b	Cytotoxicity efficacy (viability, %CTG)	Degradation potency (pDC50)c	Assay max response (%)	Notes and compound NCATS ID
MTX, 1	7.52 ±0.08	7.21 ±0.12	−68.7 ±5.7	ND	48.1 ±4.2	Classic DHFR inhibitor
FMTX, 2	7.03 ±0.08	5.98 ±0.29	−65.7 ±5.7	ND	53.9 ±5.2	MTX analog, reduced polyglutamylation
3	7.36 ±0.05	5.55 ±0.37	−72.1 ±7.9	6.27 ±0.21	−40.6 ±1.8	PROTAC, pioneer analog
Versortrexate (VSTX), 6	7.07 ±0.03	4.67 ±0.04	−45.7 ±26.9	6.43 ±0.18	−70.1 ±0.4	PROTAC, active acid, NCATS-SM8586
VSTX-IE3L, 7	7.60 ±0.10	ND	6.2 ±4.4	ND	39.3 ±6.6	PROTAC, CRBN E3L non-binder, NCATS-SM8588
VSTX-αE3L, 8	5.14 ±0.39	ND	−2.0 ±16.2	ND	21.2 ±2.5	DHFR non-binder, α-amide, NCATS-SM8589
VSTX-ME, 9	ND	5.24 ±0.30	−26.8 ±22.7	7.56 ±0.29	−71.9 ±2.1	PROTAC, methyl ester 6 prodrug, NCATS-SM8587

^aFP, mean ±SD, from three replicates.

^bDetermined by CTG in HBL-1 cells after 72 h.

^cDHFR-HiBiT, 24 h, 30°C, HBL-1 cells, mean ±SD, from three independent experiments. ND, not determined (inactive).

On-target DHFR degradation by MTX-based PROTACs.

A fluorescence polarization (FP)³¹ assay was developed to measure PROTAC interaction with a recombinant human DHFR•NADPH complex through competition with a far-red MTX fluorescent probe (Figure S3). A new Cy5-based probe was developed instead of using commercially available MTX-fluorescein³² probes due to spectral interference between the PROTAC thalidomide and MTX-fluorescein at higher MTX-PROTAC concentrations (Figure S3). Under optimized conditions in a 1536-well format, the assay detected low nanomolar binding of unlabeled MTX (IC₅₀ = ~30 nM), confirming sensitivity in the range of those values reported in the literature.^33,34 This miniaturized, robust assay measured the nanomolar binding affinity of the prototypical DHFR inhibitors MTX and FMTX (pIC₅₀ = 7.52 ±0.08 and 7.03 ±0.08, respectively; Figure 1B, top panel). PROTACs 3–9 were then assessed for their ability to displace the MTX-Cy5 probe (Tables 1 and S1). Notably, PROTAC 6 (versortrexate, VSTX) showed nanomolar binding (pIC₅₀ = 7.07 ±0.03; Figure 1B, top panel), as did the non-E3 ligase binding analog 7, VSTX-IE3L. In addition, we synthesized a VSTX ‘metabolic precursor’, VSTX-ME (9), by esterifying the α-carboxylate of VSTX to potentially improve cell permeability and function as a prodrug ester, which should be hydrolyzed to VSTX within the cell.^35,36 As anticipated, neither the prodrug ester 9 (VSTX-ME), nor the α-amide 8 (pIC₅₀ = 5.14 ±0.32) bound recombinant DHFR•NADPH with high affinity, as determined by the FP assay due to the importance of the free α-carboxylic acid group required for folate and MTX binding to DHFR (Figure 1B, top panel).³⁰

We next evaluated MTX, FMTX and the PROTACs in a 72 h cell viability assay (Figure 1B, bottom panels). MTX was potently cytotoxic (pEC₅₀ 7.21 ±0.12) toward HBL-1 cells while FMTX which undergoes diminished cellular polyglutamylation was over an order of magnitude less cytotoxic (pEC₅₀ 5.98 ±0.29, Table 1). This is consistent with the fact that polyglutamylated MTX engages other protein targets that contributed to its higher cytotoxicity.^1,3

Poly-³-glutamylation resistant PROTAC 3, yielded a pEC₅₀ of 5.55 ±0.37, on the order of FMTX (Table 1). However, it is revealing to note that 3 also showed an order of magnitude improvement in cytotoxicity when compared to PROTAC 4, which retains similar DHFR binding affinity to 3 but is incapable of forming a ternary complex between DHFR and E3 ligase (Figure 1C). Taken together these data suggests that DHFR degradation mediated by ternary complex formation underlies the observed cytotoxic response when compared to a control PROTAC lacking the capacity for ternary complex formation. A similar trend was apparent with the corresponding VSTX analogs, though 6 displayed ~10-fold lower cytotoxicity than 3 (Figure 1B and Table 1).

This pattern was reflected in the degree of apoptosis biomarkers as measured from the time-dependent activation of caspase 3/7 activity (Figure S4). Interestingly, in MOLT-4 cells both PROTAC 3 and 4 displayed more robust CTG measured cytotoxicity (Figure S5A) reaching the efficacy of the digitonin control with corresponding caspase 3/7 activation (Figure S5B). However, PROTAC 3 is more cytotoxic than PROTAC 4 by an order of magnitude. VSTX (6) and VSTX-ME (9) also showed a full CTG response (Figure S5A), however only at significantly higher concentrations and were accompanied by marginal activation of caspase 3/7 (Figure S5B).

Overall, the reduced cytotoxicity observed from PROTACs 3 and 6 versus the respective non-E3 ligase ligand (IE3L) analogs (4 and 7), suggests in this context, the levels of DHFR depletion orchestrated by these PROTACs cannot mimic the cellular toxicity of MTX, further highlighting the complex polypharmacology of this antifolate.

The ability of MTX-based PROTACs to degrade cellular DHFR was quantified using a high-throughput DHFR HiBiT assay based on NanoLuc luciferase complementation in a miniaturized 1536-well format (Figure S3E).^37,38 In this enzyme complementation assay, the sequence for a non-perturbing 11-mer peptide fragment of NanoLuc luciferase is encoded after the C-terminus of the DHFR protein. Following compound treatment, cells are lysed, and a larger complementary luciferase protein (LgBiT) and substrate are added to produce bioluminescence proportional to the amount of cellular DHFR-HiBiT.

In MTX-sensitive HBL-1 cells (EC₅₀ < 100 nM), MTX treatment leads to a documented increase in cellular DHFR (Figure 1C).³⁹ By contrast, treatment with PROTAC 3 for 24 h decreased DHFR concentration in the HiBiT assay to ~40% of DMSO control with sub-micromolar potency (pDC₅₀= 6.27, Figure 1C). However, VSTX (6) and its corresponding α-methyl ester VSTX-ME (9) were more efficacious and nearly 10-fold more potent in degrading DHFR in HBL-1 cells (pDC₅₀ = 7.56 ± 0.29, DC_max = −71 ± 5%; Figure 1C and Table 1). As expected for a DHFR binder, the E3 ligase non-binding analog VSTX-IE3L (7) led to an increase in cellular DHFR-HiBiT like MTX (Figure 1C). By contrast, the DHFR non-binding analog VSTX-αE3L (8) did not lead to appreciable changes in DHFR-HiBiT concentration due to lack of binding to DHFR (Figure 1C).

Versortrexate degrades DHFR across a range of cell lines.

The cellular degradation of DHFR in HBL-1 cells by VSTX-ME (9) was confirmed by western blot (Figure 2A). Analysis using additional cell types, including HEK-293, the human T lymphoblast cell lines, Jurkat, MOLT-4 and DND-41, in addition to TALL-5, a patient-derived cell line that is propagated as a xenograft in immunodeficient mice demonstrated that VSTX-ME (9) was consistently able to degrade DHFR across this cell panel at sub-micromolar concentrations. However, DHFR levels did not decrease to undetectable levels by western blot analysis (Figure 2A–F) reflective of the results observed in the DHFR-HiBiT HBL-1 cell line (Figure 1C) where degradation was observed at lower concentrations (Figure 2D, left panel) and a hook effect was observed at higher concentrations (Figure 2D, right panel). These observations may be related to a sequestered DHFR pool inaccessible to PROTAC 9 (e.g., mitochondrial DHFR) and the pharmacological hook-effect limitation^40,41 of the PROTAC, respectively.

Figure 2. MTX-based PROTAC Versortrexate-ME (9) induces DHFR degradation in multiple cell types.

Cell lines, (A) HBL-1, (B) HEK-293, (C) Jurkat, (D) MOLT-4 (right, expanded and extended titration range around grey area in left panel), (E) DND-41, and (F) human TALL-5 xenograft cells were treated with VSTX-ME (9), DMSO vehicle or media only (0% DMSO) as indicated for 24 h prior to harvesting cells and extracting protein for western blot analysis using anti-DHFR (►), -GAPDH (►), and -TYMS (▷) for blots. HBL-1, DND-41 and TALL-5 cells were also treated with MTX (1) at the indicated concentrations. Bar graphs quantify the ratio of either DHFR (black bar) or TYMS (white bar) to GAPDH. GAPDH was used as a loading control. Control levels (i.e., untreated) for DHFR (solid line) and TYMS (dashed line) are represented by the average value for the media and DMSO controls. Data shown are representative of 2-3 experiments (additional data see Figure S6). See Methods for additional details.

Further corroborating the observations in the DHFR-HiBiT HBL-1 cell line (Figure 1C), MTX (1) results in a DHFR increase relative to DMSO or media controls within the 24 h time window used in these western blot experiments (Figure 2A). Of note is the lack of substantial change in TYMS levels, a target of polyglutamylated MTX, when measured in this cell line panel compared to DHFR (Figure 2A–F, open vs black arrowheads).

Mechanistic evaluation of versortrexate.

The kinetics of DHFR degradation by VSTX (6) or VSTX-ME (9) were observed to lead to robust DHFR-HiBiT degradation after 5 h with the maximum depletion occurring at 24 h and not significantly changing out to 72 h treatment in HBL-1 cells (Figure 3A). By contrast, DHFR levels continue to rise throughout the MTX (1) and FMTX (2) treatments while cytotoxicity remains essentially similar for 24, 48, and 72 h.

Figure 3. DHFR degradation by MTX-based PROTACs are consistent with a CBRN E3 ligase-mediated mechanism of action.

(A) Time course of DHFR-HiBiT levels and corresponding cell viability measured by antifolates and MTX-PROTAC 6 and 9. (B, left panel) Co-treatment (5 h) with CRBN ligand pomalidomide (POM, or MTX (1) prevents DHFR degradation by 9 in HBL-1 cells, (right panel) Versortrexate-IE3L (7), does not efficiently bind CRBN, thereby preventing ternary complex formation in HBL-1 cells as measured after 5 h treatment. (C) Co-treatment of HBL-1 cells with MG-132 demonstrates DHFR degradation by Versortrexate-ME (9) is proteosome-dependent. (D) DHFR degradation by 9 is neddylation-dependent, as illustrated by MLN-4924 pre-treatment (2 h) of HBL-1 cells to prevent 9 mediated DHFR degradation at 24 h. (E) DHFR degradation in HBL-1 cells by 3 and 9 treatment is DHFR transcription-independent as measured by qRT-PCR after 5 h (left) or 24 h (right). Data, normalized to USF2, are mean ± SD from three technical replicates. (F, left) DHFR degradation by 9 in HBL-1 and MOLT-4 cells is modestly suppressed in HBL-1 cells by cellular folic acid, and enhanced (right) by the addition of the translation inhibitor cycloheximide (CHX). See Fig. S9 for densitometry. (G) Matrix testing with 9 or 1, where the top row represents 100% (black) to 0% (red) cell viability. Max concentrations are listed in μM and are titrated as 1:2 dilutions, see Supplementary Fig. S10 for individual matrix concentrations. Bottom row represents Δ excess HSA synergy metrics. Scale indicates positive (antagonistic, blue), near zero (white), and negative (synergistic, red) excess HSA, respectively. Data are mean from three inter-plate technical replicates.

Several experiments were next performed with 9 and its analogs to assess if these MTX-based compounds function by a PROTAC mechanism of action (MOA; Figure 3B–D) involving a protein of interest (POI) target • E3 ligase complex.⁴² Here the formation of the POI•PROTAC•E3 ligase ternary complex is supported by two lines of evidence. First, the degradation of cellular DHFR by 9 was reduced by co-treatment with either MTX (1) or the CRBN ligand pomalidomide (POM, Figure 3B, left panel). Second, treatment with 7 (an analog of VSTX that binds DHFR but not E3 ligase) did not reduce cellular DHFR, an observation that supports an E3 ligase-dependent MOA (Figure 3B, right panel).

Consistent with a proteasome-dependent MOA, degradation of DHFR by 9 was attenuated by the proteasome inhibitor MG-132 (Figure 3C). A similar effect was observed for 9 when treating HBL-1 cells with the NEDD8-activating enzyme inhibitor MLN-4924 (Figure 3D). This latter observation demonstrates that DHFR degradation by 9 requires neddylation, another critical component of the Cullin-RING E3 ubiquitin ligase mediated ubiquitin-proteasome pathway.⁴³ Importantly, the neddylation inhibitor MLN-4924 itself did not grossly perturb cellular DHFR concentrations in HBL-1 cells (Figure S7).

Additionally, qRT-PCR demonstrated DHFR expression by HBL-1 cells in the presence of 6 or 9, indicating the reduced cellular DHFR protein from PROTAC treatment is unlikely due to changes at the transcriptional level (Figure 3E).^44,45 This observation and the effect of MTX (1) to upregulate DHFR levels are mutually compatible with a known regulatory mechanism whereby DHFR protein binds to its mRNA to inhibit DHFR translation.⁴⁶ These collective data indicate the prototypical MTX-based PROTAC 9 decreases cellular DHFR concentrations in a manner consistent with a PROTAC MOA. Characterization and mechanistic experiments with 3, another DHFR degrader with a PEG-based linker, yielded similar findings, albeit with weaker degradation potency (Figure S8).

The effect of folic acid on DHFR degradation was investigated, given this cofactor, a DHFR substrate and cell culture media supplement would be expected to compete with the binding of de-esterified 9 (e.g., 6) to DHFR. Indeed, as observed by western blot, the degradation of DHFR by 9 was modestly enhanced when HBL-1 cells were deprived of folic acid (Figure 3F, left and Figure S6), a point that will be more deeply studied in the following section. The degradation of DHFR by 6 or 9 was further enhanced when HBL-1 or MOLT-4 cells were co-treated with the translation inhibitor cycloheximide (CHX, Figure 3F, right). The MTX (1)-mediated upregulation of DHFR levels were similarly depressed by CHX treatment, suggesting that de novo DHFR synthesis in response to a DHFR PROTAC may follow a mechanism analogous to that observed with MTX (Figure S6). This data suggests potential approaches for degrading residual DHFR after PROTAC treatment.

To further contrast the mechanistic pharmacology of an MTX-PROTAC and classic antifolate we compared VSTX-ME (9) to MTX in a matrix cross-titration cell viability assay (Figure 3G). The experiment confirmed several salient antagonistic combinations (Figure 3G, bottom panels, blue synergy scores) explainable by a DHFR target-based E3 ligase-mediated process, including folinate, a MTX antagonist and a reduced tetrahydrofolate that does not require DHFR for conversion to other reduced folic acids; POM, a CRBN ligand; and MLN-4924, a neddylation inhibitor, all used to investigate the MOA (Figure 3B, C and F). The folinate effect (antagonism of antifolate-mediated cytotoxicity) was more pronounced in the MTX vs VSTX-ME (9) titration, as reflected by the magnitude of the excess highest single agent (HSA) score, which reflects the added effect over either single agent alone.⁴⁷ This observation is likely influenced by the higher assay sensitivity toward MTX. No significant synergy or antagonism of cytotoxicity was observed for MTX in combination with VSTX-ME, consistent with our current mechanistic understanding of these agents.

Versortrexate cell toxicity is highly folate dependent.

As shown in Figures 1 and 2, VSTX (6) and VSTX-ME (9) did not show a pronounced effect on cell viability, although they efficiently degrade DHFR in several cell lines. This was also observed in a broad 49 cancer cell line profile measuring cell count and apoptosis induction after 5 days as compared to our control, MTX and the potent apoptotic inducer staurosporine⁴⁸ (Figure S11). To understand better the relationship between DHFR degradation by PROTAC 6 and 9 and their effect on cell viability, additional functional experiments were performed.

The effect of these DHFR-based PROTACs was further examined by live-cell imaging. In HEK-293 cells, nanomolar concentrations of MTX (1) and low micromolar concentrations of FMTX (2) reduced cellular confluence beginning after 48 h of treatment, recapitulating previous observations^15,49, whereas even micromolar concentrations of 6 and 9 did not grossly affect cellular confluence (Figure 4A). These data from an independent and adherent cell line corroborate the cytotoxicity trends observed in the nonadherent suspension HBL-1 cell line, with 6 and 9 minimally perturbing cellular growth at only micromolar compound concentrations after several days of treatment. The precipitous apparent drop in confluence observed at 25 μM 9 is likely due to measurement interference resulting from the lower solubility of this α-methyl ester analog.

Figure 4. Versortrexate cellular toxicity is folate dependent.

(A) HEK293 cell growth kinetics by live-cell imaging. Data are representative results from one of three independent experiments. (B) PROTAC 6 and 9 demonstrate a greater exogenous folic acid dependence on HBL-1 cell viability, as measured by intracellular ATP, than MTX or FMTX. Data are mean ±SD of eight technical replicates. Potency comparison at 72 h for folate depleted media: EC₅₀ (6) = 2.46 μM; EC₅₀ (9) = 0.26 μM; EC₅₀ (2) = 0.32 μM; EC₅₀ (1) = 0.013 μM. The aberrant measurement occurring with 25 μM VSTX-ME is likely due to its lower aqueous solubility.

Expanding on the observed folate sensitivity of VSTX-ME (9) to DHFR degradation (Figure 3F and G), the effect of supplemental folic acid in cell culture media was examined in greater detail given the importance of reduced folate levels in MTX-based cytotoxicity. After culturing HBL-1 cells in folic acid-free or folic acid-replete RPMI media and 10% dialyzed fetal bovine serum for 3 days, cells were treated with either MTX (1), VSTX 6, or α-carboxyl methyl ester 9. Compared to folic acid-containing controls, there was a time- and concentration-dependent increase in cytotoxicity for MTX (1), FMTX (2), VSTX (6), and VSTX-ME (9) grown in folic acid-free conditions (Figure 4B). Since folic acid and MTX bind to DHFR, not surprisingly, MTX treated cells showed higher cytotoxicity towards HBL-1 cells in folic acid-free media. MTX and its polyglutamylated congeners inhibit DHFR and TS simultaneously to induce a cytotoxic response. Folic acid starved cells when switched to folic-acid replete media at the time of MTX addition showed a slight decrease in cytotoxicity suggesting direct competition of folic acid and MTX. Similar results were observed with FMTX treatment (Figure 4B). However, compared with MTX and FMTX, a dramatically more pronounced shift in cytotoxicity of 6 and 9 in folate-depleted vs. folate-containing media was observed, highlighting effects of endogenous folic acid on VSTX-based cell toxicity. For example, while the cytotoxicity of MTX (1) increased by < 2-fold in folate-depleted vs. folate-containing media, that of 6 and 9 increased consistently over time to 2-orders of magnitude (Figure 4B and Supplementary Table S1 and Figure S12). Folic acid-starved cells switched to folic acid-replete media at the time of compound addition (Figure 4B, ‘Rescue’) showed intermediate cytotoxicity profiles in terms of potency and efficacy. VSTX-IE3L (7), that binds to DHFR but cannot promote ternary complex formation, did not exhibit appreciable toxicity in any of the conditions, suggesting that DHFR degradation has an improved phenotypic response over inhibition of DHFR alone. Furthermore, DHFR binding-incompetent VSTX-αE3L (8), capable of binding to E3 ligase machinery, showed no cytotoxicity and suggests that antagonizing the E3 ligase component alone does not contribute to cell cytotoxicity.

DHFR degradation by versortrexate-ME escapes MTX resistance (9).

To determine how acquired mechanisms of MTX resistance would affect VSTX-mediated DHFR degradation, we generated an HBL-1 MTX-resistant (MTX^R) cell line by prolonged culturing of HBL-1 cells in the presence of increasing concentrations of MTX. In this cell line, after withdrawal of the 1 μM MTX selection for 10 days to clear any highly retained MTX polyglutamates^15,50, the potency of MTX (1) right-shifted by greater than 100-fold, while the modest effect on cell viability of 9 is unchanged, between the parental (MTX^S) and MTX^R HBL-1 cell lines (Figure 5A). However, VSTX-ME (9) remained capable of degrading DHFR in the MTX^R cells, with a potency and efficacy similar to that of the MTX^S HBL-1 cell line (Figure 5C, compare black bars). Interestingly, the otherwise robust increase in DHFR protein levels induced by MTX (1) in the MTX^S HBL-1 cell line was no longer observed in the MTX-resistant cell line (Figure 5C, compare grey bars).

Figure 5. Versortrexate retains DHFR-degrading activity in MTX-resistant cells.

(A) Effect of digitonin (left), VSTX-ME (9, center) and MTX (1, right) on the cell viability of MTX sensitive (MTX^S, ○) and MTX resistant (MTX^R, ●) HBL-1 cells. MTX EC₅₀ values are 45.2 nM (○) and EC₅₀, 4.64 μM (●). Data are mean of two to three replicate titrations; error bars are SD from replicates. (B) Western blot analysis of DHFR levels in response to either 9 or 1 on MTX^S (left) and MTX^R (right) HBL-1 cells. Data shown are representative of 3 experiments (additional data see Figure S13). (C) Densitometry quantification of western blot.

Cellular proteomics and metabolomics of DHFR PROTACS.

The effect of PROTACs 3 and VSTX-ME (9) on the HBL-1 cellular proteome was investigated by label-free quantitative proteomics analysis.⁵¹ VSTX-ME 9 showed significant degradation of cellular DHFR in HBL-1 cells after 24 h, while the decrease in DHFR mediated by 3 compared to 9 was considerably less (Figure 6A, left upper panels) and paralleled the observation from the DHFR-HiBiT assay (Figure 1C). By contrast, treatment with the conventional DHFR enzymatic inhibitor MTX (1) led to a significant ~2-fold upregulation of cellular DHFR levels under identical conditions (Figure 6A, left upper panels), in agreement with our other data (Figures 1 and 2). Normalizing the effects of 3 and 9 to MTX (1) further separates the degradation of DHFR from the folate cofactor-dependent one-carbon metabolism enzymes, which were not significantly changed in relative abundance (Figure 6A, left lower panels). Overall, PROTAC 9 resulted in the most pronounced effect on DHFR levels as assessed by this proteomics analysis.

Figure 6. MTX-based PROTACs produce distinct changes in the HBL-1 proteome.

(A) HBL-1 cells were incubated with DMSO, MTX (1), 3, or versortrexate-ME (9), followed by label-free quantitative proteomics analysis as described in Supplementary Methods. PROTAC 3 and 9 (1 μM) show significant degradation of cellular DHFR in HBL-1 cells after 24 h. Treatment with MTX (1) leads to significant upregulation of cellular DHFR levels under identical conditions. One-carbon folate metabolism proteins include: FPGS, GGH, DHFR, TYMS, GART, ATIC, SHMT1/2, MTHFD1/2/L, MTHFR, MTHFS, MTFMT, and FOLH1. Labeled on the upper left and upper right are the number of proteins with a measured FC of 0.01 or 100 respectively. Dashed lines indicate p-value = 0.05 and FC = ±2. (B) VSTX-ME (9) displays the expected effects on cellular dUMP and purine nucleotide levels for DHFR selective inhibition. Metabolite levels by targeted HPLC-MS/MS analysis were scaled and normalized to protein concentration of the sample. Ratio between scaled metabolite signal and protein concentration is displayed on the y-axis. Four technical replicates for each sample were run and p-values calculated by t-test.

Because the affinity of MTX for TYMS is poly-γ-glutamylation dependent, increasing by nearly three orders of magnitude for MTX-glu₅⁵², poly-γ-glutamylation-resistant VSTX, not surprisingly, has no effect on TYMS abundance. Proteomics analysis further indicated that MTX (1), FMTX (2), 3, and 9 shared a common set of proteins that were significantly increased, while there was more inter-compound variability for proteins that were significantly decreased with respective treatment.

The effects of MTX (1) and VSTX-ME (9) were also investigated with targeted metabolomics by measuring biomarkers tied to thymidine and purine biosynthesis. As observed in western blot analysis, cellular TYMS abundance was unchanged by VSTX-ME treatment (Figure 2A), therefore we measured cellular dUMP levels to determine if TYMS function was impacted by VSTX-ME compared with MTX. As expected for a thymidylate synthase inhibitor MTX treatment resulted in a significant increase in dUMP levels, while VSTX-ME treatment led to only a modest increase (Figure 6B). These observations further support a specific effect of VSTX-ME (9) on DHFR in contrast to MTX (1). Correspondingly, the purine nucleotides ATP and GTP, decreased with VSTX-ME treatment relative to untreated control, as would be expected for reduced cellular DHFR activity, though not reaching the level observed with the MTX treatment (Figure 6B). The larger effect of MTX on purine nucleotides may be due to MTX blockade of early steps of purine synthesis.⁵³ Interestingly, VSTX-ME (9) had a larger effect on the levels of purine nucleotides at 0.1 μM vs the 0.5 μM PROTAC concentration, possibly related to the intrinsic hook effect phenomenon seen with PROTAC molecules at higher concentrations where the binary complex is favored over the ternary complex needed for DHFR E3 ligase engagement.⁴¹

Discussion

A series of MTX-based PROTACs were designed and characterized for their ability to selectively degrade cellular human DHFR (Figure 1). The exemplar compound 6 we have named versortrexate (VSTX) binds DHFR and degrades cellular DHFR at nanomolar compound concentrations while sparing other primary one-carbon pathway targets (e.g., TYMS; Figures 2 and 6), and can do so without causing substantial cytotoxicity in the presence of folate-replete media (Figure 4). The VSTX α-methyl ester, VSTX-ME (9), a metabolic precursor of VSTX, demonstrates improved potency over VSTX and maybe more useful for in vitro cell culture studies. An accompanying pair of inactive controls versortrexate-IE3L (7, VSTX-IE3L) and versortrexate-αE3L (8, VSTX-αE3L) do not bind the CRBN E3 ligase and the DHFR POI, respectively (Table 1).

Biological characterization revealed several important pharmacological differences between MTX and MTX-based PROTACs. First, treatment with MTX leads to the previously described upregulation of cellular DHFR³⁹, whereas treatment with MTX-based PROTACs leads to pronounced (but not complete) depletion of cellular DHFR. Second, the MTX-PROTACs (3, 6, 9) are significantly less cytotoxic compared to MTX, as determined by the CTG and Caspase Glo 3/7 assays (Figures 1B, 3A, 4B, S4 and S5), a property that is folate-dependent (Figure 4B).

For MTX to exert an antiproliferative effect, the molecule must efficiently enter cells through the folate transporter (RFC-1) or via folate receptor-mediated endocytosis and subsequently become polyglutamylated by FPGS⁵⁴. MTX-polyglutamates are highly retained within cells, maintain high affinity for DHFR, and by virtue of the polyglutamylated state become potent inhibitors of additional folate cofactor-requiring enzymes (e.g., TYMS and ATIC)⁴.

Treatment of cells with MTX results in diminished purine/pyrimidine synthesis and paradoxically increased translation and accumulation of cellular DHFR.³⁹ Furthermore, MTX-polyglutamylates are potent inhibitors of TYMS, blocking dUMP methylation to suppresses thymidine synthesis. Taken together, these two pathways along with others contribute to the profound cytotoxicity of MTX (Figure 7A). On the other hand, the γ-fluoro-D,L-glutamic acid MTX analog, FMTX (2), is a poor FPGS substrate and undergoes significantly diminished polyglutamylation¹⁵, inhibits DHFR similarly to MTX (Figure 1B), but has far less of an effect on TYMS. As a result, the general cellular toxicity profile mimics MTX at higher concentration (Figures 4A), consistent with residual prodrug properties where TYMS is partially inhibited (Figure 7B). By contrast, VSTX, devoid of a γ-glutamyl carboxylate required for FPGS-dependent polyglutamylation, cannot strongly interact with TYMS, a fact supported by a lack of cytotoxic agonism or antagonism when combined with 5-fluorouracil or Raltitrexed, agents which target the TYMS dUMP and 5, 10-methylene THF binding sites, respectively (Supplementary Figure S14). However, when exogenous folate levels are depleted (Figure 4B) in the presence of VSTX-ME (9), a cytotoxic cellular response resembling FMTX treatment arises (Figure 7C), suggesting that significantly depleted cellular DHFR, in the presence of low folate levels, can phenocopy the essential nature of DHFR (DepMap, accessed 13Oct2021) to cellular viability.⁵⁵

Figure 7. Differential mechanisms between MTX antifolate analogs.

(A) MTX exerts potent cytotoxicity through a polypharmacological effect by directly inhibiting DHFR and via polyglutamylation by FPGS to become a TYMS inhibitor. (B) FMTX, a poor FPGS substrate, inhibits DHFR similar to MTX, but has less effect on TYMS, and consequently is less cytotoxic than MTX. MTX and FMTX treatment result in an increase in cellular DHFR levels. (C) VSTX-ME, hydrolyzed to VSTX upon cellular uptake, does not contain a free γ-carboxylate needed for FPGS-dependent polyglutamylation, therefore does not inhibit TYMS. In folate-replete media, VSTX-mediated DHFR degradation dramatically lowers cellular DHFR levels while leaving cells viable, whereas in folate-depleted media cellular toxicity is exacerbated. Black, grey, and white ovals signify strongly, weakly, and not inhibited enzymes, respectively. CellTiter-Glo viability pEC₅₀ (72 h) from Figure 4b

γ-polyglutamylation-resistant MTX-PROTAC 3 and 6 are not subject to retention by cells through FPGS activity. Therefore, MTX-PROTACs of this design would be expected to be unaffected by FPGS mutations and upregulated γ-glutamyl hydrolase activity that decrease MTX efficacy.⁷ In general, acquired mechanisms of resistance that decrease MTX cytotoxicity may not impact the ability of VSTX to degrade DHFR, as we observe in MTX^R HBL-1 cells (Figure 5).

This report describes the initial development and characterization of MTX-based PROTACs targeting DHFR. This probe set was characterized according to current best practices, including target engagement and selectivity, mechanistic experiments, and phenotypic characterization.^40,56 These heterobifunctional compounds can potently and specifically degrade cellular DHFR by a UPS-dependent MOA. This specificity can help dissect the complex MTX polypharmacology and allow for a more focused interrogation of DHFR-based pathways and one-carbon metabolism. The use of these chemical probes, versortrexate 6 and its corresponding ester 9, including the accompanying inactive analogs VSTX-IE3L (7), and VSTX-αE3L (8), should complement conventional non-selective DHFR enzymatic inhibitors (e.g., MTX) and targeted genetic perturbation platforms as valuable tools for studying one-carbon biochemistry.⁵⁷ For example, while RNAi-mediated target depletion may have less potential off-target activity than a small molecule PROTAC, the advantages of MTX-based PROTACs over a DHFR-targeting siRNA include a rapid onset of activity of hours vs days, and the titratable properties of a pharmacological agent, which can be rapidly washed-out⁵⁸, unlike siRNAs or antifolates that undergo prolonged accumulation via polyglutamate formation. Further, direct application of MTX-based PROTACs to microtiter plate-cultured suspension or adherent cells, and ready uptake by all cell types so far examined provide additional practical experimental advantages. In contrast, the duration of siRNA-mediated gene silencing is governed by cell division-based dilution with transfection efficiently dependent on cell type, with leukemia cell lines particularly refractory, requiring electroporation techniques.⁵⁹

Limitations of the study

Sequestered DHFR pools inaccessible to the PROTACs (e.g., mitochondrial DHFR) and the pharmacological hook-effect which occurs at high PROTAC concentrations are current potential limitations of MTX-based PROTAC chemical probes. Although versortrexate degrades DHFR in all the cell lines tested in our study we don’t know about the overall capability of DHFR degradation in other cell lines. Versortrexate and its analogs serve as tool compounds, and our experimental results, derived from various cancer cell lines, do not provide enough information to predict in vivo potential. Versortrexate is a specific DHFR degrader and spares TYMS as observed by western blot analysis and other one carbon pathway proteins as observed by unbiased whole cell proteomics studies; however, time dependent studies or ternary complex formation studies were not performed in the current manuscript.

STAR Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to the lead contact, James Inglese (jinlgese@mail.nih.gov).

Materials availability

The E. coli expression plasmid used to prepare human DHFR-His10, pET21a HsDHFR-AviTag-His10 (ID 193883), and the plasmid used to prepare the HBL-1 DHFR HiBiT cell line, pBMN-1 DHFR-HiBiT (ID 179731) have been deposited into Addgene.

The HBL-1 DHFR-HiBiT and HBL-1 MTX (1μM)-resistant cell lines developed for this study are available under an MTA.

MTX-PROTACs and MTX-Cy5 are available under an MTA unless becoming commercially available.

Data and code availability

• The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (proteomecentral.proteomexchange.org/)⁶⁰ via the PRIDE partner repository with the dataset identifier PXD030140 and DOI 10.6019/PXD030140.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETIALS

Cell lines and cell culture.

Cells were obtained from the following sources: HBL-1 (Louis Staudt, NCI, NIH, USA), MOLT-4 (ATCC, cat # CRL-1582), HEK293 (ATCC, cat # CRL-1573) Jurkat (ATCC, Clone E6-1), DND-41 (DSMZ, ACC-525). These cell lines, including TALL-5 cells, were selected because of their various sensitivities to methotrexate. Cell line identities were confirmed by short tandem repeat profiling. Cell cultures were routinely tested for Mycoplasma contamination using the MycoAlert PLUS Mycoplasma Detection Kit (Lonza Bioscience, cat # LT07) according to manufacturer protocol. Cell counts were measured by Countess automated cell counter (Invitrogen) and 0.4% trypan blue solution (Invitrogen, cat # T10282).

Human embryonic kidney HEK293T cells were maintained in DMEM (Gibco, cat # 11965) supplemented with 10% fetal bovine serum (Hyclone, cat # SH30071.03), 2 mM L-glutamine, 1 mM HEPES (Gibco, cat # 15630) and 100 U/mL penicillin/streptomycin (Invitrogen, cat # 15140). Unless noted otherwise, the lymphoma cell line HBL-1 were maintained in RPMI (Life Technology, cat # 11875) supplemented with 10% fetal bovine serum (v/v; Hyclone, cat # SH30071.03), 2 mM L-glutamine, 1 mM HEPES (Gibco, cat # 15630) and 100 U/mL penicillin/streptomycin (Invitrogen, cat # 15140). For efficient retroviral infection and transduction, HBL-1 cells were engineered to express the murine ecotropic retroviral receptor.⁶¹ To generate MTX-resistant cells (“HBL-1R”), HBL-1 cells were cultured with increasing MTX concentrations over several months starting with 0.05 μM of MTX. Endpoints were 1, 5, and 10 μM MTX-resistance. All cells were cultured in incubators maintained at 37 °C, with 5% CO₂, and 95% humidity unless otherwise stated.

Methotrexate resistant (MTX^R) cell line development.

HBL-1 (Human diffuse large B-cell lymphoma) cells were cultured in RPMI 1640 growth medium (Gibco) supplemented with 10% FBS (HyClone), 1% Pen/Strep (Gibco), and 1% HEPES Buffer Solution (Gibco) and incubated at 37 °C with 5% CO₂ at 95% relative humidity. MTX was added to the growth culture medium at 100 nM and replaced twice a week for four weeks until cells started proliferating again. The concentration of MTX in the growth medium was increased approximately every two weeks (200 nM, 500 nM, 1 uM) with cells passaged and medium replaced as needed. Cells developed resistance to MTX at 1 μM after three months.

METHOD DETAILS

Biology Methods

Compounds and reagents.

Test compounds were typically prepared as 10 mM stock solutions dissolved in neat DMSO and stored at < −30 °C until use. All compounds were subjected to internal quality control; most demonstrated greater than 95% purity and detection of an expected parent ion by UPLC-MS (ESI).

DHFR expression and purification.

Hs-DHFR-AviTag-His10 (NCBI Gene ID 1719) cDNA was synthesized and cloned into pET21a by Bio Basic (Amherst, NY). The expression plasmid was transformed into BL21(DE3) E. coli competent cells and 50 μL of cell mixture was plated on an LB agar plate containing 50 μg/mL ampicillin. Starter culture (50 mL LB with 50 mg/mL ampicillin, “LB/AMP50”) was inoculated with a streak of colony and the culture was allowed shake overnight at 37 °C. Four 1 L LB/AMP50 in 2800 mL flasks were each inoculated with 10 mL overnight starter culture and the culture were shaken at 200 rpm and 37 °C until OD600 = 0.4-0.5. The cultures were cooled to 18 °C for 15 min, then induced with IPTG (0.5 mM, final concentration) and allowed to shake for overnight at 200 rpm and 18 °C. Cells were then harvested by centrifugation for 10 min at 7000 rpm and 4 °C, and the resulting cell pellet was stored at −80 °C until purification.

Frozen cell pellets were resuspended in 30 mL Buffer A (20 mM sodium phosphate, pH 7.4, 300 mM sodium chloride, 10 mM imidazole) and one EDTA-free protease inhibitor cocktail tablet (Millipore Sigma, cat # 5056489001) was added. The cells were lysed by six sonication cycles (30 sec on, 60 sec off). The resulting lysate was centrifuged for 25 min at 20000 rpm and 4 °C to separate the insoluble cell debris from soluble proteins.

The supernatant was then loaded onto a preequilibrated His Trap FF prepacked column (with 5 mL bed volume) with 1.5 mL/min flow rate, 5 column volumes of buffer A and 5 column volumes of 92% Buffer A and 8% Buffer B (20 mM sodium phosphate, pH 7.4, 300 mM sodium chloride, 500 mM imidazole). The target protein was eluted with a linear gradient of 10-100% Buffer B (50-500 mM imidazole). Protein fractions containing the target protein were pooled, concentrated by Amicon centrifugal concentrators (10 KDa MWCO, cat # UFC901024), and further purified by size-exclusion chromatography. Concentrated protein (approximately 2 mg/mL) was loaded onto a HiLoad 16/60 Sephadex prepacked column (120 mL bed volume) with a 1 mL/min flow rate, and the target protein were eluted with storage buffer (150 mM Tris HCl, pH 8.0, 25 mM MgSO₄, 100 mM potassium chloride) with a 0.5 mL/min flow rate. Protein purity was determined by denaturing 4-20% SDS-PAGE (expected MW 30.9 KDa). Protein concentration was determined using a NanoDrop spectrophotometer (extinction coefficient 25444 cm⁻¹ M⁻¹). Concentrated protein solutions were supplemented with glycerol (20%) for cryoprotection, aliquoted into 20 μL fractions, flash frozen in liquid nitrogen, and stored at −80 °C until further use. The expression plasmid is available at Addgene.org (plasmid ID 193883).

Fluorescence polarization assay.

All reagents were kept on ice until dispensed. DHFR enzyme was expressed and purified as above. DHFR enzyme was dispensed at 4 μL/well in 1536-well black, solid-bottom microplates (Greiner Bio-One, cat # 789176-F) via BioRaptr FRD (Beckman Coulter) in 1X PBS assay buffer containing 0.05% IGEPAL (v/v) and 5 μM NADPH at a final concentration of 20 nM enzyme. Enzyme-free buffer was plated as above in column 1 of each microplate as complete inhibition control. Methotrexate control compound (Sigma Aldrich, cat # M9929) and test compounds were then transferred to respective wells of each microplate at 23 nL/well via Wako pin-tool in 11-point, 1:3 titrations for a final compound concentration range of 0.65 nM to 38.3 μM. Final DMSO concentration was 0.4%. Microplates were incubated for 20 min at ambient temperature. Next, 2 μL of Cy5-labeled methotrexate in assay buffer was dispensed into each microplate well via BioRaptr FRD for a final assay concentration of 20 nM probe. Microplates were then centrifuged at 2000 RPM for 1 min, and equilibrium between enzyme and ligand was obtained over a 30 min incubation at ambient temperature. After 30 min, fluorescence polarization was then measured on the Spark multimode microplate reader (Tecan) with the following optical settings: monochromator excitation wavelength 645 nm (10 nm bandwidth filter), emission wavelength 670 nm (10 nm bandwidth filter), 30 flashes/well and manual settings of G-factor 1.0, Z-position 20000, optimal gain, and automatic mirror. Data were normalized to no enzyme control as −100% activity and enzyme + DMSO vehicle control wells as neutral control. K_D values were constrained based on the amount of enzyme and ligand required for detection in the system and were approximately four times higher than the values reported for sensitive enzyme kinetic assays.⁶² Using MTX and 3 as control compounds, the minimum significance ratio (“MSR”) = 4.9.⁶³ See also accompanying Table S2.^64,65

DHFR HiBiT assay.

The human DHFR-HiBit fusion protein was constructed to have the HiBiT tag sequence (VSGWRLFKKIS) located after the C-terminus of DHFR.³⁷ The human DHFR-HiBiT construct was synthetized (Bio Basic) and cloned into the retroviral vector pBMN-Ires-Lyt-2 (provided by G. Nolan, Stanford University, Stanford, CA) which expresses the coding region of mouse CD8a (Lyt-2) using the restriction sites BamH1 / Not1. All constructs were sequenced to validate authenticity (Eurofins Genomics). Infected cells were positively selected based on murine CD8a expression using magnetic beads (Miltenyi Biotech). The plasmid is available at Addgene.org (plasmid ID 179731).

Transfections were performed using Lipofectamine 2000 (Invitrogen, cat # 11668) according to manufacturer protocol. For retroviral infections, viruses were produced in HEK293T cells and used to infect the HBL-1 cell line. The retroviral construct delivering DHFR-HiBiT was co-transfected into 293T cells with the mutant ecotropic envelope-expressing plasmid pHIT/EA6x3* and the MLV gag-pol expression plasmid pHIT60n (Louis Staudt, NCI, NIH, USA) as previously reported.⁶¹ Supernatants containing the retrovirus were collected and filtered at 48 and 72 h post-transfection. HBL-1 cells were centrifuged and then resuspended with the retroviral supernatant with 8 μg/mL of polybrene. The cells were spin-infected twice on consecutive days at 2500 rpm at ambient temperature for 90 min.

For the assay, 4 μL of cells (cell density 500 cells/μL; 2000 cells total per well) were plated into columns 1-48 of white 1536-well microplates (Greiner, cat # 789173-F) via Multidrop dispenser (Thermo Fisher) and small-volume metal tip dispensing cassette (Thermo Fisher, cat # 24073295). Test compounds (columns 5-48) or DMSO controls (columns 1-4) were then transferred to respective wells of each microplate at 23 nL/well via Wako pin-tool in 16-point, 1:2 titrations for a final compound concentration range of 2 nM to 57.5 μM. The final DMSO concentration was 0.58%. Cells were incubated for 5, 24, 48, or 72 h at 30 °C with 5% CO₂ and 95% humidity. The Nano-Glo HiBiT Lytic Reagent (Promega, cat # N3030) containing furimazine substrate and LgBiT was then used to quantify HiBiT-labeled DHFR according to manufacturer instructions. To microplate wells containing compound-treated cells, 3 μL of LgBiT reagent was added via BioRAPTR FRD. Microplates were then incubated under reduced lighting at ambient temperature for 10 min. Luminescence was then measured with a ViewLux 1430 Ultra HTS (PerkinElmer) with the following optical settings: exposure = 50 sec, gain = high, speed = slow, binning = 2X. Data were normalized to DMSO-only control wells. See also Table S3.^64,65

Western blotting.

For suspension cells, 5 x10⁵ cells in 500 μL media were seeded in 12-well microplates. Samples were topped with 500 μL media for a final volume of 1.0 mL. For adherent cells, 5 x10⁵ cells in 1.0 mL were seeded in 12-well microplates. Test compounds or solvent control (DMSO) were then manually added to cells. The final DMSO concentration was 0.5%. Cells were then incubated with compounds at 37 °C, 5% CO₂, and 95% relative humidity.

At the indicated time points, cells were harvested and homogenized in lysis buffer (Cell Signaling Technology, cat # 9803) with freshly added protease/phosphatase inhibitors (Cell Signaling Technology, cat # 5872). Cell debris was removed by centrifugation. Approximately 100 μg of total protein was used per sample for analysis. Proteins were sized on NuPAGE 12% Bis-Tris Gel (Thermo Fisher, cat # NP0341) or SurePAGE 4-12% Bis-Tris (Genescript, cat # M00652) precast gels. Proteins were then transferred to 0.2 μm nitrocellulose membranes (Novex, cat # LC2000) for 60 min followed by blocking with 5% milk in Tris-buffered saline-Tween 20 (TBST) for 1 h at ambient temperature with gentle shaking. Membranes were first incubated with one of the following primary antibodies for 16 h at 4° C with shaking: monoclonal anti-human DHFR antibody (1:2000 dilution, Abcam, EPR5285, cat # ab124814, host rabbit), monoclonal anti-human TS antibody (1:1000 dilution, Abcam, EPR4545, cat # ab108995, host rabbit), monoclonal anti-ubiquitin antibody (1:1000 dilution, Cell Signaling Technology, P4D1, cat # 3936, host mouse), polyclonal anti-PARP antibody (1:1000 dilution, Cell Signaling Technology, cat # 9542, host rabbit), monoclonal anti-β-actin antibody (1:1000 dilution, Cell Signaling Technology, 8H10D10, cat # 3700, host mouse), or monoclonal anti-CUL-2 antibody (1:1000 dilution, Santa Cruz Biotechnology, C-4, cat # sc-266506, host mouse).

Membranes were then washed three times with TBST at ambient temperature and incubated with 1:10000 horseradish peroxidase secondary antibody (goat anti-mouse IgG, Promega, cat # W402B; or goat anti-rabbit IgG, Invitrogen, cat # A16096) for 1 h at ambient temperature, then washed three times with TBST and developed with SuperSignal West Dura reagent (Thermo Fisher, cat # 34076) for 5 min. Immunoreactive bands were then imaged on a ChemiDoc Imaging System (Bio-Rad) with 20 s chemiluminescence exposure. The ECL Rainbow Marker (Amersham, cat # RPN800E) was used to estimate protein molecular weights. To also detect HiBiT-tagged proteins, membranes were rinsed for 1 h in TBST and then incubated in LgBiT/buffer solution overnight. The membrane was incubated with Nano-Glo Luciferase Assay Substrate furimazine (Promega, cat # N113) and incubated for 5 min, and the blot luminescence was then measured by the imaging system.

CellTiter-Glo viability and Caspase-Glo 3/7 assays.

CellTiter-Glo Luminescent Cell Viability Assay (“CTG”; Promega, cat # G7572) was used to quantify cellular ATP, and Caspase-Glo 3/7 Assay System (“Caspase-Glo” Promega, cat # G8090) was used to quantify caspase-3/7 activation using a DEVD-aminoluciferin substrate according to manufacturer protocols. Briefly, 4 μL of cells (cell density 500 cells/μL; 2000 cells total per well) were plated into white 1536-well microplates (Greiner, cat # 789173-F) in two intra-plate technical replicates. Cells were transferred to microplates via Multidrop liquid dispenser (Thermo Fisher) and small-volume metal tip dispensing cassette (Thermo Fisher, cat # 24073295). Test compounds or DMSO controls were then transferred to respective wells of each microplate at 23 nL/well via Wako pin-tool in 16-point, 1:2 titrations for a final compound concentration range of 2 nM to 57.5 μM. The final DMSO concentration was 0.58%. After incubating cells for 72 h, 2 μL of filtered CTG or Caspase-Glo reagent was added via BioRAPTR FRD to microplate wells. Microplates were then incubated under reduced lighting at ambient temperature for 10 min. Luminescence was quantified using a ViewLux 1430 Ultra HTS (PerkinElmer) with the following optical settings: exposure = 1 sec, gain = high, speed = slow, binning = 2X. Data were normalized to DMSO-only control wells. Data are from at least three independent experiments performed with two intra-plate technical replicates. See also Table S4.^64,65

Live-cell imaging assays.

Approximately 3,000 HEK293 cells were dispensed into 96-well tissue culture-treated black clear-bottom polystyrene microplates (PerkinElmer, cat # 6005225) in 100 μL media per well (DMEM; supplemented with 10% FBS, 1 mM HEPES, 100 U mL⁻¹ penicillin, 100 μg mL⁻¹ streptomycin) via multichannel pipette. Seeded microplates were then incubated for 24 h at 37 °C, 5% CO₂, and 95% relative humidity. The following day, 100 μL of media was added to each well by multichannel pipette. Cells were then immediately treated with 1.0 μL of compound DMSO stock solutions or vehicle controls dispensed by multichannel pipette. The final DMSO concentration was 0.5% (v/v). Following compound addition, cells were incubated at 37 °C, 5% CO₂, and 95% relative humidity for 120 h and imaged every 4 h with an Incucyte SX5 Live-Cell Analysis System (Essen Biosciences) utilizing a 10X objective. Live-cell images were processed in Incucyte Analysis Software (Essen Biosciences) using top-hat background correction.

Oncopanel cytotoxicity assay.

Cells were grown in RPMI 1640, 10% FBS, 2 mM L-alanyl-L-glutamine, 1 mM pyruvate, or a special medium. Cells were seeded into 384-well plates and incubated in a humidified atmosphere of 5% CO₂ at 37 °C. Compounds were tested in 10-point, 1:3 titrations for a final compound concentration range of 1.5 nM to 30 μM. The final DMSO concentration was 0.3%. Compounds were added the day following cell seeding. At the same time, a time zero untreated cell plate was generated. After 5 d incubation, cells were fixed and stained with fluorescently-labeled antibodies and nuclear dye to allow imaging of nuclei, apoptotic cells, and mitotic cells. Automated fluorescence microscopy was carried out using an ImageXpress Micro XL high-content imager (Molecular Devices), and images were collected with a 4X objective. 16-bit TIFF images were acquired and analyzed with MetaXpress 5.1.0.41 software. Cell proliferation was measured by the fluorescence intensity of an incorporated nuclear dye. An antibody to activated caspase-3 was used to label cells from early to late-stage apoptosis. The concentration of the test compound that caused a 5-fold induction in the caspase-3 signal is reported, indicating a significant apoptosis induction. Curve-fitting, calculations, and report generation were performed using a custom data reduction engine and MathIQ based software (AIM).

Mechanistic experiments.

Cell culture and compound treatments were performed as described above with the following modifications:

• Proteosome dependence. HBL-1 cells were seeded to 12-well microplates at 5 x10⁵ cells in 1.0 mL media. Proteosome inhibitor MG-132 (1 μM, final concentration) followed by test compounds (1 μM, final concentration) or solvent control (DMSO) were then manually added to cells. The final DMSO concentration was 0.5%. Cells were then incubated with compounds at 37 °C, 5% CO₂, and 95% relative humidity for 5 or 24 h. Cells were harvested, and their lysates analyzed by western blotting as described above.

• Neddylation dependence. Neddylation dependence was adopted from a published procedure.⁶⁶ Briefly, HBL-1 cells were seeded to 12-well microplates at 5 x10⁵ cells in 1.0 mL media. Neddylation inhibitor MLN4924 (1 μM, final concentration) was manually added to cells, and the cells were incubated at 37 °C, 5% CO₂, and 95% relative humidity for 1 h. Test compounds (1 μM, final concentration) or solvent control (DMSO) were then manually added to cells, followed by incubation at 37 °C, 5% CO₂, and 95% relative humidity for 24 h. The final DMSO concentration was 0.5%. Cells were harvested and their lysates analyzed by western blotting as described above.

• E3 ligase and POI competition dependence. The dependence of E3L and DHFR was assessed by a series of competition-based experiments. Briefly, HBL-1 cells were seeded to 12-well microplates at 5 x10⁵ cells in 1.0 mL media. Competitive compounds MTX (1), VSTX-IE3L (7), or pomalidomide (1 μM, final concentration) followed by test compounds (1 μM, final concentration) or solvent control (DMSO) were then manually added to cells. The final DMSO concentration was 0.5%. Cells were then incubated with compounds at 37 °C, 5% CO₂, and 95% relative humidity for 5 or 24 h. Cells were harvested, and their lysates analyzed by western blotting as described above.

• Folate cytotoxicity dependence. HBL-1 cells were cultured in either (1) folic acid-replete RPMI (Gibco, cat # 11875093) supplemented with 10% dialyzed FBS (v/v; Gibco, cat # 26400044), 100 U mL⁻¹ penicillin, 100 μg mL⁻¹ streptomycin, and 1 mM HEPES, or (2) folate acid-free RPMI (Gibco, cat # 27016021) supplemented with 10% dialyzed FBS (v/v), 100 U mL⁻¹ penicillin, 100 μg mL⁻¹ streptomycin, and 1 mM HEPES for 3 d at 37 °C, 5% CO₂, and 95% relative humidity in T75 flasks. Cells were pelleted by gentle centrifugation. Folic acid-replete cells were resuspended in the same folic acid-replete complete media, while folic acid-starved cells were resuspended in either (1) folic acid-free complete media as before, or (2) folic acid-replete complete media as a rescue. Next, 4 μL of cells (cell density 500 cells/μL; 2000 cells total per well) were plated onto white 1536-well microplates (Greiner, cat # 789173-F) via Multidrop liquid dispenser (Thermo Fisher) and small-volume metal tip dispensing cassette (Thermo Fisher, cat # 24073295). Test compounds or DMSO controls were then transferred to respective wells of each microplate at 23 nL/well via Wako pin-tool in 16-point, 1:2 titrations for a final compound concentration range of 2 nM to 57.5 μM. The final DMSO concentration was 0.58%. Compound-treated cells were then incubated at 37 °C, 5% CO₂, and 95% relative humidity. After incubating cells for 24, 48, 72, or 96 h post-compound treatment, cellular viability was quantified using the CTG procedure. Data were normalized to DMSO-only and digitonin (115 μM) negative and positive control wells, respectively. Data are from four intra-plate technical replicates.

qRT-pCR.

Approximately 5 x10⁵ cells in 500 μL media were seeded in 12-well microplates, followed by the addition of 500 μL media for a total volume of 1.0 mL. Test compounds or solvent control (DMSO) were then manually added to cells. The final DMSO concentration was 0.5%. Cells were then incubated with compounds at 37 °C, 5% CO₂, and 95% relative humidity. At the indicated time points, total RNA was then isolated from HBL-1 cells using RNeasy Mini Kit (Qiagen, cat # 74106) as per manufacturer protocol. 250 ng of RNA was used for a reverse transcriptase reaction and subsequent real-time quantitative PCR on the Applied Biosystems VIIA7 detection system using the Taqman RNA-to-Ct 1-step kit (Thermo Fisher, cat # 4392938). Validated Taqman probes for DHFR (Hs00758822_s1) and USF2 (Hs01100995_g1) were obtained from Applied Biosystems. Ct values for DHFR were normalized to USF2 and subsequently normalized to untreated (solvent-control) conditions.

Combination screening.

Compounds and solvent controls were transferred to white/solid-bottom tissue-culture treated 1536-well microplates (Aurora E8, improved resin grey color) by Echo 550 acoustic dispenser (Labcyte) at 20 nL/well in 10-point, 1:2 titrations for a final compound concentration range of 4 to 1000 nM. Test compounds and combination compounds were dispensed in a 10x10-well array format. Assay plates were sealed in plastic wrap and frozen until ready for use. Next, 5 μL of cells (cell density 200 cells/μL; 1000 cells total per well) in media were added to these microplates via Multidrop liquid dispenser (Thermo Fisher) and small-volume metal tip dispensing cassette (Thermo Fisher, cat # 24073295). Microplates were then sealed using stainless steel gasketed lids and the cells were incubated for 72 h at 37 °C, 5 % CO₂, and 95% relative humidity. CellTiter-Glo Luminescent Cell Viability Assay (“CTG”; Promega, cat # G7572) was then used to quantify cellular ATP according to manufacturer protocol. After incubation, 3 μL of filtered CTG reagent was added via BioRAPTR FRD to microplate wells. Microplates were then incubated under reduced lighting at ambient temperature for 15 min. Luminescence was quantified using a ViewLux 1430 Ultra HTS (PerkinElmer) with the following optical settings: exposure = 1 sec, gain = high, speed = slow, binning = 4X. Data were normalized to DMSO-only control wells. Data are from three inter-plate technical replicates.

Data analysis and figure preparation.

For data analysis, the compound signal was normalized to DMSO control for 100% viability and no-cell control for 0% viability. Drug combination synergy metrics were calculated for each matrix block individually and in triplicate using the ncgcmatrix R package (github.com/ncats/ncatsr/tree/master/ncgcmatrix).

Proteomics analysis.

HBL-1 cells were seeded in T25 culture flasks at 5 million cells in 10 mL RPMI media (cell density 500,000 cells/mL) supplemented with 10% FBS (v/v, HyClone, cat # SH30071.03) and 1% penicillin/streptomycin antibiotic (v/v, Gibco, cat # 10378-016). Test compounds were manually added to cells in triplicate at a final concentration of 1 μM. The final DMSO concentration was 0.01%. Cells were then incubated with compounds at 37 °C, 5% CO₂, and 95% relative humidity for 24 h. Cells were collected on ice then centrifuged at 900 RPM for 5 min at 4 °C. The media was subsequently removed, and cells were then washed three times with 1X PBS on ice. Cell pellets were frozen at −80 °C until lysis. Cell pellets thawed on ice and NP-40 lysis buffer (ThermoFisher, cat # FNN0021) with 1X mammalian cell protease inhibitor cocktail (Sigma Aldrich, cat # P8340) was added to pellets at 5X volume. Cells were resuspended in lysis buffer by pipetting up and down ten times, then incubated with inverted rocking for 1 h at 4 °C. Samples were centrifuged at 13,000 RPM at 4 °C for 15 min. Supernatants were transferred to new Eppendorf tubes and the protein concentration of respective lysates were determined with BCA (ThermoFisher, cat # 23252).

Fifteen samples representing five groups (DMSO, MTX (1), FMTX (2), 3, and Versortrexate-ME (9)) with three biological replicates were prepared for proteomic digestion by trypsin prior to HPLC-MS analysis.⁵¹ Around 30 μg of protein estimated from the BCA results were added RapidGest (0.1% w/v) and TCEP (20 mM) followed by reduction incubation (55 °C, 850 rpm, 45 min). After reduction incubation, the samples were centrifuged and iodoacetamide was then added to each sample (15 mM) for incubation (30 min, under dark environment). Chilled acetonitrile (−20 °C) was then added to each sample in a 9:1 v/v ratio. The samples were then immediately incubated on ice then centrifuged (4 °C, 18,000 g, 10 min). As much acetonitrile as possible was then removed without disturbing the precipitated protein pellet. The samples were then evaporated on the Centrivap Complete Vacuum Concentrator for 10 min. A volume of 30 μL of 0.02 mg/mL trypsin was then added to each sample then transferred to the Thermal Mixer for overnight incubation (37 °C, 850 rpm, 15-18 h). Formic acid was added to each sample to quench the reaction prior to HPLC-MS/MS analysis.

All proteomic HPLC-MS/MS analysis was performed using an UltiMate 3000-nano LC system coupled to the Orbitrap Fusion Lumos Tribrid mass spectrometer equipped with the Nanospray Flex ion source (Thermo Fisher). Peptides were loaded onto the trap column (Acclaim PepMap 100 C18, 75 μm x 2 cm, particle size: 3 μm, 100 Å) and separated with an analytical column with the spray tip (75 μm x 30 cm, 1.7 μm, 100 Å; CoAnn Technologies) using a 200 min method (~180 min gradient). Peptides were loaded onto the trap column by autosampler using loading solvent (2% acetonitrile in 98% UHPLC-grade water) at a flow rate of 4 μL/min. Elution of peptides from the analytical column was performed using a 180 min gradient (including sample loading and re-equilibration) starting at 98% A (0.1% formic acid in UHPLC-grade water) at a flow rate of 300 nL/min. The mobile phase was maintained at 2% B (80% acetonitrile, 19.9% water, 0.1% formic acid) for 5 min, 2-9% B for 4 min, 9-38% B for 141 min, 38-50% B for 25 min, 50-90% B for 3 min, and maintained at 90% B for 10 min, followed by re-equilibration of the column with 2% B for 10 min. Column oven parameters were set as follows: temperature, 40 °C.

The mass spectrometer was operated in positive-ionization mode with the Nanospray Flex ion source with spray voltage set at 1800 V, and ion transfer tube temperature set at 250 °C. The MS scan was operated at data-dependent acquisition mode, with full MS scans over a mass range of m/z 375-1,800 with detection in the Orbitrap (120 K resolution) and with auto gain control (AGC) set to 1.0 x 10⁶. The fragment ion spectra were acquired in Orbitrap (15 K resolution) with a normalized collision energy of 28% at HCD activation mode. In each cycle of data-dependent acquisition analysis, the most intense ions above were selected for the MS/MS analysis, and the cycle time for MS and MS/MS analysis was set as 2 s. The AGC for MS/MS was set as Standard and a maximum injection time was 22 ms. Precursor ions with charges of +2 to +7 were isolated for MS/MS sequencing. The MS/MS isolation window was 1.2 Da, and the dynamic exclusion time was set at 60 s (after one MS/MS acquisition) with a mass tolerance of ± 10 ppm.

Proteome Discoverer software suite (version 2.4, Thermo Fisher) with Sequest algorithm were used for peptide identification and quantitation. The MS raw data were searched against a Swiss-Prot human database (version Jan 2019, reviewed database) consisting of 20,350 entries using the following parameters: precursor ion mass tolerance of 10 ppm and a fragment ion mass tolerance of 0.02 Da. Peptides were searched using fully tryptic cleavage constraints and up to two internal cleavage sites were allowed for tryptic digestion. Fixed modifications consisted of carbamidomethylation of cysteine. Variable modifications considered were oxidation of methionine residues and N-terminal protein acetylation. Peptide identification false discovery rates (FDR) were limited to a maximum of 0.01 using identifications from a concatenated database from the non-decoy and the decoy databases. Label-free quantification analysis used the “Precursor Ions Quantifier” node from Proteome Discoverer and normalized by total peptide amount.

For each of the four drugs, pathway enrichment for significant (padj < 0.05) proteins was determined using Fisher’s exact test (R, version 4.1.3). Pathway annotations from KEGG (305 annotations), Reactome Pathway Database (1,389 annotations), and WikiPathway (334) were tested.

Data analyses and figure preparation.

All graphical data are expressed as individual replicates or mean ± standard deviation (SD) unless stated otherwise. Graphing and statistical analyses were performed using GraphPad Prism (version 9.0.0). Final figures were prepared in Adobe Illustrator (version 25.0.1).

Data availability.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (proteomecentral.proteomexchange.org)⁶⁰ via the PRIDE partner repository with the dataset identifier PXD030140 and DOI 10.6019/PXD030140. All other relevant data are available from the authors without restriction.

Targeted metabolomics.

HBL-1 cells were cultured and expanded in standard growth media (1640 medium (Gibco Ref #11875-093) + 1mM HEPES, 100 U/mL penicillin, 100 μg/mL streptomycin, 10% FBS (v/v; Biowest, cat # S1620)) at 37°C, 5% CO₂. Cells were split into five T75 flasks per treatment (four for metabolomics, one for protein quantification for normalization) in folate-replete media for 24 hours pre-treatment. On the day of treatment, cells were centrifuged and moved into new folate-replete (1640 medium (Gibco Ref # 11875-093), 1mM HEPES, 100 U/mL penicillin, 100 μg/mL streptomycin, 10% dialyzed FBS (v/v; Biowest, cat # S162D)) and treated for 24 hours with the following, MTX (1) 0.1 or 0.5 μM, VSTX-ME (9) 0.1 or 0.5 μM, or equivalent volumes of DMSO vehicle control.

Each flask was collected and centrifuged at 4000 rpm for 10 minutes at 4°C. 1.5 ml supernatant was collected and stored at −80°C. The remaining supernatant was discarded, and the pellet was resuspended in 500 μL ice cold 80% methanol (kept on dry ice throughout the procedure). The resuspended cells were moved to a microcentrifuge tube and frozen in liquid nitrogen (LN2) for 20 mins. The tubes were removed and vortexed for one minute each. The sample was then centrifuged at 5000 rpm for 10 minutes at 4°C. The supernatant was collected into a new tube labeled ‘for metabolomics, 80% methanol.’ This process was repeated three times (1.5 ml methanol total). The samples were then dried using the speed vacuum and stored at −80°C. The isolated cells were collected in 2 ml tubes using centrifuge at 4000 rpm, for 10 min at 4°C. Intracellular metabolites were then extracted from the cells using 1.5 ml cryogenically cold 80% methanol containing internal standard mixture (choline-d9). the pellet was resuspended in 500 μL ice cold 80% methanol (kept on dry ice throughout the procedure). Resuspended cells were moved to a microcentrifuge tube and frozen in liquid nitrogen (LN2) for 20 mins. The tubes were removed and vortexed for one minute each. The sample was then centrifuged at 5000 rpm for 10 minutes at 4°C. The supernatant was collected into a new tube labeled ‘for metabolomics, 80% methanol.’ This process was repeated three times (1.5 mL methanol total). The samples were then dried using the speed vacuum and stored at −80°C and resuspended in 50 μL methanol/water (50/50) for polar metabolomics analysis. The reconstituted sample mixture was then centrifuged at 15,000 rpm for 10 min at 4 °C. The supernatant was then transferred to an LC–MS vial and ready for LC-MS/MS analysis.

The measurements of targeted polar metabolites of one-carbon metabolism were performed using Waters Acquity UPLC coupled to a Xevo TQ-Abs mass spectrometer with the positive and negative switching mode and scheduled selected reaction monitoring (MRM) methods.⁶⁷ Briefly, the metabolites were separated using a Waters UPLC BEH amide column (2.1 x 150 mm, 1.7μM) (PN: 186004801, SN: 01953221418650). The gradient buffer that contains Buffer A (100% Acetonitrile) and Buffer B (5% Acetonitrile, 95% water, 20 mM Ammonium acetate, pH 9.0) was used to elute the metabolites from the column. The gradient was as follows. 0 min, 85% A; 3 min, 84% A; 7 min, 65% A; 12 min, 60% A; 15 min, 55% A; 17 min, 50% A; 19 min, 50% A; 22 min, 70 % A; 23 min, 85% A; 30 min, 85% A.

The metabolite areas were integrated using Skyline (MacCoss Lab Software) and normalized to the respective cell protein concentration. Further analysis of the relative quantification of the resultant peak areas was performed using Metaboanalyst 5.0 software (www.metaboanalyst.ca).

Synthetic Chemistry Methods

Reagents and Methods:

All air or moisture-sensitive reactions were performed under a positive pressure of nitrogen or argon with oven-dried glassware. Anhydrous solvents and bases such as dichloromethane, N, N-dimethylformamide (DMF), acetonitrile, ethanol, DMSO, dioxane, and DMF were purchased from Sigma-Aldrich and used as such. Preparative purification was performed on a Waters semi-preparative HPLC system using a Phenomenex Luna C18 column (5-micron, 30 x 75 mm) at a flow rate of 45 mL/min. The mobile phase consisted of acetonitrile and water (each containing 0.1% trifluoroacetic acid). A gradient of 10% to 50% acetonitrile over 8 minutes was used during the purification. Fraction collection was triggered by UV detection (220 nm). Analytical analysis was performed on an Agilent LC/MS (Agilent Technologies, Santa Clara, CA).

Method 1: A 3-minute gradient of 4% to 100% Acetonitrile (containing 0.025% trifluoroacetic acid) in water (containing 0.05% trifluoroacetic acid) was used with a 4.5-minute run time at a flow rate of 1 mL/min. A Phenomenex Gemini Phenyl column (3-micron, 3 x 100 mm) was used at a temperature of 50°C. Purity determination was performed using an Agilent Diode Array Detector for both Method 1 and Method 2

Method 2: A 7-minute gradient of 4% to 100% Acetonitrile (containing 0.025% trifluoroacetic acid) in water (containing 0.05% trifluoroacetic acid) was used with an 8-minute run time at a flow rate of 1 mL/min. A Phenomenex Luna C18 column (3-micron, 3 x 75 mm) was used at a temperature of 50°C. Mass determination was performed using an Agilent 6130 mass spectrometer with electrospray ionization in the positive mode. ¹H NMR spectra were recorded on Varian 400 MHz spectrometers. Chemical shifts are reported in ppm (DMSO-d₆ at 2.49 ppm, CD₃OD at 3.31 ppm, and CDCl₃ at 7.26 ppm). All the analogs tested in the biological assays have purity greater than 95%, based on both analytical methods. High-resolution mass spectrometry was recorded on Agilent 6210 Time-of-Flight LC/MS system. Confirmation of molecular formula was accomplished using electrospray ionization in the positive mode with the Agilent Masshunter software (version B.02).

General methods of syntheses for the key intermediates and final compounds.

(2S)-2-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-4-fluoropentanedioic acid (2):

Compound 10 (50 mg, 0.154 mmol, 1.0 equiv.) was dissolved in DMF (1 mL) in a round-bottom flask, followed by the addition of EDC (35.4 mg, 0.184 mmol, 1.2 equiv.), and 2,3,4,5,6-pentafluorophenol (33.9 mg, 0.184 mmol, 1.2 equiv.). The solution was stirred at ambient temperature for 2 h. The crude product was washed with brine and extracted with ethyl acetate. The organic layer was evaporated and the crude perfluorophenyl 4-(((2,4- diaminopteridin-6-yl)methyl)(methyl)amino)benzoate was dissolved in DMF (1 mL) followed by the addition of 2-amino-4-fluoropentanedioic acid (30.5 mg, 0.184 mmol, 1.2 equiv.) and DMAP (37 mg, 0.30 mmol, 2.0 equiv.). The reaction was stirred overnight, and the crude product was filtered and then purified by reverse-phase HPLC to yield fluoromethotrexate 2 (mixture of diastereoisomers) as a yellow salt. ¹H NMR (400 MHz, Methanol-d₄) δ 8.64 (s, 1H), 7.74 (d, J = 8.6 Hz, 2H), 6.85 (d, J = 8.6 Hz, 2H), 5.20 (dd, J = 7.6, 4.3 Hz, 1H), 5.08 (dd, J = 7.6, 4.3 Hz, 1H), 4.79 (dd, J = 8.0, 5.6 Hz, 2H), 3.27 (s, 3H), 2.64 (ddt, J = 25.6, 14.9, 5.0 Hz, 1H), 2.48 – 2.35 (m, 1H). ¹³C NMR (101 MHz, Methanol-d₄) δ173.15, 168.60, 152.36, 151.51, 148.96, 128.89, 121.44, 111.28, 87.16, 85.34, 55.20, 49.08, 49.05, 38.32, 33.93, 33.72. ¹⁹F NMR (376 MHz, Methanol-d₄) δ −77.12. HPLC t_R = 2.72 min (Method 2). ESI-LRMS [M + H]⁺ = 473.2 (found). ESI-HRMS [M + H]⁺ = 473.1602 (calcd C₂₀H₂₁FN₈O₅ 472.1619).

(19S)-19-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-1-((2-(2,6-dioxo piperidin-3-yl)-1,3-dioxoisoindolin-4-yl)oxy)-2,16-dioxo-6,9,12-trioxa-3,15-diazaicosan-20-oic acid (3):

Commercially available 11 (300 mg, 0.903 mmol, 1.0 equiv.) and tert-butyl (2-(2-(2-(2-aminoethoxy)ethoxy)ethoxy)ethyl) carbamate (290 mg, 0.993 mmol, 1.1 equiv.) were dissolved in acetonitrile (10 mL) in a round-bottom flask, followed by the addition of TCFH (381 mg, 1.354 mmol. 1.5 equiv.) and 1-methyl-1H-imidazole (222 mg, 2.71 mmol, 3.0 equiv.). The reaction was stirred for 2 h. The solvent was evaporated under reduced pressure, then the dried product was redissolved in ethyl acetate and washed with brine. The organic layer was dried over MgSO₄ and the solvent was removed under reduced pressure, and then purified by silica gel column chromatography (10 % methanol in ethyl acetate) to yield 12 (500 mg, 91% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 11.10 (s, 1H), 7.98 (t, J = 5.6 Hz, 1H), 7.79 (t, J = 7.9 Hz, 1H), 7.64 (d, J = 11.0 Hz, 1H), 7.48 (d, J = 7.2 Hz, 1H), 7.38 (d, J = 8.5 Hz, 1H), 6.71 (t, J = 5.8 Hz, 1H), 5.10 (dd, J = 12.9, 5.3 Hz, 1H), 4.77 (s, 2H), 3.84 (s, 1H), 3.44 (d, J = 5.4 Hz, 1H), 3.32 (dt, J = 13.3, 5.9 Hz, 5H), 3.03 (q, J = 6.0 Hz, 2H), 2.95 – 2.81 (m, 1H), 2.64 – 2.44 (m, 3H), 2.03 (ddq, J = 8.0, 5.6, 3.1, 2.3 Hz, 1H), 1.34 (s, 9H). ¹³C NMR (101 MHz, DMSO-d₆) δ 173.23, 170.31, 167.36, 167.18, 165.90, 156.04, 155.41, 137.40, 136.23, 133.48, 123.57, 120.79, 120.21, 117.20, 116.51, 78.06, 70.18, 70.06, 69.93, 69.60, 69.25, 67.95, 49.26, 40.54, 40.34, 40.13, 39.92, 39.71, 39.50, 39.29, 38.86, 38.68, 35.84, 31.39, 28.66, 22.44. HPLC t_R = 3.00 min (Method 1). ESI-LRMS [M + H]⁺ = 607.2.

Compound 12 was dissolved in DCM (5 mL) in a round-bottom flask, followed by the addition of TFA (2 mL). The reaction was stirred at ambient temperature for 3 h. The solvent was removed under reduced pressure to yield crude 13 (470 mg, 0.757 mmol, 92% yield) which was used as such in the next step. ¹H NMR (400 MHz, DMSO-d₆) δ 11.10 (s, 1H), 9.01 (s, 1H), 8.00 (t, J = 5.6 Hz, 1H), 7.78 (d, J = 7.6 Hz, 1H), 7.68 – 7.59 (m, 1H), 7.48 (d, J = 7.3 Hz, 1H), 7.38 (d, J = 8.5 Hz, 1H), 5.09 (dd, J = 12.9, 5.4 Hz, 1H), 4.77 (s, 2H), 3.84 (s, 2H), 3.61 – 3.48 (m, 12H), 3.45 (t, J = 5.8 Hz, 2H), 3.30 (q, J = 5.8 Hz, 2H), 2.95 (q, J = 5.6 Hz, 2H), 2.93 – 2.81 (m, 1H), 2.63 – 2.48 (m, 2H). HPLC t_R = 2.51 min (Method 1). ESI-LRMS [M + H]⁺ = 507.2.

In a round-bottom flask, compound 14 (500 mg, 0.806 mmol, 1.0 equiv.) was dissolved in anhydrous acetonitrile (10 mL) followed by the addition of 13 (411 mg, 0.806 mmol, 1.0 equiv.), TCFH (454 mg, 1.612 mmol, 2.0 equiv.) and 1-methyl-1H-imidazole (331 mg, 4.03 mmol, 5.0 equiv.). The reaction was stirred for 1 h and the solvent was removed under reduced pressure. The crude product was partially purified by silica gel column chromatography (20% methanol in ethyl acetate) to yield tert-butyl (19S)-19-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-1-((2-(2,6-dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)oxy)-2,16-dioxo-6,9,12-trioxa-3,15-diazaicosan-20-oate (740 mg, 0.665 mmol, 83% yield) as a red gel. HPLC t_R = 2.70 min (Method 1). ESI-LRMS [M + H]⁺ = 999.4. The aforementioned crude product was dissolved in DCM (15 mL) in a round-bottom flask and cooled to 0 °C followed by the addition of TFA (3 mL). The reaction was stirred at ambient temperature for 3 h. The solvent was removed under reduced pressure. The dried product was then redissolved in DMSO (approximately 2 mL), and then the crude product was purified by reverse-phase HPLC to yield 3 as a TFA salt. ¹H NMR (400 MHz, DMSO-d₆) δ 12.99 (s, 1H), 12.45 (s, 1H), 11.09 (s, 1H), 9.24 (s, 1H), 9.04 (s, 1H), 8.70 (s, 1H), 8.57 (s, 1H), 8.28 (d, J = 7.4 Hz, 1H), 7.98 (t, J = 5.7 Hz, 1H), 7.88 (t, J = 5.6 Hz, 1H), 7.79 (t, J = 7.9 Hz, 1H), 7.72 (d, J = 8.7 Hz, 2H), 7.48 (d, J = 7.3 Hz, 1H), 7.38 (d, J = 8.5 Hz, 1H), 6.81 (d, J = 8.8 Hz, 2H), 5.09 (dd, J = 12.9, 5.4 Hz, 1H), 4.85 (s, 2H), 4.77 (s, 2H), 4.29 – 4.24 (m, 1H), 3.50 – 3.40 (m, 9H), 3.36 – 3.27 (m, 4H), 3.23 (s, 3H), 3.15 (q, J = 6.0 Hz, 2H), 2.96 – 2.81 (m, 1H), 2.60 – 2.52 (m, 1H), 2.18 (t, J = 7.6 Hz, 2H), 2.06 – 1.99 (m, 2H), 1.94 – 1.85 (m, 1H). HPLC t_R = 2.49 min (Method 1). ESI-LRMS [M + H]⁺ = 943.3. ESI-HRMS [M + H]⁺ = 943.4574 (found), (calcd C₄₃H₅₀N₁₂O₁₃ 943.3693).

(19S)-19-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-1-((2-(1-methyl-2,6-dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)oxy)-2,16-dioxo-6,9,12-trioxa-3,15-diazaicosan-20-oic acid (4):

Commercially available 15 (500 mg, 1.823 mmol, 1.0 equiv.) was dissolved in DMF (5 mL) in a round-bottom flask, followed by the addition of K₂CO₃ (378 mg, 2.73 mmol, 1.5 equiv.) and tert-butyl 2-bromoacetate (261 μL, 2.006 mmol, 1.1 equiv.). The reaction was stirred for 12 h and then diluted with ethyl acetate, washed with brine, and dried with MgSO₄. The organic solvent was removed under reduced pressure. The dried product was then redissolved in DCM, and then the crude product was purified by silica gel column chromatography (80% ethyl acetate in hexane) to give 16 (670 mg, 95% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 11.09 (s, 1H), 7.93 (s, 1H), 7.83 – 7.74 (m, 1H), 7.47 (d, J = 7.2 Hz, 1H), 7.36 (d, J = 8.5 Hz, 1H), 5.09 (dd, J = 12.9, 5.3 Hz, 1H), 4.95 (s, 2H), 2.87 – 2.83 (m, 1H), 2.63 – 2.50 (m, 2H), 2.02 (tt, J = 7.3, 3.8 Hz, 1H), 1.41 (d, J = 1.4 Hz, 9H). ¹³C NMR (101 MHz, DMSO-d₆) δ 173.22, 170.33, 167.59, 167.17, 166.70, 165.57, 162.75, 155.50, 137.21, 133.71, 120.43, 116.91, 116.35, 82.36, 65.97, 49.26, 36.22, 31.40, 31.21, 28.12, 22.42, 14.90. HPLC t_R = 3.08 min (Method 1). ESI-LRMS [M – tert-Bu]⁺ = 333.1.

Compound 16 (670 mg, 1.725 mmol, 1.0 equiv.) was dissolved in DMF (10 mL) in a round-bottom flask, followed by the addition of K₂CO₃ (477 mg, 3.45 mmol, 2.0 equiv.) and iodomethane (537 μL, 8.63 mmol, 5.0 equiv.). The reaction was stirred at 50 °C for 24 h. The reaction was then cooled to room temperature and then diluted with ethyl acetate, washed with brine, and dried with MgSO₄. The solvent was removed under reduced pressure to yield a white solid which was dissolved in 10 mL of DCM, followed by the addition of 2 mL of TFA. The reaction mixture was stirred for 3 h, after which the solvent was removed under reduced pressure to yield 17. ¹H NMR (400 MHz, DMSO-d₆) δ 13.22 (s, 1H), 7.83 – 7.74 (m, 1H), 7.46 (d, J = 7.2 Hz, 1H), 7.38 (d, J = 8.5 Hz, 1H), 5.15 (dd, J = 13.0, 5.4 Hz, 1H), 4.97 (s, 2H), 3.00 (s, 3H), 2.93 (td, J = 13.9, 13.4, 6.9 Hz, 1H), 2.74 (dt, J = 17.3, 3.9 Hz, 1H), 2.60 – 2.49 (m, 1H), 2.10 – 2.00 (m, 1H). ¹³C NMR (101 MHz, DMSO-d₆) δ 172.23, 170.12, 169.93, 167.16, 166.76, 165.61, 155.60, 137.25, 133.69, 120.35, 116.76, 116.23, 65.45, 49.80, 40.61, 40.40, 40.19, 39.98, 39.77, 39.56, 39.35, 31.54, 27.05, 21.63, 15.04, 14.98. HPLC t_R = 2.64 min (Method 1). ESI-LRMS [M + H]⁺ = 347.2.

Compound 17 (180 mg, 0.391 mmol, 1 equiv.) was dissolved in acetonitrile (5 mL) in a round-bottom flask, followed by the addition of tert-butyl (2-(2-(2-(2-aminoethoxy)ethoxy)ethoxy)ethyl)carbamate (126 mg, 0.430 mmol, 1.1 equiv.), TCFH (165 mg, 0.587 mmol, 1.5 equiv.), and 1-methyl-1H-imidazole (96 mg, 1.173 mmol, 3.0 equiv.). The reaction was stirred for 1 h, and the solvent was then removed under reduced pressure. The crude product was dissolved in DCM (2 mL), followed by the addition of 2 mL of TFA. The reaction was then stirred for 3 h. The solvent was removed under reduced pressure. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by reverse-phase HPLC to yield 18 as a TFA salt. HPLC t_R = 3.11 min (Method 1). ESI-LRMS [M + H]⁺ = 521.2.

In a round-bottom flask, compound 14 (121 mg, 0.236 mmol, 1.0 equiv.) was dissolved in anhydrous acetonitrile (10 mL), followed by the addition 18 (150 mg, 0.236 mmol, 1.0 equiv.), TCFH (133 mg, 0.473 mmol, 2.0 equiv.) and 1-methyl-1H-imidazole (97 mg, 1.182 mmol, 5.0 equiv.). The reaction was stirred for 1 h. The solvent was removed under reduced pressure and the crude product was dissolved in DCM (5 mL), followed by the addition of 2 mL of TFA. The reaction was continually stirred for 3 h. The solvent was evaporated. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by reverse-phase HPLC to yield 4 as a TFA salt. ¹H NMR (400 MHz, DMSO-d₆) δ 13.03 (s, 1H), 12.46 (s, 1H), 9.26 (s, 1H), 9.05 (s, 1H), 8.69 (s, 1H), 8.59 (s, 1H), 8.26 (dd, J = 14.4, 7.5 Hz, 1H), 7.98 (t, J = 5.6 Hz, 1H), 7.88 (t, J = 5.6 Hz, 1H), 7.79 (dd, J = 8.5, 7.3 Hz, 1H), 7.76 – 7.67 (m, 2H), 7.56 (s, 1H), 7.48 (d, J = 7.2 Hz, 1H), 7.38 (d, J = 8.5 Hz, 1H), 6.85 – 6.73 (m, 2H), 5.16 (dd, J = 13.0, 5.4 Hz, 1H), 4.85 (s, 2H), 4.76 (s, 2H), 4.27 (ddd, J = 9.4, 7.4, 4.9 Hz, 1H), 3.48 – 3.40 (m, 6H), 3.38 – 3.20 (m, 6H), 3.15 (q, J = 3.8, 3.4 Hz, 3H), 3.00 (s, 3H), 2.99 – 2.87 (m, 1H), 2.75 (ddd, J = 17.2, 4.5, 2.6 Hz, 1H), 2.60 – 2.49 (m, 1H), 2.18 (t, J = 7.7 Hz, 2H), 2.04 (ddt, J = 13.7, 6.5, 2.7 Hz, 1H), 1.90 (ddd, J = 13.1, 8.9, 6.7 Hz, 1H). HPLC t_R = 3.58 min (Method 2). ESI-LRMS [M + H]⁺ = 957.3. ESI-HRMS [M + H]⁺ = 957.3623 (found); (calcd C₄₄H₅₂N₁₂O₁₃: 957.385).

(17S)-17-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-1-((2-(2,6-dioxo piperidin-3-yl)-1,3-dioxoisoindolin-4-yl)oxy)-2,16-dioxo-6,9,12-trioxa-3,15-diazaicosan-20-oic acid (5):

Compound 10 (100 mg, 0.307 mmol, 1.0 equiv.) was dissolved in 1 mL of DMF in a round-bottom flask, followed by the addition of TEA (64.3 μL, 0.461 mmol, 1.5 equiv.) and PyBOP (192 mg, 0.369 mmol, 1.2 equiv.). The mixture was stirred at ambient temperature for 2 h, resulting in a dark brown solution. To this solution was added (S)-2-amino-5-(tert-butoxy)-5-oxopentanoic acid (88.0 mg, 0.369 mmol, 1.2 equiv.). The reaction mixture was then stirred for 18 h and the reaction mixture was then directly purified by reverse-phase HPLC (10 – 100% CH3CN in water, 0.01% TFA) to yield 19. ¹H NMR (400 MHz, DMSO-d₆) δ 12.50 (s, 1H), 9.24 (s, 1H), 9.04 (s, 1H), 8.70 (d, J = 1.5 Hz, 1H), 8.57 (s, 1H), 8.18 (d, J = 7.8 Hz, 1H), 7.73 (d, J = 8.5 Hz, 2H), 6.80 (d, J = 8.5 Hz, 2H), 4.86 (s, 2H), 4.33 (td, J = 8.8, 4.9 Hz, 1H), 3.23 (s, 3H), 2.27 (t, J = 7.5 Hz, 2H), 2.01 (p, J = 7.3, 6.9 Hz, 1H), 1.88 (dd, J = 15.1, 7.9 Hz, 1H), 1.35 (s, 9H). HPLC t_R = 2.68 min (Method 1). ESI-LRMS [M + H]⁺ = 511.3.

Compound 19 (30 mg, 0.059 mmol, 1.0 equiv.) was dissolved in DMF (1 mL) in a round-bottom flask, followed by the addition of 13 (29.8 mg, 0.059 mmol, 1.0 equiv.), 2-(3H-[1,2,3]triazolo[4,5-b]pyridin-3-yl)-1,1,3,3-tetramethylisouronium hexafluorophosphate(V) (33.5 mg, 0.088 mmol, 1.5 equiv.), and N-ethyl-N-isopropylpropan-2-amine (51.3 μL, 0.294 mmol, 5.0 equiv.). The reaction was stirred for 3 h and the solvent was evaporated. The crude product was dissolved in DCM (2 mL) followed by the addition of 1 mL TFA. The reaction mixture was then stirred for 3 h. The solvent was then removed under reduced pressure. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by reverse-phase HPLC to yield 5 as a TFA salt. ¹H NMR (400 MHz, DMSO-d₆) δ 12.97 (s, 1H), 12.01 (s, 1H), 11.09 (s, 1H), 9.25 (s, 1H), 9.05 (s, 1H), 8.69 (s, 1H), 8.58 (s, 1H), 7.99 (q, J = 6.7, 5.5 Hz, 2H), 7.89 (q, J = 7.6, 5.7 Hz, 1H), 7.77 (dd, J = 17.8, 9.8 Hz, 2H), 7.72 (s, 1H), 7.47 (d, J = 7.3 Hz, 2H), 7.37 (d, J = 8.5 Hz, 1H), 6.79 (d, J = 8.6 Hz, 2H), 5.09 (dd, J = 12.9, 5.4 Hz, 1H), 4.85 (s, 2H), 4.77 (s, 2H), 4.36 (td, J = 8.5, 5.3 Hz, 1H), 3.46 (d, J = 7.1 Hz, 11H), 3.44 – 3.33 (m, 3H), 3.29 (q, J = 5.7 Hz, 2H), 3.22 (s, 2H), 3.27 – 3.08 (m, 2H), 2.94 – 2.81 (m, 1H), 2.62 – 2.51 (m, 2H), 2.23 (td, J = 8.9, 4.3 Hz, 2H), 2.08 – 1.89 (m, 2H), 1.83 (dq, J = 15.4, 8.3 Hz, 1H). HPLC t_R = 2.46 min (Method 1). ESI-LRMS [M + H]⁺ = 943.3. ESI-HRMS [M + H]⁺ = 943.3563 (found); (calcd C₄₃H₅₀N₁₂O₁₃: 943.3693).

N2-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzoyl)-N5-(8-((2-(2,6-dioxopiperidin −3-yl)-1,3-dioxoisoindolin-4-yl)amino)octyl)-L-glutamine (6):

This compound was synthesized with 14 and 20 using similar conditions reported for PROTAC 5. ¹H NMR (400 MHz, DMSO-d₆) δ 12.98 (s, 1H), 12.44 (s, 1H), 11.07 (s, 1H), 9.21 (s, 1H), 9.00 (s, 1H), 8.69 (s, 1H), 8.28 (d, J = 7.4 Hz, 1H), 7.78 (t, J = 5.6 Hz, 1H), 7.72 (d, J = 8.7 Hz, 2H), 7.55 (dd, J = 8.6, 7.1 Hz, 1H), 7.06 (d, J = 8.6 Hz, 1H), 7.00 (d, J = 7.0 Hz, 1H), 6.80 (d, J = 8.6 Hz, 2H), 6.49 (t, J = 6.0 Hz, 1H), 5.03 (dd, J = 12.8, 5.4 Hz, 1H), 4.85 (s, 2H), 4.27 (ddd, J = 9.4, 7.4, 4.8 Hz, 1H), 3.26 (t, J = 6.7 Hz, 2H), 3.22 (s, 3H), 2.98 (q, J = 6.5 Hz, 2H), 2.86 (ddd, J = 17.4, 14.0, 5.4 Hz, 1H), 2.62 – 2.50 (m, 2H), 2.16 (t, J = 7.5 Hz, 2H), 2.06 – 1.98 (m, 2H), 1.96 – 1.83 (m, 1H), 1.57 – 1.50 (m, 2H), 1.37 – 1.20 (m, 12H). HPLC t_R = 4.36 min (Method 2). ESI-LRMS [M + H]⁺ = 837.4. ESI-HRMS [M + H]⁺ = 837.3588 (found); (calcd C₄₁H₄₈N₁₂O₈: 837.3791).

N2-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzoyl)-N5-(8-((2-(1-methyl-2,6-dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)amino)octyl)-L-glutamine (7):

2-(2,6-dioxopiperidin-3-yl)-4-fluoroisoindoline-1,3-dione (21, 500 mg, 1.810 mmol, 1.0 equiv.) was dissolved in cooled DMF (5 mL) in a round-bottom flask, followed by the addition of sodium hydride (109 mg, 2.72 mmol, 1.5 equiv.). The reaction was stirred for 30 min, followed by the addition of iodomethane (170 μL, 2.72 mmol, 1.5 equiv.). The reaction was stirred for an additional 18 h. The crude mixture was diluted with ethyl acetate and washed with brine, dried with MgSO₄, and concentrated under reduced pressure. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by silica gel chromatography (10% methanol in DCM) to yield 22 (310 mg, 1.07 mmol, 59% yield). 1H NMR (400 MHz, DMSO-d₆) δ 7.94 (td, J = 7.8, 4.4 Hz, 1H), 7.77 (d, J = 7.3 Hz, 1H), 7.72 (t, J = 8.9 Hz, 1H), 5.21 (dd, J = 13.1, 5.3 Hz, 1H), 3.01 (s, 3H), 2.97 – 2.87 (m, 1H), 2.76 (dt, J = 17.3, 3.7 Hz, 1H), 2.54 (td, J = 13.2, 4.5 Hz, 1H), 2.11 – 2.01 (m, 1H). ¹³C NMR (101 MHz, DMSO-d₆) δ 172.15, 169.87, 166.91, 166.54, 166.51, 164.40, 158.59, 155.98, 138.57, 138.49, 133.89, 123.58, 123.38, 120.53, 120.49, 117.53, 117.41, 55.36, 50.12, 31.50, 27.08, 21.48, 15.12, 15.05. HPLC t_R = 2.75 min (Method 1). ESI-LRMS [M + H]⁺ = 291.1.

In a microwave vial, 22 (100 mg, 0.345 mmol, 1.0 equiv.) was dissolved in DMF (2 mL), followed by the addition of tert-butyl (8-aminooctyl)carbamate (101 mg, 0.413 mmol, 1.2 equiv.) and N-ethyl-N-isopropylpropan-2-amine (301 μL, 1.723 mmol, 5.0 equiv.). The reaction mixture was heated in a microwave reactor at 110 °C for 2 h. The crude product was then diluted with ethyl acetate and washed with NaHCO₃ and brine. The organic layer was dried over MgSO₄ and concentrated under reduced pressure. The crude product was purified by silica gel chromatography (70% ethyl acetate in hexane) to give 23 (72 mg, 0.140 mmol, 41% yield). ¹H NMR (400 MHz, Chloroform-d) δ 7.49 (dt, J = 15.5, 7.6 Hz, 1H), 7.07 (t, J = 8.3 Hz, 1H), 6.86 (d, J = 8.5 Hz, 1H), 6.21 (t, J = 5.6 Hz, 1H), 4.99 – 4.86 (m, 1H), 4.52 (s, 1H), 3.24 (q, J = 6.6 Hz, 2H), 3.19 (s, 3H), 3.08 (d, J = 5.5 Hz, 3H), 3.02 – 2.88 (m, 1H), 2.85 – 2.67 (m, 2H), 2.14 – 2.01 (m, 1H), 1.64 (p, J = 7.1 Hz, 2H), 1.42 (s, 9H), 1.32 – 1.20 (m, 6H). ¹³C NMR (101 MHz, Chloroform-d) δ 171.26, 169.68, 169.04, 167.79, 155.97, 150.43, 146.95, 136.03, 134.97, 134.43, 132.52, 122.19, 116.55, 113.90, 111.26, 109.89, 79.00, 49.86, 49.60, 43.36, 42.60, 40.55, 31.91, 30.02, 29.18, 29.11, 28.41, 27.23, 26.83, 26.67, 22.13, 22.02. HPLC t_R = 3.55 min (Method 1). ESI-LRMS [M + H]⁺ = 537.3.

Compound 23 (72 mg, 0.140 mmol, 1.0 equiv.) was dissolved in dry CH₂Cl₂ (5 mL) in a round-bottom flask, followed by the addition of 4N HCl (350 μL, 1.399 mmol, 10 equiv.) in dioxane. The reaction was then stirred overnight. The solvent was then evaporated to yield 24 (50 mg, 0.111 mmol, 79% yield) and used without further purification in the next step. HPLC t_R = 2.93 min (Method 1). LRMS (ESI, +) [M + H]⁺ = 415.2.

Compound 7 was synthesized using 14 and 24 using similar conditions reported for 5. ¹H NMR (400 MHz, DMSO-d₆) δ 13.00 (s, 1H), 12.43 (s, 1H), 9.23 (s, 1H), 9.01 (s, 1H), 8.69 (s, 1H), 8.55 (s, 1H), 8.27 (d, J = 7.4 Hz, 1H), 7.78 (t, J = 5.6 Hz, 1H), 7.72 (d, J = 8.5 Hz, 2H), 7.56 (t, J = 7.8 Hz, 1H), 7.06 (d, J = 8.6 Hz, 1H), 7.00 (d, J = 7.0 Hz, 1H), 6.80 (d, J = 8.5 Hz, 2H), 6.49 (t, J = 5.9 Hz, 1H), 5.09 (dd, J = 13.0, 5.4 Hz, 1H), 4.85 (s, 2H), 4.26 (td, J = 8.4, 8.0, 4.9 Hz, 1H), 3.25 (d, J = 6.6 Hz, 2H), 3.22 (s, 3H), 2.98 (d, J = 8.5 Hz, 5H), 2.98 – 2.86 (m, 1H), 2.73 (dt, J = 17.3, 3.5 Hz, 1H), 2.54 (dd, J = 13.0, 4.3 Hz, 1H), 2.49 (s, 3H), 2.16 (t, J = 7.4 Hz, 2H), 2.02 (qt, J = 7.8, 4.2 Hz, 2H), 1.90 (dt, J = 14.0, 7.2 Hz, 1H), 1.53 (p, J = 7.1 Hz, 2H), 1.33 (d, J = 6.3 Hz, 1H), 1.33 – 1.24 (m, 4H), 1.23 (d, J = 12.8 Hz, 6H). HPLC t_R = 4.51 min (Method 2). ESI-LRMS [M + H]⁺ = 851.4. ESI-HRMS [M + H]⁺ = 851.3716 (found); (calcd C₄₂H₅₀N₁₂O₈: 851.3947).

4-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)-5-((8-((2-(2,6-dioxo piperidin-3-yl)-1,3-dioxoisoindolin-4-yl)amino)octyl)amino)-5-oxopentanoic acid (8):

The PROTAC 8 was synthesized using 19 and 20 using similar conditions reported for PROTAC 5. ¹H NMR (400 MHz, DMSO-d6) δ 12.95 (s, 1H), 11.07 (s, 1H), 9.26 (s, 1H), 9.06 (s, 1H), 8.69 (s, 1H), 8.60 (s, 2H), 7.98 (d, J = 7.9 Hz, 1H), 7.82 (t, J = 5.7 Hz, 1H), 7.73 (d, J = 8.5 Hz, 2H), 7.55 (t, J = 7.8 Hz, 1H), 7.47 (s, 1H), 7.05 (d, J = 8.5 Hz, 1H), 6.99 (d, J = 7.0 Hz, 1H), 6.79 (d, J = 8.6 Hz, 2H), 6.49 (d, J = 6.0 Hz, 1H), 5.03 (dd, J = 12.9, 5.3 Hz, 1H), 4.85 (s, 2H), 4.33 (td, J = 8.5, 5.4 Hz, 1H), 3.24 (d, J = 13.6 Hz, 5H), 3.12 – 3.04 (m, 1H), 3.00 (td, J = 6.6, 3.7 Hz, 49H), 2.92 – 2.81 (m, 1H), 2.57 (d, J = 17.6 Hz, 2H), 2.27 – 2.18 (m, 2H), 2.01 (dt, J = 12.4, 4.6 Hz, 2H), 1.94 (t, J = 7.3 Hz, 1H), 1.86 – 1.78 (m, 1H), 1.77 – 1.66 (m, 47H), 1.52 (q, J = 7.2 Hz, 3H), 1.36 (t, J = 6.2 Hz, 2H), 1.28 (s, 5H), 1.28 – 1.20 (m, 7H). HPLC t_R = 4.22 min (Method 2). ESI-LRMS [M + H]⁺ = 837.4. ESI-HRMS [M + H]⁺ = 837.3627 (calcd C₄₁H₄₈N₁₂O₈: 837.3791).

Methyl N2-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzoyl)-N5-(8-((2-(2,6-dioxo piperidin-3-yl)-1,3-dioxoisoindolin-4-yl)amino)octyl)-L-glutaminate (9):

Compounds 20 (200 mg, 0.389 mmol, 1.0 equiv.) and (S)-4-(((benzyloxy)carbonyl)amino)-5-methoxy-5-oxopentanoic acid (115 mg, 0.389 mmol, 1.0 equiv.) was dissolved in acetonitrile (5 mL) in a round-bottom flask, followed by the addition of TCFH (142 mg, 0.505 mmol, 1.3 equiv.) and 1-methyl-1H-imidazole (96 mg, 1.166 mmol. 3.0 equiv.). The reaction was stirred for 1 h. The solvent was then removed under reduced pressure and the crude product was then dissolved in CH₂Cl₂ and washed twice with saturated NH₄Cl solution and brine. The organic solvent was dried over Na₂SO₄ and the solvent was removed under reduced pressure. The crude product was purified by silica gel chromatography to yield 25 (210 mg, 0.310 mmol, 80% yield) as a light green gel. ¹H NMR (400 MHz, DMSO-d₆) δ 11.07 (s, 1H), 7.79 – 7.69 (m, 2H), 7.55 (t, J = 7.8 Hz, 1H), 7.32 (q, J = 8.6, 7.9 Hz, 4H), 7.06 (d, J = 8.6 Hz, 1H), 6.99 (d, J = 7.1 Hz, 1H), 6.51 (d, J = 6.8 Hz, 1H), 5.73 (s, 1H), 5.07 – 4.99 (m, 1H), 5.00 (s, 2H), 4.00 (td, J = 8.6, 5.1 Hz, 1H), 3.60 (s, 2H), 3.25 (s, 1H), 2.97 (q, J = 6.5 Hz, 2H), 2.93 – 2.79 (m, 1H), 2.67 (s, 6H), 2.61 – 2.50 (m, 2H), 2.48 (d, J = 5.7 Hz, 2H), 2.12 (t, J = 7.5 Hz, 2H), 1.96 (ddt, J = 34.0, 13.2, 5.5 Hz, 2H), 1.80 – 1.66 (m, 1H), 1.52 (dt, J = 17.0, 8.5 Hz, 2H), 1.35 (s, 1H), 1.31 (dd, J = 14.2, 7.9 Hz, 5H), 1.22 (d, J = 6.0 Hz, 5H). HPLC t_R = 3.32 min (Method 1). ESI-LRMS [M + H]⁺ = 678.3.

The intermediate 25 (210 mg, 0.310 mmol, 1.0 equiv.) was dissolved in ethanol (5 mL) in a round-bottom flask, followed by the addition of 10% Pd-C (65.9 mg, 0.062 mmol, 0.2 equiv.). The reaction was vacuumed three times followed by purging with H₂ gas. The reaction was stirred for an additional 3 h under H₂ atmosphere. The reaction was then filtered through celite to yield 26 (140 mg, 0.26 mmol, 83% yield). ¹H NMR (400 MHz, Methanol-d₄) δ 7.53 (dd, J = 8.5, 7.2 Hz, 1H), 7.01 (d, J = 7.7 Hz, 2H), 5.05 (dd, J = 12.6, 5.5 Hz, 1H), 3.71 (s, 2H), 3.47 (dd, J = 7.2, 5.9 Hz, 1H), 3.30 (d, J = 14.0 Hz, 1H), 3.14 (t, J = 7.0 Hz, 2H), 2.93 – 2.83 (m, 1H), 2.80 (s, 9H), 2.79 – 2.63 (m, 2H), 2.33 – 2.18 (m, 2H), 2.15 – 1.79 (m, 2H), 1.65 (p, J = 7.0 Hz, 2H), 1.48 (t, J = 6.8 Hz, 1H), 1.46 – 1.31 (m, 6H). HPLC t_R = 2.92 min (Method 1). ESI-LRMS [M + H]⁺ = 544.3.

In a round-bottom flask, intermediate 10 (14.96 mg, 0.046 mmol, 1.0 equiv.) was dissolved in DMF (1 mL) followed by the addition of TEA (19.23 μL, 0.138 mmol. 3.0 equiv.) and PyBOP (47.9 mg, 0.092 mmol, 2.0 equiv.). The reaction mixture was stirred for 30 min, followed by the addition of 26 (25 mg, 0.046 mmol, 1.0 equiv.). The reaction mixture was stirred overnight, and the solvent was then evaporated. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by reverse-phase HPLC to yield 9 as a light-yellow solid. ¹H NMR (400 MHz, DMSO-d₆) δ 12.98 (s, 1H), 11.06 (s, 1H), 9.22 (s, 1H), 9.01 (s, 1H), 8.69 (s, 1H), 8.54 (s, 1H), 8.40 (d, J = 7.1 Hz, 1H), 7.78 (t, J = 5.6 Hz, 1H), 7.72 (d, J = 8.6 Hz, 2H), 7.55 (t, J = 7.8 Hz, 1H), 7.48 (s, 1H), 7.05 (d, J = 8.6 Hz, 1H), 6.99 (d, J = 7.0 Hz, 1H), 6.80 (d, J = 8.5 Hz, 2H), 6.48 (t, J = 5.9 Hz, 1H), 5.02 (dd, J = 12.9, 5.4 Hz, 1H), 4.85 (s, 2H), 4.31 (dt, J = 8.8, 5.6 Hz, 1H), 3.59 (s, 3H), 3.24 (d, J = 14.0 Hz, 5H), 2.99 (td, J = 6.6, 3.8 Hz, 9H), 2.93 – 2.79 (m, 1H), 2.62 – 2.53 (m, 1H), 2.53 (d, J = 10.0 Hz, 1H), 2.48 (s, 4H), 2.47 (s, 1H), 2.16 (t, J = 7.4 Hz, 2H), 2.08 – 1.95 (m, 2H), 1.95 – 1.84 (m, 1H), 1.78 – 1.66 (m, 6H), 1.53 (p, J = 7.1 Hz, 2H), 1.38 – 1.28 (m, 4H), 1.24 (d, J = 17.2 Hz, 8H). ¹³C NMR (101 MHz, DMSO-d₆) δ 173.25, 171.67, 170.53, 169.38, 167.73, 166.60, 163.10, 156.33, 151.63, 151.16, 149.23, 146.85, 145.74, 136.75, 132.62, 129.42, 122.64, 121.57, 117.63, 111.58, 110.83, 109.42, 55.27, 52.86, 52.23, 48.98, 40.72, 40.52, 40.31, 40.10, 39.89, 39.68, 39.47, 32.26, 31.46, 29.49, 29.13, 26.80, 26.76, 22.61. HPLC t_R = 4.49 min (Method 2). ESI-LRMS [M + H]⁺ = 851.4. ESI-HRMS [M + H]⁺ = 851.3857 (found); (calcd C₄₂H₅₀N₁₂O₈: 851.3947).

1-(6-((3-((S)-4-carboxy-4-(4-(((2,4-diaminopteridin-6-yl)methyl)(methyl)amino)benzamido)butanamido) propyl)amino)-6-oxohexyl)-3,3-dimethyl-2-((1E,3E)-5-((E)-1,3,3-trimethyl-5-sulfoindolin-2-ylidene)penta-1,3-dien-1-yl)-3H-indol-1-ium-5-sulfonate (MTX-Cy5):

Compound 14 (4.02 mg, 0.0079 mmol, 1.1 equiv.) was dissolved in DMF (0.5 mL) in a small glass vial, followed by the addition of Cy5 Amine (5.0 mg, 0.0072 mmol, 1.0 equiv., https://fluoroprobes.com/product/cy5-amine/, Catalog # 1323-5), 2-(3H-[1,2,3]triazolo[4,5-b]pyridin-3-yl)-1,1,3,3-tetramethylisouronium hexafluorophosphate(V) (5.44 mg, 0.014 mmol, 2.0 equiv.), and N-ethyl-N-isopropylpropan-2-amine (6.25 μL, 0.036 mmol, 5.0 equiv.). The reaction was stirred for 18 h and the solvent was evaporated. LC-MS retention time: 2.55 min (method 1). ESI-LRMS [M+H]+ found 1192.2.(calcd C59H74N12O11S2: 1191.43). The crude product was dissolved in DCM (1 mL) followed by the addition of 1 mL TFA. The reaction mixture was then stirred for 3 h. The solvent was then removed under reduced pressure. The dried product was then redissolved in DMSO (approximately 1 mL), and then the crude product was purified by reverse-phase HPLC to yield MTX-Cy5 (3.35 mg, 40% over two steps) as a TFA salt. 1H NMR (400 MHz, Methanol-d4) δ 8.60 (s, 1H), 8.27 (td, J = 13.0, 5.0 Hz, 2H), 7.91 – 7.82 (m, 4H), 7.75 (d, J = 8.7 Hz, 2H), 7.29 (dd, J = 8.6, 2.3 Hz, 2H), 6.85 – 6.78 (m, 2H), 6.64 (t, J = 12.4 Hz, 1H), 6.30 (dd, J = 13.7, 4.1 Hz, 2H), 4.57 – 4.50 (m, 1H), 4.10 (t, J = 7.3 Hz, 2H), 3.62 (d, J = 10.3 Hz, 3H), 3.25 (d, J = 15.5 Hz, 3H), 3.19 – 3.06 (m, 4H), 2.37 (t, J = 7.1 Hz, 2H), 2.26 (dq, J = 13.0, 7.1 Hz, 1H), 2.17 (t, J = 7.0 Hz, 2H), 2.09 (dt, J = 14.4, 7.5 Hz, 1H), 1.82 (d, J = 7.8 Hz, 2H), 1.71 (d, J = 2.3 Hz, 13H), 1.57 (q, J = 6.9 Hz, 2H), 1.41 (q, J = 8.1 Hz, 1H). LC-MS retention time: 2.33 min (method 1), 3.20 (method 2). ESI-LRMS [M+H]+ found 1135.2.(calcd C55H66N12O11S2: 1134.4415).

QUANTIFICATION AND STATISTICAL ANALYSIS

Graphing and statistical analyses were performed using GraphPad Prism (version indicated in specific method). Final figures were prepared in Adobe Illustrator (version 25.0.1). All graphical data are expressed as individual replicates or mean ± standard deviation (SD) unless stated otherwise in Figure Legends. Statistical tests were applied as described in Figure Legends.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
DHFR	Abcam	cat # ab124814; Clone# EPR5285; RRID:AB_10975115
TYMS (TS)	Abcam	cat # ab108995; Clone# EPR4545; RRID:AB_10864610
ubiquitin	Cell Signaling Technology	cat # 3936; Clone# P4D1; RRID:AB_331292
PARP	Cell Signaling Technology	cat #9542; polyclonal; RRID:AB_2160739
b-actin	Cell Signaling Technology	cat # 3700; Clone# 8H10D10; RRID:AB_2242334
CUL-2	Santa Cruz Biotechnology	cat # sc-266506; Clone# C-4; RRID:AB_2230072
goat anti-mouse IgG, HRP Conjugate	Promega	cat # W402B; polyclonal secondary
goat anti-rabbit IgG, HRP Conjugate	Invitrogen	cat # A16096; polyclonal secondary; RRID:AB_2534770
Chemicals, peptides, and recombinant proteins
furimazine	Promega	cat # N113
Compounds 1 - 26	This article	N/A
Cy5 Amine	Fluoroprobes	cat # 1323-5
MG-132	Millipore Sigma	cat # 474787, Cas No # 133407-82-6
MLN-4924	Millipore Sigma	cat # 5054770001, Cas No # 951950-33-7
Pomalidomide (POM)	Chemshuttle	cat # 6462-3, Cas No # 19171-19-8
Cycloheximide (CHX)	Millipore Sigma	cat # 01810, Cas No # 66-81-9
5-Fluorouracil	Cayman Chemical	cat # 14416, Cas No # 51-21-8
Raltitrexed	Cayman Chemical	cat # 26079, Cas No # 112887-68-0
Methotrexate	MedChem Express	cat # HY-14519, Cas No # 59-05-2
Critical commercial assays
Nano-Glo HiBiT Lytic Reagent	Promega	cat # N3030
CellTiter-Glo Luminescent Cell Viability (CTG)	Promega	cat # G7572
Caspase-Glo 3/7 Assay System	Promega	cat # G8090
OncoPanel, Full Panel: Multi-Parameter High Content Profiling Assay, 5-day	Eurofins	cat # PSOP0015
Deposited data
mass spectrometry proteomics data	This paper	ProteomeXchange PRIDE partner repository; PXD030140
Experimental models: Cell lines
HBL-1 (human, Human diffuse large B-cell lymphoma) gender: male	Kind gift from L. Staudt, NCI, NIH	RRID:CVCL_4213
HEK293 (human, Embryonic Kidney) gender: female	ATCC, CRL-1573	RRID:CVCL_0045
MOLT-4 (human, T acute lymphoblastic leukemia) gender: male	ATCC, CRL-1582	RRID:CVCL_0013
Jurkat (human, T-cell acute lymphoblastic leukemia) gender: male	ATCC, Clone E6-1	RRID:CVCL_0065
DND-41 (human, T-cell acute lymphoblastic leukemia) gender: male	DSMZ, ACC-525	RRID:CVCL_2022
TALL5 (human, T-cell acute lymphoblastic leukemia)	Kind gift from Dr. C. Civin, U. Maryland School of Medicine.	N/A
Recombinant DNA
pET21a HsDHFR-AviTag-His10	Addgene	193883
pBMN-Ires-Lyt2	Kind gift from G. Nolan, Stanford U.	N/A
pBMN-1 DHFR-HiBiT	Addgene	179731
pHIT/EA6x3*	Kind gift from L. Staudt, NCI, NIH	N/A
pHIT60n	Kind gift from L. Staudt, NCI, NIH	N/A
Software and algorithms
Prism 8 Version 8.1.2	GraphPad Software Inc.	https://www.graphpad.com/features
Prism 9 Version 9.0.0	GraphPad Software Inc.	https://www.graphpad.com/features
Illustrator 2023 Version 25.0.1	Adobe	https://www.adobe.com/products/illustrator.html
BioRender	BioRender	https://www.biorender.com/

Significance.

The antifolate, methotrexate (MTX), is a DHFR inhibitor among other one-carbon pathway proteins. MTX has been used for decades as a potent anti-cancer agent and immune system suppressant despite its significant toxic side effects. Efforts to improve its medicinal properties have been hampered by complex and poorly understood polypharmacology. The PROTAC strategy involves the degradation of a target protein by hijacking cellular quality control machinery. We have leveraged the PROTAC modality to develop a selective DHFR degrader which we have named versortexate (VSTX). Since MTX undergoes poly-γ-glutamylation in cells which contributes to its off-target cellular toxicity, we blocked this site by conjugating an E3 ligase ligand. To facilitate this characterization of VSTX, we developed several assays. These include (1) an MTX-Cy5 ligand enabling a far-red fluorescence polarization assay to assess the binding of VSTX and its analogs to DHFR in the absence of interferences from the extended absorbance spectrum of thalidomide-based PROTACs, and (2) a bioluminescence cell-based assay to monitor DHFR abundance. Our studies revealed key pharmacological differences between MTX and VSTX. MTX treatment upregulates DHFR levels whereas VSTX potently degrades DHFR in a dose- and time-dependent manner across multiple cell lines. Intriguingly, VSTX showed a less lethal phenotype when compared with MTX, presumably due to a more restricted cellular pharmacology, supported by our unbiased global proteomics and targeted metabolomics studies. Whereas, VSTX showed a strong folate-dependent cellular toxicity, significantly greater than observed for MTX. Furthermore, we demonstrate that VSTX-mediated DHFR degradation is unaffected by the mechanisms giving rise to MTX resistance. Thus, VSTX and its analogs will be useful chemical probes to help dissect the biology of one-carbon metabolic pathways associated with autoimmune and neoplastic disease. Therefore, we foresee these probes having broad applications and will likely be of interest to the scientific and medical communities.

Highlights.

• MTX-PROTAC, Versortrexate, selectively degrades dihydrofolate reductase in vitro

• Versortrexate retains DHFR-degrading properties in an MTX-resistant HBL-1 cell line

• Versortrexate-mediated cell toxicity is highly folate-dependent.