Metabolic path toward TCF3 inactivation in Burkitt lymphoma

In this issue of Blood, Wilke et al¹ identify a key enzyme of one-carbon metabolism, serine hydroxymethyltransferase 2 (SHMT2), as a potential drug target in Burkitt lymphoma (BL). They show that inhibition of SHMT2 triggers autophagic degradation of a major oncogenic transcriptional regulator in BL, TCF3, which controls tonic B-cell receptor (BCR) signaling.

BL is an aggressive lymphoma that can be effectively treated by intensive chemotherapy in most pediatric patients.² However, these approaches are often poorly tolerated in adults, and although current clinical trials have been promising, relapsed/refractory BL and the toxicity from chemotherapy are significant clinical challenges.^3,4 Recent advances in deciphering the mutated BL genome have raised expectations that understanding the molecular mechanisms of BL pathogenesis can reveal potential tumor cell vulnerabilities and inform the development of innovative therapies.^5,6 For example, contributing to the long-established oncogenic function of MYC in BL pathogenesis⁷ are somatic mutations in the genes encoding the transcription factor TCF3 and its negative regulator ID3, leading to aberrant activation of a TCF3-controlled transcriptional network downstream of the BCR (see figure panel A).^5,6 However, these new insights have not yet changed anti-BL therapy, which in part reflects the difficulties in therapeutically targeting transcription factors.

Wilke et al sought to identify potential vulnerabilities in BL by using genome-wide CRISPR-Cas9 loss-of-function screens in BL cell lines. The screens uncovered essential roles for 2 seemingly functionally unrelated categories of genes for BL cell survival that caught their attention. First, the survival of BL cell lines was dependent on activators of the phosphoinositide 3-kinase (PI3K) pathway, but not on NF-κB or JAK1/STAT3 signaling, which supports the evidence that the survival of BL cells is mainly promoted by prosurvival tonic BCR signaling through the PI3K pathway.^5,6,8 Second, the screens identified several genes acting in one-carbon metabolism, including the key mitochondrial enzyme SHMT2. This finding was particularly compelling because methotrexate (MTX), an effective ingredient of the cytostatic drug armamentarium used against BL, is a folate antagonist that inhibits dihydrofolate reductase, an important regulator of one-carbon folate metabolism. Wilke et al then focused on characterizing the function of SHMT2 in BL pathogenesis because this enzyme constituted a nonessential gene in unrelated cell types.

First, Wilke et al validated SHMT2 as a drug target in BL cell lines by using SHMT inhibitors. They then determined by mass spectrometry-based metabolome profiling that SHMT2 downregulation led to a reduction in glycine and formate levels, in accordance with its enzymatic functions in one-carbon metabolism (see figure panel B). To get an idea of how the SHMT2-associated metabolic changes lead to growth impairment, Wilke et al turned to stable isotope labeling by amino acids in cell culture (SILAC)–based mass spectrometry. Several dozen proteins were downregulated upon SHMT2 knockdown, many interesting enough to spark their own enticing story about a potential involvement in BL pathogenesis. However, what really attracted their attention was the fact that one of those proteins was TCF3. In a battery of well-conducted experiments, Wilke et al then illuminated this newly identified SHMT2-TCF3-BCR axis in BL cells from every possible angle. They demonstrated that SHMT2 knockdown induced apoptosis at least in part by reducing TCF3 protein amounts and drew a direct link between SHMT2-mediated TCF3 degradation and abrogation of PI3K signaling.

Still, at this point it remained unclear how the metabolic changes associated with functional SHMT2 ablation ultimately led to the destruction of an oncogenic transcription factor. In subsequent hypothesis-driven experiments, Wilke et al identified degradation through the autophagosome as the causative mechanism that depletes TCF3 (see figure panel B). The link between SHMT2 inhibition–induced metabolic changes and autophagic TCF3 destruction was then established by demonstrating that reduced glycine and formate levels inhibit the mTOR pathway, a critical cellular nutrient sensor whose blockade induces autophagy.

The discovery of an SHMT2-TCF3-BCR axis in BL cells has identified SHMT2 inhibition as a conceivable therapeutic strategy to ablate TCF3-induced oncogenic BCR signaling. Indeed, Wilke et al show that currently available SHMT inhibitors do work to some extent in BL cell lines and in a case of primary BL with mutations in ID3 propagated for analysis using a newly developed in vitro culture system,⁹ as well as in an MYC-dependent lymphoma animal model. Subsequent high-throughput drug screening efforts could identify drugs that synergize with SHMT inhibitors in impairing BL cell growth, particularly the combination with dihydrofolate reductase inhibitors, including MTX. Interestingly, although both drugs act within one-carbon metabolism, only the SHMT inhibitor led to TCF3 degradation, which suggests independent modes of action.

This gripping story by Wilke et al is an alluring example of how the adept use of large-scale screening approaches can identify vulnerabilities in tumor cells and subsequently decipher the underlying molecular mechanism. By successively using loss-of-function screens, metabolomic profiling, and mass spectrometry, they were able to link the inhibition of a metabolic enzyme tha was newly identified as critical for tumor cell growth to specific metabolic changes. These, in turn, activated the fundamental cellular process of autophagy, causing the degradation of an oncogenic transcription factor. Moreover, the work is a precedent for the feasibility of degrading transcription factors that are often considered undruggable via targeting upstream components.

Wilke et al also report that inhibition of the SHMT2-TCF3-BCR axis was preferentially toxic to BL compared with several diffuse large B-cell lymphoma (DLBCL) cell lines. Intriguingly, a recent study found SHMT2 as a recurrent target of gene amplifications in these tumors whose genetic and pharmacologic inhibition induced loss of proliferation and viability.¹⁰ Moreover, SHMT2 was found to cooperate with BCL2 in the development of lymphomas that display features of follicular lymphoma and DLBCL, which follow a different pathogenic mechanism compared with BL.¹⁰ The jury is still out regarding the exact spectrum of lymphoproliferations that critically depend on SHMT2 activity. However, the identification of the SHMT2-TCF3-BCR axis in BL cells spotlights the possible importance of one-carbon metabolism in the physiology of the presumed BL normal cellular counterpart, the centroblast. Such knowledge may potentially be exploited for modulating antibody responses.

What is the translational potential? More research on genetically characterized primary BL samples would be required to determine the extent to which co-inhibition of SHMT2 and dihydrofolate reductase are effective in the real world. There seems to be sufficient rationale to generate a faithful BL preclinical model along the lines of the Myc/P110* model in which MYC and an active form of PI3K are expressed in germinal center B cells,⁸ perhaps by combining the Myc allele with a constitutively active Tcf3 allele. The availability of such a model may accelerate the identification of effective drugs that inhibit one-carbon metabolism. Above all, the intriguing data presented by Wilke et al provide confidence that clinical translation of these findings is a promising route to follow.