Energy overpowers sweet tooth in FL

In this issue of Blood, Haebe et al¹ demonstrate by single-cell sequencing that an accepted dominant driver of follicular lymphoma (FL) (ie, stimulation by its clonotypic B-cell receptor) can apparently be outcompeted by upregulation of generic metabolic pathways.

As a malignancy of the adaptive immune system, FL is a paradigmatic and fascinating disease for exploring lymphomagenesis: FL frequently does not behave like a “true” malignancy with seeming dormancy for years, and FL histopathology and evolution remain dependent on the germinal center microenvironment. Therefore, a comprehensive description of its peculiar pathophysiology, which evidently still follows many rules of the normal adaptive immune system, requires integration of functional genetics of oncogenic mutations with mechanistic understanding of physiological and aberrant immune signaling in the germinal center. In other words, cancer geneticists and immunologists have to integrate their knowledge and skills to fully understand FL. The work by Haebe et al provides an intriguing example of such an integrated approach.

Functional genetics has successfully identified and characterized recurrent mutations in FL, ranging from the seminal B-cell lymphoma 2 (BCL2) gene translocation occurring during VDJ recombination as the common founding event, followed by the chromatin-modifying variants that prolong or increase the sojourn of the cells in the germinal center, to numerous activating mutations in signaling pathways (see figure).^2,3 Although we have just begun to decipher the complex interactions between FL cells and the cellular germinal center microenvironment at the molecular level,⁴ we consider FL as neoplastic B cells that are arrested at the germinal center stage with continuous activity of activation-induced deaminase, the enzyme initiating somatic hypermutation.⁵ 

Irrespective of oncogenic mutations, simple but meticulous analyses of the mutational patterns in the clonal B-cell receptor expressed by FL cases led to a striking discovery over 20 years ago⁶: Most FL cases express a B-cell receptor (BCR) that has acquired ≥1 N-linked glycosylation motifs in their Fab portions by somatic hypermutation. High-mannose residues attached at these N motifs render FL cells susceptible to BCR-mediated stimulation by lectins presented by the FL microenvironment.^7-9 This purely immunologic mechanism exerts an apparently decisive selective advantage. With a similar approach used by Haebe et al, a recent case report demonstrated induction of a dark zone–like gene expression profile and increased somatic hypermutation by acquisition of a Fab N-motif in FL.¹⁰ 

The findings by Haebe et al, however, challenge the dominance and power of BCR-driven lymphomagenesis in FL. First, the Levy group, who wrote the current article, deserves credit for conducting a prospective trial with repetitive and extensive cytologic sampling of several anatomic FL localizations over time. Second, they leveraged single-cell RNA sequencing of fresh aspirates with highly consistent quality criteria with respect to sequencing accuracy and potential batch effects. Finally, they integrated gene expression profiling data with BCR sequencing of individual cells and sequencing of recurrent FL driver genes.

On the basis of this approach, they identified 2 of 17 patients whose lymphoma clone was composed of 2 major subclones, one of which expressed a BCR with a particular N motif and the other that lacked that motif. Acquisition of additional or alternative N motifs did not appreciatively influence the stability of the subclones. In contrast to the previous report,¹⁰ a notable aspect of the patients described here was segregation of N-motif–discrepant subclones by anatomic location. Stability of anatomic segregation over time suggests a relatively early segregation event. Unexpectedly, the subclones lacking an N motif were able to outcompete their Fab-glycosylated counterpart, including by anatomic progression with infiltration of additional lymph nodes.

The challenging question then becomes how FL cells that apparently fail to receive continuous BCR-mediated stimulation from the germinal center microenvironment can outcompete their Fab-glycosylated kin. Targeted panel sequencing failed to yield any candidate subclonal genetic driver. However, coexistence of N-motif–discrepant subclones within the same patient provided the essential experimental leverage to uncover differential gene expression profiles between these subclones.¹⁰ By comparative gene expression profiling, Haebe et al first confirmed that the transcriptional program of N-motif–carrying FL cells is characterized by increased expression of immune signaling pathways, including the BCR but also cytokine signaling. In marked contrast, FL cells lacking Fab glycosylation had higher expression of metabolism gene sets and energy sources, including glycolysis, oxidative phosphorylation, fatty acid metabolism, and mammalian target of rapamycin complex 1 signaling.

These findings create a new integrative model of FL pathogenesis that incorporates genetics, immune signaling, and now metabolism (see figure). First, the new model demonstrates that BCR-mediated “paraphysiological or superphysiological” signaling does not necessarily provide a decisive selective advantage over time, as hitherto assumed. Second, the results delineate coexisting FL subclones with clearly different needs for expansion that should lead to equally differentiated therapies that target subclone-specific vulnerabilities.

As predictable for experimental findings that open a new perspective, several research questions arise: Do the anatomic locations influence the gene expression pattern, or do they simply keep the subclones spatially separated? Is the increased metabolism in itself sufficient to provide competitive advantage, or is it the reaction to an as yet unknown other mechanism that drives expansion and therefore increases demands for energy consumption? Are the different drivers detectable by proteomics (eg, phosphorylation in signal transduction pathways) or metabolomic changes (eg, cellular energy consumption)?

Novel questions to be addressed with respect to clinical translation of the results of Haebe et al will include determining whether the identified gene expression profiles dictate also a higher clinical aggressiveness, as suggested by outcompetition. As mentioned by the authors, a true challenge is to develop diagnostic tests that reliably indicate the prevalent gene expression profile without the need to repetitively sample several lymphoma locations. Turning finally to the minority of FL cases that entirely lack Fab glycosylation: Are these the most innocuous cases, or are they also dominated by increased metabolism, perhaps having definitively outcompeted any N-motif–carrying emerging subclones during tumor evolution before clinical diagnosis? As mentioned, these analyses are more challenging because they would require interpatient and intercase comparisons. The new perspective opened by Haebe et al will certainly provide further revelations into the pathogenesis of FL for the foreseeable future.

Conflict-of-interest disclosure: The author declares no competing financial interests.