In this study we have characterized the global intracellular proteome of 75 early-stage, treatment-naïve CLL patients. We describe the downstream effects on protein expression of recurrent molecular features of CLL and use the unique case-control design of our early-stage cohort to identify THEMIS2 as novel, independent biomarker associated with TTFT.
THEMIS2, alternatively known as ICB-1, is a newly described member of the Themis family of proteins. Whereas THEMIS1 plays a nonredundant role in positive selection of CD8+CD4+ T cells in the thymus, THEMIS2 is mainly expressed in monocytes, follicular B cells and B1 B cells(29, 31). The latter observation is noteworthy, as human B1-like B cells have been proposed to be the cell of origin in CLL(32). Although THEMIS2 is not essential for B cell biology, it increases B cell receptor (BCR) sensitivity to low-avidity, but not high-avidity antigens by constitutive binding to Grb2, Lyn and PLCγ2(29). Crucially, THEMIS2 may increase constitutive BCR activation in the absence of antigen(29). Whether the relationship between THEMIS2 and survival is purely associative, or whether THEMIS2 may play a role in the pathophysiology of CLL, remains to be determined.
The association between THEMIS2 and outcome was present at both mRNA and protein level, as demonstrated by qPCR, mass spectrometry and ELISA. External validation was performed by analysis of two independently generated datasets, published by Herbst et al. and Knisbacher et al., which demonstrated an inverse association between THEMIS2 gene and protein expression and TFS, FFS and OS(22, 30). Of note, the association between THEMIS2 and TTFT or FFS was independent of IGHV mutational status and TP53 aberrancies. Moreover, incorporation of THEMIS2 protein levels in a model containing all components of the IPS-E index(28), a validated prognostic model in the context of early-stage CLL, improved model performance. As such, high intracellular THEMIS2 concentration, as measured by clinically applicable techniques such as ELISA, could be putatively positioned as biomarker of poor prognosis in early-stage CLL, complementary to IGHV and TP53 mutational profiling. Defining the optimal measurement modality and threshold value for THEMIS2-based prognostication requires additional work in a larger cohort of patients.
Recently, several mass spectrometry screens have been published in the context of CLL. Beckmann et al. compared the proteome and phosphoproteome in a small series of three U-CLL and three M-CLL patients, using isobaric tags for relative and absolute quantification (iTRAQ)-based mass spectrometry(26). Meier-Abt et al. used data-independent acquisition (DIA)-mass spectrometry to characterize the proteome of 117 CLL patients(27). Lastly, Herbst et al., using in-depth high-resolution isoelectric (HiRIEF)-mass spectrometry, characterized protein expression in a cohort of 68 CLL patients(22). Comparing the result of these contemporaneous, independently performed mass spectrometry screens allows for an assessment of the validity and reproducibility of proteome profiling in CLL.
There was high concordance between all four reported screens. Similarly to Meier-Abt et al. and Herbst et al., in our screen, the molecular features with the highest impact on the proteome were IGHV mutational status and trisomy 12(refs. 22,27). The proteomic effects of other features, specifically IGLV3-21R110 light chain use, del(13q14), del(11q22) and TP53 aberrancies, were negligible. Importantly, 72% of DEPs associated with IGHV mutational status and 85% of DEPs associated with trisomy 12 identified in our analysis were reported in at least one of the three screens from the literature, underscoring the robust nature of these mass spectrometry screens from a Bayesian perspective.
Differential expression of four proteins was consistently associated with IGHV mutational status across all four screens: ZAP70 and Lamin type A/C, with high expression in U-CLL, and the heavy and light chain of ferritin, with high expression in M-CLL. Although the association between ZAP70 expression and IGHV mutational status has been widely established, to our knowledge, the role of these other proteins has not been studied in the context of CLL. Lamin A/C belongs to the lamin family of proteins, components of the nuclear envelope. Lamin B1, a similar perinuclear protein, has been implicated in the regulation of SHM by binding to the immunoglobulin domains: LMNB1 expression is decreased in germinal centers, induction of SHM lowers Lamin B1 expression and suppression of Lamin B1 grossly increases SHM rates(33). Of note, in contrast to Lamin type A/C, LMNB1 expression was upregulated in M-CLL, and low LMNB1 expression was associated with inferior survival(33). Whether LMNA is similarly involved in the molecular machinery of SHM warrants further investigation. The heavy and light chains of ferritin, FTH1 and FTL, form polymers of 24 subunits to comprise ferritin, which is responsible for the intracellular sequestration of iron. Free intracellular iron is cytotoxic, as it catalyzes the formation of free radicals in the Fenton reaction. As such, the high expression of ferritin may, speculatively, render M-CLL cells differentially susceptible to ferroptosis. Of note, a recent study correlated the expression of eight ferroptosis-related genes to the clinical outcome of CLL patients, and to the predicted in vitro susceptibility of CLL cells to fludarabine, cyclophosphamide and ibrutinib, further indicating that ferroptosis may play an important role in the context of CLL(34).
Our study has some limitations. We confined the molecular characterization of our cohort to those features commonly determined in clinical practice. As such, we were unable to study the proteomic effects of other recurrent mutations, such as lesions affecting NOTCH1 and SF3B1, although in other screens, their effects on global protein expression were found to be minimal(22, 27). In addition, as biobank participation terminated upon treatment initiation, the effects of protein expression on treatment efficacy and OS could not be evaluated in this CLL biobank cohort. Finally, our cohort was powered and matched to evaluate differences using a TTFT threshold of 24 months. Consequently, proteins identified as DEPs at this threshold might not have the same discriminatory capacity at shorter (i.e., 6 to 12 months) or longer (i.e., 36 or 48 months) TTFT intervals. That said, our survival analyses demonstrate that the discriminatory capacity of THEMIS2 persists across the entire length of follow-up time in our cohort.
In summary, we provide a comprehensive characterization of the global proteome of early-stage CLL. We identify high expression of THEMIS2 as a biomarker of short TTFT, independent of IGHV mutational status or TP53 aberrations.