Functional Intraclonal Heterogeneity in Chronic Lymphocytic Leukemia: Proliferation vs. Quiescence

Abstract

Chronic lymphocytic leukemia (CLL) is an indolent malignancy with modest proliferation in the lymph nodes and accumulation of quiescent B cells in the peripheral blood. Targeted agents, including BTK inhibitors such as ibrutinib and the BCL2 antagonist venetoclax, have transformed therapy by disrupting proliferation, survival, and lymph node retention of CLL cells, yet CLL remains incurable. Recent studies reveal that CLL cells exist along a spectrum of proliferating, activated, and quiescent states, with dynamic transitions that shape intraclonal behavior. Whilst proliferation occurs mainly in lymph nodes, most emigrant cells in the peripheral blood become quiescent, with only a minority remaining activated. Quiescent, activated, and proliferating fractions display distinct phenotypes and CXCR4 and CD5 levels can be used to distinguish these states in the CLL life cycle. While proliferating and activated cells are more susceptible to BTK inhibition, quiescent subsets show greater sensitivity to BCL2 blockade. These functional differences, together with emerging evidence that phenotypic markers may correlate with residual disease activity, point to potential translational significance. Understanding how CLL cells switch between proliferative, activated and quiescent states will be important to uncover novel vulnerabilities and inform rational treatment strategies.

1. Introduction

Historically, chronic lymphocytic leukemia (CLL) was considered a disease of accumulating cells with defective apoptosis. Over the last decades, in vivo labeling studies using deuterated water have revealed birth rates estimated between 0.1% and 1% of the clone per day, indicating substantial proliferation of CLL cells albeit modest compared to high grade lymphomas [1]. Higher birth rates (>0.35%) inversely correlate with somatic hypermutations [2] and these patients are more likely to develop disease progression [1]. At the other end of the spectrum, there is a large fraction of quiescent CLL cells, In addition to being quiescent, these cells often display reduced B cell receptor (BCR) responsiveness [3], a state called anergy that has been extensively reviewed elsewhere [4,5]. Anergic CLL cells are characterized by low levels of surface IgM and have reduced Ca²⁺ responses when stimulated via the BCR in vitro. However, we poorly understand which CLL cells are able to proliferate, which CLL cells remain quiescent, and how CLL cells switch between these two states. This is important as the most effective (combination) therapy may need to target all CLL cells, regardless of activation state.

2. Activated, Proliferative and Quiescent States of CLL Cells

CLL cell division rates are highest in the lymph nodes (LNs), followed by peripheral blood (PB) and bone marrow [2,6], with the spleen notably understudied. Within LNs the highest proliferation is observed in proliferation centers (PCs), which are pseudofollicular structures enriched for Ki67⁺ para-immunoblasts. However, only a fraction of LN CLL cells resides in proliferation centres, and an even smaller fraction of LN CLL cells expresses proliferation markers [7,8]. LN CLL cells display higher IgM levels and elevated BCR signaling, indicating these cells are not anergic [9,10]. Together, several microenvironmental stimuli (reviewed elsewhere) lead to NFkB activation, driving proliferation [9,11,12].

Recently, a single cell transcriptomics study quantified 0.4–1% of LN CLL cells actively dividing, with 2–4% of LN CLL cells activated, and the majority of LN CLL cells in a resting state [13]. Of note, a similar spectrum of cell states exists in the PB, where we recently used flow cytometry to identify a small fraction of actively dividing cells in M-G2 phase of the cell cycle (±0.2% of PB CLL cells), whilst a larger fraction of activated CLL cells were in G1 and the majority of cells were in G0 [14]. It is unclear to what extent LN dividing, activated, and resting CLL cells differ from dividing, activated, and resting counterparts circulating in the PB. This is an important question as most experimental drugs are tested on PB CLL cells in vitro, as LN sampling is unfeasible for such studies.

2.1. Migration of Activated and Proliferative CLL Cells

In vivo labeling of dividing cells has revealed that recently divided LN emigrants in the PB have a distinct CXCR4^loCD5^hi phenotype [15,16], a fraction we found enriched for cells in G1 of the cell cycle, identified by high ki67 expression and 2n of DNA [14] (Figure 1). The CXCR4^loCD5^hi recent LN emigrant fraction displays an activated phenotype with higher levels of CD5, CD38, CD23, AID, IgD, and IgM and is receptive of BCR stimulation in vitro [3,14,16,17], indicating these cells are not anergic. The CXCR4^loCD5^hi fraction also expresses higher levels of genes associated with cell cycle entry and anti-apoptotic pathways and is enriched for epigenetic marks associated with progressive disease [16,18].

Surprisingly, this fraction in the PB contained very few cells that are in S or G2 phase of the cell cycle [14]. We found that PB CLL cells in S and G2 phase express high levels of CXCR4 and CD5, suggesting that the phenotype of CLL cells may change over the course of the cell cycle. This was supported by the Eµ-TCL1 mouse model, in which dividing cells also expressed high levels of CXCR4 and CD5 In accordance, CD40L-stimulated PB CLL cells showed increased CXCR4 levels when progressing from G1 to S and G2 phase in vitro. Based on these in vitro and in vivo observations, we hypothesize that only the most activated CXCR4^loCD5^hi recent LN emigrants progress through the cell cycle, leading to upregulation of CXCR4. Of note, the PB CXCR4^hiCD5^hi fraction that is enriched for actively dividing cells expressed high levels of IgM, CXCR4, CXCR5, and CCR7 and migrated efficiently in vitro [14], suggesting these cells may be able to return to the LNs swiftly for further rounds of activation and proliferation, and where the lifecycle of the cells may start over.

As the LNs are a major site for proliferation and survival signals [9,11,12], most studies have focused on LN retention rather than egress of CLL cells. It remains unclear why CLL cells egress from the LNs in steady state conditions. It has been postulated that CXCL12-mediated CXCR4 downregulation facilitates CLL cell egress; however, CXCR4 is also rapidly downregulated following activation through CD40, the B cell receptor (BCR), IL-4, and other receptors implicated in the LN tumor microenvironment [14,19,20]. This reduction in CXCR4 levels may only be temporary as CXCR4 quickly resurfaces when CLL cells progress through the cell cycle or when stimulation ceases [14,21]. CXCR4 is one of several markers whose expression changes upon CLL cell activation in the LNs, indicating that its levels may correlate with, rather than cause, tumor cell egress. Indeed, a range of migration markers have been found differentially expressed on LN CLL cells and CXCR4^loCD5^hi recent LN emigrants, including increased CXCR3, CD49d, and decreased CCR7 [14,16,22,23]. Moreover, the migration potential of CLL cells may not be solely dependent on expression levels of chemokine receptors, as we found activated and actively dividing cell fractions both migrated equally in response to CCL21 despite having differential CCR7 expression levels, whilst quiescent cells migrated poorly [14]. This suggests that the ability to migrate to and from activating niches in the LNs may depend on a combination of chemokine receptor expression levels and the activation status of CLL cells.

CLL cells in the LNs display elevated levels of BCR signaling [9], known to facilitate LN retention by downregulating sphingosine-1-phosphate receptor 1 (S1PR1) and upregulating adhesion molecules, such as integrins, in healthy B cells and CLL cells alike [24,25,26]. In accordance with this, disruption of BCR signaling using BTK inhibitors leads to CLL cell egress and lymphocytosis [27,28,29]. It is possible that in some LN CLL cells, disruption of BCR signaling leads to downregulation of adhesion molecules and upregulation of egress-promoting chemokine receptors such as S1PR1 [30]; however, this is at odds with the activated phenotype of CXCR4^loCD5^hirecent LN emigrants [16]. Alternatively, the loss of normal LN architecture and the sheer number of CLL cells may result in disrupted retention signals. In this scenario, LN egress is not an active process dictated by chemokine receptor levels but the consequence of ligands for LN retention no longer being accessible or not sufficiently present to support all CLL cells. Potentially, newly developed 3D in vitro culture systems, where the tissue microenvironment may be more faithfully recapitulated, can be utilized to address the question of CLL cell egress from LNs [31,32].

2.2. Re-Activation of Quiescent Cells

Following egress, only a small fraction of activated LN emigrants may progress through the cell cycle [14], while most cells may become quiescent CXCR4^hiCD5^lo cells due to a reduction in activating signals in the PB. What drives the transition to quiescence is unclear. It may be related to transformation into an anergic state. Indeed, CXCR4^hiCD5^lo cells have lowest IgM levels and reduced Ca²⁺ responses when stimulated in vitro [3,14]. Epigenetic regulation of gene transcription may also play a role, as activated CXCR4^loCD5^hi cells are enriched for epigenetic modifications associated with longevity and disease progression, whilst the quiescent CXCR4^hiCD5^lo fraction is enriched for epigenetic marks associated with indolent disease [18,33]. Of note, both bulk surface IgM levels and the balance between progression and indolence-associated epigenetic modifications have been identified as independent prognostic markers [18,34]. It is possible these parameters reflect the ratio between quiescent and activated/proliferative CLL cells.

Anergy is considered a reversible state that reverses spontaneously in vitro or can be overcome within 24 h by microenvironmental stimuli, such as IL-4 [34,35,36]. In contrast, transversing from a quiescent state to a proliferative state takes 3–5 days of persistent mitogen stimulation in vitro to allow cells to build up the molecular machinery for cell division. It is unclear to what extent quiescent CLL cells can transverse to a proliferative state in vivo. When PB CLL cells are stimulated in vitro using strong proliferation-inducing signals provided by CD40L-expressing fibroblasts, IL-4 and IL-21, ki67 levels and proliferation rates of CXCR4^hiCD5^lo cells trail behind [14], suggesting the quiescent state may not be easily overcome in a substantial fraction of PB CLL cells. This notion is supported by ex vivo trajectory analysis of single cell transcriptomics data using the velocity algorithm, suggesting that the transition between cell states was unidirectional, starting in the proliferating state, then transitioning into the activated state, and ending in the quiescent state [13]; the notion was also supported by mathematical modeling of the intraclonal dynamics of in vivo-labeled CLL cells of patients with indolent disease, which indicated quiescent cells did not return to a proliferative state within months [37].

Additionally, the induction of proliferation of CLL cells will require proliferative signals in the LN tumor microenvironment. As quiescent PB CLL cells may migrate inefficiently, they may not be able to return to the LNs without prior stimulation. In CLL with unmutated IGHV (U-CLL), Toll-like receptor stimulation in the PB may activate and mobilize cells for a return to the LNs [38]. In CLL with mutated IGHV (M-CLL), activating signals in the PB may be sparser, resulting in circulation and accumulation of quiescent cells in the PB until cell death ensues. Alternatively, quiescent cells may be able to enter the LNs but fail to reach activating niches. This could explain the presence of a large fraction of quiescent LN CLL cells. Whilst out-competed by activated cells during steady-state disease, these quiescent cells may be able to enter activating niches when therapy has wiped out activated and dividing cells, potentially driving remission.

3. Discussion

The identification of distinct subpopulations of CLL cells carries potential translational significance, as quiescent and activated/dividing cells may require distinct signals for survival. Current therapeutic approaches target distinct vulnerabilities with BTK inhibitors targeting proliferation and LN retention, whilst BCL2 inhibitors disrupt anti-apoptotic signals. As a result, subfractions of CLL cells may respond differently to these therapies. Indeed, recent LN emigrant CXCR4^loCD5^hi cells initially respond better to BTK inhibition in vivo [39], after which compensatory BTK-independent signaling pathways may take over [40,41]. Quiescent CXCR4^hiCD5^lo cells are more vulnerable to BCL2 inhibition in vitro [42], whilst activating signals in the LN microenvironment are thought to confer resistance to BCL2 inhibitors by upregulating compensatory anti-apoptotic molecules [43]. The discovery that actively dividing CLL cells in S/G2/M-phase exhibit a distinct CXCR4^hiCD5^hi phenotype and express migratory receptors such as CCR7 and CXCR5 [14] raises the possibility that this fraction contributes to tissue trafficking and disease persistence in LNs. This may help explain why combination strategies, which simultaneously interrupt survival, activation, and migratory pathways, achieve deeper remissions than monotherapies.

4. Future Directions

The studies described here warrant further research into how IGHV mutation status, disease stage, cytogenetics, epigenetics, and prior therapies influence the lifecycle of CLL and vice versa how the lifecycle of CLL affects the clinical course and response to therapy. Additionally, these observations support further exploration of phenotypic markers as dynamic biomarkers in clinical trials. The expansion of CXCR4^loCD5^hi and CXCR4^hiCD5^hi cells at relapse [14,44], suggests that these populations may reflect residual proliferative activity not captured by current measurable residual disease (MRD) assays. Given that surface phenotypes are detectable by standard flow cytometry, there is potential to incorporate phenotypic analyses into translational substudies of ongoing and future trials, particularly those testing time-limited or response adapted strategies.

In conclusion, recent findings revise the current model of CLL intraclonal dynamics and highlight functionally distinct subpopulations with clinical relevance. Further single cell analyses are required to delineate how CLL cells switch between proliferative and quiescent states and vice versa, which may expose novel vulnerabilities and inform new treatment rationales.