Differentiation resistance through altered retinoblastoma protein function in acute lymphoblastic leukemia: in silico modeling of the deregulations in the G1/S restriction point pathway

Abstract

Background: As in many cancer types, the G1/S restriction point (RP) is deregulated in Acute Lymphoblastic Leukemia (ALL). Hyper-phosphorylated retinoblastoma protein (hyper-pRb) is found in high levels in ALL cells. Nevertheless, the ALL lymphocyte proliferation rate for the average patient is surprisingly low compared to its normal counterpart of the same maturation level. Additionally, as stated in literature, ALL cells possibly reside at or beyond the RP which is located in the late-G1 phase. This state may favor their differentiation resistant phenotype. A major phenomenon contributing to this fact is thought to be the observed limited redundancy in the phosphorylation of retinoblastoma protein (pRb) by the various Cyclin Dependent Kinases (Cdks). The latter may result in partial loss of pRb functions despite hyper-phosphorylation.

Results: To test this hypothesis, an in silico model aiming at simulating the biochemical regulation of the RP in ALL is introduced. By exploiting experimental findings derived from leukemic cells and following a semi-quantitative calibration procedure, the model has been shown to satisfactorily reproduce such a behavior for the RP pathway. At the same time, the calibrated model has been proved to be in agreement with the observed variation in the ALL cell cycle duration.

Conclusions: The proposed model aims to contribute to a better understanding of the complex phenomena governing the leukemic cell cycle. At the same time it constitutes a significant first step in the creation of a personalized proliferation rate predictor that can be used in the context of multiscale cancer modeling. Such an approach is expected to play an important role in the refinement and the advancement of mechanistic modeling of ALL in the context of the emergent and promising scientific domains of In Silico Oncology and more generally In Silico Medicine.

Fig. 1

Main parts of the biochemical regulation of the G1/S restriction point in normal (non-leukemic) cell cycle. The presence of growth factors leads to the constitutive activation of Cyclin D:Cdk4,6 complexes, which in turn favors the hypo-phosphorylation of retinoblastoma protein (pRb) in early G1 phase. The hypo-phosphorylated pRb maintains the ability to inhibit E2F transcription factors. Growth factors also stimulate the metabolic machinery of the cell, leading its mass to gradually grow. When cell growth reaches a critical threshold, the Cyclin E:Cdk2 and Cyclin A:Cdk1,2 complexes are activated resulting in hyper-phosphorylation of pRb in late G1 phase (where the differentiation potential is lost), liberation of E2F transcription factors and increased Cyclin A (and Cyclin E, E2F) expression, whose levels are indicative of the passage to the S-phase

Fig. 2

G1/S restriction point alterations and deregulations in BCP-ALL. In contrast with the normal cell cycle pathway (Fig. 1), Cyclin D:Cdk4,6 complexes except for hypo-phosphorylating pRb may also lead the protein to an intermediate phosphorylation status (termed “pseudo-hyper-phosphorylated”) which retains the ability to inhibit E2F transcription factors, although its phosphorus content is increased. This version of the protein is believed to have lost differentiation related functions [32], therefore its accumulation implies that the cell resides at or beyond the restriction point. Only when Cyclin E:Cdk2 and Cyclin A:Cdk1,2 complexes are activated, could the hypo-phosphorylated and pseudo-hyper-phosphorylated versions of the protein become hyper-phosphorylated and consequently liberate E2F transcription factors. The metabolism-mediated activation of these complexes is believed to exhibit differential time-course among patients due to differences in metabolism rates

Fig. 3

Simulation results of the reference model [40] for 900 min (15 h). Human HCT116 colon carcinoma cells grown under conditions of constant growth factor exposure. (a) The levels of hypo-phosphorylated pRb (hypo-pRb, purple) rapidly rise during the first min of G1 phase, due to phosphorylation of un-phosphorylated pRb species (pRb, grey) by CyclinD:Cdk4,6, and remain steadily high until the activation of the metabolism-related activating modifier switch (at 240 min); this in turn activates Cyclin E:Cdk2 and Cyclin A:Cdk1,2 complexes (Cyclin E:Cdk2 not shown in the figure). As a consequence, the majority of hypo-pRb is transformed to hyper-phosphorylated pRb (hyper-pRb, light purple). (b) In the time interval during which the levels of pRb and hypo-pRb are significant, free E2F transcription factors (E2F, green) are predominantly bound to these versions of pRb (orange). When hyper-pRb starts to dominate the levels of retinoblastoma protein, E2F is liberated and the levels of free transcription factors quickly elevate. (c) Cyclin A levels (red) show steady or even decreasing trends until the activation of the modifier switch. After this activation they gradually rise, due to E2F liberation, reaching the indicative of S-phase passage 300 (molecules/cell) threshold at approximately 600 min (indicated in green). (d) Cyclin D levels (cyan) do not show any significant variation during the execution of the model

Fig. 4

Part of the structure of the newly proposed model in SBGN format (only modified regions in relation to the reference model of [40]). The newly introduced species are encircled by a thicker frame and the new reaction arrows are colored red. The (M) addition appended after the species/complexes names indicates activation by the activating modifier

Fig. 5

Semi-quantitative/qualitative criteria used for the calibration of the newly proposed model by optimization. The criteria are defined for (a) the phosphorylation status of the different pRb forms and (b) the levels of central model species

Fig. 6

Simulation results of the newly proposed model after estimating its parameters for the mean case. (a) Hypo-phosphorylated retinoblastoma protein (hypo-pRb, purple) although rapidly formatted at the start of the G1-phase, maintains significant levels only for a limited period of time. (b) Hyper-phosphorylated forms of retinoblastoma protein (hyper-pRb all forms, dark yellow) rapidly dominate the total levels of the protein in contrast with the un-phosphorylated form (pRb, grey) which is quickly consumed. (c) Pseudo-hyper-phosphorylated retinoblastoma protein (pseudo-hyper-pRb, dark green) is directly formulated from the very first hours of the G1-phase and exclusively represents the hyper-phosphorylated forms of the retinoblastoma protein until the Modifier Activation time point, after which hyper-phosphorylated retinoblastoma (hyper-pRb, light purple) prevails. (d) Significant free E2F levels (E2F, green) are appointed only after the Modifier Activation. However, adequate levels of E2F are bound to E2F inhibiting pRb forms (orange) for a substantial time interval. (e) Cyclin A levels behavior (red) is consistent with the criteria set, showing decreasing or steady trends for the first hours of the simulation and increasing ones till its end, reaching the G1/S Transition threshold in 5200 min. (f) Cyclin D (light blue) shows insignificant variation in its levels for the entire time course of the simulation

Fig. 7

Random sampling and parameter scan result for pr _bD4 and Modifier Time parameters. (a) By randomly sampling the two parameters, the model predicts a wide range of values for the G1/S Transition Time-point. (b) Parameter value samples taken during the random sampling procedure (c) Relationship between the observed G1/S Transition Time and pr _bD4 values in random sampling parameter scan (d) Relationship between the observed G1/S Transition Time and pr _bD4 values for constant value of Modifier Time (e) Relationship between G1/S Transition Time and Modifier Time for constant value of pr _bD4 (f) Relationship between hypo-pRb detection time span and Modifier Time for constant value of pr _bD4

Fig. 8

Parameter Scan and Simulation results for the drug administration scenario. (a) Relationship between Cyclin D levels (cyan) at the end of the simulation and drug administration rate (r _drug parameter value) (b) Relationship between the ratio of pRb vs. hyper-pRb (all forms) (dark purple) and r _drug parameter value. For an intermediate rate of drug administration: (c) free Cyclin D levels are rapidly diminished and hypo-pRb (purple) is predicted to show significant levels only for a limited period of time, (d) hyper-phosphorylated pRb species (all forms) (dark yellow) are rapidly de-phosphorylated leading to the increase, (e) Cyclin A levels fail to reach adequate for G1/S transition levels, (f) Relationship between G1/S Transition Time and r _drug parameter value. Small rates of drug administration leads to accelerated G1/S transition, however, after a certain threshold the cell cycle execution is slowed down and finally completely inhibited