ORMAT Supplementary Material
Hang Su1$, Gaowei Wang1$, Ruoshi Yuan2, Junqiang Wang1, Ying Tang3, Ping Ao1,2,4,5*, Xiaomei Zhu5*
1 Key Laboratory of Systems Biomedicine(Ministry of Education), Shanghai Center for Systems Biomedicine, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai 200240, China
2 School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
3 Department of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai 200240, China
4 State Key Laboratory for Oncogenes and Related Genes, Shanghai Cancer Institute, Shanghai Jiao Tong University School of Medicine, Shanghai 200240, China
5 Shanghai Center for Quantitative Life Sciences and Physics Department, Shanghai University, Shanghai 200444, China
$ Co-first author
* Co-senior author
Email:
Content
1. The construction of the core endogenous network of early myeloid cell-fate determination process
1.1 Cell cycle
1.2 Apoptosis
1.3 Metabolism
1.4 Differentiation
1.5 Signal transduction
2. Quantitative analysis of the endogenous network
2.1 ODE dynamical system analysis
2.1.1 Formulation of the ODE dynamical system
2.1.2 Algorithms for fixed points calculation in the ODE dynamical system
2.2 Boolean network analysis
2.2.1 Boolean network formulation
2.2.2 Algorithm for stable states calculation in Boolean network
2.3 State interconnection analysis
2.4 Significance of multi-stability of endogenous network
3. Module analysis of the robust stable states and their biological meanings
3.1 Module status of each stable state
3.2 The biological meanings of each robust stable state at modular level
4. Module analysis of the robust unstable states and their biological meanings
4.1 Module status of each unstable state
4.2 Biological meanings of the robust unstable states were discussed at modular level
5. Validation of the biological meanings of the robust fixed points at molecular level
6. Supplementary Figures
7. Supplementary Tables
1. The construction of the core endogenous network of early myeloid cell-fate determination process
The main approach of constructing the core endogenous network of early myeloid cell-fate determination was described as follows. Firstly, the core endogenous network was formed by a minimal set of modules. Specifically, the cell fate options involved in hematopoietic development, including self-renewal, cell death, and differentiation are tightly regulated to maintain the homeostasis of blood system. Recently, evidences suggests that the metabolites also influence the regulation of cell-fate [1]. Accordingly, a set of essential modules were selected to capture the core features of the myeloid cell-fate determination process, including cell cycle, metabolism, apoptosis and differentiation. These four modules were interconnected with each other via signaling pathways and gene regulations to form the whole endogenous network (Fig. 1, Table. S1). Secondly, as it is difficult to depict every detail of each module at the first stage due to its complexity, each module was specified by a minimal set of key agents to capture the basic functional status according to the accumulated molecular knowledge of these modules [2] (Table. S1-S2). The selected agents were characterized by their evolutionary conservation. Thirdly, the casual interactions among agents were summarized from the well-documented knowledge in literature confirmed by multiple experimental efforts, which suggests a solid bio-chemical basis of the endogenous network (Table. S3). Fourthly, the construction of endogenous network emphasized the significance of the feed-back loops in biological system, which led to an autonomous system.
The details of each module were elaborated in the following parts.
1.1 Cell cycle
The cycle of cell duplication and division is known as the cell cycle [3, 4]. Cell cycle is tightly regulated in hematopoiesis to maintain the HSC pool size throughout life, supplying the required numbers and types of mature blood cells needed by the organism. Certain characterisitics that a cell has to perform during the cell cycle are conserved in eukaryote,which is shaped by million years of evolution [5]. The eukaryotic cell cycle is commonly divided into four sequential phases: G1, S, G2, and M. The two gap phases (G1 and G2 phase) provide time for a cell to monitor the internal and external environment to ensure that conditions are suitable before the cell commits itself to divide. Among them, G1 phase is especially important in this respect. At the first stage, we mainly focused on describing the ‘on’ and ‘off’ status of cell cycle, while the subtle control to achieve the precise oscillation of cell cycle have not been considered in the core endogenous network yet. Therefore, we assumed that a cell will either stay quiescent in G0 phase or pass the restriction points in G1 phase to complete the cell cycle. The key regulators of the restriction point in G1 phase were selected to describe the core regulatory mechanisms of cell cycle according to the well-documented molecular biological knowledge.
In late G1, Cyclin D-CDK4/6 complex is responsible for initiating pRb phosphorylation [6]. The phosphorylation is extended by Cyclin E-CDK2 which drives pRb phosphorylation to its hyper-phosphorylated state [7]. When pRb are in their unphosphorylated state, they bind E2Fs to inhibit its activity; however, when hyperphorylated, pRb dissociate from E2Fs [6], results in activation of series of genes that are expressed specifically in late G1 and S-phase entry, including the gene encoding Cyclin E. Cyclin D-CDK4/6 serves as a signal sensor of intracellular and extracellular factors such as mTor[8], b-catenin[9], NFkB[10], Erk[11], etc. The actions of these two classes of Cyclin-CDK complexes are regulated by CDK inhibitors, nicely illustrated by p21. Outside the cell, TGF-β represents a major proliferation-inhibitory signal by inducing CDK inhibitors, such as p21[12]. p53 can up-regulate the transcription of p21 [13]. Lineage-specific transcription factors will arrest cell cycle, for example, C/EBPα can mediate CDK, E2F by directly protein-protein interactions [14, 15]. The activation of GSK-3β is expected to lead to a reduction of Cyclin D1 at both transcriptional level and protein degradation level [16].
1.2 Apoptosis
Programmed cell death plays a crucially important part in quality-control process in development, eliminating cells that are abnormal, misplaced, non-functional. In most case, these cell deaths occur by apoptosis: the cells shrink, condense and frequently fragment, and neibor cells or macrophages rapidly phagocytose the cells or fragments before there is any leakage of cytoplasmic contents [17]. Apoptosis occurs at a staggeringly high rate in the adult human bone marrow where most blood cells are produced [2]. This apparently futile cycle of production and destruction serves to maintain a ready supply of short-lived specialized cells that can be rapidly mobilized to fight infection or severe bleeding wherever it occurs in the body. Here, we mainly focus on the two best understood signaling pathways that can activate a caspase cascade leading to apoptosis, which are called the intrinsic pathway and the extrinsic pathway.
Caspases are members of a distinct family of cysteine proteases that share the ability to cleave their substrates on the carboxyl side of aspartate residues[18]. According to their functions, mammalian caspases can be classified into two groups: the “effector caspases” such as Caspase 3 and the “initiator caspases” such as Caspase-8 or Caspase-9. Firstly, cells can activate the caspases from the inside of a cell. In response to injury or other stresses, such as DNA damage or lack of oxygen, cells can activate the intrinsic pathway of apoptosis, which depends on the release of mitochondrial proteins that normally reside in the intermembrane space of these organelles into the cytosol. A crucial initiator is procaspase-9, which is activated by a set of proteins, such as Bid/Bad or Bax/Bak, to open channels in the outer membrane of the mitochondria and thereby release cytochrome c and activate the downstream reactions [19, 20]. The apoptosis inhibitors, illustrated by Bcl-2/Bcl-XL, work oppositely to keep channels closed [21]. The IAP family proteins such as XIAP, cIAP1 and cIAP2 serve as direct inhibitors of the members in caspase family, such as Caspase 3 and suppress different apoptotic pathways by inhibiting distinct caspases[22]. Secondly, extracellular signal proteins, binding to cell-surface death receptors domain, trigger the extrinsic pathway of apoptosis. The receptors are homotrimers and belong to the tumor necrosis factor (TNF) receptor family, which includes a receptor for TNF itself and the Fas death receptor[2]. When activated, the death receptors can recruit intracellular adaptor proteins, which in turn rectuit initiator procaspases, such as procaspase-8 [23]. The extrinsic pathway may recruit the intrinsic apoptotic pathway to amplify the caspase cascade in order to kill the cell. TNF receptors can also activate the NF-kB pathway which can promote cell survival [24].
1.3 Metabolism
Metabolic flexibility fuels divergent cell-fates, which include proliferation and self-renewal to maintain stem and progenitor pools, lineage specification for tissue regeneration and cell death to replace the damaged or abnormal cells [25]. Beyond a role in energetic support, new evidence implicates that nutrient-responsive metabolites serve as mediators of crosstalk between metabolic flux, cellular signaling, and epigenetic regulation of cell fate[1]. Proliferation cells acquire alterations to the metabolism of all four major classes of macromolecules: carbohydrates, proteins, lipids and nucleic acids [26]. Here, we mainly focused on describing the energy metabolic alteration between aerobic metabolism and anaerobic metabolism in hematopoietic development. We chose several essential regulatory enzymes, PKM2, GLS to represent anaerobic energy supplement achieved by glycolysis, PDH and IDH to represent aerobic energy supplement achieved by TCA and oxidative phosphorylation. Metabolic pathways active in proliferating cells are directly controlled by signaling pathways, such as mTor, AMPK, etc[25]. Besides, some key transcriptional regulators such as Myc, HIF are intimately linked to metabolic pathways through transcriptional regulation of metabolic enzymes.
1.4 Differentiation :
Lineage commitment and cellular differentiation of multipotent cells could be triggered by extrinsic signals or intrinsic signals or by both at different development stage, which are ultimately controlled by transcription factors (TFs). Therefore, a small number of critical TFs, which have been elucidated to govern the direction of differentiation, were recognized as ‘master regulators’ of lineage-specification (reviewed in [27-29]). Multiple transcription factors with mainly cell-type-restricted expression patterns are usually analyzed by the loss-of-function and force-to-express studies and have been recognized as indispensable component in the orchestration of myeloid-cell maturation. Each of these nuclear proteins drives expression of a characteristic set of lineage-specific target genes, thereby instructing to adopt a certain lineage-specific programs dependent upon TF’s expression levels and timing. For example, GATA switch served as a well-studied genetic circuit to describe the maintenance and differentiation of HSCs: GATA2 plays significant roles in the emergence and maintenance of HSCs; At the stage of CMPs (common myeloid progenitors), competition between PU.1 and GATA1 led to the binary-decision into the bi-potent progenitors GMPs (granulocyte/macrophage progenitors) or MEPs (megakaryocyte/erythrocyte progenitors). Further, the antagonism between Gfi-1:Egr/Nab in GMP and the EKLF:Fli-1 in MEP promote cell-fate specification towards four distinct cell types: granulocytes, monocytes/macrophages, erythrocytes and megakaryocytes [29]. Therefore, Eight lineage-specific TFs, including GATA2, GATA1, PU.1, C/EBPs, EKLF, Fli-1, Egr/Nab and Gfi-1, were selected to describe the core structure of the differentiation module of the endogenous network. The developmental stage of a cell can be identified by the expression of these lineage-specific TFs [29, 30]. In this context, HSCs were characterized by the highly expressed GATA2 without other-lineage-specific TFs expressed; neutrophils were characterized by the highly expressed C/EBPa and Gfi-1, as well as the low expressed PU.1; monocytes were characterized by co-expressed PU.1, Egr/Nab and C/EBPa; erythrocytes were characterized by the co-expressed GATA1 and EKLF; megakaryocytes were characterized by the co-expressed GATA 1 and Fli-1; the GMPs were characterized the co-expressed PU.1, C/EBPa, Egr/Nab and Gfi-1; while the MEPs were characterized by the co-expressed GATA-1 , EKLF and Fli-1in their differentiation module.
1.5 Signal pathways
To make a multi-cellular organism, cells must communicate with each other. Complex intracellular mechanisms are needed to control which signals are emitted at what time and to enable the signal-receiving cell to interpret those signals and use them to guide its behavior. Reception of the signals depends on receptor proteins, usually at the cell surface, which bind the signaling molecules. The binding activates the receptor, which in turn activates one or more intracellular signaling pathways. These relay chains of molecules—mainly intracellular signaling proteins—process the signal inside the receiving cell and distribute it to the appropriate intracellular targets. These targets are generally effector proteins, which are altered when the signaling pathway is activated and implement the appropriate change of cell behavior. Here, several conserved signaling pathways, including PI3k-Akt pathway, MAPK pathway, Wnt pathway, TGFb pathway, Notch pathway, AMPK pathway, p53 regulation pathway, etc, were selected to describe the crosstalk between the four modules
2. Quantitative analysis of the endogenous network
The core endogenous network was quantified and analyzed by using a set of ordinary differential equations (ODEs) and Boolean dynamics respectively.
2.1 ODE dynamical system analysis
2.1.1 Formulation of the ODE dynamical system
The core endogenous network was quantified by a set of ordinary differential equations (ODEs) [31, 32]. As many detailed mechanisms and parameters of interactions in the network are still unknown, we considered a trade-off between model tractability and detail, and used a dimensionless modeling framework here by normalizing the agent values that range from 0 to 1. “0” represented inactivated state, while “1” represented full activated state. The sigmoid-shaped Hill functions, which have been commonly used to model many biological processes including the transcriptional regulations and signaling transductions[33, 34], were adopted in this quantitative method to capture the core feature of transcriptional and post-transcriptional activation/inhibition among agents. For this level of analysis, the microscopic details underlying the forms of equation and kinetic parameter values were not required to be precise.
The dynamics of the activation/expression level of each protein, taking protein x as an example, was given by a standard way in equation 1:
Where represented the activation/expression level of protein x at time t. represented the linear degradation rate of protein x at time t. represented the degradation constant. represented the nonlinear integrated production rate of protein x. was influenced by the non-linear interactions from other agents towards protein x. We chose two standard and widely used Hill equation forms to describe the integrated production rate of protein x.