**2. Developing oxidative status pathways-based systemic biomarkers**

Since systemic biomarkers are sets of cellular factors that may represent functioning of a given pathway, development of such biomarkers starts with interactomic data - signaling pathways maps. There are numerous available solutions, including Reactome [29], BioSystems [30], GenAtlas [31], GeneGo [32], KEGG [33], etc. We have previously developed our own interactomic system dedicated exclusively to human oxidative status - Oxidative Status Interactome Map (OSIM) [34]. These interactomic systems differ in interface, data selection strategy and depth of interactome coverage. Importantly, quality of an interactomic map used for biomarker development significantly affects its specificity and sensitivity due to inherent biological overlap of targets control and multi-pathway signal reception.

oxidized NF-κB proteins are unable to transactivate the targets [42]. Interestingly, activated NF-κB protein RELA controls nuclear import of the above considered KEAP1 protein [41].

Oxidative Status Pathways: Systemic Biomarkers http://dx.doi.org/10.5772/intechopen.75543 101

All the events of pathways activation and subsequent signal transduction have different rates in different pathways. These temporal characteristics can be utilized to collect additional

Even when oxidative status pathways are activated by the same signal, ultimate effects are achieved in different time. Moreover, even within the same pathway, targets are activated at different rate. It appears that this is dictated by the complexity of signal transduction, including presence and characteristics of signal amplification circuits. It has striking effects on pathway performance. For examples, in the above-mentioned NFE2L2 pathway, some targets are activated as early as in 5 hours, while other genes fail to increase expression rate until

Another example of significantly prolonged effects of a pathway is seen in HIF1A pathway run cycle. Activated HIF1A induces expression of NF-κB genes and proteins REL, RELA and NFKB1, and DICER1. In turn, activated NF-κB (and this activation is context-dependent) transactivates genes of miRNAs *MIR-93* and *MIR-199A-5P* that are further processed by DICER1. These mature miRNAs are HIF1A suppressors that eventually block the pathway activation. Due to extremely complex chain of events, this variant of HIF1A pathway run

Temporal characteristics of the pathways analyzed should be accounted when developing and applying a systemic biomarker. This can partly be achieved by selecting appropriate ana-

Oxidative status pathways contain numerous types of molecules that can be used as biomark-

Just as signal reception and transduction are multi-step processes, so is the development of the cellular reaction. Upon decoding of the stimulus by transcription factors, the first step of the cellular reaction takes place - this is preparation to transcription initiation (interactions between transcription factors [43], competition [39], nuclear import of transcription factors [44], modulation of epigenetic machinery [45], etc.). Even this first step is complex, and subsequent stages occurring in the nucleus are legion [46]: these include DNA binding by transcription factors, RNA polymerase engagement, transcription, RNA-protein interactions, splicing, RNA modification, RNA stabilization, storage, degradation and cytoplasmic export. Further, numerous cytoplasmic processes provide or accompany generation of mature protein, and another plethora of events finalize the cellular reaction (e.g. the protein is modified, re-distributed within the cell or secreted). Remarkably, all these processes, dozens of them,

ers. These are protein-coding RNAs, miRNAs, proteins, and small molecules.

diagnostic data.

lytical level.

**2.4. Analytical levels**

**2.3. Temporal considerations**

approximately 24 hours [23].

takes considerable amount of time [25].

are affected by cellular signaling background.

## **2.1. Signal specificity**

There are dozens of oxidative status signaling pathways operating in the human cell, and most of these pathways have hydrogen peroxide or other ROS as the primary triggering signal. However, as it was shown above, ROS originate in different cellular compartments in different cellular contexts, and this is one of the basic principles of disambiguation in signal reception [35, 36]. Further divergence of pathways is achieved by their signaling background dependence: different oxidative status pathways require different kinases for functioning, while these kinases are often redox-sensitive [37, 38]. Finally, oxidative status pathways are highly specialized toward various ROS-inducing agents, which can be both chemical and physical. For example, electrophilic compounds have critically distinct effects on the AP-1 and NFE2L2 sub-pathways of the NFE2L2/AP-1 pathway (often referred to as the NRF2 pathway) [39]. An example of physical signal specificity is thioredoxin 1 triggering by ionizing radiation, UV and ultrasound [40].

Although triggering signals significantly overlap for different pathways, cellular effects of these signals greatly depend on their cellular location - and these effects are mediated by compartment-specific sensors.

#### **2.2. Sensor location and type**

Protein sensors of oxidative status pathways are ample and extremely diverse. Different pathways are greatly dissimilar in mechanisms of activation by even the same basic stimulus - hydrogen peroxide. Moreover, the same pathway may have several sensors working in tandem. For example, the above-mentioned NFE2L2 sub-pathway has a primary hydrogen peroxide sensor in cytoplasm - it is KEAP1 protein. But in addition to this cytoplasmic pool of KEAP1, the same protein is also present in the nucleus, where it acts as the second-line sensor [41]. The same pathway is also characterized by a third-line sensor - BACH1 protein, a transcriptional repressor active in reduced state [36]. Thus, depending on cellular context, NFE2L2 activity may be attenuated to different extent and so as to result in activation of different sets of target genes, since BACH1, for instance, represses only a portion of NFE2L2 dependent genes.

Another peculiar example of sensor location-induced pathway functioning modulation is known from the NF-κB pathway. Under unstimulated conditions, NF-κB proteins are bound by I-κB proteins in the cytosol preventing NF-κB nuclear import. Upon ROS formation, NF-κB proteins are oxidized, change conformation and cannot be immobilized by the I-κBs. Further developments depend on nuclear redox context and cellular signaling background, since oxidized NF-κB proteins are unable to transactivate the targets [42]. Interestingly, activated NF-κB protein RELA controls nuclear import of the above considered KEAP1 protein [41].

All the events of pathways activation and subsequent signal transduction have different rates in different pathways. These temporal characteristics can be utilized to collect additional diagnostic data.

#### **2.3. Temporal considerations**

GenAtlas [31], GeneGo [32], KEGG [33], etc. We have previously developed our own interactomic system dedicated exclusively to human oxidative status - Oxidative Status Interactome Map (OSIM) [34]. These interactomic systems differ in interface, data selection strategy and depth of interactome coverage. Importantly, quality of an interactomic map used for biomarker development significantly affects its specificity and sensitivity due to inherent biologi-

There are dozens of oxidative status signaling pathways operating in the human cell, and most of these pathways have hydrogen peroxide or other ROS as the primary triggering signal. However, as it was shown above, ROS originate in different cellular compartments in different cellular contexts, and this is one of the basic principles of disambiguation in signal reception [35, 36]. Further divergence of pathways is achieved by their signaling background dependence: different oxidative status pathways require different kinases for functioning, while these kinases are often redox-sensitive [37, 38]. Finally, oxidative status pathways are highly specialized toward various ROS-inducing agents, which can be both chemical and physical. For example, electrophilic compounds have critically distinct effects on the AP-1 and NFE2L2 sub-pathways of the NFE2L2/AP-1 pathway (often referred to as the NRF2 pathway) [39]. An example of physical signal specificity is thioredoxin 1 triggering by ionizing

Although triggering signals significantly overlap for different pathways, cellular effects of these signals greatly depend on their cellular location - and these effects are mediated by

Protein sensors of oxidative status pathways are ample and extremely diverse. Different pathways are greatly dissimilar in mechanisms of activation by even the same basic stimulus - hydrogen peroxide. Moreover, the same pathway may have several sensors working in tandem. For example, the above-mentioned NFE2L2 sub-pathway has a primary hydrogen peroxide sensor in cytoplasm - it is KEAP1 protein. But in addition to this cytoplasmic pool of KEAP1, the same protein is also present in the nucleus, where it acts as the second-line sensor [41]. The same pathway is also characterized by a third-line sensor - BACH1 protein, a transcriptional repressor active in reduced state [36]. Thus, depending on cellular context, NFE2L2 activity may be attenuated to different extent and so as to result in activation of different sets of target genes, since BACH1, for instance, represses only a portion of NFE2L2-

Another peculiar example of sensor location-induced pathway functioning modulation is known from the NF-κB pathway. Under unstimulated conditions, NF-κB proteins are bound by I-κB proteins in the cytosol preventing NF-κB nuclear import. Upon ROS formation, NF-κB proteins are oxidized, change conformation and cannot be immobilized by the I-κBs. Further developments depend on nuclear redox context and cellular signaling background, since

cal overlap of targets control and multi-pathway signal reception.

**2.1. Signal specificity**

100 Biomarker - Indicator of Abnormal Physiological Process

radiation, UV and ultrasound [40].

compartment-specific sensors.

**2.2. Sensor location and type**

dependent genes.

Even when oxidative status pathways are activated by the same signal, ultimate effects are achieved in different time. Moreover, even within the same pathway, targets are activated at different rate. It appears that this is dictated by the complexity of signal transduction, including presence and characteristics of signal amplification circuits. It has striking effects on pathway performance. For examples, in the above-mentioned NFE2L2 pathway, some targets are activated as early as in 5 hours, while other genes fail to increase expression rate until approximately 24 hours [23].

Another example of significantly prolonged effects of a pathway is seen in HIF1A pathway run cycle. Activated HIF1A induces expression of NF-κB genes and proteins REL, RELA and NFKB1, and DICER1. In turn, activated NF-κB (and this activation is context-dependent) transactivates genes of miRNAs *MIR-93* and *MIR-199A-5P* that are further processed by DICER1. These mature miRNAs are HIF1A suppressors that eventually block the pathway activation. Due to extremely complex chain of events, this variant of HIF1A pathway run takes considerable amount of time [25].

Temporal characteristics of the pathways analyzed should be accounted when developing and applying a systemic biomarker. This can partly be achieved by selecting appropriate analytical level.

## **2.4. Analytical levels**

Oxidative status pathways contain numerous types of molecules that can be used as biomarkers. These are protein-coding RNAs, miRNAs, proteins, and small molecules.

Just as signal reception and transduction are multi-step processes, so is the development of the cellular reaction. Upon decoding of the stimulus by transcription factors, the first step of the cellular reaction takes place - this is preparation to transcription initiation (interactions between transcription factors [43], competition [39], nuclear import of transcription factors [44], modulation of epigenetic machinery [45], etc.). Even this first step is complex, and subsequent stages occurring in the nucleus are legion [46]: these include DNA binding by transcription factors, RNA polymerase engagement, transcription, RNA-protein interactions, splicing, RNA modification, RNA stabilization, storage, degradation and cytoplasmic export. Further, numerous cytoplasmic processes provide or accompany generation of mature protein, and another plethora of events finalize the cellular reaction (e.g. the protein is modified, re-distributed within the cell or secreted). Remarkably, all these processes, dozens of them, are affected by cellular signaling background.

Generally, it is possible to take a "snapshot" of any of these phases of the cellular reaction and use the data to decode the initial signal. However, the more elaborate the product of the cellular reaction one analyzes, the harder it is to trace back the stimulus.

**2.6. Biomarkers quantity**

[62, 63].

tests in oncology.

details in the cited publications).

the traditional biomarker approaches.

Careful selection of systemic biomarkers candidates and their quantity are critical, since, on average, each gene is controlled by a great number of transcription factors (both transactivators and repressors) belonging to different signaling pathways. This implies a significant limitation—so that even some well-known targets cannot be used for diagnostic purposes involving pathway activation analysis. There are many examples of genes that are controlled by multiple and functionally opposite pathways even within the oxidative status systems

Oxidative Status Pathways: Systemic Biomarkers http://dx.doi.org/10.5772/intechopen.75543 103

Systemic biomarkers approach is a relatively new area of biomedicine. However, over two decades of its existence [64], significant advances have been made. Great effort spent in this area not only improved analytical algorithms, but also underlined the importance of personalized approach. For example, many prognostic markers have been suggested for breast cancer in the literature, in particular for predicting survival. But data collected in separate studies led to striking discovery of the lack of overlap of the predictive genes in most of these studies. This emphasized the need of personalized approach even within tumor groups that share the same histomorphology [65]. The reason for discrepancies is debatable—divergent patterns of expression profiles might have been due to several analytical factors considered in the present chapter, but the solution holds the same: systemic biomarkers are only informative when patterns of pathways activation, rather than changes in individual genes expression, are analyzed. This idea led foundation to development of several analytical tools and panels. Our lab developed an NFE2L2/ AP-1 pathway-based systemic biomarker for assessing slight changes in physiological parameters of the human organism using peripheral blood leukocytes as the preferred sample type [66]. The same systemic approach utilizing another set of oxidative status markers was successfully used for unveiling features of uterine cervical incompetence patients [67]. Other labs also successfully apply pathway activation-based technologies in various field and other sample types, with special attention paid to fresh solid tumors samples and paraffin blocks [68]. Of note, Oncofinder technology [69, 70] and Oncotype DX assay [71] are among the most effective interactomics/multi-gene analysis-based

In **Table 1**, some examples of suitable NFE2L2/AP-1 targets and complex markers are given along with their diagnostic properties (only area under the curve (AUC) is given, please see

As seen from **Table 1**, not only the markers may highly vary in nature, but they have different receiver operator characteristics. Notably, for each model to be studied, it is possible to choose or find a set of markers having extremely high AUCs that are hardly achievable using

**3. Current advances in developing systemic biomarkers**

Thus, most existing systemic biosensors rely on mRNA level. mRNA expression analysis is a reasonable approach to assess whether a stimulus has affected the cell or there has been a dysfunction in cellular signal transduction or decoding processes: there are only few steps between transcription factor activation (signal decoding) and mRNA maturation. However, there are still more steps that can easily be affected by the cellular functional context [47–51]. Thus, pre-mRNA might serve as a valuable alternative or addition to mRNA analysis. In one of our previous studies, we assessed whether pre-mRNAs can be used for diagnostic purposes. It appeared that two of three pre-mRNAs of single transcript variant-encoding genes had sensitivity and specificity comparable to that of the respective mRNAs [52]. Comparing such diagnostic characteristics is challenging when genes coding for more than one transcript variant are considered.

In this case, individual transcripts analysis is a great alternative to standard mRNA analysis. A pathway may control transcripts' fate individually on several levels. First, transcription factors of the pathway can directly induce individual transcripts [53]. Second, transcription factors, being central to some pathways, can attract and regulate splicing machinery themselves [54]. Third, other pathway components can easily regulate splicing machinery together with promoting target gene transcription [55]. Fourth, cellular pathways have all capabilities to individually control degradation or long-term storage of mature mRNA variants of a single gene [56–58]. Individual transcript expression-based studies are rare due to technical and interpretative difficulties [59, 60], but previously we demonstrated that this approach is highly promising in case of oxidative status pathways used for development of systemic biomarkers [28].

Proteins are also sometimes used as biomarkers in systemic diagnostics and systemic pathophysiology approaches. The drawback of this approach appears to be in decreased signalto-noise ratio leading to significant information losses. Although our lab mostly focuses on RNAs for biomarker development, we performed several attempts to use proteome as analytical level in systemic approaches [61].

#### **2.5. Signal amplification and autoregulatory blocking**

As it was mentioned above, signaling pathways are optimal for diagnostic properties only if they contain signal amplification and abruption circuits, yet they are not too complex to hinder data interpretation, and their action rate is comparatively slow.

Human oxidative status pathways are rich in signal amplification and autoregulatory blocking circuits. In one of our previous works, we discussed 15 such experimentally proven circuits of just one of the pathways - the NFE2L2/AP-1 pathway [10].

Such circuits greatly help in interpreting data and choosing time points for sample collection in repetitive measurements that greatly improve the signal decoding procedure for diagnostic purposes.

#### **2.6. Biomarkers quantity**

Generally, it is possible to take a "snapshot" of any of these phases of the cellular reaction and use the data to decode the initial signal. However, the more elaborate the product of the cel-

Thus, most existing systemic biosensors rely on mRNA level. mRNA expression analysis is a reasonable approach to assess whether a stimulus has affected the cell or there has been a dysfunction in cellular signal transduction or decoding processes: there are only few steps between transcription factor activation (signal decoding) and mRNA maturation. However, there are still more steps that can easily be affected by the cellular functional context [47–51]. Thus, pre-mRNA might serve as a valuable alternative or addition to mRNA analysis. In one of our previous studies, we assessed whether pre-mRNAs can be used for diagnostic purposes. It appeared that two of three pre-mRNAs of single transcript variant-encoding genes had sensitivity and specificity comparable to that of the respective mRNAs [52]. Comparing such diagnostic characteristics is challenging when genes coding for more than one transcript

In this case, individual transcripts analysis is a great alternative to standard mRNA analysis. A pathway may control transcripts' fate individually on several levels. First, transcription factors of the pathway can directly induce individual transcripts [53]. Second, transcription factors, being central to some pathways, can attract and regulate splicing machinery themselves [54]. Third, other pathway components can easily regulate splicing machinery together with promoting target gene transcription [55]. Fourth, cellular pathways have all capabilities to individually control degradation or long-term storage of mature mRNA variants of a single gene [56–58]. Individual transcript expression-based studies are rare due to technical and interpretative difficulties [59, 60], but previously we demonstrated that this approach is highly promising in case of oxidative status pathways used for development of systemic biomarkers [28]. Proteins are also sometimes used as biomarkers in systemic diagnostics and systemic pathophysiology approaches. The drawback of this approach appears to be in decreased signalto-noise ratio leading to significant information losses. Although our lab mostly focuses on RNAs for biomarker development, we performed several attempts to use proteome as ana-

As it was mentioned above, signaling pathways are optimal for diagnostic properties only if they contain signal amplification and abruption circuits, yet they are not too complex to hin-

Human oxidative status pathways are rich in signal amplification and autoregulatory blocking circuits. In one of our previous works, we discussed 15 such experimentally proven cir-

Such circuits greatly help in interpreting data and choosing time points for sample collection in repetitive measurements that greatly improve the signal decoding procedure for diagnostic

lular reaction one analyzes, the harder it is to trace back the stimulus.

variant are considered.

102 Biomarker - Indicator of Abnormal Physiological Process

lytical level in systemic approaches [61].

purposes.

**2.5. Signal amplification and autoregulatory blocking**

der data interpretation, and their action rate is comparatively slow.

cuits of just one of the pathways - the NFE2L2/AP-1 pathway [10].

Careful selection of systemic biomarkers candidates and their quantity are critical, since, on average, each gene is controlled by a great number of transcription factors (both transactivators and repressors) belonging to different signaling pathways. This implies a significant limitation—so that even some well-known targets cannot be used for diagnostic purposes involving pathway activation analysis. There are many examples of genes that are controlled by multiple and functionally opposite pathways even within the oxidative status systems [62, 63].
