Systems Biology Approaches for Host-Pathogen Interaction Analysis

System Biology Approaches for Microbial Pathogenesis Interaction Analysis aids biological researchers to expand their research scope using piled up data generated through recent technological advancement. In addition, it also opens avenues for bioinformatics and computer science researchers to utilize their expertise in biological meaningful ways. It also covers network biology approaches to decipher complex multiple host-pathogen interactions in addition to giving valuable coverage of artificial intelligence.

The host-pathogen interactions are generally considered as highly specific interactions leading to a variety of consequences. The utilization of data science approaches has revolutionized scientific research including host-pathogen interaction analyses. Data science approaches coupled with network biology has taken host-pathogen interaction analysis from specific interaction to a new paradigm of understanding consequences of these interaction in the biological network. Unfortunately, basic biological researchers are mostly unaware of these advancements. In contrast, data scientists are not familiar with biological aspects of such data. System Biology Approaches for Microbial Pathogenesis Interaction Analysis will bridge these gaps through a new paradigm of understanding consequences of interaction in biological networks.

• Covers network biology approaches to decipher complex multiple host-pathogen interactions
• Gives the biological researcher insights into artificial intelligence, providing an additional competitive edge
• Provides a new paradigm for understanding the consequences of interactions in biological networks

• Language: English
• Publisher: Academic Press
• Release date: Feb 16, 2024
• ISBN: 9780323958912

Book preview

Chapter 1

Host-pathogen interactions: a general introduction

Rabbani Syed¹, Fahad M. Aldakheel², Shatha A. Alduraywish³, Ayesha Mateen², Hadeel Alnajran² and Huda Hussain Al-Numan²,    ¹Department of Pharmaceutics, College of Pharmacy, King Saud University, Riyadh, Saudi Arabia,    ²Department of Clinical Laboratory Sciences, College of Applied Medical Sciences, King Saud University, Riyadh, Saudi Arabia,    ³Department of Family and Community Medicine, College of Medicine, King Saud University, Riyadh, Saudi Arabia

Abstract

The development of any infectious disease is determined by a complex set of host-pathogen interactions (HPIs). The microbial infection involves a set of molecular processes: (1) Microbial invasion into the host crossing the primary barriers; (2) Evasion of local and tissue-specific host defenses by the microbes; (3) Microbial replication inside the host body; and (4) Suppression of host immune mechanisms. Decoding this complex array of molecular interactions between the host and the pathogen holds the key to understand the infectious disease, as well as to plan its treatment and prevention. The complex mechanisms through which a pathogen enters and proliferates inside the host body can be understood by investigating and analyzing the various stages of infection. Over the past few decades, numerous computational resources providing microbial genomic, proteomic, and metabolomics data have been developed. In this context, the introduction of artificial intelligence-based web servers has enabled us to map the phylogenetic relationship between different microbial species and to study the pathogenic proteins at the 3D level. This chapter provides some insights into molecular dynamics underlying molecular mechanisms ranging from the entry of intracellular pathogens to escaping the immune response and establishing the infection. It also introduces the readers to different computational programs or web servers, which can be used to study the host-pathogen relationship mechanisms.

Keywords

Host-pathogen interaction; infectious diseases; microbial metabolism; computational biology

1.1 Introduction

Host-pathogen interactions (HPIs) can be analyzed at cellular, molecular, organismal, or population levels. These interactions update us how bacteria or viruses maintain and survive themselves within host organisms. However, not all microorganisms have the potential to cause illness in hosts.

1.1.1 Role of pathogen

Identifying the molecular interactions between the host and the pathogen is crucial for understanding infectious disease, as well as to plan its treatment and prevention. The mechanisms through which pathogen enters and proliferates in the host body can be best understood by investigating and analyzing the various stages of infection (Varghese et al., 2020). These stages refer to the microbial invasion, evasion of host immune defenses, pathogen reproduction ability, and the host immune status. The combination of the pathogen’s own prowess, or its virulent (or toxic) components that harm the host, and the weakness of host’s immune system determines the fate of the infection in vivo. In general, depending on how the pathogen interacts with the host, there are three possible scenarios. Commensalism is the term used when the pathogen gains from a contact, but the host does not. For instance, the human digestive tract has Bacteroides thetaiotaomicron, which has no recognized advantages (Gordon and Hooper, 2001). Mutualism happens when both the pathogen and the host get benefited from the encounter. Numerous bacteria use human bodies as their home and help the host break down nutrients (Backhed et al., 2005). Parasitism is the term that means a pathogen gains from a connection while the host suffers. For instance, malaria in humans is brought on by the unicellular Plasmodium falciparum.

1.1.2 Host-pathogen relationship and mechanisms

Deciphering the invariable features of both the host genome and the microbial genomes is helpful in mapping different molecular risk factors and pathogen virulence factors. A brief understanding of the microorganism is required to comprehend the intricate relationship and how genomic and proteomic analyses might aid in understanding the pathophysiology of disease, as well as improving the diagnosis and its management (Deigner and Kohl, 2018).

A microbial infection is often regarded as a host-pathogen molecular connection, where pathogen is defined as any organism that is sheltered or feeds in another creature. The term pathogen suggests a successful conclusion for the microorganism, which is not always the case. A pathogen’s connection with the host it infects is complicated and variable based on the mode and stages of infection. The three main stages of host-pathogen relationship include prepenetration stage, penetration stage and postpenetration stage.

1.1.3 Classification of host-pathogen interactions

The infection process begins with microbial invasion of the host through the host’s main defense barriers, followed by the pathogen evading host barriers, reproducing inside the host, and the host’s immune system trying to manage or eliminate the pathogen. The severity of the illness pathogenesis and patient recovery have frequently been well connected with the pathogen’s virulence feature (Fig. 1.1).

Figure 1.1 Mechanism of host-pathogen interactions.

1.1.3.1 Invasion of host through breach of primary barriers

The invasion of the host through bacterial extracellular proteins involves breaking down of primary or secondary nonspecific defenses. Pathogen’s invasion is the common term used by medical microbiologists in the field of medical microbiology. Most biomolecules involved in invasion process are the proteins (enzymes) that operate locally to harm host cells, and/or they often have the direct effect in aiding the pathogen’s growth and dissemination. The damage that this invasive ability causes to the host may become part of the pathophysiology of an infectious disease (Todar, 2016).

Protein secretion is a key mechanism through which infections enter the host. Secreted proteins are capable of exploiting host cells for pathogen’s survival by playing a role in pathogen-host interactions. Several of these proteins participate in a variety of molecular pathways, including the control of gene expression, the suppression of defensive responses, and adjustments to host vesicular trafficking (Clark et al., 2014).

1.1.3.2 Evasion of host defenses by pathogens

Microbial infections have evolved a different mode to interfere with macrophage functions and avoid host defenses in order to efficiently colonize the host. Protein toxins and effectors released by specialized bacterial secretion systems, such as type I-VII secretion apparatuses, are the main weapons used by bacterial pathogens (Escoll, 2017). The macrophage is an important component of the host’s defense mechanism. On the one hand, this cell’s scavenger activity facilitates the uptake and death of bacteria in phagolysosomes, and on the other hand, it ensures the activation of the adaptive immune system through the presentation of bacterial antigens. From a bacterial standpoint, entering a phagolysosome is usually lethal, and in many diseases, bacteria have devised strategies to avoid this organelle’s harmful environment. Normal host cell function is frequently exploited in such evasion processes. Understanding these survival strategies can help one better understand infection etiology and host cell biology.

1.1.3.3 Pathogen replication in host

Infectious diseases are caused by pathogens, such as bacteria, fungus, protozoa, worms, viruses, and even infectious proteins known as prions. The tiniest and most basic infectious agents known to man are prion proteins. They are uncommon, improperly folded proteins that unintentionally lead other proteins in the host that have the same basic amino acid sequence to fold incorrectly. Both viral and bacterial pathogens require the right host for its effective replication and proliferation during infection process. All infection agents must be able to infiltrate their host and evade being instantly eliminated by the host immune system. For the most part, microorganisms do not harm people; the ones that possess unique virulence genes mediate interactions with the host and induce the host cells to behave in ways that promote the pathogen’s reproduction and spread.

During the course of infection, pathogenic fungus, protozoa, and other eukaryotic parasites frequently change between different forms; the capacity to flip between various forms is often necessary for the parasites to be able to remain in a host and cause disease. Some parasites, including those that cause malaria, need to go through numerous different host species in order to complete their life cycles. Unlike bacteria and eukaryotic parasites, viruses are unable to synthesize the proteins encoded by their DNA or RNA genomes due to lack of the required metabolic machinery. They completely rely on hijacking the host cell’s machinery to make their proteins and copy their genomes.

1.2 Methods for prediction of host-pathogen interactions

Understanding pathogen-host interactions between species is critical for developing a solution plan for infectious diseases. In vitro approaches require a long time to detect interactions and provide just a subset of the possible interaction combinations. As a result, modeling protein interactions has required the development of computational approaches. The major goal of such studies is to increase the prediction success rate of unknown pathogen-host interactions by combining known protein interactions between the host and pathogen (Fig. 1.2).

Figure 1.2 Methods of host-pathogen interaction predictions.

1.2.1 Ortholog-based protein interaction detection

Interolog is the word for the process of finding homologous interacting sequences for two proteins (Yu et al., 2004).

1.2.2 Domain-based detection of protein interaction

Although predictions are made using the domain-domain interaction (DDI) databases rather than the protein-protein interaction (PPI) databases, the domain-based approach uses a similar logic to interolog. This technique searches DDI databases for domain pairs between two supposedly interacting proteins once domain structures are determined from protein sequences (Yoon, 2009).

Multiple HPI characteristics are involved in microbial pathogenesis, these include the microbial proteins that target host subcellular compartments during infection, and they have an impact on host physiology. Thus, a critical step in HPI investigations is establishing the subcellular localization of proteins. Protein components involved in HPIs can be studied by PSORTdb, which is an automated subcellular localization predictor for bacterial proteins (Rey et al., 2005). Two significant databases, UniProt (weblink) and the Human Protein Atlas (https://www.proteinatlas.org/), were searched to determine the subcellular localization of human proteins (Thul and Lindskog, 2018). A human protein localization prediction tool, Hum-mPLoc 3.0 (weblink), is often used to detect the proteins for which we were unable to locate the information in these databases (Zhou et al., 2017).

The subcellular localization data for the pathogen proteins was initially gleaned from the accessible UniProt annotations. The localization information from the Burkholderia Genome Database (https://www.burkholderia.com/), which was predicted using the PSORTdb web server, is used to map pathogen proteins with no known localization information (Yu et al., 2010).

1.2.3 Biological reasoning-based prediction of host-pathogen interactions

1.2.3.1 Homology-based predictions

The anticipation of conserved interactions between a pair of proteins that have interacting homologs in a different species is the justification for these strategies. It is known as Interolog that refers to the conserved sequences. To identify Interologs, one can use the following straightforward method: The homologs a′ in the host and b′ in the pathogen, and then determine the template PPI pair (a, b) source species interactions. The fundamental benefits of this approach are its simplicity and the obvious biological basis.

Krishnadev and Srinivasan (2008) have presented a homology identification technique for predicting PHI pairs. This technique uses the template PPI databases DIP (Salwinski et al., 2004) and iPfam (Finn et al., 2014).

1.2.3.2 Structure-based predictions

Many studies rely on structural similarity and use template pathogen protein interactions (PPIs) for finding related host and pathogen interacting protein pairs. Davis et al.’s, 2007, basic theory of comparative modeling was built on their earlier research (Davis et al., 2006). They start with a set of proteins from the host and the pathogen and use sequence matching tools to look for patterns between the proteins of the host or the pathogen that are known to have known structures or known interaction protein partners. In order to forecast interacting partners statistically, sequence similarity score is only employed when structure information is lacking.

The final stage involves filtering the list of possible interactions using a network-level filter and the biological contexts of proteins. This procedure has the effect of roughly five orders of magnitude lessening the likelihood of pathogen-host interactions (PHIs). This fundamental flaw in this method is that not all pathogen proteins will show substantial structural similarity to proteins with known structures. Therefore, the usefulness of this strategy would be constrained by the lack of spatial structure information. Additionally, they can only gather a certain number of benchmark PPIs from the literature to assess the accuracy of their predictions.

1.2.4 Domain/motif interaction-based predictions

The concept of predicting PPIs for a particular species by employing domains, the basic building blocks of proteins, has been extensively studied (Wojcik and Schächter, 2001; Pagel et al., 2004). The method outlined in Dyer et al. (2007) is one of the first published research for PHI prediction. They outline a strategy that combines protein domain profiles with interactions between proteins from the same organism in order to map interactions between host and pathogen proteins. For each pair of existing functional domains (d, e), the likelihood of an interaction between protein pairs (g, h) is calculated using Bayesian statistics. The authors determine the likelihood of an interaction for every pair of host and pathogen proteins that has at least one domain in order to apply this idea to a pathogen-host system. The following formula is used to determine the possibility that a protein (g, h) will interact with a different protein (Mg):

Equation

1.2.5 Machine learning-based predictions of host-pathogen interactions

Single-character sequences are generally used to represent genetic or molecular information, and streamlined molecular-input line-entry systems are frequently used to express information about tiny molecules. Patterns in the interactions between host and pathogen chemicals and genes are revealed by machine learning (Weininger, 1988).

Machine learning (ML) can assist with assessments of RNA and DNA accessibility, transcription analysis, protein-protein interactions, and sequence-based predictions of the host organism or receptor analysis (Zou et al., 2019; Karabulut et al., 2021; Mock et al., 2021). In particular, straightforward ML methods like the multilayer perceptron (MLP) or kernel-based SVM (Crammer and Singer, 2001) deliver astounding results in problems with clear objectives. Ho first presented random forest (RF), a machine learning classification method in 1995. For instance, Karabulut et al. show that kernel-based SVM performs with a 0.96 F1 score and 0.89 area under the receiver operating characteristics curve (AUC) on the task of predicting the genus of viral infection, with RF and MLP algorithms trailing considerably close (Karabulut et al., 2021).

In these conditions, sophisticated algorithms like DL are unlikely to offer a significant additional improvement. However, in situations outside of the highly specialized data set, these approaches might increase generalization. Additionally, DL may improve the discovery of genetic variations in host or pathogen genomes that promote pathogenicity (Zou et al., 2019). Clinical metagenomics, base calling, and SNP analysis (Edge and Bansal, 2019) are two other examples of successful ML uses for HPIs (Chiu and Miller, 2019). Recurrent neural networks (RNNs), including the long short-term memory (LSTM) RNN architecture, are among the most used algorithms (Hochreiter and Schmidhuber, 1997).

More specialized CNNs are employed in some situations (Tampuu et al., 2019), or even a sequential CNN and LSTM combination (Veltri et al., 2018). RNN (and occasionally CNN) architectures have historically dominated the algorithmic scene for sequence-based analysis and provided cutting-edge performance. However, the recently released transformer architecture (Devlin et al., 2018), which is currently dominating natural language ML, is steadily spreading into the field (Zhang et al., 2021). To do this, they developed the framework known as molecule transformer—drug target interaction. Researchers found a potential candidate for a treatment for the human immunodeficiency virus using this technology (Beck et al., 2020).

1.3 Online repositories for host-pathogen interactions

The pathogen-host interactions database (PHI-base) provides biological and molecular details on genes that have been demonstrated to influence the interactions between pathogens and hosts. Phenotypes are categories that can be applied to the outcomes of such interactions. A gene is listed in PHI-base after its function in a pathogenic process is investigated utilizing gene disruption and/or transcript-level change assays. These genes are known as pathogenicity genes owing to their qualitative (disease/no disease) effect on the phenotype. The genes are referred to be virulence/aggression genes if the effect is quantitative. Effector genes were known as virulence genes in the past (Urban et al.,