Intrinsic Disorder Angle of the HCV Co-Infection and Super-Infection with Other Hepatitis Viruses

1. Introduction

Among various diseases affecting liver, a very significant place is hold by viral hepatitis caused by a viral infection leading to liver inflammation and damage, which, being untreated is often associated with serious health issues, such as liver scarring or cancer. Although there are five different types of the hepatotropic hepatitis viruses, A, B, C, D, and E (HAV, HBV, HCV, HDV, and HEV) the most common types of viral hepatitis are hepatitis A, hepatitis B, and hepatitis C. In fact, it is estimated that there are more than 100 million HAV infections (∼1.5 million of HAV clinical cases occurring globally every year) [1], with hepatitis A caused by HAV being globally the most common form of the acute viral hepatitis [2]; 400 million people worldwide are chronically infected with HBV [3,4]; whereas the estimated global prevalence of HCV infection mounts to 2.2%, which corresponds to ~130 million HCV-positive persons worldwide [5], with the highest prevalence of HCV infection (15%-20%) being reported from Egypt [6,7,8]. However, the most severe form of viral hepatitis is associated with the HBV-HDV co-infection, which is observed in ~20 million people [9,10,11]. Furthermore, it was estimated that ~ 939 million (i.e., 1 in 8 individuals globally) have ever experienced HEV infection, with 15-110 million individuals having recent or ongoing HEV infection [12].

Although infection with all hepatitis viruses leads to the development of disease, and although they are not easily distinguishable clinically from each other, not all types of hepatitis are equally serious, being manifested with different characteristics. Hepatitis A, being transmitted mainly by direct contact with patients who have been infected or by ingesting contaminated water or food, is very common in n low-income countries but typically does not lead to serious complications or long-lasting illness and does not cause chronic infections [13]. Hepatitis B, also known as post-transfusion hepatitis, being transmitted through contact with infected blood or body fluids (i.e., injection drug use, unscreened blood transfusions, unsafe health care, unsafe injection practice, birth, sexual practices that lead to exposure to blood, tattoos and body piercings) [14], can be very serious if the initial viral infection progresses to a chronic infection, which is a common cause of liver cirrhosis and hepatocellular carcinoma (HCC) worldwide [15]. While no initial symptoms are typically caused by HCV infection, which is most efficiently transmitted through large or repeated direct percutaneous exposures to blood (e.g., transfusion or transplantation from infectious donors, injecting drug use) [16], it is estimated that a lifelong, chronic infection can be developed in 60-80% of patients, and can be associated with 27% of cirrhosis and 25% of HCC worldwide [8]. Being a sub-viral agent, whose genome codes for the only virus protein, the Delta antigen, for its replication and spread, HDV depends on the presence of HBV in the infected cells [11,17,18]. Similar to HBV, HDV is transmitted through contact with infected blood or blood products or broken skin (via injection, tattooing etc.), or through personal contacts [19]. Severity of hepatitis dramatically increases as a result of the HDV-HBV co-infection, as in HDV/HBV carriers, HDV causes threefold and twofold increase of the HCC and mortality risks, respectively, in comparison with the chronic carriers of HBV alone [11,20,21]. Finally, an emerging zoonotic pathogen HEV, being known worldwide as a leading cause of acute viral hepatitis, typically results in the asymptomatic or self-limiting infection in the general population but might show a high risk of developing chronic infection in immunocompromised patients or severe clinical outcomes in pregnant women [12,22].

An important feature of hepatitis viruses is their propensity for co-infection and super-infection. Both of these phenomena reflect the presence of more than one type of hepatitis virus in an individual, but they are based on different mechanisms, where co-infection corresponds to a situation, where a person is infected with multiple types of viruses simultaneously, whereas super-infection refers to a case, where primary infection with one type of hepatitis virus is followed by a secondary infection with a different hepatitis virus type. Since often it is difficult to unequivocally determine how hepatitis viruses of different types ended up in one individual, we will not differentiate between these mechanisms. For HCV, co-infections were found with HAV [23,24,25,26], HBV [27,28], HBV-HDV [29,30,31,32,33,34,35,36,37], and HEV [38,39,40,41]. Co-infections typically cause more severe illness, as evidenced by increased morbidity and mortality in chronic HCV patients co-infected with HAV and HBV [24] and already emphasized dramatically increased severity of hepatitis in the HDV-HBV co-infection [11,20,21].

It is known that intrinsic disorder is commonly present in viral proteins, where it plays a number of crucial functional roles [42,43,44,45,46,47,48,49]. It was pointed out that intrinsic disorder acts both as an armor and weapon of viruses [45,47]. On the other hand, intrinsic disorder in host proteins is crucial for the appropriate response to infection and also mediates interactions between virus and host cell proteins [50,51,52]. Generally, the protein-protein interactions (PPIs) between hosts and viruses are known to play important roles in host immune responses and define the peculiarities of viral infections [53]. It is not surprising therefore that virus-host interactomes have being established and intensively analyzed for many viruses, including hepatotropic viruses, such as the HBV [54,55,56,57,58], HCV [59,60,61,62,63,64,65,66,67], and HEV [68,69,70]. Although many facets of virus-host interactions are known, important aspects related to the prevalence and functionality of intrinsic disorder in viral interactomes remain mostly unexplored. In fact, there are only a few studies looking at the roles of intrinsic disorder in interactions between the host proteins and hepatotropic viruses [50,62]. Even less is known about prevalence and roles of intrinsic disorder in proteins related to the hepatotropic co-infections. The goal of this study is to fill this gap by conducting the bioinformatics analysis of intrinsic disorder in human proteins interacting with hepatotropic viruses, with special focus on host proteins that can interact with more than one type of hepatitis viruses.

2. Materials and Methods

2.1. Protein Datasets

Information about proteomes of human hepatitis viruses A, B, C, D, and E was retrieved from UniProt [197]. For HAV, we analyzed genome polyprotein from human hepatitis A virus genotype IB (isolate HM175) (HHAV) (human hepatitis A virus (isolate Human/Australia/HM175/1976) (UniProt ID: P08617). For HBV, we used hepatitis B virus genotype D subtype ayw (isolate France/Tiollais/1979) (HBV-D) (Proteome ID: UP000007930). For HCV, genome polyprotein from Hepatitis C virus genotype 1a (isolate H77) (HCV) (UniProt ID: P27958) was used. For HDV, we analyzed S-HDAg (UniProt ID: P0C6L3) and L-HDAg (UniProt ID: P29996) from hepatitis delta virus genotype I (isolate D380). Hepatitis E virus genotype 1 (isolate Human/China/HeBei/1987) was selected to analyze its non-structural poly-protein ORF1 (UniProt ID: Q81862), pro-secreted protein ORF2 (UniProt ID: Q81871), and protein ORF3 (UniProt ID: Q81870).

To get host proteins interacting with hepatotropic viruses, we used the STRING_Viruses platform (http://viruses.string-db.org/). Resulting HAV-host network included 7 HAV proteins and 486 Pan troglodytes proteins. Note that with the settings used in this study, instead of human proteins, STRING_Viruses pulled out Pan troglodytes proteins for the HAV-host interactome. Since the euchromatic regions of chimpanzee (Pan troglodytes) genome share ~98% sequence similarity with the human (Homo sapiens) [86], the derived set of the host proteins interacting with HAV can be used for the purposes of this study. PPI network linking HBV proteins with human proteins contained 5 HBV proteins (proteins P, X, external core antigen, middle envelope protein, and capsid) and 78 human proteins. Human interactome of HCV proteins included 8 viral and 202 human proteins. In HDV-human PPI network, there are 2 HDV and 11 human proteins. In the PPI network between the HEV and human proteins there were 9 human proteins. All retrieved host proteins interacting with hepatotropic viruses were used in global disorder analysis.

Assuming that in co-infection (or super-infection), different hepatotropic viruses infecting the same cell can be involved in interaction with same set of host proteins, we looked for host proteins shared by different hepatitis viruses within the corresponding virus-host interactome. This analysis retrieved 33 such proteins. Their IDs and major properties are listed in Table 1.

The interactability of these proteins within the set of shared host proteins and within individual PPI networks was also evaluated in terms of internal and global node degree, respectively. Next, we selected a subset of 11 host proteins interacting with at least three different hepatitis viruses and conducted more focused functional disorder analysis.

2.2. Computational Analyses of Viral and Host Proteins

The sequence-based and structure-based analyses of proteins in this study were conducted using the web-based computational tools, such as UniProt [197] and AlphaFold [198]. Disorder-based analysis was done using the RIDAO platform [199]. For functional disorder analysis we used the D²P² database [200], whereas liquid-liquid phase separation potential of query proteins was evaluated by the FuzDrop platform [147,148,201]. The STRING database [202] in multi-protein and single-protein modes was used for evaluation of the protein-protein interactions (PPIs) and analysis of functional enrichment in the resulting PPI networks and interactomes.

3. Results and Discussion

3.1. Intrinsic Disorder in Proteomes of Human Hepatitis Viruses

We started with the evaluation of the intrinsic disorder status of the proteomes of five human hepatitis viruses.

3.1.1. Hepatitis A Virus

HAV is a non-enveloped picornavirus containing a positive-sense, single-stranded RNA genome, which is packaged in a protein shell [71] and encodes a single genome polyprotein of 2,227 residues. At maturation, polyprotein is cleaved into functional viral proteins, such as capsid proteins VP1 (residues 492-765), VP2 (residues 24-245), and VP3 (residues 246-491) that form an icosahedral capsid enclosing the viral genome [72], capsid protein VP0 precursor that serve as a component of immature procapsid (residues 1-245) [72], capsid protein VP4 (residues 1-23) that participates in the assembly of an icosahedral capsid [73], and a capsid protein VP1-2A (residues 492-836) containing an assembly signal (residues 766-836) [73]. In addition to capsid proteins, genome polyprotein includes protein 2B (residues 837-1,087) acting as a viroporin among other functions [74], protein 2BC (residues 837-1,422) affecting membrane integrity and increasing the membrane permeability [74,75], protein 2C (residues 1,088-1,422) that has RNA-binding activity [76] and causes structural rearrangements of intracellular membranes [75], precursor protein 3ABCD (residues 1,423-2,227), precursor protein 3ABC (residues 1,423-1,738) that is targeted to the mitochondrial membrane where the viral protease 3C cleaves and inhibits the host antiviral protein MAVS [77] and also can efficiently cleave the 2BC precursor [78], protein 3AB (residues 1,423- 1,519) interacts with the 3CD precursor and shows RNA binding activity [79], protein 3A (residues 1,423-1,496) that anchors the 3AB and 3ABC precursors to the membrane via its hydrophobic domain [80] and shows RNA binding activity [79], viral protein genome-linked (residues 1,497-1,519) acting as a primer for the viral RNA replication [81], serine protease 3C (residues 1,520-1,738) that is responsible for generation of the mature viral proteins from the precursor polyprotein [82], protein 3CD (residues 1,520-2,227) interacting with the 3AB precursor and RNA [79], and RNA-directed RNA polymerase 3D-POL (residues 1,739-2,227) that replicates genomic and antigenomic RNA by recognizing replications specific signals.

Figure 1 represents intrinsic disorder profile generated for genome polyprotein from human hepatitis A virus genotype IB (isolate HM175) (HHAV) (human hepatitis A virus (isolate Human/Australia/HM175/1976) (UniProt ID: P08617) and shows that this genome polyprotein is moderately disordered. This is based on the accepted classification of proteins based on their percent of predicted intrinsically disordered residues (PPIDR) values as highly ordered, moderately disordered, and highly disordered, if their corresponding PPIDR values are below 10%, between 10% and 30%, and above 30%, respectively [83,84]. Additional classification mode is given by the analysis of the protein average disorder score (ADS) values, as proteins ADS < 0.15, 0.15 ≤ ADS < 0.5, and ADS ≥ 0.5 are considered as highly ordered, moderately disordered/flexible, or highly disordered, respectively. Since polyprotein has PPIDR _{PONDR VSL2} of 16.3% and ADS _{PONDR VSL2} of 0.306, it is classified as moderately disordered. Importantly, Figure 1 shows that most of the cleavage site are located within the intrinsically disordered regions (IDRs, i.e., regions with the disorder scores above 0.5 threshold). Furthermore, according to this analysis, mature HAV proteins show different levels of intrinsic disorder and based on their ADS_{PONDR VSL2} values, can be grouped as follows: VP4 (0.891±0.025) > VP0 (0.692±0.094) > VP2 (0.671±0.072) > VP3 (0.501±0.032) > VP1 (0.409±0.025) > 2A (0.397±0.032) > 2B (0.304±0.020) > 2C (0.239±0.022) > 3A (0.192±0.003) > 3B (0.184±0.001) > 3C (0.161±0.014) > 3D (0.087±0.033). Therefore, VP0, VP2, VP3, and VP4 are classified as highly disordered proteins. Remaining HAV proteins, with the only exception for the RNA-directed RNA polymerase 3D-POL, are moderately disordered. Detailed discussion of the functional roles of intrinsic disorder in individual HAV proteins and analysis of the variability of their disorder predispositions in different human HAV genotypes and subtypes are outside the scopes of this work and will be a subject of dedicated study.

Next, we looked at the HAV-host interactome via the STRING_Viruses platform (http://viruses.string-db.org/) using low confidence 0.15 for minimum required interaction score (to ensure maximal inclusion of HAV proteins in the corresponding virus-host PPI network). Based on these settings, the platform generated a dense network containing almost 500 host proteins (see Figure 2).

FuzDrop analysis indicated that although neither the whole-length genome polyprotein nor mature HAV proteins can undergo spontaneous liquid-liquid phase separation (they are characterized by the probability of spontaneous liquid-liquid phase separation, p_LLPS, below the 0.6 threshold), capsid proteins VP0 (p_LLPS = 0.2160), VP1 (p_LLPS = 0.3729), and VP2 (p_LLPS = 0.1640), can act as droplet-client proteins as they contain droplet-promoting regions, which can induce their partitioning into condensates. These observations suggest that HAV can somehow affect cellular LLPS processes, and corresponding distortions can be related to the pathogenic mechanisms of this virus. This is an intriguing hypothesis in lights of the recently recognized importance of such capability for many viruses [87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107], including HBV (see below) [108,109].

3.1.2. Hepatitis B Virus

HBV is a partially double-stranded DNA virus of the genus Orthohepadnavirus and a member of the Hepadnaviridae family of viruses. HBV genome that replicates through an RNA intermediate, encodes seven proteins. Based on the antigenic epitopes found on its envelope proteins, HBV can be divided into four major serotypes (adr, adw, ayr, and ayw), which, based on the overall nucleotide sequence variation of the genome, are further divided into ten genotypes (A–J) and forty subgenotypes [110]. Since the evaluation of the related variability in the intrinsic disorder predispositions of the HBV proteins is outside the copes of this study, we present below results of the intrinsic disorder analysis of proteins in just one variant, namely hepatitis B virus genotype D subtype ayw (isolate France/Tiollais/1979) (HBV-D) (Proteome ID: UP000007930). The large envelope protein encoded by the gene S is a transmembrane protein that exists in two topological conformations, external and internal, Le-HBsAg and Li-HBsAg, respectively, with Le-HBsAg being responsible for initiating infection via attachment to the cell receptors, and with Li-HBsAg being involved in virion morphogenesis and acting as a matrix protein by mediating the contact with the nucleocapsid. Three isoforms are produced by alternative splicing and alternative initiation, a 389-residue-long canonical form known as large envelope protein (L, LHB, and L-HBsAg), a 281-residue-long middle envelope protein (M, MHB, or M-HBsAg), which is different from L by missing residues 1-108, and a 226-residue-long small envelope protein (S, SHB, or S-HBsAg), which is different from L by missing residues 1-163. Capsid protein (183-residue-long) and external core antigen (212-residue-long) are both encoded by the same gene C via the alternative initiation. Capsid protein is responsible for formation of an icosahedral capsid, whereas external core antigen can regulate the immune response to the intracellular capsid [111]. A 153-residue-long protein X is a multifunctional protein with roles in promoting viral transcription, genome replication, and silencing host antiviral defenses [112]. Protein P is an 832-residue-long multifunctional enzyme responsible for the conversion of the viral RNA genome into dsDNA in viral cytoplasmic capsids, which also possesses DNA polymerase activity, being capable of copying either DNA or RNA templates, and shows a ribonuclease H (RNase H) activity utilized for cleavage of the RNA strand of RNA-DNA heteroduplexes [111]. HBV genome also encodes two putative uncharacterized proteins of 10.4 and 15.3 kDa. Figure 3 represents results of the intrinsic disorder analysis of these proteins and shows that all of them contain noticeable levels of disorder. Based on their ADS_{PONDR VSL2} values, these proteins can be grouped as follows: putative uncharacterized 10.4 kDa protein (0.88±0.13) > protein X (0.48±0.28) > capsid protein (0.44±0.34) > putative uncharacterized 15.3 kDa protein (0.43±0.25) > large envelope protein (0.42±0.31) > protein P (0.38±0.27) > external core antigen (0.37±0.34) > middle envelope protein (0.31±0.28) > small envelope protein (0.25±0.28).

Figure 4 represents the PPI network linking HBV proteins with human proteins. This network, which was generated by STRING_Viruses platform (http://viruses.string-db.org/) using medium confidence of 0.4 for the minimum required interaction score, contains 5 HBV proteins (proteins P, X, external core antigen, middle envelope protein, and capsid) and 78 human proteins.

Analysis of the LLPS predisposition of HBV proteins indicated that all of them are somehow related phase separation. In fact, two HBV proteins, large envelope protein (p_LLPS = 0.9812, 5 DPRs, residues 1-23, 44-108, 124-153, 201-230, and 265-285) and putative uncharacterized 10.4 kDa protein (p_LLPS = 0.9938, one DPR, residues 1-74) are expected to serve as droplet drivers, since they have high probability of spontaneous liquid-liquid phase separation. Furthermore, all the remaining HBV proteins (protein P (p_LLPS = 0.4030, one DPR, residues 162-313); capsid protein (p_LLPS = 0.3354, one DPR, residues 130-183); protein X (p_LLPS = 0.2173, one DPR, residues 18-56); external core antigen (p_LLPS = 0.2376, one DPR, residues 159-212); and putative uncharacterized 15.3 kDa protein (p_LLPS = 0.3429, one DPR, residues 1-29)) are potential droplet-clients, since although their pLLPS values are below the 0.6 threshold, all of them contain droplet-promoting regions. Importantly, the role of LLPS in HBV infection and development of HBV-driven hepatocellular carcinoma (HCC) were shown to be dependent on LLPS. In fact, it was shown that protein X binds to CREB1 and increases the cellular prion protein (PrPC) expression promoting PrPC phase separation and thereby activating the NF-κB signaling pathway and promoting tumor progression [109]. Another side of the HBV-specific LLPS mechanism is given by the roles G-quadruplexes of the HBV covalently closed circular DNA (cccDNA) in virus replication via phase separation [108].

3.1.3. Hepatitis C Virus

HCV, with at least 6 genotypes and numerous subtypes, is a member of the hepacivirus genus from the Flaviviridae family. This enveloped virus has a positive-strand single-stranded RNA genome encoding a genome polyprotein of ~3,000 residues, which is cleaved at maturation to 10 functional proteins (core protein C, and glycoproteins E1 and E2, integral membrane protein p7, and nonstructural proteins NS2, NS3, NS4A, NS4B, NS5A, and NS5B). Prevalence and functionality of intrinsic disorder in 32 HCV polyproteins across the 18 genotypes with 8 isolates for genotype 1b, 2 isolates for genotypes 1a, 1c, 2a, 2b, 3a, 5a, 6a, and 1 isolate for genotypes 2c, 2k, 3b, 3k, 4a, 6b, 6d, 6g, 6g, and 6k were analyzed in a previous study [43]. Using a wide spectrum of bioinformatics techniques it was shown that HCV proteins contain noticeable levels of intrinsic disorder, and that the associated structural flexibility is crucial for many of their functions [43]. In line with those earlier observations, Figure 5 represents the RIDAO-generated per-residues disorder profile of genome polyprotein from Hepatitis C virus genotype 1a (isolate H77) (HCV) and shows that the entire polyprotein can be classified as moderately disordered. In fact, it is characterized by PPIDR _{PONDR VSL2} of 19.5% and ADS _{PONDR VSL2} of 0.32±0.24. Based on their disorder status measured in terms of the ADS_{PONDR VSL2} values, HCV proteins can be arranged as follows: C (0.60±0.30) > NS5A (0.53±0.31) > NS3 (0.33±0.19) > NS5B (0.31±0.17) > E2 (0.25±0.15) > NS4B (0.24±0.13) > E1 (0.19±0.12) > NS4A (0.17±0.16) > NS2 (0.13±0.09) > p7 (0.10±0.08). When using PPIDR-based classification, these proteins for the following order: C (65.9%) > NS5A (46.7%) > NS3 (20.3%) > NS5B (14.6%) > NS4A (7.4%) > NS4B (4.6%) > E2 (2.3%) > E1 (1.6%) > NS2 (0.0%) > p7 (0.5%).

Human interactome of HCV proteins is shown in Figure 6. It includes 8 viral and 202 human proteins. FuzDrop analysis revealed that the HCV genome polyprotein has a noticeable LLPS potential, possessing the p_LLPS of 0.7034 and containing 7 droplet-promoting regions (residues 1-97, 96-21, 1206-1217, 1485-1495, 2191-2212, 2304-2418, and 2682-2693). Mature HCV proteins C (p_LLPS = 0.9364, three DPRs, residues 1-28, 36-81, and 96-116) and NS5A (p_LLPS = 0.6757, two DPRs, residues 219-240 and 332-444) are predicted as droplet-drivers, proteins E2 (p_LLPS = 0.1706, one DPR, residues 1-13), serine protease/helicase NS3 (p_LLPS = 0.1873, two DPRs, residues 179-194 and 459-469), and RNA-directed RNA polymerase (p_LLPS = 0.2638, one DPR, residues 262-273) are acpected to serve as droplet-clients, whereas proteins E1 (p_LLPS = 0.1119, no DPRs), viroporin p7 (p_LLPS = 0.1152, no DPRs), protease NS2 (p_LLPS = 0.1031, no DPRs), and NS4A (p_LLPS = 0.0990, no DPRs), and NS4B (p_LLPS = 0.1223, no DPRs) are predicted to be unrelated to the cellular phase separation processes. HS5A of HCV was shown to colocalize with Ras-GTPase-activating protein-binding protein 1 (G3BP1) in the HCV replication complex (RC) [113].

Although HCV RCs, which are virus-induced structures that serve as factories for viral RNA synthesis, are not membrane-less organelles, being instead membrane-associated structures [114], it is tempting to hypothesize that they might represent an example of the 2D LLPS (i.e., LLPS happening in or on the membrane) [115]. This hypothesis is in line with the notion that the viral replication complexes also known as viral inclusion bodies can be considered as biomolecular condensates formed through phase separation [116]. Both C and NS5B were shown to be associated with the processing body (P-body) biogenesis by relocalizing the P-body-specific protein Mov10 to circular structures surrounding cytoplasmic lipid droplets with NS5A and C proteins [117]. It was also speculated that NS5A phosphorylation might affect assembly and disassembly of stress granules and other membrane-less organelles [102].

3.1.4. Hepatitis D Virus

HDV, being a sub-viral agent whose replication and spread depends on the presence of HBV in the infected cells [11,17,18], is an enveloped virus with the negative-sense, single-stranded, covalently closed circular RNA genome classified as the genus Deltavirus, within the realm Ribozyviria. Virus is characterized by considerable genetic variability, existing as eight major HDV genotypes [118] showing specific geographic distributions and being linked to distinctive clinical outcomes and disease progression [119]. Genome of this virus encodes the only virus protein, the hepatitis delta antigen (HDAg). However, its envelope containing host phospholipids has three proteins taken from the HBV, L-HBsAg, M-HBsAg, and S-HBsAg. The delta antigen exists in two forms generated from the same viral genome, small (S-HDAg, 195 residue-long) and large (L-HDAg, 213 residue-long) [120], with different functions, with the small delta antigen being crucial for viral RNA replication, and with the large delta antigen acting as a potent trans-dominant inhibitor for HDV replication [121,122] being primarily involved in the assembly of new viral particles [11,123]. The difference between the S-HDAg and L-HDAg is the presence of 19 extra residues added to the C-terminal region of L-HDAg via an RNA editing mechanism taking place during the late stages of the replication cycle and causing conversion of an amber stop codon into a tryptophan codon thereby leading to the addition of 57 nucleotides to the ORF and consequently 19-residue extension to the protein [11,124,125].

Figure 7 represents RIDAO-generated disorder profiles of S-HDAg and L-HDAg and shows that both forms of delta antigen are highly disordered. The proteins are predicted to have several disorder-based binding sites, molecular recognition features, MoRFs, which are disordered segments that become ordered at interaction with binding partners. In S-HDAg, these MoRFs are regions comprising residues 40-62, 67-96, 106-123, and 143-195. In L-HDAg, C-terminal MoRF is extended to cover residues 143-210. Therefore, despite of (or due to) their highly disordered status, both proteins contain important functional regions, such as the dimerization coiled-coil region (residues 12-60), nuclear localization signal (residues 66-75), two RNA-binding regions (residues 97-107 and 136-146), and, in S-HDAg, region involved in interaction with host RNA polymerase II complex (residues 130-195). C-terminal region of L-HDAg contains a prenylation site required for viral particle assembly with HBsAg [126,127].

Both proteins are predicted to be characterized very high probability of spontaneous liquid-liquid phase separation, p_LLPS of 0.9914 and 0.9927 for S-HDAg and L-HDAg, respectively. Therefore, it is tempting to hypothesize that virus-driven LLPS could be related to the mechanism of the HDV pathogenicity. In line with this hypothesis, it was reported that HDAg and viral RNA accumulate in the nucleus of infected cell in the form of massive RNA-protein hubs that often occupy the majority of the nuclei [128]. It was emphasized that the HDV genome is exclusively enriched in G-quadruplex forming sequences, with density of these sequences being more than four times greater than that of the human genome [129]. Since in HBV, the G-quadruplex forming sequences were linked to the efficient liquid-liquid phase separation associated with virus replication [108], it is likely that similar LLPS mechanism can also be applicable to HDV as well.

Figure 8 shows that S-HDAg has more host protein partners than L-HDAg does (10 vs. 3). Overall, there are 11 human proteins that can interact by the HDV proteins.

3.1.5. Hepatitis E Virus

The HEV belongs to the genus Orthohepevirus from the Hepeviridae family. Its genome is a single-stranded, positive-sense RNA that contains three open reading frames (ORFs), ORF1, ORF2, and ORF3 [130,131]. Of seven HEV genotypes infecting various animal species, HEV-1, HEV-2, HEV-3, HEV-4 [132], and HEV-7 [133] are able to infect humans. ORF1 encodes a non-structural polyprotein pORF1, which is a 1,693-residue-long multifunctional protein containing a methyltransferase domain (residues 1-240), Y-domain (residues 241-439), which is involved in gene regulation and membrane binding in intracellular replication complexes [134], a papain-like cysteine protease (residues 442-509), zinc-binding region (residues 510-691), a proline-rich disordered hypervariable hinge region (residues 712-770), a macro domain also known as X-domain (residues 775-921), a ⁺RNA virus helicase ATP-binding domain (residues 934-1082), an NTPase/helicase domain (residues 960-1204), a C-terminal ⁺RNA virus helicase domain (residues 1083-1216), and an RNA-directed RNA polymerase (residues 1207-1693) containing RdRp catalytic domain (residues 1454-1565) [135,136]. ORF2 encodes a pro-secreted protein pORF2, which is considered as a canonical pORG2 form containing a signal peptide (residues 1-19) and a pro-peptide (residues 20-33), removal of which generate secreted protein ORF2s (ORF2g). Because of alternative initiation, a capsid protein (ORF2c or ORF2i) is synthesized as well, which is different from the canonical form by missing residues 1-15. pORG2 contains an Arginine-Rich Motif (ARM, residues 28-33) acting as a nuclear localization signal, a region responsible for particle formation (residues 368-394) and an oligomerization region (residues 585-610). The major role of secreted protein ORF2s is the inhibition of the host antibody-mediated neutralization, whereas capsid protein ORFc is responsible for the formation of an icosahedral capsid. ORF3 encodes the ORF3 multifunctional phosphoprotein also known as Vp13 with crucial roles in counteracting host innate immunity, as well as virion morphogenesis and egress [137,138,139,140]. Despite its small size (114 residues), pORF3 contains a multitude of functional regions requires for membrane association (residues 1-28), induction of host signal regulator protein alpha (SIRPA) expression required for the type I interferon down-regulation [141], interaction with host HPX (residues 28-68), interaction with the the ORP2c (residues 48-72), homodimerization and interaction with host AMBP/bikunin (residues 72-114), as well as interaction with host FYN, GRB2, HCK, PIK3R3, and SRC (residues 95-104), and PSAP/PTAP motif required for the virion release from infected cells (residues 96-99) [139].

Figure 9 illustrate the intrinsic disorder status of these proteins and shows that all of them are expected to contain noticeable levels of disorder, being classified as moderately or highly disordered. In fact, pORF1 is characterized by ADS_{PONDR VSL2} of 0.30±0.19 and PPIDR_{PONDR VSL2} of 14.9%, being, therefore, moderately disordered, whereas both pORF2 and pORF3 are highly disordered, as their ADS_{PONDR VSL2} values are 0.47±0.25 and 0.74±0.18, and PPIDR_{PONDR VSL2} values are 36.5% and 87.7%, respectively.

An important feature of pORF2 and pORF3 is that they are encoded by the bicistronic mRNA, where ORF3 substantially overlaps the 5′ coding sequence of ORF2 in a separate reading frame [131]. Therefore, it is not surprising that both of these proteins are highly disordered, as this is in line with well-established fact that intrinsic disorder is commonly found in viral proteins produced from overlapping reading frames [45,48,142,143,144].

An obvious consequence of high intrinsic disorder status of HEV proteins is their aforementioned multifunctionality and binding promiscuity. This is illustrated by Figure 10 representing PPI network of HEV and human proteins and showing that the most disordered pORF3 is highly connected.

FuzDrop-based analysis of the LLPS predisposition of HEV proteins revealed that the pro-secreted protein pORF2 can act as droplet driver (p_LLPS = 0.7344, three DPRs, residues 1-53, 57-126, and 444-465), whereas both pORF1 (p_LLPS = 0.2675, four DPRs, residues 488-501, 633-643, 672-694, and 711-773) and pORF3 (p_LLPS = 0.4066, one DPR, residues 85-114) are expected to be droplet clients. Although involvement of LLPS in the HEV life cycle was not discussed in the literature, it is very likely that similar to other hepatotropic viruses (as well as viruses in general), HEV might use phase separation either for replication or for fighting against host immune response. Clearly, further analysis of these possibilities is needed.

3.2. Intrinsic Disorder Status of Host Proteins Interacting with Hepatotropic Viruses

It is known that host cells use numerous advantages of IDPs and IDRs to fight “flexible” invaders, viruses, and to successfully overcome the viral invasion [52]. Therefore, we decided to check the intrinsic disorder status of human proteins involved in interaction with different hepatitis viruses. To be consistent with selection of human proteins for such analysis, we used datasets generated by STRING_Virus. The corresponding PPI networks were shown and briefly described in the corresponding sections above. We present results of the analysis of global intrinsic disorder predisposition of host proteins interacting with different hepatotropic viruses below.

Figure 11 and Figure 12 show results of the evaluation of global intrinsic disorder propensity of host proteins in all analyzed datasets; i.e., 486, 78, 202, 11, and 9 host proteins interacting with HAV, HBV, HCV, HDV, and HEV proteins, respectively. As a reminder, STRING_Virus generated networks for human proteins interacting with HBV, HCV, HDV, and HEV, whereas HAV-host interactome represents proteins from Pan troglodytes, despite our request to show HAV-human interactome. As we already indicated, since chimpanzee (Pan troglodytes) and human (Homo sapiens) genomes share ~98% sequence similarity [86], the derived set of the chimpanzee proteins interacting with HAV can be used for the purposes of this study. Figure 11A represents the outputs of this analysis as the PONDR^® VSL2 Score vs. PONDR^® VSL2 (%) plot, which can be used for the classification of proteins based on their Average Disorder Score (ADS) and Percentage of Predicted Disordered Residues (PPDR) values. Based on the outputs of these analysis, the proteins were classified as highly ordered if they had a PPDR of less than 10% and/or an ADS of less than 0.15. Proteins with 10% ≤ PPIDR < 30% and/or 0.15 ≤ ADS < 0.5 were considered moderately disordered. Proteins with PPDR ≥ 30% and ADS of ≥ 0.5 were defined as highly disordered. This classification is in line with the practice accepted in the field [83]. To make results of this analysis more visual, the ADS vs. PPIDR plot has differently colored blocks, where regions containing mostly ordered, moderately disordered (), or mostly disordered proteins are shown by blue/light blue, pink/light pink, and color, respectively. If the two parameters agree, the corresponding part of the background is dark (blue or pink), whereas light blue and light pink reflect regions, where the ADS and PPIDR disagree with each other. Figure 11A shows that host proteins interacting with hepatotropic viruses are characterized by noticeable variation in their intrinsic disorder status.

Another accepted way of global classification of protein intrinsic disorder utilizes position of query proteins within the CH-CDF phase space [145,146] originating from the combined consideration of two binary predictors, the charge-hydropathy (CH) plot and cumulative distribution function (CDF) analysis. CH-plot classifies proteins based on their absolute mean charge and mean normalized hydrophobicity, as highly disordered proteins are often characterized by low hydropobicity and high net charge [147,148]. The CDF describes the cumulative frequency of disordered residues along the length of a given protein, and position of the resulting CDF curve relative to the order-disorder boundary is used for protein classification [147]. Combination of the outputs of these binary predictors produces CH-CDF classifier that allows more detailed characterization of the disorder status of query proteins based on the agreement between the predictors generating four quadrants [145,146]. In the resulting ΔCH-ΔCDF plot, Quadrant 1 contains proteins that are likely to be structured (both predictors agree, CH^ordered and CDF^ordered). Quadrant 2 includes proteins that are either molten globular or hybrid; i.e., proteins that are compact yet lack a distinctive 3D structure or contain noticeable levels of ordered and disordered residues (predictors disagree, CDF^disordered but CH^ordered). Quadrant 3 contains highly disordered proteins (both predictors agree, CH^disordered and CDF^disordered), whereas Quadrant 4 contains proteins that are predicted to be disordered according to the CH-plot yet ordered according to the CDF-plot [145,146]. Figure 11B represents the results of global disorder analysis of the host proteins interacting with different hepatitis viruses in the form of the ∆CH-∆CDF plot and shows that these proteins are spread between all four quadrants, further reflecting variability of their global disorderedness.

Quantification of distribution of host proteins within the different areas the ADS_{PONDR VSL2} vs. PPID_{RPONDR VSL2} plot and the ΔCH-ΔCDF plot is illustrated by Figure 12 showing contents of different areas of these plots. Figure 11A and Figure 12A show that the vast majority of host proteins are predicted as moderately or highly disordered (none of them is found in the blue area, and less than 10% of them are located within the cyan area). Most of the host proteins are within the pink or light pink areas, being therefore predicted as moderately disordered. At least 30% of host proteins interacting with hepatitis viruses are expected to be highly disordered, as they are located within the red area of the ADS_{PONDR VSL2} vs. PPID_{RPONDR VSL2} plot. In comparison with the entire human proteome, more host proteins interacting with hepatitis viruses were predicted as moderately disordered.

Figure 12. Quantification of host proteins interacting with various hepatitis viruses based on their positions within the ADS_{PONDR VSL2} vs. PPIDR_{PONDR VSL2} plot (A) and ΔCH-ΔCDF plot (B). Analogous data for the entire human proteome are shown for comparison as well (black bars). Corresponding data are taken from [149].

Figure 12. Quantification of host proteins interacting with various hepatitis viruses based on their positions within the ADS_{PONDR VSL2} vs. PPIDR_{PONDR VSL2} plot (A) and ΔCH-ΔCDF plot (B). Analogous data for the entire human proteome are shown for comparison as well (black bars). Corresponding data are taken from [149].

Figure 11B and Figure 12B show that many host proteins interacting with various hepatitis viruses are expected to be mostly ordered (~60% are located within the Q1). All HEV-interacting proteins are predicted as either ordered/compact or molten globule/hybrids. Altogether, their distribution within the ΔCH-ΔCDF plot is not too different from the distribution of entire human proteome.

3.3. Intrinsic Disorder Status of Host Proteins Shared by Different Hepatotropic Viruses

Since co-infection and super-infection reflects simultaneous presence of more than one type of virus in the infected cells, it is logical to assume that the co-infecting viruses might interact with some common host proteins. Analysis of the host interactomes of HAV, HBV, HCV, HDV, and HEV revealed the presence of 33 shared proteins. Interestingly, 32 of these shared host proteins are found in the HBV and HCV interactomes, whereas interactomes of HAV, HDV, and HEV contains 11, 5, and 1 shared host proteins, respectively. These host proteins are listed in Table 1, which also contains information on their physico-chemical properties, disorder and LLPS propensities.

We also looked if these shared host proteins are involved in interaction with each other using the STRING platform. Figure 13A summarize the results of this analysis and shows that most of these proteins do form a dense PPI network. In this network, 28 human proteins are connected by 214 interactions. Four proteins, GDI1, CDSN, TBXT, and FUT1 are loners (i.e., do not interact with other members of the network), whereas STRING does not have information for VEGFA. The average node degree of this network is 15.3, indicating, than on average, each member of network interacts with 15 partners. The average local clustering coefficient of this network, which is a measure of the degree to which nodes in a network tend to cluster together, is 0.859, indicating that the network has densely connected clusters (note that value of 1 represents the highest possible level of clustering within a network, and if every node has a clustering coefficient of 1, then immediate neighbors of each node are completely connected to one another, resembling a clique). PPI enrichment p-value of PPI network connecting shared host proteins is 1.54×10^-13, indicating that this network has significantly more interactions among themselves than what would be expected for a set of proteins of the same size and degree distribution randomly drawn from the genome, and thereby suggesting that the proteins in this network can be biologically connected, as a group.

We also used STRING-embedded routing to visualize the functional enrichment of this network based lowest values of false discovery rate (which is the measure describing the significance of the enrichment using p-values corrected for multiple testing within each category using the Benjamini–Hochberg procedure [150]. Results of this analysis are shown in Figure 13B-G as plots representing enriched GO biological processes (Figure 13B), GO molecular functions (Figure 13C), GO cellular components (Figure 13D), disease-gene associations (Figure 13E), KEGG pathways (Figure 13F), and reactome pathways (Figure 13G). From the viewpoint of biological process enrichment, the most members of this group of proteins are involved in cell surface receptor signaling pathway, animal organ development, cellular response to chemical stimulus, positive regulation of cell population proliferation, regulation of cell population proliferation, and regulation of cell death (Figure 13B). Their most common molecular function is protein binding (Figure 13C). Most often they are found in protein-containing complexes, endomembrane system, and intracellular organelle lumen (Figure 13D). They are most commonly associated with liver diseases, gastrointestinal system diseases, gastrointestinal system cancer, intestinal diseases, disease of cellular proliferation, organ system cancer, and respiratory system diseases (Figure 13E).

Among most enriched KEGG pathways, in which members of this network can be found, are hepatitis B, pathways in cancer, AGE-RAGE signaling pathway in diabetic complications, Epstein-Barr virus infection, measles, human T-cell leukemia virus 1 infection, hepatitis C, and Kaposi sarcoma-associated herpesvirus infection (Figure 13F). Finally, many members of this network are significantly enriched in signaling by interleukins, cytokine signaling in immune system, immune system, signal transduction, and developmental biology (Figure 13G).

Since binding promiscuity and multifunctionality of proteins are frequently associated with intrinsic disorder, at the next stage, we analyzed intrinsic disorder predisposition of shared host proteins. Figure 14A represents results of this analysis and shows none of the shared host proteins was predicted as mostly ordered, indicating that all these proteins were moderately or highly disordered. In fact, among 33 shared host proteins, 12 (36.4%) were characterized by 10% ≤ PPIDR_{PONDR VSL2} < 30% and 0.15 ≤ ADS _{PONDR VSL2} < 0.5, 11 (33/3%) had PPIDR_{PONDR VSL2} ≥ 30% and 0.15 ≤ ADS_{PONDR VSL2} < 0.5, and 10 (30.3%) possessed PPDR_{PONDR VSL2} ≥ 30% and ADS_{PONDR VSL2} ≥ 0.5 (see Figure 14A).

Figure 14B represents an interesting observation on the correlation between the internal node degree (i.e., connectivity of shared host proteins within the PPI network they form according to STRING (see Figure 13A)) and the global node degree of individual proteins (i.e., their interactability in the STRING-generated networks centered at each of these proteins individually). Figure 14B shows that the global interactivity of the shared host proteins measured as their node degree within their individual PPI networks ranged 42 to 469 partners. Furthermore, this global node degree seems to increase exponentially with the increase in the internal node degree, being virtually independent on the internal node degree ranging from 0 to 15 but experiencing dramatic increase at higher internal node degree values.

Figure 14C indicates that both global and internal interactivity of shared host proteins show weak positive correlation with the intrinsic disorder propensity of these proteins. This behavior is not surprising, since all these proteins contain noticeable levels of disorder. Curiously, although the probability of host proteins to be shared by two or three different types of hepatitis viruses is almost independent on the host protein disorder status, two of the three proteins shared by four viruses (CDSN and TBXT) are highly disordered (see Table 1), with CDSN being the most disordered protein in the set. On the other hand, FUT1, which is also shared by for different hepatitis viruses is the most ordered member of the set of shared proteins.

We also used FuzDrop [151] to analyze the predisposition of shared host proteins to undergo spontaneous LLPS and found that 8 of these proteins (CDSN, CREB1, JUN, MYC, VEGFA, TP53, PRDM10, and TBXT) can serve as droplet drivers. There were also six proteins (CD4, SMYD3, HBE1, AFP, PSMB7, and GDI1) that were not related to cellular LLPS processes being characterized by low p_LLPS value and not possessing DPRs. Remaining 19 shared proteins were predicted to act as droplet-clients (see Table 1). Generally, a significant part of cellular processes is known to be determined by the functioning of liquid-droplet-like condensates – membrane-less organelles (MLOs), which are very diverse and commonly found in cytoplasm, nucleus, and mitochondria of various eukaryotic cells [152,153]. Figure 14D shows that there is a strong positive correlation between PPIDR_{PONDR VSL2} and p_LLPS, and that all 8 proteins predicted as droplet drivers (i.e., proteins characterized by p_LLPS ≥ 0.60) are also predicted to be highly disordered. The intracellular LLPS processes, also known as liquid-liquid demixing phase separation, were recognized as important biogenesis drivers of various MLOs [154,155]. These processes are strongly dependent on protein intrinsic disorder [156,157]. In fact, many of the MLO resident proteins are IDPs or contain IDRs, and the formation of all the MLOs analyzed so far relies on IDPs/IDRs, indicating that intrinsic disorder is important for MLO biogenesis [154]. Furthermore, the fact that many shared host proteins potentially associated with various hepatitis co-infections can participate in LLPS is in line with a recent observation that the liquid-liquid phase separation process is emerging as an important cellular mechanism in liver innate immunity [158].

3.4. Functional Intrinsic Disorder in Most Shared Host Proteins

Next, we looked at functionality of intrinsic disorder in 11 proteins shared by four or three different hepatitis viruses. The results of these analyses are presented below, with corresponding sections being ranged based on the intrinsic disorder status of the discussed proteins.

3.4.1. Corneodesmosin (CDSN, UniProt ID: Q15517; PPIDR: 94.52%) Shared by HAV, HBV, HCV, and HDV

Corneodesmosin, being important for the epidermal barrier integrity, is one of the crucial extracellular constituents of corneodesmosomes, which are the major intercellular adhesion structures in the outermost layer of the skin known as stratum corneum and which help maintain the protective barrier of the skin [159]. The stratum corneum is characterized by relative impermeability to water and water soluble substances. These properties are defined by the unique structure of the stratum corneum, which is composed of corneocytes, “mummified” dead and flattened cells containing a cornified envelope that provides extreme individual resistance [160]. In humans, corneodesmosomes are located in the epidermis, the hard palate epithelium, and the inner root sheath of the hair follicles [161]. The crucial role of corneodesmosomes and corneodesmosin in cell adhesion is illustrated by the facts that a complete loss of the CDSN expression is associated with the generalized peeling skin disease, which is an autosomal-recessive ichthyosiform erythroderma characterized by lifelong patchy peeling of the skin [162] and that the mutations the CDSN gene located in the major psoriasis-susceptibility locus (PSORS1) may be responsible for the susceptibility to psoriasis [163]. This is also in line with the observations that the Cdsn gene ablation in mice induced neonatal death as a result of epidermal tearing upon minor mechanical stress and hair-follicle degeneration related to the desmosome dysfunction [164] and that grafting of the Cdsn-deficient skin onto immunodeficient mice resulted in rapid hair loss together with epidermal abnormalities resembling psoriasis [163].

Potential link of corneodesmosin to hepatitits is determined by the fact that there is a potential link between hepatitis and psoriasis. In fact, these two conditions could be linked interacting with each other and exacerbating each other’s symptoms. For example, it was noted that adults with psoriasis have higher rates of the HCV infection compared to those who do not have psoriasis, and HCV-infected patients with moderate-severe psoriasis are characterized by a higher rate of hepatic decompensation [165]. Curiously, a patient with chronic HCV and psoriasis showed remission of psoriasis after receiving antiviral treatment for the HCV infection [166]. Furthermore, hepatitis is known to be manifested by some extrahepatic simproms, such as mixed cryoglobulinemia, lichen planus, and porphyria cutanea tarda reflecting skin involvement [167]. The CDSN interactors from HAV proteome are P2C, P3C, P3D-POL, VP3, and VP2. In HBV-host PPI interactome, CDSN interacts with proteins X and capsid. In the case of HCV, CDSN interacts with core p19, envelope glycoprotein E2, NS4A, NS4B, NS5A, NS5B/RNA-directed RNA polymerase, and serine protease NS3. It also binds both S-HDAh and L-HDAg of HDV.

Despite obvious importance of CDSN for healthy skin, not much is known about structure of this 529-residue-long protein. It is discussed that CDSN is a 52-56 kDa basic phosphoprotein characterized by very high serine and glycine content (27.5% and 16%, respectively), especially within its N- and C-termini of the protein (residues 43-222 and 385-470) [160,164]. Capability of CDSN to act as an adhesive molecule is attributed to the presence of “glycine loops” in this protein, which are formed as a result of the association of interspersed aromatic or aliphatic residues [168]. The presence of such “glycine loops” mediates intermolecular adhesion via the Velcro-type action [168]. The progressive proteolysis of CDSN in the stratum corneum represents another important feature of this protein [160].

Figure 15 represents the results of functional disorder analysis of human CDSN. It is clear that this protein is highly disordered (Figure 15A–C). CDSN utilizes its disordered segments for protein-protein interactions, as evidenced by the presence of 12 molecular recognition features (MoRFs), which are spread throughout its entire sequence (see Figure 15B). The CDSN-centered PPI network includes 113 members linked by 1549 interactions. This protein is characterized by the p_LLPS of 1.00 (see Table 1), and therefore is predicted as droplet-driver, which is a protein that can spontaneously undergo liquid-liquid phase separation.

3.4.2. Myc Proto-Oncogene Protein (MYC, UniProt ID: P01106; PPIDR: 79.74%) Shared by HAV (H2QWQ2), HBV and HCV

Neither MYC (also known as proto-oncogene c-Myc and class E basic helix-loop-helix protein 39 (bHLHe39)) nor its disorder status require special introduction. In fact, since MYC drives growth and proliferation of cells, this transcription factor is considered as critical oncogene involved in the majority of human tumors [169]. The link of MYC to hepatitis is manifested by significantly elevated levels of this protein in liver cell infected with hepatitis viruses [170,171,172,173,174,175]. It was also reported that HCV may accelerate breast cancer progression by increasing circulating levels of c-Myc oncoprotein [176]. MYC is one of the proteins which are crucially related to the transition from cirrhosis to hepatocellular carcinoma in HCV-infected patients [177]. The intrinsic disorder nature of this important transcription factor was discussed in several dedicated studies (e.g., see [169,178,179,180,181,182,183,184,185]).

Figure 16 shows that this 454-residue-long protein is characterized by high levels of intrinsic disorder. It is predicted to have three MoRFs (residues 208-214, 229-355, and 385-394) and p_LLPS of 0.9563. The capability of MYC to undergo LLPS and be involved in formation of biomolecular condensates was reported in several recent studies, e.g. it was found in nuclear puncta induced by the Epstein-Barr virus (EBV) proteins EBNA2 and EBNALP [186], was included to a LLPS-related gene-based risk index associated with HCC [187], as well as to the list of LLPS-associated proteins in rheumatoid arthritis based on the single-cell sequencing and transcriptome analyses [188]. Furthermore, it was pointed out that LLPS of MYC, which is related to the cell fate regulation, its dependent on the disordered domains of this protein [189].

Human MYC forms a very dense PPI network containing 349 members connected by 3,103 interactions (see Figure 16C). Among most enriched biological processes associated with the members of this network are regulation of gene expression, regulation of primary metabolic process, regulation of macromolecule metabolic process, and regulation of cellular metabolic process. Their most enriched molecular functions are binding, protein binding, DNA binding, heterocyclic compound binding, nucleic acid binding, organic cyclic compound binding, transcription factor binding, and transcription regulator activity. They are mostly enriched in the following cellular components: nucleoplasm, nuclear lumen, intracellular organelle lumen, nucleus, intracellular membrane-bounded organelle, chromosome, chromatin, protein-containing complex, intracellular non-membrane-bounded organelle, and transcription regulator complex. The most enriched disease-gene associations are cancer, disease of cellular proliferation, organ system cancer, cell type cancer, carcinoma, gastrointestinal system cancer, nervous system cancer, hepatobiliary system cancer, and hepatobiliary disease. Most enriched KEGG pathways associated with the members of the MYC-centered PPI network are pathways in cancer, micrornas in cancer, prostate cancer, viral carcinogenesis, cell cycle, human t-cell leukemia virus 1 infection, cellular senescence, human papillomavirus infection, FoxO signaling pathway, and hepatitis. Among the most enriched reactome pathways associated with these proteins are generic transcription pathway, RNA polymerase II transcription, gene expression (Transcription), signal transduction, disease, ESR-mediated signaling, signaling by nuclear receptors, estrogen-dependent gene expression, cellular responses to stress, and cellular Senescence.

3.4.3. Cellular Tumor Antigen p53 (TP53, UniProt ID: P04637; PPIDR: 68.11%) Shared by HAV, HBV and HCV

Being one of the most well-studied proteins (e.g., as of January 29, 2025, more 123,500 papers talking about different aspects of p53 were reported in PubMed), cellular tumor antigen p53 (TP53) requires even less introduction than MYC. In fact, this protein is broadly known as a guardian of the genome, mutations in which are involved in ~60% of human cancers [190]. Among functions ascribed to p53 are negative regulator of cell cycle progression via growth arrest [191] and apoptosis [192], as well as a tumor suppressor, a regulator of senescence and differentiation [193], and a guardian of the genome responsible for guiding DNA repair [194]. TP53 gene is expressed as multiple isoforms (with sometimes antagonistic functions) due to alternative splicing, alternative promoter usage and alternative initiation of translation [195]. In one of the comprehensive reviews dedicated to p53 and published in 2016 [196], it was pointed out: “…with its countless biological functions and well-known capability to interact with a myriad of unrelated binding partners, p53 acts as a polyfunctional multibinder, whose functional molecular mechanisms are clearly opposed to the lock-and-key-like functionality described for many globular proteins. This rich functional spectrum of p53 is a reflection of richness and complexity of its structure with multiple proteoforms generated due to the presence of intrinsically disordered regions, numerous posttranslational modifications, and multiple isoforms created by alternative splicing, alternative promoter usage, or alternative initiation of translation, and the ability of p53 to homotetramerize. All these factors play a role in defining the biological multifariousity of this protein. Therefore, for understanding its multifunctionality and roles in carcinogenesis, p53 should be considered through the prism of the proteoform-based protein structure-function continuum, and not in the line of the classical lock-and-key model” [196].

There is a strong link between p53 and pathogenesis of viral hepatitis. For example, p53 is frequently mutated in HCC, which is associated with the HCV and HBV infection [197,198,199,200,201,202,203]. In addition to point mutations, allelic deletions of p53 have been commonly found in human HCCs triggered by hepadnavirus infection [204]. It was also shown that the HBV X protein disrupts the normal p53 function, thereby enhancing the HBV-related carcinogenesis [205,206,207,208,209,210,211,212]. In HCV as well, the core, NS5A, and NS3 proteins were shown to interact with p53 and prevent its correct function [213,214,215,216,217].

In the realm of IDPs, p53 serves as an illustrative example of the importance of intrinsic disorder example for protein functionality and dysfunction [196,218,219]. Figure 16 demonstrate this notion by showing that human TP53 is characterized by highly disordered N- and C-terminal domains, contains multiple MoRFs, numerous PTMs and forms very dense PPI network that includes 470 proteins linked by 5,562 interactions. Most enriched biological processes associated with the members of this network are positive regulation of macromolecule metabolic processes, cellular response to stress, and negative regulation of cellular process. Their most enriched molecular functions are protein binding, transcription factor binding, heterocyclic compound binding, DNA binding, and nucleic acid binding. They are most enriched components of nucleoplasm, nuclear lumen, intracellular organelle lumen, nucleus, intracellular membrane-bounded, organelle, intracellular non-membrane-bounded organelle, chromosome, protein-containing complex, and chromatin. Most enriched disease-gene associations are cancer and disease of cellular proliferation. Most enriched KEGG pathways are pathways in cancer, microRNAs in cancer, p53 signaling pathway, FoxO signaling pathway, cellular senescence, hepatitis B, apoptosis, human T-cell leukemia virus 1 infection, Epstein-Barr virus infection, and viral carcinogenesis.

The predicted high potential of p53 to undergo spontaneous LLPS (p_LLPS = 0.9848) is supported by numerous experimental observations reporting engagement of this protein in formation of nuclear biomolecular condensates [220,221,222,223,224,225,226,227,228,229,230,231]. In the norm, such liquid-like condensates can bind to DNA and perform transcriptional activity, whereas cancer-associated p53 mutants form condensates with increased rigidity and impaired DNA binding ability [221].

Figure 17. Functional disorder analysis of human Cellular tumor antigen p53 (TP53, UniProt ID: P04637). A. Per-residue disorder profile generated by RIDAO. B. Functional disorder profile generated by D²P². Here, the IDR localization predicted by IUPred, PONDR^® VLXT, PONDR^® VSL2, PrDOS, PV2, and ESpritz are shown by 9 differently colored bars on the top of the plot, whereas the blue-green-white bar in the middle of the plot shows the agreement between the outputs of these disorder predictors, with disordered regions by consensus being shown by blue and green. The two lines with colored and numbered bars above the disorder consensus bar show the positions of functional SCOP domains [232,233] predicted using the SUPERFAMILY predictor [234]. Positions of the predicted disorder-based binding sites (MoRF regions) identified by the ANCHOR algorithm are shown by yellow zigzagged bars [235]. Locations of the sites of different posttranslational modifications (PTMs) identified by the PhosphoSitePlus platform [236] are shown at the bottom of the plot by the differently colored circles. C. 3D structural model generated by AlphaFold. D. TP53-centered PPI network generated by STRING using confidence level of 0.815 for minimum required interaction score (permalink https://version-12-0.string-db.org/cgi/network?networkId=b8c6hliT45pI).

Figure 17. Functional disorder analysis of human Cellular tumor antigen p53 (TP53, UniProt ID: P04637). A. Per-residue disorder profile generated by RIDAO. B. Functional disorder profile generated by D²P². Here, the IDR localization predicted by IUPred, PONDR^® VLXT, PONDR^® VSL2, PrDOS, PV2, and ESpritz are shown by 9 differently colored bars on the top of the plot, whereas the blue-green-white bar in the middle of the plot shows the agreement between the outputs of these disorder predictors, with disordered regions by consensus being shown by blue and green. The two lines with colored and numbered bars above the disorder consensus bar show the positions of functional SCOP domains [232,233] predicted using the SUPERFAMILY predictor [234]. Positions of the predicted disorder-based binding sites (MoRF regions) identified by the ANCHOR algorithm are shown by yellow zigzagged bars [235]. Locations of the sites of different posttranslational modifications (PTMs) identified by the PhosphoSitePlus platform [236] are shown at the bottom of the plot by the differently colored circles. C. 3D structural model generated by AlphaFold. D. TP53-centered PPI network generated by STRING using confidence level of 0.815 for minimum required interaction score (permalink https://version-12-0.string-db.org/cgi/network?networkId=b8c6hliT45pI).

3.4.4. PR Domain Zinc Finger Protein 10 (PRDM10, UniProt ID: Q9NQV6; PPIDR: 64.34%) Shared by HAV, HBV and HCV

PR domain zinc finger protein 10 (PRDM10, also known as tristanin) is a 1,147-residue long transcriptional activator belonging to a small group of zinc-finger transcription factors known as positive regulatory domain members (PRDM). Members of this family serve as important transcriptional regulators controlling cell fate specification in various developmental contexts. They are responsible for the transduction of signals that control cell proliferation and differentiation and consequently neoplastic transformation [237]. PRDMs are characterized by the presence of the PR/SET domain (PRDI-BF1 and RIZ1 homology domain closely related to the lysine methyltransferase SET (Suppressor of variegation 3–9, Enhancer of zeste and Trithorax) domain) followed by a set of C₂H₂-type zinc finger repeats mediating the sequence-specific DNA binding, as well as interactions with other protens and RNA [238,239]. Although in humans, there are 17 PRDM genes [240], because of alterative splicing or utilization of alternative promoters, members of the PRDM family are expressed in multiple molecular forms [237,239,240,241]. Often, PRDM members are expressed as PR-plus and PR-minus forms, which are different by the presence or absence of the PR domain and typically have opposite functional roles [242].

Similar to the other PRDM family members, PRDM10 was shown to be mutated in multiple different cancers [242]. Furthermore, PRDM10 mutations are associated with Birt-Hogg-Dube syndrome 2, an autosomal dominant disease characterized by fibrofolliculomas, pulmonary cysts, pneumothoraces, and renal cell carcinomas [243]. Association of PRDM10 with hepatitis is based on the analysis of the aberrant DNA methylation profile associated with HBV infection, where this protein was shown to be important transformation from the normal to chronic infection [244].

In comparison with other members of the PRDM family, this protein remains rather understudied. Human PRDM10 protein contains a degenerated PR/SET domain (residues 208-326), which possibly has lost the methyltransferase activity, followed by 10 C₂H₂-type zinc finger domains (residues 355-377, 530-552, 560-582, 588-610, 616-639, 644-666, 672-695, 727-750, 772-795, and 834-857). Figure 18A–C show that human PRDM is predicted to contain high level of disorder. Since it has ten C₂H₂-type zinc finger domains, its cysteine content (28) is high (2.44% of its sequence are cysteine residues, which is almost 2-fold higher than the average cysteine content in UniProt (1.37%)). Since cysteine is considered as one of the strongest order-promoting residues [245,246,247,248], and since its oxidative state can contribute to protein structural stability, we also checked PRDM10 for the presence of redox-sensitive disordered regions (i.e., regions that undergo order-disorder transition at reduction of disulfide bonds). Figure 18B shows that PRDM contains 3 such regions. Although human PRDM10 is predicted to be characterized by high LLPS potential (p_LLPS = 0.9912), current literature does not have data demonstrating involvement of this protein (and as a matter of fact of any human PRDM family member) in cellular phase separation processes. However, it was reported that Caenorhabditis elegans protein SET-17 with a PRDM9/7-like SET domain localizes to chromatin-associated foci and promotes spermatocyte gene expression, sperm production and fertility [249]. It is tempting to speculate that the LLPS of PRMD10 can be related to its physiological and pathological functions.

Figure 18C shows that PRDM10 forms a network containing 43 members linked by 117 interactions. Members of this network are involved in histone lysine methylation, peptidyl-lysine modification, histone modification, methylation, peptidyl-lysine trimethylation, rhythmic process. Their most enriched molecular functions are histone-lysine N-methyltransferase activity, methyltransferase activity, ion binding, and metal ion binding. The most enriched disease-gene association of these proteins are urinary bladder cancer, hepatocellular carcinoma, hepatobiliary system cancer, urinary system disease, and endocrine gland cancer. They are most enriched in the following KEGG pathways: mitophagy, ErbB signaling pathway, thyroid hormone signaling pathway, hepatitis b, viral carcinogenesis, kaposi sarcoma-associated herpesvirus infection, lysine degradation, platinum drug resistance, EGFR tyrosine kinase inhibitor resistance, and axon guidance.

3.4.5. T-Box Transcription Factor T (TBXT, UniProt ID: O15178; PPIDR: 60.00%) Shared by HAV, HBV, HCV, and HDV

T-box transcription factor T (TBXT, also known as T or brachyury) is a 435-residue-long transcription factor regulating genes required for mesoderm formation and differentiation. TBXT activates gene transcription when bound to a palindromic T site 5’-TTCACACCTAGGTGTGAA-3’ DNA sequence. TBXT plays crucial role at early stages of embryonic development, being exquisitely orchestrated and controlled to transform the nascent dorsal mesoderm into the notochord, the rudimentary axial skeleton [250,251]. TBXT is believed to define the midline of a bilaterian organism and the establishment of the anterior-posterior axis [252]. In the human fetus, TBXT is silenced at ~ 12 weeks of development [253], and in adults is minimally detected in the pituitary gland, thyroid, and testes [254]. However, TBXT became expressed in several cancers of epithelial origin (lung, prostate and breast cancer), promoting tumor progression and epithelial-mesenchymal transition [255]. Furthermore, TBXT is specifically expressed in chordomas [256,257], which are rare malignant tumors originating from the notochordal remnants in the axial bones [258]. Therefore, it is used as one of the most important markers in the diagnosis of chordoma [257]. Mutations in TBXT are associated with neural tube defects, which are congenital malformations of the central nervous system and adjacent structures related to defective neural tube closure during the first trimester of pregnancy [259], as well as with sacral agenesis with vertebral anomalies (SAVA) [260]. In the interactomes of different hepatitis viruses with the host, TBXT/brachyury interacts with P2C, P3C, P3D-POL, VP2, and VP3 of HAV; HBeAG, HBsAg, capsid, and protein X of HBV; core p19, envelope glycoprotein E2, NS4A, NS4B, NS5A, NS5B/RNA-directed RNA polymerase, and serine protease NS3 of HCV; and S-HDAh and L-HDAg of HDV.

TBXT/brachyury contains DNA-binding domain (residues 42-219), structure of which was resolved (e.g., PDB ID: 8FMU, [261]). Figure 19A–C show that both N- and C-terminal regions of this protein are predicted as disordered. There are 9 MoRFs in human TBXT/brachyury, which also has several phosphorylation sites. In line with the predicted high LLPS potential of this protein (p_LLPS = 0.9862), brachyury was shown to form liquid-like transcriptional condensates in nucleus [254].

Figure 19D represents the STRING-generated PPI network centered at human TBXT. This network includes 46 proteins connected by 415 interactions. Among the enriched biological processes associated with the members of this network are regionalization, anterior/posterior pattern specification, embryo development, embryonic morphogenesis, anatomical structure formation involved in morphogenesis, positive regulation of transcription by RNA polymerase II, circulatory system development, gastrulation, cell fate commitment, and embryonic organ development. Their enriched molecular functions are cis-regulatory region sequence-specific DNA binding, RNA polymerase II cis-regulatory region sequence-specific DNA binding, DNA-binding transcription factor, activity, RNA polymerase II-specific, transcription regulator activity, DNA binding, transcription factor binding, RNA polymerase II-specific DNA-binding transcription factor binding, DNA-binding transcription factor binding, DNA-binding transcription activator activity, and RNA polymerase II-specific co-SMAD binding. The enriched cellular components associated with this network are chromosome, chromatin, intracellular non-membrane-bounded organelle, nucleus, transcription regulator complex, RNA polymerase II transcription regulator complex, nucleoplasm, intracellular organelle lumen, activin responsive factor complex, and heteromeric SMAD protein complex. The most enriched disease-gene associations are endocrine system disease, microphthalmia, coloboma, right atrial isomerism, holoprosencephaly, embryonal carcinoma, teratoma, carcinoma, cell type cancer, and syndromic microphthalmia. The most enriched KEGG pathways are signaling pathways regulating pluripotency of stem cells, Hippo signaling pathway, gastric cancer, adherens junction, pathways in cancer, colorectal cancer, hepatocellular carcinoma, TGF-beta signaling pathway, Wnt signaling pathway, and Basal cell carcinoma.

3.4.6. Human Serum Albumin (ALB; UniProt ID: P02768; PPIDR: 32.02%) Shared by HAV, HBV and HCV

Being one of the most studied proteins, human serum albumin (HSA) is the most abundant protein in plasma. It is synthesized in the liver and exported as a single non-glycosylated chain, reaching a blood concentration of about 7×10^-4 M. It is a monomeric multi-domain macromolecule, and the main plasma component determining plasma oncotic pressure (also known as colloid osmotic pressure) [262]. Albumin is also the main regulator of fluid distribution between compartments of the body [263,264,265]. It shows a very high ligand binding ability, serving as a depot and carrier for a wide variety of compounds that may be available in quantities well beyond their solubility in plasma. In fact, the physiological importance of HSA stems from the fact that it is involved in bioregulatory and transport phenomena. It binds various metal ions: Ca²⁺ [266], Cu²⁺ and Ni²⁺ [267,268], Zn²⁺ [269], Mn²⁺, Mg²⁺, Co²⁺, and Cd²⁺ [270], and many others [262]. HSA takes part in transport and storage of different fatty acids [271,272,273]. It also binds bilirubin [274], steroids [275,276], amino acids [277], and many other ligands, usually with hydrophobic moieties [262,278]. Curiously, in human plasma, about 90% of β-amyloid peptide is carried by HSA [279]. This unique property enables HSA to fulfil a fundamental biological role as a universal carrier and reservoir in blood plasma, tissues, and secretions throughout the mammalian body [280]. Being the main carrier for fatty acids, serum albumin is known to affect pharmacokinetics of numerous drugs [281] and serves as a regulator of the metabolic modification of some ligands. It is a valuable biomarker of many diseases, including cancer, rheumatoid arthritis, and ischemia [282,283,284,285].

In relation to subject of this study, low albumin levels serve as a strong predictor of acute liver failure in patients with acute HAV [286], can be a sign of cirrhosis in HCV patients [287,288], and serve as indicator of higher hepatocellular carcinoma recurrence in HBV patients [289]. Albumin can be incorporated into the HBV coat, likely via the disulfide bonding with the hepatitis B surface antigen [290]. Polymerized HSA may play a role in the infection of hepatocytes by hepatitis B virus, as HBsAg particles contain specific binding sites for transglutaminase-cross-linked HSA, suggesting that the albumin polymers may play a role in the attachment of HBV to hepatocytes [291].

Although HSA contains disordered or structurally flexible regions (see Figure 20A,B), which are heavily decorated by various PTMs (Figure 20B), it has a well-defined structure (see Figure 20C). Predicted levels of intrinsic disorder in this protein (30.2%) correlate well with the content of random coil structure in this protein (29.1%) evaluated by Fourier-transform infrared (FTIR) spectroscopy [292]. Prevalence and functionality of intrinsically disordered regions in serum albumins was a subject of a dedicated study [293]. Among various other observations, it was indicated that structural flexibility is required for interaction of serum albumins with their partners [293]. Although HSA cannot undergo LLPS spontaneously (its p_LLPS of 0.1403 is too low), it still can act as a droplet client, since it contains one disorder-promoting region (DRP, residues 210-220).

In line with these considerations, Figure 20D shows that human serum albumin forms a densely connected PPI network, where 241 partners are linked by 3,009 edges. Multifunctionality of the members of this network is reflected in the fact that their most enriched molecular functions include signaling receptor binding, molecular function regulator activity, signaling receptor regulator activity, signaling receptor activator activity, receptor ligand activity, growth factor activity, protein-containing complex binding, protein binding, glycosaminoglycan binding, and growth factor receptor binding. Their most enriched biological processes are response to stimulus, response to stress, regulation of multicellular organismal process, cellular response to chemical stimulus, response to chemical, response to organic substance, biological process involved in interspecies interaction between organisms, regulation of cell population proliferation, defense response, and positive regulation of protein phosphorylation. Among enriched cellular components of these proteins are extracellular space, extracellular region, blood microparticle, extracellular vesicle, extracellular exosome, vesicle, cell surface, secretory granule, platelet alpha granule, and collagen-containing extracellular matrix. The most enriched disease-gene associations of these proteins are disease of anatomical entity, vascular disease, cardiovascular system disease, hematopoietic system disease, respiratory system disease, artery disease, disease by infectious agent, immune system disease and urinary system disease. The most enriched KEGG pathways linked to these proteins are pathways in cancer, PI3K-Akt signaling pathway, Rap1 signaling pathway, melanoma, gastric cancer, tuberculosis, complement and coagulation cascades, MAPK signaling pathway, RAS signaling pathway, and malaria.

3.4.7. T-Cell Surface Glycoprotein CD8 Alpha Chain (CD8A; UniProt ID: P01732; PPIDR: 26.38%) Shared by HAV, HBV and HCV

T-cell surface glycoprotein CD8 (cluster of differentiation 8) alpha chain (CD8A also known as T-lymphocyte differentiation antigen T8/Leu-2) is a 235-residue-long transmembrane protein on the surface of T cells with a crucial role in the immune response by helping T cells to response to pathogens [294]. CD8 acts as a co-receptor for the T-cell receptor. CD8 exists as a homo- or heterodimer of CD8 alpha and beta chains (CD8A and CDB, respectively) encoded by different genes, with three different dimers, CD8AA, CD8AB and CD8BB being reported [295]. CD8A, being primarily expressed on the surface of cytotoxic T cells (in fact, CD8 is a marker for cytotoxic T cell population), can also be found on cortical thymocytes, dendritic cells, and natural killer cells. This protein is responsible for the cell-mediated immune defense and T-cell development [296,297], also acts as a co-receptor for MHC class I molecule:peptide complex [298]. CD8A can serve as a diagnostic and prognosis marker for several diseases, such as inflammatory disorders and tumors, including chronic rhinosinusitis [299], rheumatoid arthritis [300], lung adenocarcinoma [301], and was proposed as a biomarker associated with immunocytes infiltration in hyperoxia-induced bronchopulmonary dysplasia [302]. Furthermore, the CD8A expression levels are positively correlated with the degree of liver inflammation and fibrosis [303], and CD8A was proposed to be used as a potential diagnostic biomarker for HBV-induced liver fibrosis in chronic hepatitis B patients [304].

Being a type I membrane protein, CD8A has four major modules, an N-terminal extracellular immunoglobulin variable (IgV)-like domain (residues 22–135), a highly dynamic proline-rich stalk region known as a hinge region (residues 136-182), a single transmembrane helix (residues 183–203), and a small cytoplasmic region (residues 204–235). IgV-like domain and hinge region together constitute an ectodomain. Structurally, the extracellular IgV-like domain is characterized by a typical IgV fold containing two β-sheets (e.g., PDB ID: 8EW6, [305]). Figure 21A–D show that the extracellular IgV-like domain and the transmembrane region of human CD8A are predicted as mostly ordered, whereas its hinge/stalk region and cytoplasmic C-tail are both expected to be disordered. In line with these observations, based on the results of NMR analysis it was reported that the hinge/stalk of human CD8A represents an intrinsically disordered region characterized by a dynamic exchange that includes proline cis-trans isomerization [306]. Furthermore, CD8A contains several important cysteine residues playing essential structural and functional roles. Since this protein contains 9 cysteine residues, its cysteine content of 3.83% almost 3-fold exceed that of UniProt (1.37%). Therefore, it is expected that it might possess some redox-sensitive disordered regions. Validity of this assumption is illustrated by Figure 21B showing that the C-tail of human CD8A indeed represents a redox-sensitive region. Curiously, this tail also contain phosphorylation sites. Furthermore, a flexible region located within the IgV-like domain (residues 74-88, see Figure 21A–C) corresponds to the functionally important CDR2 loop, which is responsible for interaction with MHC-I, and whose structure was not resolved (it is a region with missing electron density in PDB ID 8EW6 [305]). With p_LLPS of 0.2472 and with presence of one DPR (residues 135-155) that is included into the hinge/stalk region, human CD8A can serve as a droplet-driver.

Figure 21E represents the STRING-generated PPI network centered at human CD8A that includes 263 partners connected by 5,362 interactions. Among the most enriched biological processes of the members of this network are immune system process, immune response, regulation of immune system process, positive regulation of immune system process, leukocyte activation, cell activation, regulation of leukocyte activation, regulation of cell activation, regulation of lymphocyte activation and defense response. Some of their most enriched molecular functions are signaling receptor binding, signaling receptor activity, cytokine receptor binding, cytokine activity, immune receptor activity, protein binding, transmembrane signaling receptor activity, cytokine receptor activity, antigen binding, and cytokine binding.

These proteins are most enriched in the following cellular components: cell surface, external side of plasma membrane, side of membrane, plasma membrane, cell periphery, intrinsic component of membrane, integral component of membrane, intrinsic component of plasma membrane, integral component of plasma membrane, and extracellular region. Their most enriched disease-gene associations are immune system disease, primary immunodeficiency disease, autoimmune disease, autoimmune disease of musculoskeletal system, lymphatic system disease, hematopoietic system disease, disease of anatomical entity, hematologic cancer, disease by infectious agent, and organ system cancer. Most enriched KEGG pathways associated with these proteins are cytokine-cytokine receptor interaction, viral protein interaction with cytokine and cytokine receptor, hematopoietic cell lineage, natural killer cell mediated cytotoxicity, cell adhesion molecules, inflammatory bowel disease, JAK-STAT signaling pathway, Epstein-Barr virus infection, Th17 cell differentiation, and measles.

3.4.8. α-Fetoprotein (AFP; UniProt ID: P02771; PPIDR: 20.53%) Shared by HAV, HBV and HCV

α-Fetoprotein (AFP), a 590-residue-long glycoprotein (contains 4.5% carbohydrates), being a major protein of the embryonic plasma [307,308,309] produced by the yolk sac and the fetal liver during fetal development and showing high sequence and structural similarity to HSA [310,311], is commonly considered as an embryonic albumin or embryonic α-globulin. In fact, similar to HSA, AFP serves as a major carrier of bilirubin, dioxin, dyes, fatty acids, flavonoids, heavy metals, phytoestrogens, retinoids, steroids, and various drugs [310,311,312,313]. Although in fetus, AFP is present at a high concentration of 1-10 mg/ml, its content decreases abruptly soon after the birth, as only a trace amount of this protein (typically <10 ng/mL) can be detected by the end of second month postpartum, and AFP is replaced by HSA as the major serum protein [307,309,310,314]. However, being an oncofetal protein, AFP reemerges under several pathological conditions, and increased levels of this protein in the plasma of adults is the hallmark of the development of hepatocellular carcinoma (HCC) and other malignant tumors [315,316,317,318] including germ cell tumors of the testis [319], acinar cell carcinoma of the pancreas [320], intracranial yolk sac tumor [321], and many others. Furthermore, variation in the AFP levels during pregnancy are related to various fetal abnormalities, such as Down’s syndrome, as well as open neural tube defects, such as spina bifida [310,322,323,324,325]. Decrease in serum AFP levels serves as a predictor of poor prognosis of acute hepatic failure in patients with chronic hepatitis B [326]. Importantly, changes in the AFP levels have intricate connections to HCV and HBV infections as well. In fact, AFP levels were shown to be elevated in some patients with chronic viral hepatitis and cirrhosis who do not have HCC [327,328,329,330,331].

Figure 22A–C show that human AFP is a mostly ordered protein that possesses several disordered/flexible regions. This protein was shown to be characterized by the extremely high conformational stability, which was attributed to the large number of disulfide bonds (this protein, being 590 amino acid residues long, contains 15 S–S bridges) [311]. Therefore, it is not surprising that it has 4 redox-sensitive disordered regions (see Figure 22B). AFP has several PTM sites. In addition to shown in Figure 22C three phosphorylation and acetylation sites, AFP can be phosphorylated at Ser111, Ser115, Ser117, Ser344, Ser444, and Ser445 by FAM20C, which is a kinase that phosphorylates S-x-E/pS motifs in the majority of secreted proteins [332]. Furthermore, AFP is glycosylated at Asn251 [333,334], and is characterized by molecular microheterogeneity determined by differences in the attached carbohydrate moiety [335,336]

Despite its obvious functional and diagnostic importance and its remarkable similarity to HSA, no structural information about AFP was available until quite recently. In fact, the first 3D structure of this protein resolved by cryo-electron microscopy (cryo-EM) was reported in 2023 (PDB ID: 7YIM, [337]) (see Figure 22D). AFP is characterized by the V-like (or a heart-like) shape with three well-defined domains (domains I, II, and III), with each domain consisting of ~195 residues with 4, 5, and 6 disulfide bonds being positioned in domains I, II and II, respectively [338]. Since human AFP does not have DPRs and shows low p_LLPS of 0.1510, this protein is not related to the cellular LLPS processes.

Besides being a carrier, AFP has several other functions, possessing immunosuppressive activity and playing a role in regulation of cell proliferation [313], and, therefore, can interact with multiple partner proteins. This is illustrated by Figure 22E presenting a PPI network centered at AFP containing 239 proteins linked by 6,092 interactions. This network is characterized by the average node degree of 51. Members of this network are engaged in various biological processes, such as response to stimulus, response to organic substance, cellular response to chemical stimulus, response to chemical, regulation of multicellular organismal process, animal organ development, negative regulation of biological process, multicellular organismal process, positive regulation of metabolic process, and positive regulation of macromolecule metabolic process. Their most enriched molecular functions are protein binding, signaling receptor binding, identical protein binding, binding, signaling receptor regulator activity, signaling receptor activator activity, receptor ligand activity, growth factor activity, molecular function regulator activity, and protein-containing complex binding. They are enriched within the following cellular components: extracellular region, extracellular space, blood microparticle, extracellular exosome, cell surface, vesicle, platelet alpha granule lumen, platelet alpha granule, endoplasmic reticulum lumen, and vesicle lumen. Among the most enriched disease-gene associations of the members of this network are disease of cellular proliferation, cancer, organ system cancer, cell type cancer, hepatobiliary disease, gastrointestinal system disease, disease of anatomical entity, carcinoma, reproductive system disease, gastrointestinal system cancer, reproductive organ cancer, endocrine gland cancer, female reproductive system disease, and liver disease. These proteins are most enriched in the following KEGG pathways: pathways in cancer, PI3K-Akt signaling pathway, proteoglycans in cancer, gastric cancer, melanoma, breast cancer, MAPK signaling pathway, colorectal cancer, FoxO signaling pathway, maturity onset diabetes of the young, focal adhesion, hepatitis B, EGFR tyrosine kinase inhibitor resistance, AGE-RAGE signaling pathway in diabetic complications, and toxoplasmosis. The most enriched Reactome pathways associated with these proteins are regulation of insulin-like growth factor (IGF) transport and uptake by insulin-like growth factor binding proteins (IFGBPs), post-translational protein phosphorylation, hemostasis, platelet degranulation, platelet activation, signaling and aggregation, interleukin-4 and Interleukin-13 signaling, cell surface interactions at the vascular wall, immune system, metabolism of proteins, signaling by receptor tyrosine kinases, developmental biology, diseases of signal transduction by growth factor receptors and second messengers, cytokine signaling in immune system, negative regulation of the PI3K/AKT network, and extracellular matrix organization.

3.4.9. Proteasome Subunit Beta Type-7 (PSMB7; UniProt ID: Q99436; PPIDR: 17.33%) Shared by HAV, HBV and HCV

Proteasome subunit beta type-7 (PSMB7, also known as macropain chain Z, multicatalytic endopeptidase complex chain Z, proteasome subunit Z, and proteasome subunit beta-2) is a 277-residue-long component of the 20S core proteasome complex possessing a trypsin-like activity. PSMB7, being a member of the proteasome B-type family, also known as the T1B family, is one the essential subunits of the 20S core proteasome, which is composed of four rings of 28 non-identical subunits, with two rings being composed of seven alpha subunits (alpha-1 to -7) and two rings being composed of seven beta subunits (beta-1 to -7). It was reported that PSMB7 can be phosphorylated, and phosphorylation/dephosphorylation of this subunit can be related to the regulation of the adjacent beta-1 subunit [339], which interacts with PSMB7 and has the peptidylglutamyl-peptide hydrolyzing (PGPH) activity [340]. It was reported also that the levels of the phosphorylated PSMB7 were noticeable reduced in the proteasomes isolated from tumor cells as compared to normal cells lines [339]. PSMB7 is overexpressed in colorectal carcinomas [341], with elevated levels of both PSMB7 protein as well as PSMB7 mRNA showing associations with lower overall survival [342]. In breast cancer, overexpression of the PSMB7 gene was shown to be associated with poor prognostic outcome and with anthracycline resistance [343]. PSMB7 was one of the differently expressed proteins in an experimental mesangial proliferative nephritis model [344]. In multiple myeloma, PSMB7 was one of the strongest bortezomib sensitizers; i.e., genes, silencing of which is not directly cytotoxic but synergistically potentiate the growth inhibitory effects of bortezomib [345]. Furthermore, PSMB7 was recognized as a key gene involved in the development of multiple myeloma and resistance to bortezomib [346]. In HCC, PSMB7 was found as one of the proteins involved in 5-fluorouracil resistance [347]. Based on the transcriptomic analyses of thousands of samples from The Cancer Genome Atlas it was concluded that the PSMB7 expression is increased in most cancer types [348]. In Alzheimer’s disease (AD), expression of PSMB7 was shown to be significantly decreased compared to the control group, with more noticeable decrease being associated with the high Braak stages [349]. Inherited and/or de novo loss-of-function mutations in the PSMB7 gene are linked to proteasome-associated autoinflammatory syndromes (PRAAS) [350].

Analysis of the mechanisms behind the interferon lambda (IFN-λ)-induced inhibition of the HBV replication in HepG2.2.15 cells revealed that PSMB7 was one of the proteins up-regulated by IFN-λ3 treatment [351]. Furthermore, hepatitis B virus X protein was shown to interact with PSMB7, with this interaction being functionally important in the pleiotropic effect of this viral protein [352,353,354], and affecting HBV replication through the proteasome-dependent pathway [355]. PSMB7 is one of the proteins involved in interaction with the HCV proteins [356].

Figure 23 shows that although human PSMB7 is expected to be mostly ordered, it has several disordered or flexible regions, including N- and C-terminal regions. In agreement with these observations, in a cryo-EM resolved structures of 26S proteasome (e.g., PDB ID: 5LN3 [357]; 5L4G [358]) two regions of PSMB7 corresponding to its N- and C-termini (residues 1-43 and 264-277) were unmodeled, indicating their highly flexible nature. Human PSMB7 has multiple phosphorylation and ubiquitination sites (Figure 23B). With low p_LLPS of 0.1497 and lack of DPRs, human PSMB7 is not expected to be related to the cellular LLPS processes.

Figure 23D shows that human PSMB7 forms a very dense PPI network with average node degree of 97.1 that contains 295 proteins connected via 14,323 interactions. Members of this network are most enriched in biological processes, such as proteolysis involved in protein catabolic process, protein catabolic process, modification-dependent protein catabolic process, proteasomal protein catabolic process, ubiquitin-dependent protein catabolic process, proteasome-mediated ubiquitin-dependent protein catabolic process, macromolecule catabolic process, cellular macromolecule catabolic process, proteolysis, and cellular macromolecule metabolic process. Their most enriched molecular functions are ubiquitin-like protein ligase binding, ubiquitin protein ligase binding, enzyme binding, protein binding, structural constituent of ribosome, proteasome binding, kinase binding, protein kinase binding, threonine-type endopeptidase activity, and metal-dependent deubiquitinase activity. Most enriched disease-gene associations of these proteins are nervous system cancer, retinal cancer, sensory system cancer, cancer, infratentorial cancer, medulloblastoma, brain cancer, central nervous system cancer, basal cell carcinoma, and organ system cancer. These proteins are most enriched in the following KEGG pathways: proteasome, ubiquitin mediated proteolysis, Epstein-Barr virus infection, Parkinson disease, spinocerebellar ataxia, prion disease, amyotrophic lateral sclerosis, Alzheimer disease, cell cycle, and pathways in cancer.

3.4.10. RAB GDP Dissociation Inhibitor Alpha (GDI1; UniProt ID: P31150; PPIDR: 17.23%) Shared by HAV, HBV, HCV, and HDV

Guanosine diphosphate dissociation inhibitor 1 (GDI1, also known as RAB GDP dissociation inhibitor alpha, RAB GDI alpha, oligophrenin-2, and protein XAP-4) is a 447-residue-long regulator of GDP/GTP exchange if most RAS-related proteins in brain (RAB proteins). RAB proteins constitute a family of small GTPases (G proteins) with a number of crucial roles in regulating intracellular membrane trafficking in eukaryotic cells [359]. GD11 inhibits the GDP dissociation from and GTP binding to RAB proteins. It can promote dissociation of RAB1A, RAB3A, RAB5A, and RAB10 from membranes [360]. GDI1 is one of the proteins involved in nonspecific X-linked mental retardation [361]. Mutations in GDI1 can disturb vesicle trafficking and thereby act as one of the causes of the synaptic vesicle cycling (SVC) disorders [362]. Other pathological conditions associated with distorted GDI1 include Alzheimer’s disease [363], colorectal cancer [364], isolated adrenocorticotropin deficiency (IAD) [365], and oligodendrogliomas [366]. In HCC, loss of GDI1 was shown to serve as a prognostic factor, being closely related to long-term survival of patients with HCC [367,368]. In patients with chronic HCV infection, GDI1 was one of the proteins whose levels were increased in the circulating extracellular vesicles (EV) in comparison to the EVs from healthy donors [369].

No actual structural information is currently available for human GDI1. Figure 24A–C show that it is predicted as mostly ordered protein with several disordered segments. Despite its mostly ordered nature, GDI1 has multiple PTM sites (see Figure 24B). This protein is not related to cellular LLPS processes, since it does not have DPRs and shows p_LLPS of 0.1538. In GDI1-centered PPI network, there are 163 protein connected by 1,736 interactions. Members of this network are entiched in various biological processes, such as vesicle-mediated transport, cellular localization, intracellular transport, establishment of localization in cell, establishment of localization, transport, localization, protein transport, establishment of protein localization, and nitrogen compound transport. Their most enriched molecular function are GTPase activity, GTP binding, nucleoside-triphosphatase activity, purine ribonucleoside triphosphate binding, purine ribonucleotide binding, nucleotide binding, guanyl-nucleotide exchange factor activity, GTPase regulator activity, GDP binding, and small GTPase binding. The most enriched cellular components associated with these proteins are cytoplasmic vesicle, vesicle, endosome, cytosol, Golgi apparatus, endomembrane system, endocytic vesicle, bounding membrane of organelle, cytoplasmic vesicle membrane, and TRAPP complex. Endocytosis, salmonella infection, vasopressin-regulated water reabsorption, bacterial invasion of epithelial cells, synaptic vesicle cycle, AMPK signaling pathway, tuberculosis, and autophagy are among the most enriched KEGG pathways associated with these proteins. Some of the most enriched reactome pathways linked to these proteins are RAB regulation of trafficking, RAB GEFs exchange GTP for GDP on RABs, membrane trafficking, RAB geranylgeranylation, post-translational protein modification, TBC/RABGAPs, COPII-mediated vesicle transport, intra-Golgi and retrograde Golgi-to-ER traffic, retrograde transport at the trans-Golgi-network, and intra-Golgi traffic. Some of the most enriched disease-gene associations are X-linked monogenic disease, intellectual disability, developmental disorder of mental health, disease of mental health, non-syndromic X-linked intellectual disability, Warburg micro syndrome 1, Warburg micro syndrome 2, Warburg micro syndrome 3, Martsolf syndrome, and Rett syndrome.

3.4.11. Galactoside Alpha-(1,2)-Fucosyltransferase 1 (FUT1; UniProt ID: P19526; PPIDR: 10.14%) Shared by HAV, HBV, HCV, and HDV

Galactoside alpha-(1,2)-fucosyltransferase 1 (FUT1, also known as alpha(1,2)FT 1, blood group H alpha 2-fucosyltransferase, fucosyltransferase 1, GDP-L-fucose:beta-D-galactoside 2-alpha-L-fucosyltransferase 1, and type 1 galactoside alpha-(1,2)-fucosyltransferase FUT1) is a 465-residue-long enzyme catalyzing the L-fucose transfer from a guanosine diphosphate-beta-L-fucose to the terminal galactose residue of glycoconjugates through an alpha(1,2) linkage [370]. This results in the synthesis of the H antigen, a precursor to the A and B antigens, which determine a person’s ABO blood type [370]. Several mutations in FUT1 are linked to para-Bombay phenotype, which is a rare blood type that occurs when red blood cells lack the H antigen [371,372].

Changes in FUT expression were shown to have a number of significant roles in cancer development and malignancy [373], with FUT1 being highly expressed in various types of cancers [374]. For example, in ovarian carcinoma cells, overexpression of FUT1 positively regulates 215 genes [375]. In breast cancer, FUT1-driven fucosylation of the lysosome-associated membrane proteins 1 and 2 (LAMP-1 and LAMP-2) leads to the formation of the alpha1,2-fucosylated Lewis Y (LeY) antigens, whose levels were substantial in less invasive ER+/PR+/HER- breast cancer cells (MCF-7 and T47D) but negligible in highly invasive triple-negative MDA-MB-231 cells [374]. FUT1 downregulation resulted in the perinuclear localization of LAMP-1 and LAMP-2 and correlated with the increased rate of autophagic flux by decreasing mTOR signaling and increasing autolysosome formation [374].

High FUT1 expression was significantly associated with HCC advanced stages and poor outcome [376]. FUT1, being highly upregulated in acute HCV infection, was shown to regulate and facilitate HCV replication in hepatocytes [377]. HCV infection increased the expression of fucosylated epitopes, resulting in the reorganization of the glycocalyx (which is a thin, carbohydrate-rich layer covering the surface of epithelial and endothelial cells) that favors the spread of viral progeny and alters various biological functions mediated by glycans, including recognition by the adaptive immune system [378]. In HBV-associated HCC, upregulation of FUT1, 4, and 8 resulted in elevated levels of fucosylated N-glycans as evidenced by integrating transcriptomics, glycomics, and glycoproteomics [379].

Despite its physiological and pathological importance, no structural information is currently available for human FUT1. Figure 25 shows that this protein is predicted to be mostly ordered with several disordered/flexible regions. Being a single-pass type II membrane protein, FUT1 has a transmembrane helical region (residues 9-25), which, together with the most N-terminal positively charged residues forms a signal-anchor domain. Based on the evaluation of its LLPS potential, FUT1 can serve as a droplet-client, since with low p_LLPS of 0.2215, it possess one DPR (residues 58-69).

FUT1-centered PPI network contains 53 members connected by 491 interactions. Most enriched biological processes associated with the members of this network are protein glycosylation, carbohydrate derivative biosynthetic process, organonitrogen compound biosynthetic process, cellular biosynthetic process, organic substance biosynthetic process, protein O-linked glycosylation, carbohydrate metabolic process, oligosaccharide biosynthetic process, carbohydrate biosynthetic process, and fucosylation. Their most enriched molecular functions are glycosyltransferase activity, hexosyltransferase activity, transferase activity, and catalytic activity. The cellular components most enriched in these proteins are Golgi membrane, Golgi apparatus, organelle membrane, endomembrane system, integral component of membrane, cytoplasm, Golgi cisterna membrane, Golgi cisterna, Golgi apparatus subcompartment, and organelle subcompartment. Among the most enriched reactome pathways associated with these proteins are blood group systems biosynthesis, Lewis blood group biosynthesis, metabolism of carbohydrates, metabolism, O-linked glycosylation of mucins, O-linked glycosylation, keratan sulfate biosynthesis, post-translational protein modification, asparagine N-linked glycosylation, and N-glycan antennae elongation in the medial/trans-Golgi.

4. Conclusions

The bioinformatics analysis performed on the proteins of different hepatitis viruses and on the host proteins interacting with different hepatitis viruses provided important observations regarding the roles of intrinsic disorder in the life cycle of hepatitis viruses. Many of the viral and host proteins were predicted to be related to cellular LLPS processes, serving either as droplet drivers or droplet clients. These observations suggest that phase separation and formation of biomolecular condensates are important for the viral replication and fight with host immunity on one side and for the host response to the viral infection on the other side.

Host proteins interacting with hepatotropic viruses were shown to be characterized by noticeable variation in their intrinsic disorder status. Although global disorder distribution within the sets of host proteins interacting with hepatitis viruses was not too different from that of the entire human proteome, more host proteins interacting with hepatitis viruses were predicted as moderately disordered in comparison with the entire human proteome.

Analysis of the host interactomes of HAV, HBV, HCV, HDV, and HEV revealed the presence of 33 shared proteins. Interestingly, 32 of these shared host proteins are found in the HBV and HCV interactomes, whereas interactomes of HAV, HDV, and HEV contains 11, 5, and 1 shared host proteins, respectively. All the host proteins shared by several hepatotropic viruses were predicted as noticeably disordered.

Detailed bioinformatics analysis of the 11 host proteins shared by four or three different hepatitis viruses indicated that intrinsic disorder can play several important roles in functionality of these proteins. It is also very likely that intrinsic disorder of the host proteins shared by different hepatotropic viruses can be related to the hepatitis co-infection, joint pathogenicity of co-infecting viruses, modulating their mechanisms of immune evasion, promoting the development of antiviral drug resistance and, thereby, contributing to the virus adaptability and evolution.