The Hepatitis B Virus Interactome: A Comprehensive Overview

Despite the availability of a prophylactic vaccine, chronic hepatitis B (CHB) caused by the hepatitis B virus (HBV) is a major health problem affecting an estimated 292 million people globally. Current therapeutic goals are to achieve functional cure characterized by HBsAg seroclearance and the absence of HBV-DNA after treatment cessation. However, at present, functional cure is thought to be complicated due to the presence of covalently closed circular DNA (cccDNA) and integrated HBV-DNA. Even if the episomal cccDNA is silenced or eliminated, it remains unclear how important the high level of HBsAg that is expressed from integrated HBV DNA is for the pathology. To identify therapies that could bring about high rates of functional cure, in-depth knowledge of the virus’ biology is imperative to pinpoint mechanisms for novel therapeutic targets. The viral proteins and the episomal cccDNA are considered integral for the control and maintenance of the HBV life cycle and through direct interaction with the host proteome they help create the most optimal environment for the virus whilst avoiding immune detection. New HBV-host protein interactions are continuously being identified. Unfortunately, a compendium of the most recent information is lacking and an interactome is unavailable. This article provides a comprehensive review of the virus-host relationship from viral entry to release, as well as an interactome of cccDNA, HBc, and HBx.


INTRODUCTION
Hepatitis B virus (HBV) is a member of the Hepadnaviridae family which is transmitted via bodily fluids as well as by vertical transmission (Davis et al., 1989;Schweitzer et al., 2015). The outcome of HBV infection is determined by multiple host and viral factors, and determines whether the infection will be acute, chronic, or occult (Fanning et al., 2019). Despite the availability of a prophylactic vaccine and potent antiviral treatments, chronic hepatitis B (CHB) infection affects 292 million individuals worldwide (Lazarus et al., 2018). The current standard of care is treatment with nucleos(t)ide analogs (NUCs) (i.e., lamivudine, adefovir, entecavir, telbivudine, and tenofovir), that inhibit the HBV polymerase reverse transcription (Liang et al., 2015). These therapies lead to suppression of viral replication, visible by a decrease in viral load, the normalization of serum alanine transaminase and improvement of liver histology (Bitton Alaluf and Shlomai, 2016). However, even prolonged treatment with NUCs rarely results (<10%) in functional cure of CHB and most often leads to virological relapse after treatment cessation (Liang et al., 2015;Kim, 2018).
Also pegylated interferon alpha (peg-IFNα) is approved for use in CHB patients although it is not the preferred therapy due to the occurrence of side effects. Furthermore, it is counter indicated for some patients such as those with liver cirrhosis (Saracco et al., 1994).
Untreated or off-treatment chronic patients are at risk to develop life threatening conditions such as fibrosis, cirrhosis, liver failure, and hepatocellular carcinoma (HCC). In 2015, 887,000 people died from HBV-related cirrhosis and liver cancer alone (WHO, 2017). The ultimate therapeutic goal in CHB is preventing these life-limiting outcomes and to achieve a functional cure characterized by the loss of surface antigen (HBsAg) and HBV-DNA in the blood off-treatment.
Hepatitis B virus functional cure will be achieved when the high viral load, the antigen burden and inadequate host immune responses are overcome and thus may need a broader therapeutic approach involving multiple targets, both viral and host. With regard to the latter, in-depth knowledge of the HBV life cycle is indispensable for identifying mechanisms, that are targetable with new therapeutics.
Part of the therapeutic approach may be to target the interface between viral proteins and cellular targets. The HBV viral proteins have pluripotent functions and our understanding of how they interact with host proteins is continuously evolving. The interactions of these viral factors with the host cell proteome are complex and helps to shape the cellular environment for the virus to replicate. In addition, cccDNA, the template of all viral mRNAs, behaves as a minichromosome and attracts a multitude of protein partners. However, all these reported interactions are scattered in literature, and currently there is no overview bringing together the interactome of HBV. This review aims to provide such an overview, from entry to viral release, it summarizes the known interactions between viral proteins and host proteins. Because cccDNA, HBc, and HBx have been described in many interactions, we focused the construction of an interactome network around these three entities.

INTERACTIONS DURING THE EARLY PHASES OF HBV INFECTION
The HBV particle consists of an incomplete 3.2 kb doublestranded (ds)DNA genome [relaxed circular DNA (RC-DNA)] packaged together with the viral polymerase in an icosahedral capsid assembled by HBV core (HBc) proteins (Summers et al., 1975). This nucleocapsid is enveloped by a lipid membrane studded with three forms of HBV surface antigen protein (collectively referred to as HBsAg) to compose the virus or Dane particle [reviewed by Bruss (2004)].
The life cycle of HBV begins upon its interaction with heparan sulfate proteoglycans (HSPGs) and subsequent binding to the sodium taurocholate co-transporting polypeptide (NTCP) receptor on the surface of the hepatocyte (Watashi et al., 2014;Yan et al., 2014; Figure 1). The interaction between virus and cell induces conformational changes of the membrane embedded myristoylated N-terminal preS1-domain of the viral large surface protein (L-HBsAg) leading to exposure of the receptor binding site for the NTCP receptor, which enables binding of the virus and entrance into the cell (Schulze et al., 2007(Schulze et al., , 2010Yan et al., 2012Yan et al., , 2013Yan et al., , 2014Nkongolo et al., 2014;Watashi et al., 2014). Recently, a crucial role in mediating HBV-NTCP internalization of epidermal growth factor receptor (EGFR) was published (Iwamoto et al., 2019). Besides the NTCP receptor, squamous cell carcinoma antigen 1 (SCCA1) and ferritin light chain (FTL) have also been identified as HBV co-receptors (Figure 1). Triple complexes of preS1, FTL, and SCCA1 were observed and overexpression assays with these proteins showed increased infection rates both in vitro and in vivo (Hao et al., 2012). The prevention of entry has been of interest as an antiviral target to circumvent viral spread by blocking de novo infection. In recent years molecules such as Myrcludex B (also known as bulevirtide), ezetimibe, cyclosporin derivates (CsA), and monoclonal antibodies against HBsAg epitopes were identified to interfere with this process (Gripon et al., 2005;Lucifora et al., 2013;Shimura et al., 2017).
The virus enters the cell by inducing endocytosis via caveolinmediated endocytosis or via clathrin-mediated pathways (Macovei et al., 2010;Umetsu et al., 2018; Figure 1). In differentiated HepaRG cells, HBV infection has shown to be dependent on caveolin-mediated endocytosis. However, in Umetsu et al. (2018), the formation of a complex between the L-HBsAg, the clathrin heavy chain (CHC) and the clathrin adaptor protein-2 (AP-2) was described, suggesting an alternative endocytosis pathway (Figure 1). Indeed, inhibition of the clathrin-mediated pathway by silibinin and chlorpromazine has been reported to impair HBV uptake . Further work will be needed to understand the relative importance of these two pathways. After endocytosis, subsequent movement of the virus through the endocytic pathway is regulated by Rab proteins. These are guanosine triphosphatases (GTPases) that occupy specific endocytic compartments and direct endocytic vesicles to different cellular compartments. Silencing of Rab5 or Rab7, in contrast with Rab9 and Rab11, resulted in the inhibition of the early stages of HBV infection implying that the transport of virus to late endosomes is important for a successful infection (Macovei et al., 2013; Figure 1).
The precise location and timing of nucleocapsid release from the envelope remains unclear, but this process is required prior to nuclear entry. Transport of the nucleocapsid to the nucleus is facilitated by the microtubule network and the dynein L11 motor proteins through a direct interaction with the capsid (Osseman et al., 2018; Figure 1). In the nucleocapsid "uncoating" process, phosphorylation of the C-terminus of HBc destabilizes the capsid and allows the binding of importins α and β (Kann et al., 1999;Barrasa et al., 2001;Nguyen et al., 2008). Although a direct interaction has not been established, a number of kinases including core associated kinase (CAK), SR protein-specific kinase 1 (SRPK1) and SR protein-specific kinase 2 (SRPK2), have been reported to be involved in this phosphorylation process (Kau and Ting, 1998;Daub et al., 2002; Figure 1). Once the nucleocapsid arrives at the nuclear pore complex (NPC), it can pass the complex as an intact particle (Pante and Kann, 2002;Fay and Pante, 2015). Interestingly, HBV seems to utilize a unique way of triaging immature from mature capsids at the level of the NPC as only mature capsids disassemble. In this process, importin β and Nup153 play a role via direct interaction with the capsid (Schmitz et al., 2010; Figure 1). Once through the NPC, the capsid is deposited in the nuclear basket where only mature capsids can pass. In the nucleus, the final uncoating, where capsid structures and viral DNA separate, takes place in an importin α and β-dependent manner (Gallucci and Kann, 2017).

THE cccDNA MINICHROMOSOME
Once inside the nucleus, the RC-DNA is converted into cccDNA (Summers et al., 1975;Tuttleman et al., 1986;Wu et al., 1990;Lieberman, 2016). Early research using duck hepatitis B virus (DHBV) showed that the cccDNA was in fact organized as a minichromosome similar to host chromatin and SV40 (Newbold et al., 1995). Further DHBV studies showed that in vitro between 1 and 56 copies cccDNA reside in the nuclei of infected cells (Kock et al., 2010). These copy numbers were slightly lower (1-17 copies/cell) in in vivo studies in ducks. Further it was determined that the half-life of DHBV cccDNA is between 35 and 57 days (Addison et al., 2002;Zhang et al., 2003) although shorter half-lives have described (Tuttleman et al., 1986;Wu et al., 1990;Newbold et al., 1995). In vitro kinetic studies were also done using HBV, cccDNA formation is an early life cycle event (Tuttleman et al., 1986) and it was shown that the cccDNA pool grows over the course of 3 days after which a stable pool is reached (5-12 copies/cell) with a half-life of about 40 days (Ko et al., 2014). Similar findings were done using woodchuck HBV (Dandri et al., 2000). Patient samples of HBV infected individuals showed that cccDNA copy numbers were much lower in vivo ranging from 0.01 to 9 copies/cell but at the same time had a much longer half-life of months to a year (Werle-Lapostolle et al., 2004;Bourne et al., 2007;Boyd et al., 2016;Huang et al., 2021). Interestingly, the size and half-life of the cccDNA pool in patients has been suggested to depend on the antigen status (Lythgoe et al., 2021) as much more cccDNA has been shown in HBeAg positive patients while only 0.002 copies/cell were observed in patients that showed HBsAg seroclearance (Werle-Lapostolle et al., 2004).
The cccDNA genome is transcribed to different viral RNAs coding for HBx (0.7-kb RNA), three forms of HBsAg (2.4-kb RNA encoding the large and 2.1-kb RNA encoding the middle and small HBsAg), pre-core protein or HBeAg (3.5-kb RNA) and the core and polymerase protein (pre-genomic RNA or pgRNA, 3.5-kb). This pgRNA also becomes incorporated in the nucleocapsid thereby providing the template for the viral polymerase to produce RC-DNA.

Host Factors Involved in cccDNA Formation
Little is known about the host factors involved in the formation of the cccDNA. The L-HBsAg is not directly involved in cccDNA formation, but is part of a negative feedback mechanism in which high levels of surface protein shut down nuclear shuttling of mature nucleocapsids and direct the cell to produce virions instead . HBc is suggested to be present during the cccDNA formation process (Kock et al., 2010;Schreiner and Nassal, 2017) which is further evidenced by the fact that capsid modifiers inhibit cccDNA formation (Berke et al., 2017).
Several host factors have been reported to interact with HBV cccDNA during its formation and have quite diverse roles. The Flap endonuclease 1 (FEN1), an endonuclease that plays a role in DNA replication and repair, was shown to interact with RC-DNA in the nucleus and additionally could promote cccDNA formation in vitro (Kitamura et al., 2018; Figure 1). The discovery of a protein partner involved in DNA damage repair is coherent with the previous finding that this machinery is exploited by viruses to their own benefit (Schreiner and Nassal, 2017). Ku80, a component of non-homologous end joining DNA repair pathway, was essential for synthesis of cccDNA from dsDNA, but not from RC-DNA  Figure 1). In these processes, HBx could be an adaptor to link cccDNA formation with DNA damage response pathways, under the assumption that HBx is already present in the cell when cccDNA is being formed (Hodgson et al., 2012;Guo et al., 2014;Murphy et al., 2016;Niu C. et al., 2017). The link with the host DNA damage and repair machinery does not end with this interaction, the tyrosyl-DNA-phosphodiesterase (TDP2) also plays a partial role in cccDNA formation by releasing the viral transcriptase from the RC-DNA (Koniger et al., 2014;Cui et al., 2015; Figure 1). The host DNA polymerases K (POLK), H (POLH), and L (POLL) have all been reported to have a positive impact on cccDNA formation, however, the exact mechanism(s) is (are) not yet clear (Qi et al., 2016; Figure 1). In addition to DNA polymerases, knockout experiments showed the importance of cellular DNA ligase 1 and 2 in cccDNA formation (Long et al., 2017). Recently, it was shown that the plus-strand and the minus-strand require different cellular proteins. The plus-strand repair required proliferating cell nuclear antigen (PCNA), replication factor C (RFC) complex, DNA polymerase delta (POLδ), flap endonuclease 1 (FEN1), and DNA ligase 1 (LIG1) while the repair of the minus-strand only required FEN1 and LIG1 (Wei and Ploss, 2020). Also cellular DNA topoisomerases are required for cccDNA formation and amplification (Sheraz et al., 2019). Finally, pre-mRNA processing factor 31 (PRPF31) was identified as a cccDNA-associating factor involved in cccDNA formation (Kinoshita et al., 2017; Figure 1).

The Interactome of the cccDNA
Similar to a cellular chromosome, the cccDNA is bound to histones to form a minichromosome. These host-derived histones (H2A, H2B, H3, and H4) provide, together with the viral HBc, the stable scaffold for the cccDNA to be supercoiled (Newbold et al., 1995;Chong et al., 2017). That being said, the role of HBc in both cccDNA formation and maintenance is still under investigation. For example, despite their involvement in several processes regarding cccDNA formation, maintenance and transcription, capsid modifying compounds do not eliminate the cccDNA pool (Berke et al., 2017) nor is HBc essential for transcription .
On the cccDNA of Duck hepatitis B virus (DHBV), nucleosomes are non-randomly positioned, suggesting that, like host cellular chromatin, positioning of the nucleosomes and histone modifications of the cccDNA may regulate cccDNA transcription (Bock et al., 1994;Pollicino et al., 2006). Methylation, acetylation, phosphorylation or other posttranslational modifications (PTMs) of these cccDNA-bound histone tails can fine tune the gene expression by altering the chromatin structure (Tropberger et al., 2015). This change in structure can wind the chromatin more tightly to prevent access of transcription factors and repress gene transcription. On the other hand, histone modifications can also result in increased DNA accessibility, transcription factor binding and therefore promoting gene activation (Li et al., 2007;Voss and Hager, 2014). In addition, the minichromosome attracts several other partners, many of which are transcription factors that further determine whether the cccDNA is transcriptionally active or inactive ( Table 1).
As previously mentioned, HBx and HBc proteins are bound to cccDNA. HBc has been described to modulate transcription from the cccDNA. Zlotnick et al. showed that the presence of HBc on a CpG island in the cccDNA can be linked to increased cccDNA activity, while methylation of the CpG island correlated with decreased cccDNA activity . In addition, the presence of HBc appears to have a role in the maintenance of the structure of the cccDNA (Bock et al., 2001). Together these data suggest that HBc contributes to the epigenetic regulation of the cccDNA, which in turn contributes to its longevity.

Modalities Acting on cccDNA
A role in viral rebound made cccDNA a target for new antiviral drug development. Success of such tactics relies on complete inhibition of cccDNA throughout the lifespan of the hepatocyte. A first approach is to target the formation of cccDNA, although it can be questioned how much benefit CHB patients will have of such a therapy in the event the cccDNA does not become reduced. Several molecules reported to act through this mechanism have been described in literature. However, to date, these molecules have either been stopped at pre-clinical stage or did not progress far in clinical trials (Cai et al., 2012;Liu et al., 2016). The only assets which encompass this capacity and are still under clinical investigation are the entry inhibitor bulevirtide and capsid assembly modulators. The latter are small molecules that accelerate capsid formation but turned out to have a dual mode of action in preventing cccDNA formation when added in vitro at early stages of infection (Berke et al., 2017;Vandenbossche et al., 2019). Secondly, a number of molecules have been described that silence the cccDNA, either by inhibiting cccDNA transcription [e.g., Tamibarotene (Nkongolo et al., 2019)] or by diminishing HBV RNA levels post-transcription (e.g., RNA destabilizers such as RG7834 (Mueller et al., 2019); RNA interference). Tamibarotene never made it to clinical trials for HBV, while RG7834 was stopped in Phase I. Transcriptional control of cccDNA expression may also be achieved by interfering with the function of HBx, HBc or an interaction partner. An example is the interference between HBx and DNA damagebinding protein 1 (DDB1). HBx was found to hijack DDB1  Associated with supressed Replication -Potential Interactions
SIRT 3 Mediates cccDNA transcription. Repression lifted by HBx. Ren et al., 2018 Verified interactions are those protein-protein interactions that were identified using proteomics methods such as pull downs or yeast-2-hybrid. Potential interactions are those which have been shown using methods that strongly suggest an interaction (e.g., co-localization) but were not verified using pull-down methods.
Recently, epigenetic modifiers that specifically target viral factors involved in the regulation of cccDNA expression have been described and are currently being evaluated. Several selective inhibitors (e.g., C646) for histone acetyltransferase like CBP and P300 have been used to study the inhibitory effect on HBV transcription (Tropberger et al., 2015). The prodrug GS-5801 has also been shown to inhibit transcription from cccDNA by blocking the activity of lysine demethylase 5 (KDM5) (Gilmore et al., 2017). Although these observations show that silencing of HBV transcription is possible, the main throwback of most of these targets is the lack of desired selectivity for cccDNA and their potential to impact cellular processes. Complete elimination of cccDNA by compromising the stability or the half-life of the molecule is often dubbed the "Holy Grail" of HBV research. Many molecules have been described that phenotypically reduce the quantity or transcription level of cccDNA. Recently, a small molecule, ccc_R08, with an unknown mode of action was shown to decrease the pool of cccDNA together with a decrease in viral transcripts and viral antigens in primary human hepatocytes (PHH) and in an HBV minicircle mouse model (Wang et al., 2019). In most instances, information on the exact mechanism of such molecules is lacking implying a need to conduct target deconvolution studies to identify the respective interaction partner or process. We created a cccDNA network map, not only to visualize the currently known cccDNA interacting proteins but also to be put alongside such exercises (Figure 2). Proteins that were shown to interact with more than two of the founding nodes (cccDNA, HBc, and HBx) are depicted in orange. proteins that were extrapolated to connect to one or more interacting proteins are shown in blue.

THE HBx INTERACTOME
The interactome of HBx extends beyond its interaction with the cccDNA and associated proteins. Besides nuclear interaction partners, HBx also interacts with various proteins in the cytoplasm, the endoplasmic reticulum (ER) and the mitochondria (Henkler et al., 2001;Huh and Siddiqui, 2002;Belloni et al., 2009;; Figure 1). This may explain why this small viral protein (17-kDa) is not only involved in HBV replication, but is also shown to contribute to the development of HCC and interfere with cell cycle regulation, glucose metabolism, oxidative stress, calcium signaling, apoptosis and DNA repair (Luber et al., 1993;Waris et al., 2001;Bouchard et al., 2006;Benhenda et al., 2009; Table 2). The pivotal nature of HBx is demonstrated by Table 2 in which more than 250 HBx interaction partners are summarized. However, it does need to be mentioned that some of these interactions may be very weak or very brief and their relevance may be limited.

THE HBc INTERACTOME
HBc is mostly known as the building block of the HBV capsid (Summers et al., 1975) but in recent years it has been shown that its function is not limited to this and also plays a role in cccDNA stability, transcription and epigenetic regulation (Newbold et al., 1995;Bock et al., 2001;Zlotnick et al., 2015;Chong et al., 2017), evasion of antiviral mechanisms (Lucifora et al., 2014), reverse transcription , cellular trafficking (Schmitz et al., 2010;Yang et al., 2014), genomic replication (Lott et al., 2000), and viral egress (Bardens et al., 2011). The field is also discovering more and more that HBc expression is extensively regulated by core promotor regulation, core mRNA modulation and post-translational modifications which highlights its importance in the life cycle (Buckwold et al., 1997;Sohn et al., 2006;Kohno et al., 2014;Qian et al., 2015;He et al., 2016;Bartusch et al., 2017;Lubyova et al., 2017;Heger-Stevic et al., 2018;Makokha et al., 2019). Initially, the impact on the capsid made HBc an appealing drug-target (Berke et al., 2017). However, given that there is also an interplay with cccDNA and HBx these molecules may have more far-reaching consequences. As more protein interactions between HBc and the host are elucidated, we also compiled the interactome of the core protein and linked it to the HBx and cccDNA interactomes (Figure 2).
A cccDNA AND HBx GENE ASSOCIATION NETWORK: EXPANDING THE POTENTIAL cccDNA AND HBx INTERACTOME Tables 1-3 summarize what is currently known in the literature (manual curation) about cccDNA, HBx protein, and HBc protein-DNA and protein-protein interactions, respectively. However, to utilize this information to predict and potentially identify new protein interactions, network pathway analysis was performed (Ingenuity Pathway Analysis, IPA, Qiagen). IPA enables gene network generation from the Ingenuity Knowledge Base, a data repository of biological interactions and functional annotations.
If proteins selected as network "seeds" were not apparently connected or networks had less than 35 gene products, IPA added proteins from the IPA Knowledge data base to maximize the connectivity of the "seed" molecules within the filter limits. We also filtered out those proteins that were only interacting with either cccDNA, HBx, or HBc and had no extrapolated nodes. This kept the networks to a manageable size and reduced redundancy while deriving as much as possible biological context from the analysis. When adding molecules from the knowledge database, IPA uses a connectivity metric (edge-weighted spring layout) that prioritizes molecules that have the greatest overlap with the existing network. This means that the organization of the network in clusters is not based on proteins sharing similar pathways but is based on the number of described interactions in between those proteins. Upon completion of the IPA network generating  HBc reduced the DNA-binding ability of E2F1 to the binding site of the p53 promoter and respectively represses expression of p53. Kwon and Rho, 2003 SIRT7 cccDNA expression modulation, HBc functions as a bridge between cccDNA and SIRT7. Yu et al., 2021 HBs Interaction prevents nuclear translocation to HBc. Zajakina et al., 2014 algorithm, two networks were produced showing both "direct" and "indirect" relationships for either HBx, HBc, or HBV cccDNA interacting proteins. These were then merged and exported to Cytoscape 3.7.1 using an edge weighted spring layout to create the final network illustration showing the relationship between HBx, HBc, and HBV cccDNA interactingproteins (Figure 2).
In the network, proteins that interact with HBc are depicted in red, those interacting with cccDNA in pink and those interacting with HBx in green. We also highlighted those proteins that were described to interact with 2 or more of our founding (HBc, HBx, and cccDNA) nodes (Figure 2). Nineteen proteins are identified (P300, TBP, PIN4, CBP, SPIN1, CEBPB, SP1, CRTC1, RXRA, NR1H4, KDM1A, HSPA1A, APOBEC3G, APOBEC 3B, CREB1, PRMT5, HDAC1, E2F1, and SIRT1) as interacting proteins of HBx, HBc, and cccDNA. These are interesting because these components may be a driving force in cccDNA transcription and maintenance. Moreover, these may be interesting proteins for further functional research as they seem to play a connecting role in the viral life cycle (Figure 2). Most of these proteins, 12 out of 19, are regulators of transcription, for example, TBP and CRTC1 are both involved in transcription initiation; CREB1 and E2F1 are enhancers of transcription; P300, CBP, SPIN1, SP1, PRMT5, HDAC1, and SETD1B are all epigenetic modifiers that can influence the chromatin to a specific transcriptionally accessible, active state. Finally, PIN4 was described as a chromatin remodeler. Any of these proteins could be a potentially interesting target to influence transcriptional status of cccDNA. Notable is that all these transcription-related proteins have interactions with HBc, hereby confirming a role for HBc beyond capsid assembly. Also interesting is the occurrence of two APOBEC proteins as partners for HBx, HBc and cccDNA. APOBEC proteins play a role in anti-viral immunity (Stavrou and Ross, 2015) and is a means of the cell to counteract the effect of infection (Lucifora et al., 2014).
Through the IPA approach, the network was expanded from just those proteins with known interactions with HBx, HBc, or cccDNA to an additional 210 proteins (indicated in light blue) which may play a role in protein-protein or protein-cccDNA interactions. While the interactions itself are verified in literature, their involvement in the HBV pathology and viral replication cycle is not confirmed yet. Hence, these proteins provide an interesting starting point for further research. Analysis of the network shows that via these interacting proteins, HBV also taps into host pathways such as cell cycle, cell signaling, DNA repair, transcription regulation and apoptosis. This is not surprising as many of these processes have been described in relation to HBV already. However, this is the first time description of proteins which may be involved in these processes and how they relate to the cccDNA minichromosome, HBc or HBx. An additional interesting observation is that many heat shock proteins (HSP) (Hsp70, Hsp90, Hsp27, HSPD1, HSPA1A, HSPA1B, HSPA8, HSPA9, HSPA1L, HSP90AB1, HSPA5, and HSPA6) were observed as interacting proteins in this network. Literature has already described that viruses rely on host HSPs for viral protein folding and induce overexpression of HSPs in the infected cells (Bolhassani and Agi, 2019). Moreover, several HSPs were associated with some viral particles (Fust et al., 2005;Bolhassani and Agi, 2019). In HBV, downregulation of Hsp70 and Hsp90 by small interfering RNA significantly inhibited HBV production. Furthermore, also a significant reduction of HBV secretion could be observed in HepG2.2.15 cells treated with an Hsp90 inhibitor (Liu et al., 2009;Bolhassani and Agi, 2019). Further research will be required to confirm the additional protein partners identified in this network analysis.

INTERACTIONS DURING THE LATE PHASES OF HBV INFECTION
In HBV infection, besides the budding of virions, there is also the shedding of an excess amount of subviral particles (Figure 1). These particles are non-infectious 22 nm spheres or filaments of variable length consisting solely of the HBsAg envelope protein, which may be expressed from either cccDNA or HBV DNA that is integrated into the human genome (Heermann et al., 1984; Figure 1). Budding of infectious virus and shedding of subviral particles happen via distinct pathways (Selzer and Zlotnick, 2015).
Although redundant in viral assembly, the M-protein and its interaction with calnexin has been shown to be involved in the secretion of subviral particles (Werr and Prange, 1998). In the cytoplasm, HBsAg interacts with cyclophilin A (CypA) and stimulates the extracellular secretion of CypA (Tian et al., 2010; Figure 1). Interestingly, it seems that the presence of CypA reciprocally stimulates HBsAg secretion, as inhibitors against CypA reduce the amount of secreted HBsAg (Phillips et al., 2015).
To construct new virions, the pgRNA is packaged together with the viral polymerase in the nucleocapsid, which is formed in the cytoplasm by assembly of 120 HBc dimers (Lambert et al., 2007). Although not well understood, the interaction between HBc dimers and cellular protein nucleophosmin (B23) was shown to promote this assembly (Jeong et al., 2014; Figure 1). This nucleocapsid is surrounded by a cellular lipid layer embedded with three viral S glycoproteins, which originate from the endoplasmic reticulum (Bruss, 2007). Virion assembly depends solely on the L-protein, whereas the S-protein is required but not sufficient, and the M-protein is redundant (Bruss, 2004). To aid in building this unusual composition, Hsp70, and mammalian BiP were described as interaction partners of the L-protein in vitro and in vivo (Loffler-Mary et al., 1997;Lambert and Prange, 2003;Wang Y.P. et al., 2010; Figure 1). In the assembly of the mature virion, the S-protein needs to interact with the nucleocapsid (Loffler-Mary et al., 2000).
Once the mature virion is formed, it is ready to bud on the surface of the cells. The whole orchestration of this process is not clear at all, let alone accurately described in terms of interacting proteins. HBV makes use of the ESCRT, a machinery essential for the sorting of cellular cargo proteins in multivesicular bodies (Bardens et al., 2011). In this process, aryl hydrocarbon receptor interacting protein (AIP1)/ALIX and vacuolar protein sorting 4 homolog B (VPS4B) were found to colocalize with HBV particles (Kian Chua et al., 2006;Watanabe et al., 2007; Figure 1). Also, expression of dominant negative mutants of ESCRT-III complex-forming charged multivesicular body protein (CHMP) proteins (CHMP3, 4B, and 4C), as well as vacuolar protein sorting 4 homolog A (VPS4A) or VPS4B mutants, and knockout of γ2-adaptin blocked HBV assembly and egress (Hartmann-Stuhler and Prange, 2001;Rost et al., 2006;Lambert et al., 2007; Figure 1). However, the manipulation of these proteins did not alter the secretion of subviral particles. Also involved in viral egress is Neural precursor cell Expressed, Developmentally Down-regulated 4 (NEDD) E3 ubiquitin protein ligase, which appears to control virus production by binding to the late assembly domain-like PPAY motif of HBV capsids (Rost et al., 2006;Garcia et al., 2013). It is also known that at some point, autophagy is involved in HBV production as the S-protein was shown to interact with the autophagy factor LC3 and manipulations to the pathway result in changes in HBV secretion .

CONCLUDING REMARKS
The interactome we build of the cccDNA, HBc and HBx protein in this review emphasizes the vast amount of knowledge there is about the interactions between HBV proteins and in particular HBx, HBc and the cccDNA. To our knowledge, this is the first time the information has been brought together in a comprehensive overview. Bringing this information together, it shows that there are still clear gaps in knowledge. For example, the network shows that several proteins were only described in a single publication as an interacting protein of cccDNA, HBc, or HBx. Further characterization of this kind of interactions and potentially understanding the reason behind these interactions will greatly benefit the understanding of HBV-related processes. In addition, through analysis of the known interacting proteins, we predicted 210 proteins which potentially interact with either cccDNA, HBx, HBc, or with multiple key modalities of HBV.
Experimental verification of these proteins can lead to the discovery of novel mechanisms and expansion of known protein interaction networks.
Being able to position cccDNA, HBc and HBx in the greater whole of the cellular environment is paramount to better understand how HBV hijacks the cellular environment.

AUTHOR CONTRIBUTIONS
EVD, JV, LV, and FP conceived, designed, and wrote the manuscript. BS performed the network pathway analysis, designed the gene association network, and contributed to scientific discussions about the generated network. All authors have read and edited the manuscript.