HCMV, the prototypic β-herpesvirus, represents a major human pathogen and is characterized by a multifaceted mode of virus-host interaction. HCMV seroprevalence in the adult population ranges between approximately 40% to 90% and reaches even higher levels, of more than 95%, in countries with a low socio-economic standard [1]. HCMV exerts a strict species specificity and a comparably slow replication cycle spanning approximately three days in vitro [2,3]. Viral genomic DNA replication takes place in the nucleus and the double-stranded viral genome is packaged into capsids, which then undergo nuclear egress and budding through the nuclear membranes [4,5]. In the cytoplasmic virion assembly complex (cVAC), capsids are assembled with tegument proteins, before fully enveloped virus particles of approximately 150–200 nm are formed in the trans-Golgi network and released from the cell by final transition through the cytoplasmic membrane [2,6,7]. In addition to highly productive lytic infection of major target cells, such as fibroblasts, smooth muscle cells, endothelial and epithelial cells [8,9,10,11,12], HCMV causes life-long persistence by latent infection of minor target cells, such as monocytes/macrophages and CD34⁺ hematopoietic stem cells, in which latent HCMV may undergo reactivation resulting from immune insult, allogenic stimulation or differential signals (reviewed in [13]).

Due to the fact that primary and nonprimary infections (i.e., reactivation or reinfection) are mostly asymptomatic in healthy, immunocompetent individuals, HCMV infection usually remains clinically unrecognized. In contrast, patients with a compromised immune system, such as transplant recipients or AIDS patients, severely suffer from HCMV-related diseases, such as interstitial pneumonia, retinitis, gastroenteritis, esophagitis and organ failure, resulting in an increased mortality and morbidity [14,15,16,17]. Importantly, the immature immune system is a high risk factor for congenital cytomegalovirus infection (cCMV) of embryos and infants; thus, HCMV represents the most frequent cause for pathogen-derived developmental defects triggering mental retardation, loss of hearing or vision and microcephaly [18,19,20,21,22,23,24]. HCMV is one of few viruses that are able to cross the placenta efficiently, i.e., at least 33% of all primary infections during pregnancy of seronegative mothers, and an additional lower percentage of nonprimary infections undergo virus transmission resulting in cCMV infection of the unborn [25,26]. Thus, in Germany, approximately 3500 out of 700,000 newborns acquire cCMV per year [19]. Because of the lack of comprehensive HCMV screening, it is understood that approximately 10% of these are symptomatic at birth, including cases of stillbirth, and another 10%–15% may acquire symptoms at a later onset. HCMV can be transmitted by various body fluids, such as saliva, breast milk, vaginal secretions, semen and leukocytes containing blood and urine [27,28,29,30,31].

Until today, no vaccine has been approved to control HCMV infections. Despite 60 years of intensive HCMV research, only a few antiviral drugs have been approved, which mostly interfere with the viral DNA polymerase pUL54, i.e., nucleoside/nucleotide analogs, such as the gold standard ganciclovir (GCV), its prodrug valganciclovir (VGCV), cidofovir (CDV) and the pyrophosphate analog foscarnet (FOS). Unfortunately, these drugs frequently cause severe side-effects, such as myelotoxicity, anemia and nephrotoxicity, or show poor bioavailability, which drives the selection of drug resistant virus variants [32,33,34,35,36,37]. In 2017, letermovir (LMV), the first anti-HCMV drug that targets the viral terminase complex consisting of pUL56, pUL89 and pUL51 core-subunits, was successfully assessed in clinical trials. Currently, LMV is approved for HCMV prophylaxis in hematopoietic stem cell transplantation recipients. LMV also represents a promising candidate for future combination therapies or even options of cCMV control [38,39,40,41]. However, based on the occurrence of LMV-resistant viral mutants [42] and the present lack of an approved treatment option for infants, the requirement of new antiviral drugs is still emphasized. This situation underlines the necessity of basic research to refine the understanding of the manifold and complex HCMV-host interplay and antiviral targeting strategies.

pUL97 is a tegument protein, which is packaged into virions and is expressed with early-late kinetics [43]. The 707-amino acid protein exists in three isoforms due to alternative initiation of translation at residues M1, M74 or M157, resulting in protein varieties of approximately 100 kDa, 80 kDa and 70 kDa, respectively (Table 1, Figure 1) [44]. The full-length kinase possesses two NLS sequences in the poorly structured N-terminus, which mediate the predominantly nuclear localization of pUL97 [45,46]. The kinase domain was assigned to the globular C-terminal part, amino acids 337–651, based on sequence homologies or extended to 337–706, based on biochemical validation [47,48,49,50]. An invariant lysine residue at position 355 is essential for kinase activity, thus leading to the catalytically inactive K355M mutant [51,52,53]. Dimers and oligomers are formed via the self-interaction domain (amino acids 231–280) of pUL97 [54]. Interestingly, the direct association of pUL97 with human cyclins has been demonstrated and, hereby, the core region responsible for cyclin T1 binding proved to be identical with the pUL97 self-interaction domain [55], thus illustrating a functional role of cyclins in pUL97 dimerization/oligomerization [52,56,57,58]. Concerning the properties of protein interaction and substrate phosphorylation of pUL97, a number of viral as well as cellular proteins have been identified thus far [see references in legend of Figure 1]. The functionality of these substrates spans various regulatory aspects of viral replication, such as nuclear egress, intrinsic immunity, genome replication and gene expression (Table 1, Figure 1). Notably, several of the pUL97-specific substrate proteins also represent substrates of cellular CDK-cyclin complexes and may thus underlie a process of dual phosphorylation through these two different kinds of protein kinases in HCMV-infected cells. While sequence conservation between the open reading frame ORF-UL97 and other kinases is generally low, functional and structural similarities have been identified between pUL97 and CDKs, so that pUL97 was termed as a multifunctional viral CDK ortholog (vCDK). Importantly, both deletion of ORF-UL97 or pharmacological inhibition of pUL97 activity resulted in a strong delay of HCMV replication [52,59,60,61], likewise explained by the fact that the kinase exerts many regulatory functions during viral replication (Table 1). On this basis, pUL97 could be validated as an interesting target for novel antiviral strategies and a panel of small molecule-type inhibitors of pUL97 activity belonging to different chemical classes has been described during the last years (see below, Section 3, Section 4, Section 5 and Section 6).

The interaction between HCMV pUL97 and human cyclins of the types B1, T1 and H has been described in our earlier reports [47,55,82]. The three cyclins obviously possess different affinities in terms of strength of pUL97 binding detected by coimmunoprecipitation (CoIP)- and mass spectrometry (MS)-based analyses. In case of cyclin B1, a requirement of catalytic activity of pUL97 for cyclin binding was identified, whereas in case of cyclin H, pUL97 interaction was found dependent on the environment of HCMV replication [82]. Recently published data indicate a substrate-bridging function of cyclin(s) for the binding of pUL97 to its substrate pp65, as determined with a pp65 mutant lacking a putative cyclin-docking motif [83].

 

Previous investigations led to the postulate of a substantial relevance of pUL97-cyclin interactions, as characterized by the following findings: (i) The HCMV kinase pUL97 acts as a structural CDK ortholog originally based on our bioinformatic modeling and biochemical analyses. (ii) Our initial report on pUL97-cyclin T1 interaction could be extended to additional types such as cyclins B1 and H [47,55,56,82]. (iii) The interaction pUL97-cyclins B1/T1/H was confirmed by several methods including highly sensitive mass spectrometry-based proteomics. (iv) Specifically, the interaction pUL97-cyclin B1 was found to be phosphorylation-dependent for both proteins. In addition, cyclin B1 (but not H) was phosphorylated by pUL97 in vitro [56]. (v) Using a protein assembly-based CoIP assay, the formation of binary and ternary complexes involving pUL97, cyclin H and CDK7 was identified, thus suggesting a cyclin bridging concept [125]. A central finding was that regions responsible for cyclin T1 interaction of pUL97 and pUL97-pUL97 self-interaction showed an overlap in N-terminal amino acids 231-280 (Figure 1; [54,55]). These data strongly suggest that cyclin binding is involved in pUL97-pUL97 self-interaction and very recent findings specified this activity for cyclin types T1 and H (but not B1), thus confirming the bridging function of cyclins T1/H in pUL97 dimerization or hetero-oligomerization. This self-interaction property is known to be a factor required for developing full catalytic activity of the pUL97 kinase [see references in Table 1]. The amino acid region 231–280 of pUL97 is considered as a minimal binding region for cyclin T1, which may be complemented by the additional binding of globular domain interfaces of pUL97 in the further C-terminal region, contributing to cyclin binding in a type-specific manner (cyclin T1, amino acids 361–532; cyclin B1, 363–647; cyclin H, 328–532; Figure 1; [56,82]).

 

In order to address the question of which spectrum of different types of human cyclins may associate with the viral pUL97 kinase, two specific experimental approaches have recently been performed. Firstly, a recombinant HCMV expressing a Flag-tagged version of pUL97 (namely the largest, fully functional isoform M1 of pUL97 encoded by HCMV AD169-UL97(Mx4)-Flag; [44]) was used for Flag-specific coimmunoprecipitation settings. The CoIP samples were then applied in a mass spectrometry-based (MS) proteomic assessment of pUL97-associated viral (Table S1) and cellular proteins (Table S2). HCMV AD169, expressing untagged pUL97, was used as a CoIP/MS specificity control. The identified viral proteins included several known interactors and/or substrates of pUL97 and showed a substantial overlap with those detected in our similar approach performed earlier, as based on the CoIP of pUL97-cyclin complexes using cyclin-specific antibodies [82]. Cellular proteins identified by this approach contained cyclins, CDKs and additional host proteins confirming earlier findings of pUL97-specific protein complexes. Notably, cyclins T1 and B1 were again safely detected, as those types of cyclins had been found by a variety of methodological approaches before (summarized in Table 2). Secondly, a panel of cyclin-specific antibodies were employed in a broader setting of CoIP analysis to learn more about the overall spectrum of pUL97-cyclin interaction. Representative members of the functional groups of cyclin types have been chosen, i.e., B-like, C-like and Y-like cyclins (Table 2, Supplementary Materials Figure S1). To this end, the cyclin-specific CoIP of pUL97 was then performed, again on the basis of total lysates prepared from HCMV-infected primary fibroblasts, followed by a quantitative assessment based on densitometric measurements (in duplicates, using two series of stained CoIP/Wb filters). The results, on the one hand, confirmed our earlier postulate that pUL97 strongly interacts with cyclin types B1, T1 and H (the latter primarily with pUL97 expressed in HCMV-infected cells, but very poorly with pUL97 transiently expressed in transfection-based settings; [56,82]). On the other hand, even more types of human cyclins could be additionally detected, either with moderate/weak (cyclins E, F and Y) or strong (cyclins B2 and K) properties of pUL97 interaction (Figure S1, Table S3 and Table 2). This topic of cyclin specificity of pUL97 and its functional relevance for HCMV replication will be further investigated by the use of recombinant HCMVs expressing mutant versions of pUL97 carrying cyclin-binding defects.

 

Notably, pUL97 phosphorylates several viral and cellular proteins (see horizontal bars in Figure 1 for those binding regions within pUL97 that have been mapped thus far), including the viral DNA polymerase cofactor pUL44 [122], viral RNA transport factor pUL69 [79], major tegument protein pp65 [95], nuclear egress core proteins pUL50-pUL53 [99,127], cellular multiligand binding protein p32/gC1qR [98,122], tumor suppressor protein Rb [75], nuclear lamins A/C [57,84,94,98,129], RNAP II [100], translation factor EF-1δ [63,101,130], interferon-inducible, intrinsic immune restriction factors IFI16 [96] and SAMHD1 [105] (Figure 2; Table 3; compare with Tables S1–S3).

It should be emphasized that the pUL97 substrate proteins belong to several functionally different groups (Table 3), thus underlining the multifunctional nature of this singly expressed viral protein kinase. Viral proteins interacting with and being phosphorylated by pUL97 span the regulatory areas of viral nuclear egress (pUL50-pUL53 core NEC), genome replication (pUL44), tegumentation and immune-regulatory functions (pp65), viral RNA transport (pUL69) and the pUL97-pUL97 autophosphorylation/autoregulation associated with the formation of dimers and oligomers. As far as cellular substrates are concerned, the following regulatory areas are addressed: nuclear egress (lamins A/C, p32/gC1qR), cell cycle control (Rb, cyclins), intrinsic immune regulation (IFI16, SAMHD1) and transcription/translation (RNAP II, EF-1δ). The entity of this spectrum of pUL97-driven processes in virus-infected cells illustrates the functional importance of pUL97 for a high efficiency of viral replication, as demonstrated by the defects of recombinant viruses carrying UL97 deletions/mutations (up to factor 100–1000). Interestingly, the dimension of a replication defect resulting from drug-inhibited pUL97 was demonstrated to be more drastic in non-cycling compared to cycling cells [133], probably referring to the crosstalk and functional complementation between active cellular CDK-cyclin complexes and the vCDK. Moreover, the complex patterns of protein-protein interactions (PPI) undergone by pUL97 have recently been revealed by the use of highly sensitive mass spectrometry-based proteomic and phosphoproteomic approaches [56,66,82,98]. These findings make the occurrence of higher-order, pUL97-associated PPI complexes seem highly likely.

Most pUL97-related herpesviral kinases function as viral CDK orthologs (vCDKs). They were also termed conserved herpesviral protein kinases (CHPKs), as encoded by a gene conserved throughout the family Herpesviridae (e.g., prototype pUL97 and homologous kinases). Despite conservation of the UL97 gene locus, substantial variation of the primary coding sequence has been identified between herpesviruses. In addition to CHPKs, a second protein kinase is encoded by an additional non-conserved gene restricted to the subfamily α-Herpesvirinae (e.g., prototype pUS3 kinase of herpes simplex virus). CDK activity has been shown to be involved in multiple steps during HCMV infection [143]. vCDKs phosphorylate typical CDK substrates such as Rb and lamins A/C and show CDK activity in a yeast complementation assay [57,75,84,129]. The Saccharomyces cerevisiae mutant lacking activity of its sole CDK, cdc28, shows growth arrest in the early S/late G1 phase, which is overcome by CDK1 (human), pUL97 (HCMV), pU69 (HHV-6 and -7) and BGLF4 (EBV) expression [57]. In addition, pUL97 and CDK share substrate proteins, such as pUL69, RNAP II and EF-1δ [78,79,101,130]. Of note, pUL97 and CDKs phosphorylate Rb at the same residues (S780, S807, T821), leading to the inactivation of the cell cycle-inhibitory and tumor suppressor functions of Rb [75,144,145] (Table 4). In addition, the suppression of CDKs 1, 2, 5 and 9 by indirubin-derivatives increases the HCMV-inhibitory effect of maribavir (MBV), a potent pUL97 inhibitor [58]. Thus, pUL97 and CDKs possess at least partially overlapping functions.

 

The HCMV-encoded CDK ortholog pUL97 has significance in the therapy of HCMV infections, as it is responsible for the phosphorylation-mediated activation of GCV/VGCV, still representing the therapy gold standard and, similarly, additional nucleosides such as acyclovir (ACV), penciclovir (PCV) and others [69,113,174]. Hereby, the specific role of pUL97 is that nucleoside analogs have to be initially monophosphorylated in a step catalyzed by pUL97 kinase [68]. Thereafter, the active triphosphate metabolites have to be generated in a series of steps of further phosphorylation catalyzed by human guanylate kinase, dGMP kinase, phosphoglycerate kinase and potentially other host kinases [25]. In the triphosphate form, these analogs represent the active antiviral determinants, then acting as a substrate of the HCMV DNA polymerase, ultimately inhibiting the elongation of viral genome synthesis.

Genetic mutation studies showed that pUL97 plays a rate-limiting regulatory role for the replication efficiency of HCMV and virus titers were reduced by orders of magnitude when the coding sequence was disrupted [59,61]. Moreover, pharmacological inhibition of pUL97 activity by small molecules derived from various chemical classes blocked viral replication in a manner corresponding to the pUL97 null phenotype and thus proved to be a potent antiviral targeting strategy [175]. Since then, the pharmacologic inhibition of pUL97 activity together with genetic techniques have helped to characterize the mechanisms of pUL97 supporting the viral replication and virus–host kinase interactions [52,53,56,176]. A number of inhibitors of pUL97 kinase activity have been identified that exert potent antiviral activity against HCMV [64,119,175,177]. These include indolocarbazoles [51,109,120], quinazolines [64,123,178] and benzimidazole analogs [175] (Figure 3). A number of detailed investigations, both on cell culture-based in vitro and preclinical in vivo animal models, underlined the high value of this antiviral approach (reviewed in [25,53,132,179]). Thus far, however, with the exception of maribavir, none of these compounds has progressed to clinical studies.