Herpesviruses are enveloped viruses with large, double-stranded DNA genomes (1
). Herpes simplex viruses (HSVs) replicate lytically in epithelial cells near their site of entry into the body (2
). The virus then enters sensory nerves and establishes latent infection of the neurons, where it remains latent for the life of the host. Episodic reactivation from latency causes lytic replication at mucosal surfaces, triggering recurrent disease and providing the opportunity for transmission to uninfected individuals (3
The eight human herpesviruses cause an array of severe diseases associated with primary and recurrent infections. Herpes simplex virus 1 (HSV-1) and HSV-2 are closely related viruses with colinear genomes. HSV-1 infects more than half of all Americans (4
) and causes gingivostomatitis, cold sores, encephalitis, herpetic stromal keratitis, and an increasing proportion of anogenital lesions (2
). HSV-2 infects nearly one in five of Americans (4
) and is the primary cause of ulcerative anogenital lesions (4
). It also increases the risk of human immunodeficiency virus (HIV) acquisition (7
). HSV-1 and HSV-2 can be transmitted from a pregnant woman to her child during birth, often causing potentially fatal disseminated disease in the newborn (9
Treatment of herpesvirus infections primarily relies on nucleoside analog inhibitors of the viral DNA polymerase, including acyclovir (ACV), penciclovir, ganciclovir, valaciclovir, valganciclovir, brivudine, and famciclovir (10
). Several newer agents are undergoing clinical development (11
), but none of them can fully suppress herpesvirus infections (12
). Viral strains resistant to the current drugs exist and are especially common in immunocompromised individuals (11
), but they are also significant in patients with ocular infections and in children (14
). Cross-resistance to existing nucleoside analog drugs is common, because these drugs depend on the viral thymidine kinase (TK) and/or polymerase for their efficacy (18–20
). Thus, new drugs with a different mechanism of action are needed.
HSV genomic replication employs numerous viral enzymes. Replication (21
) begins when the viral linear double-stranded DNA genome circularizes shortly after infection, possibly via recombination (23
). DNA replication initiates at one or more of three viral origins of DNA replication and is primed by the viral helicase-primase complex (HSV-1 proteins pUL5, pUL8, and pUL52). DNA replication requires the single-stranded DNA-binding protein pUL29 (ICP8), which is predicted to contain an RNase H-like fold (25
). The viral DNA polymerase holoenzyme complex (pUL30 DNA polymerase plus pUL42) catalyzes DNA elongation by a presumed double-stranded rolling-circle mechanism. This complex possesses 5′-3′ exonuclease, 3′-5′ exonuclease, and RNase H activities (26
). DNA replication initially produces head-to-tail linear concatemers, and branched concatemers accumulate later in the replication cycle through recombination and/or reinitiation mechanisms. Formation of the branched forms via recombination is stimulated by the pUL12 exonuclease (27
). Last, the viral terminase complex (pUL15, pUL28, and pUL33) cleaves the viral DNA to unit length during packaging of the genome into viral capsids, and the crystal structure of pUL15 shows an RNase H-like fold (28
The nucleotidyltransferase superfamily (NTS) is a group of diverse enzymes whose members share a similar protein fold and enzymatic mechanisms (29
). These enzymes function in nucleic acid metabolic events, including RNA and DNA digestion, DNA recombination, DNA integration, replication fork repair, DNA repair, and microRNA (miRNA) maturation and function. NTS enzymes include Escherichia coli
RNase H 1 and 2 (30
), human RNase H 1 and 2 (32
), the RuvC Holliday junction resolvase (34
), the Argonaute RNase (35
), the hepatitis B virus (HBV) RNase H (36–38
), the HIV RNase H (39
), and the HIV integrase (40
). This class of enzymes is characterized by the spatial arrangement of three or four conserved carboxylates (the DDE or DEDD motifs) in their active sites. These carboxylates coordinate two divalent cations essential for the nucleic acid cleavage activities of the enzymes (29
). The HSVs express numerous proteins involved in nucleic acid metabolism, and they upregulate expression of cellular proteins that assist viral DNA replication. Many of these enzymes, especially the nucleases, have catalytic properties consistent with NTS members.
The HIV RNase H and integrase have attracted much attention as potential drug targets, with hundreds of compounds being reported to inhibit these enzymes (43
). Most of the inhibitors bind at the active site via interactions with the divalent cations (45–48
). Three anti-HIV integrase drugs have been approved by the Food and Drug Administration (FDA) (raltegravir, elvitegravir, and dolutegravir), but no anti-RNase H drugs have yet been approved. As predicted from their common NTS membership (29
) and similar orientation of active site cations (49
), some anti-HIV RNase H compounds inhibit the integrase, and some anti-integrase compounds inhibit the RNase H (46
). Furthermore, we recently demonstrated that the HBV RNase H is sensitive to many anti-HIV RNase H and integrase antagonists (36–38
). This capacity of NTS enzyme inhibitors to cross-inhibit distant superfamily members led us to hypothesize that HSV-1 and HSV-2 would be sensitive to NTS inhibitors because enzymes with activities typical of this superfamily play multiple critical roles during HSV replication.
The complex herpesvirus DNA replication mechanism uses enzymes with catalytic properties which imply that they would be sensitive to NTS enzyme inhibitors. Therefore, we screened 42 compounds from six chemical families that were known or predicted to inhibit HIV and/or HBV NTS enzymes for the ability to suppress HSV-1 and HSV-2 replication. Eighteen compounds (43%) inhibited HSV-1 replication by ≥10-fold in short-term cell culture experiments in Vero cells at ≤50 μM, and 19 inhibited HSV-2 (45%) (Table 1
). Six of the compounds that inhibited HSV-1 or HSV-2 at 5 μM were tropolones (compounds 46, 49, 50, 55, 56, and 59), and one inhibitor was found in each of the polyoxygenated heterocycle (compound 41) and hydroxyxanthenone (compound 30) classes. Four of these compounds inhibited HSV-1 and HSV-2 replication in Vero cells and HFFs by at least 100,000-fold at 5 μM. No hits were found among the other chemical families (see Table S1 in the supplemental material and Fig. 1
). All compounds that inhibited HSV-1 or HSV-2 by ≥1 log10
unit in Vero cells were also active in HFFs (Fig. 2
and Table 1
), demonstrating efficacy against HSV replication in a relevant human cell type. This high hit rate for a small primary screen provides support for the hypothesis that NTS enzymes are promising targets for antiherpesvirus drug development.
These NTS enzyme inhibitors function against HIV by binding to the RNase H or integrase enzymes in part through binding to the essential divalent cations within the active site (45–48
). Therefore, their presumed mechanism of action against the HSVs is to inhibit one or more NTS enzymes essential for viral genomic replication by binding to their active site(s) and masking the cations. The proposed mechanism, however, cannot be tested until the target enzyme(s) is identified.
Sufficient data exist to establish constraints on the structure-activity relationship for the tropolones against the HSVs. Comparing compound 46 with compounds 47, 48, 50, and 53 (see Fig. S1 in the supplemental material) indicates that efficient inhibition in the absence of extended R groups at the α, β, and γ positions of the tropolone ring requires three adjacent cation-binding moieties (the contiguous hydroxyl and carbonyl groups). Of the six compounds with larger R groups (compounds 49, 52, 55, 56, 59, and 62), four suppressed viral replication by ≥1 log10 unit at 5 μM (compounds 49, 55, 56, and 59). Three of these inhibitory compounds (compounds 49, 55, and 59) had only two metal-binding moieties on the tropolone ring, but compound 56 had three binding moieties and was the best inhibitor identified. These data imply that a carbonyl group and a modified hydroxyl group are insufficient to support robust inhibition and that three metal-binding moieties are superior to two. Last, four of the most effective inhibitors (compounds 49, 55, 56, and 59) had a variety of R groups opposite the metal-binding motif. This indicates that structural diversity is permitted in these elements, and the wide range of EC50s for the four most effective inhibitors (0.24 to 1.09 μM for HSV-1 and 0.22 to 4.12 μM for HSV-2) implies that the chemical elements in these positions can significantly impact compound efficacy.
Little to no cytotoxicity was observed for the HSV inhibitors in short-term experiments by either the MTT assay that measures mitochondrial function or by assessing cell rupture (Table 1
; Table S1 in the supplemental material; data not shown). Compounds 50 and 61 caused some toxicity in HFFs, but all other CC50
values were higher than the 50 μM maximal concentration used for screening. This indicates that cytotoxicity is not a confounding factor in our screening, and it is promising with regard to the drug potential of these compound classes. However, we do not wish to imply that toxicity will not be a concern during subsequent drug development. Substantial toxicity was detected for compounds 41, 56, and 59 in HFFs in the longer-term HCMV assays (Table 2
). Furthermore, tropolones can induce mitochondrial toxicity in rats (64
), and compound 46 has CC50
values of 2 to 16 μM in CEM-SS, Huh7, and HepG2 cells (37
), indicating that optimization to limit cytotoxicity in various cell types will be warranted.
Our data imply that the NTS inhibitors block more than one event in the viral replication cycle. The time-of-addition experiment (Fig. 4
) revealed that compounds 41 and 46 interfere with events occurring between 1 and 12 h postinfection. The partial loss of activity when addition of compounds 41 and 46 was delayed until 3 or 5 h postinfection is in sharp contrast to the lack of effect that delaying ACV addition had during this time. The NTS inhibitors therefore suppress an activity important for viral replication that occurs prior to the onset of viral DNA replication, perhaps genome circularization. This is also in sharp contrast to a recently identified HIV integrase inhibitor that also suppressed HSV replication, whose earliest detectable effect is at the stage of viral DNA replication (66
). The continued gradual loss of activity if addition of compounds 41 and 46 was delayed more than 1 h postinfection suggests that these NTS inhibitors additionally inhibit events that occur concurrently with viral DNA replication, possibly including DNA replication and primer removal, production of branched genomic concatemers, and/or genome monomerization. Alternatively, because viral infection and replication are somewhat asynchronous, the compounds may inhibit a very early event such as genome circularization, but the inhibitory effect appears to gradually diminish over time because a progressively larger proportion of the viral population has already initiated infection prior to compound addition at later times.
It is unknown whether the events suppressed by these NTS inhibitors represent a single target whose activity is needed at multiple stages of the viral replication cycle or multiple targets that are each needed for viral replication. However, the high hit rate in this small targeted screen, the high efficacy of the inhibitors, and the gradual loss of activity in the time-of-addition experiments lead us to favor the possibility that the NTS inhibitors act against multiple enzymes that function at different stages of the replication cycle. In this context, the gradual loss of activity with increasing time of addition would be due to exposure to compound after one or more of the targets had performed their functions. Alternatively, the events controlled by these enzymatic function(s) may occur less abruptly than onset of viral DNA replication. If antiviral efficacy is due to activity against multiple targets, it would suggest a higher barrier to resistance evolution than is often observed for drugs with a single target.
The target(s) of the NTS inhibitors could be viral and/or cellular. Candidate HSV gene products with NTS enzyme activities include the predicted RNase H activity of the ICP8 single-stranded DNA-binding protein (25
), the RNase H activity of the pUL30 DNA polymerase (67
), the 3′-5′ exonuclease activity of pUL30 (68
), and the 5′-3′ exonuclease activity of the pUL12 polymerase accessory protein (27
) that are directly involved in virus DNA replication (21
). The pUL15 terminase that cleaves the viral DNA into monomers is known to be an NTS enzyme and is also a possible target (28
). Cellular candidates include human RNase H 1 and the Fen1 endonuclease that may remove RNA primers during DNA synthesis (69
), and the double-stranded break repair enzymes Mre11, Rad50, NBS1, Rad51 (70
), and Rad52 (27
). The base excision repair enzymes SSH2 and MLH1 (71
) that form complexes that are recruited to viral replication sites and contribute to HSV genomic replication (71
) are also possible targets.
Our data indicate that the NTS inhibitors tested work by a different mechanism than the approved nucleos(t)ide analog antiherpesvirus drugs that terminate DNA elongation. We demonstrated that HSV-2 strains resistant to ACV are sensitive to NTS inhibitors in this targeted screen (Fig. 5
) and that none of the NTS antagonists employed are nucleos(t)ide analogs (see Fig. S1 in the supplemental material). Furthermore, unlike most nucleos(t)ide analogs, compounds 30, 41, and 46 do not require phosphorylation by the viral thymidine kinase to be active (Fig. 6
). This implies that NTS inhibitors could be used as salvage therapies for nucleoside analog-resistant infections, and it raises the possibility of additive or synergistic activity of NTS and nucleoside analog drugs that could significantly improve therapeutic efficacy. It also suggests that combination therapy employing nucleoside analogs and drugs that inhibit NTS enzymes could reduce the rate at which resistance would evolve to either drug class.
ACV and ganciclovir are standard treatments and prophylactic agents for herpesvirus infections. Recently, approval of other drugs (i.e., penciclovir, brivudine, valaciclovir, famciclovir, cidofovir, fomivirsen, and foscarnet) for one or more herpesviruses has increased therapeutic options. In addition, an inhibitor of the HSV-2 helicase-primase complex, pritelivir, is in clinical development (72
). Despite these expanded options, new antiherpesvirus agents are still needed to further improve efficacy and to address other clinical issues, such as safety in children and pregnant women, increased use of antiherpesvirus prophylaxis including in transplant recipients, and resistance to nucleos(t)ide analogs. Recently, Yan et al. observed that select HIV integrase inhibitors suppress replication of herpesvirus family members (66
). Specifically, two HIV integrase inhibitors reduced HSV replication in HEp-2 cell cultures by up to ∼3.9 log10
units (8,000-fold) at 10 μM (66
). In contrast, the inhibitors we identified suppress HSV replication by up to 5.95 log10
units (890,000-fold) at 5 μM and were originally identified as HIV RNase H inhibitors. Although cross-inhibition of the HIV integrase and RNase H by certain compounds occurs (46
), we do not know whether HIV integrase inhibitors act against the same or different targets as the HIV RNase H inhibitors during suppression of HSV replication. A significant difference between the assays used in the two studies is measurement of inhibition under one-step versus multistep growth conditions. Comparison of the two protocols revealed that our single-step assay likely underestimates the efficacy of the RNase H inhibitors because virus in the DMSO control cultures has not yet reached maximal titer. Thus, the high efficacy of the primary screening hits that we identified among three different chemical families of RNase H inhibitors, coupled with their different mechanism(s) of action relative to the nucleoside analog inhibitors, makes the members of the RNase H antagonist subclass of NTS inhibitors attractive candidates for novel antiherpesvirus therapies to complement the existing drugs.
The herpesvirus family contains many viruses beyond HSV-1 and HSV-2 of major medical or veterinary importance. Human pathogens include varicella-zoster virus (VZV) (73
), which causes chicken pox and shingles; HCMV, which causes mental retardation and deafness in neonates and retinitis in immunocompromised patients (63
); human herpesvirus 6 (HHV-6), which causes roseola infantum and febrile seizures (74
); Epstein-Barr virus (EBV), which causes infectious mononucleosis and is associated with cancers, including Burkitt's lymphoma (75
); HHV-7, which may stimulate HCMV reactivation and tissue transplant rejection (76
); and HHV-8, which causes Kaposi's sarcoma (77
). More than 90% of adults have been infected with and retain a latent form of one or more of these viruses. Animal herpesviruses of significant economic importance include pseudorabies virus (78
), Marek's disease virus (80
), bovine herpesviruses (BHV) (81
), and herpes simian B virus (82
). Importantly, these viruses all share the same basic genomic replication mechanisms, so they should be sensitive to NTS inhibitors if the presumed mechanism by which the NTS enzymes inhibit HSV-1 and HSV-2 is correct. This prediction is at least partially confirmed by our detection of anti-HCMV activity for five out of the six best NTS HSV inhibitors identified here and the recent identification of an HIV integrase inhibitor, XZ45, as an inhibitor of HCMV replication (66