A growing number of GPCR structures have facilitated the discovery of interacting small molecule compounds. For US28, the first compounds were discovered based on homology comparison to endogenous CKRs with known small molecule ligands as these were presented before the first US28 structure was solved. Several small molecule ligands acting as neutral antagonists or inverse agonists have displayed promising results in attenuating US28 signaling at micromolar concentrations (45). Among these, VUF2274 demonstrated the highest potency on US28 acting as an inverse agonist and interfered with CCL5 binding (45). The compound was observed to induce reactivation of latent HCMV infection potentially exposing it to the immune system (25), however, VUF2274 was originally discovered as a CCR1 antagonist (46) implying a selectivity issue if used as a drug. In search of potential drug candidates with limited cross-reactivity to endogenous receptors, two subsequent studies surveyed small molecule libraries based on VUF2274 (47, 48). Out of the latest study, several compounds emerged with agonistic or inverse agonistic profiles in Gαq-mediated signaling and capable of displacing CCL2 and CCL4, such as compound 56, 64 and 67 (48). These molecules were suggested as scaffolds for further development, but no advances on small molecules targeting US28 have been made since. Together, these studies demonstrate that US28, like other class A GPCRs, is highly targetable by small molecules though their clinical relevance as anti-HCMV therapeutics remains to be determined.

Apart from using small molecules to manipulate US28 activity, sdAbs are currently under investigation (49–52). Initially, a sdAb with sub-micromolar affinity to US28 was refined to create a bivalent sdAb with sub-nanomolar affinity. This compound partially inhibited ligand-dependent and constitutive US28 activity, leading to a reduction in US28+ GBM cell growth in vitro and in vivo (49), which shows therapeutic potential as US28 constitutive signaling can drive GBM proliferation (49). Subsequently, a US28-specific sdAb, VUN100, with nanomolar affinity for use in photodynamic therapy after conjugation with a photosensitizer was designed. Besides showing improved CX₃CL1 displacement compared to its predecessor, this compound displayed potent cytotoxicity in vitro on US28+ GBM cells (50). VUN100 was further refined into a bivalent version (VUN100bv) with improved affinity, acting as a partial inverse agonist inhibiting constitutive US28 signaling by 50% (51). This resulted in partial reactivation of HCMV in latently infected primary CD14+ monocytes, which lead to the hypothesis that VUN100bv could be used as a therapeutic in a “shock-and-kill” strategy where latent viral reservoirs are forced into lytic replication and subsequently killed by the host immune system. A fourth study described the generation of a sdAb (VUN103) targeting an intracellular epitope (“intrabody”) that through displacement of G proteins completely inhibits constitutive US28 signaling and attenuates spheroid growth of U251 glioblastoma cells (52). Together, this set of studies underlines the possibility of US28 targeting and modulation through sdAbs. In the clinic, a sdAb-based strategy could potentially be used to treat HCMV diseases through attenuation of US28 signaling, leading to partial viral reactivation. This could expose the latent infection and improve the immune system’s ability to combat the virus, potentially combined with existing anti-HCMV drugs.

A different approach to novel therapeutics for HCMV disease has focused on the development of a US28-specific FTP (an immunotoxin strategy) (53–55). In this case, the drug is not intended to modulate US28 signaling but rather to kill US28-expressing cells. To generate a US28-targeting FTP, the preferentially US28-binding chemokine CX₃CL1 was fused to a modified version of the Pseudomonas exotoxin A (PE) lacking the cell entry moiety (53). Taking advantage of US28’s constitutive internalization, the bound FTP is shuttled inside the cell where the toxin domain is released (53, 54). PE inactivates eukaryotic elongation factor-2 (eEF-2) by ADP-ribosylation which abolishes host cell protein synthesis, resulting in apoptosis and inevitable cell death (56). As CX₃CL1 also binds the endogenous receptor CX₃CR1, a mutated variant with high US28 selectivity (F49A-FTP) was generated. Exploiting the ubiquitous expression profile of US28 throughout both the lytic and latent cycle of HCMV infection, this FTP displayed potent and selective killing of infected cells in both stages (53, 55). The efficient elimination of HCMV-infected cells indicates a potential use in treatment of HCMV-associated diseases as demonstrated in patient-derived HCMV-infected CD34+ progenitor cells in vitro, forming the basis for a therapeutic strategy for eliminating latently infected cells before hematopoietic stem cell transplantation (55). Additionally, it showed efficacy on ganciclovir-resistant HCMV strains (53) thereby suggesting a use-case in clinical settings of treatment failure due to viral ganciclovir resistance. Following a successful trial-run (57), F49A-FTP was recently shown to reduce the load of latent HCMV by 80% in an ex vivo lung perfusion system (58), showcasing the potential for ex vivo elimination of HCMV in solid organ transplantations. These reports support a novel approach of eradicating latent virus reservoirs, which could prove particularly useful in organ transplantation settings provided improved clinical outcomes can be demonstrated.

The reports on US28-targeting compounds are promising but crossing the gap between laboratory observations and in human effects is notoriously challenging for HCMV. The virus is highly adapted and species-specific after millions of years of co-evolution with its host (59), making HCMV significantly different from other species’ CMVs in its genetic content and immune modulation. The lack of a proper animal model for replicating in vitro effects in vivo is a persistent challenge in the field. Transgenic animal models have been applied with success, e.g. insertion of HCMV US28 into murine CMV (53, 60–62), however, findings from transferring HCMV-specific genes to another species’ CMV are difficult to translate to humans and should always be considered with caution. Even though the distance between bench and bedside is increased by the lack of animal models for HCMV-associated diseases, US28 has the favorable position of a surface protein with basal, exploitable functionalities combined with homologies to endogenous class A GPCRs, which are inherently good drug targets (63).

Still, for transplantation-associated HCMV diseases, perhaps a better option is to utilize latently infected human organs unfit for clinical use in ex vivo systems (58). Here, the bigger hurdle is detecting and quantifying the latent HCMV load. Albeit not yet fully understood, latency is known to be established in a small fraction of CD34+ hematopoietic progenitor cells (HPCs) and CD14+ monocytes (64, 65), which are not abundant in most ex vivo organ settings. Additionally, gene transcription is minimal during latency (64). Reactivation assays have been described (58) but are time consuming, require steps of target cell extraction, viral reactivation, amplification using standardized cell cultures, and immunohistochemical staining of viral components, yielding more of an indirect measurement of HCMV activity. Modern techniques, such as RNA-seq, have shown promise in detecting latency transcripts (66) and may provide another approach to studying latency and ex vivo treatment effects. However, since reactivation is an inefficient process, genome- and transcript-based methods likely include abortive infections that will not reactivate. Alas, our current methods for detecting and quantifying the latent HCMV load and reactivation, and therefore evaluation treatment outcomes, are not ideal.

HCMV has a surprisingly diverse genome for a DNA virus displaying a high degree of sequence variability across many different genes including major immune modulators (67). For example, the chemokine-encoding UL146 gene is subject to extensive inter-strain diversity (68) that leads to structural and functional changes of the chemokine (69, 70). The US28 gene in contrast is quite conserved, strengthening its position as a therapeutic target (67). However, various genotypes have been observed (71–73), notably some with marked differences in the N-terminal (extracellular) tail of the receptor, which is important for chemokine binding. Indeed, molecular modeling has predicted changes in binding affinities of several endogenous chemokines to US28 variants (73). Variations of extracellular loops (ECLs) and the C-terminal (intracellular) tail have also been observed and, albeit less extensive, are not to be overlooked for their potential to alter US28 signaling. While the mechanism and biological significance are unclear, one study reported an increase in anti-CMV antibodies of renal transplant recipients carrying R267K-US28 (73). Additionally, antibody levels were reduced in HIV infected individuals carrying D170N-US28 and were accompanied by an increased HIV viral load and a reduction in sIFN-α/βR levels 12 months post initiation of anti-retroviral therapy (73). Functional differences between these naturally occurring variants remain unknown, but future research efforts exploring shifts in chemokine and drug binding along with signaling properties of the US28 variants will provide more knowledge. Furthermore, it is unclear to what extent these variants occur on a global scale as US28 genotyping studies and GenBank sequence deposits are limited. This combines to some uncertainties that should be addressed when progressing with US28-targeting drugs, as changes in drug affinities for US28 variants and downstream signaling can lead to altered drug effects. Additionally, treatment might induce US28 resistance mutations. These are risks that require clinical US28 sequencing before and during treatment to ensure and monitor the expected drug effects, however, clinically standardized tools to amplify and sequence US28 during latency is currently not available. Thus, these unknowns require attention when transitioning from lab to clinic.

The US28-targeting strategies discussed here rely on two distinct modes of action (MOA) ( Figure 1 and Table 1 ). For small molecules and sdAbs, a “shock-and-kill” strategy has been proposed where HCMV infection is forced from latent to lytic phase. This is achieved by inhibition of constitutive US28 signaling with an inverse agonist that leads to activation of MIEP which initiates viral reactivation (74). Once reactivation is induced, this strategy relies on the host immune system or a combination treatment with a replication inhibitor to clear the infection (25). This will in theory allow clearance of HCMV infected cells before dissemination of infection. While elegant in its conception, this strategy rises some safety concerns if used in immunocompromised patients. Forcing HCMV reactivation requires a degree of control over the infection that currently is not always possible as seen in patient groups where infection can flare up despite administration of SOC prophylaxis (15), such as transplant recipients. This potentially limits the usefulness of the “shock-and-kill” strategy to patients with somewhat competent immune systems, which could be envisioned in an early treatment for US28+ GBM tumors, or to reduce the latent HCMV load in other cancer patients and R+ transplant recipients prior to immunosuppressive therapy.

The MOA of sdAb-photosensitizer conjugates and FTPs is cell toxicity, which does not rely on a competent immune system or a combination treatment to finish the job. However, this advantage requires a highly US28-specific ligand domain to limit adverse toxic effects emerging from off-target binding. Indeed, the promiscuous binding of US28 (26) might make it an easier target but also increase the risk of non-specific compound effects from off-target receptors. Promisingly, F49A-FTP did not induce acute lung injury or changes in cytokine levels in an ex vivo lung perfusion setting (58). As for other immunotoxins, potential compartmentalization of the FTP combined with release of the toxin moiety will need to be addressed to ensure its safety. While the sdAb-photosensitizer conjugate has the benefit of requiring site-directed light activation, peripheral effects of reactive oxygen species resulting from photodynamic therapy could affect its viability in some cancer and transplantation settings. Lastly, clearing latent infection from donor organs prior to transplantation (55, 58) raises the question to what extent the viral reservoirs need to be reduced to influence the post-transplantation outcome. While early studies show that there is an association between the latent viral load and risk of recurrent infection (75, 76), it is unclear how the size of a latent HCMV reservoir in a donor correlates with the risk of reactivation and disease in the recipient, but it is not unthinkable that a small persistent HCMV pool can flare up to clinical significance in vulnerable patients. Overcoming these potential clinical challenges will be key to progression.