Restriction of Human Cytomegalovirus Infection by Galectin-9

Gal-9 inhibits HCMV infection of multiple cell types.

We sought to assess the functional consequence of Gal-9 upregulation by evaluating the impact of soluble Gal-9 on HCMV infection in permissive cells, given that exogenous galectins can both promote and inhibit a number of other viral infections (14, 15). Previous work from our laboratory has established that Gal-9 is upregulated during HCMV infection and is dependent upon interferon beta (IFN-β) induction of Gal-9 mRNA (39). We therefore sought to assess the functional consequence of Gal-9 upregulation by testing whether soluble Gal-9 could directly regulate HCMV infection. A green fluorescent protein (GFP)-expressing HCMV (Merlin-GFP) was pretreated with recombinant Gal-9 at a range of concentrations (0.25 to 100 nM) for 30 min prior to infection of human foreskin fibroblasts (HFs) at a multiplicity of infection (MOI) of 0.5. The extent of infection was assessed by flow cytometry at 72 hours postinfection (h p.i.), allowing the fold change in the percentage of infected cells to be determined. Representative scatter plots depict the percentage of GFP-positive cells in Merlin-GFP-infected cells compared to mock-infected cells (Fig. 1A). The addition of Gal-9 at each concentration tested (12.5 to 100 nM) significantly reduced the number of GFP-positive cells in a dose-dependent manner (Fig. 1C), with up to 88% inhibition at the highest concentration of Gal-9. In contrast, treatment of HCMV with Gal-1 (12.5 to 100 nM) did not significantly alter infection (Fig. 1B). These results indicate that Gal-9 is able to inhibit HCMV infection of HFs and that this is not a general property of all galectin protein family members.

Gal-9 has been associated with the induction of apoptosis in a number of cell types (40). Therefore, we assessed the ability of Gal-9 to induce cell death in HFs to determine whether the reduction in the number of HCMV-GFP-positive cells was due to the cytotoxicity of target cells, rather than infection per se. HFs were treated with Gal-9 (50 to 400 nM) for 90 min, and cell viability was determined by flow cytometry 24 h later using a LIVE/DEAD indicator dye (Zombie-NIR). In comparison to untreated cells, there was no significant difference in the percentage of live cells following Gal-9 treatment at any concentration tested (Fig. 1D). Thus, the Gal-9-mediated inhibitory effects on HCMV infection are not due to Gal-9-mediated cell death.

To determine whether the capacity of Gal-9 to inhibit HCMV infection was virus strain specific, another strain of HCMV, TB40, was assessed. TB40 was selected as it retains expression of a functional UL128-131A locus (UL128L), whereas the bacterial artificial chromosome (BAC)-derived Merlin recombinant described above has a mutated UL128L rendering it nonfunctional (41). An intact UL128L pentamer complex is nonessential for efficient HCMV infection of HFs; however, it is necessary for infection of a range of key in vivo cellular targets, including epithelial cells (42). A BAC-derived version of TB40 that had been engineered to express GFP (TB40-GFP) was used to monitor infection (43), with expression of GFP detected at 48 h p.i. by flow cytometry. Gal-9 significantly inhibited TB40 infection of HFs in a dose-dependent manner (Fig. 1E), demonstrating that the effect of Gal-9 on virus infection was not limited to the Merlin strain.

As entry mediated by the UL128L pentamer complex into epithelial cells uses a different mechanism from that with entry into HFs (endocytosis rather than direct fusion with the plasma membrane) with some viral glycoproteins key to both processes (44, 45), we expanded our analysis of Gal-9-mediated inhibition of TB40-GFP infection to retinal pigment epithelial 1 (RPE-1) cells. This analysis demonstrated a significant inhibition of TB40-GFP infection of RPE-1 cells following Gal-9 (25 to 200 nM) treatment (Fig. 1F). Taken together, these results indicate that Gal-9 can inhibit infection of HFs with low-passage-number strains of HCMV and that Gal-9 can inhibit HCMV infection of multiple clinically relevant cell types (in both an UL128L-dependent and -independent fashion).

Gal-9 lectin binding is required to inhibit HCMV infection.

To determine whether Gal-9 directly inhibited HCMV infection, Gal-9 was pretreated with a neutralizing (26) or isotype control antibody for 30 min prior to incubation with HCMV (Merlin-GFP) and infection of HFs (MOI, 0.5). Cells were then analyzed by flow cytometry at 72 h p.i. for GFP expression to assess the impact on infection efficiency. We found that pretreatment of Gal-9 with neutralizing antibody resulted in a significant reduction in its ability to restrict HCMV infection (P = 0.041, Fig. 2A), while isotype antibody treatment had no effect. As expected, direct treatment of HCMV virions with either the anti-Gal-9 or isotype antibodies had no impact on infection. Taken together, these results support a mechanism where Gal-9 binds directly to HCMV virions to inhibit their infectivity.

Galectins can mediate their effector functions by binding lectins or via protein-protein interactions. Lectin interactions are mediated by the carbohydrate recognition domain (CRD) binding to β-galactoside-containing carbohydrate ligands (46). To assess whether Gal-9-mediated inhibition of HCMV was dependent on the CRD, Gal-9 was pretreated with the disaccharide lactose (1.25 to 5 mM), which specifically blocks the Gal-9 CRD (38), for 30 min prior to incubation with HCMV (Merlin-GFP). As a control, Gal-9 was pretreated with sucrose (1.25 to 5 mM), which is a disaccharide that lacks the β-galactoside moiety required for binding to the Gal-9 CRD. Flow cytometric analysis of HCMV Merlin-GFP expression demonstrated that lactose, but not sucrose, significantly abrogated Gal-9-mediated inhibition of HCMV infection in a dose-dependent manner, and lactose or sucrose alone had no effect on HCMV infection in the absence of Gal-9 addition (Fig. 2B and C). Thus, the inhibition of HCMV infection by Gal-9 is dependent on carbohydrate binding by this lectin.

Gal-9 inhibits HCMV infection primarily through virion and not host cell binding.

As a number of cell membrane-associated ligands for galectins have been identified (47), we sought to determine whether Gal-9-mediated inhibition of HCMV infection resulted from Gal-9 binding to the host cell. HFs were pretreated with Gal-9 for 30 min, unbound Gal-9 was then removed by washing, and cells were infected with HCMV Merlin-GFP. In parallel, HCMV (Merlin-GFP) was pretreated with Gal-9 as before, and infection was assessed by flow cytometry. Pretreatment of cells with Gal-9 prior to infection had no impact on infection (Fig. 3A), while as expected, Gal-9 treatment of virions significantly decreased it (Fig. 3A). As a further test, we determined whether Gal-9 was able to inhibit HCMV when added to already-infected HFs. Cells were infected at an MOI of 0.5 for 90 min before being treated with Gal-9 for 24 h. In contrast to the inhibitory effects of Gal-9 when pretreating HCMV, incubation of preinfected cells had no significant impact on HCMV-driven GFP expression (Fig. 3B). Therefore, treatment of the virus inoculum prior to infection was at least in part necessary for the observed Gal-9-mediated inhibition of HCMV infection of HFs.

Our conclusions thus far that Gal-9 inhibited HCMV infection were based upon the detection of GFP-tagged HCMV strains where GFP is expressed from a viral IE/E promoter, and therefore, we could not report whether Gal-9 might also have an impact on infectious virus production. To assess this, we assayed the ability of Gal-9 to inhibit HCMV plaque formation. HCMV (Merlin-GFP) was pretreated with Gal-9 before infection of HFs (MOI, 0.001), and following a 90-min incubation, cells were then washed with either phosphate-buffered saline (PBS) (to remove unbound virus) or citrate buffer (to inactivate noninternalized virus). Gal-9 treatment of HCMV followed by PBS washing did result in a modest, but significant, decrease in plaque number at 12 dpi (Fig. 3C). However, washing with the more stringent citrate buffer had a dramatic impact on plaque formation, with a near-complete inhibition seen in Gal-9-treated virus (Fig. 3D).

Gal-9 inhibits HCMV by restricting viral fusion.

To elucidate at what point during viral infection Gal-9 mediated its inhibitory activity, we examined its impact on HCMV immediate early protein expression (48). HCMV (Merlin-GFP) was treated with Gal-9 for 30 min before infection of HFs seeded onto coverslips (MOI, 0.5). Cells were fixed, immunostained for immediate early protein-1 (IE-1) at 24, 48, and 72 h p.i., and visualized using fluorescence microscopy. There were no cells with IE-1 staining in the mock or isotype control samples, whereas IE-1 staining was observed in all samples infected with HCMV at 24 h p.i. (Fig. 4A). Quantification of the proportion of IE-1-positive cells revealed that Gal-9 treatment resulted in a significant decrease in the percentage of IE-1-positive cells at 24, 48, and 72 h p.i. (Fig. 4B), compared to HCMV control infection (24 h p.i, P = 0.048; 48 h p.i., P = 0.016; and 72 h p.i., P = 0.005). Thus, quantification of HCMV infection of HFs by immunofluorescence microscopy provided confirmation of Gal-9-mediated inhibition of infection and also indicated that Gal-9 inhibitory activity occurs early in HCMV infection.

The relative amount of viral DNA in infected cells, with or without Gal-9 treatment of HCMV, was then determined. Noninternalized virions were inactivated once again with citrate buffer, and whole cells were harvested at 4 h p.i. to determine the efficiency of HCMV infection/internalization but prior to the onset of viral genome replication (Fig. 4C). This analysis revealed that Gal-9 treatment significantly decreased the amount of internalized viral DNA compared to HCMV (Merlin-GFP) infection without Gal-9 treatment (n = 3, P = 0.032). Thus, the inhibitory activity of Gal-9 occurs very early in HCMV infection, prior to IE-1 expression and before internalization of the viral genome into the cell.

HCMV entry into HFs is mediated through a multistep process involving attachment of viral glycoproteins with cellular receptors, followed by direct fusion of the viral membrane with the plasma membrane (9, 44). Attachment is largely an energy-independent process, while fusion is energy dependent (49). Thus, to further explore how Gal-9 inhibited HCMV entry, temperature shift studies were performed to enable the separation of the attachment and fusion stages of infection. HCMV (Merlin-GFP) was added to HFs at 4°C for 90 min to allow virus attachment but not fusion (50). Nonadhered virus was then removed with PBS washing, Gal-9 was then added at 37°C for a further 60 min to allow for Gal-9 being present at times postbinding but prefusion, and noninternalized virus was then inactivated with citrate buffer (Fig. 5A). Notably, the addition of Gal-9 during the postbinding fusion step of infection significantly inhibited HCMV infection (Fig. 5B). In addition, lactose-mediated blockade of Gal-9 binding reversed this inhibition, with sucrose showing no impact (Fig. 5C), indicating CRD-dependent restriction of fusion.

Gal-9 secretion is upregulated in response to HCMV in vivo.

To assess the potential clinical relevance that soluble Gal-9 might have on HCMV pathogenesis in people, we assessed secretion of Gal-9 in an HCMV-infected patient cohort. In allogeneic HSCT recipients, HCMV reactivation resulting from donor (D) and/or recipient (R) HCMV seropositivity is associated with adverse outcomes posttransplantation, including severe morbidity and mortality (4). Sequential plasma samples were collected weekly from HSCT patients, and the levels of secreted/soluble Gal-9 protein were quantified. HSCT recipients were prospectively monitored for the development of HCMV reactivation by quantitative PCR (qPCR; Roche Cobas AmpliPrep CMV test). In three HSCT patients who developed posttransplant HCMV reactivation (HCMV-positive reactivators), Gal-9 concentrations were assessed by enzyme-linked immunosorbent assay (ELISA) in plasma samples taken at the initial detection of HCMV reactivation (T1), the peak of reactivation (T2), and during subsequent control of reactivation (DNAemia) following antiviral drug therapy (T3) (Table 1). In parallel, the plasma concentration of Gal-9 was assessed at matched days posttransplant from three HSCT recipients where both the donor and recipient were HCMV seronegative (D⁻/R⁻), as well as from three HSCT recipients where either the donor and/or recipient was HCMV seropositive but in whom no reactivation was detected (HCMV-positive nonreactivators).

Across the three time points examined, there was no significant change in plasma Gal-9 concentration in either of the two groups of HSCT recipients experiencing no HCMV reactivation (Fig. 6). In contrast, HSCT recipients who developed HCMV reactivation exhibited a significant increase in plasma Gal-9 levels which was associated with the peak of HCMV reactivation (T2, mean day 53 posttransplant). At this time point, the plasma Gal-9 concentration was significantly higher in patients with HCMV reactivation (24.4 ng/ml) than in both HCMV-negative nonreactivators (mean, 8.8 ng/ml; P = 0.0001) and HCMV-positive nonreactivators (mean, 14.1 ng/ml; P = 0.006). There was also a significant difference in Gal-9 plasma levels for HCMV-seropositive reactivators at the peak of reactivation (T2) compared to both its initiation (T1, P = 0.0015) and eventual control (T3, P = 0.0004). The concentration of Gal-9 at T3 in HCMV-seropositive reactivators was comparable to that detected at initial detection of reactivation (T1) and similar in patients with or without HCMV reactivation at these times. We also assessed the secretion of Gal-9 in response to in vitro HCMV infection of HFs. HCMV infection of HFs resulted in a significant increase in secreted Gal-9 at 48, 72, and 96 h p.i. compared to mock (Fig. 7A) and was recapitulated with interferon treatment alone, as expected (Fig. 7B) (39). Taken together, our results indicate that soluble Gal-9 has a high potential to regulate HCMV infection both in vitro and in vivo in the context of inflammatory responses and incidences of high viremia in immunosuppressed patients.

Plasma from HSCT patients.

Plasma was collected from whole-blood samples drawn from adult allogeneic HSCT recipients (Westmead Hospital, Australia) and stored at −80°C. Samples were collected from each recipient at weekly intervals for the first 100 days posttransplant, with HCMV serostatuses of recipients and donors assessed in the clinic prior to transplantation. Samples were collected from both HCMV donor (D)-/recipient (R)-seronegative (D⁻/R⁻) and HCMV-seropositive (D⁻/R⁺, D⁺/R⁻, D⁺/R⁺) HSCT recipients (Table 1). Recipients were prospectively monitored for the development of HCMV reactivation by qPCR (Roche Cobas AmpliPrep CMV test). HCMV reactivation was defined as at least 150 copies of HCMV DNA per 1 ml plasma in at least two consecutive samples. All seropositive recipients (R⁺) were administered ganciclovir or valganciclovir prior to transplantation, and all patients with HCMV reactivation selected for this study were treated with ganciclovir or valganciclovir in the case of posttransplant HCMV reactivation. Time points posttransplant for the assessment of Gal-9 by ELISA were determined based on the three patients who developed HCMV reactivation posttransplant (HCMV-positive reactivators), with plasma Gal-9 concentrations assessed at the initial detection of HCMV reactivation (T1, mean 38 days posttransplant), the peak of reactivation (T2, mean 53 days posttransplant), and the control of reactivation (T3, mean 84 days posttransplant). Control of reactivation was defined as the first time point postreactivation at which plasma HCMV DNA returned to a level below 150 copies per ml of plasma. Plasma samples for nonreactivating groups (HCMV negative and HCMV-positive nonreactivators) were assessed at matched days posttransplant to the HCMV reactivation group. Plasma Gal-9 concentrations were determined by ELISA (R&D Systems) according to the manufacturer’s instructions. This involved a 10-fold dilution of plasma samples in the provided calibration diluent (RD6-35). This dilution was within the recommended dilution range and ensured that all concentrations detected were within the detection limit of the kit used (0.16 to 10 ng/ml).

Cell culture and virus stocks.

Human foreskin fibroblasts (HFs) and retinal pigment epithelial 1 (RPE-1) cells were obtained from the ATCC and grown at 37°C and 5% CO₂ in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) and penicillin-streptomycin (100 units/ml). The UL32 GFP-expressing variant of the low-passage-number clinical isolate of Merlin was generated from a bacterial artificial chromosome (BAC) (41). The GFP-expressing TB40 used was also BAC derived (43). Virus stocks were propagated in HFs, and cell culture supernatants from virus-infected cells were centrifuged at 845 × g for 10 min to remove cellular debris before ultracentrifugation at 21,875 × g for 2 h to pellet virus (Sorvall WX+ ultracentrifuge; Thermo Fisher Scientific). Virus pellets were resuspended in fresh supplemented DMEM and stored at ₋80°C. Viral titers were determined by a plaque assay. Briefly, 1.5 × 10⁵ HFs were infected in six-well plates and overlaid with a 1:1 mixture of 2× Eagle’s minimal essential medium (EMEM) supplemented with l-glutamine (4 mM), 20% FCS, and penicillin-streptomycin (200 units/ml), combined with 2% Avicel. Plates were incubated at 37°C and 5% CO₂ for 12 days, plates were then washed with PBS, and plaques were enumerated under inverted light microscope (Axio, Zeiss).

Viral infections and galectin treatments.

HFs and RPE-1 cells were seeded in a 12-well plate (Corning Costar; Sigma-Aldrich) at 7.5 × 10⁴ and 1 × 10⁵ cells/well, respectively. For testing the inhibitory effects of galectins on HCMV, recombinant Gal-1 (R&D Systems) or Gal-9 (R&D Systems or BioLegend) was diluted in serum-free DMEM and mixed with HCMV at an appropriate MOI (0.5 for HFs, 1 for RPE-1) for 30 min at room temperature. The virus-galectin mixture was added to cells for 90 min at 37°C. Virus-containing medium was removed, cells were washed with PBS, and fresh supplemented DMEM was added to each well.

Flow cytometry.

For flow cytometric analysis of GFP expression, cells were detached with trypsin-EDTA at 72 h p.i. for Merlin-GFP infected HFs and 48 h p.i. for all TB40-GFP infected cells. Cells were washed with flow cytometry buffer (PBS with 1% FCS and 10 mM EDTA) and fixed with 1% paraformaldehyde (PFA). Flow cytometry was performed (LSRFortessa X-20; BD Biosciences), and data were analyzed with FlowJo (version 10; Tree Star, Inc., Ashland, OR). GFP expression data were normalized as the fold change in the percentage of infected cells (GFP positive) compared to untreated HCMV.

Cell viability.

HFs were seeded in a 12-well plate and treated with various concentrations of Gal-9 (25 to 400 nM) in serum-free medium for 90 min at 37°C. Medium was removed, cells were washed in PBS, and supplemented medium was added. Cell viability was assessed at 24 h posttreatment, and cells were detached by trypsin-EDTA, washed in PBS, and Zombie-NIR stained (BioLegend). Cells were washed in flow cytometry buffer and fixed in 1% PFA. Flow cytometry was performed with an LSRFortessa X-20 analyzer (BD Biosciences) and data analyzed with FlowJo (version 10; Tree Star, Inc., Ashland, OR). The percentage of live cells was determined by gating on the Zombie-NIR-negative population.

Immunofluorescence staining and microscopy.

HFs were seeded onto coverslips (5 × 10⁴ cells/coverslip), after which cells were treated or infected. At 24, 48, and 72 h p.i., cells were washed with PBS and fixed with 4% PFA. Cells were permeabilized in 0.1% Triton X-100 (in PBS) blocked with 20% normal donkey serum (in PBS) before being stained with anti-IE1 primary antibody (clone 8B1.2; Millipore) or isotype control (mIgG_2a). Primary antibodies were detected by staining with anti-mouse IgG-Alexa Fluor 549 (Invitrogen), and cells were counterstained with DAPI (Thermo Fisher Scientific) before imaging on a wide-field fluorescent light microscope (Zeiss). The percentage of infected cells was determined by quantification of IE1-postivive cells and total number of cells (DAPI positive) per image using ImageJ, with the mean taken across five images per replicate.

Quantitative PCR of viral genomes.

Gal-9-treated/infected HFs were harvested at 4 h p.i. by scraping and total DNA extracted using a Qiagen DNA extraction kit. qPCR was performed using Brilliant II SYBR green (Agilent Technologies) analyzed on a Roche LightCycler 480 at 10 min and 95°C for denaturation and then 50 amplification cycles of 15 s at 95°C and 45 s at 60°C; finally, melt curve data were generated through 1 min at 95°C, 30 s at 50°C, and 30 s at 95°C. Primers used targeted human albumin (ALB-F, 5′-TTTGCAGATGTCAGTGAAAGAGA-3′; ALB-R, 5′-TGGGGAGGCTATAGAAAATAAGG-3′) and HCMV US28 (US28-F, 5′-AGAACTCATGCTCGGTGCTT-3′; US28-R, 5′-GCTGGAGATCCATTTGAGGA-3′). The US28 primers were designed to amplify a conserved region in the US28 gene sequence (93). Viral DNA (US28) was normalized to cellular DNA (albumin) and relative changes from untreated HCMV determined by the ₋2ΔΔC_T method.

Antibody and lactose blocking of Gal-9.

Gal-9 (50 nM) was diluted in serum-free medium and blocked by the addition of one of the following for 30 min: anti-Gal-9 monoclonal antibody (40 ng/ml, clone 9M1-3, BioLegend), isotype control (40 ng/ml, mIgG1), lactose (1.25 to 5 mM; Sigma), or sucrose (1.25 to 5 mM; Sigma). Merlin-GFP was then added for a further 30 min before addition to HFs at an MOI of 0.5. The fold change in the percentage of infected cells was assessed by flow cytometric analysis of GFP expression at 72 h p.i., as detailed above.

Gal-9 pretreatment of HCMV and HFs and Gal-9 treatment postinfection.

To pretreat virus inoculum, Gal-9 (50 nM) was diluted in serum-free medium and added to Merlin-GFP for 30 min at room temperature. The virus–Gal-9 mixture was added to HFs at an MOI of 0.5 for 90 min at 37°C. Alternatively, to pretreat cells, Gal-9 (50 nM) was added to HFs in serum-free medium for 30 min at room temperature. Gal-9-containing medium was replaced with Merlin-GFP in serum-free medium, at an MOI of 0.5, for 90 min at 37°C. Following infection, cells were washed with PBS and fresh supplemented medium added. For postinfection Gal-9 treatment, HFs were infected with Merlin-GFP at an MOI of 0.5 in serum-free medium for 90 min at 37°C, and cells were washed in PBS and serum-free medium containing Gal-9 (50 nM) added for 24 h p.i., followed by replacement with fresh supplemented medium. Merlin-GFP infection in which no Gal-9 was introduced served as the control in all experiments. The fold change in the percentage of infected cells was assessed by flow cytometric analysis of GFP expression at 72 h p.i., as detailed above.

Plaque reduction assay.

HFs were seeded into six-well plates at a density of 1.5 × 10⁵ cells/well. Merlin-GFP was untreated or treated with Gal-9 (100 nM) for 30 min before addition to HFs at an MOI of 0.001 (150 PFU) in duplicate. Following 90 min of infection at 37°C, cells were washed with either PBS to remove unbound virus or citrate buffer (40 mM citric acid, 10 mM KCl, 135 mM NaCl [pH 3]) for 3 min at room temperature to inactivate noninternalized virus, followed by PBS washing in triplicate. Cells were overlaid with a 1:1 mixture of 2× Eagle’s minimal essential medium (2× EMEM) supplemented with l-glutamine (4 mM), 20% FCS, and penicillin-streptomycin (200 units per ml), combined with 2% Avicel. Plates were incubated at 37°C and 5% CO₂ for 12 days and washed with PBS, and plaques were enumerated using an inverted light microscope (Axio; Zeiss).

Gal-9 inhibition of fusion assay.

Merlin-GFP was added to HFs at an MOI of 5 in serum-free medium for 90 min at 4°C to allow the virus to bind to cells. The viral inoculum was removed with PBS washing and replaced with serum-free medium with or without Gal-9 (50 nM). In some instances, Gal-9 was preblocked with either lactose (5 mM) or sucrose (5 mM) for 30 min, as indicated in Fig. 5C. Cells were transferred to 37°C for a further 60 min to allow virus fusion. Cells were washed in citrate buffer for 3 min at room temperature to inactivate any noninternalized virus, followed by PBS washing in triplicate (50). Cells were cultured in supplemented medium for 72 h and then harvested for flow cytometric analysis to determine the fold change in the percentage of infected cells (GFP positive).

Statistical analysis.

Statistical analyses were performed using the GraphPad Prism software. Statistically significant differences among groups were assessed using analysis of variance (ANOVA) (Bonferroni Dunn test). Statistically significant differences between pairs were assessed using Student’s t test. In all analyses, a P value of <0.05 was considered statistically significant. Data are presented as means and standard errors of the mean (SEM).

Ethics statement.

This study was approved by the institutional ethics committees of the Sydney West Area Health Service and the University of Sydney. All donors were adults and provided written informed consent in accordance with the Declaration of Helsinki.