Latency-Associated Degradation of the MRP1 Drug Transporter During Latent Human Cytomegalovirus Infection

See allHide authors and affiliations

Science  12 Apr 2013:
Vol. 340, Issue 6129, pp. 199-202
DOI: 10.1126/science.1235047

Hide-and-Seek Virus

Human cytomegalovirus (HCMV) establishes latent infection in human progenitor dendritic cells, causing significant morbidity and mortality on reactivation, which may occur in transplantation patients who are immunosuppressed. Neither detection nor selective removal of rare latent HCMV-infected cells has been possible. Weekes et al. (p. 199) have found that the multidrug-resistant ABC transporter, multidrug resistance–associated protein-1 (MRP1) is down-regulated during latent HCMV infection. Consequently, cytotoxic MRP1-specific substrates are not exported from HCMV-infected cells and accumulate—leading to cell death, which could potentially provide a mechanism for eliminating infected cells prior to transplantation.


The reactivation of latent human cytomegalovirus (HCMV) infection after transplantation is associated with high morbidity and mortality. In vivo, myeloid cells and their progenitors are an important site of HCMV latency, whose establishment and/or maintenance require expression of the viral transcript UL138. Using stable isotope labeling by amino acids in cell culture–based mass spectrometry, we found a dramatic UL138-mediated loss of cell surface multidrug resistance–associated protein-1 (MRP1) and the reduction of substrate export by this transporter. Latency-associated loss of MRP1 and accumulation of the cytotoxic drug vincristine, an MRP1 substrate, depleted virus from naturally latent CD14+ and CD34+ progenitors, all of which are in vivo sites of latency. The UL138-mediated loss of MRP1 provides a marker for detecting latent HCMV infection and a therapeutic target for eliminating latently infected cells before transplantation.

Human cytomegalovirus (HCMV) is a ubiquitous beta-herpesvirus that infects 60 to 90% of individuals (1). After primary infection, HCMV establishes a latent infection under the control of a healthy immune system. Reactivation from viral latency to productive infection causes serious disease in immunocompromised individuals, such as transplant recipients and AIDS patients (1, 2).

Cells of the myeloid lineage, such as CD34+ bone marrow progenitors and CD14+ monocytes, are sites of latent HCMV infection (35). The viral genome persists in these cells with little gene expression and no detectable virus production (6, 7). Reactivation from latency occurs upon myeloid differentiation, resulting in chromatin-mediated activation of the lytic gene expression cascade, viral DNA replication, and the production of infectious virions (8). Latent viral infection is thus required for viral persistence. Establishing how latency is maintained and how latently infected cells avoid immune recognition is crucial to understanding how HCMV persists in vivo. Furthermore, the elimination of latently infected cells is a key target in preventing recurrent HCMV infection in immunocompromised individuals.

A limited number of viral transcripts have been identified during natural latency in myeloid cells (6, 7) and include UL138 (9, 10), which encodes a 21-kD transmembrane Golgi-associated protein (10). UL138 is expressed with early-late kinetics during productive HCMV infection (10) but is also required for efficient latent carriage in vitro (9, 10). The expression of UL138 during lytic infection results in increased tumor necrosis factor receptor 1 (TNFR1) cell surface expression (11, 12), but little is known about UL138 during latency.

To address how UL138 affects host cell surface receptor expression during latent HCMV infection, we used plasma membrane profiling (PMP) (13), a proteomic technique that employs stable isotope labeling by amino acids in cell culture–based differential analysis to compare the expression of plasma membrane (PM) proteins in the presence and absence of UL138 in undifferentiated myeloid cells. Of the 592 PM proteins isolated from the monocytic cell line (THP-1), only 3 were reproducibly affected more than twofold (Fig. 1A and tables S1 and S2). Most notable was multidrug resistance–associated protein-1 (MRP1) (down-regulated 6.7- to 10.3-fold in three independent experiments), whereas Notch-ligand Delta-like protein 1 (DLL1) was down-regulated 2.1- to 2.6-fold. As expected, cell surface expression of TNFR1 increased (2.4- to 2.8-fold) (11, 12).

Fig. 1 HCMV UL138 down-regulates cell surface MRP1 and other targets.

(A) Scatter plot of proteins identified in PMP and quantified by more than 2 unique peptides. The summed ion intensity (y axis) is shown as log10. Significance A was used to estimate P values (28). (B) Cytofluorometric analysis of the indicated proteins in THP-1 cells stably expressing HCMV-encoded UL138 (THP-UL138) and control THP-1 cells. (C and D) Immunoblot for MRP1/UL138 in control or UL138-transduced fibroblasts (C) or in control or UL138-transduced THP-1 cells (D).

These cell surface changes were confirmed by cell surface flow cytometry (DLL-1, TNFR1, and CD36) or intracellular flow cytometry (MRP1), whereas expression of the control protein (CCR7) was unaffected (Fig. 1B). UL138 down-regulated MRP1 in all four cell lines tested, including fibroblasts (Fig. 1C), HL60-ADR cells, a promyelocytic leukemia cell line that overexpresses MRP1 (14), and HeLa cells (fig. S1).

We focused on MRP1, the most dramatically down-regulated protein. In the presence of UL138, not only did MRP1 cell surface expression decrease but the protein was undetectable (Fig. 1, C and D). UL138 expression is not restricted to latent HCMV infection, is detected 6 hours after lytic infection, and accumulates over 48 hours (10). We analyzed the temporal relationship between UL138 expression and MRP1 degradation during lytic infection in human fibroblasts (HFFs) with the TB40 isolate of HCMV. The observed loss of MRP1 at 48 hours coincided with high levels of UL138 expression (Fig. 2A). UL138 is encoded at the 3′ end of a polycistronic transcript that also encodes UL133, UL135, and UL136 (10, 15). Consequently, we used Toledo UL133-UL138 and UL138 open reading frame deletion mutants of HCMV (12) to determine whether this region is necessary for MRP1 down-regulation. As expected, these viruses replicated similarly to wild-type viruses (12) (Fig. 2B). Forty-eight hours after HFF infection with the deletion viruses, no UL138 expression was detected by reverse transcription polymerase chain reaction (RT-PCR), and MRP1 expression was restored (Fig. 2B). Thus, UL138 is necessary for MRP1 down-regulation and degradation, although other HCMV genes might also target MRP1. Human monocyte–derived macrophages infected with wild-type TB40 but not TB40ΔUL138 (an additional UL138 deletion mutant) (11) also showed decreased cell surface expression of MRP1 (Fig. 2C).

Fig. 2 UL138 down-regulates MRP1 during productive HCMV infection.

(A) Mock- or TB40 IE2-eYFP–infected HFFs [multiplicity of infection (moi) 5] were analyzed by immunoblot 24 and 48 hours after infection. (B) Mock, Toledo wild type (wt), Toledo ΔUL133-138, or Toledo ΔUL138 virus-infected HFFs were analyzed by immunoblot 48 hours after infection (top three panels). UL138 and IE mRNA was analyzed by RT-PCR with GAPDH as an internal mRNA control (bottom three panels). (C) Differentiated primary monocytes were infected with wtTB40 or TB40ΔUL138. Seventy-two hours after infection, cells were stained for MRP1, IE, and DAPI before being examined by confocal microscopy.

To determine the functional consequence of UL138-mediated MRP1 down-regulation, we examined export of the fluorescent reporter 5-carboxyseminaptharhodafluor (SNARF-1), an MRP1-specific substrate (16), and leukotriene C4 (LTC4), an endogenous MRP1 substrate (17). The loss of SNARF-1 from cells loaded with the SNARF-1 ester is a robust measure of MRP1 activity (fig. S2A). Preincubation with the MRP1-specific inhibitor MK571 allowed the accumulation of SNARF-1 and slowed export (Fig. S2B). In THP-UL138 cells, SNARF-1 was exported more slowly than THP-1 controls (Fig. 3A and fig. S2C); by 8 hours, 97% of control cells, compared with 35% of THP-UL138 cells, had unloaded dye. SNARF-1 was also exported significantly more slowly from HCMV- than mock-infected fibroblasts (Fig. 3B), and the export of LTC4 was inhibited in THP-UL138 cells (Fig. 3C).

Fig. 3 UL138 targets mature MRP1 for lysosomal degradation and inhibits export of MRP1-specific substrates.

(A) THP-1 or THP-UL138 cells were loaded with SNARF-1 ester, and intracellular SNARF-1 was measured by cytofluorometry. The proportion of cells retaining SNARF-1 was plotted. (B) Fifteen hours after infection, HCMV-infected HFFs (moi 5) were analyzed for intracellular SNARF-1 (left panel) and immunoblot (right panel). Three independent replicates were used per time point. Plotted are mean ± SEM, relative to the post-load HCMV-infected sample. Two-tailed P values (*P < 0.05). (C) LTC4 export assayed in A23187-stimulated cells (fig. S3) (28), with three independent replicates per condition. Plotted: mean ± SEM and two-tailed P values (*P < 1 × 10−6, **P < 0.0005). (D) Quantitative RT-PCR (RT-qPCR) analysis of MRP1 and GAPDH (28). (E) MRP1 immunoprecipitations (with QCRL3 antibody) from cells radiolabeled and chased as indicated, with CcmA included at the 5-hour time point (5*). Total MRP1 is quantified as percentage of MRP1 at time 0. (F) Cells incubated with MG132, CcmA, or DMSO (28) and immunoblotted as indicated. (G) HA immunoprecipitation from ADR-UL138HA cells preincubated with CcmA for 24 hours to increase MRP1 expression. Anti-FLAG beads were used as a control.

Because hemagglutinin (HA) peptide–tagged UL138 did not alter MRP1 mRNA levels, despite the loss of MRP1 protein (Fig. 3D), MRP1 down-regulation by UL138 was likely to be posttranscriptional. In the presence of UL138, [35S]-methionine radiolabeled 170-kD MRP1 matured normally through the secretory pathway before rapidly degrading (Fig. 3E), with a reduction in half-life from 16 to 20 hours (18) to less than 3 hours, suggesting that MRP1 targets UL138 for degradation in the late secretory pathway. Consistent with this, the loss of MRP1 was inhibited by the vacuolar adenosine triphosphatase inhibitor concanamycin A (CcmA) but was insensitive to proteasome inhibition (Fig. 3, E and F). Furthermore, UL138-HA interacted with MRP1 in cells preincubated with CcmA, due to the increased MRP1 expression (Fig. 3G). Thus, UL138 associates with MRP1 and induces its lysosomal degradation.

CD34+ bone marrow progenitors and CD14+ monocytes are key sites of latent HCMV infection in vivo (35, 19). Experimental models of latent infection in both cell types are routinely used to analyze HCMV latency in vitro (4, 1923). To address how UL138 expression affects latent HCMV infection, we latently infected primary CD34+ progenitor cells with a green fluorescent protein (GFP)–labeled recombinant HCMV (TB40 gfp) and found that latently infected cells (GFP-positive) also showed specific loss of cell surface MRP1 (Fig. 4A).

Fig. 4 Selective vincristine-mediated depletion of HCMV-infected cells from experimental and natural latent infection.

(A) CD34+ progenitors were latently infected with TB40 gfp. Seventy-two hours after infection, cells were examined (28) by confocal microscopy (left panels). The GFP signal in latently infected cells was boosted with an FITC-conjugated antibody to GFP (28). RT-PCR confirmed latent infection (right panel). (B and C) Treatment of experimentally latently infected monocytes with vincristine reduced latent viral load as determined by detection of latently expressed UL138 mRNA by RT-qPCR, and the relative number of latently infected cells. Primary CD14+ monocytes were latently infected with TB40 gfp. After 3 days, vincristine was added at the indicated concentration (28). Four days later, the GFP+ cells were counted in five independent replicates. Three further independent replicates were analyzed by RT-qPCR for UL138, IE, and GAPDH (C). IE RT-qPCR was always below the limit of detection. Plotted are mean ± SEM and two-tailed P values (*P < 0.005, **P < 0.001, ***P < 0.05). (D) Primary CD14+ monocytes from HCMV-seropositive donor D were treated for 4 days with vincristine (28). Endogenous HCMV was reactivated by differentiation and maturation to mature DCs (22), cocultured with fibroblasts for 2 weeks (four replicates per condition), which were examined for viral IE protein, and foci were counted. Plotted are mean ± SEM % of IE + foci as compared to 0 ng of vincristine per milliliter (0 ng/ml) and two-tailed P values (*P < 0.005, **P < 0.001, ***P < 0.0005). (E) Primary CD34+ progenitors were treated for 4 days with vincristine (28). Endogenous HCMV was reactivated by differentiation and maturation to mature DCs (22), then cocultured with fibroblasts for 2 weeks. Cell supernatants were transferred onto fresh fibroblasts (eight replicates per condition), and 100 cells per replicate were examined after 4 days for viral IE protein with DNA counterstaining. Plotted are mean ± SEM and two-tailed P values (each treatment versus 0 ng/ml vincristine): *P < 5 x 10−8, **P < 5 x 10−9.

MRP1 exports a number of cytotoxic agents, including vincristine, a Vinca alkaloid mitotic inhibitor which is relatively MRP1-specific (24). We initially tested the ability of vincristine to reduce the latent load of HCMV by killing experimentally latent CD14+ monocytes in vitro. Monocytes latently infected with TB40 gfp and then treated with vincristine for 4 days showed a reduced number of latently infected (GFP-positive) cells (Fig. 4B and fig. S4) as well as a concomitant reduction in detectable latently expressed UL138 RNA (Fig. 4C), which is consistent with vincristine-mediated killing of latently infected cells and a reduction in latency-associated viral load.

Experimentally latent CD14+ monocytes or CD34+ progenitors can be induced to reactivate latent virus by differentiation to dendritic cells (DCs) and subsequent maturation (8). If vincristine was reducing latent viral load by killing latently infected cells, this should also be reflected in a reduction in reactivating virus. Indeed, treatment of experimentally latently infected cells with vincristine reduced reactivation of latent HCMV from CD14+ monocytes after their differentiation to DCs (fig. S5). CD14+ monocytes and CD34+ progenitors isolated from latently infected donors routinely reactivate infectious HCMV after differentiation and maturation to mature DCs, detected by coculture with indicator fibroblasts (Fig. 4, D and E, and fig. S6) (5). Vincristine treatment of CD14+ monocytes, from 7 out of 7 healthy HCMV-seropositive donors, as well as CD34+ cells, showed reduced reactivation of infectious virus after differentiation and maturation (Fig. 4D-E). Thus, MRP1 is a potential therapeutic target for eliminating latent HCMV-infected cells from bone marrow before transplantation.

The study of HCMV latency has been hampered by the inability to identify low-frequency latently infected cells ex vivo. The down-regulation of MRP1 by UL138 provides a novel marker of latent infection, but why is MRP1 targeted? DCs from MRP1-deficient mice fail to respond to chemotactic stimuli or migrate into afferent lymphatics (25), because the endogenous MRP1 substrate LTC4 (17) sensitizes the CCR7 chemokine receptor to CCL19 (25). UL138-mediated down-regulation of MRP1 reduced cellular LTC4 export, suggesting that UL138 could inhibit the migration of infected DCs to draining lymph nodes and impair the generation of an HCMV-specific immune response. Decreased MRP1 expression could also help maintain latent infection by inhibiting premature terminal differentiation of DC progenitors until conditions for reactivation are established, as reported for other HCMV latency proteins (UL111.5A) (26), and the terminal differentiation of DC progenitors is dependent on functional MRP1 (27).

UL138-mediated down-regulation of MRP1 was functionally significant, leading to a dramatic reduction in the export of MRP1-specific substrates and predicted that MRP1-transported cytotoxic drugs would accumulate and kill UL138-expressing cells. Indeed, vincristine treatment of experimentally latent myeloid cells and naturally latent CD14+ cells and their CD34+ progenitors decreased the latent CMV viral load. Vincristine treatment dramatically reduced levels of reactivated virus after myeloid cell differentiation and maturation to mature DCs, a well-established signal for virus reactivation (5).

Our results open up the possibility of developing strategies using MRP1-specific reagents to clear bone marrow or hematopoietic stem cells of latently infected cells before transplantation, either based on the selection of HCMV-negative cell subpopulations or the targeted killing of latently infected cells, using cytotoxic agents normally exported by MRP1.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Tables S1 and S2

References (2943)

References and Notes

  1. Information on materials and methods is available as supplementary material on Science Online.
  2. Acknowledgments: We are grateful to M. Reeves for discussions and cells. All data are tabulated in the main paper and the supplementary materials. This work was supported by a Next Generation Fellowship from the Cambridge Institute for Medical Research and a Wellcome Trust Fellowship (093966/Z/10/Z) to M.P.W.; a grant from the Agency for Science, Technology and Research, Singapore to S.Y.L.T.; an MRC Programme grant (G0701279) to J.H.S.; and a Wellcome Trust Senior Fellowship (084957/Z/08/Z) to P.J.L. This work was also supported by the NIHR Cambridge Biomedical Research Centre, and the Cambridge Institute for Medical Research is in receipt of a Wellcome Trust Strategic Award. P.J.L., J.H.S., M.P.W., and R.A. are authors on patent application PCT/GB2012/051094 filed by Cambridge Enterprise, entitled “Detection and depletion of HCMV infected cells.”

Stay Connected to Science

Navigate This Article