Human cytomegalovirus (HCMV) infection is an important cause of morbidity and mortality among both solid organ and hematopoietic stem cell transplant recipients. Identification of cells throughout the body that can potentially serve as a viral reservoir is essential to dissect mechanisms of cell tropism and latency and to develop novel therapies. Here, we tested and compared the permissivity of liver-, brain-, lung (LNG)- and bone marrow (BM)-derived perivascular mesenchymal stromal cells (MSC) to HCMV infection and their ability to propagate and produce infectious virus. Perivascular MSC isolated from the different organs have in common the expression of CD146 and Stro-1. While all these cells were permissive to HCMV infection, the highest rate of HCMV infection was seen with LNG-MSC, as determined by viral copy number and production of viral particles by these cells. In addition, we showed that, although the supernatants from each of the HCMV-infected cultures contained infectious virus, the viral copy number and the quantity and timing of virus production varied among the various organ-specific MSC. Furthermore, using quantitative polymerase chain reaction, we were able to detect HCMV DNA in BM-MSC isolated from 7 out of 19 healthy, HCMV-seropositive adults, suggesting that BM-derived perivascular stromal cells may constitute an unrecognized natural HCMV reservoir.
aortic-derived endothelial cells
bacterial artificial chromosome
brain microvascular endothelial cells
green fluorescence protein
human foreskin fibroblasts
hematopoietic stem cells
multiplicity of infection
mesenchymal stromal cells
phosphate buffered saline
polymerase chain reaction
quantitative polymerase chain reaction
Human cytomegalovirus (HCMV) infection is an important cause of morbidity and mortality among both solid organ transplant and hematopoietic stem cell (HSC) transplant recipients [1, 2], causing pneumonia, gastroenteritis, retinitis, hepatitis and encephalitis, and contributing to graft failure . Although the use of antiviral drugs and preemptive therapy has decreased the incidence, morbidity and mortality of HCMV disease in transplant recipients, still, persistent low levels of HCMV infection have been associated with deleterious consequences . Therefore, identification of cells serving as the reservoir and/or that are responsible for reactivation of the disease could potentially lead to the development of novel targeted therapies.
Although several cell types, including fibroblasts, myofibroblasts, endothelial cells, smooth muscle cells, epithelial cells, neuronal cells and brain (BRN)-derived pericytes, have been shown to be susceptible to HCMV infection [5-9], hematopoietic cells, particularly myeloid progenitors, have been implicated as the major culprit in HCMV reactivation, as they have been shown to serve as a latent reservoir of this virus [10-14]. In addition, studies using biopsy samples  and cultures of either mixed populations of bone marrow (BM) [16-18] or marrow stromal cells purified based on plastic adherence [19-21], demonstrated that HCMV was able to infect, and actively replicate in, these cell layers.
Perivascular mesenchymal stromal cells (MSC) in the BM are part of the hematopoietic microenvironmental niche and play a role in HSC maintenance and differentiation [22, 23]. Perivascular MSC in other organs [24, 25] encircle capillaries and vessels, and are in intimate contact with circulating blood cells when they extravasate across their walls into inflamed tissues . In similarity to their BM counterpart, these perivascular MSC express Stro-1+  and CD146+ [23, 24]. Moreover, all these perivascular cells express molecules for HCMV binding such as integrins beta 1 and beta 3 [28, 29]. We therefore hypothesized that, if these cells were susceptible to HCMV infection, they could potentially serve as a HCMV reservoir in transplantation patients. Here we show that MSC isolated based on Stro-1 positivity, from BM, liver (LVR), BRN and lung (LNG) , are susceptible to HCMV infection, and that the virus is able to establish a productive infection and propagate within these cells. Therefore, organ-derived perivascular MSC could play a role in HCMV infection that has thus far been overlooked.
Materials and Methods
Human foreskin fibroblasts (HF; kindly provided by Dr. Gregory Pari) were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA). Human BM cells were harvested from healthy adult donors after informed consent according to guidelines from the Office of Human Research Protection at the University of Nevada at Reno. Donors were determined in a commercial laboratory to be CMV-seronegative or -seropositive by testing for antibodies to CMV. Low-density BM mononuclear cells were separated using a Ficoll-Hypaque density gradient (1.077 g/mL; Sigma, St. Louis, MO) washed twice in Iscove's modified Dulbecco's media (Invitrogen) with penicillin (100 U/mL), streptomycin (100 mg/mL) and amphotericin B (0.25 mg/mL; Gibco Laboratories, Grand Island, NY). Stro-1+ cells were selected by magnetic cell sorting (Miltenyi Biotec, Inc., Auburn, CA) using a Stro-1 antibody (R&D Systems, Minneapolis, MN) as previously described . Human fetal LVR, BRN and LNG (18–22 weeks gestational age) were purchased from Advanced Bioscience Resources (Alameda, CA) and sorted using Stro-1 antibody (R&D Systems) as previously described [30, 31]. The purity of the sorted cells exceeded 95% and in addition to CD29+, CD44+, CD73+, CD90+, CD146+, NG2+ cells maintained Stro-1, throughout culture.
HCMV strains FIX(VR1814)-bacterial artificial chromosome (BAC)  (kindly provided by Dr. Gregory Pari) and Davis (ATCC, Manassas, VA) were propagated in HF cultures and viral titers were determined using standard plaque assays . The FIX-BAC virus was created from the VR1814 clinical HCMV isolate, by substituting BAC DNA for the US2-US6 region [32, 33]. In the original FIX-BAC virus, a cassette encoding the Escherichia coli guanine phosphoribosyltransferase (gpt) gene was included within the BAC fragment to allow for selection of recombinant viral genomes [34-37]. In the present study, we employed a derivative of this original virus in which the gpt gene was replaced by the green fluorescence protein (GFP) gene , allowing us to track and quantitate infection in real-time in living cells by visualizing green fluorescence. In this virus, expression of the GFP reporter is driven by the constitutively active HSV-TK promoter. The different stromal cell populations and control HF were infected with either strain at a multiplicity of infection (MOI) of 3 for 1 h under normal growing conditions, after which the stromal layers were washed twice with sterile phosphate-buffered saline (PBS; Sigma) to eliminate any remaining unbound viral particles, and new media was added.
DNA extraction and polymerase chain reaction
BM-MSC, BRN-MSC, LVR-MSC and LNG-MSC samples were infected with Davis strain at an MOI of 3 as described above. HF cells were also infected and used as a positive control PC. DNA was extracted from cells, prior to and after HCMV infection, using the DNeasy Blood and Tissue kit (Qiagen, Valencia, CA). The polymerase chain reaction (PCR) consisted of 1.5 mM MgCl2, 20 mM Tris–HCl (pH 8.3), 50 mM KCl, 0.2 mM dNTP mixture, 0.02 U of recombinant Taq DNA polymerase (Invitrogen), 0.004 µg/µl (final concentration) of each primer (IE2A:ATGGAGTCCTCTGCCAAGAGAAAGATGGAC and IE4B:CAATACACTTCATCTC CTCGAA, which amplify a region of the HCMV unique long [UL]123 gene) and 100 ng of template DNA. The amplification conditions were: initial denaturation at 94°C (3 min), followed by 35 cycles of denaturation at 94°C (45 s), annealing at 55°C (30 s), extension at 72°C (1 min 30 s) and a final extension of 10 min at 72°C. Amplification with primers specific to β-actin served as a control for DNA integrity. β-Actin and UL123 amplification products were electrophoresed on a 1% agarose gel in 1× Tris–acetate–EDTA buffer, visualized by UV transillumination (Biospectrum, Upland, CA) and recorded using visionWorksLS software (UVP LLC, Upland, CA). PCR controls were performed in the absence of template negative control (NC) or in the presence of viral DNA positive control (PC).
RNA extraction, DNase treatment and reverse transcriptase PCR
Total RNA was purified from each cell population before and after HCMV infection, using TRIzol® Reagent with the PureLink™ RNA Micro Kit (Invitrogen) and DNase-treated with RQ1 RNase-Free DNase (Promega, Madison, WI). To analyze presence/absence of UL123 (transcript that expresses immediate early [IE] protein) or UL83 (transcript that expresses pp65 protein), 100 ng of each RNA sample was used for cDNA synthesis using the SuperScript III first-strand synthesis system and random hexamer primers (Invitrogen). Reverse transcriptase PCR (RT-PCR) conditions were similar to those above except that the annealing temperature was raised to 57°C and primers for UL83 were UL83F: GGTATCCACGTACGCG TGAG and UL83R: CATGATGTGCGAGATCTTGC.
Negative controls (NCs) and PCs included PCR reactions in the absence of template, and in the presence of viral DNA, respectively. In addition, for each RNA sample, RT-PCR was performed in the absence of reverse transcriptase enzyme.
Transmitted light and fluorescence images of cell populations infected with FIX-BAC strain were captured with an Olympus IX-71 (Olympus America, Inc., Center Valley, PA) microscope with a 10× objective using bright field and/or a Fluorescein isothiocyanate (FITC) filter.
Calculation of infectious units
BM-MSC, BRN-MSC, LVR-MSC and LNG-MSC were infected with Davis at MOI of 3. After 3 h, 1, 3, 5 and 7 day postinfection (dpi), supernatants collected from infected cells were used to determine the number of infectious units (IU). To this end, HF cells were incubated for 1 h with serial 10-fold dilutions prepared from each of the MSC supernatants collected at the different time points. HF were washed twice with PBS, after which fresh media was added. HF layers were stained after 24 h for the presence of IE protein as described in the Supplementary Materials and Methods and as previously described [39-41].
Total DNA was extracted from the different cell populations by using the DNeasy Blood and tissue kit (Qiagen) following the manufacturer's instructions. DNA was eluted into 100 µL nuclease-free water and quantified using a NanoDrop (Thermo Fisher Scientific, Inc., Wilmington, DE). The amount of viral DNA present in the samples was determined by absolute amplification of the US6 HCMV gene in a quantitative PCR (qPCR) reaction as detailed in the Supplementary Materials and Methods using the following US6 primers: sense: AACGATACAGGGGACTGCAC; antisense: ACGGGATTTGTGCGTTTTAG. To perform the calibration curve for qPCR, we purified and performed serial 10-fold dilutions of the US6 HCMV recombinant DNA vector  containing known viral copies per microliter (from 1 × 108 to 1 × 102 viral copies per reaction) that were amplified simultaneously and under the same conditions as the MSC samples. The Ct values obtained from each standard dilution were plotted against the viral copy number to construct the standard curve. Each Ct value, obtained from US6 DNA amplification for each sample, was used to extrapolate, from the standard curve, the absolute viral copy number for that sample. Finally, the viral copy numbers for each sample were statistically compared to the numbers obtained for the HCMV negative samples. The amplification results were validated based on the quality of dissociation curves and high amplification efficiency of the primers. Each sample was analyzed at least in triplicate, along with no-template controls, serial dilutions of the solution containing known viral copies per microliter and mock infected samples.
The amount of viral DNA present in naturally infected stromal cells was determined by absolute amplification of the US6 HCMV gene in a nested qPCR reaction. Briefly, DNA purified from each sample was first amplified in the same conditions as the qPCR reaction previously described, but for 30 cycles using the following outer primers: US6 sense: GACCCGAAAACCCTTCTCTC; antisense: ACGGGATTTGTGCGTTTTAG. After the first round of qPCR, a second 25 µL mix of the same composition was added to each reaction substituting inner (nested) primers instead of outer primers, and the first round amplicon serving as template. PCR cycling conditions were the same as before but for only 20 cycles. The sequences of US6 inner primers were as follows: US6 sense: GACCCGAAAACCCTTCTCTC; antisense: GTGCAGTCCCCTGTATCGTT. The standard calibration curve for absolute qPCR was performed as described in the previous section, but the dilutions of the vector solution ranged from 1 × 104 to 1 × 102 viral copies per reaction. Each Ct value, obtained from the second round of US6 DNA amplification for each sample, was used to extrapolate from the standard curve the absolute viral copy number for that sample. Finally, the viral copy numbers for each sample were statistically compared to the values obtained for the HCMV negative samples. Each sample was analyzed at least in triplicate, along with no-template controls, serial dilutions of the solution containing known viral copies per microliter and HCMV negative samples.
Experiments were independently repeated at least three times. Results are presented as mean ± SD. One-way analysis of variance followed by Tukey's post hoc test analysis for multiple comparisons was performed in the studies addressing production and release of infectious viral particles; p-values < 0.05 were considered to be statistically significant.
Two-tailed t-tests were used to determine statistical significance between amplification values obtained from stromal cells from different donors when compared to stromal cells from CMV negative donors; p-values < 0.05 were considered to be statistically significant.
BM-, LVR-, BRN- and LNG-derived perivascular Stro-1+ MSC are permissive to HCMV infection as determined by detection of GFP-positive cells and detection of virus-induced cytopathic effects
We used the Davis  and FIX-BAC  strains of HCMV to investigate the permissivity of BM-, LVR-, BRN- and LNG-derived Stro-1+ perivascular MSC to HCMV infection. MSC were isolated from the different organs based on Stro-1 expression, as previously described . Flow cytometric analysis revealed these cells to be CD29+, CD44+, CD73+, CD90+, CD146+, NG2+ and negative for the hematopoietic markers CD34 and CD45. After infection of stromal layers with HCMV FIX-BAC at an MOI of 3, susceptibility to infection was assessed, in each one of the cell populations, by analyzing GFP expression over a 10-day period. GFP expression was readily seen by day 5 in BM-(n = 4) LVR-(n = 4), LNG-(n = 4) and BRN-(n = 4) MSC (Figure 1A).
In order to demonstrate that each of the different cell layers supported HCMV FIX-BAC replication, supernatants from infected BM-, LVR-, LNG- and BRN-MSC were collected at 10 dpi and used to infect new cell monolayers of the corresponding cell type from which each supernatant had been harvested. As can be seen in Figure 1B, GFP expression was readily observed by day 5 in all stromal cell layers, demonstrating that all phases of viral replication occurred upon initial infection.
We also infected MSC derived from the various organs with HCMV Davis at MOI of 3, and investigated the generation of cytopathic effects (CPE) in cell cultures (n = 6 for each cell type). In a 10-day period, all infected cells displayed HCMV CPEs (Figure 2), demonstrating their susceptibility to HCMV infection.
In addition, and in similarity to HCMV FIX-BAC, HCMV Davis could also be propagated in all BM-, LVR-, LNG- and BRN-MSC monolayers (data not shown).
BM-, LVR-, BRN- and LNG-derived perivascular Stro-1+ MSC produce and release infectious viral particles
We next collected at 3 h, 1, 3 and 5 dpi, supernatants from the different MSC layers infected with HCMV Davis strain at an MOI of 3, and used each of the supernatants to infect HF cells. Analysis of HF cells for the presence IE proteins at 48 h postinfection demonstrated conclusively that supernatants from each of the HCMV-infected MSC cultures contained infectious viral particles with the ability to infect HF. After calculating the number of IU present at the different time points, the values were log 10–transformed to allow statistical comparison between the number of viral particles present in the supernatants of the different MSC types (Figure 3). At 1 dpi, BM-MSC (3.70 ± 0.13; n = 3), LNG-MSC (3.55 ± 0.16; n = 3) and LVR-MSC (3.61 ± 0.03; n = 3) supernatants had similar levels of IU. BRN-MSC supernatants (3.22 ± 0.17; n = 3) had lower levels than the LVR- and BM-MSC (p < 0.05), but the difference was not significant when compared with LNG-MSC (p > 0.05).
By contrast, at 3 dpi, BRN-MSC (4.69 ± 0.12) produced significantly higher levels than the LVR- and BM-MSC (p < 0.05), and LNG-MSC formed the highest levels of viral particles (6.67 ± 0.06) when compared with the other three populations (p< 0.05). At this same time point, LVR-MSC (3.63 ± 0.01) and BM-MSC (3.83 ± 0.12) produced the smallest amount of IU, with no significant difference between the two cell populations (p > 0.05).
At 5 dpi, production of IU was maintained in BM-, BRN- and LNG-MSC. BM-MSC levels were 3.58 ± 0.04; BRN, 4.89 ± 0.05; LNG, 6.57 ± 0.05; and LVR-MSC, 6.90 ± 0.04.
BM-, LVR-, BRN- and LNG-derived perivascular Stro-1+ MSC are permissive to HCMV infection as determined by presence of the IE gene UL123 and quantitation of CMV genomes
HCMV (Davis strain)-infected MSC (n = 3 for each population at MOI of 3) were extensively washed to remove residual viral particles, and allowed to grow for 7 days with regular media changes to remove any possible residual unbound virions. Mock-infected cells were used as NCs to ascertain that the viral DNA amplified by PCR resulted from successful HCMV infection in vitro. DNA was extracted at 7 dpi and PCR was performed with primers specific for the IE gene UL123, as described in the Materials and Methods section. HCMV-infected MSC were positive for HCMV DNA, while the mock-infected cells were consistently devoid of HCMV DNA (Figure 4A), demonstrating that cultures were indeed infected by HCMV and maintained the viral genome. Further confirmation of successful HCMV infection of the various MSC was obtained by using qPCR to determine the viral copy number present in the cultured MSC at 7 dpi. As can be seen in Figure 4B, all cultures contained ≥5 × 104 viral copies/100 ng of cellular DNA. Since an MOI of 3 corresponds to 4.8 × 104 viral copies/100 ng of cellular DNA, the presence of viral copy numbers exceeding the initial viral input at 7 days after exposure to HCMV demonstrates successful infection of the MSC and production of new viral genomes. In addition, LNG-MSC contained a significantly higher number of viral copy numbers by 100 ng/DNA than any other organ-derived MSC.
Transcription of UL123 and UL83 genes after HCMV infection of Stro-1+ BM-, LVR-, BRN- and LNG-derived perivascular MSC
In order to confirm that HCMV was able to establish a productive infection, BM- (n = 3), LVR- (n = 3), BRN- (n = 3) and LNG- (n = 3) MSC, and HF (n = 3) were infected with HCMV Davis strain at MOI of 3, and analyzed by RT-PCR for expression UL123 and UL83 . In order to exclude any potential DNA contamination or viral carryover, NCs included PCR reactions in the absence of template, and for each RNA sample, RT-PCR was performed in the absence of reverse transcriptase enzyme. Analysis of mRNA at 1, 3 and 5 dpi demonstrated that UL123 and UL83 were being transcribed at most time points during the 5 days following infection (Figure 5).
BM-, LVR-, BRN- and LNG-derived perivascular Stro-1+ MSC are naturally infected with HCMV
We next investigated whether BM-MSC served as a natural reservoir for HCMV within the body by performing nested qPCR (n = 3) on MSC isolated from the BM of 19 healthy HCMV-seropositive donors. Our results showed that 7 of these 19 donors harbored HCMV DNA (Figure 6) in their BM-derived stromal cells, confirming that these cells were naturally infected in these individuals.
The same method for US6 nested PCR (n = 3) was used to determine whether LNG-, (n = 9) BRN- (n = 9) and LVR- (n = 8) MSC contained HCMV DNA sequences (Figure 6). While the serologic status or the presence of active disease was unknown for these donors, eight out of nine BRN donors, seven out of nine LNG donors and four out of nine LVR donors were found to be positive for viral DNA, demonstrating that LVR, BRN and LNG stromal cells can be naturally infected with HCMV. Of note is also that LNG-derived stromal cells, overall, had the highest copy number/100 ng of DNA of any naturally infected tissue, while BM had the least (p < 0.05) (Figure 6).
HCMV infection remains one of the major complications after both solid organ and allogeneic BM transplantation (BMT), contributing to allograft rejection and development of graft-versus-host disease . While in solid organ recipients the transplanted organ is the main target of CMV infection, in BMT recipients, interstitial pneumonia is the primary manifestation of HCMV infection . In addition, HCMV is one of the pathogens that have been convincingly implicated in the pathophysiology of cardiac allograft vasculopathy and arteriosclerosis . Although the hematopoietic myeloid lineage has been implicated as the main reservoir of the virus [14, 17, 48-51], HCMV transmission occurs following transplantation of virtually every organ; therefore, it is conceivable that tissue-derived cell populations are also able to be infected and serve as a reservoir contributing to HCMV disease . For instance, we and others have shown that, within the BM, the stromal components are susceptible to HCMV infection [17, 19, 53, 54], and that HCMV replication differs depending upon the cell type and organ of origin . When HCMV-infected BRN microvascular endothelial cells (BMVEC) and aortic-derived endothelial cells (AEC) were compared, the former underwent rapid cellular lysis, while in the latter the infection was nonlytic, but cells continued to release the virus for the lifespan of the culture. In addition, mature viral particles were only detected at significant amounts in the cytoplasm of BMVEC. This suggests that the levels of HCMV replication in endothelial cells obtained from different organs are distinct, and that in addition to hematopoietic cells, persistently infected AEC may also serve as a reservoir of the virus . When other BRN-derived cells, such as human BRN pericytes , were infected with HCMV, these cells, in similarity to BMVEC, experienced lytic infection. By contrast, neural progenitor cells and mature neural cell types were shown to be fully permissive to HCMV, but depending on the strain of HCMV used, the development of CPE varied . Therefore, identification of cells that are permissive to HCMV within the different organs could provide a means of pinpointing the source of viral transmission and/or reactivation. Here, we demonstrate that perivascular MSC, isolated from organs directly associated with HCMV pathological manifestations, are productively infected by HCMV, albeit at different levels. It has been shown that MSC actively migrate to sites of inflammation, allograft rejection  and vascular injury , and actively participate in tissue repair. Therefore, CMV-positive MSC could conceivably be mobilized into circulation and spread the virus to the transplanted tissue and thereby contribute to pathology and/or rejection.
In agreement with other studies using AEC , BM-derived stroma , BRN-derived endothelial cells  and BRN pericytes , we show that HCMV is able to infect BRN-, LVR-, LNG- and BM-derived perivascular MSC and release infectious virus particles. In addition, because all the perivascular MSC used in these studies were isolated based upon Stro-1 expression [27, 58] and displayed identical phenotypic markers , we were able, for the first time, to perform a side-by-side comparison of identical perivascular cell populations harvested from different organs, for their susceptibility to and kinetics of HCMV infection. The highest HCMV infection rate was observed in LNG-MSC, while MSC from the BM exhibited the lowest viral load. We also demonstrated conclusively that supernatants from each of the HCMV-infected MSC cultures contained infectious viral particles, with the number of infectious virions produced varying widely between the various organ-specific MSC. For instance, LVR-MSC produced high levels of virions at a later time point than LNG-MSC, and the viral copy number/100 ng of DNA of LVR-MSC was considerably lower than that of LNG-MSC, demonstrating differing abilities of these two phenotypically identical cell populations to support a productive infection. BM-MSC remained the cell type producing the smallest number of viral particles throughout culture and containing the smallest viral copy number/100 ng of DNA. However, because of their direct interaction with HSC within the niche, it is possible that they can still function as a source of HCMV infection. Indeed, in the present study, using nested qPCR, we were able to demonstrate the presence of HCMV DNA in BM-MSC from 7 out of 19 healthy, HCMV-seropositive adults; the viral copy number/100 ng DNA was significantly lower than those found after in vitro infection of the same cells. While other studies have been unsuccessful in finding HCMV DNA in BM-MSC from serologically positive human patients , the ability to detect HCMV infection is continually improving as more sensitive tests are developed, and it is likely that our use of a more sensitive nested qPCR assay contributed to our ability to detect HCMV-infected MSC within these seropositive donors.
In conclusion, organ-derived perivascular MSC are permissive to HCMV infection and have the ability to produce infectious HCMV virions, providing evidence that they may represent a previously overlooked natural HCMV reservoir for infection.
This work was supported by National Institutes of Health NHLBI: HL97623.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.