Extracellular viral microRNAs as biomarkers of virus infection in human cells

Abstract

Nucleic acid amplification tests (NAATs) have enabled fast and sensitive detection of virus infections but are unable to discriminate between live and dead/inert viral fragments or between latent and reactivated virus infections. Here, we show that extracellular viral microRNAs (viral exmiRs) are cell-free candidate biomarkers of live, latent, and reactivated virus infections, achieving fast (under 1 day) and sensitive (30 attomolar [aM]) detection by quantitative real-time reverse transcription PCR (real-time RT-qPCR). We report that spent-media-derived Epstein-Barr virus (EBV) miR-BART10-3p and herpes simplex virus 1 (HSV-1) miR-H5 are biomarkers of live EBV-2 and HSV-1 infection of T cell cultures, respectively. We identified extracellular human herpesvirus 6 (HHV-6) miR-Ro6-4 as a biomarker of endogenous latent HHV-6 in healthy human donor T cell cultures and identified human cytomegalovirus (HCMV) miR-US5-2-5p and miR-US22-5p as plasma biomarkers of endogenous latent HCMV infection. Viral exmiR profiling of spent media from EBV- and HHV-8-reactivated B cell models revealed specific signatures of elevated EBV miR-BHRF1-2-3p and HHV-8 miR-K12-10a-3p, miR-K12-10b, and miR-K12-12-3p, respectively, during virus reactivation. Our study thus suggests the utility of viral exmiR biomarkers in enabling NAAT-based detection of live, endogenous latent, and reactivated virus infections of cells.

Graphical abstract

This study harnesses extracellular viral microRNA (viral exmiR) signatures as cell-free biomarkers of live, latent, or reactivated virus infections in human cells. By leveraging fast and sensitive PCR detection, viral exmiR-based PCR assay can identify potentially dangerous viral infections at attomolar sensitivity in under 1 day.

Introduction

Virus testing is crucial to ensure the safety of advanced therapeutic medicinal products (ATMPs) and the safety of immunosuppressed patients post-treatment.^1,2,3,4 Testing for adventitious viruses—those unintentionally introduced as contaminants during the manufacturing process—is especially important in the manufacturing of cell therapy products since the removal of viral contaminants by terminal sterilization is not feasible. However, traditional, slow (14–28 days) culture-based virus tests are incompatible as in-process/product release tests for cell therapy products, e.g., chimeric antigen receptor (CAR) T cell therapies that need to be expeditiously administered to extremely ill patients. Innovation in alternative rapid and sensitive molecular-based virus tests is key in shaping the regulatory framework on viral safety testing and assurance for cell therapy products, as there are currently no regulatory requirements. One such class of molecular tests is nucleic acid amplification tests (NAATs), which are increasingly being adopted to complement culture-based virus testing but are unable to discriminate between live virus infections and dead/inert viral fragments from viral shedding or past infections.^5,6

Timely detection of viral infections is also crucial in high-risk immunosuppressed patients and transplant recipients to manage viral disease complications arising from reactivation of endogenous, childhood-acquired viruses.^3,7,8,9 Post-treatment reactivation of clinically important viruses, including human cytomegalovirus (HCMV), Epstein-Barr virus (EBV), human herpesvirus 6 (HHV-6), human adenovirus (HAdV), HHV-3, and herpes simplex virus 1 (HSV-1) has been associated with an increased risk of morbidity and mortality, as well as slower immune recovery in hematopoietic stem cell transplant patients^{10,11,12,13,14,15,16,17} and CAR-T cell recipients.^{18,19,20,21,22,23} However, current diagnoses of viral reactivation events lack well-defined criteria, which may include the onset of clinical syndromes/manifestations complemented with retrospective molecular-based testing for viral DNA.²⁴ This underscores the need for fast, sensitive, and informative viral tests to guide clinical decision-making and timely patient management of virus-associated post-treatment complications.

Viral-encoded microRNAs, or viral miRNAs, are short (19–24 nt), regulatory non-coding RNAs that are produced in virus-infected cells using the host transcription and host miRNA processing machinery.^25,26,27,28 Viral miRNAs have been shown to be expressed during viral latency in cells despite limited viral transcription, antigen production and the absence of replicating virus^29,30,31,32 and may be secreted into the extracellular milieu in both vesicle^33,34,35 and non-vesicle, protein/lipid-bound forms.³⁶ Quantitative real-time reverse transcription PCR of viral miRNAs in plasma suggests that circulating viral miRNAs may allow for more accurate assessments of latent viral infections than traditional serological methods,³⁷ which may suffer from false negatives in immunocompromised patients who do not mount a sufficient antibody response.³⁸ Specific viral miRNAs have also been reported to be expressed in varied amounts during productive, latent, and reactivated HSV infection,³⁰ highlighting the potential of viral miRNAs as informative NAAT targets of viral infections. Here, we report the use of extracellular viral miRNAs (viral exmiRs) as biomarker candidates for the cell-free and sensitive detection of viral infections within 1 day. In this proof-of-concept study, viral exmiRs were harvested from spent culture media using precipitation-based capture of both vesicle and non-vesicle-bound extracellular RNA (exRNA).³⁴ This achieved a lower limit of quantification (LLOQ) of approximately 30 attomolar (aM) using gold-standard target-specific miRNA detection by stem-loop reverse transcription coupled with hydrolysis probe-based quantitative real-time PCR (real-time RT-qPCR).^39,40 We demonstrated the successful detection of viral exmiRs from spent media of virus-infected human T cell cultures and show that HSV viral exmiRs are specific identifiers of live-virus infection in cells that could be detected as early as 8 h post-infection. Extracellular HHV-6 miR-Ro6-4 identified endogenous HHV-6 latency in HHV-6-positive healthy human donor T cells. Further, HCMV miR-US5-2-5p and miR-US22-5p in plasma identified patients with endogenous latent HCMV infection. Finally, we found significantly elevated levels of specific viral exmiRs in endogenous reactivated EBV and HHV-8 B cell models that could be detected as early as 6 h after viral reactivation. Collectively, our findings support viral exmiRs as identifiers of viral infections in cells that can discriminate between live versus dead, latent versus reactivated virus infections and highlight the potential utility of viral exmiR-based NAATs for early, cell-free virus detection in cell cultures, donor/patient endogenous virus screening, and post-treatment monitoring of viral reactivation in patients.

Results

Detecting viral exmiRs: Assay sensitivity and limit of quantification

To establish a proof-of-concept workflow to investigate viral exmiRs, the polyethylene glycol/sodium chloride overnight precipitation-based method was chosen to enable the isolation of exRNA bound in macromolecular complexes, including extracellular vesicles, lipoproteins, and ribonucleoprotein complexes in spent culture media of virus-infected cells^41,42 in a phenol-free manner while reducing hands-on time (Figure 1A). exRNA was isolated (60 min) and interrogated for specific viral exmiRs using the current gold-standard target-specific stem-loop reverse transcription (70 min) coupled with hydrolysis probe-based qPCR (30 min fast cycling) (Figure 1B). The time to result from sampling spent culture media to qPCR assay readout was approximately 20 h, including a 16 h overnight precipitation incubation and hands-on time for assay set up (Figure 1A).

Figure 1.

Viral exmiR detection by real-time RT-qPCR

(A) Extracellular RNA isolation and real-time RT-qPCR exmiR detection workflow of spent culture media in under 24 h and their potential applications. The figure was created in BioRender (https://BioRender.com/h74x736). (B) Illustration of target-specific stem-loop reverse transcription coupled with hydrolysis probe-based real-time PCR miRNA detection. (C) Serially diluted synthetic oligos corresponding to EBV miR-BART10-3p (i) and HSV miR-H5 (ii) determined by real-time RT-qPCR. (D) Table of real-time RT-qPCR assay performance characteristics showing assay efficiency, linearity, and linear dynamic range derived from the respective standard curves. (E) Real-time RT-qPCR detection of extracellular EBV miRs BHRF1-1 (i), BHRF1-2-3p (ii), BHRF1-3 (iii), BART1-5p (iv), BART6-5p (v), and BART10-3p (vi) from the spent media of mock- and EBV-2-infected healthy human PBMC-derived T cells. (F) Real-time RT-qPCR detection of extracellular HSV miRs H2 (i), H5 (ii), and H26 (iii) from the spent media of mock- and HSV-1-infected healthy human PBMC-derived T cells. Data shown are from 3 biological repeats. Error bars represent the standard error of the mean. N.D., not detected.

To determine the performance parameters of viral exmiR detection, serial dilutions of synthetic oligonucleotides corresponding to the individual nucleic acid sequence of each viral exmiR of interest were reverse transcribed and assessed by qPCR (Figures 1Ci and 1Cii). A total of 25 viral miRNAs sequences were characterized by real-time RT-qPCR for their assay efficiency, linearity, and linear dynamic range (Figure S1 and Table S1). Assay efficiency of the majority of the viral miRNAs assessed was determined to be between 90% and 110%, with linearity consistently greater than 0.99 (Table S1). Overall, the linear dynamic range of the majority of assayed viral miRNA sequences was determined to be over 5 orders of magnitude, with the LLOQ at approximately 33 aM and the upper limit of quantification (ULOQ) at approximately 3.3 pM (Table S1). For example, synthetic oligonucleotides of representative EBV miR-BART10-3p and HSV miR-H5 demonstrated assay efficiencies of 92% and 100%, respectively, with R² consistently greater than 0.99 (Figure 1D). The linear dynamic range was determined with the LLOQs at approximately 8 and 33 aM for EBV miR-BART10-3p and HSV miR-H5, respectively, and the ULOQs at approximately 3.3 pM for both viral miRNAs (Figure 1D and Table S1). The use of the gold-standard real-time RT-qPCR approach for miRNA detection therefore affords high sensitivity and specificity as well as a broad range of quantification for subsequent investigation of viral exmiRs in this study.

Detecting viral exmiRs from virus-infected ex vivo human T cell cultures

To determine if viral exmiRs can be detected from spent media of virus-infected cells, T cells from a heathy donor were infected with either T cell-tropic EBV-2 (multiplicity of infection [MOI] 10) or HSV-1 (MOI 1). Total exRNA was isolated from 1 mL of spent culture media from 1 million EBV-2- or HSV-1-infected T cells at 24 h post-infection with paired uninfected mock controls. Individual viral exmiRs were quantified by target-specific real-time RT-qPCR and normalized against the respective total exRNA concentration. To assess viral exmiR detection from the spent media of EBV-2-infected T cells, we tested six EBV miRNAs that are located either at the BHRF1 region (miR-BHRF1-1, -1-2-3p, and -1-3) or the BamHI A rightward transcripts (BART) region (miR-BART1-5p and miR-BART6-5p at cluster 1 and miR-BART10-3p at cluster 2). Real-time RT-qPCR analysis revealed the successful detection of all six EBV viral exmiRs tested (Figure 1E). Notably, four of the six EBV exmiRs were detected only in the spent media of EBV-2-infected T cells and not in the mock-infected controls: BHRF1-2-3p (2,964 copies/ng exRNA; Figure 1Eii), miR-BART1-5p (266 copies/ng exRNA; Figure 1Eiv), miR-BART6-5p (84 copies/ng exRNA; Figure 1Ev), and miR-BART10-3p (513 copies/ng exRNA; Figure 1Evi). The other two EBV exmiRs tested showed lower specificity for the respective viral exmiRs over mock-infected controls: EBV miR-BHRF1-1 (220 versus 103 copies/ng exRNA, respectively; Figure 1Ei) and miR-BHRF1-3 (411 versus 57 copies/ng exRNA, respectively; Figure 1Eiii). Similarly, real-time RT-qPCR quantification of three HSV exmiRs from the spent media of HSV-1-infected T cells showed detection of HSV miR-H2 (854 copies/ng exRNA; Figure 1Fi) and miR-H5 (140 copies/ng exRNA; Figure 1Fii) in only HSV-infected samples and not in mock-infected controls, while miR-H26 showed lower specificity for the viral exmiR target (3,530 copies/ng exRNA in HSV-1-infected samples versus 658 copies/ng exRNA in mock-infected samples; Figure 1Fiii). This initial quantification of viral exmiRs from virus-infected T cells revealed that each viral exmiR may be present at different abundances in the extracellular milieu, with some viral exmiRs demonstrating superior detection specificity over others.

To test the assay robustness, we assessed viral exmiR detection by real-time RT-qPCR in the spent media of human T cells infected with either EBV-2 or HSV-1 at different MOIs. CD8⁺ T cells infected with EBV-2 at MOIs of 10, 5, and 1 were quantified for both spent-media-derived EBV exmiRs and cellular EBV genomic DNA as a measure of virus infection (Figures 2Ai and 2Aii). Real-time RT-qPCR quantification of extracellular EBV miR-BART10-3p showed decreasing abundance of the viral exmiR with concomitant decrease in cellular EBV genomic DNA quantified by digital PCR (dPCR; R² = 0.85; Figure S2A). Separately, a Jurkat T cell line infected with HSV-1 at MOIs of 1, 0.1, and 0.01 was likewise quantified for both spent-media-derived HSV exmiRs and cellular HSV genomic DNA as a measure of virus infection (Figures 2Bi and 2Bii). Similarly, decreasing abundance of extracellular HSV miR-H5 was detected with decreasing cellular HSV genomic DNA amounts (R² = 0.77; Figure S2B). We next tested the ability to detect viral exmiRs from decreasing volumes of spent media (0.5, 0.25, and 0.1 mL) from EBV-2- and HSV-1-infected human T cells. As shown in Figures 2C and 2D, we successfully detected EBV miR-BART10-3p and HSV miR-H5 from the spent media of EBV-2- and HSV-1-infected samples down to 0.1 mL (equivalent to spent media from 100,000 cells), with R² values of 0.88 and 0.92, respectively (Figures S2C and S2D). Collectively, the findings suggest that the viral exmiR isolation and detection assay is sensitive (aM detection) and specific, requiring low amounts of starting spent media with a time to result of within 1 day.

Figure 2.

Viral exmiRs detected in the spent media of human T cell cultures

Healthy human PBMC-derived CD8⁺ T cells were infected with EBV-2 (MOI 10),⁴³ Jurkat T cells were infected with HSV-1 (MOI 1), and spent media and/or cells were harvested at 24 h post-infection. (A and B) Spent media of T cell cultures infected using different MOIs of EBV-2 (A) and HSV-1 (B) viruses were profiled for extracellular EBV miR-BART10-3p (Ai) and HSV miR-H5 (Bi), respectively, by real-time RT-qPCR. Cellular EBV (Aii) and HSV (Bii) viral genomic DNA was quantified by dPCR. (C and D) Real-time RT-qPCR detection of extracellular EBV miR-BART10-3p (C) and HSV miR-H5 (D) from different starting input volumes of spent media from the respective virus-infected T cells. (E and F) Spent media from mock-infected T cells and T cells incubated with live virus, heat-inactivated virus, or antibody pre-neutralized live EBV-2 (E) or HSV-1 (F) were quantified for extracellular EBV miR-BART10-3p (Ei) and HSV miR-H5 (Fi), respectively, by real-time RT-qPCR. Cellular EBV (Eii) and HSV (Fii) viral genomic DNA was quantified by dPCR. Data shown are from 3 biological repeats. Error bars represent the standard error of the mean. CV, coefficient of variation; N.D., not detected.

Viral exmiRs are identifiers of live virus infection in human T cells

NAATs allow for the rapid and sensitive identification of viruses by amplifying and detecting specific viral sequences of the viral genetic material but are unable to discriminate between live viruses and dead/inert viral fragments. Since viral miRNAs are produced and secreted from virus-infected cells,^26,27,33,34 we proposed that viral exmiRs could be cell-free viral nucleic acid targets that discriminate between live-virus infection of cells and dead, inert viruses/viral fragments.

To test this hypothesis, human T cells were incubated with live EBV-2 (MOI 10) or HSV-1 (MOI 1), heat-inactivated dead virus, or antibody pre-neutralized virus in which virus entry into cells is impeded. Cells and spent media were quantified for cellular viral genomic DNA and viral exmiRs, respectively, at 24 h post-infection. Real-time RT-qPCR quantification of spent-media-derived EBV miR-BART10-3p revealed an approximately 3-fold higher abundance of the viral exmiR in T cell cultures infected with live EBV-2 (347 copies/ng exRNA) compared to that of heat-inactivated dead virus (115 copies/ng exRNA) and antibody pre-neutralized virus (129 copies/ng exRNA) (Figure 2Ei). Concomitant dPCR quantification of cellular EBV genomic DNA confirmed the reduction of the EBV viral load in T cells incubated with heat-inactivated dead virus (3.7-fold) and antibody pre-neutralized virus (1.8-fold) (Figure 2Eii). Similarly, up to approximately 20-fold higher abundance of HSV miR-H5 was quantified in the spent media of T cell cultures infected with live HSV-1 (3,207 copies/ng exRNA) compared to that of heat-inactivated dead virus (160 copies/ng exRNA) and antibody pre-neutralized virus (719 copies/ng exRNA) (Figure 2Fi). A reduction in HSV viral load in T cells incubated with heat-inactivated dead virus (1,000-fold) and antibody pre-neutralized virus (1.6-fold) was confirmed by dPCR quantification of cellular HSV genomic DNA (Figure 2Fii). We note the relatively higher abundance of extracellular HSV miR-H5 compared to cellular HSV viral load in heat-inactivated dead virus and antibody pre-neutralized virus samples. A plausible explanation may be in part due to the production, secretion, and/or stability/accumulation of HSV exmiRs in the spent media, which remains to be further investigated. Collectively, these findings suggest that viral exmiRs are cell-free identifiers of live-virus infection of cells and are potential discriminants of live versus dead viruses.

Viral exmiRs are identifiers of endogenous viruses in ex vivo human T cell cultures and patient plasma

Screening for endogenous latent viruses is of particular importance for high-risk groups, including immunocompromised individuals and transplant recipients. Despite the absence of viral particle production, viral DNA replication, and restricted viral transcription during viral latency, viral miRNAs are expressed and have been detected during viral latency.^30,31,32,37 We therefore assessed the potential of viral exmiRs as cell-free nucleic acid identifiers of endogenous latent viral infections.

To test this hypothesis, ex vivo peripheral blood mononuclear cell (PBMC)-derived T cells from four healthy donors were first screened for expression of the respective latency genes of four of the most common T cell-tropic endogenous viruses: EBV (Epstein–Barr virus nuclear antigen 1, EBNA1), HHV-6B (U94), HHV-8 (latency-associated nuclear antigen, LANA), and HSV-1 (latency-associated transcript, LAT) using RT-dPCR. Absolute quantification of the viral latent genes identified donor 2 as positive for latent HHV-6, evidenced by the detection of the HHV-6 latent gene U94 (3,182 copies/μg RNA) (Figure 3A). There was no appreciable EBV, HHV-8, and HSV-1 latent gene expression detected in any of the four donor T cells (Figure 3A). To determine if viral exmiRs can be used as cell-free identifiers of HHV-6 latency, we profiled all four known HHV-6 miRs, miR-Ro6-1, -2, -3, and -4, in spent media of the four ex vivo healthy human donor PBMC-derived T cells by real-time RT-qPCR. Abundant levels of extracellular HHV-6 miR-Ro6-4 (1,304 copies/μL) were detected in HHV-6-positive donor 2 (Figure 3B). In contrast, miR-Ro6-4 was absent in the spent media of all HHV-6-negative donors (Figure 3B), demonstrating the specificity of extracellular HHV-6 miR-Ro6-4 in detecting endogenous latent HHV-6. HHV-6 miR-Ro6-1, -2, and -3 were not detected in the spent media from ex vivo T cell cultures of donor 2 (Figure 3B). Our findings suggest that extracellular HHV-6 miR-Ro6-4 is a potential cell-free identifier of endogenous HHV-6 latency and support the utility of harnessing specific viral exmiRs as cell-free biomarkers of endogenous latent viral infections.

Figure 3.

Viral exmiRs identify endogenous viruses in cells

(A) Four healthy human donor PBMC-derived T cell cultures were screened for latency genes of common endogenous viruses in T cells using RT-dPCR: EBV-2 (EBNA1), HHV-6B (U94), HHV-8 (LANA), and HSV-1 (LAT). (B) Profiling of extracellular HHV-6B miR-Ro6-1, -2, -3, and -4 from spent media of T cells from all four donors using real-time RT-qPCR. Data shown are from 3 biological repeats. Error bars represent the standard error of the mean. (C) Table of HCMV IgG, IgM, and DNA status of eight CAR-T cell therapy patients determined by in-house clinical serological assays and Roche COBAS 6800 CMV qPCR test, respectively. (D) Real-time RT-qPCR quantification of four HCMV exmiRs in plasma from all eight patients. +, positive; −, negative; N.D., not detected.

Having demonstrated that specific viral exmiRs are identifiers of endogenous latent viral infections in ex vivo human T cell cultures, we next assessed the utility of viral exmiRs in detecting endogenous viruses in clinical plasma samples. Plasma RNA was similarly isolated using the polyethylene glycol/sodium chloride precipitation-based method from eight CAR-T cell therapy patients who were seropositive for HCMV immunoglobulin (Ig)G and negative for HCMV IgM, indicative of prior HCMV exposure and harboring endogenous latent HCMV (Figure 3C). Notably, HCMV DNA was not detected in blood samples from any of the patients (Roche COBAS 6800 CMV qPCR test; Figure 3C). However, real-time RT-qPCR quantification of four HCMV exmiRs in patient plasma revealed that HCMV miR-US5-2-5p (average of 6,392 copies/200 μL plasma) and miR-US22-5p (average of 880 copies/200 μL plasma) were detected across all eight patients, while miR-US25-2-5p and miR-UL22A-5p were not detected in patient plasma, possibly due to low plasma abundance (Figure 3D). HCMV miR-US5-2-5p and miR-US22-5p assays were verified to be specific for the respective viral miRNA targets (Figure S3). This demonstrates the potential of plasma viral exmiR-based NAAT detection to identify endogenous latent HCMV infection in patients where the commercial viral DNA-based NAAT does not.

Viral exmiR signatures identify endogenous viral reactivation in human B cell models

Endogenous viral reactivation is a major complication in immunosuppressed patients, including hematopoietic stem cell transplantation recipients, and is associated with higher morbidity and mortality and slower immune recovery. However, the current diagnosis of viral reactivation is poorly defined, involving a combination of tests for seroconversion, viral antigen detection, and quantification of viral load, often with presentation of clinical syndromes. Since changes in viral miRNA levels have been reported during viral reactivation,^30,44,45 we hypothesized that altered abundances of viral exmiRs during viral reactivation, which we term viral exmiR reactivation signatures, may distinguish endogenous viral reactivation from latent infections in cells.

To test this hypothesis, we first employed three human B cell lines that harbor endogenous latent EBV: Jijoye, Raji, and Daudi cells. Viral reactivation in the cells was chemically induced using a combination of 12-O-tetradecanoylphorbol-13-acetate (TPA) and sodium butyrate. Successful EBV reactivation was achieved by 24 h post-induction as evidenced by increased BamHI Z fragment leftward open reading frame 1 (BZLF1) expression—the viral gene responsible for EBV reactivation—in induced Jijoye (83-fold; Figure 4Aii), Raji (513-fold; Figure 4Bii), and Daudi cells (13-fold; Figure 4Cii) compared to the non-induced counterparts as determined by RT-dPCR. This was in contrast to the expression of the EBV latent gene EBNA1 that remained generally unchanged during EBV reactivation (Figures 4Ai, 4Bi, and 4Ci). We next profiled eight EBV exmiRs in spent culture media from paired non-induced and 24 h-induced B cells by real-time RT-qPCR. This revealed that EBV miR-BHRF1-2-3p was consistently and significantly elevated in spent media of reactivated EBV cells across all three B cell lines tested: Jijoye (3.2-fold, p = 0.04; Figure 4Di), Raji (2.4-fold, p = 0.03; Figure 4Dii), and Daudi (2-fold, p = 0.01; Figure 4Diii). Of note, we observed significantly elevated levels of EBV miR-BART10-3p in the spent media of reactivated EBV in Raji (2.2-fold, p = 0.008) and Daudi (2.4-fold, p = 0.04) cells but not in those of reactivated Jijoye cells (p = 0.42; Figure S4). We detected elevated levels of other EBV exmiRs, including miR-BART1-5p, in the spent media of EBV-reactivated Daudi cells (1.4-fold, p = 0.04) and miR-BHRF1-1 in the spent media of EBV-reactivated Jijoye cells (2.1-fold, p = 0.03; Figure S4). EBV miR-BHRF1-3 was not detected in the spent media of Raji cells, consistent with a previous study by Pratt et al. that did not detect miR-BHRF1-3 in Raji cells (Figure S4).⁴⁶ In contrast, the abundance of extracellular EBV miR-BHRF1-3, BART6-5p, BART11-3p, and BART13-3p remained unchanged during EBV reactivation across all three cell lines (Figure S4). Together, the findings suggest that the abundance of specific EBV exmiRs is altered during EBV reactivation. Importantly, our findings reveal consistently elevated levels of extracellular EBV miR-BHRF1-2-3p during EBV reactivation, suggesting the utility of this EBV exmiR reactivation signature as a potential biomarker of EBV reactivation in cells.

Figure 4.

Viral exmiRs identify viral reactivation in cells

(A–C) RT-dPCR quantification of cellular EBV genes EBNA1 (i) and BZLF1 (ii) at 6 and 24 h post-chemical induction of EBV-harboring B cells Jijoye (A), Raji (B), and Daudi (C). (D) Real-time RT-qPCR quantification of EBV miR-BHRF1-2-3p in spent media of non-induced and 24 h post-induction Jijoye (i), Raji (ii), and Daudi (iii) cells. (E–H) RT-dPCR quantification of cellular HHV-8 genes LANA and Rta at 6 and 24 h post-chemical induction of BC-1 (E) and BC-3 (G) cells. Real-time RT-qPCR quantification of HHV-8 miR-K12-10a-3p (i), miR-K12-10b (ii), and miR-K12-12-3p (iii) from spent media of non-induced and 24 h post-induction BC-1 (F) and BC-3 (H) cells. Data shown are from 4 biological repeats. Error bars represent the standard error of the mean.

To identify HHV-8 exmiR signatures of HHV-8 reactivation, we used chemical inducers TPA and sodium butyrate to selectively reactivate HHV-8 in BC-1 B cells that harbor endogenous latent HHV-8 and EBV,⁴⁷ as well as in BC-3 cells that harbor endogenous HHV-8 alone. In BC-1 cells, we observed a significant increase in Rta expression—the viral gene responsible for HHV-8 reactivation—at 6 (14-fold) and 24 h (24-fold) post-induction using RT-dPCR (Figure 4Eii). In contrast, only a moderate increase in the HHV-8 latent gene LANA was observed at 6 (3.7-fold) and 24 h (2-fold) post-induction (Figure 4Ei). We next profiled the abundance of three HHV-8 miRNAs that have been associated with HHV-8 reactivation, namely miR-K12-10a-3p, miR-K12-10b, and miR-K12-12-3p.^44,45 Target-specific real-time RT-qPCR profiling of spent-media-derived HHV-8 exmiRs revealed a statistically significant increase in all three HHV-8 reactivation-associated miRs at 24 h post-induction of BC-1 cells: miR-K12-10a-3p (11-fold, p = 0.006; Figure 4Fi), miR-K12-10b (9-fold, p = 0.02; Figure 4Fii), and miR-K12-12-3p (13-fold, p = 0.03; Figure 4Fiii). Consistent with the findings by Miller et al.,⁴⁷ we did not observe EBV reactivation in BC-1 cells as evidenced by the lack of BZLF1 induction at 24 h post-induction (Figure S5). Similarly, successful HHV-8 reactivation was achieved in BC-3 cells as evidenced by a significant increase in Rta expression at 6 (6.4-fold) and 24 h (6.5-fold) following chemical induction using RT-dPCR (Figure 4Gii). An initial increase in the expression of the HHV-8 LANA latent gene was observed at 6 h (6.5-fold) post-induction, but this was not statistically significant at 24 h post-induction (2.2-fold, p = 0.07; Figure 4Gi). Consistent with the findings in BC-1 cells, all three HHV-8 reactivation-associated miRNAs, miR-K12-10a-3p (2.4-fold, p = 0.002; Figure 4Hi), miR-K12-10b (2.6-fold, p = 0.03; Figure 4Hii), and miR-K12-12-3p (2.6-fold, p = 0.02; Figure 4Hiii), were significantly elevated in HHV-8-reactivated BC-3 cells at 24 h post-induction. Together, the findings suggest a role for elevated levels of extracellular HHV-8 miR-K12-10a-3p, miR-K12-10b, and miR-K12-12-3p as candidate biomarkers of HHV-8 reactivation in cells and support a role for viral exmiR reactivation signatures as potential cell-free biomarkers of viral reactivation.

Viral exmiRs are early identifiers of viral infection and reactivation in cells

To assess if viral exmiRs can be early identifiers of viral infection, we profiled spent-media-derived viral exmiR abundance and cellular viral transcription at regular intervals following viral infection or reactivation. HSV-1 infection of T cells from a healthy donor was performed and profiled in two phases: (1) first 12 h post-infection and (2) 12–24 h post-infection. Viral transcription was measured using RT-dPCR, and the corresponding spent media were profiled by real-time RT-qPCR. The characteristic HSV gene expression kinetics cascade was observed where immediate-early and early genes infected cell protein 27 (ICP27) and thymidine kinase (tk) were expressed by 2 h post-infection (Figures 5A and 5B), and the late gene glycoprotein C (gC) was expressed at 5–6 h post-infection (Figure 5C). Cellular expression of all three viral genes remained elevated up to 24 h post-infection (Figures 5A–5C). Real-time RT-qPCR quantification of extracellular HSV-1 miR-H2 from the respective spent media revealed consistent detection of the viral exmiR as early as 8 h after viral infection of T cells and remained elevated up to 24 h post-infection (Figure 5D).

Figure 5.

Viral exmiRs detected as early as 8 h after virus infection of T cells and 6 h after virus reactivation

(A–C) Cellular expression of HSV-1 genes ICP27 (A), tk (B), and gC (C) at regulator intervals up to 24 h following HSV-1 infection of healthy human donor PBMC-derived T cell cultures. HSV-1 viral gene expression was quantified using RT-dPCR and normalized to β-actin. (D) Real-time RT-qPCR detection of HSV miR-H2 in spent media from the respective indicated time points over 24 h post-infection. (E) RT-dPCR quantification of cellular HHV-8 lytic switch Rta over 24 h in non-induced and chemically induced BC-1 cells. (F) Real-time RT-qPCR detection of HHV-8 miR-K12-10a-3p in spent media from the respective indicated time points over 24 h post-induction. p values derived from the comparison of HHV-8 miR-K12-10a-3p levels in paired non-induced and induced spent media at each indicated time point were corrected using the Benjamini-Hochberg procedure with FDR 0.1. Data shown are from 3 biological repeats. Error bars represent the standard error of the mean. N.D., not detected.

To determine how early viral exmiR signatures can be detected following chemically induced viral reactivation, we monitored HHV-8 reactivation in BC-1 B cells at regular intervals over two time periods: (1) first 12 h and (2) 12–24 h after treatment with TPA and sodium butyrate. RT-dPCR profiling of the HHV-8 lytic switch Rta confirmed HHV-8 reactivation with maximum Rta expression at 10 h post-induction (Figure 5E). Rta expression remained elevated up to 24 h post-induction (Figure 5E). Next, we quantified extracellular HHV-8 miR-K12-10a-3p from the respective spent media by real-time RT-qPCR and observed significantly increased HHV-8 miR-K12-10a-3p levels as early as 6 h after HHV-8 reactivation (Benjamini-Hochberg adjusted p value, false discovery rate [FDR] 0.1; Figure 5F). Extracellular HHV-8 miR-K12-10a-3p levels remained elevated up to 24 h post-induction (Figure 5F). Together, our findings suggest that viral exmiRs and viral exmiR reactivation signatures can be detected as early as 8 h after viral infection and 6 h after viral reactivation. These observations support a role for viral exmiRs as potential candidate biomarkers for the early detection of virus infection and/or reactivation.

Discussion

Rapid and sensitive virus detection is important for the prevention, control, and management of viral disease in areas such as human health, agriculture, and aquaculture. While molecular tests such as NAATs have revolutionized the speed and sensitivity at which viral infections can be detected, more informative biomarkers are needed to discriminate between live-virus infections and dead/inert viral fragments and latent versus reactivated viral infections. Our study presents evidence for viral exmiRs as a class of cell-free nucleic acid targets that add critical information of the virus of interest to fast and sensitive NAATs. We show that spent-media-derived short, non-coding viral exmiRs are (1) cell-free, early identifiers of live virus infection in T cells, (2) identifiers of latent endogenous viruses in human donor cells and patient plasma, and (3) potential biomarkers of endogenous viral reactivation (summarized in Table 1). By employing phenol-free precipitation-based exRNA isolation combined with gold-standard real-time RT-qPCR detection with fast cycling (30 min) for miRNAs, our study achieved fast and sensitive viral exmiR detection with a time to result of under 1 day and at aM sensitivity. Viral exmiR detection in total exRNA afforded superior sensitivity in identifying live-virus infection and viral reactivation events over vesicle- or non-vesicle-bound fractions (Figures S6i and S6ii). As cell-free viral nucleic acid targets of NAATs, viral exmiRs thus enable NAATs with the capability to discriminate between live versus dead and latent versus reactivated viruses by sampling spent culture media or potentially other cell-free samples.

Table 1.

Summary of viral exmiR biomarkers of live adventitious, endogenous latent, and reactivated viruses identified in this study

Virus	Live adventitious	Endogenous latent	Endogenous reactivated
HHV-4/EBV	miR-BART10-3p	miR-BHRF1-2-3p	↑ miR-BHRF1-2-3p
HHV-1/HSV-1	miR-H5	N/D	N/D
HHV-5/HCMV	N/D	miR-US5-2-5p miR-US22-5p	N/D
HHV-6	N/D	miR-Ro6-4	N/D
HHV-8	N/D	miR-K12-10a-3p, miR-K12-10b, miR-K12-12-3p	↑ miR-K12-10a-3p, ↑ miR-K12-10b, ↑ miR-K12-12-3p

N/D, no data/not tested.

Fit-for-purpose tests for adventitious virus detection are urgently needed to facilitate the development of viral safety strategies in cell therapy manufacturing. Autologous cell products demand rapid, sensitive, and informative virus detection modalities that can expedite product safety release and inform on potentially life-threatening viral contaminants prior to administration to the patient. Fast and sensitive NAATs such as PCR-based tests have been increasingly adopted to complement slow culture-based virus tests to expedite product release testing, reducing the time to result from weeks to just a few hours. However, since PCR-based methods amplify and detect specific nucleic acid sequences of the viral genomic material, such tests are unable to discriminate between live viruses versus dead or inert viral fragments. This study demonstrates the utility of viral exmiR candidate biomarkers—products of virus transcription in virus-infected cells—in conferring NAATs the ability to discriminate between potentially dangerous live-virus contaminants and dead/inert viral fragments (Figure S7), informing clinical decision-making on administering the life-saving treatment to the patient and patient management. Our findings further suggest the potential of viral exmiR-based NAATs to detect adventitious virus contamination early, as early as 8 h after infection in the case of HSV-1-infected T cell cultures. That viral exmiRs can be mined from the extracellular milieu without destroying precious cell products further offers the potential for in-process testing of adventitious virus contamination from spent culture media at critical points during the cell manufacturing process.

Several challenges remain in the development of viral exmiR-based NAATs for adventitious virus contamination detection in cell manufacturing. In general, targeted approaches for viral testing, such as NAATs, are limited by the current lack of a priori knowledge of the identities of adventitious viruses encountered in cell therapy manufacturing. This can be overcome through collaborative efforts between industry, clinical, and academic cell therapy manufacturers on their shared experiences of adventitious agent contamination in that space. Another challenge in viral exmiR biomarker development is the limited a priori knowledge of RNA virus-encoded miRNAs. Unlike DNA virus-encoded miRNAs, which are well studied, there is comparatively far less corroborative evidence for RNA virus-encoded miRNAs. To facilitate a more comprehensive detection of both DNA and RNA adventitious viral contamination in manufactured cell products, ongoing efforts are focused on characterizing other classes of extracellular viral RNA, including other non-coding RNA and subgenomic RNA, using untargeted approaches such as next-generation sequencing.^48,49,50 Additionally, as altered abundances of human host miRNAs have been implicated in viral infections,^51,52,53 an in-depth characterization of infection-associated host RNA signatures may reveal potential candidate biomarkers of pan-viral infection of cells.

Timely virus detection is especially important in high-risk groups, including in immunosuppressed patients and transplant recipients, for whom viral reactivation-associated disease causes significant morbidity and mortality. Fast PCR-based quantification of viral DNA load is commonly used in clinical practice but is limited by inter-laboratory and inter-assay variability and the lack of universal viral load thresholds to define viral reactivation, inform on the initiation of therapy, and viral clearance versus persistence. Such tests are also often performed when the patient already presents with clinical syndromes rather than as a mode of early detection for viral reactivation, which would inform clinical decision-making and improve patient management for virus-associated post-transplant/cell therapy complications. The findings of our study suggest the utility of viral exmiR signatures for the early detection of endogenous viral reactivation and suggest the merits of identifying and characterizing viral exmiR reactivation signatures in other clinically important viruses, including members of the herpesvirus family HHV-3, -5 (HCMV), -6, and -7.^{13,21,23,54,55} Having demonstrated viral exmiR detection in the plasma of patients with endogenous latent HCMV, ongoing efforts are focused on validating the findings of this study in patients and identifying plasma viral exmiR reactivation signatures of these clinically important viruses in a larger study with the respective virus-negative control groups. Future work will also focus on improving the sensitivity of viral exmiR detection in the complex matrix of patient plasma, which may inhibit NAATs, using innovative sample preparation methods to enrich for exRNA while removing such inhibitors. Viral exmiR-based reactivation biomarkers would open opportunities toward a new paradigm for holistic patient management for the early detection of viral reactivation from pre-treatment viral exmiR screening to post-treatment monitoring for viral exmiR reactivation signatures.

In summary, this study demonstrates the utility of viral exmiRs as cell-free biomarker candidates of live, latent, and reactivated viral infections that harness the speed and sensitivity of NAAT-based virus detection. Our findings form the foundation to develop rapid, sensitive, and cell-free detection of adventitious viruses and endogenous reactivated viruses to expedite in-process/release testing of cell therapy products as well as early detection of viral reactivation events in patients. Beyond its application in human health, viral exmiR-based detection may present opportunities for the prevention and control of devastating viral disease outbreaks in the agriculture and aquaculture settings. Ongoing efforts to streamline workflows and improve the cost effectiveness of viral exmiR-based detection would make it a valuable virus detection modality with potentially broad applications.

Materials and methods

Cell lines

The Vero cell line (ATCC CCL-81) was cultured in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific). Jurkat T cells, B cell lines (ATCC CCL-87, CCL-213, CCL-86, CRL-2230, CRL-3615), and the Kasumi cell line (ATCC CRL-2725) were cultured in Roswell Park Memorial Institute (RPMI) 1640 Medium (Thermo Fisher Scientific). Media were supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific), 100 IU/mL penicillin, and 100 μg/mL streptomycin. Media for the Kasumi cell line were supplemented with 20% FBS. Cells were cultured in 5% CO₂ at 37°C.

Human T cell isolation, activation, and expansion

Frozen human PBMCs from healthy donors were purchased from STEMCELL Technologies or were isolated from a cryopreserved leukopak (Lonza Bioscience) by Ficoll density gradient centrifugation (Cytiva). T cells were isolated using the EasySep Human T cell Isolation Kit (STEMCELL Technologies) and resuspended in AIM V Medium (Thermo Fisher Scientific) supplemented with 2% human male AB serum (Merck Sigma-Aldrich) and 100 IU/mL recombinant human interleukin (IL)-2 (Miltenyi Biotec). CD8⁺ T cell selection was performed using the EasySep Human CD8⁺ T cell Isolation Kit (STEMCELL Technologies). Purified T cells were activated using DynaBeads Human T-Expander CD3/CD28 (Thermo Fisher Scientific) and cultured in a G-Rex 24-well plate (Wilson Wolf). Expanded T cells were used for downstream experiments. All healthy human donor PBMC-derived T cells were negative for EBV and HSV.

Clinical plasma samples and plasma RNA isolation

Clinical plasma samples were obtained from eight CAR-T cell therapy patients (SingHealth Centralized Institutional Review Board approval no. 2022/2322) with known HCMV status. All patients were HCMV IgG positive and IgM negative as previously determined using in-house clinical serological assays. No HCMV DNA was detected in EDTA blood samples from any of the patients using the Roche COBAS 6800 CMV qPCR test. RNA was isolated from 200 μL of plasma using ExoQuick RNA isolation with thrombin pre-treatment (System Biosciences).

Viruses

HSV-1 (ATCC VR-260) was propagated using Vero cells, purified using iodixanol density gradient ultracentrifugation, and titered by plaque assay. EBV type 2 was produced by treating the Jijoye cell line (ATCC CCL-87) with 4 mM sodium butyrate and 25 ng/mL TPA and concentrated by ultracentrifugation. Genome copy numbers of EBV-2 were obtained by absolute quantification using dPCR (QIAcuity RT-dPCR System, QIAGEN) following viral DNA extraction using the QIAamp DNA Mini Kit (QIAGEN) according to the manufacturer’s instructions. HCMV (VR-538) was purchased from ATCC.

Virus infection

1 × 10⁶ healthy human donor PBMC-derived T cells or Jurkat T cells were infected with EBV-2 (MOI 10) or HSV-1 (MOI 1) for 1 h at 37°C with intermittent rocking. The following parameters were used to heat inactivate viruses: 80°C for 1 h (EBV-2) and 60°C for 20 h (HSV-1). For antibody pre-neutralization studies, viruses were incubated with the respective neutralizing antibody: rabbit polyclonal EBV GP350 antibody (PA5-81825, Thermo Fisher Scientific) or mouse monoclonal HSV1 + HSV2 gD antibody 2C10 (ab6507, Abcam) for 1 h at 37°C prior to the infection of T cells. Cells were washed with PBS, grown in fresh medium, and harvested at the indicated times.

exRNA isolation

Small extracellular vesicles and macromolecules were harvested from spent culture media by polyethylene glycol/sodium chloride-based precipitation using ExoQuick (System Biosciences) followed by RNA extraction using the Exosomal RNA Isolation Kit (Norgen Biotek) according to the manufacturer’s instructions. For experiments analyzing vesicle- and non-vesicle-bound compartments, RNA from the respective compartments was isolated using the Plasma/Serum Exosome Purification Mini Kit (Norgen Biotek) according to the manufacturer’s instructions. Total exRNA was quantified using the Quant-iT microRNA Assay Kit (Thermo Fisher Scientific) following DNase I (Thermo Fisher Scientific) digestion at 37°C for 90 min.

Viral exmiR quantification by real-time RT-qPCR

Target-specific reverse transcription was performed using the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems) with the respective TaqMan MicroRNA Assays (Applied Biosystems) according to the manufacturer’s instructions. Briefly, 5 μL of total exRNA was used in the RT reaction with the following parameters: 16°C, 30 min; 42°C, 30 min; 85°C, 5 min and then kept on ice. The RT product was diluted with an equal volume of RNase-free water. qPCR was performed using 2 μL of RT product in a 10 μL reaction volume with TaqMan Fast Advanced Master Mix (Applied Biosystems) on the StepOnePlus Real-Time qPCR System with fast cycling mode (Applied Biosystems) using the following parameters: 95°C, 20 s; 40 cycles of 95°C, 1 s and 60°C, 20 s. Serial dilutions of synthetic oligonucleotides with the nucleic acid sequence corresponding to the viral miRNA of interest were subjected to real-time RT-qPCR to obtain the respective standard curves from which the absolute number of copies of the viral miRNA of interest was determined. Viral miRNA abundance was normalized to the respective total exRNA amounts for all real-time RT-qPCR analyses unless indicated otherwise.

Cellular DNA and RNA extraction and reverse transcription

Genomic DNA was extracted using the QIAamp DNA Mini Kit (QIAGEN) according to the manufacturer’s instructions. Total cellular RNA was extracted using the RNeasy Mini Kit with DNase treatment (QIAGEN) according to the manufacturer’s instructions. Nucleic acid quantification was performed using a NanoDrop spectrophotometer (Thermo Fisher Scientific). For the reverse transcription of cellular RNA, 0.5–1 μg of total RNA was used with random hexamers in the SuperScript IV First-Strand Synthesis System (Thermo Fisher Scientific) according to the manufacturer’s instructions.

dPCR

Absolute quantification of target genes was performed using dPCR (QIAcuity Digital PCR System, QIAGEN) using either the QIAcuity EG PCR Kit or the QIAcuity Probe PCR Kit (QIAGEN) according to the manufacturer’s instructions. Primers and probes used are listed in Table S2.

Chemical induction of B cell lines

B cell lines CCL-86, CCL-87, CCL-213, CRL-2230, and CRL-3615 (ATCC) harboring EBV and/or HHV-8 were chemically induced using 25 ng/mL TPA and 4 mM sodium butyrate. Cells and spent media were harvested at the indicated time points.

Statistical analysis

All experiments were performed with a minimum of two technical replicates and 3 biological replicates. Statistical significance was determined using Student’s t test: ∗p < 0.05 and ∗∗p < 0.01. In cases of multiple hypothesis testing, p values were corrected using the Benjamini-Hochberg procedure with FDR 0.1.

Data and code availability

The authors declare that all data are available in the main text or the supplemental information. Correspondence and requests for materials should be addressed to S.L.S.

Acknowledgments

This research is supported by the National Research Foundation, Prime Minister's Office, Singapore, under its Campus for Research Excellence and Technological Enterprise (CREATE) program, through the Singapore-MIT Alliance for Research and Technology (SMART): Critical Analytics for Manufacturing Personalized-Medicine (CAMP) Inter-Disciplinary Research Group, and supported in part by the Singapore National Research Foundation and Ministry of Education under the Research Centre of Excellence Programme. The graphical abstract was created in part in BioRender (https://biorender.com/c35l993).

Author contributions

C.C. conceptualized the research and designed and planned the experiments. C.C. and J.X.Y.L. performed the experiments and analyzed the data. W.-X.S. contributed to the experimental design and planning of healthy human PBMC-derived T cell culture experiments. D.B.L.T. and W.-X.S. provided all healthy human PBMC-derived T cell cultures. Y.C., F.L.W.I.L., and C.N. provided all clinical samples, and N.K.Z.T. processed the clinical samples. C.C. wrote the manuscript, and S.L.S. and R.B.H.W. reviewed and edited the manuscript. S.L.S., R.B.H.W., and M.E.B. provided mentorship and feedback. All authors provided critical feedback and contributed to the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.