Viral opportunistic infections in Mauritian cynomolgus macaques undergoing allogeneic stem cell transplantation mirror human transplant infectious disease complications

Abstract

Allogeneic hematopoietic stem cell transplantation (HSCT) and xenotransplantation are accompanied by viral reactivations and virus-associated complications resulting from immune deficiency. Here, in a Mauritian cynomolgus macaque model of fully MHC-matched allogeneic HSCT, we report reactivations of cynomolgus polyomavirus, lymphocryptovirus, and cytomegalovirus, macaque viruses analogous to HSCT-associated human counterparts BK virus, Epstein-Barr virus, and human Cytomegalovirus. Viral replication in recipient macaques resulted in characteristic disease manifestations observed in HSCT patients, such as polyomavirus-associated hemorrhagic cystitis and tubulointerstitial nephritis or lymphocryptovirus-associated post-transplant lymphoproliferative disorder. However, in most cases, the reconstituted immune system, alone or in combination with short-term pharmacological intervention, exerted control over viral replication, suggesting engraftment of functional donor-derived immunity. Indeed, the donor-derived reconstituted immune systems of two long-term engrafted HSCT recipient macaques responded to live attenuated yellow fever 17D vaccine (YFV 17D) indistinguishably from untransplanted controls, mounting 17D-targeted neutralizing antibody responses and clearing YFV 17D within 14 days. Together, these data demonstrate that this macaque model of allogeneic HSCT recapitulates clinical situations of opportunistic viral infections in transplant patients, and provides a pre-clinical model to test novel prophylactic and therapeutic modalities.

INTRODUCTION

Infectious disease complications are a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (HSCT) and xenotransplantation {Haustein:2008ws, Boneva:2001gc, ^1,2. BK Polyomavirus (BKV), Epstein-Barr virus (EBV), and human Cytomegalovirus (HCMV) are ubiquitous, persistent, opportunistic viruses and major causative agents of disease in allogeneic HSCT patients^3-9. In the immunodeficient setting of allogeneic HSCT, lack of effective antiviral immunity can result in viral reactivation and serious virus-associated complications, including BKV-associated hemorrhagic cystitis and nephropathy, EBV-associated post-transplant lymphoproliferative disorder (EBV-PTLD), and HCMV-associated pneumonia and colitis^10-16. Limited antivirals and therapies exist to address these viral complications in allogeneic HSCT patients and are accompanied with significant toxicitiy¹². Thus, new treatments are needed to reduce opportunistic infection-mediated patient morbidity and mortality. The development of such novel antiviral therapeutics would be greatly accelerated by an animal model that recapitulates human viral reactivations in the setting of allogeneic HSCT.

We recently developed a nonhuman primate model of allogeneic HSCT using Mauritian cynomolgus macaques¹⁷, and sought to determine the competence of reconstituted immune systems post-transplant. To this end, we first examined the extent of viral reactivation and disease emergence in transplanted recipient macaques, including both SIV-naïve macaques and SIV-infected macaques maintained on antiretroviral therapy with fully suppressed SIV viremia. Macaques are ubiquitously infected with polyomaviruses, lymphocryptoviruses, and cytomegaloviruses, viruses closely related to human viruses BK, EBV, and HCMV, respectively¹⁸. Due to the close evolutionary and physiological relationship between humans and macaques and between human and macaque viruses^19,20, we hypothesized that allogeneic HSCT recipient Mauritian cynomolgus macaques would experience reactivations of macaque viruses analogous to human viruses known to reactivate and cause disease in human allogeneic HSCT recipients. Second, to directly test the competence of post-transplant reconstituted immune systems, we examined the response to live attenuated yellow fever vaccine 17D (YFV 17D) in two allogeneic HSCT recipient macaques vaccinated 26-29 months post-transplant and possessing fully donor-derived immune systems. Our results demonstrate (1) reactivation of human-analogous, clinically relevant opportunistic viruses in Mauritian cynomolgus macaques undergoing allogeneic HSCT, and (2) the ability of HSCT-reconstituted macaque immune systems to eventually control opportunistic viral replication and respond to and clear a live attenuated virus vaccine. Thus, this pre-clinical animal model can be used in the development of novel antivirals and therapeutic regimens for allogeneic HSCT-associated opportunistic infections. Furthermore, because viral reactivation and disease are major adverse events in preclinical macaque models of xenotransplantation during immunosuppression, the results presented here will benefit these translational animal studies by improving diagnosis and treatment of opportunistic viral infections.

MATERIALS AND METHODS

Animals

Mauritian cynomolgus macaques (Macaca fascicularis), 4–10 years of age, were housed at the Oregon National Primate Research Center (ONPRC) under the approval of the Oregon Health and Science University (OHSU) West Campus Institutional Animal Care and Use Committee (IACUC). All macaques in this study were managed according to the ONPRC animal care program, which is fully accredited by AAALAC International and is based on the laws, regulations, and guidelines set forth by the United States Department of Agriculture (e.g., the Animal Welfare Act and its regulations, and the Animal Care Policy Manual), Institute for Laboratory Animal Research (e.g., Guide for the Care and Use of Laboratory Animals, 8th edition), and the Public Health Service Policy on Humane Care and Use of Laboratory Animals. The nutritional plan utilized by the ONPRC is based on National Research Council recommendations and supplemented with a variety of fruits, vegetables, and other edible objects as part of the environmental enrichment program established by the Behavioral Management Unit. Paired/grouped animals exhibiting incompatible behaviors were reported to the Behavioral Management staff and managed accordingly. All efforts were made to minimize suffering through the use of minimally invasive procedures, anesthetics, and analgesics when appropriate. During the peri-transplant period, animals had indwelling subclavian catheters and were placed on a tether catheter protection system, which allowed for full range of motion as well as blood collection and drug delivery without sedation¹⁷. Animals were acclimated to the tether catheter protection system -- approximately 30-40 days prior to transplantation, prior to catheterization, animals were placed in the jacket and on tether. Where indicated, animals were painlessly euthanized with sodium pentobarbital and euthanasia was assured by exsanguination and bilateral pneumothorax, consistent with the recommendations of the American Veterinary Medical Guidelines on Euthanasia (2013).

Pre-transplant conditioning, peri-transplant care, and immunosuppression.

HSCT recipient Mauritian cynomolgus macaques were transplanted with leukapheresis product from mobilized, MHC-matched unrelated donors as previously described¹⁷. This study includes 11 HSCT recipient macaques, including 4 SIV-naïve macaques and 7 SIVmac239-infected treated with once daily injectable combination antiretroviral therapy consisting of 5.1 mg/kg tenofovir disoproxil fumarate, 40 mg/kg emtricitabine, and 2.5 mg/kg dolutegravir. In all SIVmac239-infected macaques, SIV viremia was fully suppressed throughout the study described here. Detailed transplant regimens for each HSCT recipient macaque studied are shown in Figure S4. Nine of eleven studied HSCT recipient Mauritian cynomolgus macaques (MCM) underwent reduced intensity conditioning with a previously described regimen consisting of CD3-immunotoxin (acquired through the NHP Reagent Resource, Boston, MA), busulfan, and total body irradiation (TBI, delivered at 10-25 cGy/min using an Elektra Synergy system) ¹⁷. Two of eleven studied HSCT recipient MCM underwent reduced intensity conditioning with a slightly altered regimen of CD3-immunotoxin and two doses of busulfan, without TBI. All eleven studied recipient MCM received anti-GVHD prophylaxis consisting of daily tacrolimus given as an intramuscular (IM) injection SID, with dose adjusted to maintain a tacrolimus trough level of 5-15 ng/ml in whole blood (measured by immunoassay, Abbott Architect i2000). Prior to tacrolimus discontinuation, tacrolimus dose was tapered by 20% each day until dose fell below 0.002 mg/kg, at which point administration was discontinued. Ten of eleven studied HSCT recipients also received additional anti-GVHD prophylaxis of post-transplant cyclophosphamide and belatacept. On days of cyclophosphamide treatment, animals received IV Normosol at approximately 15-20 ml/kg over 6-8 hours as well as bladder protectant MESNA (40 mg/kg IV) to prevent sterile hemorrhagic cystitis. One of eleven studied HSCT recipients (ID 35132) received an altered anti-GVHD prophylaxis regimen consisting of tacrolimus and abatacept, without post-transplant cyclophosphamide. Peri-transplant care for all recipients consisted of polymyxin B, neomycin sulfate, cefazolin, and trimethoprim-sulfamethoxazole (30 mg/kg PO, BID from day of cefazolin discontinuation to approximately day +60). Enrofloxacin was administered (10 mg/kg IV, SID) when absolute neutrophil count was less than 1,000/ml of whole blood and continued until absolute neutrophil count exceeded 1,000/ml of whole blood for two consecutive complete blood counts. Some macaques received ondansetron (0.1-0.2 mg/kg PO, SID or BID) and dronabinol (2.5 mg PO, SID to BID) to treat nausea and anorexia. Four of eleven HSCT recipients were treated with prednisone (up to 4 mg/kg/day PO) or methylprednisolone sodium succinate (up to 5 mg/kg/day) to treat suspected or confirmed GVHD. Two macaques received rhesus recombinant anti-CD8α depleting monoclonal antibody (M-T807R1, acquired through the NHP Reagent Resource). Four macaques received rhesus recombinant anti-CD20 depleting monoclonal antibody (CD20-2B8R1, acquired through the NHP Reagent Resource). Serum chemistry and complete blood counts were performed 2-3 times weekly or as clinically indicated. Leukoreduced whole blood or packed red blood cells were transfused when hemoglobin was less than 6 g/dl or hematocrit was less than 20%. Platelet-rich plasma was transfused when platelet counts were less than 20,000 cells/μl blood or if signs of coagulopathy were observed on physical examination. All blood product transfusions were matched according to the ABO blood group antigen system. Animals were treated with intravenous fluids and nephrotoxic drug doses were adjusted as needed when azotemia was detected. Animals received appropriate supportive veterinary care under the direction of an ONPRC veterinarian. Any animal suspected of being in discomfort received analgesics.

Virus detection and monitoring

To detect and measure viral DNA and/or RNA, plasma, urine, or cell culture supernatant was concentrated by centrifugation at >20,000 rcf at 4 °C for 1 h, after which supernatant was removed to leave 200 μl volume. Viral nucleic acid was extracted from plasma using QIAamp MinElute Virus Spin Kit (Qiagen, Venlo, Netherlands) according to manufacturer’s instructions. Polyomavirus DNA was initially detected by polymerase chain reaction (PCR), which was carried out in 25 μl reactions by combining 8.5 μl extracted viral nucleic acid, 12.5 μl Phusion Hot Start Flex 2X mastermix (New England Biolabs, Inc., Ipswich, MA), and 2 μl of each 5 μM primer (forward: 5′-TAGGTGCCAACCTATGGAAC-3′; reverse: 5′-GGAAAGTCTTTAGGGTCTTCTACC-3′) with the following cycling conditions: 98°C 02:00/40 cycles: 98°C 00:15, 52°C 00:20, 72°C 01:00/72°C 3:00. Primers were designed based on previously published sequences for cynomolgus polyomavirus²¹. Reactions were run on a 2% agarose gel alongside TrackIt 100 bp DNA ladder (Invitrogen, Carlsbad, CA). For sequencing, the 182-bp bands corresponding to polyomavirus large T antigen amplicons were excised, purified using the Nucleospin Gel and PCR cleanup (Macherey-Nagel, Bethlehem, PA), Sanger sequenced using the same primers, and analyzed in Geneious (Biomatters Ltd., Auckland, New Zealand). Quantitative PCR probes for polyomavirus were designed based on the resulting sequences (Table S4). Quantitative PCR primers and probes targeting CyLCV, CyCMV, and YFV 17D were designed based on previously published sequences [Genbank accession numbers KP676001.1 (CyLCV), KX689265 (CyCMV), U17066.1 (YFV 17D), Table S4]. For YFV 17D viral load, cDNA was reverse-transcribed from 10 μl extracted viral nucleic acid prior to quantitative PCR using the RevertAid first strand cDNA synthesis kit (Thermo Scientific, Waltham, MA) as per manufacturer’s instructions, after which 30 μl water was added to dilute the cDNA. Quantitative PCR targeting was carried out in 20 μl reactions by combining 5 μl of extracted viral nucleic acid (polyomavirus, CyLCV, CyCMV) or cDNA (YFV17D), 10 μl TaqMan Fast Advanced Master Mix (Life Technologies, Carlsbad, CA), 1.5 μl of each 6 μM primer, and 2 μl of 2.5 μM probe. Sample reactions were run in duplicate in MicroAmp Fast Optical 96-well reaction plates (Applied Biosystems, Foster City, CA) alongside no template control reactions and standard reactions containing ten-fold serial dilutions of standard (1 copy to 10⁷ copies per reaction, in duplicate). Plasmids (pCEP4 or pUC57) containing each target region served as standards. Plates were run on a StepOnePlus Real-Time PCR system (Applied Biosystems) with the following conditions: 95 °C 00:20, followed by 40 cycles of 95 °C 00:01, 60 °C 00:20. Quantities in each sample reaction were calculated according to the standard curve. For polyomavirus, CyLCV, and CyCMV viral loads, quantities were adjusted for the proportion of viral nucleic acid added to the reaction and initial sample volume to obtain genome copies per milliliter plasma, urine, or cell cultures supernatant. For YFV 17D plasma viral loads, quantities were adjusted for the proportion of viral nucleic acid added to the cDNA synthesis, proportion of cDNA added to the quantitative PCR reaction, and the initial plasma volume to obtain RNA copies/ml plasma. Theoretical limits of detection (LODs) are based on the ability of quantitative PCR assays to reliably detect standard reactions containing 10 copies. For CD20+ cell-associated CyLCV assays, DNA was extracted from CD20+ cells sorted from ACK-treated blood or from tissue mononuclear cell suspensions (see “Flow cytometric analysis and cell sorting” section) using Qiagen DNeasy kit (Qiagen) and eluted in 50 μl water. Quantitative PCR reactions were modified to include primers and probe targeting the cynomolgus macaque albumin gene (Genbank accession number U17066.1, Figure S3): 5 μl of extracted cell-associated DNA, 10 μl TaqMan Fast Advanced Master Mix (Life Technologies), 0.25 μl of each 36 μM primer (CyLCV-F, CyLCV-R, albumin-F, albumin-R, and 2 μl of each 2.5 μM probe (CyLCV and albumin). Cell number input for each quantitative PCR reaction was calculated by dividing the albumin quantity by 2. Cell-associated viral load (copies per 1×10⁶ cells) was calculated by dividing the CyLCV quantity by cell number input, then multiplying by 1×10⁶.

CyCMV treatment

When CyCMV plasma viral loads exceeded 1,000 copies/ml, treatment with valganciclovir (10 mg/kg PO) was initiated SID. If plasma viral loads were not decreased by the subsequent timepoint (3-4 days later), valganciclovir treatment was escalated to BID and continued until plasma viral loads fell below 1,000 copies/ml. If plasma viral loads increased above 15,000 copies/ml despite BID valganciclovir treatment, cidofovir was administered weekly (5 mg/kg IV) with SID valganciclovir until viremia remained below 15,000 copies/ml for one week, at which point cidofovir was discontinued and BID valganciclovir was administered until complete resolution of viremia. Nine HSCT recipient MCM experiencing CMV viremia of >1,000 copies/ml plasma were treated with valganciclovir. Two HSCT recipient MCM experiencing CMV viremia of >15,000 copies/ml were treated with valganciclovir and cidofovir, which was administered with kidney protectant probenecid (500 mg PO BID on the day of cidofovir treatment).

CyCMV immunity

CMV-specific T cell responses were assessed by interferon-γ enzyme-linked immunospot assay (IFN-γ ELISPOT) as previously described¹⁷ using Monkey IFN-γ ELISPOT Plus ALP kit (Mabtech, Cincinnati, OH), following manufacturer’s instructions. ELISPOT assays were performed in duplicate, stimulating 100,000 peripheral blood mononuclear cells per well with rhesus CMV viral lysate and pools of 15-mer peptides (11 amino-acid overlap, 7–13 peptides per pool) spanning the RhCMV pp65a, pp65b, IE-1, and IE-2 proteins (Genbank accession #AY186194). Negative control wells were incubated without antigen and served as background controls; positive control wells were incubated with Concanavalin A.

Flow cytometric analysis and cell sorting

Immune subset frequencies and T cell activation in HSCT recipients were monitored longitudinally by whole blood staining as previously described¹⁷. EDTA-treated whole blood (50-100μl) was washed twice with PBS and stained for CD45, CD3, CD20, CD4, CD8, CD14, CD28 and CD95, and viability (Live/dead Yellow Fixable, Invitrogen) at room temperature for 30 minutes. After staining, whole blood was resuspended in 1ml FACS Lysing Solution (BD Biosciences) to lyse red blood cells and fix remaining cells, incubated at room temperature for 8 minutes, washed three times with FACS buffer, and collected on a BD LSRII instrument. All flow cytometric analysis was performed using Flow Jo (Tree Star, Inc., Ashland, OR). CD20+ frequency was defined as the percentage of CD20+CD3− cells within the live CD45+ singlet population. CD8+ T cell frequency was defined as the percentage of CD3+CD8+CD4− cells within the live CD45+ singlet population. Absolute cell counts were calculated by multiplying whole blood staining frequencies by the white blood cell count, as determined by complete blood count performed on an additional aliquot of EDTA-treated whole blood. To assess levels of CD8+ T-cell activation in YFV 17D-vaccinated macaques, EDTA-treated whole blood (50-100μl) was washed twice with PBS and stained with two panels: (1) CD45, CD3, CD4, CD8, HLA-DR, CD38, and viability (Live/dead Yellow Fixable, Invitrogen), (2) CD45, CD3, CD4, CD8 at room temperature for 30 minutes. After staining, whole blood was resuspended in 1ml FACS Lysing Solution (BD Biosciences) to lyse red blood cells and fix remaining cells, incubated at room temperature for 8 minutes, and washed three times with FACS buffer. Tubes stained with panel 2 were stained intracellularly as follows: Pellets were resuspended in 500 μl FACS perm (FACS Lysing solution with 0.05% Tween-20), incubated for 10 minutes, washed three times with FACS buffer, and stained for Ki67 and Bcl2 for 45 minutes. After staining, cells were washed once with FACS buffer. Stained samples were collected on a BD LSRII instrument and analyzed using Flow Jo (Tree Star, Inc.). HLA-DR+CD38+ CD8+ T-cell frequency was defined as the percentage of HLA-DR+, CD38+, CD3+, CD8+, CD4− cells within the live CD45+ singlet population. Ki67+Bcl2− CD8+ T-cell frequency was defined as the percentage of Ki67+, Bcl2-, CD3+, CD8+, CD4− cells within the live CD45+ singlet population. Absolute counts of HLA-DR+CD38+ CD8+ T-cells and Ki67+Bcl2− CD8+ T-cells were calculated by multiplying whole blood staining frequencies by the white blood cell count, as determined by complete blood count performed on an additional aliquot of EDTA-treated whole blood. To assess levels of CD20+ cell-associated LCV, whole blood cells or tissue mononuclear cell suspensions were sorted on a BD FACS Aria Fusion instrument. Whole blood was treated with ACK Lysing Solution (Fisher Scientific, Hampton, NH) to lyse red blood cells, incubated at 5 minutes at room temperature, and washed twice with PBS. Tissues were processed into mononuclear cell preparations, as previously described²². ACK-treated whole blood or tissue mononuclear cells were stained for CD45, CD3, CD20, and viability (Live/dead near-IR, Invitrogen) for 30 minutes at room temperature and subsequently washed with PBS prior to sorting. Cells were kept at 4°C until sorting. CD20+ cells were defined and sorted as live, CD45-hi, SSC-lo, CD3−, CD20+ events. For each sorted population, 12,000-200,000 cells (>95% pure) were collected and used for DNA extraction. The following monoclonal antibodies were used in these studies: a) from BD Biosciences, D058-1283 (CD45; APC), SP34-2 (CD3; Alexa700, PacBlu), 2H7 (CD20; APC-H7, FITC), L200 (CD4; PerCP-Cy5.5, Alexa700), RPA-T8 (CD8; PacBlu), SK1 (CD8, TruRed, APC-Cy7), 28.2 (CD28; PE), DX2 (CD95; FITC, PE-Cy7), G46-6 (HLA-DR; FITC), B56 (Ki67, Per-CP-Cy5.5), Bcl-2/100 (Bcl2, PE-CF594), b) from Beckman Coulter, RMO52 (CD14; ECD), and (c) from NHP Reagent Resource, OKT10 (CD38; PE).

Histology

Immunohistochemistry was performed on the intelliPATH FLX® automated slide stainer (Biocare Medical, Pacheco, CA). Deparaffinized slides were pretreated with 3% H₂O₂/diH₂O for 10 minutes, and then antigen was retrieved using microwave heated citrate buffer (pH 6.0). Immunohistochemistry for SV40 was performed using the ABC-Immunoperoxidase procedure as follows. On the intelliPATH FLX®, slides were then incubated with avidin (Biocare Medical), biotin (Biocare Medical), and protein block (Dako; Agilent, Santa Clara, CA) for 10 minutes each, respectively. Anti-SV40 T Antigen (Ab-2) Mouse monoclonal antibody (clone PAb416, Sigma-Aldrich/Millipore Sigma, Burlington, MA) was diluted to a working 1:12,500 dilution with Dako Antibody Diluent with background reducing components (Dako) and incubated for 2 hours at room temperature. Following this, slides were incubated for 30 minutes at room temperature with Biotinylated Horse Anti-Mouse IgG (Vector Laboratories; Maravai LifeSciences, Burlingame, CA). Peroxidase step was performed using Vectastain ABC HRP standard kit (Vector Laboratories). Finally, the reaction was visualized by 3,3′-diaminobenzidine (DAB) and counterstained with hematoxylin. Multiplex immunohistochemistry for EBNA2/CD20 was performed using the multiplex micro-polymer HRP and AP detection systems. After the pre-treatment and antigen retrieval steps described above, slides were incubated with protein block (Dako) for 10 minutes on the intelliPATH FLX®. Mouse Monoclonal Anti-EBNA2 antibody (EBNA2) (clone PE2, Abcam Inc., Cambridge, MA) was diluted to a working 1:1000 dilution with Dako Antibody Diluent with background reducing components (Dako) and applied for 30 minutes at room temperature. Following this, Mach 2 Mouse HRP-polymer (Biocare Medical) was applied for 30 minutes. EBNA2 was then visualized by Betazoid DAB (Biocare Medical) for 5 minutes. Next, slides were eluted using a denaturing solution (diluted 1 part A with 2 parts B) (Biocare Medical) for 5 minutes. Monoclonal Mouse Anti-Human CD20cy (clone L26, Dako) was diluted to a working 1:125 dilution with Dako Antibody Diluent with background reducing components (Dako) and incubated for 2 hours at room temperature. Slides were then treated with Mach 2 Mouse AP-polymer (Biocare Medical) for 30 minutes. CD20 was visualized via intelliPATH™ Warp Red™ for 7 minutes (Biocare Medical) and counterstained with hematoxylin.

Yellow fever 17D vaccination and neutralizing antibody detection

Macaques were vaccinated with 1×10⁵ plaque-forming units of yellow fever strain 17D subcutaneously. The preparation of yellow fever strain 17D (substrain 17DD) used in this study was obtained from David Watkins (University of Miami) and used previously in NHP studies^23,24. To detect neutralizing antibodies in serum of macaques, a YFV foci reduction neutralization assay was performed. Briefly, YFV-17D at 1.2×10³ FFU/ml was pre-incubated with serial dilutions of heat inactivated macaque serum at 37°C for 2 hours in Dulbecco modified Eagle medium with 5% fetal bovine serum (FBS), penicillin, and streptomycin. Virus-serum mixtures containing 30-40 FFU were added in to individual wells of 96-well tissue culture plates containing Vero cell monolayers at ∼90% confluence (prepared by seeding 1.5×10⁴ cells in 100 μl/well one day prior to assay setup) and incubated at 37°C for 1 hour. The wells were overlaid with 200 μl of 1% methylcellulose in DMEM containing 2% FBS. After 3 days of incubation at 37°C, the methylcellulose overlay was removed by gently pipetting and the wells were washed two times with 200 μl of PBS. The wells were fixed by incubating with 200 μl of 1% formaldehyde in PBS for 10 minutes at room temperature followed by washing with 200 μl of PBS. The cells were permeabilized by washing with 200 μl of Permwash consisting of 0.1% saponin, 0.1% FBS and 0.1% sodium azide in PBS. The cells were stained intracellularly for 2 hours with 50 μl/well of biotinylated YFV-specific mouse monoclonal antibody (clone 3A8.B6) prepared in Permwash. After washing with 200 μl of Permwash twice, the wells were incubated for 1–2 hours with horseradish peroxidase (HRP)-conjugated streptavidin (Invitrogen) at 0.31 μg/ml. The wells were washed with 200 μl of Perm wash buffer two times and the foci were visualized by adding 50 μl of ImmPACT VIP Peroxidase Substrate (Vector Laboratories) according to manufacturer’s instructions. The plates were dried and foci were counted by visual inspection under a stereomicroscope after drying the plates for 2-4 hrs. The 50% focus reduction neutralization titer (NT₅₀) was calculated using the reciprocal of the serum dilution that reduced the number of foci of infected cells by at least 50%.

RNA-sequencing and gene-wise differential expression analyses

Per sample, RNA was extracted from 1 million PBMCs using the Qiagen RNAeasy kit (Qiagen, Venlo, Netherlands) according to the manufacturer’s protocol. An Agilent Bioanalyzer 2100 was used to evaluate RNA quality and RIN score (Agilent, Santa Clara, CA). RNA-seq library construction was performed by MedGenome (MedGenome, Inc., Foster City, CA), using the TruSeq v2 mRNA kit (Illumina, San Diego, CA), followed by sequencing on an Illumina HiSeq (Illumina, San Diego, CA). An average of 52,133,947 single end 100bp reads were obtained per sample (range: 39,990,202 to 60,708,782). The files were imported into ONPRC’s DISCVR-Seq²⁵, LabKey Server²⁶ based system, which managed raw data and most analyses. Trimmomatic²⁷ was used to remove any remaining Illumina adapters and trim the 3’ read ends. Reads were aligned to the rhesus macaque reference genome (rheMac8, NCBI assembly GCF_000772875.2) using the STAR aligner in two-pass mode with default parameters²⁸. Ensembl gene annotations were used (release 87). The per-sample gene count tables generated by STAR were merged to form a single gene count matrix. We performed differential expression (DE) analyses using three experimental contrasts: A) Controls post-vaccination (day 3 and 14 post-HSCT) vs. controls at day 0, B) HSCT macaques post-infection vs. controls at day 0 and C) HSCT macaques post-infection vs. HSCT macaques at day 0. These analyses were performed directly with the raw gene counts using R version 3.5.1^29,30. We performed DE analysis using two commonly used methods Limma³¹ and DESeq2³², and combined the significant genes from each. These methods rely on different mathematical theories and assumptions, and we expect each method to capture different projections of the true biology. The Limma package fits multiple gene-wise linear models using weighted least squares, while DESeq2 fits multiple gene-wise negative binomial generalized linear models. This combined approach was done for two main reasons: first, in this pilot study a significance threshold was not estimated a priori and second, there was a small number of animals in each group. For each DE algorithm, we selected a reasonable p-value (<0.001) threshold heuristically to achieve none-zero gene sets and then combined across all of the contrasts. The union of the two methods ensures the capture of the majority of possible DE genes. To remove false-positive genes, we derived the final set by intersecting our day 3 and day 14 genes which yielded the 72-gene signature. External gene names as well as descriptions pertaining to these genes were identified using the useEnsembl() function of biomaRT package³³.

RESULTS

Polyomavirus is commonly shed by HSCT recipient macaques post-transplant and can be associated with hemorrhagic cystitis and nephritis

To assess the frequency of polyomavirus shedding in HSCT recipient macaques and identify macaques at risk of polyomavirus-associated complications post-HSCT, we monitored urine and plasma of HSCT recipient macaques for polyomavirus large T antigen DNA by PCR and quantitative PCR using primers previously reported to amplify cynomolgus polyomavirus (CPV) DNA in immunosuppressed cynomolgus macaques²¹. Overall, polyomavirus shedding was detected in 67% of analyzed recipients post-HSCT (4/6 macaques, Figure 1A, Table S1), in line with previously reported frequencies of BKV shedding in HSCT patients^11,34-38. Cases ranged in severity from asymptomatic viruria only to viremia accompanied by hemorrhagic cystitis and nephritis (Figure 1A, 2).

Figure 1. Frequency of viral reactivations in HSCT recipient Mauritian cynomolgus macaques.

Pie charts summarize findings among HSCT recipient macaques analyzed for A, Polyomavirus, B, Lymphocryptovirus, and C, Cytomegalovirus reactivation. Numbers on each slice correspond to the number of macaques belonging to each group as described in the corresponding legend. Numbers in the middle of each chart represent the number of recipients analyzed. Viremia and viruria were defined as >60 copies/ml plasma or urine, the limit of detection for quantitative PCR assays. Polyomavirus symptoms include hematuria, hemorrhagic cystitis, and/or kidney dysfunction, defined as a serum blood urea nitrogen and/or creatinine levels exceeding the normal range of 10 to 21 and 0.6 to 1.3 milligrams per deciliter, respectively. Post-transplant lymphoproliferative disorder (PTLD) was diagnosed by tissue histology.

Figure 2. Polyomavirus reactivation varies among HSCT recipient macaques and can result in hemorrhagic cystitis and nephropathy.

A, Weight changes in four HSCT recipient macaques with detectable polyomavirus. B, Longitudinal blood urea nitrogen (BUN) and creatinine (CREA) in serum of HSCT recipient macaques shown in part A. Gray boxes indicated the normal ranges. C, SV40 T antigen immunohistochemistry of 33460 kidney at time of euthanasia. Main (10X magnification): Multifocally, nuclei within infected epithelial cells are immunoreactive to SV40 T antigen. Inset (40X magnification): Infected renal tubular cells exhibit nuclear positivity. D, Amino acid sequence alignment of large T antigens from human polyomaviruses (BK, JC), previously identified macaque polyomaviruses (SV40, CPV), and polyomaviruses from HSCT recipient macaques shown in part A. Sequences shown correspond to amino acids 89 to 131 of SV40 large T antigen. Genbank accession numbers for SV40, CPV, BK virus Dunlop strain, and JC virus are provided in Materials & Methods section. Dashes indicate identical amino acids; asterisks indicate deletions. E, Longitudinal Polyomavirus viral loads in plasma (colored circles) and urine (black squares) of HSCT recipient macaques shown in part A, measured by quantitative PCR. Undetectable viral loads are graphed as 60 DNA copies/ml, the limit of detection for the assay, indicated by horizontal dotted lines. Based on sequence variation among detected Polyomavirus amplicons (see part E), a distinct probe sequence was required for quantitative PCR of 33460, described in Materials & Methods section. Arrows indicate intravenous cidofovir for treatment of CMV reactivation. Gray trace on 33460 and 33459 graphs indicates daily doses of oral prednisone (pred.). Labeled black arrows indicate treatment with depleting antibodies targeted CD20 and CD8α. Labeled brown arrows indicate timepoint 33459 and 36484 were paired. Green cross indicates timepoint of 33460 euthanasia.

In the first case, HSCT recipient macaque 33460 presented with gross hematuria, anorexia, and sustained weight loss around 30 days post-HSCT (Figure 2A). Hematuria was confirmed microscopically and by urine dipstick, and serum chemistry showed elevated levels of blood urea nitrogen (BUN) and creatinine (CREA) indicative of renal dysfunction (Figure 2B). Despite multiple rounds of intravenous fluid diuresis, symptoms persisted, and the macaque became lethargic and was euthanized due to weight loss (>20%, Figure 2A) and persistent cystitis. Gross and histological analysis of necropsy tissues demonstrated active tubulointerstitial nephritis and hemorrhagic cystitis (Figure S1). In addition, intranuclear inclusions indicative of viral infection were observed in renal tubule cells of both kidneys. While the macaque had experienced CyCMV reactivation, CyCMV plasma viremia was minimal and resolved with valganciclovir treatment by time of euthanasia, and tissue sections did not stain positive for CMV antigens via immunohistochemistry (data not shown). Instead, renal epithelial cells in 33460 kidney sections stained positive for SV40 T antigen by immunohistochemistry (Figure 2C), suggesting ongoing polyomavirus replication. In contrast, kidney sections of a non-transplanted control cynomolgus macaque did not show any intranuclear inclusions or SV40 T antigen reactivity (Figure S1). We detected polyomavirus large T antigen DNA in 33460’s plasma post-HSCT, but not in plasma from prior to transplant (Figure S2). These results were confirmed by quantitative PCR, which detected moderate levels of polyomavirus in 33460 plasma at 37 and 45 days post-HSCT, but not in pre-HSCT plasma (Figure 2E).

Polyomavirus shedding was detected in three additional macaques post-transplant: 33459, 36478, and 36484 (Figure 1A, 2E, Figure S2). In each case, no polyomavirus viremia was detected prior to HSCT. Macaque 33459 presented with mild clinical symptoms, including weight loss and low-grade, intermittent hematuria between 14 and 50 days post-HSCT. Two rounds of intravenous fluid diuresis were administered as supportive care around 30 days post-HSCT. Unlike the case of macaque 33460, weight loss and hematuria in macaque 33459 resolved, and no other clinical symptoms of polyomavirus reactivation were observed throughout the monitoring period (Figure 2A, B). Polyomavirus reactivation in HSCT recipient macaques 36478 and 36484 was not accompanied by weight loss or other clinical symptoms of polyomavirus reactivation (Figure 2A, B). With the exception of 33460, all HSCT recipient macaques eventually controlled polyomavirus replication to undetectable levels (<60 copies per milliliter plasma and urine, Table S1), suggesting effective antiviral immunity was eventually restored.

In the process of designing probes for polyomavirus quantitative PCR assays, we sequenced large T antigen PCR amplicons from macaques 33460, 33459, 36478 and 36484. Interestingly, sequencing revealed differences among the polyomavirus large T antigens detected in HSCT recipient cynomolgus macaques as well as differences between these sequences and the previously described CPV large T antigen (Figure 2D). While large T antigen sequences in 33459 and 36484 were 100% identical at the nucleotide level, this sequence differed from those found in macaques 36478 and 33460. In macaques 33459/36484 and 36478, the amplified large T antigen regions were very similar to each other (95% nucleotide identity, 98% amino acid identity), and were most similar to BKV Dunlop strain (84% and 87% amino acid identity, respectively). However, this region in macaque 33460 was divergent from macaques 33459/36484 and 36478 (70% and 74% nucleotide identity, and 69% and 71% amino acid identity, respectively). In addition, 33460’s sequence matched more closely with macaque polyomaviruses SV40 and CPV (93% and 90% amino acid identity, respectively). Thus, polyomavirus sequences vary among individual cynomolgus macaques, a factor which must be considered in viral nucleic acid detection assays.

Development of post-transplant lymphoproliferative disorder is associated with cynomolgus lymphocryptovirus reactivation

In order to examine the incidence of LCV reactivation among HSCT recipient cynomolgus macaques in this study, we developed a quantitative PCR assay to detect cynomolgus lymphocryptovirus (CyLCV) by adapting a previously reported quantitative PCR assay for the rhesus lymphocryptovirus EBER-1 coding region^39,40. We detected cynomolgus LCV (CyLCV) plasma viremia in 67% (6/9) of studied HSCT recipient cynomolgus macaques post-HSCT (Figure 1B, Table S1), which is in line with EBV reactivation rates reported in T cell-depleted HSCT patients⁴¹.

Due to the ability of lymphocryptoviruses to latently infect and transform B-cells, immunosuppressed transplant patients are at high-risk for developing EBV-associated post-transplant lymphoproliferative disorder (PTLD), a spectrum of disorders ranging from B-cell hyperplasia to monoclonal B-cell lymphoma⁴². In this study, 50% (3 of 6) of cynomolgus macaques positive for CyLCV plasma viremia developed PLTD, as diagnosed by tissue histology (Figure 1B, 3, Table S1). A representative case of LCV-PTLD is shown in Figure 3. Approximately 45 days post-HSCT, recipient macaque 35132 presented with anorexia, axillary and inguinal lymphoadenopathy, and splenomegaly. Abdominal ultrasound at 53 days post-HSCT showed enlarged, nodular spleen and enlarged, irregular mesenteric lymph nodes. In addition, complete blood count revealed thrombocytopenia, anemia, and 11% lymphoblastic cells. The macaque was euthanized due to suspicion of PTLD. Lymphoma consistent with monomorphic PTLD was found in the spleen, liver, kidney, adrenal glands, pancreaticosplenic, and several periaortic lymph nodes (kidney shown in Figure 3A). Some tissues contained both overtly neoplastic lymphocytes forming masses and infiltrates more consistent with polymorphic PTLD, with mixed populations of lymphoid cells. By immunohistochemistry, the majority of infiltrating neoplastic cells in kidney stained positive for B-cell marker CD20 and lymphocryptovirus Epstein Barr Nuclear Antigen-2 (EBNA-2) (Figure 3B), demonstrating lymphocryptovirus-associated B-cell lineage lymphoproliferative disease and consistent with the vast majority of PTLD cases in transplant patients and previously described cases of PTLD in transplanted cynomolgus macaques^43-45. In contrast, kidney sections from a non-transplanted control cynomolgus macaque did not show infiltrating neoplastic cells and did not stain positive for CD20 or EBNA-2 (Figure S3).

Figure 3. Post-transplant lymphoproliferative disorder is associated with lymphocryptovirus reactivation.

A, Hematoxylin and eosin stain of HSCT recipient macaque 35132 kidney. Left (10X magnification): Sheets of neoplastic lymphocytes infiltrate and expand the renal interstitium. Right (63X magnification): Infiltrating neoplastic lymphocytes have scant amounts of basophilic cytoplasm and round to oval nuclei with clumped chromatin and large, often multiple, eosinophilic nuclei. There is mild anisokaryosis. B, Immunohistochemistry of HSCT recipient macaque 35132 kidney for CD20 (red) and lymphocryptovirus EBNA-2 (brown) at 10X (left) and 40X (right) magnification. The majority of infiltrating neoplastic lymphocytes are both immunoreactive for CD20 (red membranous and cytoplasmic staining) and EBNA-2 (brown nuclear staining), indicating B-cell lineage lymphocytes infected with LCV. C, Longitudinal CyLCV plasma viral loads in 35132 (top) compared to 33459 (bottom), an HSCT recipient macaque who did not develop PTLD, as measured by quantitative PCR. Undetectable plasma viral loads are graphed as 60 DNA copies/ml, the limit of detection (LOD) for the assay, indicated by horizontal dotted lines. Yellow box indicates period where clinical symptoms consistent with PTLD were observed, including lymphadenopathy and splenomegaly. Horizontal solid lines indicate period of oral prednisone treatment. Dark blue cross indicates timepoint of 35132 euthanasia. D, CyLCV CD20+ cell-associated viral loads in 35132, shown longitudinally in blood and lymph node (top) and in tissues at time of euthanasia (bottom). LOD for cell-associated viral loads varied among assays due to variable cell number input. Of note, blood CD20+ cell-associated LCV was undetectable 5 days pre-HSCT and 25 days post-HSCT (LODs 7,324 and 4,048 DNA copies per 1 × 10⁶ CD20+ cells, respectively). E, Longitudinal blood CD20+ cell counts in HSCT recipients shown in part C. F, CyLCV viral loads in supernatants from cultures of single cell suspensions isolated from necropsy tissues.

Quantitative PCR showed detectable CyLCV plasma viremia in macaque 35132 at 42 days post-HSCT, which continued to increase until time of euthanasia (Figure 3C). A low level of CyLCV DNA was found in pre-HSCT plasma, raising the possibility of ongoing lytic viral replication in this macaque prior to transplantation. In contrast to macaque 35132, HSCT recipient macaque 33459 did not exhibit any clinical signs of PTLD and had only one small blip of detectable CyLCV viremia (<200 copies per milliliter plasma) throughout the monitoring period. Next, we measured cell-associated CyLCV in isolated CD20+ B-cells from 35132’s blood and tissues. Levels of cell-associated CyLCV in blood and lymph node CD20+ cells increased after transplant and until time of euthanasia, and high levels of CD20+ cell-associated CyLCV were detected in blood, lymph nodes, bone marrow, and spleen at time of euthanasia (Figure 3D). Consistent with B-cell-origin lymphoproliferative disease, longitudinal absolute CD20+ cell counts in blood of macaque 35132 revealed a large increase in blood B-cells at time of euthanasia, in contrast to the slow B-cell reconstitution that typically occurs post-HSCT and exemplified by macaque 33459 (Figure 3E). To further demonstrate the existence of LCV-transformed cells in macaque 35132, we cultured single cell suspensions from necropsy tissues in R10 medium without any additional cytokines or growth factors. After over two months of culture with passage and media replacement once or twice weekly, all cultures resembled B-lymphoblastoid cell lines and clarified supernatants were positive for CyLCV DNA (Figure 3F). Cultures continued to grow and secrete CyLCV indefinitely, over 20 months. Together, these data demonstrate the development of CyLCV-associated PTLD in a cynomolgus macaque undergoing HSCT.

Cytomegalovirus reactivates early post-HSCT, but is subsequently controlled

To assess cynomolgus CMV (CyCMV) reactivation in HSCT recipient cynomolgus macaques, we developed a quantitative PCR for CyCMV UL121 DNA based on the sequence of CyCMV isolates from Mauritian cynomolgus macaques⁴⁶. CyCMV plasma viremia was detected in 100% of analyzed recipient macaques within 30 days post-transplant (Figure 1C, 4A, Table S1). Representative macaques are shown in Figure 4A. Plasma viral load in 2 of 11 macaques remained below 1,000 copies per milliliter (copies/ml) and dropped below detectable levels within 30 days of transplant (Figure 4A, 36478). As standard preemptive therapy, treatment with HCMV antiviral valganciclovir was initiated when CyCMV plasma viral load exceeded 1,000 copies/ml (9/11 macaques) and continued until viremia was no longer detectable. Valganciclovir administration alone led to sustained decreases in CyCMV plasma viral load in 7 of 9 treated macaques. CyCMV plasma viral loads fell below detectable levels within 50 days of treatment in six treated macaques (Figure 4A, 36484); one macaque was euthanized prior to complete CyCMV control due development of LCV-PTLD (Figure 4A, 35132). In 2 of 9 treated macaques, valganciclovir alone was unable to control CyCMV replication -- CyCMV plasma viral loads increased above 15,000 copies/ml despite twice daily valganciclovir administration (Figure 4A, 33459). This was associated with lower absolute counts of blood CD8+ T cells early post-HSCT in macaque 33459 compared to macaques that did not require cidofovir treatment to control viremia (Figure S5). These macaques received weekly infusions of intravenous cidofovir until viremia remained below 15,000 copies/ml for one week. Combination valganciclovir and cidofovir treatment effectively controlled CyCMV plasma viremia to undetectable levels in both macaques. After treatment discontinuation, all recipients controlled CyCMV plasma viremia to low levels (<500 copies/ml) and eventually achieved complete control of viremia within 30 days of treatment interruption and for the remainder of the monitoring period. No symptoms of CMV disease were noted in any HSCT recipient macaque throughout the monitoring period.

Figure 4. Cytomegalovirus consistently reactivates early post-transplant, but is subsequently controlled.

A, Longitudinal cynomolgus Cytomegalovirus (CyCMV) plasma viral loads in four representative HSCT recipient macaques, measured by quantitative PCR (limit of detection 60 DNA copies/ml). Gray boxes indicate periods of oral valganciclovir treatment. Arrows indicate intravenous cidofovir treatment. Colored asterisks indicate timepoints at which post-HSCT PBMC was tested for CMV-specific T cell responses, shown in part B. Dark blue cross indicates timepoint of 35132 euthanasia. SID = once daily, BID = twice daily. B, CMV-specific T cell responses measured by interferon-gamma ELISPOT of peripheral blood mononuclear cells (PBMC) stimulated with CMV lysate or pools of 15-mer peptides spanning IE-1, IE-2, and pp65 (11 amino acid overlap). Graphs show results for HSCT recipient macaques shown part A, pre-HSCT (black) and post-HSCT (colored), and their HSCT donors (gray). SFCs = spot-forming units.

The observed control of CyCMV viremia after reactivation suggested successful reconstitution of effective anti-CMV immunity post-HSCT. Virus-specific T-cells are critical for controlling CMV replication and disease in humans and macaques^47-52. Thus, we used interferon-gamma ELISPOT to screen for T cell responses against CMV lysate and pools of 15-mer peptides spanning CMV antigens IE-1, IE-2, and pp65 (Figure 4B). Indeed, we readily detected CMV-specific T cell responses in PBMC of recipient macaques post-HSCT as well as PBMC of recipients pre-HSCT and donor macaques. In some cases, we observed CMV peptide-specific T cell responses in recipients post-HSCT that were not detectable prior to HSCT or in the HSCT donor, raising the possibility that new CMV-specific T cell responses were primed after transplant.

HSCT recipient macaques respond to live attenuated yellow fever virus 17D vaccination indistinguishably from non-transplanted controls

The observed reactivation and subsequent control of polyomavirus, CyLCV, and CyCMV in HSCT recipient cynomolgus macaques suggested post-transplant reconstitution of effective antiviral immunity. However, as not all macaques possessed a fully donor-derived immune system at the timepoints studied above (data not shown), we could not rule out the possibility that post-transplant antiviral immunity had expanded from surviving recipient-derived immune responses. To directly test the ability of HSCT recipient macaque immune systems to prime new immune responses, we used live attenuated yellow fever vaccine 17D (YFV 17D). In healthy humans and rhesus macaques, YFV 17D vaccination results in minimal, short-term viremia and induces vaccine-specific neutralizing antibody and T cell responses^23,24,53-57. Here, we compared the response to YFV 17D vaccination in six Mauritian cynomolgus macaques: two HSCT recipient macaques (862 and 806 days post-HSCT at time of vaccination) and four non-transplanted control macaques. Of note, the immune systems of both HSCT recipient macaques were 100% donor-derived prior to vaccination¹⁷. First, we assessed the ability of the macaques to control YFV 17D replication. Quantitative reverse transcription PCR detected low levels of YFV 17D RNA in plasma of 1 of 2 HSCT macaques and 3 of 4 control macaques within 7 days of vaccination (Figure 5A). By 14 days post-vaccination, YFV 17D RNA was no longer detectable in the plasma of any macaque, suggesting all macaques had cleared YFV 17D virus. Indeed, all macaques mounted a YFV 17D-targeted neutralizing antibody response by 14 days post-vaccination, and we found similar titers of neutralizing antibodies in the serum of all six macaques at 14 and 28 days post-vaccination (Figure 5B). In addition, we did not observe any differences in CD8+ T cell activation in blood between HSCT and control groups, as measured by HLA-DR+CD38+ and Ki67+Bcl2− CD8+ T-cells (Figure 5C).

Figure 5. Long-term engrafted HSCT recipient macaques respond to live-attenuated yellow fever 17D vaccination similarly to untransplanted macaques.

A, Longitudinal yellow fever (YFV) plasma viral loads in two long-term engrafted HSCT recipient macaques (blue) compared to four control untransplanted Mauritian cynomolgus macaques (red) after live-attenuated yellow fever 17D vaccination. Undetectable plasma viral loads graphed as 300 RNA copies/ml, the limit of detection (LOD) for the assay, indicated by horizontal dotted line. B, Longitudinal serum YFV 17D neutralizing antibody titers. NT50 = highest fold serum dilution where YFV 17D plaque number was reduced by at least 50%. The lowest dilution of serum tested was 10-fold, indicated by the horizontal dotted line. C, Longitudinal counts of activated CD8+ T-cells in blood post-vaccination, as measured by HLA-DR and CD38 (left) and Ki67 and Bcl2 (right). No statistically significant difference in plasma viral loads or CD8+ T cell activation were observed between the HSCT and control groups, analyzed by comparing areas under the curve (AUC) using the Mann-Whitney test (p-value displayed on graphs). No statistically significant difference in neutralizing antibody titers were observed between the HSCT and control groups at any timepoint, analyzed using repeated measures ANOVA (overall group factor p-value displayed on graph, Sidak-adjusted p-values for timepoints 0, 14, and 28 are >0.99, 0.76, and 0.37, respectively).

Previous studies have reported transcriptomic changes following vaccination with YFV 17D in humans and rhesus macaques^55,58. To determine whether a gene expression signature was present in our animals following vaccination, and to determine if this signature differed between HSCT and control macaques, we performed bulk RNA-seq on peripheral blood mononuclear cells (PBMC) from vaccinated macaques at days 0, 3 and 14 post-vaccination. These data revealed a total of 655 differentially expressed gene (DEG) candidates across the experimental contrasts in this study (Table S2); the intersect of these DE gene sets highlights a signature of 72 differentially expressed genes (DEGs) (Figure 6, Table S3) with two distinct patterns across the subjects. The first pattern contained the majority of DEGs (59 of 72) and separated the pre- and post-vaccination samples (Figure 6, clusters 1 and 3). For all but one of these genes, expression was consistently increased (n=30) or decreased (n=28) after vaccination in both HSCT and control groups (Figure 6, clusters 1 and 3). Expression of one gene, 6-phosphofructo-2-kinase (PFKFB1), an activator of the glycolysis pathway, was elevated in the HSCT group compared to the control group prior to vaccination, but decreased to similar levels to controls after vaccination. PFKFB1 expression in the control group did not change after vaccination. The second, minor pattern contained genes (13 of 72) that were significantly different between HSCT and controls prior to vaccination (Figure 6, clusters 2 and 4). Gene expression was increased (n=6) or decreased (n=7) in the HSCT macaques relative to the controls, but did not consistently change after vaccination in either group. Thus, while HSCT recipient macaques have some baseline gene expression differences from non-transplanted controls, changes in gene expression induced by YFV 17D vaccination in control macaques also occurred in HSCT recipient macaques, suggesting the vaccine response in HSCT recipient macaques was not impaired. Together, these data demonstrate priming of effective donor-derived antiviral immunity in HSCT recipient macaques, who by all measures responded to YFV 17D vaccination indistinguishably from non-transplanted controls.

Figure 6: HSCT and control macaques experience similar changes in gene expression after YFV 17D vaccination.

A heatmap of differentially expressed genes (rows) and samples (columns) demonstrating the identified 72-gene transcriptomic signature on the rlog-transformed expression matrix; pair-wise hierarchical clustering with Ward.D2 using Euclidean distances. There are 2 experimental groups (HSCT vs. controls), collected at 3 timepoints: day 0, 3, and 14, which are labeled according to the colors in the legend. The genes are obtained by identifying the overlap of 3 contrasts pertaining to the experimental design: A) Control macaques post-vaccination (Days 3 and 14) versus pre-vaccination (Day 0), B) HSCT macaques post-vaccination versus control macaques pre-vaccination (day 0), and C) HSCT macaques post-vaccination (Days 3 and 14) versus pre-vaccination (Day 0). Genes grouped into four clusters based on their expression pattern across groups.

DISCUSSION

In this study of viral reactivations in Mauritian cynomolgus macaques undergoing allogeneic HSCT, we focused on macaque analogues of BKV, EBV, and HCMV, which are among the most common viral etiologies of infectious disease complications in allogeneic HSCT patients⁵⁹. These viral pathogens are also significant obstacles in solid organ xenotransplantation^18,60. Multiple polyomaviruses have been reported to ubiquitously infect macaques, including cynomolgus polyomavirus (CPV) and simian virus 40 (SV40) ^21,61. Lymphocryptoviruses (LCVs) analogous to EBV ubiquitously infect many species of non-human primates, with virtually all macaques harboring LCV by 2 years of age^62,63. Species-specific CMVs analogous to HCMV have been isolated from many nonhuman primate species, and virtually all individuals are infected within the first year of life^46,64-66. We observed frequent reactivations of polyomavirus, CyLCV, and CyCMV in transplanted cynomolgus macaques, which occasionally resulted in clinical manifestations identical to those observed in human patients, including polyomavirus-associated hemorrhagic cystitis and nephropathy and lymphocryptovirus-associated PTLD. To our knowledge, this is the first report of polyomavirus and lymphocryptovirus reactivation and disease in a macaque model of HSCT.

Similar to allogeneic HSCT in the clinic⁵⁹, viral reactivation and disease in this study typically occurred early post-transplant, and in most cases, was eventually brought under control, presumably due to reconstitution of effective antiviral immunity. In support of this, we found CMV-specific T cell immunity after resolution of post-transplant CMV viremia in all studied allogeneic HSCT recipient cynomolgus macaques. Associations of antiviral T-cells with viral control post-transplant have been reported for CMV, BKV, EBV, and adenovirus^67-77, and numerous studies have demonstrated safety and efficacy of allogeneic virus-specific T cell infusion in treating viral reactivations with minimal or no exacerbation of GVHD^78-88. Due to the simplified MHC genetics of Mauritian cynomolgus macaques^89-91, this model provides an opportunity to conduct detailed investigations of T cell-based therapies for viral complications of allogeneic HSCT.

Finally, in further support of reconstitution of functional immunity in our macaque model of allogeneic HSCT, we show safe and effective vaccination of HSCT recipient macaques with live attenuated virus vaccine YFV 17D, in line with previous reports of YFV 17D vaccination in allogeneic HSCT patients^92-94. Further, whole transcriptome RNA-seq of YFV 17D-vaccinated HSCT recipient and non-transplanted control macaques showed identical changes in gene expression after immunization. Previous studies of gene expression changes after YFV 17D vaccination in rhesus macaques were conducted with microarray technology and reported significant changes in 39 genes following YFV-17D vaccination⁵⁸, and our DEG set overlapped with 7 of these: IFIT3, MARKS, HERC6, HERC5, CMPK2, RSAD2, EIF2AK2, LILRA2. This limited overlap is likely due to the very distinct technologies of microarray versus whole transcriptome RNA-seq where discrepancies are not uncommon⁹⁵. Regardless, the detection of consistent transcriptional changes between groups of macaques in this study suggests that the response to YFV 17D vaccination in HSCT recipient macaques is not impaired.

Despite vast improvements in managing infectious disease complications of allogeneic HSCT, many challenges still exist. Between 2015 and 2016, approximately 11 to 24% of deaths after allogeneic HSCT were due to infection^1,2. While implementation of viral surveillance methods has improved early diagnosis and intervention for virus-associated complications, most measurements have poor predictive value for disease. Namely, patients with high viral loads do not always progress to disease, and disease can occasionally manifest without high viral loads in blood^37,41,96-102. In addition, there are major challenges in treating viral complications, including lack of effective treatments, drug toxicity, and simultaneous management of GVHD, which is treated with intensified immunosuppression that can exacerbate viral reactivation and disease^103-105. Similar challenges exist in solid organ and xenotransplantation, where intensified immunosuppression is often required to prevent rejection^18,60. Overall, the findings in this study demonstrate that this Mauritian cynomolgus macaque model recapitulates viral disease and immunity observed in human transplant patients. Therefore, this macaque model of allogeneic HSCT can be used to explore novel strategies for early diagnosis, prevention, and treatment of viral complications after allogeneic and xenogeneic transplantation.

ABBREVIATIONS

HSCT

hematopoietic stem cell transplantation

MHC

major histocompatibility complex

LCV

lymphocryptovirus

CyLCV

cynomolgus lymphocryptovirus

CMV

cytomegalovirus

CyCMV

cynomolgus cytomegalovirus

HCMV

human cytomegalovirus

BKV

BK virus

EBV

Epstein-Barr virus

PTLD

post-transplant lymphoproliferative disorder

SIV

simian immunodeficiency virus

YFV 17D

yellow fever virus strain 17D vaccine

GVHD

graft-versus-host disease

PBMC

peripheral blood mononuclear cells

BKV-HC

BK virus-associated hemorrhagic cystitis

BKVN

BK virus-associated nephropathy

PCR

polymerase chain reaction

CPV

cynomolgus polyomavirus

SV40

simian virus 40

HLA

human leukocyte antigen

ATG

antithymocyte globulin

EBNA-2

Epstein-Barr nuclear antigen 2

DEG

differently expressed gene

SID

once daily

BID

twice daily

IV

intravenous

PO

per os

IM

intramuscular

SC

subcutaneous