I have read the journal's policy and have the following conflicts: Dr. Greenberg has received honoraria from the MSAA. He has received consulting fees from Sanofi Aventis, The Greater Good Foundation, Biogen Idec, Elan, Applied Clinical Intelligence and DioGenix. He holds equity in DioGenix, Inc. He receives grant support from the Guthy Jackson Charitable Foundation, Amplimmune and Accelerated Cure Project. Dr. Frohman has received speaker and consulting fees from Biogen Idec, TEVA, Novartis, and Acorda; and consulting fees from Abbott and Genzyme. This does not alter our adherence to all PLoS Pathogens policies on sharing data and materials.
Conceived and designed the experiments: BG DCD EF EOM MRP XX. Performed the experiments: AR CK CL CR DH GR HLE JA JCL MCGM MRP PNJ SI TA. Analyzed the data: CK CL CR GR JA JCL MCGM MRP NLM TA. Wrote the paper: DCD EOM MRP. Gathered patient data and samples: BG CK CL CR EOM EF GR JA NM TA.
Progressive multifocal leukoencephalopathy (PML) induced by JC virus (JCV) is a risk for natalizumab-treated multiple sclerosis (MS) patients. Here we characterize the JCV-specific T cell responses in healthy donors and natalizumab-treated MS patients to reveal functional differences that may account for the development of natalizumab-associated PML. CD4 and CD8 T cell responses specific for all JCV proteins were readily identified in MS patients and healthy volunteers. The magnitude and quality of responses to JCV and cytomegalovirus (CMV) did not change from baseline through several months of natalizumab therapy. However, the frequency of T cells producing IL-10 upon mitogenic stimulation transiently increased after the first dose. In addition, MS patients with natalizumab-associated PML were distinguished from all other subjects in that they either had no detectable JCV-specific T cell response or had JCV-specific CD4 T cell responses uniquely dominated by IL-10 production. Additionally, IL-10 levels were higher in the CSF of individuals with recently diagnosed PML. Thus, natalizumab-treated MS patients with PML have absent or aberrant JCV-specific T cell responses compared with non-PML patients, and changes in T cell-mediated control of JCV replication may be a risk factor for developing PML. Our data suggest further approaches to improved monitoring, treatment and prevention of PML in natalizumab-treated patients.
Progressive multifocal leukoencephalopathy (PML) is a complication of treatment with natalizumab in patients with multiple sclerosis (MS) and Crohn's disease. PML results from a failure of the immune system to control replication of JC virus (JCV) in the brain. We studied the T cell responses of 8 patients with MS who were starting treatment with natalizumab, 10 healthy volunteers, and 4 patients with natalizumab-associated PML. The magnitude and quality of JCV-specific immune responses remained unchanged after starting natalizumab. However, applying the same methods and antigens, we found that immune responses in the individuals who developed PML differed from those in the MS patients and healthy volunteers. In the four patients with PML from whom the laboratory had identified JCV DNA in the cerebrospinal fluid (CSF), two had no measurable T cell response to JCV and two had T cells that produced IL-10, an anti-inflammatory mediator. Furthermore, we studied the CSF of 10 patients with natalizumab-associated PML and 10 patients on natalizumab who had similar symptoms but did not have PML. We found that IL-10 was detectable in the CSF of half of the individuals with PML but none of the control group. These findings shed light on the mechanisms that lead to PML in a subset of patients treated with natalizumab and have implications for therapeutic and preventative measures.
Progressive Multifocal Leukoencephalopathy (PML) is a demyelinating disease of the central nervous system caused by JC virus (JCV)
Previous studies have measured CD4 and CD8 T cell responses to JCV in individuals with PML who had not received natalizumab therapy
Two recent studies yielded disparate results when examining longitudinal T cell responses to JCV in individuals with MS after natalizumab treatment
We characterized the T cell immune response to JCV in eight patients with MS initiating natalizumab and ten healthy volunteers. As T cell responses have not been described to JCV proteins other than VP1, we sought to determine whether either CD4 or CD8 T cells were directed against the other portions of the virus. We stimulated PBMC with 5 pools of peptides covering the entire JCV proteome: large T antigen, small t antigen, VP1, VP2 and agnoprotein. The peptides were 15mers overlapping by 11 amino acids to optimize coverage of both CD4 and CD8 T cell epitopes, and included additional peptides to cover JCV sequence variants identified in the literature (as described in the
Both CD4 and CD8 memory T cell responses were readily measured
PBMC from individuals with MS treated with natalalizumab who did not have PML were stimulated with JCV peptide pools or costimulatory molecules alone (negative control) for 6 hours. Panel A shows memory CD4 T cells from 3 samples; Panel C shows memory CD8 T cells from 3 samples. The fluorescence intensity of IFNγ and TNF are shown on the X and Y-axes, respectively. Panels B and D show baseline pre-treatment responses from all eight longitudinal subjects with MS, with the background-subtracted magnitude of the response to each JCV protein depicted by colored bars. Responses were measured by production of any combination of IFNγ, TNF and IL-2, using Boolean gates and then background subtracting from each Boolean population.
PBMC from individuals with MS treated with natalalizumab who did not have PML were stimulated with JCV peptide pools, CMV pp65 peptide pool or costimulatory molecules alone (negative control) for 6 hours. A response was considered positive if the frequency of memory T cells producing IFNγ, TNF or IL-2 was higher in the peptide-stimulated cells than in those stimulated with costimulatory molecules alone. Response size was calculated by measuring the frequency of cells producing each Boolean combination of cytokines, and subtracting the frequency of these cells in the negative control. Summing the background-subtracted Boolean subsets gave the total frequency of cytokine-producing memory T cells specific for the peptide pool. The total response to JCV (Panel A) was calculated by summing the frequency of cells specific for each of the 5 JCV peptide pools. The functional profile of the response is shown in the pie charts above, with the blue slice representing the proportion of responding cells that produce all 3 cytokines, the red slices representing the proportion of cells that produce a combination of 2 cytokines, and the green slices representing the proportion of cells that produce only 1 cytokine. Panel B shows the frequency of CD4 (left) and CD8 (right) memory T cells responding to the CMV pp65 peptide pool.
To determine the impact of natalizumab after short-term treatment, when risk of PML is low, we compared T cell responses to JCV and cytomegalovirus (CMV) before and after the initiation of natalizumab. The eight individuals with MS were examined at baseline (month 0) before natalizumab therapy, one month after the first infusion of natalizumab (month 1), and at the latest timepoint available (after 3–7 monthly infusions).
The total CD4 and CD8 T cell responses to JCV, calculated as the sum of the memory response to each of the five JCV peptide pools, did not change significantly at different timepoints (
CMV was chosen as a control antigen for JCV because it is a prevalent DNA virus that, like JCV, is neurotropic and establishes latent infection. The CMV pp65-specific CD4 and CD8 memory T cell responses did not change in magnitude or functional profile between the three timepoints (
To gauge the impact of natalizumab on the overall memory T cell response to stimulation we measured cytokine production after stimulation with the mitogen SEB. No difference was observed in the magnitude or profile of IFNγ, TNF and IL-2 produced at the different time points in response to SEB stimulation (data not shown). We also measured IL-10 production because this regulatory cytokine is associated with poor control of persistent viral infections
PBMC from all studied MS patients were stimulated with SEB. Background-subtracted frequency of memory CD4 T cells producing IL-10 is shown.
In order to determine whether MS patients with natalizumab-associated PML had similar T cell responses, we also examined the response to mitogenic stimulation in T cells from four patients whose CSF samples were previously tested in our laboratory for JCV DNA. One of these patients, PML-4, was one of the first cases diagnosed in 2005, who was followed for more than 5 years and never cleared JCV from the brain
Approximately 2.1/1000 MS patients treated with natalizumab develop PML
In two subjects, PML-1 and PML-2, no JCV-specific CD4 or CD8 T cell responses were observed that were significantly above background (
Panel A shows the total response to JCV. For each of 4 subjects with PML, the summed frequency of memory CD4 (left) and CD8 (right) T cells producing IFNγ, TNF, IL-2 or IL-10 (indicated by colors) in response to all JCV peptides is shown. Red bars indicate frequency of cells producing IFNγ, including those that produce any combination of IL-2 and TNF in addition to IFNγ. The responses for subjects PML-1 and PML-2 are not significantly above background in these samples. Panel B shows the JCV-specific IL-10 response in subjects PML-3 and PML-4. Subjects PML-1, 2, 3 and 4 were sampled 2 weeks, 2 months, 4 months and 5 years, respectively, after diagnosis with PML.
In PML-3 and PML-4, the JCV-specific T cell response had a markedly different functional profile than was observed in any of the non-PML or healthy subjects. These subjects, who were diagnosed with PML 4 months and 5 years prior to sampling, respectively, had JCV-specific CD4 T cells that produced IL-10 (
As JCV-specific T cells in the blood of the four individuals with PML were either absent or of unusual functionality, we next examined the cytokine profile within CSF samples from 10 individuals with natalizumab-associated PML, including subjects PML 1–4, at the time of initial diagnosis. In addition, a second CSF sample from a later time point was available for 8 of these subjects. The samples were tested for the presence of 27 cytokines and other markers and were compared to diagnostic CSF samples from 10 individuals who had suspected natalizumab-associated PML which was subsequently ruled out by a negative PCR for JCV. Twelve of the molecules were undetectable in the vast majority of CSF samples, including IL-1β, IL-2, IL-4, IL-12p70, IL-17, Eotaxin, FGF basic, GM-CSF, IFNγ, MIP1α, RANTES, and TNF. An additional 12 molecules, IL-1ra, IL-6, IL-7, IL-8, IL-9, IL-13, G-CSF, IP-10, MCP-1, PDGF-β, MIP1β and VEGF, were measured in the CSF samples but did not vary in PML and non-PML samples. Although the assay was not sensitive enough for accurate quantification of low levels of IL-10, it could readily distinguish whether IL-10 was detectable above background. Thus, we found that 50% of the PML CSF samples had detectable IL-10, while none of the non-PML samples had detectable IL-10 (
Diagnostic CSF samples taken from subjects with natalizumab-associated PML early in disease course were compared to samples from subjects who had suspected natalizumab-associated PML but tested negative for JCV DNA. Panel A shows the proportion of CSF samples with detectable IL-10 in diagnostic samples from subjects in whom natalizumab-associated PML was confirmed or ruled out, and the proportion of CSF samples in each group with IL-5 levels above the lower limit of quantification (1.1 pg/mL).
The factors that lead to the development of PML in individuals treated with natalizumab need to be investigated in more detail, particularly immune responses to the virus. Consequently, we investigated the T cell immune response to the entire JC virus proteome longitudinally in subjects with MS who were initiating therapy with natalizumab and in subjects who had natalizumab-associated PML. The principal findings were: 1) T cell responses were identified against all JC virus proteins and could be measured
Whether JCV-specific T cell responses can be reliably measured
However, in the subjects with PML, JCV-specific CD4 T cell responses were either undetectable or uniquely dominated by IL-10. Importantly, it has been shown that IL-10 is detrimental to the clearance of lymphocytic choriomeningitis virus (LCMV) infection because of its inhibitory effect on virus-specific memory CD4 T cells
Notably, we found that the frequency of memory CD4 T cells that produce IL-10 upon mitogenic stimulation is transiently increased after the first dose of natalizumab. It is tempting, therefore, to speculate that natalizumab may skew the CD4 T cell response toward IL-10 production and away from production of IFNγ, TNF, and IL-2. This suggests a possible mechanism by which natalizumab treatment could lead to PML, as 50% of the subjects with natalizumab-associated PML that we studied produced IL-10 in response to JCV, and the other 50% had no measurable T cell response to the virus. There are a number of mechanisms by which natalizumab treatment could potentially skew the CD4 T cell response toward IL-10 production, including increasing antigen load through mobilization of infected CD34+ cells
The low probability that any of the natalizumab-treated MS study subjects could go on to develop PML is consistent with the lack of JCV-specific IL-10 production in those subjects who did not have PML. Furthermore, in this study we were only able to measure T cell responses in the peripheral blood and were not able to sample the CNS of subjects without PML. A previous study showed that after 12 months of natalizumab treatment, levels of IL-10 mRNA were increased in bulk CSF cells, while remaining unchanged in PBMC
Taken together, our data provide a framework for understanding immune control of JC viremia and the development of PML and suggest avenues of investigation to allow the better monitoring, treatment and prevention of PML in natalizumab-treated people. First, our finding that subjects with PML lacked JCV-specific T cell responses or produced IL-10 in response to stimulation suggests that immune monitoring might identify natalizumab-treated individuals who are at risk of developing PML, by screening subjects prior to treatment or while on treatment. None of the MS patients without PML or healthy subjects included in our study showed an absent or IL-10 producing T cell response similar to that observed in the subjects with PML, and this suggests that individuals with these phenotypes are relatively rare and could be identified by immune monitoring prior to treatment. The potential of such screening of JCV-specific T cell responses to identify a small number of individuals at risk for the development of PML could be complementary to stratification strategies based on antibody levels that are currently being tested to identify approximately 50% of treated individuals who are at increased risk
The study was approved by the IRB of the University of Texas Southwestern Medical Center. Written informed consent was obtained from all study subjects.
Eight individuals were recruited who were initiating natalizumab therapy (Tysabri, Biogen Idec) at the University of Texas Southwestern Medical Center Multiple Sclerosis Clinic. Subjects underwent a washout period of at least two weeks and had blood samples taken immediately before the first dose and each subsequent monthly infusion. Ten healthy blood donors were recruited at the NIH Vaccine Research Clinic. Blood samples from 4 subjects with natalizumab-associated PML were obtained, and CSF samples from these individuals and 6 others, as well as 10 control subjects who had suspected PML but a negative test for JCV were supplied by the NIH Laboratory of Molecular Medicine and Neuroscience clinical testing lab.
Peripheral blood mononuclear cells (PBMC) were isolated from whole blood by ficoll-hypaque density centrigugation (GE Heathcare Life Sciences). PBMC were cryopreserved in freezing media containing 90% fetal bovine serum and 10% DMSO for use in T cell assays.
Frozen PBMC were thawed and washed twice with RPMI 1640 supplemented with 10% heat inactivated fetal calf serum, 100 U/ml penicillin G, 100 U/ml streptomycin sulfate, and 1.7 mm sodium glutamine (R-10) containing 10 U/ml DNase I (Roche Diagnostics). Cells were then rested for two hours before being washed and then plated in 96-well plates in 200 uL final volume of R10 with DNase I. All experiments were done at 5×106 PBMC/ml in the presence of 1 µg/ml each of αCD28 and αCD49d (BD Bioscience), in the absence or presence of peptide antigens or SEB. Cells were stimulated for 6 hours, with 10 µg/ml brefeldin A (BFA) (Sigma Chemical Company) added after 1 hour. For subjects PML-1, PML-2 and PML-3, freshly isolated PBMC were stimulated for 16 hours with BFA added after 1 hour. For subject PML-4 frozen PBMC were stimulated for 6 hours as described in parallel with MS and healthy subjects.
Directly conjugated monoclonal antibodies (mAbs) specific for the molecules listed were obtained from the following: CD3 APC-H7, TNF Cy7PE, IFNγ V450, IL-10 APC, IL-10 PE, IL-2 FITC, CD3 Alexa700, IFNγ FITC, IL-2 APC from BD Biosciences; CD45RO-TRPE, CD27-Cy5PE from Beckman Coulter; CD4-Cy55PE from Caltag, CD8 QD705 from Invitrogen, and IL-2 Alexa700 from BioLegend. The following antibodies were conjugated in our laboratory according to standard protocols (
Stimulated PBMC used for intracellular cytokine staining were washed and pre-stained for 10 minutes with a pre-titered amount of LIVE/DEAD fixable aqua dead cell stain (Molecular Probes). Cells were then surface stained with a mixture of pre-titered amounts of directly conjugated antibodies to CD27, CD45RO, CD57, CD4, CD8, CD19, and CD14 made to a total volume 100 µl with Delbecco's phosphate buffered saline (PBS). Cells were stained for 30 min at 4°C in the dark. Cells were then washed and permeabilized using the cytofix/cytoperm kit (BD Biosciences) according to the manufacturer's instructions. After intracellular staining for CD3, IFNγ, TNF, IL-2 and IL-10 cells were washed one final time and fixed in PBS containing 1% paraformaldehyde and then stored at 4°C. Flow cytometric analysis was done within 24 h of staining.
Cells were analyzed with a modified LSRII (BD Immunocytometry Systems) equipped for the detection of 18 fluorescence parameters. Between 500,000 and 1,000,000 events were collected for each sample. Electronic compensation was conducted with antibody capture beads (BD Biosciences) stained separately with individual mAbs used in test samples. All analytical gating was performed using FlowJo version 9.0.1 (Tree Star, Inc. Ashland, OR).
CD4 and CD8 memory T cells were identified by sequential gating, using the same gating scheme for all analyzed samples. Cells were identified as lymphocytes by Side Scatter Area (SSC-A) and Forward Scatter Area (FSC-A), and as singlets by Forward Scatter Area (FSC-A) and Forward Scatter Height (FSC-H). Memory CD4 and CD8 cells were defined as Aqua LIVE/DEAD stain−, CD14−, CD19−, CD3+, CD4+ and CD8− or CD8+ and CD4−, and memory cells were defined as CD45RO+ or CD45RO− and CD27−. Cells positive for IFNγ, TNF, IL-2 and IL-10 were expressed as a percentage of either memory CD4 or memory CD8 T cells. In all samples other than that from subject PML-3, IL-10+ cells were defined as cells that were positive for both anti-IL-10 PE and anti-IL-10 APC. In PML-3, IL-10+ cells were defined by anti-IL-10 PE alone.
An amino acid sequence for the entire JCV coding region was constructed based on Mad1 that also included any common variants from NCBI and the literature
Cytokines were measured using the Bio-Plex Pro Human Cytokine 27-plex Assay according to manufacturer's instructions (Bio-Rad Laboratories, Hercules, CA). CSF samples were assayed at a dilution of 1∶2 due to sample volume limitations.
Statistical comparisons were performed using Prism (GraphPad Software, San Diego, CA). Experimental variables were analyzed using Fisher's Exact test, Mann-Whitney U test or Wilcoxon matched-pairs signed rank test. Bars depict median values. P-values <0.05 were considered significant. Analysis and graphical representation of cytokine production was conducted by using the data analysis program Simplified Presentation of Incredibly Complex Evaluations (SPICE version 5.05013 Beta)
(PDF)
(PDF)
(DOCX)
(DOCX)
(DOCX)
The authors wish to thank Dr. Mario Roederer, Dr. Pratip Chattopadhyay and Joanne Yu for antibody conjugation. They are also grateful to Chris Harp, Ann Ligocki and Jason Mendoza for assistance in blood processing.