PBMC isolated from 5–10 ml of heparinized venous blood by Ficoll–Hypaque density gradient centrifugation (Sigma-Aldrich) were washed twice with PBS and were plated in 24-well tissue culture plates (Costar, NY, USA) at 4 × 10⁶ cells/ml and incubated at 37°C and 5% CO₂ in RPMI1640 medium supplemented with 2mM l-glutamine, gentamycin (50 µg/ml), and 10% heat-inactivated fetal calf serum (Gibco, NY, USA).

Human T-cell lymphotropic virus-1 matrix protein p19 was quantified in cell-free supernatant of HAM/TSP patients’ PBMC and AC and HC using RetroTek HTLV-1/2 p19 Antigen ELISA kit (ZeptoMetrix) after 48 h of in vitro culture. Proviral load (PVL, i.e., viral DNA integrated into the host genome) in HAM/TSP patients and AC was quantified as published (30, 31).

For phenotypic analysis, PBMC were resuspended at a density of 200,000 cells in 50 µl of 1% BSA, 0.1% NaN₃ in PBS (+20% human serum to block Fc receptors), and incubated for 30 min on ice with mAbs specific for CD3, CD4, CD8, CD80, CD86, CD95/Fas, and HLA-DR and corresponding isotype controls (BD Biosciences). For total Fas surface quantification and apoptosis, a minimum of 100,000 events/sample were stained and acquired with FACSort and FACSCanto II flow cytometers (BD Biosciences) and analyzed using CellQuest and Diva software, respectively.

Lymphoproliferation was quantified by [³H]-thymidine incorporation and flow cytometry [as described in Ref (29, 32)], the initial stage of apoptosis was analyzed using annexin V staining, whereas cells in the late/final stage of apoptosis were identified as a sub-diploid population by flow cytometry. Nuclear fragmentation was quantified by fluorescence microscopy and ELISA (Cell Death Detection plus, Boehringer Mannheim, Germany).

PBMC were cultured as above for 48 h in the presence or absence of agonist or antagonist anti-Fas mAbs (1 µg/ml, Alexis Biochemicals) or anti-CD3 mAb (Butantan Institute, Sao Paulo, Brazil) as a positive control for in vitro apoptosis.

Total RNA was extracted from PBMC according to manufacturer’s protocol (QIAgen, Venlo, The Netherlands). Whole genome microarray was performed at VIB Nucleomics (Leuven, Belgium) using GeneChip^® Human Gene1.0 ST Array (Affymetrix, Santa Clara, CA, USA), according to manufacturer’s specifications. Data were analyzed using Bioconductor limma package, using a moderated t-test, resulting p-values were corrected for genome-wide testing (5% FDR). All microarray raw data are available at Gene Expression Omnibus database (GEO, http://www.ncbi.nlm.nih.gov/geo/) series accession number GSE82160.

The use of parametric (t-test, Pearson correlation) or non-parametric (Mann–Whitney or Spearman rank correlation) tests was based on normal distribution as determined by Kolmogorov–Smirnov test (all GraphPad Prism v5.0 or v6.0). A p-value of <0.05 was considered significant for all statistical tests. Transcriptome-wide correlation of FAS mRNA expression levels was calculated using Spearman rank correlation test, with stringent correction for multiple testing (5% FDR).

In a first cohort, we quantified surface Fas levels as well as apoptosis by flow cytometry, ex vivo in PBMC from HC (HTLV-1-negative, n = 14), AC (HTLV-1-positive, n = 30), and HAM/TSP patients (n = 18). We observed a significant increase in ex vivo levels (%) of Fas⁺ lymphocyte in AC (1.8-fold) as well as in HAM/TSP patients (2.1-fold), when compared to HC (Kruskal–Wallis, Dunn’s posttest, p < 0.05, p < 0.001, respectively, Figure 2A). Moreover, lymphocyte Fas level on a per-cell basis, expressed as mean fluorescence intensity (MFI), revealed an eight-fold increase in AC and a striking 19-fold increase in HAM/TSP (Kruskal–Wallis, Dunn’s post-test, p < 0.001), when compared to HC, but also when compared to AC (p < 0.05, Figure 2B), indicating that clinical progression to HAM/TSP is characterized by a predominant Fas^hi lymphocyte population, possibly primed for apoptosis. To confirm the two-step model of Fas increase, we performed a post hoc test for linear trend, which was highly significant (p < 0.001) for both % (slope 18.8) and MFI (slope 64.1).

Next, we proceeded to examine Fas expression in CD4, CD8, and B cell subsets in more detail in an independent second cohort of HC (n = 7), AC (n = 6), and HAM/TSP patients (n = 9). There was no difference in the percentage of cells expressing Fas between the three clinical groups for either cellular subset (Figure 2C). However, we observed a small but significant linear trend in Fas MFI of CD4⁺ T cells with clinical status (ANOVA p = 0.067, post-test for linear trend p < 0.05, slope 349.2), but not in CD8⁺ T cells or B cells. Thus, the strongest difference between the clinical groups was in total Fas⁺ lymphocytes rather than specific subsets, revealing a Fas^hi phenotype in HAM/TSP. To verify if this Fas^hi phenotype might be shared among neuroinflammatory disorders, we compared Fas expression between HAM/TSP and MS patients. As shown in Figure 2D, we found a significant 1.6-fold increase in % of ex vivo Fas⁺ lymphocytes in HAM/TSP (Mann–Whitney, p = 0.03), as well as a 2.4-fold increase in Fas MFI, which approached statistical significance (Mann–Whitney, p = 0.08).

Finally, ex vivo spontaneous apoptosis in HAM/TSP and AC, as measured by DNA degradation (quantified as sub-diploid cells in flow cytometry) occurred at very low levels (<0.2% of PBMC, data not shown). Therefore, we questioned if the observed ex vivo increase in lymphocyte Fas surface expression in HAM/TSP reflected the immunological, virological, or clinical status of HAM/TSP patients, rather than an apoptosis-prone status.

To explore possible clinical relevance of this increased lymphocyte Fas in HAM/TSP patients, we correlated ex vivo Fas surface expression to patient demographic and clinical data. We observed that, in HAM/TSP, ex vivo lymphocyte Fas (% or MFI) was not correlated to age, gender, disease duration, or severity. In addition, ex vivo lymphocyte Fas was not significantly correlated to PVL in AC or HAM/TSP (p > 0.05). However, ex vivo Fas levels (%) correlated significantly to lymphocyte activation markers HLA-DR and CD86 (Figures 3A,B), implying that increased Fas expression may be coupled to immune activation and/or inflammation in HAM/TSP.

Upon quantification of in vitro Fas⁺ lymphocyte expression in HC, AC, and HAM/TSP patients by flow cytometry, we again observed a two-step increase in % Fas⁺ lymphocytes: two-fold in AC and 3.4-fold in HAM/TSP vs. HC (post-test for linear trend, p = 0.0001, slope 27.0) (Figure 4A). In HAM/TSP, in vitro Fas levels per-cell (MFI) were even more pronounced, with an eight-fold increase over HC. Hence, clinical status impacts both ex vivo (Figures 2A,B) and in vitro (Figure 4A) Fas expression. In addition, Fas in vitro levels showed a significant negative correlation to age of disease onset in HAM/TSP patients (p = 0.019, Pearson’s r = −0.69, n = 11) (Figure 4B), but not to age, disease duration, and gender, suggesting Fas^hi phenotype predisposes to earlier, aggressive disease manifestation. Further, in vitro Fas expression neither correlated to viral p19 protein level (p = 0.41), nor to PVL (p = 0.14) in HTLV-1-infected individuals (data not shown).

In agreement with its role as a death receptor in immune homeostasis, Fas surface expression positively correlates with spontaneous in vitro apoptosis in HC, while this correlation was lost in AC (data not shown). Surprisingly, ex vivo Fas expression correlated negatively (Figure S1 in Supplementary Material) to spontaneous in vitro apoptosis in HAM/TSP. Furthermore, in vitro Fas level (MFI) also correlates negatively to lymphocyte apoptosis in HAM/TSP (Figure 5A). This negative correlation was confirmed by fluorescence microscopy. As shown in Figure 5B, Fas^hi cells are negative for annexin V staining and display normal nuclear morphology, whereas Fas^lo cells were seen to undergo apoptosis by both annexin V staining and nuclear condensation/fragmentation, occasionally triggering phagocytosis by macrophages, emphasizing their apoptotic nature. Since resistance to Fas induced apoptosis has been observed in vitro in lymphocytes from MS patients (33), we compared in vitro lymphocyte Fas expression and apoptosis between HAM/TSP and MS patients. As shown in Figure 5C, there was a significant increase (2.4-fold, Mann–Whitney test, p = 0.019) in Fas MFI in HAM/TSP as compared to MS patients, but not apoptosis (as measured by annexin V staining, Mann–Whitney test, p = 0.84). In contrast to HAM/TSP, no correlation was observed between Fas MFI and apoptotic cells in MS patients (p = 0.35, data not shown). Taken together, the significant negative correlations between ex vivo and in vitro Fas lymphocyte expression and in vitro apoptosis observed only in HAM/TSP, suggest a possible selective defect in Fas-mediated apoptosis. Hence, we next aimed to comprehensively explore non-apoptotic Fas signaling in HAM/TSP.

We quantified in vitro spontaneous lymphoproliferation by [³H]-thymidine incorporation in HAM/TSP patients. Surprisingly, we found that Fas expression positively correlates to spontaneous lymphoproliferation in vitro (Figure 6A), which might imply that the observed defect in Fas-mediated proapoptotic signaling in HAM/TSP might be explained as a bias in Fas signaling toward proliferation rather than apoptosis. Therefore, we hypothesized that Fas^hi cells might be already proliferating in vivo in HAM/TSP although at a very low level. We thus extended our previously described (29) sensitive flow cytometry assay to quantify Fas⁺ diploid vs. tetraploid (proliferating) lymphocytes ex vivo in HAM/TSP patients, stained immediately after PBMC isolation, without in vitro culture. As shown in Figure 6B, virtually all of the proliferating cells were Fas^hi (99.2 ± 0.8%), as compared to non-proliferating lymphocytes (69.4 ± 5.9%, Paired t-test, p = 0.0082).

We then examined if this apparent defect in Fas-mediated apoptosis might be reversible by stimulating with agonist anti-Fas mAb, and if blocking with antagonist anti-Fas mAb could reveal ongoing Fas-FasL signaling in HAM/TSP. Hence, we treated PBMC in vitro with anti-Fas mAb (agonist or antagonist) or anti-CD3 mAb as a positive control. No decrease in spontaneous apoptosis was observed upon treatment with antagonist anti-Fas mAb, confirming our hypothesis of inactive Fas-FasL signaling in vitro in HAM/TSP. Interestingly, treatment with agonist anti-Fas mAb resulted in significantly increased apoptosis (1.7-fold, p < 0.05), similar to treatment with anti-CD3 mAb (positive control, 1.8-fold, p < 0.01) (Figure 7A). These results imply that agonist anti-Fas mAb treatment can restore the apparent defect in apoptosis in HAM/TSP, at least in vitro.

Considering the significant correlation between in vitro Fas expression to age of onset in HAM/TSP, we resorted to genome-wide transcriptional analysis of PBMC treated in vitro with agonist or antagonist Fas mAb, to explore the broad pro/antiapoptotic, inflammatory, proliferative, and immunoregulatory Fas signaling pathways specifically triggered in HAM/TSP. Microarray analysis revealed that in vitro treatment with agonist anti-Fas mAb, significantly downregulated 190 genes and upregulated 59 genes (Tables S1A,B in Supplementary Material), while treatment with antagonist anti-Fas mAb downregulated 38 genes and upregulated 18 genes (Tables S1C,D in Supplementary Material). Thus, triggering Fas signaling effects a broader gene spectrum than inhibiting it. This was also evident from Ingenuity^® pathway analysis (IPA), since no biological functions were significantly associated with antagonist anti-Fas mAb treatment, whereas treatment with agonist anti-Fas mAb resulted in 22 significantly associated biological functions (5% FDR-adjusted and a stringent cut-off of at least five enriched molecules per pathway) (Table S2 in Supplementary Material). The top 10 biological functions activated by agonist anti-Fas mAb (Table S2 in Supplementary Material), highlight cellular migration, especially of myeloid cells. In addition, IPA network analysis (Figure 7B) of Fas-triggered gene expression reveals a central role for NFκB pro-survival signaling, connecting several upregulated proliferative and inflammatory molecules (TNF, JNK, RNA Polymerase II, POLR2D, HIST1H3A, HIST1H2AB) as well as downregulated anti-proliferative genes (L3MBTL2, CARD6). This central role for NFκB signaling was confirmed by ingenuity upstream regulator analysis, identifying RelA as the top upstream regulatory molecule upon triggering Fas signaling (target genes: BCL2A1, CASR, CXCL3, ICAM1, L3MBTL2, PTGES, TGM2, TNF, and TPMT; p = 0.000032). Again, blocking Fas signaling did not yield any significantly enriched upstream regulators (using the same stringent cut-off of five enriched molecules/pathway, data not shown).

Finally, we used a pathway-based data mining approach, to test our hypothesis of biased Fas signaling, and to possibly extend our findings by including additional pro- and anti-apoptotic genes (e.g., TRAIL, cFlip, etc.). For this purpose, we explored possible interactions of Fas mRNA within the ex vivo global gene expression profile in PBMC of HAM/TSP patients (n = 6). Using transcriptome-wide correlation, 4,554 genes significantly correlated to Fas transcript levels (Table S3 in Supplementary Material), after stringent FDR-correction for multiple testing. Using annotated ingenuity pathways, we found a significant enrichment for proliferation-related genes (159 of 4,554 genes, p = 0.023). However, apoptosis, as defined by IPA, was not enriched amongst the ex vivo Fas-correlating genes (71 genes out of 4,554 genes, p = 0.10).