Fig 1.
Intense transcription of the HTLV-1 sense strand in PBMCs identified by smFISH.
(A) HTLV-1 transcripts, smFISH probes and sample preparation for smFISH. Three sets of probes for smFISH are indicated that hybridize to HTLV-1 transcripts: sense-strand in the pX region (Q670, red stars), HBZ (Q570, yellow stars), and unspliced sense transcripts containing gag region (FAM, green stars). (B) Representative images of HTLV-1+ PBMCs with the sense-strand transcripts at the indicated time points. Blue area indicates the nucleus stained with DAPI, and red spots indicate the HTLV-1 sense transcripts. Scale bar (white) = 5 μm. (C) A site of intense transcription identified by the probes for gag. The image on the left shows the gag staining (green). The image overlaid with sense transcript spots (red, shown in the middle) is presented on the right. The cell presented is from a HAM patient coded TDZ sampled at 4 hours of incubation.
Fig 2.
Sense transcript count per HTLV-1+ cell: Progression over time.
(A) Schematic diagram of the identification of the HTLV-1+ population for analysis. The entire population (in the black and grey rectangles) are the cells identified and analyzed by microscopy. The size of the HTLV-1-infected population (black rectangle) was estimated by multiplying the total cell count by the proviral load as measured by ddPCR. Also see Materials and methods. (B) Sense-transcript count per HTLV-1+ cell (bin size = 20, note the units) at each time point for the two patients with HAM (TBC and TDZ) and two with ATL (LGZ and LGB). Note that the first few bins in each case are out of scale on the y-axis. In the inset for each histogram is indicated the number of cells analyzed. (C) The mean and variance of the sense transcript count per HTLV-1+ cell obtained from four patients, each containing nine time points.
Fig 3.
Rate of onset of HTLV-1 reactivation.
(A) Increase in the fraction of cells that carry ≥3 sense transcripts over time (red dots). The black line for each HTLV-1+ subject indicates a fitted curve with a cumulative gamma distribution. (B) Probability density of the gamma distribution obtained from panel A for each HTLV-1+ subject. The inset numbers indicate the mean time of the onset of reactivation after the start of ex vivo culture.
Fig 4.
Stochastic simulation of the transient HTLV-1 expression in PBMCs ex vivo.
(A) Kinetic model of HTLV-1 proviral transcription, with three promoter states: Off, On, and On-Tax. (B) Stochastic simulation of the model presented in panel A. The best 50 fits (mean and standard deviation shown in black) are superimposed on the smFISH data from the ATL patient LGZ (red bars, reproduced from Fig 2B). (C) The marginal distributions of the best 50 parameters (posterior; green bars); the respective prior distributions are shown in blue. The mean and standard deviation are indicated in the inset. (D) The interval between the initiation of the On-Tax promoter state and the direct transition from the On-Tax to the Off state in the simulation. The frequency of the interval was fitted with an exponential decay, and its probability density function is presented. The mean intervals (τ) ± standard deviation for the prior set I (Table 2) and the prior set II (S4 Fig) are indicated in the inset.
Fig 5.
Identification of HTLV-1 antisense transcripts by smFISH.
(A) Typical image captured by smFISH of HBZ mRNA (yellow spot) in a cell (from a patient with HAM, coded TDZ) without ex vivo incubation. Blue indicates the nucleus stained with DAPI. Scale bar (white) = 5 μm. (B) Frequency distribution of the number of HBZ mRNA molecules carried per HTLV-1+ PBMC. Yellow bars show the frequencies observed by smFISH; the number of HTLV-1+ cells analyzed (N) and the average number of HBZ molecules per HTLV-1+ cell (<m>) are indicated in the inset. Black lines indicate the expected frequencies from a Poisson distribution with the parameter <m>. (C) The kinetic model for the antisense transcription. HBZ mRNA is transcribed at the rate k, and degraded at the rate δ.
Table 1.
HTLV-1+ subjects and PBMCs examined in this study.
Table 2.
Parameters used in the stochastic simulation.