M-Sec induced by HTLV-1 mediates an efficient viral transmission

Human T-cell leukemia virus type 1 (HTLV-1) infects target cells primarily through cell-to-cell routes. Here, we provide evidence that cellular protein M-Sec plays a critical role in this process. When purified and briefly cultured, CD4+ T cells of HTLV-1 carriers, but not of HTLV-1- individuals, expressed M-Sec. The viral protein Tax was revealed to mediate M-Sec induction. Knockdown or pharmacological inhibition of M-Sec reduced viral infection in multiple co-culture conditions. Furthermore, M-Sec knockdown reduced the number of proviral copies in the tissues of a mouse model of HTLV-1 infection. Phenotypically, M-Sec knockdown or inhibition reduced not only plasma membrane protrusions and migratory activity of cells, but also large clusters of Gag, a viral structural protein required for the formation of viral particles. Taken together, these results suggest that M-Sec induced by Tax mediates an efficient cell-to-cell viral infection, which is likely due to enhanced membrane protrusions, cell migration, and the clustering of Gag.

Introduction genes, including the gene encoding the trans-activating protein Tax [17,18]. In this study, we found that when sorted and cultured (Fig 1A), CD4 + T cells of most HTLV-1 carriers tested expressed M-Sec (Fig 1B). Such a change was not observed in CD4 + T cells of HTLV-1individuals. M-Sec induction in CD4 + T cells of carriers was detected as early as 4 h after the beginning of culture and increased at least up to 48 h (Fig 1C). Although we analyzed carriers whose viral load was similar (6.3-13.1%, 9.3 ± 2.6%, n = 7), M-Sec induction was weak or undetected in several carriers (Fig 1B). Similar variability was observed in viral transcripts among HTLV-1 + individuals [19].
Furthermore, HTLV-1 + Tax + T cell lines (SLB-1 and MT-2) expressed M-Sec protein, the level of which was comparable to that of monocyte-derived macrophages (Fig 2A), which are typical M-Sec + cells [3,8]. SLB-1 and MT-2 cells also formed many membrane protrusions ( Fig  2B), the hallmark activity of M-Sec [3]. The HTLV-1cell line (Jurkat) and HTLV-1 + Taxcell lines (S1T, ED, and KK-1) did not show clear M-Sec expression and membrane protrusions. The CD3 + CD4 + CD14cells in the live cell gate were sorted from peripheral blood mononuclear cells (PBMCs). The profile of an HTLV-1 carrier is shown as an example. CD14 was used to exclude monocytes, which abundantly express M-Sec [3,8]. (B) The sorted CD3 + CD4 + CD14cells of HTLV-1individuals (HTLV-1−; n = 7) or HTLV-1 carriers (HTLV-1+; n = 7) were analyzed for the expression of M-Sec mRNA by using qRT-PCR, before or after culturing for 18 h. The expression levels are shown by setting the value of un-cultured cells as 1. � p < 0.05. (C) The sorted CD3 + CD4 + CD14cells of HTLV-1individuals (HTLV-1−; #1 and #2) or HTLV-1 carriers (HTLV-1+; #1 and #2) were cultured for the indicated periods and analyzed as in (B). The expression levels are shown by setting the value of un-cultured cells as 1. Interestingly, KK-1 cells expressed Tax, but the number of Tax + KK-1 cells in the culture was small at any given time because of the sporadic on/off switching of Tax [18] (S1 Fig). The minor Tax + fraction, but not the major Taxfraction, highly expressed M-Sec (Fig 2C). Moreover, in a mouse model of HTLV-1 infection [20], in which humanized mice were inoculated with MT-2 cells, only Tax high human CD3 + cells expressed M-Sec at a detectable level (S1 Table, mouse #3 and #4).
To further analyze M-Sec expression by HTLV-1, we used JEX22, a Jurkat cell-based HTLV-1-infected line that produces HTLV-1 upon stimulation with phorbol 12-myristate The indicated T cell lines were co-stained with phalloidin (green, to visualize F-actin) and DAPI (blue). Scale bar: 10 μm. (C) KK-1 cells expressing GFP in a Tax-dependent manner were subjected to sorting (Taxor Tax + fraction) followed by RNA-seq (accession number in NCBI GEO database: GSE108601) [18]. The RNA-seq data were analyzed for the expression of M-Sec mRNA in those fractions. FPKM, Fragments Per Kilobase of exon per Million mapped reads. https://doi.org/10.1371/journal.ppat.1010126.g002

PLOS PATHOGENS
M-Sec in HTLV-1 transmission 13-acetate (PMA) and ionomycin [20]. When stimulated, JEX22 cells expressed Tax and M-Sec (Fig 3A, right). Such a change was not observed in uninfected control cells (JET35; Fig  3A, left). When expressed in Jurkat cells, Tax induced M-Sec ( Fig 3B). Interestingly, a Tax mutant M22, which can activate the HTLV-1 LTR promoter but not the NF-κB promoter [21] (S2 Fig), failed to induce M-Sec ( Fig 3B). Collectively, these results suggest that Tax, which is expressed by the viral plus-strand transcriptional burst, induces M-Sec in CD4 + T cells through potent activation of the NF-κB pathway.

M-Sec mediates efficient HTLV-1 infection in co-cultures
We performed co-culture assays to examine how M-Sec contributes to HTLV-1 infection. MT-2 or SLB-1 was used as infected cells, and Jurkat carrying the HTLV-1 LTR promoterdriven luciferase gene was used as the target cells. M-Sec knockdown MT-2 or SLB-1 bulk culture cells were prepared (MSec-KD; S3 Fig), and M-Sec inhibitor was also used (MSec-i; S3  Fig). It has been shown that MSec-i reduces the formation of tunneling nanotubes in several cell lines engineered to express M-Sec [8,22] and the production of HIV-1 in the culture of macrophages without affecting podosome formation and phagocytic activity of the cells [8,23]. Control MT-2 or SLB-1 cells were prepared by transducing non-targeting shRNA (Cr; S3 Fig). We found that M-Sec knockdown in MT-2 or SLB-1 cells, or MSec-i addition to control MT-2 or SLB-1 cells reduced viral infection to target cells (Fig 4A), albeit modestly when compared to the anti-envelope (Env) antibody [24] (Cr/Env Ab). A group of small GTPases is a possible downstream effector of M-Sec, as both small GTPases and M-Sec regulate actin cytoskeleton remodeling [3,6,25]. Among the inhibitors of small GTPases tested [26,27], a Cdc42 inhibitor (ZCL278) reduced viral infection to target cells in MT-2-based co-culture ( Fig 4B, upper), whereas a Ral inhibitor (BQU57) reduced viral infection in both MT-2-and SLB-1-based co-cultures ( Fig 4B). This result was consistent with the finding that M-Sec-induced membrane protrusions in macrophage-like RAW264 cells were strongly inhibited by a dominant-negative Ral, and modestly inhibited by a dominantnegative Cdc42 [3].
The results of co-cultures using cell lines prompted us to test the effect of the M-Sec inhibitor (MSec-i) and Ral inhibitor (Ral-i) on a primary cell-based co-culture. HTLV-1-infected CD4 + T cells were enriched in the CADM1 + fraction [28]. When sorted and cultured (S4A Fig), approximately half of the CD3 + CD4 + CADM1 + cells of carriers expressed M-Sec at a detectable level, albeit modestly when compared to monocytes (S4B and S4C Fig). Thus, we co-cultured the CD3 + CD4 + CADM1 + cells of carriers and the CD3 + CD4 + cells of HTLV-1individuals as target cells, and confirmed that both M-Sec inhibitor and Ral inhibitor reduced the number of proviral copies in the co-culture (Fig 5). The timing (day 2 or 4) and extent of the inhibitory effect of those inhibitors were variable among carriers (S5 Fig), as observed with the M-Sec induction in CD4 + T cells of carriers ( Fig 1B). As an obvious increase in the number of sorted CD3 + CD4 + CADM1 + cells during the culture period was not observed in a microscopic analysis, the increase in the number of proviral copies might mainly reflect de novo infection. M-Sec knockdown did not affect the proliferation of MT-2 or SLB-1 cells, and

M-Sec mediates efficient HTLV-1 infection in a mouse model
To examine how M-Sec contributes to in vivo HTLV-1 infection, we next utilized a mouse model [20]. When un-humanized immunodeficient mice were inoculated intraperitoneally with control-or M-Sec knockdown (MSec-KD) MT-2 cells, the number of MT-2 cells in the spleen did not differ between the two groups (S7 Fig). Under these conditions, the provirus in the tissues, including the spleen, was below the detection limit (Un-humanized; Fig 6A). However, once humanized immunodeficient mice were used, the provirus became detectable (Humanized; Fig 6A), indicating de novo infection of reconstituted human T cells. Of note, under these conditions, the number of proviral copies in the liver, spleen, bone marrow, and peripheral blood of the M-Sec knockdown group were lower than that of the control group ( Fig 6A). In mouse models of HTLV-1 infection, the number of human cells in tissues correlated with that of proviral copies [29,30]. For instance, Percher et al. reported the proliferation of infected human T cells in spleens of MT-2-inoculated humanized mice, the extent of which correlated with the level of plasma viral load [30]. Consistent with this finding, the number of human CD45 + cells or CD3 + T cells in the liver, spleen, and bone marrow of the M-Sec knockdown group were lower than that of the control group (Fig 6B and 6C). These results further support the idea that M-Sec mediates efficient cell-to-cell HTLV-1 infection.

M-Sec not only enhances membrane protrusions and cell migration, but also regulates the intracellular distribution of Gag
We next examined how M-Sec contributes to in vitro and in vivo HTLV-1 infection. M-Sec has been reported to enhance the proliferation of several cancer cells [6,7]. However, we did The CD3 + CD4 + CADM1 + cells in the live cell gate were sorted from the PBMCs of HTLV-1 carriers (n = 5). The CD3 + CD4 + cells were also sorted from PBMCs of HTLV-1 − individuals as target cells. They were mixed and co-cultured in the absence (None) or presence of M-Sec inhibitor (MSec-i) or Ral inhibitor (Ral-i) for 2 or 4 days. The number of proviral copies in the coculture was quantified using qPCR and is shown as percentages relative to that of the co-culture with no inhibitor on day 4 (the third bar from the right). � p < 0.05.  Fig 7A). M-Sec knockdown or inhibition also reduced plasma membrane protrusions ( Fig 7B) and the migratory activity of MT-2 cells (Fig 7C). If a cell had one or more F-actin + protrusions longer than 2 μm, it was considered protrusion + ( Fig 7B). The migration of MT-2 cells toward the chemokine SDF-1/CXCL12 was also impaired by M-Sec knockdown or inhibition ( Fig 7C). The enhanced migration and membrane protrusions of infected cells may increase the likelihood of encountering target cells and contact between those cells, thereby contributing to efficient cell-to-cell infection. However, these well-known functions of M-Sec might not be sufficient to explain the potent effect of M-Sec knockdown in the mouse model (Fig 6). Gag is a viral structural protein that localizes to the inner leaflet of the plasma membrane [12, [31][32][33][34][35]. In both MT-2 and SLB-1 cells, the signal of Gag did not overlap with that of Calnexin, an endoplasmic reticulum marker (S10 Fig). The signal of Gag partially overlapped with that of GM130 (a Golgi marker) in MT-2 cells, but not in SLB-1 cells (S10 Fig). Thus, Gag appears to mainly localize to the inner leaflet of the plasma membrane in these cells. Gag is also known to form puncta, which is an indicator of their assembly requisite for the formation of viral particles [12, [31][32][33][34][35]. In fact, both control (S1 Video) and M-Sec knockdown MT-2 cells (S2 Video)

PLOS PATHOGENS
M-Sec in HTLV-1 transmission had many puncta. Of note, we found that control MT-2 cells had a large cluster of Gag in addition to those puncta ( Fig 8A, yellow arrowheads for large clusters), and that M-Sec knockdown (MSec-KD) or inhibition (Cr/MSec-i) reduced the percentage of large Gag cluster + cells ( Fig  8A and 8B, upper). This change was associated with an increase in the number of small clusters of Gag per cell (Fig 8B, lower), which was consistent with an unchanged total amount of Gag per cell (Fig 8C). It was likely that puncta grew into a large cluster by recruiting additional puncta because the large cluster was composed of many puncta (S11 Fig). M-Sec knockdown SLB-1 cells also showed a reduced percentage of large Gag cluster + cells (S12A and S12B Fig,  upper) and increased number of small clusters of Gag per cell (S12B Fig, lower). These results suggest that M-Sec mediates an efficient clustering of Gag. Consistent with the idea, when

Discussion
It has been believed that M-Sec is not related to an infection in T cells because of the lack of their expression under physiological conditions. In this study, we revealed that M-Sec plays a critical role in HTLV-1 infection in CD4 + T cells. Our study suggests that Tax expressed by the viral plus-strand transcriptional burst induces M-Sec through a potent activation of NF-κB pathway, and that M-Sec mediates an efficient cell-to-cell infection of HTLV-1 likely due to enhanced membrane protrusions, cell migration, and the clustering of Gag.
We demonstrated that M-Sec is a Tax-inducible protein. A series of experiments using CD4 + T cells from HTLV-1 carriers (Fig 1), HTLV-1 + cell lines, including SLB-1, MT-2, and KK-1 (Fig 2), and Tax-expressing Jurkat cells (Fig 3) support the idea that Tax translated by the viral plus-strand transcriptional burst induces the expression of M-Sec in CD4 + T cells. Tax-mediated M-Sec induction required activation of the NF-κB pathway (Fig 3B). Tax is known to activate NF-κB [36], while the effects of HIV-1 on NF-κB activation depend on the type of cells or their activation state [37], which explains why M-Sec was not induced in HIV-1-infected Jurkat cells despite active viral replication [8]. The finding that LMP-1, an oncoprotein of Epstein-Barr virus, upregulates M-Sec in nasopharyngeal carcinoma cells through NF-κB activation [5] also supports M-Sec induction in CD4 + T cells by the Tax-NF-κB cascade.
Tax is essential for de novo infection of HTLV-1 as it induces the expression of viral genes, but it simultaneously allows the recognition of infected cells by cytotoxic T lymphocytes [38]. Thus, the transient or intermittent expression of Tax by the viral plus-strand transcriptional burst [17,18] may be a strategy for HTLV-1 to maintain a balance between escape from the immune system and de novo infection [38]. To ensure viral infection within its limited period of expression, Tax may induce cellular proteins including M-Sec, which mediates an efficient transmission of HTLV-1 as demonstrated by our analyses of co-cultures (Figs 4 and 5) and the mouse model (Fig 6).
The ability of M-Sec to enhance migratory activity of infected cells (Fig 7C) can increase the likelihood of encountering target cells, thereby contributing to an efficient cell-to-cell infection. In fact, among Tax-inducible cellular proteins that facilitate contact between HTLV-1-infected cells and target cells [39][40][41], Gem, a member of the small GTP-binding proteins, not only enhances the migratory activity of infected cells but also mediates an efficient viral infection from infected cells to target cells in a co-culture assay [41]. The ability of M-Sec to enhance the formation of membrane protrusions (Fig 7B) can facilitate contact between the infected cells and target cells. The protrusions of MT-2 or SLB-1 cells (Fig 2B) appear to be smaller in length but larger in number than those of HIV-1-infected macrophages or U87 cells [8,9]. Furthermore, the ability of M-Sec to facilitate the clustering of Gag (Figs 8, S12, and 9) may be beneficial for viral transmission because Gag is the key driver for the formation of viral particles [31][32][33][34][35]. Consistent with this idea, an alanine-scanning mutagenesis analysis of Gag has identified several mutants that not only fail to form puncta but also produce lesser amounts of viral particles when compared to the wild type Gag [34].
M-Sec inhibition in macrophages or knockdown in U87 glioma cells reduced the production of HIV-1 [8,9]. However, such viral reduction became less obvious over time in the culture [9], implying that M-Sec mainly contributes to the initial phase of HIV-1 transmission. In the co-culture of the present study, reduced viral transfer from M-Sec knockdown MT-2 cells to Jurkat cells was observed at 16 h (Fig 4A, upper), but not at 36 h (S13 Fig). In addition, in several cases in the co-culture using CADM1 + T cells of HTLV-1 carriers, the reduced proviral copies by M-Sec inhibition found at 2 days were lost at 4 days (S5 Fig, #4 and #5). Meanwhile, the extent of the reduced proviral copies by M-Sec knockdown in the mouse model found at 4 weeks (Fig 6) appears to be more obvious than that in the co-cultures. Thus, to clarify whether

PLOS PATHOGENS
M-Sec in HTLV-1 transmission M-Sec mainly contributes to the initial phase of HTLV-1 transmission, detailed time-course analyses will be necessary for the assays involving the mouse model.
M-Sec expression in several cancer cells enhances their invasion/metastasis [5][6][7] or proliferation [6,7]. In our cultures, M-Sec knockdown or inhibition in MT-2 cells reduced their migration (Fig 7C), but not proliferation (S6 and S8 Figs). When intraperitoneally injected into immunodeficient mice, the numbers of M-Sec knockdown MT-2 cells in the spleen were not different from those of the control MT-2 cells (S7 Fig). These results suggest that M-Sec knockdown does not affect the proliferation or survival of MT-2 cells in mice or the migration from the peritoneal cavity to the spleen. Thus, it is tempting to speculate that a weak intra-tissue migration of M-Sec knockdown MT-2 cells and their reduced membrane protrusions/Gag clustering could explain the reduced viral infection in the mouse model, although further studies including the identification of M-Sec mutants which lack selected functions are necessary to prove this hypothesis.
How different M-Sec functions are related to each other is still unclear. How these functions contributes to HTLV-1 transmission and the extent to which each function contributes remain unexplored. Despite these unresolved questions, the present study revealed the importance of M-Sec for HTLV-1 transmission. M-Sec is a new and useful tool to further clarify the process of cell-to-cell infection of HTLV-1.

Ethics statement
All protocols involving human subjects were reviewed and approved by the institutional review board of Kumamoto University. The protocols were reviewed and approved by the ethical committee of Imamura General Hospital. Written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki. The research protocols for mouse experiments were approved by the animal center and the ethical committee of the National Institute of Infectious Diseases and were performed according to the institutional guidelines for the experimental use of animals.

PLOS PATHOGENS
M-Sec in HTLV-1 transmission CADM1 (3E1; MBL, Nagoya, Japan). Total RNA and genomic DNA were isolated using the RNeasy micro kit and QIAamp DNA micro kit (both from Qiagen), respectively. The RNA and DNA samples were used for qRT-PCR (M-Sec mRNA) and qPCR (provirus), respectively. The cells were also used for immunofluorescence (Gag).

HTLV-1 + T cell lines
MT-2 cells were obtained from the Japanese Collection of Research Bioresources Cell Bank. S1T, KK-1, and SLB-1 cells were provided by N. Arima (Kagoshima University, Japan), H. Hasegawa (Nagasaki University, Japan), and M. Fujii (Niigata University, Japan), respectively. All the cells were maintained in RPMI 1640 medium containing 10% FCS. KK-1 cells were cultured in the presence of 10 ng/mL rhIL-2. The RNA-seq data of KK-1 cells expressing GFP in a Tax-dependent manner (accession number in NCBI GEO database: GSE108601) [18] were also analyzed.

M-Sec knockdown in MT-2-and SLB-1 cells
The knockdown of M-Sec in MT-2 or SLB-1 cells was performed using MISSION shRNA lentiviral transduction particles (Sigma). pLKO.1-puro-CMV-tGFP shRNA targeting human M-Sec (TRCN0000330220) or a scrambled nontargeting control was used. Control cells expressing non-targeting siRNA or cells expressing M-Sec-targeting siRNA were selected under the culture containing 0.3 μg/mL puromycin and used without cloning to exclude the possibility of clonal variation. GFP expression in these cells and reduced M-Sec expression in M-Sec knockdown cells were confirmed using flow cytometry and western blotting, respectively.

Jurkat cells
The Jurkat cell-based JEX22 cell line and its parental JET35 cell line [20,44] were used. JEX22 had been infected with the HTLV-1 molecular clone pX1 MT-M [45] and produced HTLV-1 upon stimulation with 50 ng/mL PMA and 1 μM ionomycin [20]. Jurkat cells (clone E6-1) were obtained from the American Tissue Culture Collection. The cells were nucleofected with Tax plasmid by using the Cell Line Nucleofector kit V and Nucleofector II (both from Lonza; the program T-014). The wild type or M22 mutant Tax derived from pX1 MT-M were flagtagged and sub-cloned into the pIRES-neo3 vector (Invitrogen). The nucleofected cells were subjected to qRT-PCR for M-Sec or Tax mRNA. Jurkat reporter cells were also used in the coculture assay (see below).

Inhibitors
The M-Sec inhibitor (MSec-i) was identified by an affinity-based chemical array screening [8]. In brief, compounds were arrayed onto the photoaffinity linker-coated slides, which were incubated with the lysates of 293T cells expressing either DsRed or DsRed-M-Sec fusion proteins. The fluorescence signals were quantified, and the identified MSec-i was synthesized at Pharmeks (Moscow, Russia). BQU57 (Ral inhibitor), ZCL278 (Cdc42 inhibitor), and NSC23766 (Rac1 inhibitor) [26,27] were purchased from Sigma-Aldrich. MSec-i, BQU57, and ZCL278 were dissolved in DMSO, and NSC23766 was dissolved in H 2 O. These inhibitors were added to cultures at a final concentration of 10 μM (0.1% v/v), and the same volume of DMSO or H 2 O was added as a vehicle control.

qPCR for provirus
The number of proviral copies was quantified using qPCR and a 7500 Fast Real-Time PCR system, as described previously [20]. The pX region of the provirus was amplified using the following primers and probe: 5'-CGGATACCCAGTCTACGTGTT-3' (forward), 5'-CAGTAGG GCGTGACGATGTA-3' (reverse), and 5'-CTGTGTACAAGGCGACTGGTGC C-3' (probe) [47]. Human RNase P, a single copy gene, was also amplified as an internal control [20,47], using primers and probes purchased from Applied Biosystems. When humanized mouse tissues were analyzed, the human ribosomal protein S19 (RPS19) gene was used as an internal control [48], along with the following primers and probe: 5'-GGAACGGTGTCAGGATTCAAG-3' (forward), 5'-AGCGGCTGTACACCAGAAATG-3' (reverse), and 5'-TCTGACTGCTCTGGGCGCTA GTCCC-3' (probe). The assay was performed in triplicate, and the copy number of the pX region in each sample was determined relative to that in the reference TL-Om1 cells [20,47,49]. The number of proviral copies of samples was calculated as ([2 x copy of pX]/[copy of RNase P or RPS19]) x 100, and expressed as copies per 100 cells.

Western blotting for M-Sec and Tax
Western blotting was performed as described previously [8,9]. The antibodies used were as follows: anti-M-Sec (F-6; Santa Cruz Biotechnology), anti-Tax (Lt-4) [50], and anti-α-tubulin (DM1A; Santa Cruz Biotechnology). Detection was performed with HRP-labeled secondary antibodies (GE Healthcare), Immunostar LD Western blotting detection reagent (Wako, Japan), and an image analyzer (LAS-3000; FujiFilm). Macrophages obtained by culturing monocytes of healthy volunteers in the presence of the cytokine M-CSF [8] were analyzed as a reference.

Co-culture using cell lines
Jurkat cells carrying the firefly luciferase gene under the control of the HTLV-1 LTR promoter [51] were used as target cells. The cells were also transfected with the pRL-SV40 control Renilla luciferase plasmid (Promega). MT-2 or SLB-1 cells were used as infected cells. These cells were suspended in RPMI 1640 containing 10% FCS, mixed at a ratio of 1:1, seeded into 24-well plates (total 1 x 10 5 cells per well), and cultured for 16 h. Luciferase activity was measured using a dual-luciferase reporter assay system and a Glomax 96 microplate luminometer (Promega), and firefly luciferase activity was normalized to control Renilla luciferase activity in the same sample. In an assay in which the effect of chemical inhibitors was tested, MT-2-or SLB-1 cells were pre-treated with each inhibitor for 48 h and used for the co-culture in which fresh antibodies were re-added. In a selected assay, MT-2-or SLB-1 cells were pre-treated with

PLOS PATHOGENS
M-Sec in HTLV-1 transmission 20 μg/mL of anti-gp46 Env neutralizing antibody LAT-27 [24] for 1 h and used for the co-culture in which fresh antibodies were re-added.

Co-culture using CD4 + T cells of HTLV-1 carriers
CD3 + CD4 + CADM1 + cells were sorted from PBMCs of HTLV-1 carriers as infected cells. Carriers with detectable levels of M-Sec expression in CD4 + T cells after ex vivo culture were selected for this assay. CD3 + CD4 + cells were also sorted from the PBMCs of HTLV-1individuals as target cells. They were mixed at a ratio of 1:10 to 1:30, seeded into 96-well plates (3,000-5,000 cells/well for CD3 + CD4 + CADM1 + cells), and cultured with RPMI 1640/10% FCS/rhIL-2 for up to 4 days in the absence or presence of M-Sec-or Ral inhibitor. The genomic DNA of the cells in the co-culture was isolated and analyzed for proviral copies, as described above.

Mouse model of HTLV-1 infection
Humanized mice were prepared by transplanting human hematopoietic stem cells (HSCs) into immunodeficient mice [20]. Immunodeficient NOD/SCID Jak3 null (NOJ) mice [52] were obtained from Kyudo (Saga, Japan). Human CD133 + HSCs [53] were enriched from cord blood (the Japanese Red Cross Cord Blood Bank, Tokyo, Japan), using the MicroBead kit (Miltenyi Biotec). Their purity was routinely > 90%. HSCs were intrahepatically transplanted into newborn NOJ mice (1 × 10 5 cells/mouse). After 3 months, the number of human cells was measured using AccuCheck counting beads (Thermo Fisher). Human CD45 + cells in the peripheral blood are usually > 40%, in which human CD4 + T cells are > 20%. Mice with similar numbers of human cells were used for the inoculation of MT-2 cells.
The control or M-Sec knockdown MT-2 cells were irradiated using an MBR1505R2 X-ray cabinet system (70 Gy; Hitachi Power Solutions, Japan), and injected into the peritoneal cavity of the humanized mice (2 x 10 6 cells/mouse). After 4 weeks, the cells in the peripheral blood, liver, spleen, and bone marrow were subjected to the isolation of genomic DNA, followed by proviral qPCR, as described above. Tissue samples were subjected to immunohistochemistry. The excised tissues were fixed in 10% phosphate-buffered formalin and embedded in paraffin. The paraffin blocks were cut into 3-μm-thick sections and mounted on silane-coated glass slides. The sections were stained with hematoxylin and eosin and processed for immunohistochemistry with anti-CD3 (2GV6; Roche) or anti-CD45 antibodies (NCL-L-LCA; Dako). For antigen retrieval, the sections were immersed in citrate buffer (pH 6.0) and heated for 10 min at 121˚C. Specific antigen-antibody reactions were visualized using the EnVision+ system

Cell migration
The migration of MT-2 cells was measured using a transwell assay with 5-μm pore size inserts (Corning). The inserts were placed into 24-well plates containing 600 μL RPMI1640/10% FCS in the absence or presence of rhSDF-1 (BioLegend) at a final concentration of 20 or 100 ng/ mL. Then, cells were added to the inserts (2.5 x 10 5 cells in 100 μL RPMI1640/10% FCS) and incubated at 37˚C for 24 h. The number of cells that migrated through the inserts was enumerated using the trypan blue dye exclusion method.

Membrane protrusion and cell morphology
Membrane protrusions were assessed using immunofluorescence, as described previously [8,9]. Cells were seeded onto BioCoat Poly-D-Lysine culture slides (Corning) and incubated for 30 min at 37˚C to allow for adherence to the glass surface. The cells were then fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and incubated with phalloidin conjugated to AlexaFluor488 and DAPI (both from Molecular Probes) to visualize F-actin and nuclei, respectively. Signals were visualized using an FV1200 confocal laser-scanning microscope (Olympus), and image processing was performed using the FV Viewer ver. 4.1 soft ware (Olympus). If a cell had one or more F-actin + protrusions longer than approximately 2 μm, it was considered protrusion + . The cell surface area, height, and diameter were also quantified using ImageJ 1.52n software (NIH) [9].

Analyses of Gag
The intracellular distribution of Gag was visualized using immunofluorescence. In brief, cells were fixed, permeabilized, and stained with anti-p24 Gag antibodies (6G9; Santa Cruz Biotechnology) followed by anti-mouse IgG-AlexaFluor568 or anti-mouse IgG-AlexaFluor633 (both from Molecular Probes). The nuclei were stained with DAPI. Signals were visualized using an FV12 00 or FV3000 confocal laser-scanning microscope (Olympus), and image proce ssing was performed using FV Viewer ver. 4.1 or FV31S-SW software (both from Olympus). The amount of intracellular Gag was quantified using flow cytometry. The antibodies used were anti-p24 Gag (6G9) and APC-anti-mouse IgG (Molecular Probes).

Statistics
Differences between groups were determined using the Student's t-test. For multiple compariso ns, a two-way ANOVA with Sidak's multiple comparisons test was used. p < 0.05 was considered significant. These analyses were performed using Prism software (GraphPad Software).
Supporting information S1 Table. Provirus, Tax mRNA, and M-Sec mRNA in human CD3 + cells of MT-2-inoculated humanized mice. (PPTX) S1 Fig. KK-1 Fig 2C). The reporter cassette expressing GFP under the control of the Tax-responsive element is schematically shown in the upper panel [18], and an example of GFP expression in KK-1 cells carrying the reporter cassette is shown in the lower panel.  Fig 3B). To confirm that the Tax mutant M22 used in Fig 3B does not activate the NF-κB promoter [21], a co-transfection experiment using 293A cells and LipofectAMINE3000 (Invitrogen) was performed. The plasmids used were as follows: the empty vector, Tax expression plasmid (the wild type or M22), firefly luciferase reporter plasmid (NF-κB-Luc; a gift from H. Iha, Oita University, Japan), and pRL-SV40 control Renilla luciferase plasmid (Promega). Luciferase activities were measured as described in Materials and Methods section, and are shown by setting the value of the empty vector/luciferase plasmid-transfected cells as 1 (n = 3).

PLOS PATHOGENS
M-Sec in HTLV-1 transmission