REVIEW Molecular and Cellular Aspects of HTLV-1 Associated

Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux1,3, A-S Gabet1 and E Wattel1,2

1Unite´ d’Oncogene`se Virale, UMR5537 CNRS-Universite´ Claude Bernard, Centre Le´ ´on Be´rard, Lyon, France; and 2Service d’He´matologie, Pavillon E, Hoˆpital Edouard Herriot, Lyon, France

Most cancers and leukemias are preceded by a prolonged per- leukemia/lymphoma (ATLL).2,3 Furthermore, this virus has iod of clinical latency during which cellular, chromosomal and been associated with the development of a chronic progress- molecular aberrations help move normal cell towards the malig- ive neuromyelopathy (tropical spastic paraparesis (TSP)/HTLV- nant phenotype. The problem is that premalignant cells are 4 usually indistinguishable from their normal counterparts, ther- 1-associated myelopathy (HAM)), and, to a lesser extent, to eby ruling out the possibility to investigate the events that a variety of inflammatory diseases.5–9 govern early leukemogenesis in vivo. Adult T cell Among the 15–25 million individuals infected worldwide, leukemia/lymphoma (ATLL) is a T cell malignancy that occurs approximately 3% to 5% will develop ATLL, depending on as after a 40–60-year period of clinical latency in about 3–5% of yet unknown cofactors. ATLL harbors different clinical fea- HTLV-1-infected individuals. ATLL cells are monoclonally expanded and harbor an integrated provirus. A persistent tures resulting in a division of the spectrumof the disease into 10,11 oligo/polyclonal expansion of HTLV-1-bearing cells has been four clinical subtypes referred to as acute, lymphoma, shown to precede ATLL, supporting the fact that in ATLL tumor chronic and smoldering subtypes. Infection early in life is cru- cells arise from a clonally expanding non-malignant cell. It is cial in the development of ATLL.12 The prolonged premalig- possible to isolate infected, ie preleukemic, cells during the nant asymptomatic phase that precedes leukemia is charac- premalignant asymptomatic phase of the infection, thus provid- terized by the persistent oligoclonal expansion of HTLV-1 ing an exceptional system to study the mechanisms underlying 13–21 human cancers. Here we review some of the consequences of infected cells. This results fromthe action of the HTLV-1 22,23 HTLV-1 on its host cell in vivo, at different stages of infection. Tax protein that triggers T cell proliferation and negatively Leukemia (2003) 17, 26–38. doi:10.1038/sj.leu.2402777 interferes with the main cellular pathways involved in the Keywords: adult T cell leukemia; HTLV-1; genetic variability; maintenance of genomic integrity.22–30 The mutagenic effect leukemogenesis; mutation of Tax generates a hitherto undescribed mechanism of retrovirus genetic variability that results from somatic mutations of the proviral sequence rather than fromreverse transcription. 18 Introduction In fact, HTLV-1 genetic variability and infected cell genetic instability go hand in hand, thereby establishing a pro- The fact that the DNA of cancer cells is different fromthe cancer genotype. DNA of normal cells indicates that carcinogenesis involves substantial errors in DNA replication, deficits in DNA repair, and alterations in chromosomal segregation.1 Given the rarity Reverse transcription is an error-prone process of mutations in normal cells and the large numbers of mutations observed in human cancers, it has been proposed RNA viruses are unique in that their polymerases, although that the spontaneous mutation rate in normal cells is not suf- clearly marked by the same basic quaternary structure of the ficient to account for the number of mutations found in human polymerization domain common to all nucleic acid poly- tumors. In other words, cancer is assumed to result from an merases, lack a particular domain encoding a 3Ј exonucle- increased somatic mutations frequency, ie from a mutator ase.31,32 Furthermore, host exonucleases are not incorporated phenotype. DNA damage, combined with loss of DNA repair into specific virions. Consequently, any polymerization error and clonal selection, is thought to help move the normal cell becomes immortalized. Not surprisingly, the highest mutation towards such a phenotype. rates observable for microbes are to be found in RNA The main human tumor viruses include human T cell leuke- viruses.33 Higher mutation rates – referred to hypermutations – mia viruses, Epstein–Barr virus, human papilloma virus, can theoretically and practically be encountered. However, Kaposi sarcoma-associated herpes virus and human hepatitis there comes a threshold beyond which the mutation rate is B and C viruses. These virus provide exceptional systems to deleterious to the survival of the virus.33 Retroviruses are study the mechanisms underlying human cancers. Among somewhat schizophrenic: on the one hand they undertake these, human T cell leukemia viruses type 1 (HTLV-1), the first DNA polymerization, while on the other hand their reverse human pathogenic retrovirus isolated, is the etiologic agent of transcriptase is devoid of any 3Ј exonucleolytic activity. Thus the malignant CD4 T lymphoproliferation adult T cell the genetic variability typical of RNA viruses becomes part of their way of life. For some reason that is not yet evident, the average mutation rate of retroviruses is approximately 20-fold Correspondence: E Wattel, Unite´ d’Oncogene`se Virale, UMR5537- lower than that of an ensemble of familiar RNA viruses.34 CNRS-Universite´ Claude Bernard, Centre León Be´rard, 28, rue Lae¨n- nec 69373 Lyon cedex 08 France; Fax: 33 4 78 78 27 17 3 Present address: National Fund for Scientific Research, Faculte´ Univ- HTLV-1 is genetically stable ersitaire des Sciences Agronomiques, Biologie Molećulaire, 13 avenue Marećhal Juin, B5030 Gembloux, Belgium 35 The first two authors contributed equally to this work The first complete HTLV-1 sequence emerged in 1983. Received 11 June 2002; accepted 31 July 2002 Since then, a forcible accumulation of sequence data has Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 27 shown that HTLV-1 is extraordinarily stable genetically. While some segments are slightly more variable than others, geo- graphically widely dispersed viruses are <10% divergent at the nucleic acid level. This includes isolates fromtribes and aboriginal populations isolated from the mainstream of human movement for thousands of years.36 Intrapatient variability is usually <0.5%, a value significantly lower than that observed with lentiviruses, such as HIV-1. The intrapatient variation of the HIV-1 envelope protein 5 years post infection is greater than that of all HTLV-1 envelope proteins to date.37 Even antigenically stable RNA viruses, such as measles and hepatitis A virus, have, in this PCR era, shown more variability than expected. Thus HTLV-1, and its close or distant cousins, HTLV-2 and bovine leukemia virus (BLV) respectively, constitute remarkable exceptions, unique among the genetically promiscuous RNA viruses.

Figure 1 Schematic representation of HTLV-1 replication in vivo. Highcirculating proviral loads characterizeHTLV-1 infection

The non-malignant neurological disorder TSP/HAM is accompanied by very high proviral loads, sometimes approaching 1/5 peripheral blood mononuclear cells.38–43 Consequently, substantial proportions of the CD4 T cells in the periphery may be infected. Proviral loads during the long asymptomatic phase of infection are lower than in TSP/HAM, although they may be as high as 1/25 PBMCs.38–43 Here, there is evidence of an extensive replication, yet HTLV-1 is genetically more stable than any other RNA virus. HIV proviral loads in late-stage disease, AIDS, may reach up to 1/100 to 1/30 in the periphery. In other words HIV-1 also shows signs of extensive viral replication, although the maximum HIV-1 proviral load is lower than that observed for HTLV-1. Yet HIV-1 varies considerably, as one would expect.37 By contrast, HTLV-1 remains stable while displaying very high proviral loads. How is this possible? The misincorporation rate of HTLV-1 reverse transcriptase turns out to be on the lower side (7 × 10−6/base/cycle),44,45 while that of HIV-1 is on the upper side (3.5 × 10−5/base/cycle).46 Only one log discriminates between them. Of all the components underlying viral variation, the Figure 2 Distribution of HTLV-1 infected clones in two patients mutation rate is among the least important while the number with TSP/HAM (patients P1 and P2). DNA fromPBMCs was analyzed of successive cycles of replication is probably the most by IPCR. Each signal on the gel corresponds to a cluster of >100 provi- important.47 How can the virus remain stable while rep- ruses with the same flanking sequence, ie belonging to the same cellu- licating intensely? lar clone. Quadruplicate experiments permitted estimation of the abundance of cells within circulating clones, most abundant cells being the most frequently detected after four experiments. M, molecular weight marker; NC, negative control (adapted from Ref. 18). Persistent clonal expansion of infected cells, a way of life for HTLV-1 via mitosis rather than via reverse transcription. Analyzing the The paradoxical combination of an elevated proviral load pattern of HTLV-1 replication over time revealed that the Tax- with a remarkable genetic stability suggested that the virus driven expansion of T cells may persist for considerable per- replicated in concert with cell mitosis (Figure 1).19 This was iods of time, despite strong cellular immunity. Hence, most confirmed for all cases of HTLV-1 infection, notably asympto- circulating HTLV-1-positive clones persist over time. How- matic carriers, TSP/HAM and ATLL.13 In ATLL, tumor cell ever, the abundance of these persistent clones, ie the number clones were identified against a general background of oligo- of cells within clones, is shown to fluctuate over time in some or polyclonal expansion of infected, but non-transformed infected individuals, suggesting that the degree of infected T cells.14 The number of clones detected varied considerably cell proliferation (or elimination) varies over time. Interest- but remained between five and 60 clones per sample (Figure ingly, in asymptomatic carriers, infected cells accumulate 2). The fact that there may be hundreds or more clones within within circulating clones, since detected clones are more an individual clearly indicates that HTLV-1 does replicate via abundant in persons with longstanding infection.13 reverse transcription in vivo. However the bulk of the proviral The HTLV-1 proviral load of TSP/HAM patients may draw load is made up of the most abundant clones, which are rela- near 10% of PBMCs, or approximately 3 × 109 copies (the tively few (<10–15) in each sample. Consequently we con- proviral burden in secondary lymphoid organs is unknown). clude that HTLV-1 in vivo replication predominantly occurs Assuming a minimum of 10 HTLV-1-driven proliferating

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 28 clones and that proliferation is not unduly abnormal, ie the clones divide approximately every 24 h, it would be possible to attain such a proviral load after approximately 28 divisions. Even if the proviral load was 10-fold greater, only 31 divisions would be needed. Clearly, as only few HTLV-1 carriers ulti- mately develop non-malignant disease after 20–40 years of infection, clonal expansion must be actively restrained, other- wise CD4 lymphocytosis would be obvious within weeks post infection. As clonal expansion is held in check, the question arises of how. Anti-HTLV-1 cellular immunity is the obvious explanation. It is present in asymptomatic carriers48 and is parti- cularly intense among TSP/HAM patients49 in whomthe proviral load is much greater than in the former group. Indeed, anti-tax activity is especially intense in TSP/HAM patients, easily detectable in direct CTL assays. Although tax expression directly induces the cell to proliferate and up-regulates a number of cytokines and transcription factors, it is antigenically Figure 3 Schematic representation of HTLV-1 replication over non-self, which is detrimental to clonal survival. time. The gray curve represents the early and transient step of vertical replication through reverse transcription and integration that allows allogeneic cells to infect the new organism. Black curves represent the persistent clonal expansion of newly infected cells. The degree of Two-step nature of HTLV-1 primary infection in vivo expansion varies between clones and/or between infected organisms. As there is no evidence for cell-free virus in vivo, HTLV-1 transmission must involve infected allogeneic cells. Reverse HTLV-1-infected individuals with no malignancy. Indeed, the transcription must follow soon, as allogeneic cells are ephem- number of infected cells within circulating clones ranges from eral. However, once a few rounds of reverse transcription are 1/3000 to у1/150 PBMCs. As proviral loads are higher in over, the clonal expansion of infected T cells predominates. TSP/HAM than in asymptomatic carriers, both the mean num- The squirrel monkey (Saimiri sciureus) constitutes an interest- ber of clones and the abundance of cells within theses clones ing model for investigating the tempo of HTLV-1 replication have been found to be significantly higher in TSP/HAM. during primary infection. After experimental infection of squir- In the spinal cord of patients with HAM/TSP, infiltrating rel monkeys (Saimiri sciureus) with HTLV-1 infected cells, the CD4-positive lymphocytes seem to be the major reservoir for virus appeared to be transcribed only transiently into the ani- the virus.54 This is thought to play a central role in the patho- mal circulating blood, spleen and lymph nodes.50 A stable dis- genesis of the disease. An oligoclonal proliferation of these appearance of the viral expression occured at 2–3 weeks from HTLV-1-infected CD4 lymphoid T cells is present in the cer- inoculation, coinciding with the development of the anti- ebrospinal fluid of the patients.55 Furthermore, clonal popu- HTLV-1 immune response and the persistent detection of the lations of HTLV-1-infected lymphocytes sharing identical provirus in PBMCs. Analyzing the HTLV-1 replication pattern HTLV-1 proviral flanking sequences (ie integration sites in the over time in PBMCs and various organs from HTLV-1 exper- cellular DNA) and thus derived froma single HTLV-1-infected imentally infected monkeys revealed that PBMCs and lymph- progenitor, are detected, for a given patient, in both the cer- oid organs constitute the major reservoirs for HTLV-1.50–53 A ebrospinal fluid and the peripheral blood. This indicates that pattern of persistent clonal expansion of infected cells, ident- HTLV-1 crosses the blood–brain barrier by way of the ical to that observed in humans, was evidenced in the PBMCs migration of HTLV-1-infected lymphocytes in vivo. Similarly, of experimentally infected animals.51 The dissemination of the the dissemination of the virus throughout the different body virus throughout the different body compartments appeared to compartments has been evidenced in humans as in animal result fromthe cellular transport of the integrated provirus. models (Figure 4). The circulating proviral burden increased as a function of time Monitoring the clonality of malignant cells over time in in one animal studied over a period of 4 years. The high pro- patients treated for ATLL has been found to be a useful tool to viral loads observed in the last samples resulted from the assess treatment efficacy.14,56 Figure 5 represents the clonality accumulation of infected cells via the extensive proliferation pattern of HTLV-1-infected cells over time in a 60-year-old of a restricted number of persistent clones on a background of woman treated for smoldering ATLL with skin infiltration. polyclonally expanded HTLV-1-positive cells. Therefore, the Southern blotting showed that this infiltration corresponded to animal model revealed that HTLV-1 primary infection would tumor cells bearing a single HTLV provirus. Local chemo- be a two-step process that includes a transient phase of therapy induced a near complete remission of skin lesions. A reverse transcription followed by the persistent multiplication second skin biopsy of a remaining lesion was performed. Fig- of infected cells. As represented in Figure 3, the degree of ure 5 shows the HTLV-1 integration patterns of both samples. T cell proliferation may vary between animals. These results A single, intense band at ~365 bp dominated the tumor suggest that when attempting to block HTLV-1 replication the samples. Nonetheless, there were 38 additional bands of choice of the target might depend on the stage of infection. lower frequency, some of which only came up once. The number of detectable bands in the post-therapy sample was Pathogenic and clinical implications of HTLV-1-infected T much lower. Only nine were detected, although the ~365 bp cell proliferation band, presumably corresponding to the tumor cells, was found in 3/4 samples, suggesting a frequency of ~1/300 of all Although flower cells, typical of ATLL, could be detected on cells recovered. Six courses of systemic chemotherapy blood smears, hyperlymphocytosis is a very rare event in resulted in complete clinical remission. A PBMC sample taken

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 29

Figure 4 Quadruplicate linker-mediated PCR analysis of HTLV-1 integration sites in DNA samples derived from a cell line and from various organs of an experimentally infected squirrel monkey. DNA was analyzed by quadruplicate LMPCR. Ampliﬁed products were submitted to run- off analysis with an HTLV-1 3Ј LTR speciﬁc oligonucleotide. Run-off products were resolved on a sequencing gel. M, molecular weight marker. E1540 cell line was used for experimental infection. Note that common clones are present in different body compartments (adapted from Ref. 51).

Figure 5 Run-off analysis of IPCR products of tumor cell DNA from skin biopsies of a patient with ATLL (A) and after local chemotherapy (B). Lane C: analysis of IPCR products derived from the PBMC sample taken 6 months after the second skin biopsy. After local chemotherapy, the number of clones detected decreased from 39 to nine. The major integration site, identiﬁed by a horizontal arrow, was detected 4/4 times in the ﬁrst sample; 2/4 replicas were still detected in the second biopsy and third sample, respectively (adapted from Ref. 14).

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 30 at remission was analyzed, revealing clonally expanding ively. Neither pX-I nor pX-II proteins are required for virus HTLV-1 T cells. The ~365 bp band was also present at a fre- replication in vitro.67 However, both are important in vivo, quency of 4/4 samples or у1/150 PBMCs. since the HTLV-1 p12I protein and the Tof protein of HTLV- Analyzing additional patients under treatment revealed that 2 are required for the establishment of a persistent infection the persistence of tumor clones at high frequencies (>1/300 in rabbits.68,69 In humans, circulating specific pX-I and pX- PBMCs) was commonly observed, even in complete II CD8+ T lymphocytes are regularly observed, whatever the responders. This persistence was invariably correlated with clinical status.70 This indicates that pX-I and pX-II are synthe- relapse and/or poor outcome.56 A fluctuating frequency of sized in vivo. Recent works have demonstrated that by activat- some tumor clones was observed, with evidence for clonal ing nuclear factor of activated T cells, p12I is critical for viral change under treatment in one patient, indicating that treat- infectivity in quiescent primary lymphocytes.71–73 ment of ATLL can result in the selection of resistant clones. Finally, allogeneic bone marrow transplantation (BMT) using an HTLV-1-infected sibling as donor was found to be associa- Somatic mutations of the provirus ted with the long-lasting disappearance of tumor clones and a possible cure of the disease. A long-termpersistent clonal The data summarized above indicate that in vivo, HTLV-1 rep- expansion of circulating HTLV-1 bearing T cells, deriving from licates mainly through persistent host cell proliferation in a the donor bone marrow, was evidenced in this patient. Data context of Tax-induced genetic instability. In a recent work, indicate that the variable success of ATLL treatments is prob- we investigated whether this could account for HTLV-1 gen- ably due to the clonal heterogeneity of malignant cells that etic variability in vivo.18 To this end, we used a strategy able results in the selection of resistant clones. Semi-quantitative to distinguish reverse transcriptase (RT)-associated mutations assessment of infected T cell clonality should be able to pre- fromsomaticsubstitutions. Figure 6 represents two clones of dict treatment failure. Accordingly, additional therapy may be HTLV-1-infected cells. All the cells of a given clone derive tailored to the clonality pattern observed after first-line froma single infected progenitor and therefore share the same therapy. integration site. As there is no preferential target site for HTLV- 1 integration in vivo, the characterization of an HTLV-1 integration site can be used as the specific and unique signature Tax triggers T cell proliferation and genetic instability of a given clone. Consequently, the comparison of HTLV-1 proviral sequences bordered by different flanking sequences There is ample literature documenting the influence of the reveals RT-associated mutations, while the comparison of HTLV-1 Tax protein on the up-regulation of transcription fac- HTLV-1 proviral sequences sharing the same integration site, tors and lymphokines and their receptors (for recent review, ie belonging to the same clone, reveals somatic mutations. see Ref. 57). In a recent work, Hanon et al58 demonstrated An inverse polymerase chain reaction (IPCR) strategy was that Tax expression was the normfor infected lymphocytes designed to distinguish somatic mutations from RT-associated that derived frominfected individuals without malignancy. substitutions (Figure 7). The proviral sequences were isolated, Therefore, the pleiotropic function of Tax is thought to cooper- together with flanking cellular sequences. A total of 208 ate in promoting the proliferation of infected T cell. Indeed, clones encompassing the 3Ј RU5 sequences of HTLV-1 (379 Tax protein induces the expression of numerous genes bp, ~80 kb of data) and their flanking cellular sequences involved in the differentiation and the proliferation of T (mean, 119 bp; range 7 to 362 bp, ~23 kb of data) were cells.59,60 Through the functional inactivation of P16INKA4,61 sequenced. Sequences obtained could be arranged into 29 and the activation of cyclin D262 and cyclin D3,63 Tax inter- cellular clones, depending on their cellular flanking venes directly in the pathway controlling T cell proliferation. sequences. For 60% of the clones, 8% to 80% of infected cells Furthermore, the N terminus of Tax has recently been found harbored a mutated HTLV-1 provirus, without evidence of to directly and specifically interact with CDK4, resulting in a reverse transcription-associated mutations. Typical results of Tax/CDK holoenzyme complex that capably phosphorylates sequence alignment corresponding to a DNA sample derived the Rb protein.64 froma patient with TSP/HAM are shown in Figure 8. Overall, In addition to its positive effect on cell cycling, Tax protein not all RU5 sequences derived fromthe samecellular clone negatively interferes with some DNA repair functions of the host cells. Indeed, Tax inactivates in trans the promoter of the human ␤ polymerase22 and disrupts other prominent cellular DNA repair pathways.24,26 By competing with p53 for p300/CBP binding, Tax influences the transition fromG1 to S phase and impairs the activity of the DNA-damage sentinel at this junction.64,65 Furthermore, Tax functionally inhibits the human mitotic checkpoint protein MAD1.23 These functional characteristics of Tax may explain its mutagenic effect on cellular chromosomal DNA, as recently evidenced in vitro.25 In addition to Tax, the more recently described accessory proteins of HTLV-1 might also account for the pathogenesis of the infection.66 Indeed, open reading frames (ORFs) pX-I and pX-II encode proteins produced fromboth single- and double-spliced transcripts. The double-spliced pX-I and pX-II transcripts encode Rof and Tof proteins, respectively, whereas I Figure 6 RT-associated mutation versus somatic mutation. Two the single-spliced pX-I and pX-II RNAs encode the p12 pro- clones of HTLV-1-infected cells are represented. Bottom: two 3Ј II tein consisting of the last 98 residues of Rof, and the p13 extremities of the HTLV-1 provirus flanked by their integration sites protein corresponding to the last 87 residues of Tof, respect- (see text for details).

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 31

Figure 7 LMPCR and IPCR analyses of HTLV-1 integration and clonal expansion.

Figure 8 Somatic mutations of the HTLV-1 3Ј RU5 sequence in vivo in a patient with TSP/HAM. Seven distinct HTLV-1 3Ј integration sites were isolated fromthis patient; the corresponding ﬂanking cellular sequence hexamericrepeats, generated during integration, are given on the right. RU5 sequences were aligned according to patient RU5 sequence. Deletions are denoted by a dash (-). Coordinates of the sequence are those of the ATK-1 sequence.35 Each cluster of RU5 sequences sharing a common integration site, and therefore belonging to a unique clone of expanded T cells, is identiﬁed by its cellular clone number. For each cellular clone, the number of non-unique RU5 consensus sequences is indicated between brackets (adapted fromRef. 18).

were identical – 18 of 29 clones (~60%) were not homo- above indicates that those mutations did not correspond to geneous. The number of variants per RU5 sequence ranged minus-strand synthesis-associated RT errors, which are from0 to 4. A total of 38 single-base substitutions and eight expected to result in a homogeneous population of RU5 single-base deletions were scored. Some sequences harbored sequences (Figure 9B). Similarly, the presence of at least one up to 3–4 mutations. patient consensus sequence within each of the 29 clones dem- Figure 9 summarizes the different steps at which a mutation onstrated that none of the 46 RU5 substitutions corresponded can occur during the synthesis of HTLV-1 provirus. Accord- to a plus-strand synthesis-associated mutation corrected ingly, the distribution of the 46 RU5 mutations described before the newly infected host cell divided. These substi-

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 32

Figure 9 HTLV-1 3Ј RU5 sequence mutations at different stages of HTLV-1 replication. Different steps in HTLV-1 RU5 sequence replication that may affect retroviral mutation rates are illustrated. (A) Reverse tanscription and clonal expansion without mutation. (B) Reverse transcriptase (RT)-associated substitution during minus-strand synthesis. The plain circle represents a mutation that has occurred during the synthesis of the 5Ј RU5 of the minus strand. This substitution appears to be harbored by both strands of the two LTR of the integrated provirus. Accordingly, all infected cells from the corresponding clone (identified by their common integration site) harbor a provirus with the same substitution. As a consequence, all the sequences from this clone obtained after inverse PCR harbor the same mutation at the same position. (C) RT-associated substitution during plus-strand 3Ј RU5 synthesis, in the absence of DNA mismatch repair before the first division of the newly infected CD4+ T cell. Such substitution is harbored by only one strand of the integrated provirus (black square). Accordingly, after the first mitosis and in the absence of DNA repair (such repair being expected to result in the absence of mutation or in a pattern identical to that of (B)), such substitution results into two populations of infected cells: one harbors the RU5 substitution, the other one harbors the native RU5 sequence. As a consequence, two sequence populations are obtained after inverse PCR (line 10). (D) Somatic mutation of the 3Ј RU5 sequence in the absence of RT-associated substitution. In this case, no substitution occurs during provirus synthesis. After integration, a 3Ј RU5 substitution that has occurred during the second mitosis is represented at the bottom. As is the case for plus-strand substitution, such somatic mutation results in a double population of RU5 sequences. The distinction between somatic mutations and plus-strand-associated substitutions is explained in the text.

tutions, all of which were harbored by only a subset of that consist of somatic mutations of the proviral sequence sequences within clones, might have theoretically resulted occurring during clonal expansion. from somatic mutations, or from RT-associated substitutions occurring during a single cycle of the plus-strand synthesis of the provirus, in the absence of DNA mismatch repair before Consequences of somatic mutations for the virus the first division of the newly infected CD4+ T cell (Figure 9). However, two aspects of our results rule out the second Mutations were found to be generally distributed in a random possibility. First, the actual RU5 mutation frequency is incom- manner across the HTLV-1 RU5 region, with no evidence of patible with those RT-associated errors. By using a HTLV-1 hot spots. The rex responsive element (RXRE) is a cis-acting single cycle replication assay for mutation detection, the in RNA element required for Rex function that maps to the U3R vivo mutation rate for HTLV-1 was determined by Mansky et region of the LTR.74–78 HTLV-1 Rex protein acts at the post- al to be 7 × 10−6 mutations per target base pair per replication transcriptional level to induce the appearance of unspliced cycle.44 Indeed, if the mutations shown in Figure 8 corre- and singly spliced viral mRNA in the cytoplasm.79 Rex action sponded to plus-strand synthesis-associated errors, they would requires both the overall secondary structure intrinsic to the necessarily be the result of a single cycle of reverse transcrip- RXRE and specific sequences fromone smallregion of this tion. Accordingly, the average RU5 mutation frequency was large structure74,80,81 which forms a binding site for the pro- ~4.2 × 10−3 substitution per replication cycle per base (46 tein. Mutational analyses of RXRE have demonstrated that in distinct substitutions in 29 RU5 sequences of 379 bp), which vitro Rex binding correlates with in vivo Rex function.76,81 is about 600 times higher than for HTLV-1 RT44 and 100 times Of the 46 mutations, 22 mapped to 21 sites in the RXRE. higher than for HIV RT.46 Second, some RU5 sequences harb- As can be seen from Figure 10A, some mapped to the Rex ored two to four substitutions, a mutation frequency not binding site or else disrupted RNA secondary structures, as observed to date for a single step of RT-mediated DNA elong- evidenced by de novo calculations using the variant ation. Therefore, it appears that most substitutions are not due sequences (Figure 10B). As RXRE plays a crucial role in the to plus-strand synthesis-associated RT errors nor to PCR arti- expression of HTLV-1, it is likely that a number of these vari- facts or recombinations. They result from somatic mutations ants will have impaired the expression of viral structural pro- during cellular replication. These data indicate that HTLV-1 teins. It will be interesting to confirmthese results by testing in vivo variation results mainly from post-integration events the efficiency of present RXRE sequences for exporting mRNA.

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 33

Figure 10 Somatic mutations within the RXRE secondary structure. (a) RXRE corresponds to consensus sequence of patient P1 (see Figure 8). Position 1 corresponds to position 8615 of the ATK sequence.35 The three Rex binding motifs are boxed74–78 (adapted fromRef. 18). The Figure represents the 22 substitutions (del, deletion). (b) Mutated RXRE RNA secondary structures, as predicted by de novo calculations with ESSA-SAPSSARN software101,102 using the variant sequences.

Somatic mutation of cellular sequences are mutation frequencies, not mutation rates, as it is not possible to estimate the average number of mitoses in any lineage. Frequent somatic mutations were also found in the flanking If the same mutation frequency applied across the whole gen- sequences, suggesting that both the provirus and the cellular ome, one would predict a remarkable mutation load of ~1.7 genome undergo the same process of genetic instability. A × 106 mutations per diploid cell (~2.8 × 10−4–6 × 109), or total of seven substitutions were found in ~24 kb of cellular approximately one mutation per 3.5 kb. DNA, representing a frequency of 2.8 × 10−4 per bp A cellular clone derived froma TSP/HAM sampleresulted sequenced. This is remarkably similar to results obtained for in five sequences; the proviral flanking sequences showed vir- the HTLV-1 RU5 region (5.8 × 10−4 per bp sequenced), parti- tual identity with the promoter region of the human ␣-, or cularly in light of the very different base composition of the non-neuronal, enolase gene.82 Proviral integration occurred two regions (HTLV-1 40% AT, cellular 59% AT). These values 146 bp 5Ј to the TATA box so uncoupling it froma number

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 34 of transcriptional motifs such as CCAAT box, AP1 and PEA2 sites. All five sequences harbored a T→C transition at position 1217 with respect to the published sequence, while one car- ried two additional purine–purine transitions. The T1217C transition was found to be restricted to clone P1-7, indicating that it represented a somatic mutation. PBMC DNA from this patient was amplified on two occasions, 4 and 6 years later. The clone with the T1217C substitution was identified on both occasions, indicating that the somatically mutated clone persisted over time.

Somatic mutation frequency is proportionate to the degree of infected T cell proliferation

It has been clearly established that genomic instability is characteristic of some cancers.1 However, somatic mutations are not restricted to ATLL samples. When a somatic mutation is associated with DNA replication, its extent generally depends on the number of rounds of mitosis, which may be reflected by the clonal frequency in vivo. It is possible to get an approximate estimate of clonal frequencies from quadruplicate inverse PCR.14,83 Figure 11a shows that the average number of mutations per kilobase in 57 RU5 sequences belonging to six HTLV-1 clones with a detection frequency of Ն1 per 150 PBMCs was more than twice that of the remaining 151 sequences derived from23 clones with a detection frequency of <1 per 150 PBMCs. This difference was statistically significant (P = 0.037). Indeed, since the frequency of abundant circulating clones varies in the order ATLL > TSP/HAM > asymptomatic carriers, the number of acquired somatic mutations was higher in HTLV-1-associated disease than in virus carriers. As shown in Figure 11b, PBMC DNA from ATLL, TSP/HAM, and asymptomatic carriers harbored a mutation frequency of 0.71, 0.52 and 0.22 substitution per kb of sequence (RU5 sequences plus integration sites), respectively. The difference between asymptomatic carriers and ATLL patients was statistically significant (P = 0.048). However, the frequency of somatic mutation in the 45 sequences derived fromthe six non-malignantATLL clones was identical to that of TSP/HAM sequences: 0.5 substitution/kb of sequence (RU5 sequences plus integration sites). Figure 11 Increased somatic mutations in RU5 and flanking sequences associated with disease. (a) Frequency of somatic mutations in RU5 sequences according to the abundance of circulating clones. The detection frequency of circulating HTLV-1 positive HTLV-1 infection as a model of premalignant condition clones was estimated by quadruplicate inverse PCR analysis (Figure 2). (b) Distribution of the frequency of somatic mutations along both Figure 12 summarizes the data detailed above. The link the RU5 and flanking sequences according to the clinical status (from between mutation frequency and the degree of infected T cell Ref. 18). proliferation fits well with a somatic process. As shown in the figure, both result fromthe effect of Tax, which is also the main target for the robust cellular immune response that characterizes HTLV-1 infection in vivo. The proliferation of infected cells increases the proviral load and allows the virus to disseminate through body compartments such as the CNS in the case of TSP/HAM. Somatic mutations account for HTLV-1 genetic variability and infected cell genetic instability. HTLV-1 variations might modify Tax epitope and thereby gen- erate mutants with a proliferative advantage. Such mutants were recently isolated in the tumor DNA from ATLL patients.84 Cellular genetic instability constitutes a propitious background for the accumulation of critical mutations leading to malignancy. The proposed model implies that any factor impairing cellular immune response triggers both infected T cell proliferation and genetic instability, thereby increasing the risk of developing ATLL. This is supported by clinical observation Figure 12 Pathogenesis of HTLV-1 replication in vivo.

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 35 and experimental investigation showing that the suppression ible under effective treatment of Ss infection. This will help of HTLV-1-speciﬁc cellular immune response leads to the conﬁrmthat the down-regulation of both HTLV-1 proviral development of ATLL.85–89 In a more general manner, any fac- load and infected T cell proliferation under treatment results tor that stimulates T cell proliferation in asymptomatic carriers fromparasite clearance. Validating this hypothesis will allow may constitute a cofactor of ATLL. Recent works suggest that strongyloidiasis treatment as an effective prophylaxis of ATLL Strongyloides stercoralis might be one of these factors. in HTLV-1 carriers.

Strongyloides stercoralis as a cofactor of ATLL Conclusions and perspectives

Strongyloides stercoralis (Ss) causes strongyloidiasis, a During normal somatic cell mitosis, the parental cell dupli- chronic, usually asymptomatic, gastrointestinal infection,90,91 cates its genome, containing approximately 6 × 109 nucleo- found in the same endemic area as HTLV-1.92 The incubation tides, then partitions its DNA equally between the two newly period that precedes ATLL is significantly shortened in HTLV- generated daughter cells. Errors occurring in the course of 1 carriers with Ss infection, suggesting that Ss could be a these processes result in mutations, which is infrequent since cofactor of ATLL. We recently investigated the effect of Ss normal human somatic cells accurately replicate their DNA infection on infected T lymphocytes in HTLV-1 asymptomatic every time they divide.1 The overall mutation rate of somatic carriers in vivo. After real-time quantitative PCR, the mean human cells has been estimated at 1.4 × 10−10 circulating HTLV-1 proviral load was more than five-fold nucleotides/cell/division or 2.0 × 10−7 mutations/gene/cell higher in HTLV-1 carriers with strongyloidiasis than in HTLV- division.97 The mean mutation frequency of 106 per HTLV-1 1+ individuals without Ss infection (P < 0.009). This increased infected lymphocytes18 might account for the oncogenic proviral load was found to result fromthe extensive prolifer- potential of the virus. Yet, a conundrumremains:given the ation of a restricted number of infected clones, ie from oligo- mutational pressure of elevated levels of somatic mutations, clonal expansion, as evidenced by the semiquantitative ampli- why is the lifetime risk of ATLL as low as 3–5%? The fication of HTLV-1 flanking sequences. The positive effect of expression of HTLV-1 proteins would mark the infected cell Ss on clonal expansion was reversible under effective treat- as non-self. Besides, the frequency of neo-antigen formation ment of strongyloidiasis in one patient with parasitological could be higher than usually found in other malignancies. cure, whereas no significant modification of the HTLV-1 repli- Both features might allow robust control by host cell-mediated cation pattern was observed in an additional case with stron- immunity. As detailed above, suppressing the anti-HTLV-1 gyloidiasis treatment failure. Therefore, Ss stimulates the oligo- cellular immune response contributes to the development of clonal proliferation of HTLV-1-infected cells in HTLV-1 ATLL, whereas restoring it might have a curative effect in ani- asymptomatic carriers in vivo. Since the degree of infected mal models,98 as in humans.56 How could somatic mutations T cell proliferation correlates with the frequency of somatic help an infected cell escape CTL response? The generation of mutations, this finding is thought to account for the increased Tax escape mutants is the first explanation that comes to mind. risk of ATLL and the shortened period of latency observed in The expression of HTLV-1 proteins marks out the cell as non- patients with strongyloidiasis. self and as a target for cell-mediated immunity, which is parti- The positive effect of a parasitic infection on the develop- cularly intense.49 Only deletions of single nucleotides were ment of a lymphotropic virus-associated lymphoma is remi- noted. However, if they occurred in the viral open reading niscent of the well-known synergetic association between frames they would considerably limit protein expression. Epstein–Barr virus (EBV) infection and malaria in the genesis Negative selection then ensures the maintenance of the Tax+ of Burkitt lymphoma (BL).93,94 Just as ATLL and HTLV-1, BL phenotype. As Tax is a target for cytotoxic T lymphocytes, has a complex, multistep pathogenesis, in which EBV is the some mutations in HLA-class I-restricted CTL epitope are best documented, but almost certainly not the only, contribu- likely to be compatible with Tax function, thus allowing the ting factor. Epidemiologically, hyperendemic malaria cellular clone to escape cell-mediated immunity.99 This is increases the incidence of BL by a factor of 20 in Africa.93 possible as long as most other HTLV-1 protein sequences are Concerning HTLV-1, the incidence of strongyloidiasis seems inactivated first, as revealed by the observation that cases of to be higher in ATLL patients than in asymptomatic HTLV-1- CTL escape seem to occur only when high cellular immune infected individuals. In Martinique, the French West Indies, responses are focused on a single target.100 Indeed, tax the prevalence of Ss infection in the general population aver- sequences corresponding to escape mutants have been ages 2%, whereas 20% of Ss-infected individuals are coin- observed in the tumor DNA from ATLL patients,84 suggesting fected by HTLV-1.95 In the same area, the prevalence of Ss that, in these cases, the tumor clone would have derived from infection in ATLL patients is 42% versus 23% in patients with an infected cell having escaped CTL response. In addition to other T cell lymphomas.96 Pathogenically, EBV-associated the generation of Tax escape mutants, the huge excess of lymphomagenesis is assumed to result from the malaria- somatic cellular mutations might alter the expression of genes associated chronic stimulation of germinal center cell prolifer- involved in antigen presentation. Indeed, one can speculate ation, together with a possible acquired EBV-specific T cell that the CTL pressure can select clonally expanded cells with immune deficiency. In the case of ATLL, present results show acquired mutations impairing the proteasome, the transport of that Ss increases the proliferation of HTLV-1 infected T cells the peptide or MHC–peptide interactions. Future studies allow by a mechanism which remains to be elucidated. exploration of this question. Present data emphasize the need to assess the effect of other endemic pathogens on the clonality pattern of HTLV-1- infected cells, especially in areas where ATLL incubation per- Acknowledgements iod is short. Strongyloidiasis is asymptomatic in about 50% of the cases.90,91 Future experiments are needed in order to con- This work was supported by grants fromthe Association pour firmthat the positive effect of Ss on clonal expansion is revers- la Recherche sur le Cancer, the Fondation Contre la Leuce´-

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 36 mie, the Comite´sDe´partementaux du Rhoˆne, de la Saone et 17 Furukawa Y, Fujisawa J, Osame M, Toita M, Sonoda S, Kubota Loire, de l’Arde`che et de la Savoie de la Ligue Nationale Con- R, Ijichi S, Yoshida M. Frequent clonal proliferation of human T- tre le Cancer, and from the Programme de Recherche Fonda- cell leukemia virus type 1 (HTLV-1)-infected T cells in HTLV-1- associated myelopathy (HAM-TSP). Blood 1992; 80: 1012–1016. mentale en Microbiologie et Maladies Infectieuses et Parasit- 18 Mortreux F, Leclercq I, Gabet A, Leroy A, Westhof E, Gessain aires of the French Ministry of Health. FM was the recipient A, Wain-Hobson S, Wattel E. Somatic mutation in human T-cell of bursaries fromthe Ministe`re de l’Enseignement Supe´rieur leukemia virus type 1 provirus and flanking cellular sequences et de la Recherche, fromthe Fondation pour la Recherche during clonal expansion in vivo. J Natl Cancer Inst 2001; 93: Me´dicale and fromthe Fond National pour la recherche scien- 367–377. tifique (Belgium). ASG is supported by a grant from the Centre 19 Wattel E, Vartanian JP, Pannetier C, Wain-Hobson S. Clonal expansion of human T-cell leukemia virus type I-infected cells León Be´rard. We thank Christiane Pinatel, Agnes Lançon, in asymptomatic and symptomatic carriers without malignancy. Ce´line Paul and Marie-Dominique Reynaud for assistance. J Virol 1995; 69: 2863–2668. 20 Wattel E, Cavrois M, Gessain A, Wain-Hobson S. Clonal expansion of infected cells – a way of life for HTLV-1. J Acquir Immune References Defic Syndr Hum Retrovir 1996; 13 (Suppl. 1): 92–99. 21 Wattel E. Proliferation of HTLV-1 infected cells in vivo: patho- 1 Loeb LA. Cancer cells exhibit a mutator phenotype. Adv Cancer genic implications in leukemogenesis and neuropathogenesis. Res 1998; 72: 25–56. ABI Professional Publications: Arlington, 1999. 2 Poiesz BJ, Ruscetti FW, Gazdar AF, Bunn PA, Minna JD, Gallo 22 Jeang KT, Widen SG, Semmes OT, Wilson SH. HTLV-I trans-acti- RC. Detection and isolation of type C retrovirus particles from vator protein, tax, is a trans-repressor of the human beta-poly- fresh and cultured lymphocytes of a patient with cutaneous T- merase gene. Science 1990; 247: 1082–1084. cell lymphoma. Proc Natl Acad Sci USA 1980; 77: 7415–7421. 23 Jin DY, Spencer F, Jeang KT. Human T cell leukemia virus type 3 Yoshida M, Mioshi I, Hinuma Y. Isolation and characterisation 1 oncoprotein tax targets the human mitotic checkpoint protein of a retrovirus from cell lines of human adult T-cell leukemia and MAD1. Cell 1998; 93: 81–91. its implication in the disease. Proc Natl Acad Sci USA 1982; 79: 24 Kao SY, Marriott SJ. Disruption of nucleotide excision repair by 2031–2040. the human T-cell leukemia virus type 1 Tax protein. J Virol 1999; 4 Gessain A, Barin F, Vernant JC, Gout O, Maurs L, Calender A, 73: 4299–4304. de The´ G. Antibodies to human T-lymphotropic virus type-I in 25 Miyake H, Suzuki T, Hirai H, Yoshida M. Trans-activator Tax of patients with tropical spastic paraparesis. Lancet 1985; 2: 407– human T-cell leukemia virus type 1 enhances mutation fre- 410. quency of the cellular genome. Virology 1999; 253: 155–161. 5 LaGrenade L, Hanchard B, Fletcher V, Cranston B, Blattner W. 26 Philpott SM, Buehring GC. Defective DNA repair in cells with Infective dermatitis of Jamaican children: a marker for HTLV-I human T-cell leukemia/bovine leukemia viruses: role of tax gene. infection. Lancet 1990; 336: 1345–1347. J Natl Cancer Inst 1999; 91: 933–942. 6 Mattos K, Queiroz C, Pecanha-Martins AC, Publio L, Vinhas V, 27 PiseMasison CA, Choi KS, Radonovich M, Dittmer J, Kim SJ, Melo A. Lymphocyte alveolitis in HAM/TSP patients. Preliminary Brady JN. Inhibition of p53 transactivation function by the human report. Arq Neuropsiquiatr 1993; 51: 134–136. 7 Mochizuki M, Tajima K, Watanabe T, Yamaguchi K. Human T T-cell lymphotropic virus type 1 Tax protein. J Virol 1998; 72: lymphotropic virus type 1 uveitis. Br J Ophthalmol 1994; 78: 1165–1170. 149–154. 28 Ressler S, Morris GF, Marriott SJ. Human T-cell leukemia virus 8 Sato K, Maruyama I, Maruyama Y, Kitajima I, Nakajima Y, Higaki type 1 Tax transactivates the human proliferating cell nuclear M, Yamamoto K, Miyasaka N, Osame M, Nishioka K. Arthritis antigen promoter. J Virol 1997; 71: 1181–1190. in patients infected with human T lymphotropic virus type I. 29 Majone F, Jeang KT. Clastogenic effect of the human T-cell leuke- Clinical and immunopathologic features. Arthritis Rheum 1991; mia virus type I Tax oncoprotein correlates with unstabilized 34: 714–721. DNA breaks. J Biol Chem 2000; 275: 32906–32910. 9 Sherman MP, Amin RM, Rodgers-Johnson PE, Morgan OS, Char 30 Uittenbogaard MN, Giebler HA, Reisman D, Nyborg JK. Tran- G, Mora CA, Iannone R, Collins GH, Papsidero L, Gibbs CJ Jr, scriptional repression of p53 by human T-cell leukemia virus type Poiesz BJ. Identification of human T cell leukemia/lymphoma I Tax protein. J Biol Chem 1995; 270: 28503–28506. virus type I antibodies, DNA, and protein in patients with poly- 31 Domingo E, Holland JJ. Complications of RNA heterogeneity for myositis. Arthritis Rheum 1995; 38: 690–698. the engineering of virus vaccines and antiviral agents. In: Setlow 10 Takatsuki K, Yamaguchi K, Kawano F, Hattori T, Nishimura H, JK (ed.). Genetic Engineering, Principles and Methods. Plenum Tsuda H, Sanada I, Nakada K, Itai Y. Clinical diversity in adult Press: New York, 1992, pp 13–31. T-cell leukemia–lymphoma. Cancer Res 1985; 45: 4644S–4645S. 32 Holland JJ, de le Torre JC, Steinhauer DA. RNA virus populations 11 Shimoyama M, Kagami Y, Shimotohno K, Miwa M, Minato K, as quasispecies. Curr Top Microbiol Immunol 1992; 176: 1–20. Tobinai K, Suemasu K, Sugimura T. Adult T-cell 33 Eigen M. Viral quasispecies. Sci Am 1993; 269: 42–49. leukemia/lymphoma not associated with human T-cell leukemia 34 Drake JW, Holland JJ. Mutation rates among RNA viruses. Proc virus type I. Proc Natl Acad Sci USA 1986; 83: 4524–4528. Natl Acad Sci USA 1999; 96: 13910–13913. 12 Blattner WA. Human T-lymphotrophic viruses and diseases of 35 Seiki M, Hattori S, Hirayama Y, Yoshida M. Human adult T-cell long latency. Ann Intern Med 1989; 111: 4–6. leukemia virus: complete nucleotide sequence of the provirus 13 Cavrois M, Gessain A, Wain-Hobson S, Wattel E. Proliferation genome integrated in leukemia cell DNA. Proc Natl Acad Sci of HTLV-1 infected circulating cells in vivo in all asymptomatic USA 1983; 80: 3618–3622. carriers and patients with TSP/HAM. Oncogene 1996; 12: 36 Gessain A, Gallo RC, Franchini G. Low degree of human T-cell 2419–2423. leukemia/lymphoma virus type I genetic drift in vivo as a means 14 Cavrois M, Wain-Hobson S, Gessain A, Plumelle Y, Wattel E. of monitoring viral transmission and movement of ancient human Adult T-cell leukemia/lymphoma on a background of clonally populations. J Virol 1992; 66: 2288–2295. expanding human T-cell leukemia virus type-1-positive cells. 37 Pedroza Martins L, Chenciner N, Wain-Hobson S. Complex Blood 1996; 88: 4646–4650. intrapatient sequence variation in the V1 and V2 hypervariable 15 Cavrois M, Leclercq I, Gout O, Gessain A, Wain-Hobson S, Wat- regions of the HIV-1 gp 120 envelope sequence. Virology 1992; tel E. Persistent oligoclonal expansion of human T-cell leukemia 191: 837–845. virus type 1-infected circulating cells in patients with Tropical 38 Wattel E, Mariotti M, Agis F, Gordien E, Le Coeur FF, Prin L, spastic paraparesis/HTLV-1 associated myelopathy. Oncogene Rouger P, Chen IS, Wain-Hobson S, Lefrere JJ. Quantification of 1998; 17: 77–82. HTLV-1 proviral copy number in peripheral blood of symp- 16 Etoh K, Tamiya S, Yamaguchi K, Okayama A, Tsubouchi H, Ideta tomless carriers from the French West Indies. J Acquir Immune T, Mueller N, Takatsuki K, Matsuoka M. Persistent clonal prolifer- Defic Syndr 1992; 5: 943–946. ation of human T-lymphotropic virus type I-infected cells in vivo. 39 Albrecht B, Collins ND, Newbound GC, Ratner L, Lairmore MD. Cancer Res 1997; 57: 4862–4867. Quantification of human T-cell lymphotropic virus type 1 pro-

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 37 viral load by quantitative competitive polymerase chain reaction. for adult T-cell leukaemia/lymphoma (ATLL). Br J Haematol J Virol Meth 1998; 75: 123–140. 1999; 105: 743–751. 40 Kubota R, Umehara F, Izumo S, Ijichi S, Matsumuro K, Yashiki S, 57 Yoshida M. Multiple viral strategies of HTLV-1 for dysregulation Fujiyoshi T, Sonoda S, Osame M. HTLV-I proviral DNA amount of cell growth control. Annu Rev Immunol 2001; 19: 475–496. correlates with infiltrating CD4+ lymphocytes in the spinal cord 58 Hanon E, Hall S, Taylor GP, Saito M, Davis R, Tanaka Y, Usuku frompatients with HTLV-I-associated myelopathy. J Neuroimmu- K, Osame M, Weber JN, Bangham CR. Abundant tax protein nol 1994; 53: 23–29. expression in CD4+ T cells infected with human T- cell lympho- 41 Kira J, Koyanagi Y, Yamada T, Itoyama Y, Goto I, Yamamoto N, tropic virus type I (HTLV-I) is prevented by cytotoxic T lympho- Sasaki H, Sakaki Y. Increased HTLV-I proviral DNA in HTLV-I- cytes. Blood 2000; 95: 1386–1392. associated myelopathy: a quantitative polymerase chain reaction 59 Fujii M, Niki T, Mori T, Matsuda T, Matsui M, Nomura N, Seiki study (published erratumappears in Ann Neurol 1991; 29, 363). M. HTLV-1 Tax induces expression of various immediate early Ann Neurol 1991; 29: 194–201. serumresponsive genes. Oncogene 1991; 6: 1023–1029. 42 Nagai M, Usuku K, Matsumoto W, Kodama D, Takenouchi N, 60 Inoue J, Seiki M, Taniguchi T, Tsuru S, Yoshida M. Induction of Moritoyo T, Hashiguchi S, Ichinose M, BanghamCR, IzumoS, interleukin 2 receptor gene expression by p40x encoded by Osame M. Analysis of HTLV-I proviral load in 202 HAM/TSP human T-cell leukemia virus type 1. Embo J 1986; 5: 2883–2888. patients and 243 asymptomatic HTLV-I carriers: high proviral 61 Suzuki T, Kitao S, Matsushime H, Yoshida M. HTLV-1 Tax pro- load strongly predisposes to HAM/TSP. J Neurovirol 1998; 4: tein interacts with cyclin-dependent kinase inhibitor p16INK4A 586–593. and counteracts its inhibitory activity towards CDK4. Embo J 43 Gabet AS, Mortreux F, Talarmin A, Plumelle Y, Leclercq I, Leroy 1996; 15: 1607–1614. A, Gessain A, Clity E, Joubert M, Wattel E. High circulating pro- 62 Santiago F, Clark E, Chong S, Molina C, Mozafari F, Mahieux R, viral load with oligoclonal expansion of HTLV-1 bearing T cells Fujii M, Azimi N, Kashanchi F. Transcriptional up-regulation of in HTLV-1 carriers with strongyloidiasis. Oncogene 2000; 19: the cyclin D2 gene and acquisition of new cyclin-dependent kin- 4954–4960. ase partners in human T-cell leukemia virus type 1-infected cells. 44 Mansky LM. In vivo analysis of human T-cell leukemia virus type J Virol 1999; 73: 9917–9927. 1 reverse transcription accuracy. J Virol 2000; 74: 9525–9531. 63 Neuveut C, Low KG, Maldarelli F, Schmitt I, Majone F, 45 Mansky LM. Sorting out mutations in human T-cell leukemia Grassmann R, Jeang KT. Human T-cell leukemia virus type 1 Tax virus type 1 proviruses during in vivo clonal expansion. J Natl and cell cycle progression: role of cyclin D-cdk and p110Rb. Cancer Inst 2001; 93: 336–337. Mol Cell Biol 1998; 18: 3620–3632. 46 Mansky LM, Temin HM. Lower in vivo mutation rate of human 64 Haller K, Wu Y, Derow E, Schmitt I, Jeang KT, Grassmann R. immunodeficiency virus type 1 than that predicted from the fid- Physical interaction of human T-cell leukemia virus type 1 Tax elity of purified reverse transcriptase. J Virol 1995; 69: 5087– with cyclin-dependent kinase 4 stimulates the phosphorylation 5094. of retinoblastoma protein. Mol Cell Biol 2002; 22: 3327–3338. 47 Coffin JM. Genetic variation in avian retroviruses. Dev Biol Stand 65 Van PL, YimKW, Jin DY, Dapolito G, KurimasaA, Jeang KT. 1990; 72: 123–132. Genetic evidence of a role for ATM in functional interaction 48 Parker CE, Daenke S, Nightingale S, BanghamCR. Activated, between human T-cell leukemia virus type 1 Tax and p53. J Virol HTLV-1-specific cytotoxic T-lymphocytes are found in healthy 2001; 75: 396–407. seropositives as well as in patients with tropical spastic parapar- 66 Koralnik IJ, Gessain A, Klotman ME, Lo Monico A, Berneman esis. Virology 1992; 188: 628–636. ZN, Franchini G. Protein isoforms encoded by the pX region of 49 Jacobson S, Shida H, McFarlin DE, Fauci AS, Koenig S. Circulat- human T-cell leukemia/lymphotropic virus type I. Proc Natl Acad ing CD8+ cytotoxic T lymphocytes specific for HTLV-I pX in Sci USA 1992; 89: 8813–8817. patients with HTLV-I associated neurological disease. Nature 67 Derse D, Mikovits J, Ruscetti F. X-I and X-II open reading frames 1990; 348: 245–248. of HTLV-I are not required for virus replication or for immortaliz- 50 Kazanji M, Ureta-Vidal A, Ozden S, Tangy F, de Thoisy B, Fiette ation of primary T-cells in vitro. Virology 1997; 237: 123–128. L, Talarmin A, Gessain A, de The´ G. Lymphoid organs as a major 68 Cockerell GL, Rovnak J, Green PL, Chen IS. A deletion in the reservoir for human T-cell leukemia virus type 1 in experimen- proximal untranslated pX region of human T-cell leukemia virus tally infected squirrel monkeys (Saimiri sciureus): provirus type II decreases viral replication but not infectivity in vivo. expression, persistence, and humoral and cellular immune Blood 1996; 87: 1030–1035. responses. J Virol 2000; 74: 4860–4867. 69 Collins ND, Newbound GC, Albrecht B, Beard JL, Ratner L, Lair- 51 Mortreux F, Kazanji M, Gabet AS, de Thoisy B, Wattel E. Two- more MD. Selective ablation of human T-cell lymphotropic virus step nature of human T-cell leukemia virus type 1 replication in type 1 p12I reduces viral infectivity in vivo. Blood 1998; 91: experimentally infected squirrel monkeys (Saimiri sciureus). J 4701–4707. Virol 2001; 75: 1083–1089. 70 Pique C, Ureta-Vidal A, Gessain A, Chancerel B, Gout O, Tam- 52 Kazanji M, Moreau JP, Mahieux R, Bonnemains B, Bomford R, ouza R, Agis F, Dokhelar MC. Evidence for the chronic in vivo Gessain A, de The´ G. HTLV-I infection in squirrel monkeys production of human T cell leukemia virus type I Rof and Tof (Saimiri sciureus) using autologous, homologous, or heterologous proteins fromcytotoxic T lymphocytesdirected against viral pep- HTLV-I-transformed cell lines. Virology 1997; 231: 258–266. tides. J Exp Med 2000; 191: 567–572. 53 Kazanji M, Tartaglia J, Franchini G, de Thoisy B, Talarmin A, 71 Lairmore MD, Albrecht B, D’Souza C, Nisbet JW, Ding W, Bartoe Contamin H, Gessain A, de The´ G. Immunogenicity and protec- JT, Green PL, Zhang W. In vitro and in vivo functional analysis of tive efficacy of recombinant human T-cell leukemia/lymphoma human T cell lymphotropic virus type 1 pX open reading frames I virus type 1 NYVAC and naked DNA vaccine candidates in squir- and II. AIDS Res Hum Retrovir 2000; 16: 1757–1764. rel monkeys (Saimiri sciureus). J Virol 2001; 75: 5939–5948. 72 Albrecht B, Collins ND, Burniston MT, Nisbet JW, Ratner L, 54 Hollsberg P, Hafler DA. Seminars in medicine of the Beth Israel Green PL, Lairmore MD. Human T-lymphotropic virus type 1 Hospital, Boston. Pathogenesis of diseases induced by human open reading frame I p12(I) is required for efficient viral infec- lymphotropic virus type I infection. N Engl J Med 1993; 328: tivity in primary lymphocytes. J Virol 2000; 74: 9828–9835. 1173–1182. 73 Albrecht B, D’Souza CD, Ding W, Tridandapani S, Coggeshall 55 Cavrois M, Gessain A, Gout O, Wain-Hobson S, Wattel E. Com- KM, Lairmore MD. Activation of nuclear factor of activated T mon human T cell leukemia virus type 1 (HTLV-1) integration cells by human T-lymphotropic virus type 1 accessory protein sites in cerebrospinal fluid and blood lymphocytes of patients p12(I). J Virol 2002; 76: 3493–3501. with HTLV-1-associated myelopathy/tropical spastic paraparesis 74 Askjaer P, Kjems J. Mapping of multiple RNA binding sites of indicate that HTLV-1 crosses the blood–brain-barrier via clonal human T-cell lymphotropic virus type I rex protein within 5Ј- and HTLV-1-infected cells. J Infect Dis 2000; 182: 1044–1050. 3Ј-Rex response elements. J Biol Chem 1998; 273: 11463– 56 Leclercq I, Mortreux F, Morschhauser F, Duthilleul P, Desgranges 11471. C, Gessain A, Cavrois M, Vernant JP, Hermine O, Wattel E. 75 Ahmed YF, Hanly SM, Malim MH, Cullen BR, Greene WC. Struc- Semiquantitative analysis of residual disease in patients treated ture–function analyses of the HTLV-I Rex and HIV-1 Rev RNA

Leukemia Molecular and cellular aspects of HTLV-1 associated leukemogenesis in vivo F Mortreux et al 38 response elements: insights into the mechanism of Rex and Rev Wada C, Uchida H, Ozawa K, Asano S. Adult T-cell leukemia action. Genes Dev 1990; 4: 1014–1022. developing during immunosuppressive treatment in a renal trans- 76 Ballaun C, Farrington GK, Dobrovnik M, Rusche J, Hauber J, plant recipient. Am J Hematol 1992; 41: 292–294. Boˆhnlein E. Functional analysis of human T-cell leukemia virus 89 Zanke BW, Rush DN, Jeffery JR, Israels LG. HTLV-1 T cell lym- type I rex-response element: direct RNA binding of Rex protein phoma in a cyclosporine-treated renal transplant patient. Trans- correlates with in vivo activity. J Virol 1991; 65: 4408–4413. plantation 1989; 48: 695–697. 77 Hanly SM, Rimsky LT, Malim MH, Kim JH, Hauber J, Duc Dodon 90 Goyal SB. Intestinal strongyloidiasis manifesting as eosinophilic M, Le SY, Maizel JV, Cullen BR, Greene WC. Comparative analy- pleural effusion. South Med J 1998; 91: 768–769. sis of the HTLV-I Rex and HIV-1 Rev trans- regulatory proteins 91 Molbak K, Wested N, Hojlyng N, Scheutz F, Gottschau A, Aaby and their RNA response elements. Genes Dev 1989; 3: 1534– P, da Silva AP. The etiology of early childhood diarrhea: a com- 1544. munity study from Guinea-Bissau. J Infect Dis 1994; 169:581– 78 Toyoshima H, Itoh M, Inoue J, Seiki M, Takaku F, Yoshida M. 587. Secondary structure of the human T-cell leukemia virus type 1 92 Gelber M, Rodrig J. Strongyloides stercoralis hyperinfection. Har- rex-responsive element is essential for rex regulation of RNA pro- efuah 1996; 130: 90–92. cessing and transport of unspliced RNAs. J Virol 1990; 64: 93 de The´ G. Epstein–Barr virus and associated diseases. Course of 2825–2832. Medical Virology, Institut Pasteur, 1995/1996. Ann Med Interne 79 Inoue J, Yoshida M, Seiki M. Transcriptional (p40x) and post- 1997; 148: 357–366. transcriptional (p27x-III) regulators are required for the 94 Facer CA, Playfair JH. Malaria, Epstein–Barr virus, and the gen- expression and replication of human T-cell leukemia virus type esis of lymphomas. Adv Cancer Res 1989; 53: 33–72. I genes. Proc Natl Acad Sci USA 1987; 84: 3653–3657. 95 Patey O, Gessain A, Breuil J, Courillon-Mallet A, Daniel MT, 80 Baskerville S, Zapp M, Ellington AD. High-resolution mapping Miclea JM, Roucayrol AM, Sigaux F, Lafaix C. Seven years of of the human T-cell leukemia virus type 1 Rex-binding element recurrent severe strongyloidiasis in an HTLV-I-infected man who by in vitro selection. J Virol 1995; 69: 7559–7569. developed adult T-cell leukaemia. Aids 1992; 6: 575–579. 81 Bogerd HP, Tiley LS, Cullen BR. Specific binding of the human 96 Agape P, Copin MC, Cavrois M, Panelatti G, Plumelle Y, T-cell leukemia virus type I Rex protein to a short RNA sequence Ossondo-Landeau M, Quist D, Grossat N, Gosselin B, Fenaux P, located within the Rex-response element. J Virol 1992; 66: Wattel E. Implication of HTLV-I infection, strongyloidiasis, and 7572–7575. P53 overexpression in the development, response to treatment, 82 Giallongo A, Oliva D, Cali L, Barba G, Barbieri G, Feo S. Struc- and evolution of non-Hodgkin’s lymphomas in an endemic area ture of the human gene for alpha-enolase. Eur J Biochem 1990; (Martinique, French West Indies). J Acquir Immune Defic Syndr 190: 567–573. Hum Retrovirol 1999; 20: 394–402. 83 Cavrois M, Wain-Hobson S, Wattel E. Stochastic events in the 97 DeMars R, Held KR. The spontaneous azaguanine-resistant amplification of HTLV-I integration sites by linker-mediated PCR. mutants of diploid human fibroblasts. Humangenetik 1972; 16: Res Virol 1995; 146: 179–1784. 87–110. 84 Furukawa Y, Kubota R, Tara M, Izumo S, Osame M. Existence 98 Hanabuchi S, Ohashi T, Koya Y, Kato H, Hasegawa A, Takemura of escape mutant in HTLV-I tax during the development of adult F, Masuda T, Kannagi M. Regression of human T-cell leukemia T-cell leukemia. Blood 2001; 97: 987–993. virus type I (HTLV-I)-associated lymphomas in a rat model: pep- 85 D’Incan M, Combemale P, Verrier B, Garin D, Audoly G, Brunot tide-induced T-cell immunity. J Natl Cancer Inst 2001; 93: J, Desgranges C, Flechaire A. Transient adult T-cell 1775–1783. leukemia/lymphoma picture during varicella infection in an 99 Nowak MA, Bangham CR. Population dynamics of immune HTLV-1 carrier. Leukemia 1994; 8: 682–687. responses to persistent viruses. Science 1996; 272: 74–79. 86 Hanabuchi S, Ohashi T, Koya Y, Kato H, Takemura F, Hirokawa 100 Borrow P, Lewicki H, Wei X, Horwitz MS, Peffer N, Meyers H, K, Yoshiki T, Yagita H, Okumura K, Kannagi M. Development Nelson JA, Gairin JE, Hahn BH, Oldstone MB, Shaw GM. Anti- of human T-cell leukemia virus type 1-transformed tumors in rats viral pressure exerted by HIV-1-specific cytotoxic T lymphocytes following suppression of T-cell immunity by CD80 and CD86 (CTLs) during primary infection demonstrated by rapid selection blockade. J Virol 2000; 74: 428–435. of CTL escape virus. Nat Med 1997; 3: 205–211. 87 Jenks PJ, Barrett WY, Raftery MJ, Kelsey SM, van-der-Walt JD, 101 Chetouani F, Monestie P, Thebault P, Gaspin C, Michot B. ESSA: Kon SP, Breuer J. Development of human T-cell lymphotropic an integrated and interactive computer tool for analysing RNA virus type I-associated adult T-cell leukemia/lymphoma during secondary structure. Nucleic Acids Res 1997; 25: 3514–3522. immunosuppressive treatment following renal transplantation. 102 Gaspin C, Westhof E. An interactive framework for RNA second- Clin Infect Dis 1995; 21: 992–993. ary structure prediction with a dynamical treatment of con- 88 Tsurumi H, Tani K, Tsuruta T, Shirato R, Matsudaira T, Tojo A, straints. J Mol Biol 1995; 254: 163–174.

Leukemia