Characterization by molecular hybridization of two viral populations derived from a radiation leukemia virus.

Two leukemogenic viral populations were derived from a radiation leukemia virus of the C57BL mouse. One (FB), in which only B-tropic virus could be detected, was obtained in vivo by serial passage of cell-free extract in newborn rats. The second (3C), a complex containing at least B-tropic and xenotropic viruses, was produced in vitro by a permanent cell line (13-3C) established from the spleen of a virus-infected C57BL mouse. In molecular hybridization experiments, the 70S RNA of Gross leukemia virus hybridized 96 and 78% of FB and 3C radioactive complementary DNA's, respectively, with a relatively high thermal stability of the duplexes formed. In contrast, the 70S RNA of Rauscher leukemia virus hybridized 23 and 20% of the FB and 3C DNA probes, respectively, with a low thermal stability. The rat-grown FB virions exhibited 50% genome homology with the viruses produced in vitro on the 13-3C cells. Finally, hybridizing the FB and 3C probes with normal or leukemic mouse spleen DNA's resulted in 89 to 100% homology. The rat-grown virions did not appear to contain detectable rat cellular DNA sequences, while about 20 complete copies of their nucleotide sequences were detected in covalent linkage with FB-infected rat spleen DNA. These findings strongly support the endogenous murine origin of the investigated virions.

Antigenic variability of an immunodominant region (239-261) of the surface glycoprotein of HTLV-I.

In the 239-261 region of the surface glycoprotein of HTLV-I, we delineated five epitopes recognized by antibodies present in sera from HTLV-I infected patients. Three epitopes are located between the amino acids 252 and 261, one is of a linear type and the two others of a conformational type. One epitope is comprised between amino acids 244 and 249. The last epitope described lies between the amino acids 244 and 257 and strictly requires the presence of a serine at position 250 for its recognition. When patients are infected by viral isolates presenting a substitution of the serine at position 250 by a proline, no antibodies recognizing the 244-257 region could be found. Altogether, our data demonstrate that the immunodominant 239-261 region is complex and is subject to antigenic variability.

Identification of malignant cell clones in radio-induced murine thymic lymphomas by viral and cellular probes.

The role of B ecotropic recombinant retroviruses in the emergence and the progression of radio-induced thymic lymphomas was evaluated by analyzing the cell populations present in nine primary and in in vivo propagated tumors. For this, tumor DNAs were analyzed by the Southern method using probes specific for newly acquired proviral sequences, T-cell receptor beta-chain, and immunoglobulin heavy chain genes. Our results show that primary radio-induced tumors are composed of several tumoral cell clones but do not support that malignant cell transformation and proliferation are conferred, solely, by the newly acquired ecotropic recombinant retroviral sequences.

Recognition by human sera of a variable region of the surface glycoprotein of HTLV-I.

The region comprised between the amino acids 175 and 199 of the HTLV-I envelope surface glycoprotein is one of the immunodominant domains of this molecule. In this region, which is well recognized by sera from HTLV-I infected patients, a substitution of the proline at position 192 by a serine has been described in some isolates. Because this mutation could modify the secondary structure of the glycoprotein molecule, we studied the inference of the presence of proline or serine on the recognition of the region 175-199 by human sera. For this, three peptides have been synthetized (a 25-mer 175-199 corresponding to the sequence of the ATK prototype, and two internal 10-mer 190-Pro-199 and 190-Ser-199 having a proline or a serine at position 192) and tested by immunosorbent assay. While most sera reacted with 190-Pro-199 and with 190-Ser-199 synthetic peptides, a differential recognition was observed according to the pathology associated to HTLV-I infection. Moreover sera corresponding to patients infected with a virus harboring a serine at position 192 were found to recognize only the 10-mer with a serine. These data indicates that HTLV-I is subject to antigenic variability.

Characterization of monoclonal antibodies directed against the gp46 of human T-cell leukemia virus type I.

Essential HTLV-I biological functions depend on the structural motives of the surface glycoprotein (gp46). Monoclonal antibodies (mAbs) have been generated in order to identify functional regions of gp46. We obtained three monoclonal antibodies (3F3F10, 4F5F6 and 7G5D8) by immunizing Balb/c mice with beta-propiolactone inactivated HTLV-I producing cells and partially purified gp46. The mAbs are of the IgG 1 subclass. They have been characterized by western blot analysis, reactivity with HTLV-I and HTLV-II producing cells and ELISA binding assays using synthetic peptides. The immunoblot analysis performed with sheets prepared with the virus released by HUT 102 and 2060 cells (an HTLV-I virus producing cell line established in our laboratory) indicate that the three mAbs recognize a 46 kDa product as did the anti -gp46 mAb 0.5 alpha (18). Reactivity of the three mAbs with various cell lines was examined by indirect immunofluorescence assay. The mAb 7G5D8 stained strongly the membrane of all HTLV-I producing cells (MT2, C91/PL, HUT102 and cells of seven lines established in our laboratory and by A. Gessain); uninfected lymphoid cells (HSB-2, MOLT 4, CEM and PHA activated lymphocytes from normal donors) were negative. Interestingly cells of a HTLV-II producing line (344 MO) were positive. The mAbs 3F3F10 and 4F5F6 reacted with the same cells as did 7G5D8 but the fluorescence intensity was much lower than that observed with this later. A long synthetic peptide corresponding to the immunodominant region of the gp46 defined by the amino acids 175-199 and 10-mer peptides overlapping this region were used in an approach to identify the recognized epitope(s). The long 175-199 peptide was recognized by the three mAbs. 3F3F10 and 4F5F6 recognized none of the 10-mer peptides whereas 7G5D8 bound to 186-195 and 182-191 peptides. In addition 7G5D8 did not inhibit either syncytia formation or virus infection. In view of the data concerning the previously described mAbs 0.5 alpha, LAT 27 (5) and KE36-11 (6), our results suggest that the epitope recognized by 7G5D8 is different from those recognized by the former ones. As the 183-191 sequence corresponds to a region in which HTLV-I and HTLV-II harbour six common amino acids and two similar ones, this is consistent with the observation that 7G5D8 stained the HTLV-II producing cells 344 MO as well as all HTLV-I producing ones. Altogether our data support the hypothesis whereby this epitope recognized by 7G5D8 is contained within a sequence defined by amino acids 183-191.

Antibodies directed against a variable and neutralizable region of the HTLV-I envelope surface glycoprotein.

The majority of neutralizing antibodies of HTLV-I are directed against linear epitopes of the envelope surface glycoprotein (gp46) in the immunodominant region 175-199. Although gp46 presents a remarkable degree of conservation, the substitution of the proline at position 192 by a serine is described for 10 isolates among the 54 sequenced ones. This amino acid substitution is known to induce an important change in the orientation of the exposed residues of this region and has drastic consequences on the immunogenicity of the neutralizable epitopes located in this region. We developed monoclonal antibodies directed against epitopes located in this region containing a proline or a serine at position 192. The six monoclonal antibodies obtained recognize the gp46 at the surface of living HTLV-I producing cells, two of them are specific of a 190-197 epitope with a serine at position 192. This demonstrates that the antigenicity of this epitope differs depending on the presence of a proline or a serine at position 192. Altogether, these results demonstrate that the immunodominant neutralizable region 175-199 is antigenically variable.

Modelling, synthesis and biological activity of a BLV proteinase, made of (only) 116 amino acids.

Bovine leukaemia virus (BLV) is the aetiological agent of Leukosis enzootica bovis [Viral Oncology (1980), G. Klein (Ed.) Raven Press, New York, pp. 231-238], a widely spread disease in cattle. BLV is reported as the animal model of human T-cell leukaemia virus (HLTV) which is the causative agent of adult T-cell leukaemia and tropical spastic paraparesis. Like the viruses themselves, the two retroviral proteinases (PR) are very closely related [Virology 142 (1985) 357-377]. BLV and HTLV-I PR are reported as putative proteins made of 126 [J. Virol. 57 (1986) 826-832] and 125 [FEBS Lett. 293 (1991) 106-110] amino acids, respectively (long sequences), belonging to the aspartyl proteinase family [Nature 329 (1987) 351-354], with the aid of molecular modelling, we show that BLV and HTLV-I proteinases made of only 116 and 115 amino acids, respectively (short sequences), display three-dimensional structures similar to that observed for other retroviral aspartyl proteinases. The models are based on three-dimensional structures of Rous sarcoma virus (RSV PR) and the human immunodeficiency virus (HIV-1 PR). We used solid phase peptide synthesis to produce the putative proteolytic enzyme of BLV (116 amino acids). In this study, we show that the folded synthetic protease accurately hydrolyzes a decapeptide corresponding to the sequence of the Matrice-Capside (MA/CA) cleavage site of the gag polyprotein. In addition, the proteolytic activity is inhibited by a statine ((4S,3S)-4-amino-3-hydroxyl-6-methylheptanoic acid) containing an analogous sequence.

Inhibition of activity of the protease from bovine leukemia virus.

In view of the close similarity between bovine leukemia virus (BLV) and human T-cell leukemia virus type I (HTLV-I) we investigated the possibility of developing specific inhibitors of the proteases of these retroviruses using the purified enzyme from BLV. We tested the ability of this protease to specifically cleave various short oligopeptide substrates containing cleavage sites of BLV and HTLV-I proteases, as well as a recombinant BLV Gag precursor. The best substrate, a synthetic decapeptide bearing the natural cleavage site between the matrix and the capsid proteins of BLV Gag precursor polyprotein, was used to develop an inhibition assay. We determined the relative inhibitory effect of synthetic Gag precursor-like peptides in which the cleavable site was replaced by a non-hydrolyzable moiety. The encouraging inhibitory effect of these compounds indicates that potent non-peptidic inhibitors for retroviral proteases are not unattainable.

A new leukemogenic retrovirus isolated from tumor cells derived from a radio-induced lymphoma of C57BL/6 mice: analysis of the env and LTR sequences.

We report the cloning and characterization of a new ecotropic provirus encountered in a radio-induced thymic lymphoma of the C57BL/6 mouse. The provirus with an abnormally long LTR was inserted in the chromosomal DNA within the Pvt-1/MLVi-1/Mis-1 region which is a common integration site for MCF virus in mice and for Mo-MuLV in rats. This new ecotropic provirus was molecularly cloned and found to be infectious and competent for replication after transfection of murine cells. The recovered virus termed T3651/B was B-ecotropic, T-lymphotropic (in vivo) and highly leukemogenic for newborn C57BL/6 mice and for adult mice provided they were submitted to a subleukemogenic dose of irradiation. As compared to the AKV prototype N-ecotropic endogenous retrovirus, the T3651/B env proteins are only affected by few scattered point mutations. In contrast, the LTR has five repeats of enhancer sequences containing consensus motifs specific of the nuclear factors NF1-like, LVa, LVb and SEF1. Since a virus with such properties was encountered only once in 31 radio-induced tumors and isolated at a fourth tumor passage, a direct role of T3651/B virus in tumor genesis after irradiation is uncertain. Nevertheless, it is clear that T3651/B virus is a new leukemogenic retrovirus with a particular LTR structure which fits well with the model proposed by Rassart et al. (J. Virol. 58, 96-106, 1986) for the emergence of a thymotropic highly leukemogenic RadLV.

Establishment of HTLV-I-infected cell lines from French, Guianese and West Indian patients and isolation of a proviral clone producing viral particles.

Human T-cell leukemia virus (HTLV-I) induces adult T cell leukemia/lymphoma (ATL) and a chronic neurological disease named either tropical spastic paraparesis (TSP) or HTLV-I associated myelopathy (HAM). We report here the establishment and characterization of eight HTLV-I-infected lymphoid cell lines derived either from patients with TSP (5) or from asymptomatic carriers (1). Southern blot analysis of T cell beta chain gene rearrangements indicates that all cell lines are composed of clonal populations. The same type of analysis performed with HTLV-I-specific probes showed that they harbor 1 to 5 copies of full length proviruses often associated with deleted proviruses with a restriction map for BamHI, HindIII, PstI and SacI restriction enzymes resembling those of HTLV-I previously isolated from Japan and Caribbean area. One of the cell lines, 2060, derived from a TSP patient was shown to express a relative large amount of virus easily transmissible to fresh peripheral and cord blood lymphocytes. The full length proviral genome contained in this cell line was cloned and used in transient expression experiments. We showed that the cloned provirus was able to direct the synthesis of the major structural viral proteins, the protease and the tax and rex regulatory proteins. The structural viral proteins could be assembled into free particles detected in the culture medium of transfected cells. Although the infectivity of these viral particles remains to be determined, this new clone can be employed to examine the cell types in which this TSP-derived provirus directs viral protein synthesis and eventually replicates. It should also prove of value in studies on the early cellular events induced by viral products.

Sequence variations in the amino- and carboxy-terminal parts of the surface envelope glycoprotein of HTLV type 1 induce specific neutralizing antibodies.

The surface envelope glycoprotein gp46 of the human T cell leukemia virus type 1 elicits a strong immune response. Its protective role against HTLV-1 infection in animal models is well established, suggesting that recombinant envelope glycoproteins or synthetic peptides could be used as an effective vaccine. However, reports have indicated that some variations in envelope sequences may induce incomplete cross-neutralization between HTLV-1 strains. To identify amino acid changes that might be involved in induction of specific neutralizing antibodies, we studied sera from three patients (2085, 2555, and 2709) infected by HTLV-1 with surface glycoprotein gp46 harboring variations in amino acid sequence at positions 39, 72, 265, and 290. Inhibition of syncytia induced by parental, chimeric, or point-mutated envelope proteins indicated that sera 2555 and 2709 primarily recognized neutralizable epitopes located in N- and C-terminal parts of the gp46 glycoprotein. Amino acids changes at positions 39, 265, and 290 greatly impaired recognition of neutralizing epitopes recognized by these two sera. These results demonstrate that amino acid changes in envelope glycoprotein gp46 can induce strain-specific neutralizing antibodies in some patients. On the other hand, the neutralizing activity of serum 2085 was not affected by amino acid changes at positions 39, 265, and 290, suggesting that the neutralizing antibodies present in this serum were directed against epitopes located in other parts of the molecule, possibly those located in the central domain of the molecule, which has the same amino acid sequence in the three viruses.

Influence of amino acid substitutions on antigenicity of immunodominant regions of the HTLV type I envelope surface gylcoprotein: a study using monoclonal antibodies raised against relevant peptides.

By the use of sera of human T cell leukemia virus type I (HTVL-I)-infected individuals it was shown that amino acid substitutions at positions 192 (proline to serine) and 250 (serine to proline) in major immunodominant regions (175-199 and 239-261) of the surface envelope glycoprotein (gp46) of the virus may influence the humoral response. Since human sera are polyclonal in nature, one cannot readily discriminate between an immunoglobulin-specific recognition and multiple bindings of diverse antibodies. To overcome this difficulty we generated murine monoclonal antibodies to synthetic peptides mimicking all or portions of these gp46 regions. The reactivity of some of these antibodies to synthetic peptides harboring (or not harboring) the preceding amino acid substitutions at position 192 or 250, to denatured gp46 by Western blotting, and to live (variously substituted) HTLV-I-infected cells, combined with blocking experiments with various peptides, allow us to conclude that the major epitopes (positions 183-191, 190-197, 190-199, and 246-252) in the two immunodominant regions may elicit different antibody responses according to their sequences. It is worth noting that in a reporter gene inhibition assay, it was found that a neutralizing monoclonal antibody (MF1), the epitope for which is located between residues 190 and 197, had a high level of activity when cells (2060) harboring a gp46 with proline at position 192 were used and had no activity toward cells (1010) with a serine at this position. Therefore our results establish that certain amino acid substitutions of gp46 may drastically affect the antigenicity of the molecule and the biological activity of the antibodies elicited.

Course of specific T lymphocyte cytotoxicity, plasma and cellular viral loads, and neutralizing antibody titers in 17 recently seroconverted HIV type 1-infected patients.

Relationships were sought between specific anti-HIV cytotoxic T lymphocyte (CTL) responses (against structural and regulatory proteins of the HIV-1 LAI isolate) and plasma and cellular viral loads (VLs) in 17 recently HIV-1-infected patients including 3 displaying asymptomatic primary infection (PI) followed up for 12 months. Plasma VL was correlated directly with CD8 counts and inversely with CD4 counts. Cytotoxic reactions were observed in all patients and directed mainly against structural proteins. The earliest CTL responses were against Gag and Env proteins detected in 87 and 75% of the subjects, respectively, within the first month following PI. Anti-Env and Gag cytotoxic responses were inversely correlated with the plasma VL. Reactions against the pol gene products were thought to be either less involved in or less efficient for the initial decrease of viremia. Responses against regulatory gene products were weak and variable, apart from Nef, which was recognized by half of the subjects. Neutralizing antibodies were not detected before month 3, and were found only in six patients at subsequent times. Two of three patients with asymptomatic PI had a low viral burden and either a delayed response or one limited to a few protein CTL responses, suggesting that the magnitude of the CTL response depends on the initial plasma VL. The third patient displayed viral and CTL parameters identical to those of the patients with symptomatic PI. However, two subjects with symptomatic PI exhibited similarly low plasma VL and moderate CTL responses. Overall, the results suggest that the CTL response may not be the sole factor controlling viremia.

Immunogenicity of variable regions of the surface envelope glycoprotein of HTLV type I and identification of new major epitopes in the 239-261 region.

The reactivity of sera of 96 individuals infected with human T-cell leukemia virus type I (HTLV-I) was tested against various synthetic peptides corresponding to the gp46 immunodominant antigenic domains: residues 86-107, 175-199, and 239-261. The frequency of reactive sera was higher for 175-199 (93%) than for 239-261 (78%) or 86-107 (24%) with some variations in geographical regions and in diseases. The region 239-261 was extensively analyzed and five (linear or conformational) epitopes were found. The reactivity of sera toward functional or immunodominant domains may depend on the sequence of the infecting virus, and the role of three frequent substitutions (asparagine by tyrosine, proline by serine, and serine by proline or leucine at positions 93, 192, and 250 respectively) was established. Finally, the role of the genetic background of the host may condition the humoral immune response as individuals infected by HTLV-Is harboring the same predicted gp46 peptide sequence may recognize one, several, or all regions examined.

Thymic lymphosarcomas obtained by X-rays in association with a weakly leukemogenic virus: cellular studies.

The association in C57BL/6 mice of a subleukemogenic radiation dose (1.75 Gy X 2) which induces 7% of thymic lymphosarcomas (TL) with the injection of a weakly oncogenic B-tropic retrovirus responsible for 5% of TL resulted in a higher incidence of TL (31%) than expected from a simple cumulative effect when the viral injection preceded the irradiations (VX protocol). When virus was injected after irradiation (XV protocol) TL incidence (19%) was not significantly different from that of a cumulative phenomenon. The B-tropic virus used (1223) was isolated from RadLV-Rs extract and cloned. The TL incidence correlates with the presence of virus firstly in the thymus and bone marrow (BM) during the preleukemic period, secondly in the cell lines established in vitro from TL obtained in both protocols. This suggests that B-tropic viruses derepressed by 4 radiation doses of 1.75 Gy might be similarly implicated in the mechanism of radio-induced TL. This hypothesis is further supported by the evidence that BM restoration inhibited leukemogenesis in processes induced either by 4 radiation doses of 1.75 Gy or by the association of 2 radiation doses and viral injection whereas it has no effect on TL induced by highly oncogenic thymotropic viruses. Transplantation of BM cells from animals which had been submitted shortly before to leukemogenic VX protocol failed to induce donor type TL or leukemias in irradiated recipients suggesting that preleukemic cells either are not present or cannot be detected. However a high incidence of recipient TL was observed indicating that viruses were transferred with the grafted cells.

Evidence of recombinant ecotropic provirus integration in thymic lymphomas induced by direct or indirect radiation effects.

Several investigators described the occurrence of ecotropic recombinant proviruses in the DNA of in-vivo or in-vitro propagated radio-induced lymphomas, but such proviruses were never detected in primary tumors. To assess their biological significance in the tumorigenic process, we reinvestigated the presence of new proviruses chiefly in primary radio-induced tumors and in models of radioleukemogenesis which could give additional support for their role. Such models included thymic lymphomas originating after (i) graft of non-irradiated thymuses in thymectomized irradiated mice and (ii) the injection of a B-ecotropic retrovirus (T1223/B) in association with a subleukemogenic dose of irradiation. We report for the first time that new ecotropic proviral sequences are encountered in a significant number (30%) of primary lymphomas induced directly by irradiation or indirectly in non-irradiated thymuses grafted in irradiated hosts. The existence of a 3.5-kbp Kpn1 restriction fragment with ecotropic sequences in the digested DNA of these tumor cells indicates that these new sequences belong to an ecotropic provirus recombinant in the gag-pol region. We observed that most of the primary radio-induced tumors in which novel recombinant provirus could be detected, displayed the integration at a single or at a few sites, demonstrating their clonality with respect to viral integration. The same was observed in thymic lymphomas arising after T1223/B virus injection and irradiation and in in-vivo or in-vitro propagated tumors. Altogether, these data bring the first evidence of the integration of ecotropic recombinant proviral genomes in a significant number of primary radiation induced thymic lymphomas and of their possible role in view of their frequent occurrence in grafted thymomas.

Murine thymic lymphomas after infection with a B-ecotropic murine leukemia virus and/or X-irradiation: proviral organization and RNA expression.

The role of retroviruses in murine radioleukemogenesis was reinvestigated using a protocol associating the injection of a non-pathogenic retrovirus (T1223/B virus) and a subleukemogenic dose of X-radiation (2 X 1.75 Gy). Using the Southern blotting technique we studied MuLV proviral organization and RNA expression in thymic lymphomas induced by the combined effect of virus and irradiation or irradiation alone. A recombinant provirus was detected in the chromosomal DNA of every tumor induced by associating virus and radiation whereas it was unconstantly found in radio-induced tumors. In every instance, the provirus was not integrated at a common site. No relationship was observed between viral RNA expression and tumor induction. Trisomy 15 was observed in all metaphases irrespective of the protocol of tumor induction. The G-banding technique revealed an extra-band in several thymic lymphomas induced by irradiation and T1223/B virus injection.

Natural killer activity, production of interferon-gamma and interleukin 2 in immunodeficient C57BL/6 mice injected with RadLV-Rs viral complex.

Injection of the RadLV-Rs, a viral complex originally obtained from radio-induced thymic lymphosarcomas, into C57BL/6 mice induces a massive enlargement of spleen and lymph nodes. T- and B cell-proliferating populations display severe deterioration of their immune functions, resulting in death of the animals within three months. In contrast, the experiments reported here indicate that in such animals the natural killer (NK) activity is increased for about 2 months after viral injection. Interleukin 2 production is dramatically decreased, whereas interferon-gamma production is increased to twice the control value and thereafter decreases concomitantly with NK activity. This suggests that in RadLV-Rs-injected mice, the high NK activity is related to interferon-gamma production.

Identification of HTLV-I- or HTLV-II-producing cells by cocultivation with BHK-21 cells stably transfected with a LTR-lacZ gene construct.

The Syrian Hamster kidney cell line (BHK-21) was stably transfected with a plasmid vector containing the lacZ bacterial gene under the control of a HTLV-I-LTR promoter. In these cells termed pA18G-BHK-21, this lacZ construct is inducible by the tax protein produced by a tax expression vector. It was also shown that beta-galactosidase synthesis was detected within 48 h after cocultivation of pA18G-BHK-21 cells with HTLV-I (HUT-102, MT2, C91/PL, 2060) or HTLV-II (MoT strain) -producing cells. The number of positive cells was directly related to the number of HTLV-I or -II-infected cells seeded. In addition, the LTR transactivation observed in coculture with HTLV-I-infected cells was specifically inhibited by sera containing antibodies directed against HTLV-I proteins, but not, or only weakly, by sera containing HTLV-II antibodies. Conversely, beta-galactosidase expression induced by HTLV-II-infected cells was inhibited by sera of HTLV-II-infected individuals, but not, or only weakly, by HTLV-I-positive sera. Irrespective of the inducer cell, sera from uninfected people did not inhibit LTR-driven expression of the lacZ gene in pA18G-BHK-21 cells cocultivated with HTLV-producing cells. This assay may thus be employed profitably for the detection and quantification of both HTLV-producing cells and HTLV-specific antibodies.

Effects of bolesatine on a cell line from the SP2/O thymic lymphosarcoma.

Bolesatine, a toxic protein isolated from Boletus satanas Lenz inhibits in vitro protein synthesis in a concentration-dependent manner in a cell line from a radiation-induced thymic lymphosarcoma (SP2/O) with a 50% inhibitory concentration (IC50) of 9.5 nM (0.6 microgram/ml). In vivo, an i.p. single injection of bolesatine, corresponding to 1/6 and 1/10 of 24-h 50% lethal dose, in Balb/c mice having ascitic tumour induced by the i.p. preinjection of SP2/O cells allows a remission of 50% and 30%, respectively. Treated mice survived 120 days after the treatment, i.e. 90 days after the death of control animals.