IL-1 ANTAGONIST AND TOXICITY INDUCED BY CELL THERAPY

Abstract

The present invention relates to a IL-1 antagonist alone or in combination with other therapeutic agents and relative pharmaceutical compositions for use for the treatment and/or prevention of toxicity induced by a T cell therapy, wherein the T cell expresses at least one recombinant receptor.

Claims

1. A method for the treatment and/or prevention of toxicity induced by a T cell therapy wherein the T cell expresses at least one recombinant receptor, comprising administering an IL-1 antagonist to a patient in need thereof.

2. The method according to claim 1 wherein: (a) the administration of the IL-1 antagonist is: at a time that is less than or no more than ten, seven, six, five, four or three days after initiation of the administration of the cell therapy; and/or at a time at which the subject does not exhibit a sign or symptom of toxicity; and/or (b) between the time of the initiation of the administration of the cell therapy and the time of the administration of the IL-1 antagonist, the subject has not exhibited toxicity; and/or (c) the administration of the IL-1 antagonist is performed before or simultaneously to the T cell therapy.

3. The method according to claim 1, wherein the IL-1 antagonist is selected from the group consisting of: anakinra, rilonacept, canakinumab, gevokizumab, LY2189102, MABp1, MEDI-8968, CYT013, sIL-1RI, sIL-1RII, EBI-005, CMPX-1023, VX-765.

4. The method according claim 1, wherein the toxicity is selected from the group consisting of cytokine release syndrome, neurotoxicity, delayed toxicity.

5. The method according to claim 1, wherein the physical signs or symptoms associated with neurotoxicity, optionally severe neurotoxicity are selected from among confusion, delirium, expressive aphasia, obtundation, myoclonus, lethargy, altered mental status, convulsions, seizure-like activity, seizures (optionally as confirmed by electroencephalogram [EEG]), encephalopathy, dysphasia, tremor, choreoathetosis, symptoms that limit self-care, symptoms of peripheral motor neuropathy, symptoms of peripheral sensory neuropathy and combinations thereof; and/or the physical signs or symptoms associated with toxicity, optionally severe neurotoxicity, are associated with grade 3, grade 4 or grade 5 neurotoxicity; and/or the physical signs or symptoms associated with neurotoxicity, optionally severe neurotoxicity, manifest greater than or greater than about or about 5 days after cell therapy, 6 days after cell therapy or 7 days after T cell therapy.

6. The method according to claim 1 wherein the physical signs or symptoms associated with neurotoxicity, are selected from among acute inflammatory response and/or endothelial organ damage, fever, rigors, chills, hypotension, dyspnea, acute respiratory distress syndrome (ARDS), encephalopathy, ALT/AST elevation, renal failure, cardiac disorders, hypoxia, neurologic disturbances, and death, neurological complications such as delirium, seizure-like activity, confusion, word-finding difficulty, aphasia, and/or becoming obtunded, or fatigue, nausea, headache, seizure, tachycardia, myalgias, rash, acute vascular leak syndrome, liver function impairment, and renal failure and combinations thereof; and/or the physical signs or symptoms associated with toxicity manifest greater than or greater than about or about 5 days after cell therapy, 6 days after cell therapy or 7 days after cell therapy.

7. The method according to claim 1 wherein the T cell therapy is associated with or is capable of inducing toxicity, and wherein the T cell therapy optionally is adoptive T cell therapy and/or wherein the T cell therapy comprises administration of a dose of cells to treat a disease or condition in the subject.

8. The method according to claim 7, wherein the disease or condition is a cancer.

9. The method according to claim 1 wherein the dose of T cells comprises a number of cells between about 0.5106 cells/kg body weight of the subject and 3106 cells/kg, between about 0.75106 cells/kg and 2.5106 cells/kg or between about 1106 cells/kg and 2106 cells/kg.

10. The method according to claim 1 wherein the dose of T cells comprises a number of cells between about 1105 cells/kg and 5107 cells/kg, 2105 cells/kg and 2107 cells/kg, 2105 cells/kg and 1107 cells/kg, 2105 cells/kg and 5106 cells/kg, 2105 cells/kg and 2106 cells/kg or 2105 cells/kg and 1106 cells/kg.

11. The method according to claim 1 in combination with administering a further therapeutic agent.

12. The method according to claim 11 wherein the further therapeutic agent is a IL-6 antagonist or a chemotherapeutic agent, preferably the further therapeutic agent is selected from among tocilizumab, siltuximab, sarilumab, clazakizumab, olokizumab (CDP6038), elsilimomab, ALD518/BMS-945429, sirukumab (CNTO 136), CPSI-2634, ARGX-109, FE301, FMlOl, Hu-Mik--I, tofacitinib, ruxolitinib, CCX140-B, R0523444, BMS CCR2 22, INCB 3284 dimesylate, JNJ27141491 and RS 504393, adalimumab, certolizumab pegol, golimumab, lenalidomide, ibrutinib or acalabrutinib.

13. The method according to claim 1 wherein the recombinant receptor binds to, recognizes or targets an antigen associated with the disease or condition; and/or the recombinant receptor is a T cell receptor or a functional non-T cell receptor; and/or the recombinant receptor is a chimeric antigen receptor (CAR).

14. The method according to claim 13 wherein the CAR comprises an extracellular antigen-recognition domain that specifically binds to the antigen and an intracellular signaling domain comprising an IT AM, wherein optionally, the intracellular signaling domain comprises an intracellular domain of a CD3-zeta chain; and/or wherein the CAR further comprises a costimulatory signaling region, which optionally comprises a signaling domain of CD28 or 4-IBB.

15. The method according to claim 14 wherein the antigen is CD19 or CD 44v6.

16. The method according to claim 1 wherein the T cell is a CD4+ or CD8+ T cell.

17-20. (canceled)

Description

[0205] The present invention will be illustrated by means of non-limiting examples in reference to the following figures.

[0206] FIG. 1. Non-xenoreactive HuSGM3 T cells can be redirected against leukemia by CAR gene transfer. Cord blood (CB) human CD34.sup.+ hematopoietic stem cells (HSCs, n=5 donors) were injected intra-liver into irradiated newborn NSG (n=10, HuNSG) or SGM3 (n=10, HuSGM3) mice. After weaning, mice were monitored weekly for human lympho-hematopoietic reconstitution. (a) Mean countsSD of human (Hu) CD19.sup.+ B cells, (b) CD14.sup.+ monocytes or (c) CD3.sup.+ T cells in mice over time (weeks of age) are shown. (d) Representative plot of circulating human CD4/CD8 T cells in HuSGM3 mice at 8 weeks of age (left) and mean CD4/CD8 frequenciesSD in human peripheral blood (PB, n=16 donors), CB (n=12 donors) and HuSGM3 T cells (right) are shown. (e) Representative plot of circulating human CD45RA/CD62L T cells in HuSGM3 at 8 weeks of age (left), mean frequenciesSD of circulating CD45RA.sup.+/CD62L.sup.+ nave/stem cell memory (T.sub.Na/SCM), CD45RA.sup./CD62L.sup.+ central memory (T.sub.CM), CD45RA.sup./CD62L.sup. effector memory (T.sub.EM) and CD45RA.sup.+/CD62L.sup. effector memory RA (T.sub.EMRA) cells in HuSGM3 mice at 4, 6 and 8 weeks of age (middle) and in PB and CB (right) are shown. (f) Histology (hematoxylin and eosin, H&E) and (g) human CD3 immunohistochemistry pictures of HuSGM3 mouse thymus at 12 weeks of age (representative of n=5) are shown. (h) T cells were harvested from the spleen of 12-weeks old HuSGM3 mice (n=9) and left alone (Nil) or co-cultured with irradiated splenocytes from NSG or C57/Bl6 (B6) mice, or with irradiated human allogeneic PB mononuclear cells (Allo). Proliferation of HuSGM3 T cells was measured by CFSE-dilution. Representative plots (left) and percentages of CFSE-diluting cells in response to the different stimuli (right) are shown. Dots represent biological replicates. (i-I) 510.sup.6 HuSGM3 or PB T cells were infused into sub-lethally irradiated NSG mice (n=15 per group from three independent experiments). Mean percentagesSD of weight from initial and of circulating human CD3+ T cells over weeks from T cells are shown. (m) 510.sup.6 HuSGM3 T cells were transferred into sub-lethally irradiated NSG mice (n=18 from two independent experiments). After 24 weeks, mice were challenged with irradiated DCs from NSG mice (NSG, n=6), human allogeneic PB (Allo, n=6) or autologous CB mononuclear cells (Auto, n=6). Mice were re-challenged after 48 days (arrow). Mean percentagesSD of circulating human CD3+ T cells over days from DCs are shown. (n) HuSGM3 T cells were activated with CD3/CD28-beads and IL-7/IL-15, and RV transduced with either a CD44v6.28z, a CD44v6.BBz or a CD44v6.zOX CAR (see Methods). HuSGM3 CAR-T cells were co-cultured at a 1:10 E:T ratio with CD33.sup.+CD44v6.sup.+ THP-1 leukemic cells (upper row) or with CD19.sup.+CD44v6.sup. BV173 leukemic cells (lower row). Representative plots after 4-days co-culture (left) and mean elimination indexesSD (see Methods) by CD44v6 CAR-T cells of different design from 9 independent experiments (right panel) are shown. Results from a one- or a two-way ANOVA test are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0207] FIG. 2. Non-xenoreactive CAR-T cells cause TLS in SGM3 mice. (a) Adult SGM3 mice (8 weeks of age) were infused i.v. with 510.sup.6 (low leukemia burden) or 1010.sup.6 (high leukemia burden) CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells and after 5 weeks (low leukemia burden) or 7 weeks (high leukemia burden) with 210.sup.6 T cells from newborn HuSGM3 mice (n=3 HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (44v6.28z, n=12 mice from two independent experiments), a CD19.28z CAR (19.28z, n=12 mice) or left untransduced (CTRL, n=10 mice). Secondary recipients were followed over time by daily monitoring of weight loss and body temperature, and weekly monitoring of serum concentrations of human cytokines, mouse amyloid A (SAA), uric acid and peripheral blood leukemia. (b-c) Mean leukemic cell countsSD over weeks from leukemia challenge in mice receiving CAR-T or CTRL cells are shown. (d-e) Mean percentages of body weight variationsSD over days from CAR-T or CTRL cells are shown. Dashed lines indicate the threshold for severe weight loss (>15%). (f) Mean body temperature variationsSD over days from CAR-T cells or CTRL cells are shown. Dashed lines indicate the threshold for high fever (T>2 C.). (g) MeansSD of human IFN-, IL-2, TNF-, IL-10 or IL-6 serum concentrations measured by cytokine immunoassay 7 days after CAR-T or CTRL cells in n=4 mice with high leukemia burden per group are shown. (h-i) MeansSD of mouse SAA and uric acid serum concentrations over days from CAR-T or CTRL cells in mice with high leukemia burden are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0208] FIG. 3. Non-xenoreactive CAR-T cells induce CRS in HuSGM3 mice. (a) Adult SGM3 mice (8 weeks of age) were infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and after 4 weeks with 210.sup.6 T cells from newborn HuSGM3 mice (same HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (44v6.28z in HuSGM3, n=15 mice from three independent experiments) or a CD19.28z CAR (19.28z in HuSGM3, n=15 mice). Non HSC-humanized SGM3 mice were infused with either HuSGM3 CD44v6.28z or CD19.28z CAR-T cells as control, and results were pooled (44v6/19.28z in SGM3, n=18 mice). Secondary recipients were followed over time by daily monitoring of weight loss and body temperature and weekly monitoring of serum concentrations of human IL-6 and circulating human B cells/monocytes. (b) Mean countsSD of human CD19.sup.+ B cells or (c) CD14.sup.+ monocytes over days from 44v6.28z or 19.28.z HuSGM3 CAR-T cells are shown. (d) Mean percentages of body weight variationsSD over days from either 44v6/19.28z CAR-T cells in SGM3 mice, or from 44v6.28z or 19.28.z CAR-T cells in HuSGM3 mice are shown. Dashed lines indicate the threshold for severe weight loss (>15%). (e) Mean human IL-6 serum concentrationsSD over days from CAR-T cells are shown. (f) Mean body temperature variationsSD over days from CAR-T cells are shown. Dashed lines indicate the threshold for high fever (T>2 C.). (g) MeansSD of mouse SAA concentrations over days from CAR-T are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0209] FIG. 4. CRS severity by non-xenoreactive CAR-T cells in HuSGM3 mice correlates with leukemia burden. Adult SGM3 mice (8 weeks of age) were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells, and, after either 4 weeks (low leukemia burden) or 7 weeks (high leukemia burden), with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (44v6.28z, n=15 mice from three independent experiments) or a CD19.28z CAR (19.28z, n=15 mice). Leukemic non HSC-humanized SGM3 mice were infused with either HuSGM3 CD44v6.28z or CD19.28z CAR-T cells as control, and results were pooled (44v6/19.28z in SGM3, n=18 mice). Secondary recipients were followed over time by daily monitoring of weight loss and body temperature, and weekly monitoring of serum concentrations of human IL-6. (a-b) Mean percentages of body weight variationsSD, (c-d) mean human IL-6 serum concentrationsSD over days from CAR-T cells, (e-f) mean body temperature variationsSD over days from CAR-T cells are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001). (g-h) Kaplan-Meyer survival plots of leukemic SGM3 mice infused with 44v6/19.28z CAR-T cells or leukemic HuSGM3 mice infused with either 44v6.28z or 19.28.z CAR-T cells are shown (for survival criteria, see Methods). Results from a Mantel-Cox (log-rank) test are indicated as exact P values of 44v6.28z in HuSGM3 vs 44v6/19.28z in SGM3 (red, hazard ratio: 10.3, 1.7-61.3 95% CI) or of 19.28z in HuSGM3 vs 44v6/19.28z in SGM3 (blue, hazard ratio: 9.8, 1.9-49.9 95% CI). CRS mortality was defined as death preceded by high fever (T>2 C.) and human IL-6 serum concentration>1,500 pg/ml. (i) Adult SGM3 mice (8 weeks of age) were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells and, after 7 weeks with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (44v6.28z, n=10 mice from two independent experiments), a CD44v6.BBz CAR (44v6.BBz, n=10), a CD19.28z CAR (19.28z, n=10 mice) or a CD19.BBz CAR (19.BBz, n=10). Mean CAR-T cell countsSD, (I) mean body temperature variationsSD and (m) Kaplan-Meyer survival plots of HuSGM3 mice infused with CD44v6 CAR-T cells are shown. Results from a Mantel-Cox (log-rank) test are indicated as exact P values of 44v6.BBz vs 44v6.28z (red, hazard ratio: 0.3, 0.1-0.6 95% CI). (n) Mean CAR-T cell countsSD, (o) mean body temperature variationsSD and (p) Kaplan-Meyer survival plots of HuSGM3 mice infused with CD19 CAR-T cells are shown. Dashed lines indicate the threshold for high fever (T>2 C.). Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0210] FIG. 5. Circulating monocyte ablation by non-xenoreactive CD44v6 CAR-T cells protects HuNSG-SGM3 mice from CRS. Eight-weeks old female NSG (n=26 from 3 independent experiments, HuNSG F), male SGM3 (n=10, HuSGM3 M) or female SGM3 (n=26, HuSGM3 F) were co-infused i.v. with 10.sup.5 CB HSCs and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells and, after 5 weeks, with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with a CD19.28z CAR. Secondary recipients were followed for survival over days from CAR-T cells. (a) Mean countsSD of human CD14.sup.+ monocytes, (b) of leukemic cellsSD at 5 weeks prior to CAR-T cells are shown. Results from a one-way ANOVA test with Bonferroni correction are indicated when statistically significant (***, P<0.001). (c) Kaplan-Meyer survival plots of HuNSG F, HuSGM3 M or HuSGM3 F mice infused with CD19.28z CAR-T cells are shown. Results from a Mantel-Cox (log-rank) test are indicated as exact P values of HuSGM3 F vs HuNSG F (blue, hazard ratio: 3.8, 1.1-13.3 95% CI). (d) Leukemic HuSGM3 mice were treated or not with liposomal clodronate (LC, n=20 per group from two independent experiments) prior to the infusion of CD19.28z (n=10) or CD44v6.28z (n=10) CAR-T cells. Non HSC-humanized SGM3 mice were used as control (n=20). (d,g) Mean body temperature variationsSD over days from CAR-T cells are shown. Dashed lines indicate the threshold for high fever (T>2 C.). Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001). (e,h) Kaplan-Meyer survival plots of SGM3, HuSGM3 or HuSGM3+LC mice infused with CD19.28z or (h) CD44v6.28z CAR-T cells are shown. Results from a Mantel-Cox (log-rank) test are indicated as exact P values of HuSGM3+LC vs HuSGM3 (red, hazard ratio: 8.3, 0.9-80.5 95% CI). (f) Mean leukemic cells percentagesSD after one, 7 and 14 days from CD19.28z or (i) CD44v6.28z CAR-T cell infusion are shown. Results from a one-way ANOVA test with Bonferroni correction are indicated when statistically significant (***, P<0.001). (I) HuSGM3 mice were infused with HuSGM3 CD44v6.28z (44v6.28z, n=15 mice from three independent experiments or CD19.28z (19.28z, n=15 mice) and, after 3 weeks, with 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells. non HSC-humanized SGM3 mice were infused with either HuSGM3 CD44v6.28z or CD19.28z CAR-T cells as control, and results were pooled (44v6/19.28z in SGM3, n=17 mice). Mean body temperature variationsSD over days from leukemic cells are shown. Dashed lines indicate the threshold for high fever (T>2 C.). Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001). (m) Kaplan-Meyer survival plots of SGM3 mice infused with 44v6/19.28z CAR-T cells or HuSGM3 mice infused with either 44v6.28z or 19.28.z CAR-T cells are shown. Results from a Mantel-Cox (log-rank) test are indicated as exact P values of 19.28z in HuSGM3 vs 28z in SGM3 (blue, hazard ratio: 13.9, 1.8-105.0 95% CI). (n) Mean bone marrow (BM) leukemic cellsSD 24 weeks after CAR-T cell infusion are shown. Results from a one-way ANOVA test with Bonferroni correction are indicated when statistically significant (**, P<0.01).

[0211] FIG. 6. Monocytic cells are the key cellular sources for IL-6 and IL-1 release upon leukemia recognition by CAR-T cells. Eight-weeks old SGM3 mice (n=8 from 2 independent experiments) were co-infused i.v. with 10.sup.5 CB HSCs and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells and, after 5 weeks, with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with a CD19.28z CAR. Secondary recipients were followed over time by daily monitoring of weight loss, body temperature, and intracytoplasmic staining of human IL-1/IL-6 on peripheral blood. (a) Representative plot of human CD3+ CAR-T cells and CD14+ monocytes in leukemic HuSGM3 mice after 7 days from CD19.28z CAR-T cell infusion is shown. (b-c) Representative plots (left) and meanSD human IL-1/IL-6 production over days from CRS onset by (b) CD19.28z CAR-T cells and (c) monocytes are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001). (d) tSNE plot incorporating scRNA-Seq data of human CD45+ cells sorted from spleen of leukemic HuSGM3 mice infused with CD19.28z CAR-T cells at day 2 and day 7 of CRS (n=6511). Colors and numbers in the legend indicate transcriptionally defined clusters as well as the assigned cell type based on gene signature analyses. Representative discriminative genes are shown for clusters 5, 7, 11 and 12 (DCs and monocytes). Each dot represents an individual cell. (e) Bar plots showing mean expression (log transformed TPM values, normalized for number of cells) of the indicated genes for each cluster.

[0212] FIG. 7. Anakinra, but not tocilizumab, abolishes neurotoxicity by non-xenoreactive CAR-T cells in HuSGM3 mice. Adult SGM3 mice (8 weeks of age) were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells, and, after 7 weeks with 210.sup.6 HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with a CD44v6.28z CAR (44v6.28z, n=50 mice from three independent experiments) or a CD19.28z CAR (19.28z, n=50 mice). Just before CAR-T cells, mice received vehicle (n=14 per group), tocilizumab (n=18 per group) or anakinra (n=18 per group) and followed for CRS mortality and lethal neurotoxicity. For doses and schedule of drug administration, see Methods. CRS mortality was defined as death preceded by high fever (T>2 C.) and human IL-6 serum concentration>1,500 pg/ml. Lethal neurotoxicity was defined as death preceded by generalized paralysis or convulsions, in the absence of CRS signs. (a-b) CRS mortality over days from CAR-T cells is shown. Results from a Mantel-Cox (log-rank) test are shown as exact P values comparing tocilizumab (red, hazard ratio: 6.4, 1.6-24.7 95% CI) or anakinra (blue, hazard ratio: 3.9, 1.1-14.4 95% CI) to vehicle in mice infused with 19.28z CAR-T cells, or comparing tocilizumab (red, hazard ratio: 7.9, 2.2-29.2 95% CI) or anakinra (blue, hazard ratio: 5.3, 1.5-18.4 95% CI) to vehicle in mice infused with 44v6.28z CAR-T cells. When non visible, lines are overlapping with x axis. (c-d) Mean leukemic cells countsSD over weeks from leukemia challenge are shown. Grey arrows indicate CAR-T cell infusion. (e-f) Lethal neurotoxicity over days from CAR-T cells is shown. Results from a Mantel-Cox (log-rank) test are shown as exact P values comparing anakinra (blue, hazard ratio: 6.3, 1.1-37.1 95% CI) to vehicle in mice infused with 19.28z CAR-T cells, or comparing anakinra (blue, hazard ratio: 4.0, 0.8-20.4 95% CI) to vehicle in mice infused with 44v6.28z CAR-T cells. When non visible, lines are overlapping with x axis. (g) Histology (hematoxylin and eosin, H&E) and (h) human CD68 immunohistochemistry pictures of HuSGM3 brain (vehicle) at the time of neurotoxicity are shown. (i) Meningeal thickening quantification (0-3 score, see Methods) in HuSGM3 mice receiving vehicle (n=4), tocilizumab (n=9) or anakinra (n=7) is shown. Results from a two-tailed Mann-Whitney test are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001). (l-m) Overall survival over days from CAR-T cells is shown. Results from a Mantel-Cox (log-rank) test are shown as exact P values comparing anakinra (blue, hazard ratio: 3.9, 1.2-12.7 95% CI) to vehicle in mice infused with 19.28z CAR-T cells, or comparing anakinra (blue, hazard ratio: 3.5, 1.0-11.7 95% CI) to vehicle in mice infused with 44v6.28z CAR-T cells. (n-q) Adult SGM3 mice (8 weeks of age) were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells, and, after 7 weeks with 510.sup.6 HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with a CD19.28z CAR (19.28z, n=20 mice). At the onset of CRS symptomps, mice received vehicle (n=5 per group), tocilizumab (n=7 per group) or anakinra (n=8 per group) and followed for CRS mortality and lethal neurotoxicity. For doses and schedule of drug administration, see Methods. CRS mortality and neurotoxicity were defined above. (n) Mean body temperature variationsSD over days from CAR-T cell infusion are shown. Black arrow indicates beginning of tocilizumab/anakinra treatment. (o) CRS mortality over days from CAR-T cells is shown. Results from a Mantel-Cox (log-rank) test are shown as exact P values comparing tocilizumab or anakinra to vehicle in mice infused with 19.28z CAR-T cells. When non visible, lines are overlapping with x axis. (p) Lethal neurotoxicity over days from CAR-T cells is shown. Results from a Mantel-Cox (log-rank) test are shown as exact P values comparing anakinra to vehicle in mice infused with 19.28z CAR-T cells. When non visible, lines are overlapping with x axis. (q) Mean leukemic cells countsSD over weeks from leukemia challenge are shown. Grey arrow indicates CAR-T cell infusion. Black arrow indicates beginning of tocilizumab/anakinra treatment.

[0213] FIG. 8: Human lympho-hematopoietic reconstitution in HuSGM3 mice. (a) Mean countsSD of circulating human (Hu) CD45+ cells, (b) CD33+ myeloid cells and (c) CD15+ granulocytes in mice over time (weeks of age) are shown. (d) Mean countsSD of circulating human (Hu) CD19+ B cells, (e) CD14+ monocytes and (f) CD3+ T cells from mice over time (weeks of age) are shown. Data representative of five donors. Results from a two-way ANOVA test are indicated when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0214] FIG. 9: Human T cell repopulation of lymphoid organs in HuSGM3 mice. (a) Representative plot of circulating human CD3 T cells in HuSGM3 mice at 5 weeks of age (left) and mean CD3 countsSD in HuSGM3 mice transplanted before (n=10) or after (n=10) post-natal day 2 are shown. Results from a one- or a two-way ANOVA test are indicated when statistically significant (**, P<0.01; ***, P<0.001). (b) Representative plot of circulating human CD95+ on CD8+/CD45RA+/CD62L+ T cells at 12 weeks of age (left) and mean countsSD in human peripheral blood (PB, n=16 donors), CB (n=12 donors) and HuSGM3 T cells (right) are shown. Staining with isotype control antibody is shown in grey and specific antibody in red. (c) Representative plot of circulating human CD3 T cells in HuSGM3 mice at 5 weeks of age (left) and mean CD3 countsSD in HuSGM3 mice in the thymus, (d) spleen and (e) bone marrow are shown. Results from Mann-Whitney test are indicated when statistically significant (***P<0.001).

[0215] FIG. 10: CAR engineering of HuSGM3 T cells. (a) Representative plot (left) and frequenciesSD (right) of transduction efficiency, (b) fold increase frequenciesSD of CAR-T cells in human peripheral blood (PB, n=8 donors), CB (n=8 donors) and HuSGM3 (n=8) 15 days after activation. (c) mean human CD4/CD8 frequenciesSD and (d) mean human T.sub.Na (CD45RA+/CD62L+), T.sub.CM (CD45RA/CD62L+), T.sub.EM (CD45RA/CD62L) and T.sub.EMRA (CD45RA+CD62L) frequencies in HuSGM3 CAR-T cells before (Pre, n=8) and after (Post, n=8) ex vivo activation are shown. Results from one-way ANOVA test and Bonferroni correction are indicated when statistically significant (*P<0.05).

[0216] FIG. 11: In vitro functionality of HuSGM3 CAR-T cells. (a) HuSGM3 T cells were activated with CD3/CD28-beads and IL-7/IL-15, and RV transduced with either a CD44v6.28z, a CD44v6.BBz or a CD44v6.zOx CAR (see Methods). Mean IFN-g productionSD in HuSGM3 CAR-T cells in response to CD44v6.sup.+ leukemic cells after one day of co-culture are shown. (b) fold increase frequenciesSD of CAR-T cells after 4-days co-culture are shown. (c) PB-T cells were activated with CD3/CD28-beads and IL-7/IL-15, and RV transduced with either a CD44v6.28z, a CD44v6.BBz or a CD44v6.zOx CAR (see Methods). PB CAR-T cells were co-cultured at a 1:10 E:T ratio with CD33.sup.+CD44v6.sup.+ THP-1 leukemic cells (upper row) or with CD19.sup.+/CD44v6.sup. BV173 leukemic cells (lower row). Representative plots after 4-days co-culture (left) and mean elimination indexesSD by CD44v6 CAR-T cells of different design from 5 independent experiments are shown (right). (d) Mean IFN-g productionSD in PB CAR-T cells in response to CD44v6.sup.+ leukemic cells after one day of co-culture are shown. (e) fold increase frequenciesSD of CAR-T cells after 4-days co-culture are shown. (f) HuSGM3 T cells were activated with CD3/CD28-beads and IL-7/IL-15, and RV transduced with either a CD19.28z or a CD19.BBz CAR (see Methods). CAR-T cells were co-cultured at a 1:10 E:T ratio with CD19.sup.+ BV-173 leukemic cells (upper row) or with CD33.sup.+/CD19.sup. THP-1 leukemic cells (lower row). Representative plots after 4-days co-culture (left) and mean elimination indexesSD by CD19 CAR-T cells of different design from 5 independent experiments are shown (right). (g) Mean IFN-g productionSD in HuSGM3 CAR-T cells in response to CD19.sup.+ leukemic cells after one day of co-culture are shown. (e) fold increase frequenciesSD of HuSGM3 CAR-T cells after 4-days co-culture are shown. Results from Mann-Whitney test with Bonferroni correction are shown when statistically significant (*, P<0.05; ***, P<0.001).

[0217] FIG. 12: In vivo antileukemic effects by HuSGM3 CAR-T cells. (a) NSG mice were engrafted with CD19.sup./CD44v6.sup.+ THP-1 leukemic cells and,after one week, with 510.sup.6 CD44v6 CAR-T cells from HuSGM3 (n=16 from 3 independent experiments), PB (n=16 from 3 independent experiments), or with CTRL CD19 CAR-T cells from HuSGM3 (n=16 from 3 independent experiments). THP-1 leukemic cell progression by hepatic echography after 28 days in CTRL (HuSGM3 19.28z) or (b) CD44v6 (HuSGM3 44v6.28z) CAR-T cells are shown.(c) Antileukemia efficacy after 35 days by liver weight analysis is shown. Results from a one-way ANOVA test with Bonferroni correction are shown when statistically significant (***P<0.001). (d) Representative plots (left) and mean CAR.sup.+ cell frequenciesSD are shown. Results from a two-way ANOVA test with Bonferroni correction are shown when statistically significant (***P<0.001).

[0218] FIG. 13: Suboptimal antileukemia efficacy by CD44v6.BBz HuSGM3 CAR-T cells. (a) Representative plot (left) and mean CD19.sup.+ ALL-CM leukemic cell frequenciesSD in NSG mice (n=8) are shown. (b) Representative plots of CD44std.sup./CD44v6.sup./NGFR.sup. ALL-CM untransduced (UT, left), CD44std.sup.+/CD44v6.sup.+/NGFR.sup.+ ALL-CM (44v6.sup.+, middle) and CD44std.sup.+/CD44v6.sup.+/NGFR.sup.+ (44v6.sup., right) cells are shown. (c) Adult SGM3 mice were infused i.v. with CD19.sup.+/CD44v6.sup.+ ALL-CM leukemic cells and after 5 weeks with 510.sup.6 T cells from newborn HuSGM3 mice (n=3 HSC donors) that had been ex vivo engineered with either a CD44v6.28z (n=6), CD44v6.BBz (n=6), CD19.28z (n=6) or CD19.BBz (n=6) CAR or left untransduced (CTRL, n=6). Mean leukemic cell countsSD over weeks from tumor challenge are shown. Results from a two-way ANOVA test with Bonferroni correction are shown when statistically significant (*, P<0.05; ***, P<0.001).

[0219] FIG. 14: Deep leukemia remissions by HuSGM3 CAR-T cells. Adult SGM3 mice were infused i.v. with CD19.sup.+/CD44v6.sup.+ ALL-CM leukemic cells and after 5 weeks with 210.sup.6 T cells from newborn HuSGM3 mice (n=3 HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (n=12 from 2 independent experiments) or a CD19.28z CAR (n=12 from 2 independent experiments) or left untransduced (CTRL, n=10). (a-b) HuSGM3 CAR-T cell expansion kinetics in low (a) and high (b) tumor burden settings are shown as meanSD. (c) Representative plots after 24 weeks (left) and meanSD frequencies of BM leukemic cells in animals are shown. Threshold for minimal residual disease (MRD) identification is set at 5%. Results from one-way ANOVA test with Bonferroni correction are indicated when statistically significant (***P<0.001). MeanSD frequencies of BM leukemic cells are shown. (d) BM leukemic cells were purified from T cells and injected in SGM3 tertiary recipients. (e) Mean frequenciesSD of circulating leukemic cells in SGM3 recipients (>10%, n=10 from 2 independent experiments; 5-10%, n=6 from 2 independent experiments; <5%, n=10 from 2 independent experiments) are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (*, P<0.05; **, P<0.01; ***, P<0.001).

[0220] FIG. 15: CRS biomarkers in HuSGM3 mice infused with CAR-T cells (a) Adult SGM3 mice were infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and after 4 weeks with 210.sup.6 T cells from newborn HuSGM3 mice (same HSC donors) that had been ex vivo engineered with either a CD44v6.28z (n=15 from 3 independent experiments) or CD19.28z (n=15 from 3 independent experiments). Non HSC-humanized SGM3 mice were infused with either HuSGM3 CD44v6.28z or CD19.28z CAR-T cells as control, and results were pooled (44v6/19.28z in SGM3, n=18 mice). (a) Representative plots for CD44v6 expression on B cells and monocytes from HuSGM3 mice are shown in red. Grey histograms represent isotype control. (b) HuSGM3 CAR-T cell expansion kinetics are shown as meanSD. (c-d) Mean productionSD of human TNF-a (c) and IL-10 (d) over days from CAR-T cells are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (*, P<0.05; ***, P<0.001).

[0221] FIG. 16: CAR-T cell expansion levels in HuSGM3 mice with different leukemia burdens. (a-b) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6 CD19+CD44v6+ ALL-CM leukemic cells, and, after either 4 weeks (Low leukemia burden) or 7 weeks (High leukemia burden) with HuSGM3 CD44v6.28z (n=15) or CD19.28z (n=15) CAR-T cells. Non HSC-humanized SGM3 mice were infused with either HuSGM3 CD44v6.28z or CD19.28z CAR-T cells as control, and results were pooled (44v6/19.28z in SGM3, n=18). HuSGM3 CAR-T cell expansion kinetics are shown as meanSD. (c-d) Mean human IFN-g serum concentrationsSD over days from CAR-T cells are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (***, P<0.001). (e) Mean human mouse cytokine concentrationsSD at the peak of CRS are depicted. Results from a one-way ANOVA test with Bonferroni correction are indicated when statistically significant. (***, P<0.001).

[0222] FIG. 17: Lack of CRS in leukemic HuSGM3 mice infused with irrelevant EGFR.28z CART cells. (a) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=2 donors, HuSGM3) and 510.sup.6 CD19+CD44v6+ ALL-CM leukemic cells, and, after 7 weeks (High leukemia burden) with 210.sup.6 HuSGM3 EGFR.28z CAR-T cells (EGFR.28z, n=6 from two independent experiments). Mean percentages of body weight variationsSD over days are shown. (b) MeansSD of human IL-6 serum concentrations and (c) mean body temperature variationsSD over days are shown. (d) HuSGM3 CAR-T cell expansion kinetics are shown as meanSD. (e) CRS-free survival and (f) leukemia-free survival over days from CAR-T cells are shown.

[0223] FIG. 18: Monocyte increase in leukemic HuSGM3 mice infused with 44v6.BBz CAR-T cells. (a) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors) and 510.sup.6 CD19+CD44v6+ ALL-CM leukemic cells and, after 7 weeks with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with either a CD44v6.28z CAR (44v6.28z, n=10 mice from two independent experiments), a CD44v6.BBz CAR (44v6.BBz, n=10), or control EGFR.28z CAR (EGFR.28z, n=6 from two independent experiments). Mean HLA-DR/CD25+ percentagesSD on CAR-T cells, (b) mean countsSD of circulating leukemic cells and (c) human monocytes over days are depicted. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (**, P<0.01; ***, P<0.001). (d-m) Mean human TNF-a, IL1, IL-6, IL8, CCL2, CCL3, CCL4 and CXCL9 serum concentrationsSD at the peak of CRS are shown. Results from Mann-Whitney test with Bonferroni correction are shown when statistically significant (*, P<0.05; **, P<0.01).

[0224] FIG. 19: CAR-T cell expansions in monocyte-depleted HuSGM3 mice. (a) HuSGM3 CD19.28z CAR-T cell expansion kinetics in leukemic HuNSG female, HuSGM3 male or female mice are shown as meanSD. (b) Mean countsSD of circulating human monocytes, (c) human B cells and (d) leukemic cells before (pre) and after (post) liposomal clodronate administration are shown. Results from Mann-Whitney test with Bonferroni correction are shown when statistically significant (***, P<0.001). (f) HuSGM3 CD19.28z CAR-T cell expansion kinetics in leukemic HuSGM3 mice infused with liposomal clodronate are shown as meanSD. (g) Representative plots after 4-days co-culture (left) and mean elimination indexesSD by CD19 CAR-T cells in the presence or absence of monocytes from 5 independent experiments are shown (right). Results from Mann-Whitney test with Bonferroni correction are shown when statistically significant (**, P<0.01). (h) Mean human IL-6 serum concentrationsSD over time from leukemia challenge are depicted. (i) HuSGM3 CAR-T cell expansion kinetics in prophylactically infused HuSGM3 mice are shown. Arrow indicates leukemia challenge. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (**, P<0.01; ***, P<0.001).

[0225] FIG. 20: Monocytic cells are required for IL-6 and IL-1 release upon leukemia recognition by CAR-T cells. T cells from human peripheral blood (n=4 donors) were engineered with a CD19.28z CAR and co-cultured with CD19.sup.+ ALL-CM leukemic cells. After 48 hrs, supernatants were collected and added to PMA-stimulated THP-1 cells. (a) GM-CSF, (b) TNF-a, (c) IL-8, (d) MIP-1a, (e) IL-1b and (f) IL-6 release was measured by cytokine immunoarray after 24 hrs and is expressed as meansSD. Results from a Student's t-test are shown when statistically significant (*, P<0.05). (g) Time-course analysis of IL-1 and IL-6 release from THP-cell exposed to CAR-T cell supernatants is shown. Results from a two-way ANOVA are depicted when statistically significant (*, P<0.05; **, P<0.01; ***, P<0.001).

[0226] FIG. 21: IL-1 and IL-6 production in three-party co-cultures. T cells from human peripheral blood (n=3 donors) were engineered with a CD19.28z CAR and co-cultured with CD19.sup.+ ALL-CM leukemic cells in presence or absence of autologous monocytes. After 12, 24, or 48 hrs cells were stained for intracytoplasmic detection of human IL-1/IL-6. (a) Representative plot of three-party coculture. (b-d) Representative plots (left) and mean percentagesSD (right) of IL-1/IL6 production after 12, 24 and 48 hrs coculture by CD19.28z CAR-T cells, (c) leukemic cells and (d) monocytes. Results from a two-way ANOVA are depicted when statistically significant (*, P<0.05; **, P<0.01).

[0227] FIG. 22. IL-1 and IL-6 production in leukemic HuSGM3 mice infused with irrelevant EGFR.28z CAR-T cells. T cells from human peripheral blood (n=3 donors) were engineered with a EGFR.28z CAR and co-cultured with CD19.sup.+ ALL-CM leukemic cells in presence or absence of autologous monocytes. After 12, 24, or 48 hrs cells were stained for intracytoplasmic detection of human IL-1/IL-6. (a-b) Representative plots (left) and mean percentagesSD (right) of IL-1/IL6 production after 12, 24 and 48 hrs coculture by EGFR.28z CAR-T cells and (b) monocytes. (c) Representative plots (left) and meanSD (right) of in vivo human IL-1/IL-6 production by CD4 and CD8 CD19-28z CAR-T cells in leukemic HuSGM3 mice. Results from a Student's t-test are shown when statistically significant (***, P<0.001).

[0228] FIG. 23: Definition of human lymphoid and myeloid cell populations in HuSGM3 in CRS by scRNA-Seq. (a-b) Correlation analyses of replicate scRNA-Seq experiments, showing mean gene expression values in the indicated conditions. (c-f) tSNE plots showing single-cell gene expression levels of a T-cell signature (CD3D, CD3E, CD3G, CD27, CD28), (d) CD8/CD4, (e) B-cell (CD19, MS4A1, CD79A, CD79B, BLNK) and (f) NK-cell signature (FCGR3A, FCGR3B, NCAM1, KLRB1, KLRC1, KLRD1, KLRF1, KLRK1). Color scale reflects mean expression (log transformed TPM) across genes within each signature.

[0229] FIG. 24: Dynamic changes in the composition of human lympho-myeloid system in HuSGM3 mice during CRS. (a) Expression (scaled log transformed TPM values) of top 20 discriminative genes for each cluster is shown as heatmap. Selected representative genes for each cluster are shown on the right. Up to 200 single cells are shown for each cluster. (b) tSNE plot incorporating scRNA-Seq data of human CD45+ cells sorted from the spleen of leukemic HuSGM3 mice infused with CD19.28z CAR-T cells at day 2 and day 7 of CRS. Each dot is colored based on the respective experimental sample and replicate, as shown in the legend. Clusters, as defined in FIG. 6e, are indicated by circled numbers.

[0230] FIG. 25. Myeloid-specific expression of genes encoding for inflammatory cytokines and chemokine in leukemic HuSGM3 mice during CRS. (a-h) tSNE plots showing single-cell expression levels of the indicated genes. Color scale reflects gene expression in log(TPM+1).

[0231] FIG. 26: CAR-T cell expansion after tocilizumab/anakinra prophylaxis. (a-b) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6CD19+CD44v6+ ALL-CM leukemic cells, and, after 7 weeks with HuSGM3 T cells (same HSC donors) that had been ex vivo engineered with a CD44v6.28z CAR (44v6.28z, n=50 mice from three independent experiments) or a CD19.28z CAR (19.28z, n=50 mice). Just before CAR-T cells, mice received vehicle (n=14 per group), tocilizumab (n=18 per group) or anakinra (n=18 per group). HuSGM3 CAR-T cell expansion kinetics are shown as meanSD. (c-f) Mean human IFN-g and IL-2, concentrationsSD over days from CAR-T cells are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (*, P<0.05).

[0232] FIG. 27: CRS prevention by tocilizumab/anakinra. (a-b) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=3 donors, HuSGM3) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells, and, after 7 weeks with HuSGM3 CD44v6.28z (n=50 mice from 3 independent experiments) or CD19.28z (n=50 mice) CAR-T cells (same HSC donors). Just before CAR-T cells, mice received vehicle (n=14 per group), tocilizumab (n=18 per group) or anakinra (n=18 per group). Mean percentages of body-weight variationsSD over days from CAR-T cells. Dashed line indicate the threshold for severe weight loss (>15%). (c-d) Mean body-temperature variationsSD over days from CAR-T cells. Dashed line indicate the threshold for high fever (DT>2 C.). Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (**, P<0.01; ***, P<0.001).

[0233] FIG. 28: Cytokine/chemokine kinetics after tocilizumab/anakinra prophylaxis. (a-h) Mean human TNF-a, IL-10, IL-6, IL-1, IL-8, CXCL10, CCL3 and CCL2 serum concentrationsSD over days from CD19.28z CAR-T cells are shown. Results from a two-way ANOVA test with Bonferroni correction are indicated when statistically significant. (*, P<0.05; **P<0.01; ***,P<0.001).

[0234] FIG. 29: Lack of neurotoxicity in leukemic HuSGM3 mice infused with irrelevant EGFR.28z CAR T cells and receiving tocilizumab/anakinra prophylaxis. (a) Adult SGM3 mice were co-infused i.v. with 10.sup.5 CB HSCs (n=2 donors, HuSGM3) and 510.sup.6 CD19.sup.+CD44v6.sup.+ ALL-CM leukemic cells, and, after 7 weeks with HuSGM3 EGFR.28z (n=18 mice from 2 independent experiments) CAR-T cells (same HSC donors). Just before CAR-T cells, mice received vehicle (n=6 per group), tocilizumab (n=6 per group) or anakinra (n=6 per group). CRS-free survival, (b) neurotoxicity-free survival and (c) leukemia-free survival over days from CAR-T cells are shown.

[0235] FIG. 30: Gating strategy exemplification. Mouse peripheral blood was stained with antibodies, lysed with ACK and acquired through a FACS Canto II apparatus. Serial gating is shown for cells (upper row) and counting fluorospheres (lower panel).

DETAILED DESCRIPTION OF THE INVENTION

[0236] Methods

[0237] Generation of CAR constructs. CAR constructs were generated by gene synthesis of scFVs specific for CD44v6 (BIWA-8) or CD19 (FMC63), fused to a nerve growth factor receptor-derived spacer (NGFR), a transmembrane domain, a costimulatory endodoman from either CD28 (28z) as described in WO 2016/042461 (incorporated by reference), 4-1BB (BBz) or OX40 (zOX), and the CD3 zeta chain. In case of CD28 endodomains, the transmembrane domain was also derived from CD28. In all other cases, it was derived from CD4. All constructs were expressed in SFG RV vectors. RV supernatants were produced in 293T cells.

[0238] Cells and culture conditions. PB mononuclear cells were derived from healthy blood donors. CB mononuclear cells were supplied by commercial vendors (Lonza). CD34.sup.+ HSCs were isolated with immunomagnetic beads (Miltenyi). All procedures were approved by the Institutional Review Board (IRB number: TIGET_01) of San Raffaele University Hospital and Scientific Institute and human material obtained after written informed consent. Leukemic cell lines (THP-1, BV173) were purchased from ATCC.

[0239] THP-1 leukemia progression was followed in vivo by ultrasound imaging of the liver, where this cell line spreads forming myeloid sarcomas. ALL-CM leukemic cells were derived from patient with chronic myeloid leukemia in lymphoid blast crisis. CD44v6 was expressed in ALL-CM leukemic cells by lentiviral (LV) transduction. T cells were activated with CD3/CD28-beads (InVitrogen) at 3:1 ratio and 5 ng/ml IL-7/IL-15, and RV transduced by spinoculation at day 2 and 3. At day 6, beads were removed and T cells cultured in X-VIVO 10 (BioWhittaker) plus 10% FBS (Lonza). Transduction efficiency was determined by staining with an anti-NGFR mAb reactive with the CAR spacer. T cell expansion is expressed as fold increase: T cell numbers at day 14/T cell numbers at day 0. DCs were generated by culturing NSG mouse bone marrow, PB or CB adherent fractions with GM-CSF/IL-4 for 6 days, followed by LPS maturation overnight.

[0240] Flow cytometry. Mouse monoclonal Abs specific for human CD3 (BV510-conjugated, clone OKT3, Biolegend, lot nr. B226707; APC-Cy7-conjugated, clone SK7, Biolegend, lot nr. B225054), CD4 (PerCP-conjugated, clone SK3, BD Biosciences, lot nr. 23-5127-01), CD8 (APC-Cy7-conjugated, clone SK1, Biolegend, lot nr. B209571), CD14 (PerCP-conjugated, clone MP9, BD Biosciences, lot nr. 23-5143-01), CD15 (BV510-conjugated, clone W6D3, Biolegend, lot nr. B201379), CD19 (PE-conjugated, clone HIB19, Biolegend, lot nr. B188908), CD33 (PE-conjugated, clone WM53, Biolegend, lot nr. B195145), CD44v6 (PE-conjugated, clone 2F10, R&D, lot nr. YAV0616061; APC-conjugated, clone 2F10, R&D, lot nr. YAW0515041), CD45 (APC-Cy7-conjugated, clone HI30, Biolegend, lot nr. B214034; PE-Cy7-conjugated, clone HI30, Biolegend, lot nr. B210429), CD45RA (FITC-conjugated, clone HI100, Biolegend, lot nr. B202186), CD62L (APC-conjugated, clone DREG-56, Biolegend, lot nr. B230061), CD95 (PE-conjugated, clone DX2, Biolegend, lot nr. B2013943), NGFR (PE-conjugated, clone C40-1457, BD Biosciences, lot nr. 7068641), IL-6 (PE-conjugated, Miltenyi Biotec, lot nr. 5171106502), IL-1 (APC-conjugated, Miltenyi Biotec, lot nr. 5171106567), and a rat mAb specific for mouse CD45 (Ly5.1; PerCP-conjugated, clone 30-F11, Biolegend, lot nr. B214531) were purchased from commercial vendors. Samples were run through a FACS Canto II flow cytometer (BD Biosciences) and data were analyzed with the FlowJo software (LLC). An example of gating strategy is shown in FIG. 30.

[0241] In vitro functional assays. CAR-T cells were cultured with target cells at different E:T ratios. After 24 hrs, co-culture supernatants were collected and subsequently analyzed with the LEGENDplex bead-based cytokine immunoassay (Biolegend). After four days, surviving cells were counted and analyzed by FACS. T cells transduced with an irrelevant CAR (GD2-specific or EGFR-specific) were always used as control. Elimination index was calculated as follows: 1(number of residual target cells in presence of experimental CAR-T cells)/(number of residual target cells in presence of CTRL CAR-T cells). In CFSE-diluting assays, T cells were loaded with CFSE and stimulated with irradiated (10000 cGy) splenocytes from NSG or CD57/Bl6 mice, or with irradiated human allogeneic PB mononuclear cells at 1:5 E:S ratio. After 6 days, T cell proliferation was measured by FACS.

[0242] Mouse experiments. All mouse experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of San Raffaele University Hospital and Scientific Institute and by the Italian Governmental Health Institute (Rome, IT). Eight-to-ten weeks old female or male NSG (NOD.Cg-Prkdc.sup.scid II2rgt.sup.m1Wjl) or SGM3 mice (NSG Tg.sup.CMV-IL3,CSF2,KITLG1Eav/MloySzJ; Jackson Laboratories) were screened by PCR (according to JAX instructions and primers, stock number 013062) and ELISA (R&D systems; Catalog numbers DCK00, DGM00 and D3000 for SCF, GM-CSF and IL-3, respectively), for transgene expression and human cytokine expression. Newborn (0-2 days from birth) female of male NSG or SGM3 mice were sub-lethally irradiated (150 cGy from a linear accelerator) and injected intra-liver with 110.sup.5 human CB CD34.sup.+ cells. Adult mice were sub-lethally irradiated (200 cGy) and immediately i.v. infused with 110.sup.5 human CB CD34.sup.+ cells. For assessing X-GVHD, mice were monitored daily for hunching, activity, fur texture, skin integrity and weight loss. For studying CAR-T cell toxicities, mice were followed daily for weight loss and body temperature by rectal thermometer, and weekly for mouse SAA, uric acid and human cytokine levels by LegendPLEX bead-based cytokine immunoassay (Biolegend). For evaluating antileukemia efficacy, mice were infused i.v. with THP-1 (110.sup.6) or ALL-CM (510.sup.6 or 1010.sup.6 for mice in FIG. 2) leukemic cells and, after 5 or 7 weeks (low or high tumor burden, respectively) with 210.sup.6 CAR-T cells. Leukemic and CAR-T cell counts were monitored weekly in peripheral blood by FACS using Flow-Count Fluorospheres (BeckmanCoulter). Mice were euthanized when weight loss was >20% or when manifesting signs of inhumane suffering. For depleting phagocytes, mice were treated i.p. with liposomal clodronate ClodronateLiposomes.com) for three consecutive days prior to CAR-T cell infusion. Tocilizumab (10 mg/kg, Roactemra, Roche) or anakinra (10 mg/kg, Kineret, Amgen) were administered i.v. immediately before CAR-T cells. While tocilizumab was given only once, anakinra administration was repeated daily for 7 days because of the different pharmacokinetics.

[0243] Single-Cell RNA Sequencing

[0244] Droplet-based digital 3 end scRNA-Seq was performed on a Chromium Single-Cell Controller (10X Genomics, Pleasanton, Calif.) using the Chromium Single Cell 3 Reagent Kit v2 according to the manufacturer's instructions. Briefly, suspended single cells were partitioned in Gel Beads in Emulsion (GEMs) and lysed, followed by RNA barcoding, reverse transcription and PCR amplification (12-14 cycles). Sequencing-ready scRNA-Seq were prepared according to the manufacturer's instructions, checked and quantified on 2100 Bioanalyzer (Agilent Genomics, Santa Clara, Calif.) and Qubit 3.0 (Invitrogen, Carlsbad, Calif.) instruments. Sequenced was performed on a NextSeq 500 machine (Illumina, San Diego, Calif.) using the NextSeq 500/550 High Output v2 kit (75 cycles).

[0245] Computational Methods

[0246] Raw reads were processed and aligned to the ENSEMBL hg19 transcriptome using CellRanger version 1.3 (https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger) with default parameters. Only confidently mapped reads, non-PCR duplicates, with valid barcodes and UMIs (Unique Molecular Identifiers) were retained. The inventors filtered out low quality cells. A minimum of 500 unique genes detected for cell was required, additionally cells with a ratio of mitochondrial versus endogenous gene expression exceeding 0.1 were discarded. Resulting 6511 cells were retained for further analysis. Gene expression values were quantified in log transformed transcript per million [log(TPM+1)]. Downstream analyses were performed using the R software package Seurat version 2.1 (https://github.com/satijalab/seurat/). Cell clustering and tSNE analysis were performed on 1175 most variable genes, selected with mean expression higher than 0.01 and log transformed variance to mean ratio higher than 0.5.

[0247] Histopathological analysis. After hematoxylin and eosin staining, mouse organs including thymus and brain were blindingly and independently analyzed by at least two experienced NSG mouse pathologists (S. F and C. P.). Immunohistochemistry for TdT, human CD3 or CD68 was performed according to standard procedures. Meningeal thickening was scored according to the following arbitrary criteria: 0, normal; 1, mild; 2, moderate; 3, severe.

[0248] Statistics. Statistical analysis was performed by either one- or two-way ANOVA, by Mantel-Cox (log-rank) test or by a two-tailed Mann-Whitney test (Prism Software 5.0, Graphpad). Differences with a P value<0.05 were considered statistically significant. Sample size was calculated by power analysis with 0.05 alpha error and 0.80 power. For experiments on antileukemia efficacy (assumptions: leukemia progression in 100% of control mice vs 50% in treated mice), power analysis returned a n=11 size per experimental group. For experiments on tocilizumab or anakinra effectiveness (assumptions: CRS mortality in 35% of control mice vs 0% in treated mice), power analysis returned a n=17 size per experimental group. Before any treatment, mice were blindly randomized and no sample or animal was excluded from analysis.

EXAMPLES

[0249] T Cells from HuSGM3 Mice are Non-Xenoreactive and can be Redirected Against Leukemia by CAR Engineering

[0250] Aiming at the development of a xenograft mouse model for studying the specific contribution of myeloid cells to CAR-T cell toxicities, the inventors transplanted human cord blood (CB) hematopoietic stem cells (HSCs) by intra-liver injection into sub-lethally irradiated newborn NSG-SGM3 (HuSGM3) mice and initially profiled lympho-hematopoietic reconstitution. Compared with control HuNSG mice, HuSGM3 mice reconstituted human CD45.sup.+ hematopoiesis more rapidly (FIG. 8a), displaying lower counts of CD19.sup.+ B cells (FIG. 1a), but inversely higher counts of CD33.sup.+ myeloid cells (FIG. 8b), CD14.sup.+ monocytes (FIG. 1b), and CD15.sup.+ granulocytes (FIG. 8c). HSC humanization of newborn SGM3 mice also resulted in robust CD3.sup.+ T cell development (FIG. 1c), which contrariwise was negligible when mice were humanized in adulthood (FIG. 8d-f). The timing of HSC injection soon after birth was critical to successful human T lymphopoiesis, since a two-days delay almost completely dampened the effect (FIG. 9a). Circulating T cells in HuSGM3 mice displayed a physiological CD4/CD8 ratio (FIG. 1d) and over time appeared to differentiate from CD45RA.sup.+CD62L.sup.+ nave (T.sub.Na), to CD45RA.sup.CD62L.sup.+ central memory (T.sub.CM) to CD45RA.sup.CD62L.sup. effector memory (T.sub.EM) cells (FIG. 1e). Only a minority of CD45RA.sup.+CD62L.sup.+ T cells expressed the stem cell memory (T.sub.SCM) marker CD95.sup.39 (FIG. 9b). T cell development in HuSGM3 mice was associated with substantial thymus cellularity (at 12 weeks of age, mean 0.9910.sup.60.59 SD), including single positive CD4/CD8 T cells (FIG. 9c), and an architecture characterized by distinct cortical and medullary areas (FIG. 1f), populated with human CD3.sup.+ T cells by immunohistochemistry (FIG. 1g). Spleen (mean 3.7910.sup.61.50 SD; FIG. 9d) and bone marrow (mean 1.7110.sup.61.17 SD; FIG. 9e) were also colonized by human T cells. Intrigued by the observation of sizeable T lymphopoiesis in HuSGM3 mice, the inventors next addressed the issue of their functionality. In vitro, HuSGM3 T cells were hypo-responsive to NSG mouse antigens (I-A.sup.97), but vigorously proliferated in response to C57/Bl6 mouse antigens (I-A.sup.d) and to human alloantigens (FIG. 1h). Moreover, once i.v. transferred into sub-lethally irradiated secondary NSG recipients, HuSGM3 T cells failed to induce X-GVHD (FIG. 1i) yet persisted at low levels up to 24 weeks (FIG. 1l). The functionality of secondarily transferred HuSGM3 T cells in vivo was confirmed by expansion in response to vaccination with human allogeneic, but not with autologous CB-derived or NSG mouse dendritic cells (DCs; FIG. 1m). To establish bio-equivalence with CAR-T cells from humans, HuSGM3 T cells were activated with CD3/CD28-beads and IL-7/IL-15 ex vivo, according to a protocol that preserves early-differentiated (T.sub.SCM/T.sub.CM) memory T cells.sup.40-42 and subsequently engineered with anti-CD44v6 CARs of different designs (28z, BBz, zOX) by retroviral (RV) transduction. Transduction and expansion rates were slightly inferior to those of T cells from human peripheral blood (PB), but superimposable to those of CB T cells (FIG. 10a-b). After CAR engineering, CD4/CD8 ratios and memory differentiation phenotypes were conserved (FIG. 10c-d). HuSGM3 T cells engineered with CD44v6.28z or CD44v6.zOX CAR, but not with CD44v6.BBz CAR, specifically and effectively killed CD44v6.sup.+ THP-1 leukemic cells in vitro (FIG. 1n), produced IFN- and secondarily proliferated (FIG. 11a-b). CD44v6.BBz CAR-T cells from human PB were also weakly effective (FIG. 11c-e), indicating that suboptimal functionality was due to this particular design, rather than to T cell source. Accordingly, there were no differences between 28z and BBz designs in case of HuSGM3 T cells transduced with CD19 CARs (FIG. 11f-h). Once infused i.v. into mice previously engrafted with THP-1 leukemic cells, CD44v6.28z CAR-T cells from HuSGM3 mice were as potent as those from human PB in controlling leukemic outgrowth (FIG. 12a-d). Moreover, HuSGM3 CAR-T cells were progressively enriched for transgene expression, confirming lack of xenoreactivity (FIG. 12e).

[0251] Leukemia Clearance by CAR-T Cells in HuSGM3 Mice Associates with CRS

[0252] To evaluate the antileukemia efficacy of CAR-T cells specific for CD19 and CD44v6 in vivo using the same xeno-engrafting tumor cells, the inventors transduced patient-derived CD19.sup.+ ALL-CM leukemic cells with different CD44 isoforms containing or not the variant 6 (FIG. 13a-b). After initial remission, CD44v6.sup.+CD19.sup.+ ALL-CM leukemia-bearing mice infused with CD44v6.BBz CAR-T cells eventually relapsed (FIG. 13c), while those receiving either CD44v6.28z, CD19.BBz or CD19.CD28z CAR-T cells benefited from durable antileukemic effects (FIG. 13d). For sake of comparability, all subsequent experiments were therefore performed with either CD19.28z or CD44v6.28z CAR-T cells.

[0253] The inventors next exploited HuSGM3 CAR-T cells for mimicking early toxicities associated with antileukemic effects in the absence of confounding xenoreactivity (FIG. 2a). Adult SGM3 mice were engrafted with ALL-CM leukemic cells and later infused with either CD19.28z or CD44v6.28z CAR-T cells after 5 weeks (low leukemia burden, circulating leukemic cells: mean 20.213.1 SD; FIG. 2b) or after 7 weeks (high leukemia burden, circulating leukemic cells: mean 2811.0390.2 SD; FIG. 2c). In either setting, CD44v6.28z or CD19.28z CAR-T cells mediated rapid and long-lasting leukemia clearance in peripheral blood. However, only in case of high leukemia burden, CAR-T cells robustly expanded in vivo (FIG. 14a-b) and SGM3 mice developed a transient syndrome (median duration: 7 days, range 3-10), characterized by moderate weight loss (<15% from initial; FIG. 2d-e) and mild fever (T<2 C. from basal; FIG. 2f). These signs were paralleled by increased systemic levels of human IFN- and IL-2, but not of TNF-, IL-10 and IL-6 (FIG. 2g). The levels of serum amyloid A (SAA), murine homolog to the human CRS biomarker C-reactive protein.sup.17, whose production is under IL-6 control, were also unchanged (FIG. 2h). These data, along with a transient rise in uric acid (FIG. 2i), were therefore more indicative of tumor lysis syndrome, rather than of CRS. Long-term antileukemia efficacy by HuSGM3 CAR-T cells was confirmed by high rates of deep remission (bone marrow leukemic cells<5%; FIG. 14c) at 24 weeks from infusion, without differences between mice receiving CD19.28z (7/11 mice) or CD44v6.28z CAR-T cells (5/11). The 5% cut-off for deep remission was chosen based on subsequent experiments demonstrating lack of engraftment in tertiary recipients in case of residual bone marrow leukemic cells below this threshold (FIG. 14d-e).

[0254] Endogenous myeloid cells from immunocompromised mice derived from the NOD background are known to be functionally defective.sup.43-45. Aiming at modeling human CRS, the inventors therefore infused non-xenoreactive HuSGM3 CAR-T cells into secondary recipients previously humanized with HSCs, as a way for simultaneously providing functional myeloid cells (FIG. 3a) and antigenic CD19.sup.+ B cells or CD44v6.sup.+ monocytes (FIG. 15a). As expected, CD19.28z and CD44v6.28z CAR-T cells expanded in vivo, although with different kinetics (FIG. 15b), and induced long-lasting B cell (FIG. 3b) and monocyte (FIG. 3c) aplasia, respectively. Moreover, despite a significant difference in circulating antigenic cells before infusion (CD19.sup.+ B cells per l: mean 447.527.5 SD vs CD44v6.sup.+ monocytes per l: mean 44.13.1 SD, P<0.05 by Mann-Whitney test), CD19.28z and CD44v6.28z CAR-T cells equivalently caused a violent systemic inflammatory syndrome, highly reminiscent of human CRS and characterized by severe weight loss (>15% from initial; FIG. 3d), increased systemic human IL-6 levels (FIG. 3e) and high fever (T>2 C. from basal; FIG. 3f). Elevations of systemic human TNF- and IL-10 (FIG. 15c-d), as well as of IL-6-induced mouse SAA (FIG. 3g), closely mirrored the kinetics of the syndrome. All these signs were negative in control SGM3 mice not previously humanized with HSCs. Interestingly, it was evident that CRS by CD44v6.28z CAR-T cells was somewhat anticipated and shorter than that by CD19.28z CAR-T cells, although resulting in comparable mortality (25% vs 33.3%). At histopathology, mice dying from CRS had human CAR-T cell infiltration in the liver, often accompanied by a human histiocytic component (not shown).

[0255] Monocytes are Major Sources of IL-1 and IL-6 Induced by CAR-T Cells in HuSGM3 Mice

[0256] The inventors next examined whether leukemia presence in HuSGM3 mice, and especially its burden, could be a determinant of CRS severity by CAR-T cells, as observed in humans.sup.17. To this aim, HuSGM3 mice were co-engrafted with ALL-CM leukemic cells and later infused with non-xenoreactive HuSGM3 CAR-T cells after verifying the establishment of different leukemia burdens. CRS by either CD19.28z or CD44v6.28z CAR-T cells was more severe in case of higher leukemia burden, as revealed by more profound weight loss (FIG. 4a-b), superior systemic levels of human IL-6 (FIG. 4c-d) and higher fever (FIG. 4e-f). Consequently, CRS mortality was also significantly different (FIG. 4g-h), correlating with in vivo kinetics of CAR-T cells (FIG. 16a-b) and with systemic human IFN- elevations (FIG. 16c-d). During CRS, the majority of mouse cytokines and chemokines were undetectable (FIG. 16e), suggesting a minor contribution, if any. Leukemic HuSGM3 mice infused with irrelevant EGFR.28z CAR-T cells as control did not develop CRS (FIG. 17a-e), but conversely died from leukemia within 8 weeks (FIG. 17f).

[0257] A highly relevant question to the CAR-T cell field is whether the type of costimulatory endodomain influences CRS liability. To answer this question, the inventors compared CRS incidence and severity by either 28z or BBz CAR-T cells specific for CD19 or CD44v6 in HuSGM3 secondary recipients with high leukemia burden. Despite differences in kinetics (FIG. 4i), CD44v6.BBz CAR-T cells unexpectedly caused significantly more severe CRS than CD44v6.28z CAR-T cells, resulting in 100% mortality (FIG. 4l-m). Disproportionate CRS mortality by CD44v6.BBz CAR-T cells was associated with inferior antileukemic effects, despite greater T cell activation in vivo (FIG. 18a-b), and a paradoxical surge in human monocyte counts (FIG. 18c). Such an effect was mirrored by increased systemic levels of human inflammatory cytokines (FIG. 18d-g) and, among monocyte-derived chemokines, of IL-8 and CCL3/MIP-1 (FIG. 18h-m). In line with results in humans, CD19.BBz CAR-T cells mediated similar antileukemic effects compared to CD19.28z CAR-T cells, without inducing excessive mortality (FIG. 4n-p).

[0258] While exploring the variables influencing CRS, the inventors noticed that, due to different timing from HSC humanization (7 vs 5 weeks) at the time of CAR-T cell infusion, SGM3 mice with higher leukemia burden concomitantly displayed superior monocyte counts (mean 57.918.3 SD per l vs 26.89.8 SD per l, P<0.0001 by Mann-Whitney test). To weigh monocyte contribution to CRS, the inventors took advantage of the observation that their reconstitution in HSC-humanized mice is strain and sex-dependent (FIG. 5a), whereas leukemia engraftment (FIG. 5b) and CAR-T cell kinetics (FIG. 19a) are not. CRS mortality by CD19.28z CAR-T cells proved higher in female HuSGM3 than in female HuNSG mice (FIG. 5c), correlating with superior monocyte counts. More directly, depleting monocytes before CD19.28z CAR-T cell infusion by liposomal clodronate administration (FIG. 19b-d) had no direct effect on B cell or leukemic cell counts and completely abated CRS incidence and mortality (FIG. 5d-e and FIG. 19e). At a closer look, it was however evident that monocyte depletion had a negative impact on in vivo CAR-T cell expansion (FIG. 19f) and on the kinetics of leukemia clearance (FIG. 5f). Similar results were observed with CD44v6.28z CAR-T cells (FIG. 5g-i). The adjuvant role of monocytes on overall antileukemia efficacy by CD19.28z CAR-T cells was confirmed in vitro in three-party co-culture experiments (FIG. 19g).

[0259] To demonstrate that monocytes were primarily responsible for CRS and contributed to the antileukemic effects by CAR-T cells, the inventors used CD44v6.28z CAR-T cells as a way to ablate monocytes long term in HuSGM3 mice, and subsequently challenged them with leukemia. In agreement with the inventors' hypothesis, mice rendered monocyte aplastic by prophylactic CD44v6.28z CAR-T cells, but not mice infused with CD19.28z CAR-T cells as control, were protected from CRS (FIG. 5l-m and FIG. 19h). In the absence of monocytes, decreased secondary in vivo expansion of CD44v6.28z compared to CD19.28z CAR-T cells (FIG. 19i) was however awkwardly associated with lower rates of deep remission at sacrifice (FIG. 5n).

[0260] Although IL-6 is recognized to be pivotal for CRS pathogenesis.sup.18, it is at present unknown whether CAR-T cells themselves might be major sources of this cytokine during the syndrome. To tackle this issue, the inventors set up an in vitro cytokine release assay by co-culturing CD19.28z or control EGFR.28z CAR-T cells with leukemic cells with or without THP-1 monocytic cells. In this assay, while GM-CSF and TNF- (FIG. 20a-b) were released upon specific tumor recognition by CD19.28z CAR T cells alone, the production of IL-1, IL-6, IL-8, CCL3/MIP-1, required THP-1 cells (FIG. 20c-f). Interestingly, a time course analysis revealed that IL-1 preceded IL-6 release by approximately 24 hrs (FIG. 20g). Intracellular staining results confirmed the kinetics of IL-1/IL-6 production both in vitro, in three-party co-cultures with primary autologous monocytes (FIG. 21a-d), and in vivo in leukemic HuSGM3 mice infused with CD19.28z (FIG. 6a-c), but not with control EGFR.28z CAR-T cells (FIG. 22a-b). In vivo, transient IL-6 production was also detected in CD19.28z CAR-T cells, limitedly to the CD4 subset (FIG. 22c).

[0261] To define the cellular determinants of CRS in a broader manner, the inventors performed single-cell RNA-Sequencing (scRNA-Seq) on whole human CD45.sup.+ leukocytes isolated from leukemic HuSGM3 mice infused with CD19.28z CAR-T cells, two days after CRS onset and 5 days later. Using a microfluidics-based approach.sup.46, the inventors generated scRNA-Seq libraries from 6,511 cells and sequenced them at a median depth of 56,164 reads per cell. The average number of detected genes per cell was 1,980, with a very high correlation between replicates (R.sup.2>0.9; FIG. 23a-b). Clustering analysis, performed using a graph-based approach.sup.47,48 identified 12 clusters (cl.) encompassing the major human lymphoid and myeloid cell populations (FIG. 6d). Using unbiased gene signature analysis (FIG. 23c-f), the inventors defined populations of CD4 T cells (cl. 1 and 8), CD8 T cells (cl. 3), NK-like cells (cl. 2 4 and 9), B cells (cl. 6), as well as a poorly defined population (cl. 10). The inventors also identified monocytes (cl. 11), two sub-populations of conventional DCs, respectively expressing FCERIA and CLEC9A (cl. 5 and 7, respectively), and plasmacytoid DCs.sup.48 (cl. 12; FIG. 6d, FIG. 24a) Cell populations showed different dynamics during CRS, with monocytes and DCs being detected at both time points (FIG. 24b). As expected, cl. 6, comprising both B cells and leukemic cells, was present at the earlier time point, but disappeared later on, mirroring on-target clearance of CD19.sup.+ cells. Contrariwise, cl. 1, 8 and 3 were selectively enriched, reflecting CAR-T cell expansion. At the single-cell level, monocytes specifically expressed high levels of IL1B and IL6, as well as of IL8, CCL2, CCL8 and CXCL10 (encoding for IL-8, MCP-1, MCP-2 and IP-10, respectively; FIG. 6e). This comprehensive analysis revealed that, at least to some extent, also DCs expressed inflammatory genes, including CXCL9 and IL18 at high levels (FIG. 6e, FIG. 25a-f).

[0262] Anakinra, but not Tocilizumab, Protects HuSGM3 Mice from Lethal Neurotoxicity by CAR-T Cells

[0263] In humans, tocilizumab is often used, either alone or in combination with steroids, to manage CAR-T cell toxicities, ameliorating fever and hypotension typical of severe CRS, but apparently failing to revert severe neurotoxicity.sup.10-12,17. Despite anecdotal reports, ample data on CRS responsiveness to anakinra, an IL-1 receptor antagonist, are lacking. Motivated by the in vitro observation of early IL-1 induction in monocytes by CAR-T cells, the inventors used the inventors' xenograft mouse model of human CRS to verify whether anakinra might have some advantages over tocilizumab. At the time of CAR-T cell infusion, cohorts of leukemic HuSGM3 mice were administered either tocilizumab, anakinra, or vehicle as control. Either drug did not substantially interfere with in vivo CAR-T cell expansion (FIG. 26a-b) or in vivo IFN- and IL-2 production (FIG. 26c-f), and was effective at preventing CRS by both CD19.28z and CD44v6.28z CAR-T cells (FIG. 7a-b and FIG. 27a-d). CRS prevention by tocilizumab was associated with early normalization and a later increase in systemic human IL-6 levels (FIG. 28a-c). Initial normalization of systemic human IL-1 levels by anakinra was not followed by a similar increase (FIG. 28d), possibly due to a different pharmacology in mice compared to humans. Systemic human IL-8 and CCL3/MIP-1 levels were protractedly abated by either drug (FIG. 28e-h). Importantly, leukemia clearance by CAR-T cells in HuSGM3 mice receiving either tocilizumab or anakinra was similar to that in control mice (FIG. 7c-d).

[0264] By prolonging follow-up for detecting potential leukemia relapses, after a median of 30 days (range 27-33), in HuSGM3 mice prophylactically receiving either vehicle or tocilizumab, but not in those receiving anakinra, the inventors unexpectedly documented the occurrence of a sudden (24 hrs duration) and highly lethal neurological syndrome (FIG. 7e-f), characterized by generalized paralysis and, in some cases, by spontaneous convulsions. This form of delayed neurotoxicity was common to both CD19.28z and CD44v6.28z CAR-T cells and emerged only in mice with previous CRS (P<0.01 by Fisher's exact test, not shown). Post-mortem analysis did not reveal any sign of X-GVHD in target organs (skin and liver, not shown), but conversely showed multi-focal brain meningeal thickening, without leukemic cell infiltration in the CNS (FIG. 7g). Meningeal thickening, accompanied by human macrophage infiltration in subarachnoid space, as ascertained by scattered positivity for CD68 by immunohistochemistry (FIG. 7h), was effectively prevented by anakinra, but not tocilizumab (FIG. 7i). As a result, only anakinra prophylaxis had a statistically significant effect on overall survival (FIG. 7l-m). HuSGM3 mice infused with control EGFR.28z CAR-T cells did not develop either CRS or neurotoxicity but died from leukemia within 12 weeks (FIG. 29a-c).

[0265] The inventors finally investigated whether administering tocilizumab or anakinra to leukemic HuSGM3 mice after, rather than before, the onset of CRS by CD19.28z CAR-T cells (FIG. 7n) could revert the syndrome. Also, in this therapeutic setting, either drug was confirmed to be effective at decreasing CRS mortality, although with borderline statistical significance for anakinra (FIG. 7o). Nonetheless, anakinra treatment was uniquely associated with rescue from lethal neurotoxicity (FIG. 7p). Leukemia clearance by CAR-T cells was unaffected by either treatment (FIG. 7q).

[0266] The cellular and molecular players involved in life-threatening toxicity by cell therapy, in particular CAR-T cells in humans remain poorly understood. For gauging into its pathogenesis, the inventors used T cells derived from HSC-humanized SGM3 mice, a strain known to better support human lympho-hematopoiesis compared to NSG mice, including the development of myeloid and T cells.sup.32. Successful thymic education of human T cells in SGM3 mice was implied by their robust xenotolerance, a prerequisite for unbiased studies on CRS and neurotoxicity in secondary recipients. Although the reasons for efficient human T cell development in SGM3 mice are at present unknown, it is reasonable that transgenic expression of c-kit ligand/stem-cell factor might be key, as this cytokine is known to sustain thymopoiesis in immunocompromised mice transplanted with human HSCs.sup.49. Either transgenic expression of HLA molecules.sup.50,51 or co-transplantation of human thymic tissue.sup.52 has been successfully used for boosting thymopoiesis in xenograft models and, in the future, would be worth combining with transgenic SCF in order to further improve human T cell development in NSG mice. In the present invention, by transferring non-xenoreactive CAR-T cells in leukemic HSC-transplanted SGM3 mice, the inventors demonstrated at the single-cell level, by both scRNAseq and flow cytometry, that human circulating monocytes are primarily responsible for the systemic release of IL-6, which ultimately cause the clinical manifestations of CRS. In this human xenograft mouse model of CRS, mouse cytokines, and IL-6 in particular, did not appear to play a significant role, likely due to cytokine dysregulation inherited from the NOD background.sup.53.

[0267] In humans, circulating monocytes can be divided in different subpopulations according to their ability to phagocytose (classical monocytes, CD14.sup.+CD16.sup.), produce proinflammatory cytokines (intermediate monocytes, CD14.sup.+CD16.sup.+) or patrol endothelial integrity (non-classical monocytes, CD14.sup.loCD16.sup.).sup.54. In the inventors' model, besides proinflammatory monocytes, DCs were also involved in cytokine production, as revealed by unbiased and comprehensive in vivo scRNA-Seq analysis, underlying unexpected complexities, but also suggesting new cellular and molecular targets for therapeutic intervention. Since in the present invention the inventors have used leukemic cells that, besides obvious bone marrow homing, essentially accumulate in the circulation.sup.41, it is reasonable that intravascular leukemia recognition by CAR-T cells might have been crucial for licensing human circulating myeloid cells to produce inflammatory cytokines. Although the inventors cannot exclude that in tumors in which malignant cells do not routinely circulate in blood, e.g. lymphoma, the role of proinflammatory monocytes could be less prominent, the inventors' findings might explain the apparently higher incidence of severe CRS by CD19 CAR-T cells reported in human ALL.sup.9-12, as compared to NHL.sup.13-16.

[0268] While human T cells are known to produce IL-6, the major source of this cytokine in vivo are monocytes/macrophages.sup.55. Confirming recent findings.sup.56, the inventors found that upon tumor recognition in vitro, CAR-T cells produce negligible levels of IL-6, whose release conversely requires by-stander monocytes. Quite unexpectedly, however, the inventors also observed that monocytes are licensed by CAR-T cells to produce IL-1, with a kinetics that precedes IL-6 by many hours. Since IL-1 is capable of inducing the secretion of IL-6, as well as of its soluble IL-6R (sIL-6R).sup.55, it is tempting to speculate that CRS by CAR-T cells in HSC-humanized SGM3 mice, and in humans, might be primarily initiated by IL-1 release from circulating monocytes. The inventors' in vivo scRNA-Seq and flow-cytometry data are in line with this hypothesis. Accordingly, in the inventors' human xenograft model, antagonizing IL-1 by in vivo administration of a IL-1 antagonist, such as anakinra, was equally effective at protecting mice from CRS mortality as blocking IL-6 trans-signaling, i.e. signaling derived from IL-6 coupling to sIL-6R, through tocilizumab. Most importantly, administration of either drug did not result in decreased antileukemic effects, even if given preemptively, suggesting that pharmacological CRS prophylaxis could be routinely adopted, without jeopardizing antileukemia efficacy.

[0269] Neurotoxicity by CD19 CAR-T cells, whose acknowledgement as a separate clinical entity was initially challenged by neurological manifestations of CRS.sup.19, is becoming an emerging issue. The recent halt to some ongoing CD19 CAR-T cell trials for lethal neurotoxicity has emphasized the need of a better understanding of this severe adverse event, especially in light of further clinical development and ongoing commercialization. The inventors were surprised to find that, besides CRS, the inventors' human xenograft mouse model of CAR-T cell therapy also recapitulated neurotoxicity, which was delayed, abrupt and highly lethal, mimicking a pattern often observed in humans. Another similarity with humans was that neurotoxicity by CAR-T cells in mice was seemingly unrelated to leukemia recognition in the CNS, as indicated by no evidence of leukemic localization at brain histopathology. Instead, mice dying from neurotoxicity displayed signs of meningeal inflammation, suggesting blood-brain barrier leakage to peripherally produced cytokines, as recently described in humans.sup.57. As clinical data are accumulating, it is emerging that neurotoxicity by CAR-T cells may be more diversified than initially assumed, both in timing and relationship with CRS, possibly reflecting a combination of different mechanisms. Far from asserting that the specific type of neurotoxicity observed in the inventors' model may fit all varieties, the inventors' findings appear particularly relevant from a clinical standpoint. By analogy with humans, for example, tocilizumab did not protect mice from lethal neurotoxicity. In striking contrast, a IL-1 antagonist, anakinra, proved highly effective, either prophylactically or therapeutically, revealing IL-1 as a valuable target for global pharmacological intervention against life-threatening CAR-T cell toxicities. Selective responsiveness of neurotoxicity to anakinra is also supported by data in neonatal-onset multisystem inflammatory disease (NOMID).sup.58,59, an auto-inflammatory disease characterized by chronic aseptic meningitis, which is effectively reverted by the drug due to its CNS bioavailability.

[0270] At the current state of the art, it is debated whether CRS and neurotoxicity are restricted to CD19 CAR-T cells or, more in general, are to be expected with CAR-T cells specific for other tumor antigens. The inventors have recently developed a CD44v6-specific CAR-T cell strategy for treating AML and multiple myeloma, which express the antigen at high levels and are effectively targeted.sup.21. By using the inventors' human xenograft mouse model, the inventors here demonstrate that severe CRS and lethal neurotoxicity are likely common to all CAR-T cell antigens, provided that similarly effective in vivo tumor recognition is achieved. Interestingly, the inventors also found that in case of CD44v6 CAR-T cells, employing a BBz, rather than a 28z design, was detrimental in terms of toxicity. Differently from CD44v6.28z CAR-T cells, which rapidly ablated circulating monocytes, therefore protecting mice from CRS if given prophylactically, CD44v6.BBz CAR-T cells appeared to paradoxically induce proinflammatory monocyte licensing, resulting in 100% CRS mortality. While these findings might support the infusion of CD44v6.28z CAR-T cells soon after HSCT as a way to prevent toxicities, the observation of increased relapse rates due to prolonged monocyte aplasia warrants the implementation of a suicide gene in order to switch-off delayed unwanted effects.sup.21.

[0271] In summary, by using a newly developed xenotolerant mouse model, the inventors have demonstrated that monocyte-derived IL-1 and IL-6 are required for CRS and neurotoxicity by cell therapy, in particular CAR-T cells, and that targeted intervention against IL-1 may successfully overcome both toxicities, without interfering with antileukemia efficacy.

BIBLIOGRAPHY

[0272] 1. Sadelain, M. Curr Opin Immunol 41, 68-76 (2016).

[0273] 2. Brentjens, R. J., et al. Blood 118, 4817-4828 (2011).

[0274] 3. Savoldo, B., et al. J Clin Invest 121, 1822-1826 (2011).

[0275] 4. Kochenderfer, J. N., et al. Blood 119, 2709-2720 (2012).

[0276] 5. Porter, D. L., et al. N Engl J Med 365, 725-733 (2011).

[0277] 6. Grupp, S. A., et al. N Engl J Med 368, 1509-1518 (2013).

[0278] 7. Porter, D. L., et al. Sci Transl Med 7, 303ra139 (2015).

[0279] 8. Turtle, C. J., et al. J Clin Oncol, JCO2017728519 (2017).

[0280] 9. Brentjens, R. J., et al. Sci Transl Med 5, 177ra138 (2013).

[0281] 10. Maude, S. L., et al. N Engl J Med 371, 1507-1517 (2014).

[0282] 11. Lee, D. W., et al. Lancet 385, 517-528 (2015).

[0283] 12. Turtle, C. J., et al. J Clin Invest 126, 2123-2138 (2016).

[0284] 13. Kochenderfer, J. N., et al. J Clin Oncol 33, 540-549 (2015).

[0285] 14. Turtle, C. J., et al. Sci Transl Med 8, 355ra116 (2016).

[0286] 15. Schuster, S. J., et al. N Engl J Med 377, 2545-2554 (2017).

[0287] 16. Neelapu, S. S., et al. N Engl J Med 377, 2531-2544 (2017).

[0288] 17. Davila, M. L., et al. Sci Transl Med 6, 224ra225 (2014).

[0289] 18. Teachey, D. T., et al. Cancer Discov 6, 664-679 (2016).

[0290] 19. Lee, D. W., et al. Blood 124, 188-195 (2014).

[0291] 20. Topp, M. S., et al. Lancet Oncol 16, 57-66 (2015).

[0292] 21. Casucci, M., et al. Blood 122, 3461-3472 (2013).

[0293] 22. Gill, S., et al. Blood 123, 2343-2354 (2014).

[0294] 23. Sotillo, E., et al. Cancer Discov 5, 1282-1295 (2015).

[0295] 24. Zhao, Z., et al. Cancer Cell 28, 415-428 (2015).

[0296] 25. Long, A. H., et al. Nat Med 21, 581-590 (2015).

[0297] 26. Sommermeyer, D., et al. Leukemia 30, 492-500 (2016).

[0298] 27. Gomes-Silva, D., et al. Blood 130, 285-296 (2017).

[0299] 28. Sabatino, M., et al. Blood 128, 519-528 (2016).

[0300] 29. Bondanza, A., et al. Blood 107, 1828-1836 (2006).

[0301] 30. Mastaglio, S., et al. Blood (2017).

[0302] 31. Traggiai, E., et al. Science 304, 104-107 (2004).

[0303] 32. Billerbeck, E., et al. Blood 117, 3076-3086 (2011).

[0304] 33. Rongvaux, A., et al. Nat Biotechnol 32, 364-372 (2014).

[0305] 34. Rongvaux, A., et al. Annu Rev Immunol 31, 635-674 (2013).

[0306] 35. Kochenderfer, J. N., et al., Blood 116, 3875-3886 (2010).

[0307] 36. Davila, M. L., Kloss, C. C., Gunset, G. & Sadelain, M. PLoS One 8, e61338 (2013).

[0308] 37. Paszkiewicz, P. J., et al. J Clin Invest 126, 4262-4272 (2016).

[0309] 38. Ghosh, A., et al. Nat Med 23, 242-249 (2017).

[0310] 39. Gattinoni, L., Nat Med 23, 18-27 (2017).

[0311] 40. Kaneko, S., et al. Blood 113, 1006-1015 (2009).

[0312] 41. Bondanza, A., et al. Blood 117, 6469-6478 (2011).

[0313] 42. Cieri, N., et al. Blood 121, 573-584 (2013).

[0314] 43. Rongvaux, A., et al. Annu Rev Immunol 31, 635-674 (2013).

[0315] 44. Shultz, L. D., et al. J Immunol 154, 180-191 (1995).

[0316] 45. Takenaka, K., et al. Nat Immunol 8, 1313-1323 (2007).

[0317] 46. Zheng, G. X., et al. Nat Commun 8, 14049 (2017).

[0318] 47. Levine, J. H., et al. Cell 162, 184-197 (2015).

[0319] 48. Villani, A. C., et al. Science 356(2017).

[0320] 49. Wils, E. J., et al. J Immunol 187, 2974-2981 (2011).

[0321] 50. Koo, G. C., Hasan, A. & O'Reilly, R. J. Expert Rev Vaccines 8, 113-120 (2009).

[0322] 51. Strowig, T., et al. J Exp Med 206, 1423-1434 (2009).

[0323] 52. Lan, P., et al., Blood 108, 487-492 (2006).

[0324] 53. Bouma, G., et al. Eur J Immunol 35, 225-235 (2005).

[0325] 54. Patel, A. A., et al. J Exp Med 214, 1913-1923 (2017).

[0326] 55. Hunter, C. A. & Jones, S. A. Nat Immunol 16, 448-457 (2015).

[0327] 56. Singh, N., et al. Cytotherapy 19, 867-880 (2017).

[0328] 57. Gust, J., et al. Cancer Discov 7, 1404-1419 (2017).

[0329] 58. Goldbach-Mansky, R., et al. N Engl J Med 355, 581-592 (2006).

[0330] 59. Fox, E., et al. J Neuroimmunol 223, 138-140 (2010).