Intrathymic Notch3 and CXCR4 combinatorial interplay facilitates T-cell leukemia propagation

Notch hyperactivation dominates T-cell acute lymphoblastic leukemia development, but the mechanisms underlying “pre-leukemic” cell dissemination are still unclear. Here we describe how deregulated Notch3 signaling enhances CXCR4 cell-surface expression and migratory ability of CD4+CD8+ thymocytes, possibly contributing to “pre-leukemic” cell propagation, early in disease progression. In transgenic mice overexpressing the constitutively active Notch3 intracellular domain, we detect the progressive increase in circulating blood and bone marrow of CD4+CD8+ cells, characterized by high and combined surface expression of Notch3 and CXCR4. We report for the first time that transplantation of such CD4+CD8+ cells reveals their competence in infiltrating spleen and bone marrow of immunocompromised recipient mice. We also show that CXCR4 surface expression is central to the migratory ability of CD4+CD8+ cells and such an expression is regulated by Notch3 through β-arrestin in human leukemia cells. De novo, we propose that hyperactive Notch3 signaling by boosting CXCR4-dependent migration promotes anomalous egression of CD4+CD8+ cells from the thymus in early leukemia stages. In fact, in vivo CXCR4 antagonism prevents bone marrow colonization by such CD4+CD8+ cells in young Notch3 transgenic mice. Therefore, our data suggest that combined therapies precociously counteracting intrathymic Notch3/CXCR4 crosstalk may prevent dissemination of “pre-leukemic” CD4+CD8+ cells, by a “thymus-autonomous” mechanism.


Introduction
Malignant transformation of T-cell progenitors is causative of T-cell acute lymphoblastic leukemia (T-ALL). T-ALL accounts for 15% of pediatric and 25% of adult ALL cases, very frequently bearing somatic gain-of-function gene mutations in Notch1, as well as overexpression of Notch3 [1][2][3]. Moreover, Notch3 gene activating mutations have been recently reported in T-ALL [4]. Notch receptors regulate T-cell fate choices, dominating early steps of thymocyte differentiation [5,6]. Additionally, thymocyte turnover is regulated by natural cell competition, between "young" bone marrow (BM)-derived and "old" thymus-resident progenitors, whose impairment enables T-ALL progression via pre-malignant stages [7]. A major role is also played by the interaction between leukemia and nonleukemia cells in the microenvironment, probably dictating the survival of leukemia initiating cells.
We previously demonstrated the oncogenic potential of Notch3 in transgenic (tg) mice, overexpressing the constitutively active intracellular domain of Notch3 (N3-IC) in immature thymocytes, which develop an aggressive T-cell ALL, recapitulating most of human T-ALL features. Fourweek-old N3-ICtg mice display early precursor deregulation, by expanding the DN3 stage and increasing total thymic cellularity [18]. At 12 weeks, thymus depletion, splenomegaly, lymph nodes enlargement, and BM colonization by lymphoblastic cell population occur. Phenotypic similarities between the infiltrating lymphoma cells and the thymocytes of younger mice suggested an immature T-cell propagation [18]. Notably, a prominent feature in Notch-induced T-ALL mouse models is the circulation of CD4 + CD8 + T-cells [19,20]. Moreover, disrupted natural cell competition in the thymus may enable progression to leukemia by dissemination of pre-T-ALL CD4 lo/+ /CD8 + cells [7].
Here, we study anomalous CD4 + CD8 + T-cells propagation in Notch3-IC-induced T-ALL, by detecting atypical DP T-cells outside the thymus at early and late T-ALL stages. Notably, our results highlight that the high and combined expression of CXCR4 and Notch3 defines "preleukemic" DP-cells, precociously detected inside the thymus, and then in circulating blood and BM. Newly, by experiments of in vivo cell-transfer, we delineate the biological properties of CD4 + CD8 + Notch3 + CXCR4 + thymocytes that are fit to infiltrate peripheral organs. Notably, in young transgenic N3-ICtg mice, the in vivo administration of the CXCR4 antagonist, AMD3100, can drastically reduce the infiltration of CD4 + CD8 + Notch3 + CXCR4 + Tcells into BM. Interestingly, by ex vivo and in vitro experiments, we demonstrate that Notch3 modulates CXCR4 cell-surface expression through a β-arrestin-mediated mechanism, both in N3-ICtg mice-derived cells and in the human TALL-1 cell line, known to harbor Notch3 activating mutations [21].
Overall, our data suggest that in Notch3-induced T-ALL leukemia, high Notch3 and CXCR4 co-expression marks "pre-leukemic" DP thymocytes and allows their egression and propagation outside the thymus. This finally gives them access to the blood stream and BM and favors T-ALL progression.
SDF-1/CXCR4 axis-mediated DN/DP transition and its suggested relevance in T-ALLs [14] prompted us to analyze CXCR4 expression in abnormally represented DP T-cells in N3-ICtg mice.
First, we observed a higher CXCR4 cell-surface expression in transgenic DP thymocytes at the single cell level (Fig. 1a), as highlighted by the significant difference in the CXCR4 mean fluorescence intensity (MFI) between wild-type (WT) and N3-ICtg DP-gated thymocytes of 6-8week-old mice. Increased CXCR4 expression seems to not be transcriptionally regulated (Fig. 1b). Moreover, in DP thymocytes of N3-ICtg as well as in WT mice, SDF-1 mRNA is undetectable (data not shown). Our data propose a leukemic cell-autonomous mechanism, excluding SDF-1 autocrine production, as previously reported [24,25]. In addition, we observed a greatly decreased expression of SDF-1 mRNA in N3-ICtg as opposed to WT whole thymus (Fig. 1c).
SDF-1/CXCR4 axis-mediated chemotaxis regulates thymocyte maturation by directing DP T-cells into the cortex [26] and retaining them there. In our model, enhanced cellsurface expression correlates to CXCR4 function. In fact, in response to SDF-1, N3-ICtg DP thymocyte migration increases more than WT DP cell migration and does so in a dose-dependent manner (Fig. 1d).
Conceivably, transgenic DP thymocytes, with higher CXCR4 surface expression, are more responsive to SDF-1, whose reduced expression in the N3-ICtg thymus microenvironment (Fig. 1c) and decreased retaining function may deregulate their progressive maturation into the thymic cortex and increase their responsivity to external stimuli.
Our results suggest a dynamic trend, with a time-dependent switch of circulating DP T-cell subsets in N3-ICtg blood. Propagation of Notch3 +high CXCR4 +high DP T-cells outside the thymus may be a result of the abnormal increase within young N3-ICtg thymus (Fig. 2a) and their enhanced SDF-1dependent migratory ability (Fig. 1d), possibly related to the decreased expression of thymic SDF-1 (Fig. 1c), which may justify the decreased retention inside the thymus.
The wave of Notch3 + CXCR4 + DP T-cells progressively invades N3-ICtg BM CXCR4 expression facilitates pathogenic T-cell trafficking in the BM of murine Aplastic Anemia model [28]. Moreover, BM stroma producing SDF-1 and vascular endothelial niche are necessary for leukemic cell maintenance in T-ALL progression [25].
Newly, our results highlight the homing and expansion of Notch3 + CXCR4 + DP-cells in N3-ICtg BM, during T-ALL progression.
The association of CXCR4 expression to site-specific metastasis, into BM [29,30] of neuroblastoma-bearing patients [31], and the supportive effect of endothelial niche in Notch-dependent T-ALL cells maintenance [25,24], favor our hypothesis that Notch3/CXCR4 crosstalk directs "pre-leukemic" DP-cells to the BM.
Purified CD4 + CD8 + thymocytes of 6-8-week-old transgenic mice, immunophenotypically distinguished in Notch3 +high CXCR4 +high (N3 high ) or Notch3 +low CXCR4 +high (N3 low ) ( Figure S3B) were injected intravenously (i.v.) into NSG recipient mice. At day10 post-injection, BM cells from mice receiving N3 high cells, show a high percentage of donor DP T-cells (6.95%), mostly composed of CD4 + CD8 + Notch3 +high CXCR4 +high cells (89.7%) (Fig. 5a, upper   panels). Conversely, BM cells from N3 low -injected mice display only 0.1% of DP T-cells, essentially devoid of surface Notch3 (Fig. 5a, lower panels). The evaluation of DP T-cell numbers in BM of N3 high -injected mice (Fig. 5b) evidences their trend to greatly increase, by comparing DP cell numbers at day1 and day10 (upper graph). Already at day1, DP T-cell number is higher in N3 high -injected when compared to N3 low -injected mice. In fact, at day1, N3 low DP cells fail to home in NSG BM (Fig. 5b). Strikingly at day10, we demonstrate that N3 high DP cells achieve a greatly increased engraftment in the BM of NSG recipient mice when compared to N3 low DP cells (Fig. 5b). The difference is clearly indicated by the increased ratio of DP cell numbers (Fig. 5b, lower graph) from N3 high -injected versus N3 low -injected mice at day10. This evidence confirms the higher competence of N3 high thymocytes to home already at Selective homing and engraftment of N3-ICtg DP thymocytes, highly expressing Notch3 and CXCR4. a (CD4 + CD8 + Notch3 +high CXCR4 +high ) N3 high or (CD4 + CD8 + Notch3 +low CXCR4 +high ) N3 low thymocytes engraft in BM day10 post-intravenous injection (i.v.). Percentages of DP cells (middle) and of CD4 + CD8 + Notch3 + CXCR4 + (right). b Absolute numbers of DP infiltrating NSG BM after N3 high (purple-circle) or N3 low (blue-square) thymocytes i.v. (upper, Kruskal-Wallis test) and the ratio N3 high /N3 low DP cells at day1 (green-hexagon) and day10 (red-triangle) (lower, Mann-Whitney test) (*p<0.05). c Ex vivo experiments: percentages of AnnexinV + cells in N3 high (upper) and N3 low (lower) thymocytes in 24 h SDF-1-treated relative to saline (PBS). Data are represented as meanvalue ± SD (**p < 0.01; ns, not significant; Student's t-test). d Absolute DP T-cell numbers in spleen after N3 high (purple-circle) or N3 low (blue-square) DP thymocytes i.v. (upper) and the ratio of N3 high /N3 low DP T-cells (lower) at day1 (green-hexagon) and day10 (red-triangle) (*p<0.05; Student's t-test, Mann-Whitney). See also Figure S3A and B day1 and to engraft and expand in the BM of recipient mice 10 days post-injection. This "fitness" may rely on the enhanced responsivity of N3 high cells to BM microenvironment, releasing the pro-survival factor SDF-1. To this regard, in ex vivo experiments, 24 h after SDF-1 treatment, we observed a significant reduction of spontaneous apoptosis, as measured by AnnexinV staining (AnnnexinV + ) within N3 high DP thymocytes subset (Fig. 5c,  upper panel), when compared to N3 low DP cells, which appear quite SDF-1 unresponsive (Fig. 5c, lower panel). We hypothesize that Notch3 hyperexpression, through the increased surface expression of CXCR4 (Fig. 1a), allows a better DP settlement by potentiating SDF-1 responsiveness and hence CXCR4-mediated survival programs. Interestingly, N3 high and N3 low DP cells have a different proliferation rate. In vivo BrdU labeling demonstrates that N3 high DP cells have an enhanced proliferation rate ( Figure S4A and B). Additionally, in comparison to N3 low , N3 high DP subset displays a higher Ki67 expression as confirmed by higher Ki67 MFI ( Figure S4C).
Propagation of purified thymocytes was further analyzed in the spleen. By day1 post-injection, N3 high DP-cells invade the spleen, further increasing at day10. In contrast, N3 low cells are nearly absent at both times (Fig. 5d, upper graph). Accordingly, the ratio of Notch3 +high CXCR4 +high versus Notch3 +low CXCR4 +high (N3 high /N3 low ) DP cells increases, sustaining the competence of N3 high to be successful in infiltrating both BM and spleen (Fig. 5d, lower graph). Overall, we hypothesize that Notch3/CXCR4 crosstalk enhances the early spreading of N3 high DP thymocytes toward BM and the spleen in response to SDF-1 chemoattraction, thus evidentiating their "fitness" in dissemination and expansion.

CXCR4 in vivo pharmacological antagonism reduces BM infiltration by Notch3 overexpressing DP T-cells
AMD3100 is an extremely specific and effective CXCR4 antagonist. It has been used for hematopoietic stem cell mobilization as well as for the treatment of myeloid leukemia and solid tumors [34].
To investigate the effect of CXCR4 antagonism in abnormal Notch3 + CXCR4 + DP cell propagation, we performed a daily in vivo intraperitoneal (IP) administration of young N3-ICtg mice, just before BM colonization occurs. After 10 days, we evaluated DP T-cell numbers in BM by flow cytometry, as shown in Fig. 6. Total DP cells that infiltrate BM are greatly reduced by the administration of AMD3100, as compared to PBS-injected mice (Fig. 6a). Interestingly, infiltrating DP cells are mostly composed by Notch3 + CXCR4 + , still significantly decreased in AMD3100-treated with respect to PBS-injected mice (Fig.  6b). Notably, these results demonstrate that in vivo CXCR4 pharmacological antagonism can dramatically reduce the BM infiltration and engraftment by "pre-leukemic" DP cells. Therefore, this in vivo treatment may have significant implications for a potential therapeutic approach.

Notch3 targets CXCR4 surface expression by modulating β-arrestin1 function
For a molecular insight in Notch3-boosted CXCR4 expression, we modulated Notch3 function in a human leukemia cell-line, TALL-1 cells, characterized by CD4/ CD8/CD3 expression, Notch3 activating mutations without known alterations of Notch1 [4], and here, for the first time, by high CXCR4 cell-surface expression (Fig. 7a).
To investigate the role of β-arrestin1 in mediating the Notch3 effect on CXCR4 expression, we utilized a β-arrestin1 mutant that mimics its phosphorylated form Ser412→Asp (S412D) and acts as a dominant negative inhibitor of receptor sequestration/internalization [41]. To this purpose, we transiently cotransfected Hek293 cells with the active Notch3-IC together with either the WT (β-arrestin1-wt) or the phosphorylated mutant (β-arrestin1-S412D) ( Figure S7). As shown in Fig. 7d, we observed that, in the presence of activated Notch3, the constitutively phosphorylated form of β-arrestin1 (β-arrestin1-S412D) enhances CXCR4 surface expression more efficiently with respect to the wt form.
Together, the above results demonstrate that Notch3 is able to sustain the levels of p-β-arrestin1, which in turn cooperates with Notch3 to maintain the levels of CXCR4 surface expression.
Finally, since we observed a significant decrease of β-arrestin1 protein levels in thymocytes from N3-ICtg when compared to WT mice (Fig. 7f), which would further support its role in regulating CXCR4 expression, in order to investigate a possible translational significance of the correlation between the CXCR4 receptor and the β-arrestin1 adaptor/scaffold protein, we performed an in silico analysis of CXCR4 and β-arrestin1 gene expression in a cohort of 117 T-ALL pediatric patients [42]. Interestingly, we found a significant inverse correlation between CXCR4 and β-arrestin1 gene expression in selected patient subgroups ( Figure S9A and B).

Discussion
We characterized the impact of the relative "fitness" of thymus-derived "pre-leukemic" DP-cells on BM homing and engraftment. Still poorly understood are clues on what goes astray when in a normal thymus originates T-cell leukemia.
Gene expression profiling of human T-ALL samples, distinguishes different stages of arrest during T-cell development: (i) early immature T-ALLs at the DN stage, opposed to (ii) CD1a + /CD4 + /CD8 + or CD3 + /CD4 + /CD8 + T-cells, respectively, in early-cortical and late-cortical thymocyte T-ALLs [43]. We report here the competitive progressive expansion of CD4 + CD8 + Notch3 +high CXCR4 +high thymocytes in 3-week-old N3-ICtg thymus, peaking at 6-8 weeks and reaching a plateau at 12-14 weeks, well correlated to the anomalous propagation of DP-cells in circulating blood. Enhanced CXCR4 cell-surface expression characterizes highly migrating transgenic DP thymocytes and may interfere with normal T-cell/stroma interactions, thus relieving Notch3 +high CXCR4 +high DP cells from outside control. Our data converge on a "thymus-autonomous" mechanism also relying on reduced SDF-1-dependent retention inside the thymus of transgenic as opposed to WT intrathymic microenvironment. Hence, early detection of Notch3 +high CXCR4 +high DP in blood and BM of N3-ICtg mice at 6-8 weeks of age may suggest DP accelerated egress and thymus depletion, possibly unbalancing thymocyte subpopulations. As a possible consequence of this, T-cell depletion and subverted architecture in 12-14-week-old N3-ICtg thymus occur as previously reported [18]. Our results demonstrate the transient presence of Notch3 +high CXCR4 +high DP in circulating blood, like a wave of early propagating cells, further suggesting that Notch3 +high CXCR4 +high expression is a possible "tag" to precociously detect abnormally circulating DP-cells. In contrast, 12-14-week-old N3-ICtg BM is still largely populated by Notch3 +high CXCR4 +high DP-cells, suggesting that Notch3 makes them more fit to BM colonization. Indeed, SDF-1-producing vascular endothelial cells required in T-ALL niche [25] represents a supportive microenvironment, chemoattracting and favoring the expansion of Notch3 +high CXCR4 +high DP cells in BM vascular niches.
We demonstrate that Notch3 modulates surface expression of CXCR4 in human TALL-1, by requiring β-arrestin1, whose unphosphorylated status is necessary for β-arrestin/ src and β-arrestin/clathrin interaction and GPCR/CXCR4 receptor endocytosis [44,45]. Intriguingly, the β-arrestin1-S412D mutant, unable to bind clathrin and acting as a dominant negative inhibitor of receptor sequestration/internalization [41], enhances Notch3-induced CXCR4 cellsurface expression. We observed here that in N3-ICtg DP thymocytes, β-arrestin1 undergoes heavy Ser412phosphorylation and increased nuclear translocation, suggesting the impairment of its function toward CXCR4 internalization. Additionally to previous data [24] linking CXCR4 recycling to calcineurin-dependent expression of cortactin, we propose that Notch3 regulates function and expression of β-arrestin1 to prevent CXCR4 receptor internalization. Nevertheless, β-arrestin1 can bind and colocalize with cortactin [46], and we cannot exclude their convergent connection.
To go further beyond the Notch3 +high CXCR4 +high DP organ-wide dissemination, we investigated their migratory ability in in vivo NSG transplant experiments. Our results highlight that not simply DP, but selected N3 high DP thymocytes are efficient in positioning into BM, already at day1 post-injection. We report here for the first time how a wave of "thymus-derived" Notch3 +high CXCR4 +high DP-cells is able to home and engraft into BM of NSG recipient mice, while the ability of "bone marrow-derived" leukemic cells to infiltrate lymphoid organs and CNS in T-ALL was previously reported [24,25,47]. We show that high CXCR4 is per se insufficient, but requires an active Notch stimulus, Notch3 in our model, that possibly potentiates the SDF-1driven nest of Notch3 +high CXCR4 +high DP-cells into BM niches. Our results delineate a new mechanism by which a specific subset of DP T-cells, with a high Notch3 and CXCR4 co-expression and an elevated proliferation rate, are forced to egress from the thymus and fit to rapidly colonize BM and possibly peripheral organs (i.e., the spleen) during T-ALL progression.
In support of our hypothesis, in vivo CXCR4 antagonism by AMD3100 can contrast atypical DP infiltration in the BM of young N3-ICtg mice. Notably, the increased number of N3 high and the ratio N3 high /N3 low DP cells infiltrating the spleen further highlight the environmental requirements to sustain Notch3 +high CXCR4 +high DP "fitness" in T-cell leukemia propagation. It has been previously shown that primary explants of T-ALL are prone to apoptosis unless stromal stimuli, like Notch ligands (Delta-like1) and SDF-1, are present [25,48]. In keeping with this we observed a reduced spontaneous apoptosis in SDF-1-treated N3 high DPcells, while apoptosis of N3 low DP-cells resulted unaffected by SDF-1 treatment. The incompetence of N3 low DP T-cells to home and engraft into BM and spleen, at day1 and day10 post-injection, underlies the need of a strong Notch signal for committing CXCR4-expressing leukemic cells to propagate and expand in peripheral organs.
In conclusion, our results highlight a new perspective function of CXCR4 combined with Notch3 in selecting and mobilizing "pre-leukemic" DP cells that easily fit to disseminate in early T-ALL progression. Moreover, the Notch3/CXCR4 crosstalk suggests the attractive possibility of combined therapy protocols to precociously target T-ALL cells.

Materials and methods
Mice N3-ICtg mice [18] were bred and maintained as reported in [49]. NOD.Cg-Prkdc scid (NSG) immune-compromised recipient mice (Charles-River Laboratories). Animal experiments performed with at least three animals of each genotype. The number of used mice is reported in each Figure legend. Experimental mice groups were based on age and genotype. No mice were excluded during the experiments. All animal experiments were performed according to Italian D.Lgs. n.26/2014 and European Directive 2010/63/ UE.

Cell culture
Hek293T, TALL-1 cells were maintained as described elsewhere [4,50] and all are mycoplasma-free.

RT-PCR/qRT-PCR
Total RNA extraction and reverse transcription-polymerase chain reaction (RT-PCR) were described [52]. CXCR4 and HPRT mRNA levels were determined by TaqMan quantitative real-time RT-PCR (qRT-PCR) performed on cDNA following manufacturer's instructions (Applied-Biosystems). Data analysis by ΔΔCt method (HPRT use for normalization of mRNA levels).

Cell sorting for in vivo cell transfer experiment
Thymocytes of 8-week-old N3-ICtg mice were stained with antibodies abovementioned in flow cytometry.

In vivo administration of AMD3100
Three-to four-week-old N3-ICtg mice were IP injected daily with AMD3100 (10 mg/kg; Sigma-Aldrich) or vehicle (PBS). FACS analysis of BM cells was performed at day1 and day10. Antibodies are mentioned above in flow cytometry [53].

Statistical analysis
Comparison of two groups using unpaired t-test or Mann-Whitney test was done. Comparison of more groups using one-way ANOVA (Tukey's post-test and Kruskal-Wallis test) was done. A p-value of <0.05 was considered as statistically significant (*p < 0.05, **p< 0.01, ***p < 0.001, ****P < 0.0001; § § p < 0.01; § § § § p < 0.0001). At least three independent experiments are reported as mean ± SD (the repeat number was increased according to effect size or sample variation). Estimation of sample's size considering variation and mean of samples was done. No statistical method was used to predetermine sample size. No animals/samples were excluded from the analysis. Statistical analysis was conducted with PRISM-program (GraphPad).

In silico analysis of T-ALL patients' deposited data
Blood samples from 117 pediatric T-ALL patients [42] were selected and analyzed for correlation between CXCR4 and ARRB1. Expression probe sets 217028_at representing CXCR4 and 222912_at representing ARRB1 were used.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.