In vitro co-culture experiments provided the first demonstration that CTLA-4 and PD-1 control T-cell motility and especially the interaction time between T cells and antigen-presenting cells (APC) ^[1][2][3][42,43,44]. As stated previously, when T cells encounter sufficient activating peptide–MHC complexes they decelerate to form stable conjugates with APC. This stop signal is dependent on the strength of signaling pathways triggered by TCR engagement. Thus, any mechanism able to attenuate TCR-induced signals would in principle reduce the interaction time of T cells with APC or target cells. Accordingly, it was shown that engagement of immune checkpoint proteins with their ligands prevented T cells from forming long-lived interactions with APC cells ^[1][4][5][6][42,45,46,47]. Conversely, when the bindings of PD-1 or CTLA-4 with their ligands were blocked, T cells regain stabilized contact with APC ^[1][6][42,47]. Blocking TCR-induced pathways during a normal immune response (e.g., antibacterial or antiviral) has a clear meaning. First, it reduces a prolonged activation of T cells that might be detrimental with a risk of autoimmunity. Second, it enables lymphocytes to disengage from APC and retrieve an active scanning mode.

The impact of CTLA-4 and PD-1 on T-cell motility was then evaluated using two-photon microscopy and intravital imaging in mice with tumors or virally infected. In 2012, Mike Dustin’s lab reported the effects of anti-CTLA-4 antibody on tumor growth and on T-cell motility in tumors ^[7][48]. It was shown that the treatment of tumor-bearing mice with anti-CTLA-4 antibody failed to induce tumor regression. At the same time, this therapy reduced contact time between T cells and target cells (likely decreasing tumor cell killing), leading indirectly to an increase in intratumoral T-cell motility. However, when combined with radiation therapy, anti-CTLA-4 antibody treatment provoked both tumor regression and T cells to stop on tumor cells (so increasing contact time and favoring tumor cell killing) in an MHC-I and NKG2D-mediated manner, in line with the in vitro data ^[7][48]. The notion that immune checkpoint inhibitors (ICI) restore stabilized conjugates between T cells and tumor cells has recently been confirmed in a mouse melanoma model after the adoptive transfer of cytotoxic T cells. In such a setting, the treatment with anti-PD-L1 and CTLA-4 decreased intratumoral T-cell motility consistent with target engagement ^[8][49]. In contrast, Pentcheva-Hoang et al. have reported increased motility of T cells when anti-CTLA-4 antibodies were administered with GVax, a cellular vaccine producing GM-CSF ^[9][50]. Similar findings (i.e., an increase in T-cell motility) were reported in the spleen of mice chronically infected with LCMV and treated with anti-PD-1 antibodies ^[10][51].

Several reasons can explain these contradictory findings. First, the motility of T cells in response to ICI may differ depending on their spatial localization. In tumor islets where T cells are in contact with cancer cells, PD-1 and CTLA-4 likely act by interrupting TCR-mediated stop signals. However, this scenario might be very different for T cells located in the stroma. In this compartment, T cells interact with stromal cells including macrophages and dendritic cells which express PD-1 ligands, namely PD-L1 and PD-L2. Yet these cells do not necessarily present tumor antigens to T cells. The impact of PD-1 or CTLA-4 engagement on T-cell motility in the absence of TCR-induced signals is not known for the moment. It is possible that by reducing the signals triggered by chemokine receptors, immune checkpoint proteins mediate inhibitory effects on T-cell migration as opposed to the T-APC disengagement. In such a setting, ICI treatment would increase T-cell trafficking in the stroma and decrease it in the tumor islets.

Second, the duration of the treatment with ICI before measuring the impact on T-cell motility is another important parameter. In this regard, the pro-migratory effects described above were usually observed after several hours-days of treatment with blocking antibodies but not after an acute intervention. In such conditions, it is very difficult to make sure that the enhanced motility is a direct consequence of ICI on T cells. It is known that by relieving the brake of T-cell suppression, fully activated lymphocytes produce more inflammatory cytokines (e.g., IFNγ) known to reprogram the tumor microenvironment (TME), which might be more permissive to T-cell migration. For instance, chemokines such as CXCL9/10, upregulated by IFNγ, are known to attract activated T cells and promote their migration ^[11][12][52,53]. This is consistent with data showing that cancer patients responding to ICI exhibit increased numbers of T cells within tumors ^[13][14].

Finally, depending on their differentiation state and level of PD-1 expression, T cells may be affected differently by PD-1 blockade. For instance, discrepancies in the motility of CD4+ and CD8+ T cells following ICI blockade could be explained by their epigenetic profiles ^[14][54]. In particular, TOX and NR4A, which are epigenetic regulators of T cells exhaustion and anergy, are also known to impact immune checkpoint expression ^[15][16][55,56]. Interestingly, several studit waes suggested that PD-1 expression is associated with chronic exposure to antigens due to infections or cancers leading to TOX- and NR4A-dependent immune exhaustion of T cells ^[14][54]. However, the implications of PD-1 engagement and blockade of those T cells considered as exhausted on motility are unclear and would depend on the context (acute vs. chronic state for example). For instance, in mice with LMCV persistent infection where T cells are considered exhausted, CD8+ and CD4+ T cells are nearly static in the white pulp. In this setting, PD-1 blockade results in an increase in CD8+ T-cell motility. In addition, transcriptomic analysis performed on human TILs purified from lung tumors indicates that T cells expressing high levels of PD-1 harbor an altered expression of genes involved in cell migration ^[17][57].

Another important aspect to consider is the potential response of CAR T cells to ICI with regards to motility and interaction with target cells. Indeed, the onstudie that was cited above only focused on analyzing the impact of ICI on the dynamics of endogenous tumor-infiltrating T cells (TILs). However, as previously stated ^[18][58], signaling from CAR and TCR receptors differ greatly, and the same could be true for their response to ICI.

First, a suboptimal activation of engineered T cells due to inefficient CAR-mediated signaling compared to TCR-based T cells may impact on CAR T cell motility and their ability to form durable and efficient contacts with target tumor cells. In that respect, CAR T cells have been shown to form a disorganized synapse with tumor cells as opposed to the mature synapse of non-modified T cells ^[19][37]. With regards to T-cell motility within tumors, no direct comparison has been made between CAR T cells and TILs.

Second, although encouraging results have been observed for a combined CAR T cell and anti-PD-1 therapy ^[20][21][59,60], the effects of ICI treatments on cell motility have been mainly assessed for endogenous TILs, and their impacts on CAR T-cell motility have still to be determined, especially by comparing them to those observed for endogenous TILs.

Altogether, it wahose studies showed that th the importance of considering the timing, the location, and the type of T cells when interpreting the effects of anti-PD-1 and anti-CTLA-4 blocking antibodies.

Once matured from the thymus, naïve T cells enter the circulation and start patrolling the tissues looking for specific antigens. At this stage, these cells are considered metabolically quiescent. Indeed, they engage catabolic reactions mainly to sustain cell survival, and the overall levels of both glycolysis and oxidative phosphorylation (OXPHOS) are very low ^[22][61]. Upon antigen encounter, T cells undergo a massive metabolic remodeling, greatly increasing anabolism to sustain their clonal expansion ^[22][61]. Particularly, T cells can activate a positive feedback loop between signaling pathways and glycolytic enzymes ^[23][62], whose expression is greatly enhanced, thus favoring a fast ATP production and the generation of glycolytic intermediates used to sustain nucleotide and lipid synthesis. However, although aerobic glycolysis was long believed to be a hallmark of activated T cells, T cells also display a significant increase in OXPHOS upon activation, although to a lesser extent than glycolysis ^[24][63]. Partly, this is required to increase mitochondrial reactive oxygen species (ROS) production, which controls several signaling pathways upon activation ^[25][64]. The relative contribution of glycolysis and OXPHOS to T-cell metabolism seems to be intricately interconnected with the differentiation route of activated T cells ^[26][27][65,66]. Indeed, while in effector CD8+ T cells glycolysis is important to promote cytokine expression and effector functions ^[28][67], in activated CD8+ T cells primed towards a memory phenotype a relative higher engagement of OXPHOS is associated with the long-term survival of these cells ^[29][68]. Particularly, short-chain fatty acids (partly derived from intestine microbiota) have been shown to sustain memory differentiation by increasing OXPHOS through increased glutamine utilization and fatty acid catabolism ^[30][69]. Moreover, it has been shown that distinct modes of mitochondrial metabolism support either differentiation (through citrate export and electron transport chain (ETC) complex I) or effector functions (though ETC complex II) in CD4+ T helper1 (T_H1) cells ^[31][70].

The eukaryotic cytoskeleton is composed mainly of two protein polymers, namely actin and tubulin. These proteins can dynamically polymerize into filaments by consuming energy (ATP for actin, and GTP for tubulin). Such cytoskeletal rearrangements are crucial to sustaining T-cell motility, which therefore is a highly energy-demanding process. ATP can be generated in the cell either through glycolysis (fast rate but poor yield) or OXPHOS (low rate but high yield, although only within mitochondria). GTP can be generated by the enzyme succinyl-CoA-synthetase during the Krebs cycle, or it can be obtained from ATP thanks to the enzyme nucleoside-diphosphate-kinase (NDPK).

The link between metabolic pathways and cell migration has been well studied in a number of cells including cancer cells and spermatozoa. Interestingly, in cancer cells, several glycolytic enzymes have been described to interact with cytoskeletal proteins, such as hexokinase, phosphofructokinase (PFK), and pyruvate kinase (PK) ^[32][71]. The latter is one of the two glycolytic enzymes that produce ATP (the other one being the phospho-glycerate kinase). PFK1 (controlling the main rate-limiting step enzyme of glycolysis) can be targeted to degradation upon actin depolymerization ^[33][72]. Of note, in cells moving through actin protrusions (such as filopodia and lamellipodia of endothelial cells) at the leading edge, several glycolytic enzymes, such as PK and 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase-3 (PFKFB3), associate with the polymerized actin (F-actin) at the leading edge ^[34][73]. Overall, in cancer cells glycolysis seems to be preferred over OXPHOS to sustain a highly dynamic process (i.e., actin remodeling) occurring simultaneously throughout the whole leading edge by providing ATP locally.

By contrast, a different motility is observed in spermatozoa. In the human spermatozoa, mitochondria are enriched in the midpiece of the tail and supply the energy required for the motility together with other factors. The first is the concentration of glucose and fructose in the sperm. While glucose sustains glycolytic ATP production and increases spermatozoid mobility ^[35][74], fructose (provided by seminal fluid) appears essential to enhance a vigorous motility enabling spermatozoa to reach the egg for fertilization ^[36][75]. Second, the citric acid contained in the sperm positively influences the number and motility of spermatozoa ^[37][76]. Third, AMPK (5′-AMP-activated protein kinase), the key energy sensor promoting mitochondrial activity and OXPHOS, is essentially localized in the flagellum and acrosome ^[38][77], and it is involved in the regulation of sperm motility and acrosome reaction ^[39][40][78,79]. AMPK is also involved in the management of lipid peroxidation and gamete antioxidant enzymes ^[39][78]. In line with this, AMPK inhibition significantly decreases the percentages of motile and rapid spermatozoa ^[40][79], while AMPK activators (AICAR, metformin) significantly improved sperm motility ^[38][77].

Contrary to epithelial cell-like and traction-based mesenchymal motility, T cells rely mainly on an ameboid-like migration in tissues, which is less dependent on actin-derived filaments at the leading edge and requires a functional uropod at the rear edge, where the actomyosin motor is located ^[41][80]. Here, this motor acts as a propulsor to push the cell forward by generating a retrograde actin flow. Of note, in a 3D environment, T cells can generate such retrograde actin flow without the presence of adhesion molecules, by exploiting only the topographical features of the substrate, thus being autonomous in their locomotive behavior ^[42][81]. In addition, an efficient T-cell motility in tissue seems to be proportional to their cortical actomyosin contractility and it can be improved by direct activation of RhoA ^[43][82], a key regulator of actomyosin contractility. Remarkably, only a few studies have addressed the metabolic requirements of migrating T cells so far. Glycolysis has been reported to modulate T-cell motility. Indeed, lactate (the end-product of glycolysis) can inhibit CXCL10-driven T-cell motility. Interestingly, extracellular sodium lactate and lactic acid were shown to inhibit the motility of CD4+ and CD8+ T cells, respectively, through different subset-specific transporters which are differentially expressed by CD4+ (Slc5a12) and CD8+ (Slc16a1) T cells ^[44][83]. While lactate reduces CXCL10-driven CD4+ T-cell motility by interfering with glycolysis, its effect on CD8+ T-cell motility seems to be glycolysis-independent ^[44][83]. In addition, the CCL5 chemokine has been reported to increase glucose uptake, glycolysis, and ATP production in T cells to meet the increased energy demand during cell migration ^[45][84], presumably by upregulating mTOR, which is also required to promote the expression of chemotaxis-related proteins. Similarly, stimulation of T cells with either CXCL12 or CCL19/21 chemokine activates ERK1/2 and the mitochondrial pro-fission protein Drp1 ^[46][47][85,86], two molecules known to sustain glycolysis in T cells ^[47][48][86,87]. Moreover, glycolysis promotes cell migration of regulatory T cells, too ^[49][88]. In these cells, glycolytic enzyme glucokinase fuels cytoskeletal rearrangements by associating with actin. Apart from glycolysis, OXPHOS seems to play a positive role too in supporting T-cell migration. Mitochondria specifically localize at the uropod during T-cell migration ^[47][50][86,89]. Indeed, chemokines promote the fragmentation of mitochondria through the activation of the mitochondria pro-fission protein Drp1 ^[47][86]. Subsequently, fragmented mitochondria can be transported along microtubules towards the cell rear edge ^[47][86], where it has been hypothesized that they may act to supply ATP locally to sustain myosin activity ^[50][89]. However, during phases of slow locomotion, it has been observed that mitochondria may accumulate also at the leading edge of migrating CD4+ T cells in response to chemokine stimulation ^[51][90].

In sum, both glycolysis and OXPHOS seem to play a role during T-cell migration, although the relative contribution of the two processes is not currently understood. ItWe could be postulated that glycolysis and OXPHOS may be alternatively engaged according to the specific locomotion of a T-cell, i.e., slow locomotion (during which the T-cell needs to accurately scan the surrounding environment to find APCs or target cells) or fast motility (required to efficiently infiltrate into tissue). However, it should be noted that all these studies have been performed using in vitro 2D systems. Therefore, although some of these findings may extend also to in vivo conditions, the actual metabolic requirements of a T-cell in a real 3D environment are completely unknown.

A solid TME is characterized by different subpopulations of cells in addition to cancer cells, including both pro-tumoral and anti-tumoral immune and stromal cells. Since cancer cells are constantly trying to grow and proliferate, the TME is a tissue greedy for nutrients, which are consumed at high levels by tumoral cells. Consequently, several nutrients required for T cells to proliferate and for their effector functions may not be available in the adequate range, such as glucose, glutamine, tryptophan, and arginine ^[52][92]. In parallel, several toxic metabolites can accumulate in the TME as waste products of cancer cell metabolism (such as lactate) or actively produced by immune cells (such as the derivatives of tryptophan metabolism) ^[53][54][93,94]. Although several studit has beenes have elucidated that hhow all these alterations may negatively impact T-cell functions (for a review see ^[55][95]), either by promoting a pro-tumoral suppressive phenotype or by inhibiting effector functions, we currently, it was ignored that ignore if and how these metabolic alterations impact T-cell motility. Based on in vitro data, it can be we can speculated that the reduced availability of glucose and the consequent reduction in glycolytic rates observed in tumor-infiltrating T cells ^[56][96] may be associated with reduced motility. In addition, a defective functionality of some glycolytic enzymes in TIL has been reported that may be further responsible for reducing the glycolytic rate, and thus cell motility ^[57][97]. Both tryptophan and arginine have been described as positive regulators of cell migration ^[58][59][98,99]. Therefore, their reduced amount in the TME may potentially decrease T-cell migration, although specific onstudies are lacking. In addition, several waste products accumulating in the TME may inhibit T-cell motility, thus favoring tumor growth and tumorigenicity. For example, lactate can inhibit T-cell motility in vitro ^[44][83], and its accumulation in the TME ^[54][94] may play a similar role. Finally, the lack of oxygen, a hallmark of growing tumors, can impede the migration of T cells dependent on oxidative phosphorylation. Naive T cells in murine lymph nodes actively rely on a high level of oxygen to migrate ^[60][100]. Likewise, in melanoma mouse tumors TIL were shown to concentrate and migrate more actively around blood vessels as compared to avascular areas ^[61][101].

T cells within the TME are frequently characterized by the activation of inhibitory co-receptor pathways, such as PD-1, LAG-3, CTLA-4, and so on. It has beenetc. Several studies have shown that these signaling pathways may negatively regulate T-cell metabolism. For example, PD-1 signaling can reduce both glycolysis and OXPHOS in T cells ^[62][63][102,103]. Indeed, PD-1 engagement on T cell surface during activation can inhibit both the PI3K/Akt/mTOR and MAPK/ERK signaling pathways, both required for an efficient glycolytic engagement in T cells ^[63][64][103,104]. PD-1 has been also reported (i) to inhibit the activation of the mitochondrial pro-fission protein Drp1 ^[46][85], which is required for T cell motility ^[47][86], and (ii) to promote disassembly of the mitochondrial cristae ^[62][102], which are required for OXPHOS-dependent ATP generation. Similarly, CTLA-4 engagement on T cell surface dampens the activation of PI3K/Akt/mTOR ^[65][105], and therefore reduces glycolysis. Knock-out of TIM-3 in an in vitro T cell model reduces glucose uptake and glycolysis ^[66][106]. Overall, given the connections mentioned above between metabolism and migration in T cells, the effects of these inhibitory co-receptors on T-cell motility described above may also depend on the concomitant alteration of T-cell metabolism. However, no studies have so far no one addressed this point.

In sum, metabolic alterations in the TME may have the potential to impact T and CAR T cell motility within human solid tumors. Although specific onstudies were addressing these points are still lacking, these concepts should be considered to understand how the TME is able to reduce the effectiveness of CAR T cells and to develop new strategies to improve current immunotherapy approaches based on engineered T cells. It is also worth noting that focusstudying the influence of metabolism on T-cell motility has been performed so far on endogenous T cells. Therefore, the role of metabolism in regulating the dynamic of CAR T-cells is currently ignored, as well as the potential impact of metabolic alterations in the TME on CAR T cell motility.

As stated previously, CAR T cells need to form productive conjugates with their targets via the assembly of an immunological synapse to achieve control of tumor growth ^[67][5]. Up to now, little is known about the mechanisms that regulate CAR T-cell interaction with tumor cells. Previously, it wastudies performed with non-modified T cells have underlined that, in addition to the recognition of specific peptide-major histocompatibility complex (pMHC) molecules via the T-cell receptor (TCR), engagement of adhesion and costimulatory molecules with their respective ligands is mandatory to trigger efficacious antitumor T-cell activities. Among adhesion/costimulatory molecules, attention has mostly focused on integrins, in particular, lymphocyte function-associated antigen-1 (LFA-1, CD11a/CD18 or αLβ2) and CD103 (αEβ7), which play important roles in T-cell-target cell adhesion through interaction with their respective ligands, intercellular adhesion molecule-1 (ICAM-1 or CD54) and the epithelial cell marker E-cadherin ^[68][69][107,108]. CD2, which binds to CD58 (LFA-3) on target cells, also acts as an adhesion/costimulatory molecule that provides signals to amplify TCR signaling ^[70][109].

Although the central role of LFA-1 and CD103 in stabilizing interactions between naturally occurring T cells and tumor target cells is well established, their contribution to CAR T-cell activity is ill-defined. In native T cells, the adhesive properties of integrins are regulated via conformational activation and clustering, initiated by an “inside-out” signaling process emanating at least in part from the TCR ^[71][35]. Given the distinct nature of CAR constructs, a key open question pertains to the ability of CAR T cells to properly activate integrins upon tumor cell recognition. This is especially relevant regarding recent onstudies showing that many aspects of CAR signaling are unique and distinct from endogenous TCR signaling ^[72][110]. In particular, the finding that CAR T cells form a disorganized immunological synapse when contacting cancer cells might be a direct consequence of a suboptimal activation of leukocyte integrins ^[19][37]. Inefficient integrin activation upon CAR ligation would be particularly critical in conditions of limited expression of adhesion molecules on the surface of tumor cells, in particular ICAM-1, which is frequently downregulated by cancer cells to evade CD8 T-cell mediated destruction ^[73][111]. As important support of this model, itwe has beenve obtained that recently published data showing that CAR T efficacy is strongly dependent on the level of tumor cell ICAM-1 ^[74][112]. By monitoring intracellular Ca²⁺ responses (a proxy of T-cell activation) of CD20 CAR T cells during their interaction with malignant B cells from chronic lymphocytic leukemia patients, a clear positive correlation was observed between the percentage of activated CAR T cells and ICAM-1 density on tumor cells. A similar process was also observed with EGFR CAR T cells during their contact with carcinoma cells from several cell lines. Under low ICAM-1 expression and despite efficient target antigen expression, EGFR CAR T cells were unable to form productive conjugates with carcinoma cells. Recently, the role of LFA-1-ICAM-1 interaction in CAR T cell activation was confirmed in two independent studies. In the first, it wahe authors exploited that the property of the extracellular magnesium (Mg²⁺) to bind to LFA-1 and stabilize its active conformation. Under low Mg²⁺ levels, CAR T-cell activation and cytotoxicity against tumor cells were considerably reduced. Most importantly, in lymphoma patients treated with CD19 CAR T cells, low serum magnesium levels correlated with poor prognosis ^[75][113]. The importance of ICAM-1 expression on cancer cells for CAR T-cell activation was revealed in a CRISPR-based screen performed in a multiple myeloma cell line. Invalidation of ICAM-1 gene in tumor cells led to resistance to BCMA CAR T cells ^[76][114]. Along the same lines, Majzner’s lab has reported that the loss of CD58 at the surface of tumor cells is associated with CD19 CAR-T cell failure in patients with large B-cell lymphoma (ASH 2020 annual meeting). Thus, the expression of ICAM-1, and possibly other adhesion molecules such as CD58, emerges as a pivotal factor that sets the threshold of CAR T-cell responsiveness.