Mitochondrial C1QBP is essential for T cell antitumor function by maintaining mitochondrial plasticity and metabolic fitness

The metabolic stress present in the tumor microenvironment of many cancers can attenuate T cell antitumor activity, which is intrinsically controlled by the mitochondrial plasticity, dynamics, metabolism, and biogenesis within these T cells. Previous studies have reported that the complement C1q binding protein (C1QBP), a mitochondrial protein, is responsible for maintenance of mitochondrial fitness in tumor cells; however, its role in T cell mitochondrial function, particularly in the context of an antitumor response, remains unclear. Here, we show that C1QBP is indispensable for T cell antitumor immunity by maintaining mitochondrial integrity and homeostasis. This effect holds even when only one allele of C1qbp is functional. Further analysis of C1QBP in the context of chimeric antigen receptor (CAR) T cell therapy against the murine B16 melanoma model confirmed the cell-intrinsic role of C1QBP in regulating the antitumor functions of CAR T cells. Mechanistically, we found that C1qbp knocking down impacted mitochondrial biogenesis via the AMP-activated protein kinase (AMPK)/peroxisome proliferator-activated receptor gamma coactivator 1-alpha signaling pathway, as well as mitochondrial morphology via the phosphorylation of mitochondrial dynamics protein dynamin-related protein 1. In summary, our study provides a novel mitochondrial target to potentiate the plasticity and metabolic fitness of mitochondria within T cells, thus improving the immunotherapeutic potential of these T cells against tumors.


Introduction
Mitochondria are highly plastic organelles which are critical in modulating T cell development, fate determination, and antitumor responses [1][2][3][4]. In order to functional desirably, it is critical for the mitochondria within T cell to remain plastic and metabolically fit in response to a variety of metabolic perturbations. Mitochondrial plasticity including mitochondrial dynamics, metabolism and biogenesis is tightly linked to mitochondrial structure, function, activity, 1 3 and homeostasis [2,[4][5][6][7]. Robust mitochondrial plasticity endows T cells, and especially tumor-infiltrating T lymphocytes (TILs), with persistent and durable antitumor function, while impaired mitochondrial plasticity causes TILs to exhibit metabolic insufficiency, exhaustion, and hypofunction [6,8,9]. However, the causal relationship between T cell mitochondrial plasticity and T cell antitumor immune function in response to metabolic stress of the tumor microenvironment (TME) still remains poorly understood.
The complement C1q binding protein (C1QBP), predominantly located in the mitochondrial matrix, has been correlated with cellular bioenergetics and mitochondrial function in tumor cells. Specifically, C1QBP has been implicated in metabolic reprogramming via regulation of mitochondrial morphology and metabolism, which is further connected to tumor transformation and metastasis [10][11][12][13][14]. Nevertheless, metabolic adaptation via the mitochondria is required not only for tumor malignant progression but also for tumor-specific T cells to perform their antitumor immune responses [15][16][17]. Given that C1QBP is indispensable for mitochondrial structure and function, exploring the exact role of C1QBP in regulation of T cell mitochondrial plasticity and the subsequent impact of antitumor immunity will be of interest.
Mitochondria are constantly performing cycles of fission and fusion in order to meet the ever-changing metabolic needs of their host cells [18][19][20][21][22]. Among the many other impacts of this behavior, increasing evidence has suggested that such mitochondrial dynamics help to determine T cells' fate through metabolic programming. At the same time, it has been shown that mitochondrial fusion in T cells can improve adoptive antitumor cell therapy responses [2,3]. In this study, we wanted to interrogate the effect and mechanism of C1QBP on the mitochondrial dynamics of T cells. Besides, the previous study reported that C1QBP has also been implicated in regulation of mitochondrial oxidative phosphorylation (OXPHOS): C1qbp-deficient mouse embryonic fibroblasts (MEFs) showed impaired OXPHOS and ATP production, which resulted in the delayed embryonic development [23]. However, whether C1QBP could be involved in the manipulation of T cell metabolic fitness through mitochondria-mediated OXPHOS still remains unclear.
Mitochondrial biogenesis has not only been implicated in the maintenance of mitochondrial mass, but also in the regulation of T cell antitumor immune function. TILs notoriously exhibit persistent loss of mitochondrial biogenesis in both human patients and mouse models, while proper maintenance of mitochondrial biogenesis is required to recover T cell antitumor responses [24,25]. Although C1QBP is reported to be associated with the formation of mitochondrial ribosomes and mitochondrial DNA (mtDNA)-encoded mitochondrial protein translation [23,26], here we want to explore whether C1QBP regulates T cell mitochondrial biogenesis and mass to impact TILs exhaustion in the nutrient-and oxygen-deprived TME.
Finally, in order to assess the potential role of C1QBP in CAR T cell therapy, we evaluated the impact of C1QBP on mitochondrial plasticity and further delineated the role of C1QBP on CAR T cells tumor infiltration, central memory cell formation, and exhausted phenotype development, thus impacting their antitumor immune response.

Mice
C1qbp heterozygous mice were donated by Dr. Yanping Zhang from the Department of Radiation Oncology at Lineberger Comprehensive Cancer Center, University of North Carolina, USA. C1qbp heterozygous mice were generated as previously described [27]. All mice were housed and bred in the specific pathogen-free (SPF) animal facility of the Experimental Animal Center in Xuzhou Medical University. All experimental animal procedures were performed in compliance with institutional ethical requirements and approved by the Committee of Xuzhou Medical University for the Use and Care of Animals.

Immunoblot analysis
Lysates of T cells were prepared in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.2% Triton X-100, 0.5% NP-40, and protease inhibitor cocktail). Samples were resolved by SDS/PAGE (12% gel) and transferred to nitrocellulose membranes. The transferred membranes were blocked with 5% (w/v) non-fat dried skimmed milk powder in PBST and immunoblotted using the commercially available primary antibodies listed below. Membranes were then washed with PBST, incubated with the appropriate HRP (horseradish peroxidase)-conjugated secondary antibodies in 5% (w/v) non-fat dried skimmed milk powder/PBST, and visualized using enhanced chemiluminescence reagents (Proteintech) according to the manufacturer's instructions.

Immunofluorescence
For immunofluorescence imaging, T cells were fixed in 4% paraformaldehyde for 15 min at room temperature and permeabilized in 0.2% Triton X-100 for 5 min at 4 °C. Fixed and permeabilized cells were blocked for 30 min in 0.5% BSA blocking buffer diluted in PBS, incubated with the appropriate primary antibody overnight at 4 °C, and then incubated with Alexa Fluor secondary antibodies (Life Technology Inc.) for 1 h at room temperature. Nuclei were stained by (4′,6-diamidino-2-phenylindole) (DAPI). All stained cells were mounted with fluorescence mounting medium (Dako, Carpinteria, CA, USA) and then were observed using the ZEISS LSM 880 confocal fluorescence microscope. Images were analyzed with ImageJ software.

Flow cytometry and analyzation
For preparation of tumor single cell suspensions, tumor tissues from C1qbp +/+ and C1qbp ± mice were mechanically dissected and washed with PBS. The tumor cell suspension was passed through a sterile 40 µm Nylon Filter (BD Falcon). The single cell suspension was then stained by PerCPanti-mouse-CD4 and APC-anti-mouse-CD8α antibodies to detect the percentage of tumor-infiltrating CD4 + and CD8 + T cells. Next, in order to examine tumor-infiltrating T cell exhaustion, the single-cell suspension was also stained by PE-anti-mouse-PD-1, PE-anti-mouse-LAG-3, PE-anti-mouse-Tim-3 antibodies, respectively. In parallel, C1qbp +/+ and C1qbp ± T cells were activated to anti-CD3/ CD28 antibodies or tumor single-cell suspensions, and were stained with MitoTracker Green (M7514, ThermoFisher) or MitoTracker Deep Red (M22426, ThermoFisher) to measure mitochondrial mass. Stained cells were detected by the FACS CantoTM II Flow Cytometer (BD biosciences), and FACS data were analyzed using FlowJo software (TreeStar).

Seahorse extracellular flux analysis
For analysis of the metabolic phenotype of C1qbp +/+ and C1qbp ± T cells, oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using the Seahorse XF-96 metabolic extracellular flux analyzer (Seahorse Bioscience). FACS-sorted T cells were planted in a 24-well plate coated with anti-CD3ε/CD28 antibodies and cultured with RPMI-1640 medium for 48 h. Activated T cells were then collected and spun on poly-D-lysine coated seahorse 96 well plates (2 × 10 5 cells per well) and preincubated at 37 °C for a minimum of 45 min in the absence of CO 2 . Cellular OCR (pmol/min) was measured under basal conditions and following treatment with 1 μM oligomycin, 2 μM trifluoromethoxy carbonyl cyanide phenylhydrazone (FCCP), 0.5 μM rotenone, and 0.5 μM antimycin (all from Sigma-Aldrich). Real-time analysis of extracellular oxygen and pH was conducted in a standard humidified incubator with 5% CO 2 . Mitochondrial function parameters of basal respiratory capacity, maximal respiratory capacity, proton leak, oxidative ATP turnover, and spare respiratory capacity were determined. At the same time, ECAR (mpH/min) was recorded under basal conditions and following the addition of 10 mM glucose, 1 μM oligomycin, and 50 mM 2-deoxyglucose (2-DG).
Glycolysis, glycolytic capacity, and glycolytic reserve were determined. The related quantification was analyzed in duplicate for at least three independent experiments.

qRT-PCR analysis of mRNA
Total RNA was isolated from T cells by TRIzol reagent (Invitrogen) following manufacturer's instruction, and then reverse-transcribed into cDNA with HiScript II Q RT SuperMix (Vazyme Biotech Co.). Quantitative real-time PCR was performed on LightCycler 96 real-time PCR system (Roche life Science) with AceQ qPCR SYBR Green Master Mix kit (Vazyme Biotech Co.). The relative mRNA expression of indicated genes was calculated based on the expression of β-actin. Ct values for each analyzed gene were normalized to β-actin and adjusted for amplification efficiency.

Construction of targeted huB7-H3 CAR T cells
In order to prepare targeted huB7-H3 CAR T cells, we first constructed the retroviruses containing huB7-H3 CAR. Procedures and conditions for construction of the targeted huB7-H3 CAR retroviruses were previously described [28]. Next, murine T cells were isolated by Pan T magnetic Microbeads (Miltenyi Biotec) from splenocytes obtained from C1qbp +/+ and C1qbp ± mice and then stimulated on plates coated with anti-CD3ε/CD28 antibodies for 24 h. Activated murine T cells were transduced with the above retroviral supernatants using retronectin-coated plates (Takara Bio Inc., Shiga, Japan). The transfection of murine T cells was performed according to previous protocols [29]. After removal from the retronectin-coated plates, T cells were expanded in complete RPMI-1640 medium supplemented with 10% FBS, 4 mM L-glutamine, 1% penicillin/streptomycin, 75 μM β-mercaptoethanol, 2 ng/ 1 3 mL IL-2, 10 ng/mL IL-7, and 5 ng/mL IL-15. After 48 h, CAR T cells targeting huB7-H3 were collected for use in the following in vivo and in vitro experiments.

ELISA
CAR T cells targeting huB7-H3 were co-cultured with huB7-H3 B16 tumor cells (2 × 10 5 cells/well) in 24-well plates at different effector to target ratios (E:T = 1:2 or 1:1) without the addition of exogenous cytokines. After 18 h, the supernatant was collected and cytokines IFN-γ and TNF-α were measured in duplicate by specific ELISA kits (Proteintech) following manufacturer's instructions.

Statistical analysis
Data were analyzed using Graph Pad Prism 6.0 software (La Jolla, CA). Statistical comparisons were performed using the Student's t-test. A P-value < 0.05 was considered statistically significant. All results are presented as the mean ± standard deviation.

Compromised antitumor immune response in C1qbp heterozygous mice
Given that C1QBP homozygous deletion results in early embryonic lethality [23], we wanted to investigate whether disruption of just one allele of C1qbp would impact T cell development and homeostasis. First, we observed that C1QBP protein levels of C1qbp ± mouse were significantly decreased relative to wild-type in a variety of organs including brain, lung, liver, colon, and kidney (S. Fig. 1A). Moreover, we isolated the T cells of C1qbp ± and C1qbp +/+ mice, and found that C1qbp ± T cells also exhibited significantly decreased levels of C1QBP protein in response to stimulation with anti-CD3/CD28 antibodies (S. Fig. 1B). At the same time, we found that the relative percentages of CD4 + and CD8 + T cells in the spleen, lymph nodes, and blood were comparable in both C1qbp ± mice and their C1qbp +/+ littermates. Additionally, C1qbp ± and C1qbp +/+ mice had similar immune compartments, including CD44 hi CD62L lo effector memory cells and CD44 lo CD62L hi naïve T cells (S. Fig. 2). These above data suggest that disruption of one allele of C1qbp is unable to significantly alter T cell development and homeostasis in a healthy, non-cancerous context. However, to further evaluate whether C1QBP might regulate anti-tumor immunity, we injected murine melanoma B16 cells subcutaneously into the flanks of C1qbp +/+ and C1qbp ± mice, and observed their tumor progression. As shown in Fig. 1A-C, C1qbp ± mice exhibited significantly more severe tumor progression than their C1qbp +/+ littermates. Uncontrolled tumor progression in C1qbp ± mice was further supported by expression of important tumor cell markers, including increased levels of Ki67 (tumor proliferation) and decreased levels of Puma (apoptosis) (Fig. 1D). Together, these data suggest that a single-allele deletion of C1qbp is sufficient to compromise the antitumor immune response.

Knocking down C1qbp dampens T cell antitumor responses
To explore the precise role of C1QBP in T cell antitumor immunity, we further investigated the relative proportions of CD44 hi CD62L hi central memory and CD44 hi CD62L lo effector memory T cells in the tumor and spleen. As shown in  Fig. 2A, Knocking down C1qbp led to a moderately lower ratio of CD44 hi CD62L hi central memory to CD44 hi CD62L lo effector memory T cells in the tumor, while C1QBP deficiency did not alter their proportion in spleen (S. Fig. 3). We further noticed a drastic decrease of T cells in the tumor tissues of C1qbp ± mice, which was correlated with the low expression of C1QBP in these T cells. As shown in Fig. 2B, the percentage of tumor infiltrating CD4 + T cells in C1qbp ± mice (1.87%) is dramatically less than that in C1qbp +/+ mice (5.71%). Similarly, the percentage of tumor infiltrating CD8 + T cells in C1qbp ± mice (1.67%) is much lower than that in C1qbp +/+ mice (7.08%). Moreover, immunohistochemistry staining also confirmed that knocking down C1qbp repressed tumor infiltration by both CD4 + and CD8 + T cells (Fig. 2C).

Fig. 2
C1qbp knockdown impacts T cell antitumor response. A Proportions of CD44 hi CD62L hi central memory and CD44 hi CD62L lo effector memory TILs in B16 tumors from C1qbp +/+ and C1qbp ± mice, as detected by flow cytometry. B Relative percentages of CD4 + and CD8 + TILs in B16 tumors from C1qbp +/+ and C1qbp ± mice, as assessed by flow cytometry. C Proportions of CD4 + , CD8 + , and CD3 + TILs in B16 tumors from C1qbp +/+ and C1qbp ± mice, as detected by immunohistochemistry. Scale bars, 50 μm (200×), 20 μm (400×). D Exhaustion markers (PD-1, Tim-3, LAG-3) expressed by CD4 + and CD8 + TILs in B16 tumors from C1qbp +/+ and C1qbp ± mice, as analyzed by flow cytometry. Bar graphs show the mean ± S.D. of three independent experiments (n = 3); *P < 0.05, **P < 0.01 Further analysis demonstrated that mice with one allele of C1qbp disrupted contained more dysfunctional and exhausted T cells relative to their wild-type littermates. Our results showed that C1qbp ± TILs exhibited higher expression of inhibitory receptors like PD-1, Tim-3, and LAG-3 than the corresponding C1qbp +/+ TILs (Fig. 2D). We also assessed the expression of PD-1 and Tim-3 in CD8 + T cells from the spleen of C1qbp +/+ and C1qbp ± mice, but observed similarly low expression of such exhaustion molecules in the peripheral T cells between these mice (S. Fig. 4). This distinction suggests that the impact of C1QBP on T cell exhaustion is mainly restricted to the TME, where TILs are exposed to a great deal of metabolic stress. Our results revealed that the intact expression of C1qbp is important for prevention of TILs exhaustion and hypofunction.
Together, knocking down C1qbp dampened T cell antitumor immune responses by decreasing the proportion of CD44 hi CD62L hi central memory T cells, diminishing the tumor-infiltrating T cells, and exacerbating their exhausted phenotype.

Knocking down C1qbp impairs T cell mitochondrial metabolism
As C1QBP has been reported to regulate cellular bioenergetics and mitochondrial function [23,30,31], we hypothesized that C1QBP may also play an important role in T cell mitochondrial metabolism. By measuring OCR and ECAR after stimulation with anti-CD3/CD28 antibodies for two days, we found that C1qbp ± T cells possessed the lower basal respiration and the maximal respiration but the higher glycolytic capacity and glycotic reserve than the corresponding C1qbp +/+ T cells, indicating that knocking down C1qbp suppressed mitochondria-mediated OXPHOS rather than glycolysis occurring in the cytoplasm (Fig. 3). In addition, C1QBP deficiency resulted in T cells with more proton leak and less ATP production, implying that C1QBP may be involved with the maintenance of the integrity of the mitochondrial membrane, thus preferring the catabolic pathways rather than the anabolic pathways.
Spare respiratory capacity (SRC) is an important marker used to assess the extra mitochondrial capacity available to produce energy, especially under the conditions of metabolic stress and nutrient depletion [32][33][34]. Enhancement of SRC endows T cells with long-term survival and persistent immune function, and is particularly applicable in the metabolically stressed, nutrient-deprived TME. Here, we found that C1qbp ± T cells exhibited lower SRC levels compared to the corresponding C1qbp +/+ T cells, suggesting that down-regulation of C1QBP expression attenuated the extra mitochondrial capacity of these T cells. Therefore, knocking down C1qbp repressed T cell mitochondrial OXPHOS and ATP production, aggravated mitochondrial proton leak, and diminished SRC levels, which would collectively be expected to impair T cells adaptation to the TME.

Knocking down C1qbp impacts T cell mitochondrial morphology
Previous studies have shown that T cell metabolic programming can be attributed to mitochondrial remodeling [2,35,36]. Here, we wanted to investigate whether reduced C1QBP expression would induce the alteration of T cell mitochondrial morphology. First, we examined the mitochondrial morphology of C1qbp ± or C1qbp +/+ T cells in response to T cell receptor (TCR) stimulation with anti-CD3/CD28 antibodies, as before. Our results demonstrated that activated C1qbp +/+ T cells possessed predominantly elongated and fused mitochondria, especially after 3 days, while C1qbp ± T cells possessed more fragmented and punctate mitochondria, which was consistent with the lower fluorescence intensity of mitochondrial marker Cytochrome C (Fig. 4A, B). At the same time, C1qbp ± T cells also had less mitochondrial mass compared with C1qbp +/+ T cells, as shown in Fig. 4C.
The highly dynamic fusion and fission processes determining mitochondrial morphology are controlled by a series of regulatory proteins including dynaminrelated protein 1 (Drp-1), optic atrophy-1 (OPA1), and mitofusion1/2(Mfn1/2) [37][38][39][40]. Specifically, fission is regulated by the GTPase Drp1, which is translocated from cytoplasm to the outer mitochondrial membrane (OMM) upon phosphorylation to split mitochondria [37,38]. Conversely, fusion is controlled by three-dynamin family GTPases including Mfn1/2 on the OMM and OPA1 on the inner mitochondrial membrane (IMM) [40][41][42]. To explore the potential mechanisms by which C1QBP regulates T cells mitochondrial morphology, we detected the protein levels of mitochondrial fission protein and fusion proteins, in both C1qbp ± and C1qbp +/+ T cells. As shown in Fig. 4D, we found that C1QBP obviously enhanced not only the expression of the mitochondrial fission protein, Drp1, Fig. 4 C1qbp knockdown alters T cell mitochondrial morphologies. A T cells isolated from C1qbp +/+ and C1qbp ± mice were stimulated with anti-CD3/CD28 antibodies for varying amounts of time. Mitochondrial morphologies of C1qbp +/+ and C1qbp ± T cells were detected by confocal microscopy. Scale bar, 5 µm. B Mean Fluorescence Intensity (MFI) of the mitochondrial marker Cytochrome C in C1qbp +/+ and C1qbp ± T cells. C MFI of MitoTracker Green in C1qbp +/+ and C1qbp ± T cells after stimulation with anti-CD3/CD28 antibodies for 3 days. D Protein levels of OPA1, MFN2, Drp1, and C1QBP relative to β-actin, as well as the phosphorylation of Drp1 at Ser616, as analyzed by immunoblotting. Bar graphs show the mean ± S.D. of three independent experiments (n = 3); *P < 0.05, **P < 0.01, ***P < 0.001 ◂ but also its phosphorylation on Ser616 which is responsible for the scission of mitochondrial membrane. At the same time, C1QBP insufficiency also resulted in decreased expression of mitochondrial fusion proteins, OPA1 and Mfn2. Together, these data suggest that C1qbp knockdown resulted in a preference for mitochondrial fission through the increase of Drp1 expression and its phosphorylation on C1qbp ± and C1qbp +/+ T cells were activated with anti-CD3/ CD28 antibodies for the indicated time points. A Protein levels of AMPK, PGC1α, Complex I (ND1), Complex II (SDHC), Complex III (UQCRC1), and Complex IV (COXIV) relative to β-actin, as well as phosphorylation of AMPK at Thr172 were detected by immunoblotting. B mRNA levels of PGC1α, Complex I, II, III, and IV relative to β-actin were analyzed using qRT-PCR. Bar graphs show the mean ± S.D. of three independent experiments (n = 3); *P < 0.05, **P < 0.01 Ser616 as well as the decrease of OPA1 and Mfn2 protein levels.

C1QBP is required to maintain T cell mitochondrial biogenesis through the AMPK/PGC1α signaling pathway
Given that the regulation of mitochondrial dynamics is connected with the alteration of mitochondrial biogenesis [24], we further interrogated whether C1qbp knockdown would impact T cell mitochondrial proliferation. Peroxisome proliferatoractivated receptor γ coactivator 1α (PGC-1α) has been proposed to play a central role in regulation of mitochondrial biogenesis [43]. Here, our results showed that C1qbp knockdown dramatically diminished the transcription and translation of PGC1α, as shown in Fig. 5A-B. On the other hand, AMP-activated protein kinase (AMPK), as a conserved Ser/Thr kinase, is activated by the cellular AMP/ATP ratio and is accordingly involved with the bioenergetic switches from anabolic to catabolic pathways [44]. Moreover, AMPK activation enhances PGC1α expression to promote mitochondrial protein synthesis [45][46][47]. In this study, we also detected and compared the protein levels and phosphorylation status of AMPK between C1qbp ± and C1qbp +/+ T cells (Fig. 5A). Our data showed that AMPK protein levels gradually increased accompanying with the stimulation of anti-CD3/CD28 antibodies in both C1qbp ± and C1qbp +/+ T cells, with no detectable differences between these populations. However, C1qbp ± T cells were observed Fig. 6 C1qbp knockdown decreases the antitumor immune function of CAR T cells in vitro. A HuB7-H3-B16 cells (1 × 10 4 ) were seeded on a 16-well micro-E-plate for 24 h, and then co-cultured with nontransduced T cells (E:T = 1:1), C1qbp +/+ , or C1qbp ± CAR T cells targeting the huB7-H3 antigen (E:T = 1:2, 1:1, and 2:1) for an additional 72 h. Survival of huB7-H3-B16 cells was monitored using the real time RTCA assay. B HuB7-H3-B16 cells (2 × 10 5 ) were seeded on a 24-well plate for 24 h, and then co-cultured with C1qbp +/+ or C1qbp ± CAR T cells for an additional 6 h (E:T = 1:2, 1:1, and 2:1). The percentages of residual huB7-H3-B16 tumor cells (GFP + ) and CAR T cells (PE-Cy7-CD3 + ) were estimated by flow cytometry. C C1qbp +/+ and C1qbp ± CAR T cells targeting huB7-H3 were co-cultured with huB7-H3 B16 tumor cells (2 × 10 5 cells/well) in 24-well plates at different effector to target ratios (E:T = 1:2 or 1:1) without the addition of exogenous cytokines. After 18 h, the supernatant was collected and IFN-γ and TNF-α cytokines were measured in duplicate using specific ELISA kits. Bar graphs show the mean ± S.D. of three independent experiments (n = 3); *P < 0.05, **P < 0.01 to exhibit lower phosphorylation of AMPK on Thr172 when compared to C1qbp +/+ T cells. Therefore, it is likely that C1qbp knockdown at least partly contributed to a decrease of PGC1α via reduced phosphorylation of AMPK.
Whether the PGC1α signaling pathway regulates T cell mitochondrial protein synthesis still remains unclear. As such, we detected the transcription and translation of mitochondrial proteins, such as the electron transport chain (ETC) complexes I, II, III, and IV, between C1qbp ± and C1qbp +/+ T cells in response to the anti-CD3/CD28induced TCR stimulation. Our data showed that C1qbp knockdown impaired the transcription and translation of the mitochondrial ETC complexes I, III, and IV, which is consistent with the results of the previous studies [11,48]. Consequently, these data suggest that C1QBP pivotally supports translation of the mitochondrially encoded respiratory chain protein complexes. In order to further explore whether C1QBP could regulate mitochondrial protein synthesis through the PGC1α signaling pathway, we treated C1qbp +/+ T cells with PGC1α siRNA and observed the expression of mitochondrial ETC complexes I, III, and IV. Our results showed that PGC1α siRNA repressed the expression level of these mitochondrial proteins even in the context of intact C1qbp, as shown in S. Fig. 5. In other words, PGC1α played a pivotal role in C1QBPmediated mitochondrial biogenesis in response to TCR stimulation. Taken together, C1QBP insufficiency impaired T cell mitochondrial biogenesis and mitochondrial protein synthesis through the AMPK/PGC1α signaling pathway.

Knocking down C1qbp attenuates CAR T cells immunotherapeutic efficacy
In order to explore whether C1QBP would directly impact T cell antitumor functions, we constructed CAR T cells with one or two functional copies of C1qbp. B7-H3, as a member of the B7 family of immune co-stimulatory and co-inhibitory proteins, possesses two isoforms in humans: 2Ig-B7-H3 and 4Ig-B7-H3. B7-H3 protein exhibits limited expression in normal human tissues, such as those of the liver, breast, and colon, for example, but it exhibits abnormally high expression in the corresponding malignant tissues of these areas [49][50][51]. Moreover, CAR T cells targeting huB7-H3 were recently reported to exhibit potent antitumor effects on hematologic and solid tumors [52][53][54]. Here, we assessed C1qbp ± and C1qbp +/+ CAR T cells targeting huB7-H3 to delineate the effect of C1QBP on CAR T cell antitumor efficacy. First, we monitored CAR T-mediated killing of melanoma huB7-H3-B16 cells via the real-time cell analysis (RTCA) and found that C1qbp ± CAR T cells exhibited weaker cytotoxicity than C1qbp +/+ CAR T cells at different effector to target ratios (E:T), such as 1:2, 1:1 and 2:1, as shown in Fig. 6A. HuB7-H3-B16 cells co-cultured with huB7-H3-targeting CAR T cells also exhibited a similar trend. C1qbp +/+ CAR T cells repressed tumor cells to 18.7% (E:T = 1:2), 15.5% (E:T = 1:1) and 4.36% (E:T = 2:1), while C1qbp ± CAR T cells only controlled tumor cells to 38.8%, 21.5% and 10.4% under the same conditions (Fig. 6B).
To confirm the effect of C1QBP in CAR T cells immunotherapy in vivo, we inoculated huB7-H3-B16 cells into the flanks of C57BL/6 mice and then adoptively transferred C1qbp +/+ and C1qbp ± CAR T cells to assess the impact of this change on tumor progression of these recipient mice. First, mice receiving either of these CAR T cells treatment exhibited significant tumor suppression compared with mice receiving the control non-transduced T cells (NT), suggesting that both CAR T cells targeting huB7-H3 could exert some antitumor immune function. Importantly, though, mice receiving C1qbp ± CAR T cells exhibited relatively weaker tumor regression than those receiving C1qbp +/+ CAR T cells in both the subcutaneous xenograft model as well as the metastasis model of B16 murine melanoma, as shown in Figs. 7B-D and 8B-C. Moreover, our results confirmed that the more Ki67 expressed on tumor cells and the fewer tumor-infiltrating T cells were detected in TME from C1qbp ± CAR T group, as shown in Figs. 7E-F and 8D-E. At the same time, C1qbp ± CAR TILs possessed relatively lower mitochondrial mass compared with the corresponding C1qbp +/+ CAR TILs (S. Fig. 6). We also assessed the relative ratios of two distinct subsets of memory T cells: central memory T cells (CD62L hi ) and effector memory T cells (CD62L lo ). Notably, mice receiving C1qbp ± CAR T cells exhibited a significantly lower ratio of CD62L hi to CD62L lo than those receiving C1qbp +/+ CAR T cells, suggesting that C1qbp knockdown inhibited CAR T cells from forming into the central memory phenotype (Figs. 7G, 8F). Further, C1qbp ± CAR TILs exhibited relatively higher expression of the inhibitory receptor, PD-1, in the tumor but not in the spleen (S. Fig. 7A, B), suggesting that C1qbp knockdown aggravated the exhausted phenotype of the tumor-infiltrating CAR T cells but not that of the peripheral CAR T cells. Taken together, down-regulation of C1QBP attenuated CAR T cell tumor infiltration and central memory cell formation and exacerbated their exhausted phenotype in the TME, thus dampening their corresponding immunotherapeutic efficacy in vivo.

Discussion
Mitochondria as dynamic organelles are critical for the tuning of metabolic preferences in response to environmental stimuli, which are, in turn, critical in orchestrating T cell development, fate, and function. Indeed, Zhai et al. presented important evidence of a critical role for C1qbp in the differentiation of effector CD8 + T cells through a metabolic-epigenetic axis, suggesting that C1qbp deficiencymediated mitochondrial dysfunction can induce metabolic and epigenetic reprograming of T cell differentiation [55]. However, recent studies have also pointed out that dysfunctional mitochondria with decreased metabolic fitness can drive permanent T cell dysfunction in the TME [3,4,24]. Consequently, we sought to explore the relationship between C1QBP-mediated mitochondrial rewiring and T cell antitumor immune function. Here, mitochondrial plasticity was used to evaluate mitochondrial adaptation and metabolic fitness in response to a challenging microenvironment. Given that multiple tumor microenvironmental factors are involved in the development of TIL exhaustion, mitochondrial plasticity helps T cells to cope with a variety of metabolic stressors, including glucose competition, metabolite accumulation, chronic tumor antigenic stimulation through the TCR, hypoxia, and checkpoint signaling such as the PD-1/PD-L1 pathway. In this study, we demonstrated that C1qbp knockdown impaired mitochondrial plasticity by increasing mitochondrial fission, decreasing mitochondrial metabolism, and diminishing mitochondrial biogenesis, all of which subsequently impacted T cell exhaustion and central memory cell generation to ultimately dampen their persistent antitumor immune function.
Mitochondrial dynamics, including remodeling of mitochondrial architecture, mass, and activity, are important drivers for appropriate metabolic reprogramming in response to metabolic perturbations. Mitochondrial fusion allows T cells to support metabolic demands, while mitochondrial fission impairs T cells mitochondrial functions and represses their antitumor immunity [2]. Our data showed that the mitochondrial protein C1QBP is a critical regulator in the modulation of T cell mitochondrial morphology. C1qbp ± T cells possessed more punctate and fragmented mitochondria, while C1qbp +/+ T cells had more tubular and fused mitochondria in response to stimulation with anti-CD3/CD28 antibodies. Moreover, we found that C1qbp knockdown increased levels of the mitochondrial fission protein Drp1 as well as its phosphorylation on Ser616, and also repressed the mitochondrial fusion proteins OPA1 and Mfn2. These data suggest that reduced C1QBO expression rendered T cells prone to mitochondrial fission, which attenuates the ability of T cells to adapt to the relentlessly challenging TME.
In fact, emerging evidence has delineated the importance of this intricate balance between mitochondrial fission and fusion on T cell metabolism and immune function [56][57][58]. Pearce et al. [2] found that fused mitochondria had tight cristae, which yielded more efficient ETC activity and favored OXPHOS as well as FAO, while fragmented mitochondria had the loose cristae, which exhibited less efficient ETC activity and thus favored aerobic glycolysis. Inhibition of mitochondrial OXPHOS has also been shown to result in the suppression of CD8 + T cell memory formation as well as upregulation their exhaustion [4]. In our study, C1qbp ± T cells exhibited more fragmented mitochondria and lower OXPHOS as well as less ATP production, which decreased TIL metabolic fitness. Notably, enhanced mitochondrial SRC was also thought to be an important property for cells to achieve long-term survival and efficacy [34,59]. In this study, C1qbp ± T cells exhibited weakened SRC relative to C1qbp +/+ T cells, which suggested down-regulation of C1QBP rendered T cells with less extra mitochondrial capacity to meet bioenergetic demands. On the other hand, C1QBP insufficiency repressed the relative ratio of CD44 hi CD62L hi central memory to CD44 hi CD62L lo effector memory T cells. Similarly, tumor-bearing mice receiving tumor-specific C1qbp ± CAR T cells exhibited lower proportions of the central memory phenotype when compared to their C1qbp +/+ CAR T cells, although this effect, importantly, was only true for TILs but not for peripheral T cells. Consequently, how and whether C1QBP enhancement could promote mitochondrial metabolism and SRC to drive central memory T cell formation remains to be elucidated.
Indeed, the previous study reported that C1QBP deficient mice exhibited mid-gestation lethality due to the severe dysfunction of the mitochondrial respiratory chain  [23]. They further found that C1QBP bound RNA and interacted with mitochondrial messenger RNA species in vivo. Moreover, co-immunoprecipitation revealed the close association of C1QBP with the mitoribosome. This RNA-binding ability of C1QBP was well correlated with mitochondrial translation. However, in our study, we found that C1qbp knockdown repressed the synthesis of the mitochondrial ETC proteins, complexes I, III, and IV. Reduced C1QBP expression also resulted in the decrease of T cell mitochondrial mass in vitro and in vivo. In this regard, C1QBP not only takes a crucial role in mtDNA translation through the association with mitoribosome, but also regulates the AMPK/PGC1a signaling axis to modulate mitochondrial proliferation.
Given that tumor specific T cells frequently exhibit decreased mitochondrial biogenesis, metabolic reprogramming via enhancement of PGC1α expression was reported to be correlated with reinvigoration of mitochondrial function and improvement of T cell antitumor responses [24,60]. In this study, C1qbp knockdown exacerbated the expression of inhibitory receptors, PD-1, Tim-3, and LAG-3, in TILs. Given that TILs with a progressive loss of PGC1α have been shown to lack mitochondrial biogenesis, and that enhancement of PGC1α has been shown to contribute to metabolic plasticity and effector function [24], we presumed that C1QBP-mediated PGC1α upregulation could trigger metabolic reprogramming in TILs, which could reinvigorate their mitochondrial plasticity to resist exhaustion and hypofunction in the TME. Intriguingly, Bengsch et al. reported that PD-1 repressed PGC1α, while improving bioenergetics by overexpression of PGC1α prevented T cell exhaustion, in turn inducing the anti-PD-L1 therapeutic reinvigoration by reprogramming metabolism in the subset of PD-1 Int exhausted T cells [60]. T cells with features of more severe exhaustion exhibit a specialized DNA methylation pattern and chromatin architecture that may lock T cells in a permanently dysfunctional state, and are refractory to anti-PD-1 treatment. However, the transcription factor T cell factor 1 (TCF1) endows T cells with a stem-like phenotype by triggering the distinct transcriptomic and epigenetic regulation, which further contributes to a proliferative burst upon PD-1 blockade treatment [61][62][63]. Therefore, the development of novel, effective strategies to reprogram mitochondrial functions will improve the current response to such checkpoint blockade therapies. In this regard, how and whether C1QBP-mediated PGC1α enhancement could trigger TCF1 expression, and thus induce additive antitumor immunity when used in conjunction with anti-PD-1 checkpoint blockade, warrants further evaluation.
Taken together, as shown in Fig. 8G, C1qbp knockdown resulted in T cells with more mitochondrial fission, lower mitochondrial OXPHOS, and weaker mitochondrial biogenesis, together attenuating mitochondrial plasticity and metabolic fitness and thus exerting a negative impact on T cell tumor infiltration, exhaustion, and memory phenotype formation. In this regard, improvement of mitochondrial plasticity through C1QBP might endow T cells or CAR T cells with robust and persistent antitumor functions, which may present a novel strategy to promote their antitumor immunotherapeutic efficacy.