GITR Ligation Improves Anti-PD1-Mediated Restoration of Human MMR-Proficient Colorectal Carcinoma Tumor-Derived T Cells

Background & Aims In contrast to mismatch repair deficient colorectal carcinoma (CRC), MMR proficient (pMMR) CRC does not respond to immune checkpoint blockade. We studied immune checkpoint stimulation via glucocorticoid-induced tumor necrosis factor receptor-related protein (GITR) on ex vivo functionality of human tumor-infiltrating lymphocytes (TIL) isolated from pMMR primary CRC and liver metastases (CRLM). Methods Using lymphocytes from resected tumor, adjacent tissues, and peripheral blood mononuclear cells (PBMC) of 132 pMMR primary CRC or CRLM patients, we determined GITR expression and the in vitro T-cell agonistic activity of recombinant GITR ligation. Results Here, we show that GITR was overexpressed on TIL when compared with other stimulatory immune checkpoints (4-1BB, OX40). Its expression was enhanced in TIL compared with PBMC and adjacent tissues. Among CD4+ TIL, GITR expression was primarily expressed by CD45RA- FoxP3hi activated regulatory T cells. Within CD8+ TIL, GITR was predominantly expressed on functionally exhausted and putative tumor-reactive CD103+ CD39+ TIL. Strikingly, recombinant GITRL reinvigorated ex vivo TIL responses by significantly enhancing CD4+ and CD8+ TIL numbers. Dual treatment with GITRL and nivolumab (anti-PD1) enhanced CD8+ TIL expansion compared with GITRL monotherapy. Moreover, GITRL/anti-PD1 dual therapy further improved anti-PD1-mediated reinvigoration of interferon gamma secretion by exhausted CD8 TIL from primary CRC. Conclusions GITR is overexpressed on CD4+ and CD8+ TIL from pMMR CRC and CRLM. Agonistic targeting of GITR enhances ex vivo human TIL functionality and may therefore be a promising approach for novel monotherapy or combined immunotherapies in primary pMRR CRC and CRLM.

C olorectal cancer (CRC) is the second and third most common cancer worldwide in women and men, respectively. 1 In 2020, 1.9 million newly diagnosed patients and 935,000 disease-related deaths were reported, letting CRC account for 10% of total cancer incidence and mortality annually. 1 Even though recent diagnostic and treatment regimens have enhanced overall survival rates significantly, there remains a need for further improvement. 2 Extensive analyses on genomic, epigenomic, and transcriptomic CRC features have extended our understanding on CRC biology and its stromal-immune microenvironment, pleading for a more tailor-made, subtype-based therapeutic approach. 3,4 Enhanced inflammatory stromal immune infiltration has been linked to a more favorable clinical outcome in earlystage CRC. 5 High intratumoral and peritumoral CD8 þ cytotoxic T lymphocyte (CTL) and CD4 þ T helper 1 (Th1) content is limited to the hypermutated microsatellite instable (MSI) subtype of CRC. 3 Recent clinical trials on programmed cell death 1 (PD1)-based immune checkpoint inhibition have confirmed survival benefit in these patients, exploiting the inflammatory immune phenotype, granting Food and Drug Administration approval for the treatment of metastatic mismatch repair-deficient and MSI-high (dMMR-MSI-H) CRC. [6][7][8][9] In contrast, mismatch repair-proficient and MSIlow (pMMR-MSI-L) CRC, comprising 85% of the total CRC population, has hardly shown any response to immune checkpoint blockade (ICB). 6,7,10 Its impaired clinical activity is hypothesized to be caused by poor immune cell infiltration, decreased inhibitory ligand expression, local immune suppression, and enhanced exhaustion of tumor-infiltrating CTL. 3,11,12 Therefore, there is a need for effective immunotherapy that boosts the impaired tumor immune microenvironment in pMMR CRC.
Promoting anti-tumor immune responses via agonistic targeting of co-stimulatory receptors on tumor-infiltrating lymphocytes (TIL) is a promising alternative to current ICB therapies. In addition to T-cell receptor (TCR) signaling, costimulation is required to initiate effective T-cell activation, subset differentiation, effector function, and survival. 13 Among co-stimulatory receptors, members of the tumor necrosis factor receptor superfamily (TNFRSF) (CD30, DR3, GITR, HVEM, OX-40, TNFR3, 4-1BB) have been widely studied for their application in anti-cancer immunotherapy. 14,15 GITR is considered one of the most notable TNFRSF checkpoints because its activation has been shown to activate effector T cells, hamper regulatory T cell (Treg) functionality, and induce efficient anti-tumor responses in preclinical models using agonistic antibodies. [16][17][18][19][20][21][22][23] Recent phase 1 clinical trials have reported manageable safety profiles for GITR targeting therapy among multiple advanced solid tumors; therefore, this approach may also be relevant for patients with pMMR CRC. [24][25][26][27][28] Yet, expression and functionality of co-stimulatory checkpoint receptors including GITR on TIL isolated from CRC have not been elucidated.
In this study, we aimed to analyze the expression of TNFRSF members GITR, OX40, and 4-1BB on TIL in pMMR CRC and pMMR CRC-derived liver metastasis (CRLM) patients. Because GITR was expressed most prominently on CD4 þ and CD8 þ TIL, in functional assays we focused on costimulatory receptor targeting via agonistic GITR ligation in pMMR CRC and CRLM patients. We demonstrate that GITR is expressed by activated CD4 þ and non-terminally exhausted CD8 þ TIL. Moreover, GITR co-stimulation is found to be an effective immunomodulatory approach enhancing proliferation and function of TIL from pMMR CRC and CRLM patients. Last, GITR co-stimulation enhanced anti-PD1-mediated immune stimulation in pMMR CRC-derived TIL.

Patients
In Tables 1 and 2, an overview of patient characteristics is displayed. One hundred thirty-two tumors of pMMR CRC and CRLM patients were included between July 2016 and July 2022.

GITR Is Predominantly Expressed on Intratumoral CD4 þ Activated Th and Treg Cells From pMMR CRC and CRLM
Because we have shown previously that GITR is enriched in CD4 þ TIL from the hepatic microenvironment, we first *Authors share co-second authorship; § Authors share co-senior authorship.
To evaluate the potential of co-stimulatory receptor targeting in anti-tumor immunotherapy, TNFSRF member (GITR, 4-1BB, and/or OX40) expression was measured ex vivo on CD4 þ T cells from pMMR CRC and CRLM patients. Tumor-derived aTreg, aTh, and Th demonstrated higher costimulatory receptor expression compared with PBMC and adjacent tissues in CRC ( Figure 1D). In CRLM, co-stimulatory receptor expression was enhanced on tumor-derived aTreg and aTh compared with PBMC and adjacent tissues ( Figure 1E).
We compared expression of GITR, 4-1BB, and OX40 on tumor-derived aTreg, aTh, and Th. In CRC, aTreg, aTh, and Th demonstrated enhanced GITR expression compared with 4-1BB and OX40 ( Figure 2A). Moreover, GITR was expressed more prominently in CRLM-derived aTh and aTreg compared with OX40 ( Figure 2B). We conclude that high rates of co-expression among TNFSRF members were observed on all CD4 þ TIL subsets, with GITR being expressed most prominently, especially in CRC.
Taken together, we demonstrate that compared with adjacent tissues, expression of the co-stimulatory molecule GITR is increased most prominently of all TNFRSF members investigated on aTh and aTreg in TIL from pMMR CRC and CRLM patients and that its expression is associated with proliferation and activation markers.
GITR Delineates Activated CD103 þ CD39 þ CD8 þ TIL From pMMR CRC and CRLM Subsequently, we studied co-stimulatory molecule expression on cytotoxic CD8 þ T cells from pMMR CRC and  CRLM patients. Compared with colorectal tissues, CRLM demonstrated higher frequencies of CD8 þ T cells among total CD45 þ T cells ( Figure 4A). CD8 þ T cells were increased in total CD45 þ TIL fractions compared with adjacent tissues in CRC but not in CRLM.
GITR Expressing PD1 hi CD8 þ TIL Feature an Exhausted Phenotype With Enhanced Proliferative Capacity in pMMR CRC and CRLM Because GITR expression on CD8 þ TIL appears to be linked to a more activated, potentially tumor-reactive phenotype, we aimed to analyze the functional state of GITR þ CD8 þ TIL.
These data demonstrate that GITR is predominantly expressed by functionally impaired CD8 þ TIL characterized by enhanced inhibitory receptor and TOX expression ex vivo. Nevertheless, GITR þ PD1 hi CD8 þ TIL differ from GITR -PD1 hi CD8 þ TIL because they demonstrate an enhanced proliferative capacity based on enhanced Ki67 and TCF1 expression. Agonistic targeting of GITR might potentially functionally reinvigorate this cell population.

GITR Ligation Enhances pMMR CRC-and CRLM-Derived CD4 þ /CD8 þ TIL Expansion and Pro-inflammatory Cytokine Secretion
The immune modulatory effect of GITR ligation on TIL was tested in vitro. Primary TIL were cultured in the presence of anti-CD3/-CD28 activation beads providing initial TCR and CD28 signaling, which is required for efficient GITR-mediated co-stimulation. After 8-10 days of in vitro culture, CD4 þ and CD8 þ TIL expansion was determined by ratiometric evaluation of cell numbers by flow cytometry using counting beads ( Figure 7A).
Variable baseline proliferation rates of TIL were observed among individual patients, but overall co-culture in the presence of hexameric GITRL doubled the numbers of CD4 þ and CD8 þ pMMR CRC-and CRLM-derived TIL compared with TIL stimulated in the presence of aCD3/ CD28 alone ( Figure 7B-D). We could demonstrate a dosedependent increase for both CD4 þ and CD8 þ TIL proliferation ( Figure 7C). Moreover, proliferation marker Ki67 was significantly enhanced upon GITRL treatment for both TIL subsets ( Figure 7E). Correlating with the ex vivo GITR expression data, CD4 þ TIL demonstrated higher Ki67 frequencies compared with CD8 þ TIL upon GITR ligation. Nevertheless, expansion rates did not significantly differ between both TIL subsets (data not shown).
To study how GITR ligation affects different CD4 þ TIL subsets, we compared the expansion of FoxP3and FoxP3 þ CD4 þ TIL in vitro. Generally, after GITR ligation FoxP3 -CD4 þ TIL demonstrated enhanced expansion compared with FoxP3 þ TIL ( Figure 8A). Furthermore, FoxP3 MFI declined upon GITR ligation, suggesting a reduced immune suppressive capacity, as was described previously in liver cancers ( Figure 8B). 21 To confirm the ability of CD4 þ Th TIL to respond to GITR ligation independently of Treg, we cultured CD25-depleted TIL for 8-10 days. CD4 þ TIL demonstrated similar expansion rates compared with controls in the presence or absence of CD25 þ CD4 þ Treg TIL, suggesting that the presence of aTreg within the TIL does not negatively impact the effect of GITR ligation on T-cell proliferation ( Figure 8C).  . Mann-Whitney test (A) was used to analyze differences between 2 unpaired groups. Kruskal-Wallis test (B and C) was applied to analyze differences between more than 2 different groups. *P .05, **P .01, ***P .001, ****P .0001. Boxes and whiskers represent mean and 95% confidence interval. A, adjacent tissues; CRC, primary colorectal cancer; CRLM, liver metastasis; DP, double positive; P or PBMC, peripheral blood mononuclear cell; SP, single positive; T, tumor.
To evaluate pro-inflammatory effector cytokine secretion by CD3 þ TIL, after 8-10 days of in vitro culture with or without GITR ligation, total TIL were restimulated using PMA-ionomycin. Upon restimulation, CD3 þ TIL that were treated with GITRL demonstrated increased frequencies of both IFN-g þ and TNF-a þ cells compared with non-treated TIL ( Figure 9A). These data were confirmed by the detection of elevated IFN-g levels in the culture supernatants of pMMR CRC and CRLM TIL that were treated with GITR ligation ( Figure 9B). Furthermore, pro-inflammatory proteases granzyme A (GzmA), granzyme B (GzmB), and perforin were increased in response to GITRL in CRC-derived TIL cultures as well ( Figure 9C). Remarkably, GITR ligation not only skewed TIL to enhance pro-inflammatory cytokine secretion but also improved pro-inflammatory chemokine secretion. In CRCderived TIL cultures, we detected enhanced CCL3, CCL4, CCL17, and CXCL9 levels ( Figure 10A). CCL3, CCL4, CCL17, CXCL1, and CXCL5 levels showed a trend to increase among CRLM-derived cultures ( Figure 10B).
Taken together, these data demonstrate that GITR ligation reinvigorates CRC-and CRLM-derived CD4 þ and CD8 þ TIL functionality, leading to enhanced expansion and proinflammatory cytokine or chemokine secretion.

GITR Ligation Potentiates Immune Stimulation of Anti-PD1 in pMMR CRC-derived TIL
So far, PD1 blockade monotherapy has not shown a clinical benefit in pMMR CRC and CRLM patients. Our data demonstrate that tumor-derived GITR þ CD8 þ cells were generally concentrated among the PD1 hi CD8 þ TIL. Therefore, we studied whether co-stimulatory checkpoint targeting via GITR ligation could enhance anti-PD1-mediated Tcell reinvigoration in vitro.
In the majority of patient-derived TIL, anti-PD1 monotherapy did not enhance TIL expansion or effector cytokine production ( Figure 11). However, GITR co-stimulation in combination with PD1 blockade did enhance pMMR CRCderived CD4 þ and CD8 þ expansion ( Figure 11A and B). Conformably, proliferation marker Ki67 was significantly enhanced upon combination treatment for CD8 þ TIL ( Figure 11B). Ultimately, IFN-g production in CD4 þ and CD8 þ TIL was enhanced upon PMA-ionomycin restimulation after in vitro culture with the combination regimen as well as TNF-a production in CD4 þ TIL, indicating that GITR costimulation improves anti-PD1-mediated immune stimulation in pMMR CRC-derived TIL ( Figure 11C and D).

Discussion
Whereas clinical trials on PD1-based immune checkpoint inhibition have shown great anti-tumor efficacy in dMMR (MSI-high) CRC, pMMR (MSI-low or microsatellite stable [MSS]) tumors that occur in the majority of CRC and CRLM patients do not respond well to ICB. 6,7,10 Presently, the implementation of co-stimulatory checkpoint targeting in anti-tumor immunotherapy is widely studied in preclinical and early clinical trials as a suitable alternative or supportive approach for ICB. Yet, the biological and clinical implications of these receptors in anti-tumor immunotherapy remain elusive. In this study, we aimed to comprehensively analyze the expression of TNFRSF members GITR, OX40, and 4-1BB on CD4 þ and CD8 þ TIL and to associate their expression to T-cell activation or exhaustion status in pMMR CRC and CRLM patients. Furthermore, we studied the impact of GITR co-stimulation on pMMR-CRCand -CRLM-derived-TIL functions.
In a cohort of 132 patients with pMMR CRC or CRLM, we first demonstrated that GITR was expressed more prominently on tumor-derived TIL compared with OX40 and 4-1BB. Furthermore, GITR was predominantly co-expressed with OX40 and 4-1BB and compromised the vast majority of TIL expressing a TNFRSF member. GITR expression was enhanced on activated, proliferating CD4 þ and CD8 þ TIL. CD4 þ aTh and aTreg displayed highest GITR expression compared with their counterparts in adjacent tissues and blood. Notably, GITR expression in CD8 þ TIL delineated putative tumor reactive TIL.
Enhanced GITR expression was detected on CD4 þ CD45RA -FoxP3 hi aTreg and non-suppressive CD4 þ CD45RA -FoxP3 lo aTh. Improved activation and anti-tumor immunity via GITR engagement in CD4 þ TIL are described to be mediated either by CD4 þ Treg depletion via Fcg-receptor co-engagement on innate immune cells or CD4 þ Treg  We have confirmed previously that co-stimulation via GITR could alleviate Treg-mediated immune suppression, restoring proliferative capacity and cytokine production of non-Treg (CD4 þ CD25 -) TIL from primary and secondary liver tumors. 37 Nevertheless, multiple studies have shown a direct stimulatory effect of GITR ligation on effector TIL as well. 20,22,38,39 Moreover, hexameric GITR ligation (GITRL) in human naive PBMC was shown to induce CD4 þ T-cell proliferation in the absence of Treg. 40 In CRC, TME-residing aTh have been shown to correlate with improved survival rates because they produce pro-inflammatory cytokines and harbor anti-tumor immunity. 31,41 Although on the basis of our in vitro functional polyclonal assay we cannot distinguish between any direct or indirect effects on CD4 þ (FoxP3 þ ) or CD8 þ TIL, these findings by us and others suggest that immunotherapeutic targeting of GITR may stimulate the local immune response within the CRC-/CRLM-related TME through affecting both Treg and aTh cell function.
In CRC, Duhen et al 32 and Simoni et al 42 have demonstrated that CD103 þ CD39 þ CD8 þ TIL are enriched by tumor-reactive T cells. Here, we confirmed enrichment of CD103 and CD39 co-expression on CD8 þ T cells in TIL compared with adjacent tissues and blood (data not shown). Interestingly, GITR expression was highest on CD103 þ CD39 þ CD8 þ TIL, suggesting that tumor-reactive CD8 þ T cells within the TME display GITR expression. Furthermore, enhanced expression of Ki67, HLA-DR, and perforin illustrates the pro-inflammatory state of these putative tumor reactive GITR þ CD8 þ TIL. Yet, ex vivo CD8 þ TIL stimulation revealed impaired effector cytokine secretion in GITR þ cells compared with GITR -TIL, suggesting the former to be functionally impaired. Functional exhaustion of these cells might be caused by potential chronic T-cell receptor stimulation of CD103 þ CD39 þ CD8 þ TIL in the TME. 43 In support of this hypothesis, we observed enrichment of GITR among PD1 hi and LAG3 þ CD8 þ TIL. Interestingly, although there was enhanced transcription factor TOX expression in PD1 hi GITRand GITR þ CD8 þ TIL, the latter showed enhanced TCF1 and Ki67 expression compared with PD1 hi GITR -CD8 þ TIL. Therefore, GITR seems to delineate PD1 hi CD8 þ TIL into cells that demonstrate some similarity to exhausted T (Tex) precursor cells. In recent preclinical cancer models, these Tex precursor cells have been shown to be a crucial source for providing a robust response to ICB, suggesting that these cells may be eligible for GITRmediated reactivation as well. 44,45 Whereas TNFRSF co-stimulation was thought to affect CD8 þ TIL indirectly via its selective effect on CD4 þ TIL, direct effects of GITRL on CD8 þ TIL have been described. 20,38 We performed polyclonal functional assays testing the ability of GITRL to stimulate tumor-derived CD4 þ and CD8 þ TIL function ex vivo. We demonstrated enhanced Ki67 expression and expansion of CD4 þ and CD8 þ TIL from pMMR CRC and CRLM. In addition, tumor-derived T cells that were treated with GITRL demonstrated increased frequencies of both IFN-g þ and TNF-a þ cells and enhanced secretion of pro-inflammatory proteases and chemokines (eg, CCL3, CCL4, CXCL9), potentially driving immune cells attraction and activation in the TME in vivo. 46 In mice, dual therapy of GITR co-stimulation and PD1 blockade was shown to enhance effector T-cell function synergistically by restoring the balance of key homeostatic regulators CD226 and T-cell immunoreceptor with Ig and ITIM domains. 20 Moreover, it has been shown recently that the anti-GITR/anti-PD1 combination could restore the effector T-cell:Treg ratio, resulting in enhanced anti-tumor immunity. 25,39 In concordance with clinical studies, our data indicate that anti-PD1 monotherapy hardly stimulates the activity of tumor-derived TIL from pMMR CRC in vitro. However, the combination regimen of anti-PD1 with GITRL led to significant functional T-cell reinvigoration. Because high GITR expression is mostly restricted to the tumor site, which potentially limits systemic adverse events of GITR targeting, in our opinion GITR represents an attractive target for therapeutic immunomodulation alone or in combination with ICB in these pMMR patients. Although treatment with anti-GITR antibodies had been shown to cause toxicity upon repeated administration in preclinical animal models, recently it has been shown that a newly developed fully human agonistic GITR immunoglobulin G1 monoclonal antibody in combination with nivolumab demonstrated an Wilcoxon matched test (A and F) was used to analyze differences between 2 paired groups. Friedman test (C and D) was applied to analyze differences between more than 2 different groups. Correlation analysis was performed according to Spearman. **P .01, ***P .001, ****P .0001. Boxes and whiskers represent mean and 95% confidence interval. CRC, primary colorectal cancer; CRLM, liver metastasis; IFN-g, interferon gamma; MFI, median fluorescence intensity; TIL, tumorinfiltrating lymphocyte; TNF-a, tumor necrosis factor alpha. . GITR ligation enhances CD4 D and CD8 D TIL expansion. TIL were isolated from tumor tissues using enzymatic digestion and subsequent Ficoll density gradient centrifugation. TIL were stimulated using CD3/CD28 stimulation beads in the absence (ctrl) or presence of a hexameric GITR ligand (GITLR 1 mg/mL þ anti-HA 2.5 mg/mL). After 8 days, TIL expansion was measured and defined as fold increase over control conditions (proliferation index). (A) Flow cytometric gating strategy used for ex vivo phenotyping after in vitro polyclonal assay. Cell numbers were normalized using counting beads (red box). (B) pMMR CRC-derived CD4 þ and CD8 þ TIL expansion upon GITR ligation (n ¼ 19). (C) CD4 þ and CD8 þ TIL expansion upon various dosages of GITR ligation (0.001 mg/mL, 0.01 mg/mL, 0.1 mg/mL, 1.0 mg/mL, 10.0 mg/mL, and 100.0 mg/mL, respectively) (n ¼ 8). (D) pMMR CRLM-derived CD4 þ and CD8 þ TIL expansion upon GITR ligation (n ¼ 6). (E) Gating strategy and frequencies of Ki67 þ cells among CD4 þ and CD8 þ TIL upon GITR ligation, respectively. CRC are depicted in blue, and CRLM are depicted in red. Wilcoxon matched test (B, D, and E) was used to analyze differences between 2 paired culture conditions. Friedman test (C) was applied to analyze differences between more than 2 different culture conditions. *P .05, ***P .001, ****P .0001. Boxes and whiskers represent mean and 95% confidence interval. CRC, primary colorectal cancer; CRLM, liver metastasis; Ctrl, control; TIL, tumor-infiltrating lymphocyte.
acceptable safety profile in patients with advanced solid tumors. 26,47 In humans, GITR monoclonal antibodies have been shown to reduce circulating and intratumoral Treg. 25 Yet, efficient direct T-cell activation via TNFR requires receptor clustering to induce optimal downstream signal transduction. 19 In our study, GITR oligomerization is facilitated using a hexameric GITRL. Other strategies to induce receptor multimerization are Fc-engineering or the construction of bi-specific antibodies. 48,49 Chan et al 49 have recently shown the great potential of combining immune checkpoint stimulation (ICS) and ICB using an anti-PD-1-GITR-L bispecific. In particular, the approach of bispecific antibodies deserves great attention for development of future clinical trials on ICS because these constructs not only drive receptor multimerization but also direct GITR-mediated activation to tumor reactive PD-1 expressing cells only.
The strength of our study is the use of human patientderived TIL, but it entails some limitations as well. 1 Because all patients underwent surgical resection, our study cohort does not comprise the full spectrum of the CRC and CRLM patient population. Currently, the efficacy of immunotherapy has been mostly studied in advanced therapy resistant tumors. Nevertheless, the (neo)adjuvant application of immunotherapy in early or intermediate stage disease is being studied more and more frequently. 2 Some patients from our CRC and CRLM cohort have undergone systemic therapy before resection. Therapeutic regimen might have varied among individual patients, potentially changing the composition of the TME. To exclude any direct effects of neoadjuvant therapy on the TME, minimum time before surgery was set at 4 weeks. 3 TIL numbers isolated from individual patient tissues were highly variable. Therefore, phenotype analysis and in vitro functional assays could not be performed in all patients.
In summary, we conclude that GITR is enriched in CD4 þ aTh and aTreg TIL as well as in PD1 hi CD8 þ TIL. Furthermore, GITRL enhances pMMR CRC-and CRLM-derived human TIL functionality ex vivo. Our study provides compelling preclinical data that support agonistic targeting of GITR as part of a new immunotherapeutic approach for pMMR CRC and CRLM patients.

Patient Selection
Patients (age >18 years) undergoing surgical resection between July 2016 and November 2021 for either primary CRC (stage 1-3) or CRLM were included from 3 different hospitals. Patients having received (any) neoadjuvant treatment or immunosuppressive therapy <4 weeks and <3 months before surgical resection were excluded from our study. Peripheral blood was drawn on the day of the surgical resection before surgery. In accordance with the pathologists, fresh samples of tumor and adjacent tumor-free tissues (>10 cm and >2 cm distant of the tumor for primary CRC and liver metastasis, respectively) were obtained and processed within 24 hours. Patients' mismatch repair status was determined using immunohistochemistry and defined by the loss of expression of any of the following mismatch Wilcoxon matched test was used to analyze differences between 2 paired culture conditions. *P .05, **P .01. Boxes and whiskers represent mean and 95% confidence interval. CRC, primary colorectal cancer; CRLM, liver metastasis; MFI, median fluorescence intensity; TIL, tumor-infiltrating lymphocyte.
repair gene-related proteins: MLH1, PMS2, MSH2, and MSH6. Patient data were retrieved from electronical medical records. All study procedures were approved by the local ethics committee (NL58534.078.16; NL47888.041.14). Patients had given informed consent for tissue and blood donation as well as usage of personal data.

Cell Preparation
PBMC were isolated by Ficoll density gradient centrifugation. Mononuclear single cell suspensions from tumor and adjacent tissues were obtained by enzymatic tissue digestion. Adjacent colon tissues were cut and stirred in the presence of EDTA (10% fetal calf serum, 15 mmol/L HEPES, 1 mmol/L EDTA in phosphate-buffered saline) 4 times for 15 minutes at 37 C. Subsequently, adjacent colon as well as cut primary CRC tissues were digested in the presence of 400 U/mL collagenase type VIII (Sigma-Aldrich, St Louis, MO) and 0.2 mg/mL DNAse I (Sigma-Aldrich) in Hanks balanced salt solution for 30-60 minutes at 37 C with interrupted gently swirling. CRLM and adjacent liver tissues were cut into small pieces and then digested with 0.2 mg/ mL DNAse I (Sigma-Aldrich) and 0.125 mg/mL collagenase type IV (Sigma-Aldrich) for 40 minutes at 37 C with . GITR ligation increases CD3 D TIL pro-inflammatory cytokine secretion. TIL were isolated from tumor tissues using enzymatic digestion and subsequent Ficoll density gradient centrifugation. TIL were stimulated using CD3/CD28 stimulation beads in absence (ctrl) or presence of hexameric GITR ligand (GITRL 1 mg/mL þ anti-HA 2.5 mg/mL). After 8 days, supernatants were analyzed for soluble factors, and TIL were phenotyped after PMA/ionomycin restimulation. (A) Flow cytometric gating strategy and relative frequencies of IFN-g and TNF-a producing pMMR CRC-derived CD3 þ TIL upon PMA/ ionomycin restimulation after GITR ligation compared with control conditions (n ¼ 11). (B) Concentration of secreted IFN-g defined as fold increase over control conditions. CRC are depicted in blue (n ¼ 18), and CRLM are depicted in red (n ¼ 4). (C) Concentration of secreted cytokines defined as fold increase over control conditions. CRC are depicted in blue (n ¼ 6). Wilcoxon matched test was used to analyze differences between 2 paired culture conditions. *P .05. Boxes and whiskers represent mean and 95% confidence interval. CRC, primary colorectal cancer; CRLM, liver metastasis; Ctrl, control; IFN-g, interferon gamma; TIL, tumor-infiltrating lymphocyte; TNF-a, tumor necrosis factor alpha.
interrupted gently swirling. Cell suspensions were filtered through 70-mm pore cell strainers (BD Biosciences, San Diego, CA), and mononuclear leukocytes were obtained by Ficoll density gradient centrifugation. Mononuclear cell viability, CD45/CD3 purities, and cell numbers were determined using flow cytometry (MACS Quant; Miltenyi Biotec, Bergisch Gladbach, Germany). A minimum of 100,000 total CD3 þ TIL/1 g tissue was set for study inclusion.

Flow Cytometric Analysis
Freshly isolated PBMC, tumor-infiltrating lymphocytes, and tissue-infiltrating lymphocytes were analyzed for the expression of specific surface and intracellular markers using monoclonal antibodies (Table 3). Nonviable cells were excluded by labeling with Fixable Viability Dye Efluor 506 (eBioscience, Vienna, Austria). Cell surface staining with fluorochrome-conjugated antibodies was performed in the dark at 4 C for 30 minutes, after which cells were fixed, permeabilized using the FoxP3 staining buffer kit (eBioscience), and stained for intracellular antigens. Cells were measured using a FACSCanto II, FACSAria SORP II, or FAC-Symphony flow cytometer (BD Biosciences) and analyzed using FlowJo software version V10 (BD Biosciences). Appropriate isotype control antibodies were used for gating purposes (Table 3).

Statistical Analysis
Statistical analyses were performed using GraphPad Prism software version 9.0 (GraphPad, La Jolla, CA). A Wilcoxon matched test was used to analyze differences between 2 paired groups of data. Either a Friedman or a Kruskal-Wallis test was applied to analyze differences between more than 2 different groups. Correlation analysis was performed according to Spearman. A P value lower than .05 was considered statistically significant (*P .05, **P .01, ***P .001, ****P .0001). All authors had access to the study data and had reviewed and approved the final manuscript.