Current clinical trials of combined EGFR-tyrosine kinase inhibitors (TKI) and immune checkpoint blockade (ICB) therapies show no additional effect. This raises questions regarding whether EGFR-TKIs attenuate ICB-enhanced CD8+ T lymphocyte function. Here we show that the EGFR-TKI afatinib suppresses CD8+ T lymphocyte proliferation, and we identify CAD, a key enzyme of de novo pyrimidine biosynthesis, to be a novel afatinib target. Afatinib reduced tumor-infiltrating lymphocyte numbers in Lewis lung carcinoma (LLC)–bearing mice. Early afatinib treatment inhibited CD8+ T lymphocyte proliferation in patients with non–small cell lung cancer, but their proliferation unexpectedly rebounded following long-term treatment. This suggests a transient immunomodulatory effect of afatinib on CD8+ T lymphocytes. Sequential treatment of afatinib with anti-PD1 immunotherapy substantially enhanced therapeutic efficacy in MC38 and LLC-bearing mice, while simultaneous combination therapy showed only marginal improvement over each single treatment. These results suggest that afatinib can suppress CD8+ T lymphocyte proliferation by targeting CAD, proposing a timing window for combined therapy that may prevent the dampening of ICB efficacy by EGFR-TKIs.
This study elucidates a mechanism of afatinib-mediated immunosuppression and provides new insights into treatment timing for combined targeted therapy and immunotherapy.
EGFR-tyrosine kinase inhibitors (EGFR-TKI) markedly improved the clinical outcome of advanced non–small cell lung cancer (NSCLC) harboring activating EGFR mutations (1, 2). Afatinib, a second-generation EGFR-TKI with irreversible inhibitory ability for pan-ErbB, was developed for EGFR-mutant NSCLC (3), and is extensively used in squamous cell carcinoma of the lung, head, and neck because of its positive effect on wild-type EGFR (4). Despite the remarkable initial response, the long-term effectiveness of EGFR-TKIs is unsatisfactory (5). To improve the clinical usage and efficacy of EGFR-TKIs in cancer treatment, a better understanding of the mechanism of these drugs and strategizing new therapeutic approaches, including combinations with other treatments, are needed.
Immune checkpoint blockade (ICB) therapies have been developed to reverse cancer immunosuppression, leading to enhanced antitumor immunity. Combined therapy using EGFR-TKIs and ICBs is theoretically feasible (6, 7). However, current clinical trials show no significant improvement in treatment with EGFR-TKIs combined with ICBs (8–12). The reason for the unsuccessful results is still unclear. In addition to the possibility of low expression levels of PD-L1 or tumor mutation burdens (13), EGFR-TKIs might also directly affect immune cell functions via unknown mechanisms, because a drug including afatinib has been estimated to have 6.3 targets (14). On the other hand, a sequential combination of chemotherapy and targeted agents has shown improvements in tumor cell apoptosis and in enhancing drug sensitivity (15, 16). Understanding the mechanism of how a targeted drug acts on cancer cells or even on surrounding cells, including immune cells, will provide important information for a precise use of the drug. This may also offer an extensive opportunity to combine the drug with immunotherapy, with correct timing, for clinical benefit.
Nucleotide biosynthesis is necessary for the proliferation, maturation, and survival of T lymphocytes (17), with a higher demand for de novo pyrimidine biosynthesis than purines (18). For de novo pyrimidine biosynthesis, CAD [carbamoyl-phosphate synthetase 2 (CPS2), aspartate transcarbamoylase (ATCase) and dihydroorotase (DHOase)] is a rate-limiting and multi-functional enzyme to be used in executing the first three reactions of pyrimidine biosynthesis in eukaryotes (19). Targeting of the de novo pyrimidine pathway has been extensively developed and employed in the treatment of autoimmune diseases such as rheumatoid arthritis (RA) and the prevention of allograft rejection (20, 21). Thus, suppression of de novo pyrimidine biosynthesis can dramatically lead to immune remission, including T lymphocytes.
Our study highlights the effects of the EGFR-TKI afatinib on CD8+ T lymphocyte activation and proliferation. Using a systematic antibody-based approach, we were able to reveal the drug's targeting profile and delineate the molecular mechanism of afatinib's action on CD8+ T lymphocytes. Moreover, we observed a delicate modulation of CD8+ T lymphocyte proliferation in patients with NSCLC who received afatinib-targeted therapy for up to 48 weeks during the treatment course. During the early period of treatment, afatinib suppressed CD8+ T lymphocyte proliferation, while CD8+ T-cell adaptation to the targeted therapy occurred after long-term treatment. This clinical phenomenon encouraged us to use a sequential combination of afatinib-targeted therapy and immunotherapy in cancer treatment, in addition to a simultaneous combination of both treatments. Our data further demonstrated that a sequential combination of afatinib and anti-PD1 immunotherapy could dramatically improve therapeutic efficacy in MC38 and Lewis lung carcinoma (LLC) tumor-challenged mice. Together, the information gathered in this study not only delineates the molecular mechanisms of afatinib's inhibition of T lymphocyte proliferation, but also provides insight into T-cell modulation with EGFR-TKI treatment and into treatment timing for a sequential combination of afatinib and immunotherapy.
Materials and Methods
Written informed consent was obtained from healthy volunteers or patients with NSCLC with EGFR mutations who received afatinib as first-line treatment. The study was approved by the National Taiwan University Hospital (Taipei, Taiwan) Institutional Review Board (201604086RINA).
Six-week-old C57BL/6 mice were purchased from the National Laboratory Animal Center, Taipei, Taiwan. All animals were housed in the animal facility of National Taiwan University College of Medicine (Taipei, Taiwan) under specific pathogen-free conditions. Animal care and use corresponded to the guidelines approved by the Committee of Animal Care at our institution.
Blood sample collection and processing
Ten milliliters (mL) of blood samples from patients with lung cancer and healthy volunteers were drawn by peripheral venipuncture into BD Vacutainer Heparin Tubes. To isolate peripheral blood mononuclear cells (PBMC), 10 mL of whole blood were diluted with PBS (1:1) and gently layered on top of 10 mL Ficoll-Paque PLUS in a 50-mL tube, and centrifugated at a speed of 400 × g for 30 minutes. After the PBMCs were processed (22, 23), the cells were cultured in RPMI1640 media supplemented with 10% FBS and 1% l-glutamine. To examine the effect of afatinib on T-cell activation, PBMCs were pretreated with afatinib for 1 hour and then stimulated with anti-CD3 (1 μg/mL) and anti-CD28 (1 μg/mL) antibodies for 24 hours (early activation) to 72 hours (late activation and proliferation). To investigate whether afatinib affected T lymphocyte activation in patients with lung cancer who received targeted therapy, T cells in patients' PBMCs were stimulated by anti-CD3 (1 μg/mL) and anti-CD28 (1 μg/mL) antibodies for 24 to 72 hours. Cells were then subjected to flow cytometry analysis.
CFSE proliferation assay
Human PBMCs or mouse lymphocytes (1 × 106 cells/mL) were labeled with the vital dye carboxyfluorescein diacetate succinimidyl ester (CFSE, Sigma-Aldrich, catalog no. 21888) at a concentration of 2.5 μmol/L at 37°C for 10 minutes and kept from exposure to light. After labeling, human PBMCs or mouse lymphocytes were washed twice with complete RPMI1640 media supplemented with 10% FBS and 1% l-glutamine or complete T-cell media (CTM) to remove unbound CFSE after centrifugation at 400 × g for 3 minutes. Stained cells were resuspended in complete RPMI1640 media supplemented with 10% FBS and 1% l-glutamine or CTM, stimulated with anti-CD3 and anti-CD28 antibodies for 72 hours, and then subjected to flow cytometry analysis.
Afatinibs target protein identification using TISTA, in-gel digestion, and LC/MS-MS analysis
To identify the potential target protein(s) of afatinib in T cells, we employed the newly developed approach of “Target identification by specific tagging and antibody detection (TISTA)” (24). Briefly, the procedure is as follows: Jurkat T cells were treated with 1 μmol/L afatinib for 24 hours. Cells were then lysed with a lysis buffer (1% Triton X-100 in TBS) containing protease inhibitor cocktail (MCE, catalog no. HY-K0010). Cells lysates were used for immunoprecipitation (IP) with an anti-afatinib antibody (24). IP samples were boiled in 2× Laemmli sample buffer for 10 minutes and subjected to SDS-PAGE, Western blot analysis, and Coomassie stains. The gels with protein bands were excised from the polyacrylamide gels. The isolated proteins in gels were reduced with 1,4-dithioerythreitol (50 mmol/L) at 56°C for 45 minutes, and alkylated with iodoacetamide (100 mmol/L) at room temperature for 1 hour in the dark. Gel slices were dehydrated with 100% acetonitrile for 5 minutes and vacuum dried for 5 minutes. In-gel tryptic digestion was carried out at an enzyme-to-substrate ratio of 1:20 at 37°C for 16 hours. The tryptic peptides were extracted twice with 50% acetonitrile containing 5% formic acid under moderate sonication for 10 minutes and then vacuum dried completely. The peptide mixtures were desalted and isolated with a C18 Zip-tip (Millipore) and subjected to LC/MS-MS analysis (Thermo Fisher Scientific, LTQ-Orbitrap Velos).
ADP-Glo kinase assay
The ADP-Glo Kinase Assay (Promega) was followed by a modified standard protocol for CAD enzyme activity assay (25), and carried out in solvent containing 0.1 mol/L Tris, pH 7.5, 0.1 mol/L KCl, 2.5% glycerol, 25 mmol/L MgCl2, 1 mmol/L dithiothreitol, and 5 mmol/L NaHCO3. Purified hamster CAD proteins were obtained from Santiago Ramón-Maiques [Centro de Biología Molecular Severo Ochoa (CSIC-UAM), Madrid, Spain]. Before the enzymatic reaction, purified hamster CAD proteins were preincubated with afatinib (DMSO as a control) at 37°C for 5 minutes. The reaction was initiated by adding ATP, and then incubated at 37°C for 15 minutes. The luminescent signal was measured using an Epoch Microplate Spectrophotometer (BioTek Instruments, Inc.).
LC/MS-MS metabolomics profiling
For analysis of the metabolomics profiles, human PBMCs or mouse lymphocytes were seeded at a density of 1 × 106 cells per mL, treated with or without afatinib for 1 hour, and then T-cell activation was stimulated with anti-CD3 and anti-CD28 antibodies for 72 hours. For metabolite extraction, cells were taken, washed with ice-cold PBS, and lysed with ddH2O. Small metabolites were then extracted with acetonitrile and dried in a SpeedVac. The samples were subjected to LC/MS analysis as described previously (23). For detailed information, please see the Supplemental Materials and Methods section.
Analysis of tumor-infiltrating lymphocytes
LLC cells were inoculated subcutaneously at a density of 2 × 105 cells into the right flank of each C57BL/6 mouse. As tumor volumes reached 200 mm3, each animal was randomly assigned to treatment with afatinib (10 mg/kg) and ddH2O by oral gavage once daily for 6 days. After the mice were sacrificed, isolation of mouse tumor-infiltrating leukocytes (TIL) was performed using a standard protocol, as described previously, with some modifications (26). The detailed procedure of isolating mouse TILs is described in the Supplementary Data. The numbers of CD8+ T lymphocytes in the TILs were measured by flow cytometry. A portion of the tumor tissues was fixed using 4% paraformaldehyde and preserved for tissue embedding, dissection, and IHC staining.
For in vitro studies, we obtained afatinib compounds from LC laboratories and dissolved them in DMSO solvent with a stock concentration of 10 mmol/L for the coming experiments. Thus, DMSO was used for the control experiments. For in vivo animal studies, we were concerned that DMSO might bring on strong cytotoxicity and induce unwanted side effects in the mice. Hence, we adopted the clinical practice guideline for afatinib-targeted therapy in patients with NSCLC, and used the commercial product GIOTRIF (afatinib), which has been modified in the form of a dimaleate salt to increase solubility in water (Supplementary Fig. S1A and S1B).
Combined afatinib treatment and anti-PD1 immunotherapy using MC38- or LLC-bearing mice
MC38 or LLC cancer cells were subcutaneously inoculated at a density of 5 × 105 or 2 × 105 cells into the right flank of each C57BL/6 mouse. As tumor volumes reached approximately 100 mm3, the mice were randomly assigned to the vehicle or treatment group. The mice were then treated with a vehicle (ddH2O plus IgG2a isotype), afatinib (10 mg/kg), anti-PD1 (10 mg/kg), a combination of afatinib (10 mg/kg) and anti-PD1 (10 mg/kg), or afatinib (10 mg/kg) prior to anti-PD1 (10 mg/kg) therapy. Tumor volumes were measured every 2–3 days using a digital caliper. The study endpoint was recognized when tumor volumes reached up to 2,000 mm3 or tumor burdens were greater than 15% of body weight.
Statistical significance was determined by one-way ANOVA or Student t test using Prism 6 software (GraphPad). In all circumstances, P ≤ 0.05 were considered significant. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
To determine whether afatinib could affect CD8+ T lymphocyte activation and proliferation, we isolated human PBMCs from healthy volunteers. PBMCs were treated with afatinib in the concentrations from 0 to 1,000 nmol/L, because plasma concentrations of afatinib in patients who orally take afatinib at 40 mg/day have been shown to range from 100 to 800 nmol/L (27–29). T lymphocyte proliferation and activation were then stimulated with anti-CD3 and anti-CD28 antibodies for 24 hours (early activation) and 72 hours (late activation and proliferation). After flow cytometry, the results showed that afatinib could significantly inhibit the proliferative capability of human CD8+ T lymphocytes (Fig. 1A), and suppress T-cell activation in higher concentrations (Fig. 1B and C). Similar results were also observed in mouse T lymphocytes after afatinib treatment (Supplementary Fig. S2A–2C). To further validate the direct effects of afatinib on CD8+ T lymphocytes, mouse CD8+ T lymphocytes were isolated and then subjected to cell proliferation and apoptosis assays. The results showed that afatinib could also suppress CD8+ T lymphocyte proliferation in a dose-dependent manner (Supplementary Fig. S3A and S3B). Moreover, afatinib had no significant effect on cell apoptosis (Supplementary Fig. S3C). The results together indicate that afatinib can suppress CD8+ T lymphocyte proliferation.
To delineate the molecular mechanisms of how afatinib inhibited T-cell proliferation and activation, we used an anti-afatinib antiserum for the immunoblot and IP of afatinib-targeted proteins (24). The immunoblot results showed that afatinib could covalently modify many cellular proteins in Jurkat T cells (Fig. 2A). Immunoprecipitated proteins using an anti-afatinib antiserum were subjected to SDS-PAGE and Coomassie blue staining (Fig. 2B). The protein identities in the #1–6 regions of the gel were revealed after LC/MS-MS analysis and 29 proteins were identified in the afatinib-treated cells (Supplementary Table S1). A gene ontology analysis of the potential afatinib-targeted proteins further revealed a strong enrichment of molecular functions in the ATP and protein kinase binding categories, and in biological processes, including the biosynthetic process, cellular component biogenesis, nucleobase-containing metabolism, and mitotic cell-cycle regulation (Fig. 2C). Among them, carbamoyl-phosphate synthetase 2, aspartate transcarbamoylase, and dihydroorotase (CAD) gained our attention because of the high scores, the ATP requirement for its enzymatic reaction, and the de novo pyrimidine biosynthesis essential for T-cell proliferation and function (18). Moreover, the results showed a covalent interaction between afatinib and CAD after IP and immunoblot analyses (Fig. 2D and E). We then further constructed the plasmids encoding full-length CAD, CAD domain 1 [glutaminase and carbamoyl phosphate synthetase 2 (CPS2)], CAD domain 2 (dihydroorotase), and CAD domain 3 (aspartate transcarbamoylase; Supplementary Fig. S4A). HEK293T cells were then transfected with those plasmids, treated with or without afatinib, and then subjected to IP of CAD proteins and immunoblot analysis. The results showed that afatinib could covalently interact with the full length and domain 1 of CAD in cells (Supplementary Fig. S4B). To further identify the modification site(s) tagged by afatinib, immunoprecipitated CAD proteins from CAD-overexpressing HEK293T cells after afatinib treatment were subjected to LC/MS-MS analysis. The mass spectrometry (MS) results revealed that the amino acid residue of cysteine 758 of CAD protein was found to be modified with afatinib (Supplementary Fig. S5A), which was located in the ATP-binding motif of CAD domain 1. Moreover, we used the SwissDock molecular docking program to explore the possible interaction mode of afatinib in the binding sites of human CPS2. The results of molecular docking revealed that the amino acid residue of cysteine 758 in the ATP-biding site of human CPS2 was essential for a covalent linkage with afatinib (Supplementary Fig. S5B). Thus, the results together indicate that afatinib can bind to CAD, and CAD's amino acid residue of cysteine 758 is in an important position for interaction with afatinib.
CAD is responsible for the first three reactions of pyrimidine biosynthesis (19), first, by catalyzing the foremost ATP-dependent biosynthesis of carbamoyl phosphate from glutamine and bicarbonate; second, generating N-carbamoyl-L-aspartate from carbamoyl phosphate and aspartate; and third, producing dihydroorotate from N-carbamoyl-L-aspartate, up to UMP (Fig. 3A). To further explore whether afatinib could affect the de novo pyrimidine biosynthesis in immune cells, we employed a MS-based metabolomics approach to analyze the levels of intracellular metabolites after afatinib treatment. The results showed that afatinib could differentially suppress the production of CAD-mediated metabolites, including N-carbamoyl-L-aspartate (the domain 3 product of CAD) and dihydroorotate (the domain 2 product of CAD), and UMP in Jurkat T cells and human PBMCs (Fig. 3B–G), while carbamoyl phosphate (the domain 1 product of CAD) was undetectable due to its instability using metabolomics analysis (30). Moreover, mouse CD8+ T lymphocytes were isolated for analyzing the effects of afatinib, first-generation EGFR-TKIs (Irresa and Tarceva), and canertinib on CAD-mediated metabolites. The results showed that afatinib, rather than the three other EGFR-TKIs, could significantly reduce the amounts of CAD's products (N-carbamoyl-L-aspartate and dihydroorotate) and have a tendency to reduce its downstream metabolites (UMP and CMP), suggesting that afatinib can restrict CAD-mediated pyrimidine metabolism (Supplementary Fig. S6A–S6D). To delineate whether afatinib could directly inhibit CAD's activity, we used a luminescent ADP-Glo assay to measure ADP production after the reaction, and found that afatinib could competitively suppress the enzymatic activity of CAD to a 10:1 molar ratio (Fig. 3H and I). The interaction between afatinib and the purified CAD proteins was confirmed after Western blot analysis (Fig. 3J). Thus, afatinib can repress CAD enzymatic activity to interrupt de novo pyrimidine biosynthesis.
To investigate whether afatinib also suppresses cell-cycle progression, we performed cell-cycle analysis. The results indicated that afatinib treatment arrested the cell cycle at the G1 phase, with no significant effect on the sub-G1 phase of human T lymphocytes (Fig. 4A and B) or on the apoptosis of Jurkat T cells (Fig. 4C). Moreover, afatinib could decrease the protein levels of cyclin D3 and CDK4, and increase the protein levels of p18 CDK inhibitor (CKI), with no significant effect on the protein levels of cyclin D1 and CDK6 (Fig. 4D). We further found that afatinib could repress CAD by reducing the activating phosphorylation level of CAD's Thr456 residue (Fig. 4E). To further examine whether afatinib could reduce CAD-mediated T-cell proliferation, Jurkat cells were transiently transfected with CAD plasmids and then treated with the indicated concentrations of afatinib. The results from cell proliferation assays showed that overexpression of CAD could increase cell proliferation. CAD-increased cell growth was suppressed by 250 nmol/L afatinib, down to the control cells without afatinib treatment (Supplementary Fig. S7A and S7B). Hence, the results together indicated that afatinib can repress T lymphocyte proliferation rather than induce apoptosis, which would act via suppression of CAD.
To explore whether afatinib could impede the tumor infiltration of CD8+ T lymphocytes in vivo, we subcutaneously inoculated LLC cells into C57BL/6 mice. When the tumor volumes reached 200 mm3, the LLC-bearing mice were randomly divided into two groups: a control and an afatinib group, and kept under daily oral gavage for 6 days (Fig. 5A). The animal body weights and tumor volumes were not significantly altered with afatinib treatment (Fig. 5B and C), and IHC staining of CD8+ T lymphocytes was reduced in the afatinib-treated group (Fig. 5D). We then measured the tumor-infiltrating lymphocytes (TILs) by counting the intratumoral CD8+ T lymphocytes using flow cytometry. The results showed that afatinib treatment significantly decreased the numbers of CD45+CD8+ T lymphocytes (Fig. 5E) in tumor lesions. Moreover, afatinib could modulate the immune responses and cell proliferation of intratumoral CD8+ T lymphocytes (CD8 TILs) by downregulating the numbers of CD45+CD8+FasL+ T lymphocytes (Fig. 5F), CD45+CD8+PD1+ T lymphocytes (Fig. 5G), and CD45+CD8+Ki67+ T lymphocytes (Fig. 5H). However, the percentage of the CD8+ T lymphocyte population in the TILs was not obviously changed in the afatinib-treated group (Supplementary Fig. S8). Thus, these results imply that afatinib has an immune modulation effect on CD8 TILs and limits their lytic potency as an antitumor immune response.
To further investigate whether afatinib could also affect T lymphocytes in patients with lung cancer, we assessed CD8+ T lymphocyte responses in the serial peripheral blood of 9 patients with NSCLC who received afatinib-targeted therapy. For T-cell proliferation assays, the gating strategy of dividing CD8+ T lymphocytes after CFSE labeling and flow cytometry analysis was depicted in Fig. 6A. The results indicated that afatinib treatment decreased the proliferation capability of CD8+ T lymphocytes initially in the patients with lung cancer who received afatinib. Unexpectedly, afatinib-suppressed CD8+ T lymphocyte proliferation rebounded differentially in 2, 4, or 24 weeks after the treatment, some even with a higher proliferation rate than the individual starting point (Fig. 6B). This would suggest a modulation of CD8+ T lymphocytes during the treatment course. Similar trends could be found in in vivo experiments when C57BL/6 mice received serial time courses of 10 mg/kg/day afatinib treatment on day 1, day 2, day 7, and day 14. The results showed that afatinib treatment decreased the proliferation capability of CD8+ T lymphocytes on day 1 and day 2, while CD8+ T lymphocyte proliferation rebounded in mice on day 7 and day 14 after the treatment course (Supplementary Fig. S9A and S9B). The results further suggest that CD8+ T cells can be modulated and gradually adapt to afatinib treatment.
We also analyzed the T lymphocyte activation markers (CD69 and CD25) and inflammatory cytokine IFNγ of patients' peripheral CD8+ T lymphocytes, and found they were not significantly changed after afatinib treatment (Supplementary Fig. S10A–10C). Furthermore, to examine whether there exists an adaptive mechanism to CD8+ T lymphocytes in afatinib treatment, we recruited an additional 5 patients with NSCLC who received over 24-week afatinib treatment and 5 volunteers with NSCLC without treatment, and collected their PBMCs for examination of the growth inhibition (GI50) values of afatinib on CD8+ T lymphocyte proliferation. The results revealed that the afatinib-treated group had higher GI50 values than the untreated group (Fig. 6C). This suggested that CD8+ T lymphocytes in patients undergoing long-term afatinib-targeted therapy become less sensitive to afatinib. Together, the results indicate that afatinib exhibits an immune modulation effect on the CD8+ T lymphocyte proliferation of patients with NSCLC.
These clinical observations and the mouse CD8+ T-cell modulation to afatinib-targeted therapy encouraged us to examine whether a sequential combination of afatinib-targeted therapy and immunotherapy showed improvement in cancer treatment. To test this concept, C57BL/6 mice were inoculated subcutaneously with MC38 cells. Once the tumor sizes reached 100 mm3, the mice were randomly assigned to five groups for treatment; the groups included a control, afatinib, αPD1, simultaneous combination of afatinib and αPD1, and a sequential combination with afatinib prior to αPD1 immunotherapy. The treatment course is schemed in Fig. 7A. The results showed that sequential combination treatment with afatinib prior to αPD1 immunotherapy significantly improved therapeutic efficacy in suppressing tumor growth, compared with the control, afatinib, αPD1, and simultaneous combination of afatinib and αPD1 therapy groups (Fig. 7B and C). Moreover, combination therapy with afatinib prior to αPD1 therapy yielded better overall survival than the other treatment groups (Fig. 7D). IHC staining was performed to analyze the numbers of tumor-infiltration T lymphocytes after the mice were sacrificed. The results showed that the numbers of CD3+ and CD8+ T lymphocytes in tumor lesions were significantly increased in the sequential combination of afatinib and anti-PD1 antibodies group (Fig. 7E). Similar results were also observed in LLC tumor-challenged mice (Supplementary Fig. S11). The sequential combination with afatinib prior to anti-PD1 antibodies could enhance their efficacy against LLC tumor growth (Supplementary Fig. S11A–S11D), and increase the numbers of CD8+ T lymphocytes, proliferative CD8+ T lymphocytes (Ki67+) and activated CD8+ T lymphocytes (IFNγ+) in the tumor lesions (Supplementary Fig. S11E–S11G). Thus, the results together indicate that a sequential course of afatinib prior to anti-PD1 immunotherapy can elicit the antitumor immune response of cytotoxic CD8+ T lymphocytes, at least partly by promoting the population of tumor-infiltrating CD8+ T lymphocytes.
It has been known that EGFR signaling participates in the maintenance of regulatory T (Treg) cells in vitro and in vivo (31). EGFR-TKIs then have been considered to increase immune function by downregulation of Tregs and immune checkpoints (32). However, current results from clinical trials of combined EGFR-TKI and immunotherapy have shown no additional benefits in lung cancer treatment (8–12). These clinical trial outcomes encouraged us to ask whether afatinib could affect CD8+ T lymphocytes. In this study, we found that afatinib could inhibit CD8+ T lymphocyte proliferation and activation. Moreover, afatinib treatment could also reduce the numbers of tumor-infiltrating CD3+CD8+ T lymphocytes and the expression levels of PD1 in LLC-bearing mice. Because the amounts of tumor-infiltrating CD8+ T lymphocytes (TIL) have been shown to be strongly associated with their antitumor effects (33, 34), the reduction of TILs by afatinib suggests that afatinib has a suppression effect on the antitumor activity of T cells. The data thus provide an explanation as to how afatinib-inhibited cytotoxic T-cell proliferation, activation, and tumor infiltration may dampen the efficacy of simultaneous combinations with ICB therapies in treating human cancer.
To understand how afatinib could suppress CD8+ T lymphocyte proliferation and function, we used the TISTA approach to isolate afatinib-targeted proteins in T lymphocytes, because there is little EGFR expression in CD8+ T lymphocytes. We identified CAD to be an afatinib-targeted enzyme, and found that afatinib can inhibit CD8+ T lymphocyte proliferation via repressing CAD function and pyrimidine biosynthesis. Nucleotide metabolism is important in numerous cellular processes, including T-cell proliferation. In fact, expansion of the nucleotide pool is necessary for T lymphocyte proliferation and survival during mitogen-stimulated activation (18). Targeting de novo pyrimidine biosynthesis has been developed as an immunosuppressant strategy in the treatment of severe RA and prevention of allograft rejection (20, 21). Thus, our data provide further molecular evidence that EGFR-TKI afatinib-suppressed T-cell proliferation occurs via targeting the rate-limiting enzyme of pyrimidine biosynthesis, CAD.
How can we explain the EGFR-TKI afatinib's targeting of non-tyrosine kinases such as CAD in T cells? One possible explanation for afatinib-suppressed CAD may be the fact that afatinib was originally designed as an ATP analogue, which may also fit into CAD's ATP-binding pocket (35). Afatinib's novel targeting of CAD, in addition to pan-EGFRs, provides further evidence to support the theory that each small-molecule drug used in targeted therapy is estimated to have an average of 6.3 target proteins (14). Thus, the small-molecule drugs designed for targeted therapy may not only be specific to the original targets, but also to other unidentified targets in cells, all of which contribute to therapeutic efficacy in cancer treatment or is a reason for the side effects of the drug (36, 37). To reveal the detailed mechanisms of the actions of targeted therapy drugs in precision medicine, we developed a systematic approach to isolate afatinib-targeted protein(s) in lung cancer cells (24) and CD8+ T lymphocytes. This antibody-based approach to identifying drug targets proved to be useful to discovering the molecular target of afatinib action in the suppression of T lymphocyte growth and activation. This can serve as a quick method to reveal a drug's target profile and to help understand the detailed mechanism of drug actions in precision medicine.
Also, using TISTA, we found that afatinib has many target proteins, in addition to EGFR, in human lung cancer cells (24) and in Jurkat T cells (Fig. 2). From the list of afatinib-targeted proteins, we identified ribonucleotide reductase (RNR) to be an afatinib-targeted protein in lung cancer cells (24), but RNR was not shown in the target protein list of T cells after TISTA using an anti-afatinib antibody and LC/MS-MS analysis. Instead, we found that CAD was an important target protein of afatinib in Jurkat T cells. Thus, the results together indicate that a targeted drug exhibits multiple target proteins and has a different favorite target or a differential interactome in different cell types.
Our results from the analysis of the PBMCs of patients with NSCLC receiving afatinib treatment showed an unexpected tendency in which CD8+ T lymphocyte proliferation was initially suppressed after afatinib treatment, followed by a rebound stage after long-term follow-up (Fig. 6B). To explore possible explanations for the modulation of T lymphocytes to afatinib, we performed T-cell receptor (TCR) repertoire analysis for the patients with NSCLC who received afatinib-targeted therapy (9 naïve patients and 9 afatinib-treated patients who were in the rebound phase after treatment). The results from the repertoire overlap analysis showed that there were obvious differences in TCR, especially in shared CDR3 amino acids sequences, between the naïve and the treatment group (Supplementary Fig. S12). The results suggest that CD8+ T lymphocytes can modulate and/or rejuvenate themselves in the face of afatinib-targeted therapy. Another possible molecular mechanism for T lymphocyte modulation to afatinib may be the upregulation of multiple drug-resistant proteins, such as MDR1 and ABCG2. A third possible explanation for the restoration of T-cell proliferation after long-term afatinib treatment might be T-cell regeneration, because our data showed that some rebound timing for T-cell proliferation in the patients was more than 3 weeks (general T-cell turnover time; ref. 38). Thus, the exact mechanism CD8+ T-cell modulation to afatinib treatment needs to be further investigated. Because several recent studies have shown that the efficacy of ICBs tends to be better in patients with NSCLC after treatment with EGFR-TKIs targeted therapy (39–41), our unexpected findings of T lymphocyte modulation to afatinib-targeted therapy provide a rationale-based approach involving sequential treatment of afatinib prior to αPD1 therapy to achieve the benefit of combination therapy.
The results in Supplementary Fig. SS9, showing that CD8+ T lymphocyte proliferation in mice was suppressed after 1 and 2 days of afatinib treatment, and rebounded after 7 and 14 consecutive days of treatment, further support the concept that the modulation of CD8+ T lymphocytes to afatinib is common and is a gradual and continuous process. However, due to rapid tumor growth in the syngeneic mouse models, in which the tumor volumes of vehicle mice reached almost 2,000 mm3 of maximal tumor burdens on day 11, it was difficult to extend the time scale for the sequential treatment of afatinib prior to anti-PD1 therapy after the CD8+ T lymphocytes had fully adapted to afatinib treatment. To make the animal study feasible, we had to adjust the sequential therapeutic regimens by shortening afatinib treatment from 7 to 5 days, followed by anti-PD1 on days 6, 9, and 12. This experimental procedure follows animal usage guidelines and is useful in examining whether sequentially therapeutic intervention exhibits any improvement in therapeutic efficacy or overall survival. The choice of 5 days for afatinib treatment prior to anti-PD1 immunotherapy is due to the following two reasons: (i) 5-day afatinib treatment in mice is approximately equal to 225-day treatment in human beings, and (ii) the adaption of CD8+ T lymphocytes to afatinib treatment in patients with NSCLC often occurs 180 days after the treatment. Thus, 5-day afatinib treatment in mice can reflect the rebound phase of patients receiving 6–7 months of afatinib-targeted therapy.
Together, our results show the molecular mechanisms of afatinib's effects on CD8+ T lymphocytes via targeting the pyrimidine anabolic enzyme CAD, an unexpected observation of T-cell modulation to targeted therapy after long-term treatment, and the treatment course of afatinib prior to immunotherapy was able to improve the effects of combination therapy. The findings may provide new insight into treatment timing for combined targeted therapy and immunotherapy.
C.-C. Ho reports grants from AstraZeneca as well as personal fees from Eli Lilly, Roche/Genentech/Chugai, MSD, Boehringer Ingelheim, Novartis, BMS and Ono Pharmaceutical, and Pfizer outside the submitted work. M.-S. Lee reports grants and personal fees from Ministry of Science and Technology, Taiwan as well as grants from National Health of Research Institutes, Taiwan, National Taiwan University, and National Taiwan University Hospital during the conduct of the study. G.-D. Chang reports a patent for patent filed not yet issued pending. No disclosures were reported by the other authors.
H.-F. Tu: Data curation, formal analysis, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. C.-J. Ko: Conceptualization, data curation, investigation, methodology. C.-T. Lee: Software, formal analysis. C.-F. Lee: Conceptualization, data curation, formal analysis, methodology. S.-W. Lan: Conceptualization, formal analysis, methodology. H.-H. Lin: Conceptualization, methodology. H.-Y. Lin: Methodology. C.-C. Ku: Conceptualization, resources, data curation, software, methodology. D.-Y. Lee: Conceptualization, resources, data curation, software, formal analysis. I.-C. Chen: Formal analysis, validation. Y.-H. Chuang: Conceptualization, resources, supervision. F. Del Caño-Ochoa: Resources, software, formal analysis, supervision, methodology. S. Ramón-Maiques: Resources, software, formal analysis, supervision, methodology. C.-C. Ho: Resources, supervision, funding acquisition, validation, investigation, methodology, project administration, writing–review and editing. M.-S. Lee: Conceptualization, data curation, supervision, funding acquisition, validation, investigation, project administration, writing–review and editing. G.-D. Chang: Conceptualization, resources, supervision.
This study was supported by Taiwan Ministry of Science and Technology grants MOST 104-2320-B-002-044-MY3, MOST 106-2320-B-002-046-MY3, and MOST 108-2320-B-002-024-MY3, National Health Research Institutes grants NHRI-EX106-10401BI and NHRI-EX109-10725BI, National Taiwan University grants NTU107L890504 and NTU110L893503 to M.-S. Lee, and National Taiwan University Hospital grants 106-003451, 107-003849, 108-004269, and 109-004720 to C.-C. Ho. This work was also supported by MINECO grants BFU2016-80570-R and RTI2018-098084-B-I00 (AEI/FEDER, UE). The authors would like to thank the Laboratory Animal Core Facility at the College of Medicine, National Taiwan University for their services.
They sincerely thank Zee-Fen Chang for sharing her knowledge and experience with nucleotide metabolism and CAD enzymatic experiments. Also, they sincerely thank Shiou-Ru Tzeng for sharing her experience and assistance in molecular docking and binding mode analysis. They thank the clinical staffs at National Taiwan University Hospital and all study participants. Finally, they thank the service provided by the Flow Cytometric Analyzing of the Third Core Laboratory, National Taiwan University College of Medicine, and the laboratory animal center in the National Taiwan University College of Medicine for its animal care.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.