A TIGIT-based chimeric co-stimulatory switch receptor improves T-cell anti-tumor function
Tumors can employ different mechanisms to evade immune surveillance and function. Overexpression of co-inhibitory ligands that bind to checkpoint molecules on the surface of T-cells can greatly impair the function of latter. TIGIT (T cell immunoreceptor with Ig and ITIM domains) is such a co-inhibitory receptor expressed by T and NK cells which, upon binding to its ligand (e.g., CD155), can diminish cytokine production and effector function. Additionally, the absence of positive co-stimulation at the tumor site can further dampen T-cell response.
As T-cell genetic engineering has become clinically-relevant in the recent years, we devised herein a strategy aimed at enhancing T-cell anti-tumor function by diverting T-cell coinhibitory signals into positive ones using a chimeric costimulatory switch receptor (CSR) composed of the TIGIT exodomain fused to the signaling domain of CD28.
After selecting an optimized TIGIT-28 CSR, we co-transduced it along with tumor-specific TCR or CAR into human T-cells. TIGIT-28-equipped T-cells exhibited enhanced cytokine secretion and upregulation of activation markers upon co-culture with tumor cells. TIGIT-28 enhancing capability was also demonstrated in an original in vitro model of T-cell of hypofunction induction upon repetitive antigen exposure. Finally, we tested the function of this molecule in the context of a xenograft model of established human melanoma tumors and showed that TIGIT-28-engineered human T-cells demonstrated superior anti-tumor function.
Overall, we propose that TIGIT-based CSR can substantially enhance T-cell function and thus contribute to the improvement of engineered T cell-based immunotherapy.
KeywordsTIGIT T-cell engineering Tumor immunotherapy Costimulation Chimeric receptors
T-cell activation and function are dependent on multiple signals. First and foremost, a specificity signal mediated by the TCR (T-cell receptor) upon recognition of a specific antigenic peptide presented by MHC molecules is needed to activate the cell. In addition, co-stimulatory/inhibitory molecules can supply a second signal that can impact on T-cell function, proliferation and response. Amongst the different receptors participating in this second signal, CD28, ICOS, 4-1BB etc. are considered co-stimulatory and CTLA4, PD1, LAG3, Tim-3, and TIGIT enforce an inhibitory phenotype [1, 2]. The latter, TIGIT (T cell immunoreceptor with Ig and ITIM domains), is a checkpoint molecule that belongs to the poliovirus receptor (PVR)/nectin family and it was identified by Yu and colleagues . TIGIT is expressed by lymphocytes, mainly by NK cells, CD4+, CD8+ and regulatory T cells (Treg). It is crucial for balancing T cell activation and for protection from autoimmunity [4, 5, 6, 7].
Similarly to the antagonistic relationship of CTLA-4/CD28 with their ligands, TIGIT competes with a “positive” (stimulatory) receptor CD226 (also known as DNAM1). Both can bind to either of the two following ligands, CD155 and CD112, though TIGIT does so with a higher affinity [8, 9, 10]. It is important to mention that TIGIT also binds to CD155 with higher affinity than CD112 [4, 11]. TIGIT expression on naïve T cells is usually low, though it is upregulated following activation , and particularly on exhausted T cells in the tumor microenvironment (TME) . TIGIT was shown to inhibit T cell proliferation and activation upon binding to CD155 [13, 14]. Similarly to other immune checkpoint ligands, TIGIT ligands are often overexpressed in cancer cells [15, 16, 17, 18] while TIGIT is significantly upregulated in chronically stimulated or exhausted tumor-infiltrating T cells [14, 19, 20]. TIGIT activation can reduce NK cell cytotoxicity  and CTL proliferation and cytokine production via SHIP1-mediated mechanisms causing downstream inhibition of NF-kB, PI3K and MAPK pathways, and thereby diminishing the effectiveness of the cellular immune response [10, 13, 22, 23]. Moreover, high TIGIT expression on CD8+ T cells is associated with diverse malignancies include gastric cancer  and refractory hematological cancer and their relapse [14, 19, 20]. Thus, TIGIT represents an attractive target for immunotherapeutic intervention.
In the past decade, a tremendous progress was achieved in the treatment of cancer due to the development of immunotherapeutic approaches that include the use of checkpoint inhibitors, personalized cancer vaccines and the adoptive cell transfer (ACT) of tumor specific lymphocytes (either tumor infiltrating or genetically engineered T-cells) . T-cell engineering was primarily designed to endow T cells with novel specificities, and this can be achieved by expressing either a T cell receptor (TCR) or a chimeric antigen receptor (CAR) consisting of a targeting moiety (e.g., scFv) fused to an activation domain (that incorporate usually a co-stimulation portion and the CD3ζ intracellular domain).
An important difference between native TCR and CAR is the inclusion of co-stimulatory domain(s) in the latter. To recruit co-stimulation in the context of TCRs, it is possible to either transduced them with CD28 or 4-1BB [25, 26], provided their respective ligands are expressed by the target cells. Another approach is based on the use of chimeric costimulatory switch receptor (CSR) based on the exodomain of coinhibitory receptors and the endodomain of costimulatory ones . We and others demonstrated that CSRs based on PD1 can enhance T-cell function in the presence of inhibitory ligands expressed by tumors cells [28, 29]. Whereas therapeutic approaches that target immune checkpoint receptors such as CTLA4 and PD-1 have demonstrated unprecedented results in cancer patients, not all of them will eventually benefit from these treatments . Thus, it is desirable to assess the impact of targeting additional immune checkpoint receptors.
Herein, we aimed to develop and characterize a TIGIT-based CSR in the form of a chimeric receptor composed of TIGIT and CD28. We also describe an original in vitro model of T-cell hypofunction induction upon repetitive antigen exposure, in which this TIGIT CSR was able to enhance T-cell function. We were able to express high level of this chimeric receptor and we demonstrated its enhancing potential both in vitro, but more importantly, in a xenograft mouse model of human tumors.
Patient PBMCs and cell lines
All of the PBMCs used in this study were from healthy donors obtained from the Israeli Blood Bank (Sheba Medical Center, Tel-Hashomer, Israel). Melanoma cell lines HLA-A2+/MART-1+ (624.38) and HLA-A2−/MART-1+ (888) were generated at the Surgery Branch (National Cancer Institute, National Institutes of Health, Bethesda, MD) as described previously . 888A2 is an HLA-A2-transduced line derived from 888. SK-MEL23 is a HLA-A2+ melanoma cell line (CVCL_6027). A375 (CVCL_0132) melanoma is HLA-A2+/MART-1−. Adherent cells were cultured in DMEM (Invitrogen, Carlsbad, CA), supplemented with 10% heat-inactivated Fetal Bovine Serum (Biological Industries, Beth Haemek, Israel) and were maintained in a 37 °C and 5% CO2 incubator. CD19-expressing B-cell targets were Raji (CCL86), JY (CVCL_0108), 721.221 (CVCL_6263), Nalm6 (CVCL_0092). K562 (CCL_243; which is CD19 negative) was engineered to express the CD19 antigen following retroviral transduction with a CD19 encoding vector. Non-adherent tumor cells were cultured in RPMI (Invitrogen, Carlsbad, CA), supplemented with 10% heat-inactivated Fetal Bovine Serum (Biological Industries, Beth Haemek, Israel) and were maintained in a 37 °C and 5% CO2 incubator. Lymphocytes were cultured in BioTarget medium (Biological Industries, Beth Haemek, Israel) supplemented with 10% heat-inactivated FBS and 300 IU/ml IL-2 (Peprotech, Israel) and maintained at 37 °C and 5% CO2.
TCR and TIGIT chimeras retroviral constructs
Transduction of PBLs
For transient virus production, transfection of 2.5 × 105 293GP cells with 2 μg DNA of MSGV1-based retroviral construct and 1 μg envelop plasmid (VSV-G) was performed using JetPrime transfection reagent (Polyplus, France). After 4 h, the medium was replaced. Retroviral supernatant was collected 48 h after the DNA transfection. Freshly isolated PBLs were stimulated in the presence of 50 ng/ml OKT3 (eBioscience, San Diego, CA). 2 days after stimulation, lymphocytes were transduced consecutively, first with a TCR or CAR, and 24 h after this, with supernatant encoding the CSR or control. Transduction was performed in non-treated tissue culture dishes (Nunc, Rochester NY) that had been pre-coated with RetroNectin (Takara, Japan) and retroviral vectors as previously described .
Flow cytometry analysis and mAb
Fluorophore-labeled anti-human CD4, CD8, CD25, CD69, CD137, CD134 (OX40), TIGIT, CD155, CCR7, CD45RO and CD34 were purchased from BioLegend (San Diego, CA). Anti-Vβ12 antibody specific for F4 TCRβ was purchased from Beckman-Coulter/Immunotech (Marseille, France). Biotinylated Protein-L was purchased form Genscript (Piscata, NJ). Immunofluorescence, analyzed as the relative log fluorescence of gated live cells, was measured using a CyAn-ADP flow cytometer (Beckman Coulter, Brea). Approximately 1 × 104 to 1 × 105 cells (gated on live lymphocytes) were analyzed. Cells were stained in a FACS buffer made of PBS, 0.5% BSA, and 0.02% sodium azide.
Cytokine release assays
PBL cultures were tested for reactivity in cytokine release assays using commercially available ELISA kits for IFNγ, IL-2 and TNFα (R&D Systems, Minneapolis, MN). For these assays, 1 × 105 responder cells (PBL) and 1 × 105 stimulator cells (tumor cells) were incubated in a 0.2-ml culture volume in individual wells of 96-well plates. Stimulator cells and responder cells were co-cultured for 18 h. Cytokine secretion was measured in culture supernatants diluted to be in the linear range of the assay.
T-cell populations were separated using a magnetic beads-based approach for negative selection (EasySep TM - StemCell Technologies Inc., Canada).
Following a 30-min co-culture of 7 × 105 transduced T-cells with 3 × 105 melanoma targets, the cells were fixed with formaldehyde 5% and permeabilized using ice-cold 90% methanol for 20 min. Then, the cells were washed in FACS buffer, stained for either phosphorylated ERK (clone D13.14.4E - Cell Signaling Technology, Danvers MA) or Bcl-xL (clone 7B2.5 – Southern Biotech, Birmingham AL) expression using a specific antibody and analyzed by flow cytometry, gated on the lymphocyte population.
Cell mediated cytotoxicity assay
Target cells were co-cultured with transduced lymphocytes at 37 °C for 4 h, at E:T ratio of 1:3,1:6 and 1:12. All wells was completed to final volume 100ul. After the co-culture, equal volume of CytoTox-ONE™ (Promega, Madison, WI) according to manufacturer’s manual.
In vitro hypofunction induction upon repetitive antigen exposure assay
1 × 106 transduced lymphocytes were co-cultured with 1 × 105 tumor target cells. Every 2 days, the effector cells were transferred to a new culture vessel in which 1 × 105 tumor cells were previously seeded (Fig. 5a). This was repeated 4 times (i.e. a total of 8 days). At the end of this 8-day co-culture, these T-cells were tested in different assays as indicated.
Established tumor assay
6–12 weeks year-old NOD/SCID/Gamma mice (Harlan, Jerusalem, Israel) were subcutaneous injected with 1 × 106 SK-MEL23/155 cells resuspended in 100 μl HBSS medium (Biological Industries, Beth Haemek, Israel) and 100 μl Cultrex matrix (Trevigen). Two intravenous injections of 5 × 106 transduced lymphocytes resuspended in 200 μl HBSS medium were performed at day 7 and 10 after tumor inoculation. Tumor size was measured every 2–3 days using a caliper in a blinded fashion. All the procedures were performed according to the guidelines of the university committee for animal welfare.
Design and expression of TIGIT-chimeric constructs
The TIGIT receptor is a T-cell co-inhibitory molecule capable of downregulating T-cell function via binding to its ligands, often overexpressed by tumor cells. We aimed to take advantage of the presence of inhibitory ligands expressed by tumor cells to boost T-cell function using a costimulatory retargeting molecule. To this end, we designed and evaluated two TIGIT-based CSRs as described below. We hypothesized that such chimeric receptor could successfully convey positive signals to T cells following binding to TIGIT ligands. These TIGIT-based chimeras were constructed by fusing the extracellular domain of TIGIT to intracellular portion of the CD28 molecule (TIGIT-28) using a transmembrane (TM) portion derived from either TIGIT or CD28 (Fig. 1a). To enable the antigen-specific recognition of the tumor target cells, we utilized the MART1-specific TCR F4 previously characterized and used in clinical trials . Following transduction of the chimeras and the TCR into primary human T-cells, we tested the expression of these molecules by flow cytometry. To negate any difference in function deriving from a differential TCR expression between the examined experimental groups, we first performed a TCR transduction step and then used these cells for subsequent transduction with the TIGIT or control construct. We also carefully and constantly controlled for equal TCR expression following transduction. As depicted in Fig.1b, we were able to express both TIGIT-28 constructs (TM TIGIT and TM CD28) in human T-cells efficiently without any selection. However, TM TIGIT (i.e. that contained the native TIGIT TM domain) was better expressed than TM CD28–69% (MFI = 143) vs. 62% (MFI = 140) of positive cells (p < 0.05). As aforementioned, the levels of expression of F4-TCR were similar between all tested groups (approximately 50%, with MFI = 90). Overall, these levels of expression by transduced PBLs cultured in vitro was sustained for more than 30 days without selection (data not shown).
TIGIT has been shown to bind to two ligands - CD155 and, with a lesser affinity, CD112. CD155 is a co-inhibitory ligand expressed on multiple human malignant tumors, including melanoma cells [15, 16, 18, 33]. Thus, we examined CD155 expression level on multiple melanoma lines. As seen in Fig. 1c, all the melanoma lines we tested expressed significant levels of CD155 (ranging from MFI = 24 to 105; p < 0.05). Additionally, to examine the impact of CD155 expression on T-cell function, 888A2, SK-MEL23 and 624.38 melanoma lines were transduced to enforce CD155 expression (Fig. 1c).
Following transduction with TIGIT-28 or a control gene (tr.CD34), we measured the distribution of CD4+/CD8+ cells after a 10-day culture. As seen in Fig. 1d, we did not observe a statistically significant difference between the TIGIT-28 and control populations with an approximate CD4/CD8 ratio of 20%/80%. Similarly, we also assessed the memory phenotype of these different populations by staining them for CD45RO and CCR7 expression and dividing them into effector memory, central memory, EMRA (terminally differentiated effector memory cells re-expressing CD45RA) or naïve cell population. No significant differences were observed in the distribution of these populations between the different treatments (i.e., TIGIT-28 or controls). Finally, we also followed the cell count of these cells following transduction with TIGIT-28 + TCR F4, TCR F4 only or mock transduced. As seen in Fig. 1f, we observed a similar cellular growth pattern between the different group. Overall, the transduction of TIGIT-28 did not significantly alter CD4/CD8 ratio, memory phenotype or in vitro growth pattern of engineered T-cells.
TIGIT-28 (TM TIGIT) enhances T cell function and cytokine secretion
T-cells encounter a hostile environment when interacting with solid tumors. One of the main immunosuppressive protagonists is the cytokine TGFβ which can dampen critically T-cell function, T cell proliferation and IL-2 production [35, 36]. Since TIGIT-28 mediated an increase in cytokine secretion in the presence of CD155 (Fig. 2a-c), we sought to examine if this pattern would be preserved in the context of an additional inhibitory stimulus. We set up an overnight co-culture with melanoma cells in the presence of soluble TGFβ (1.25 ng/ml) and measured IL-2 secretion in the supernatant by ELISA. As anticipated, IL-2 secretion was reduced in the presence of TGFβ, when the target did or did not overexpress CD155 (Fig. 2d). Nevertheless, TIGIT-28 transduced T-cells maintained a higher secretion compared to the control (TCR only) in the presence of TGFβ (an average of 708 pg/ml vs. 137 pg/ml in co-culture with 888A2; p < 0.05). This suggests that TIGIT-28 transduced cells may function better in a hostile tumor microenvironment compared to unmanipulated cells. In conclusion, TIGIT-28-expressing T cells demonstrate an improved anti-tumor cytokine secretion capability.
Activation marker upregulation and increased pERK and Bcl-xL levels in TIGIT-28 transduced T-lymphocytes
Activation of the CD28 pathway has been shown to enhance the activation and survival of T-cells via several signaling molecules such as pERK and Bcl-xL [37, 38]. We thus assessed if TIGIT-28 engineered T-cell may actively augment ERK phosphorylation in co-cultures with tumor cells. TIGIT-28- or control-transduced TCR F4 T-cells were incubated with target melanoma cells for 30 min and then analyzed for intracellular pERK expression. As seen in Fig. 3d, we observed a significant increase in pERK expression (e.g. 35% vs. 22% of pERK-positive cells respectively; p < 0.05). No significant pERK elevation was observed in control co-cultures with the melanoma line 888 (not shown). We also examined whether TIGIT-28 could lead to an increased Bcl-xL expression in F4 transduced T cells. Following an overnight co-culture, we performed an intracellular staining of engineered T-cells with anti-Bcl-xL. As depicted in Fig. 3e, Bcl-xL expression in TIGIT-28/F4 was upregulated compared to that assessed in F4 control cells (29% vs 3%; p < 0.05). These results demonstrate that TIGIT-28 is capable of activating the CD28 signaling cascade.
TIGIT-28 can enhance the function of high affinity TCR expressing T-cells
TIGIT-28 can enhance the function of CAR T-cells depending on CD155 expression
In conclusion, TIGIT-28 can improve CAR-T cell function and this enhancement is dependent on CD155 expression by target cells.
TIGIT-28 can help rescuing hypofunctional T cells
As seen in Fig. 6b-c, after a period of 8 days, F4 only T-cells, which were subjected to long co-cultures with a MART1+/HLA-A2+ target cell line (888A2), displayed a dramatic reduction in cytokines secretion, akin to a hyporesponsive state; for example, when normalized to the IFNγ secretion observed by TCR-F4 only T-cells previously incubated with antigen negative targets during 8 days, that of T-cells incubated with 888A2 went down by 80% (i.e., 100% vs. 20.3% respectively) in co-cultures with the SK-MEL23/155 target. In contrast, TIGIT-28/F4 T cells exhibited much higher cytokine secretion profile (reaching in average 90.1% of the positive control; p < 0.05 – Fig. 6b). This beneficial effect mediated by TIGIT-28 was not due to a differential TCR expression following these long co-cultures as F4 TCR levels (measured by flow cytometry) were similar in T-cells exposed to antigen-negative, antigen-positive targets (Additional file 1: Figure S2). This demonstrates that TIGIT-28 is able to mitigate the effects of prolonged antigen exposure on T-cell function.
TIGIT can limit lymphocyte functionality by downregulating the expression of surface receptors such as CD226 (known also as DNAM1) which conveys positive signals . This mechanism has further ramifications as TIGIT and DNAM1 directly compete for the binding of the ligand CD155 . Thus, we also assessed DNAM1 expression in T cells subjected to this 8-day co-culture compared to that observed in T-cells before this treatment. As seen in Fig. 6d, TIGIT-28/F4 equipped cells were able to maintain, after several days in co-cultures with target cells, higher levels of DNAM1 surface expression compared to TCR F4-only T cells (87% vs. 72% of DNAM1 positive cells respectively on day 5 – Fig. 6d). Thus, TIGIT-28 may improve T-cell function and activation phenotype also in the case of continued challenge with tumor cells.
TIGIT-28 mediates superior anti-tumor cytotoxicity in xenograft model
The TIGIT/CD155 inhibitory axis is an attractive target as CD155 (PVR, necl-5) is overexpressed in multiple cancer types including colon cancer, lung adenocarcinoma, melanoma, ovarian, breast, pancreatic cancer and glioblastoma and its expression is correlated with poor prognosis and tumor proliferation [13, 17]. To derive benefit from inhibitory ligand overexpression on tumors, we designed and expressed a CSR in the form of TIGIT exodomain fused to CD28 endo-domain. While most of the CD28-based human CSRs incorporate a TM domain derived from CD28 [28, 43, 44, 45, 46], we noticed in the present case that the TM of the original exodomain, i.e. TIGIT was more advantageous than that of CD28 (Figs. 1 and 2). This observation strengthens the need to evaluate CSR design empirically similarly to CARs . Though we describe a CD28-based CSR prototype in this work, it is conceivable that the use of alternative/additional costimulatory endodomain (such as 41BB. ICOS, CD27 or OX40) to design 1st or 2nd generation CSR with poly-functionality, especially as we have showed in the past that additional co-stimulation is beneficial in the context of TCR gene transfer . Moreover, since simultaneous blockade of TIGIT and PD-1 could enhance cytokine production, proliferation and degranulation in CD8+ TILs from melanoma patients, one could surmise that the combination of multiple CSRs based on TIGIT and PD1, each with different signaling moieties may act synergistically . As relative levels of ligands, nature and affinity of the interaction, expression levels of the different chimeras can further influence T-cell function, it is nevertheless difficult to directly compare the function of PD1–28 and TIGIT-28 CSRs as they bind to different ligands (PDL1 and CD155 respectively) on the target cells. Conversely, TIGIT has been showed to bind to other ligands besides CD155, such as CD112 or CD113 (with lower affinity though) and thus, we might infer that TIGIT/28 may also be biologically functional when binding to these alternative ligands. Additionally, as TIGIT is also naturally expressed by NK cells, one may envisage that TIGIT-based CSR may improve their function, using either CD28 or perhaps a different signaling moiety such as from the DAP family .
T-cell antigen specificity may be redirected using either TCRs or CARs . While one of the advantages of CARs is the incorporation a co-stimulatory moiety, these receptors are limited to membrane antigen targets. Alternatively, TCR can target also intracellular antigen but lack built-in co-stimulatory domains . Thus, a major advantage of CSR lies with the possibility to selectively combine costimulatory signals in the context of TCR stimulation. Interestingly, we show that CSRs can be beneficial when being engaged concomitantly not only with medium-affinity but also with high-affinity TCR (e.g., using the F5 TCR, Fig. 4), though with the latter, we observed a lower enhancement of cytokine secretion (we measured an average enhancement in cytokine secretion of 162% ± 27% in co-cultures with T-cells expressing F4 TCR compared to 48% ± 13% with F5 TCR; p = 0.0042, compared using a paired Student’s t-test). Of importance, this superior function was also independent of the type of subpopulation both in CD8+ and CD4+ T cell populations (Fig. 4c-d) or of the type of antigen-targeting receptor employed; indeed, we also showed that the function of aCD19-BBz CAR could be enhanced when co-expressing TIGIT-28 (Fig. 5). In sum, these results underline the versatility of the CSR approach.
The ubiquitous nature of CD155 expression on different tumors makes TIGIT-28 a valuable switch receptor to be used in conjunction with receptors targeting tumor antigens widely expressed such as NYESO or p53 [24, 50]. Also, it was shown that TIGIT is upregulated by T-cells in chronic viral infections [13, 14]. Thus, it is possible that anti-viral strategies based on engineered T-cells may derive benefit from the use of TIGIT-28.
An additional advantage to the present strategy, when compared to antibodies/checkpoint inhibitors, lies with the permanent nature of this modification and with the fact that it provides engineered T-cells with the opportunity to activate costimulatory pathways; the latter could facilitate T-cell persistence over time, proliferation, differentiation into memory cells and improved performance in patients. It is also reasonable to conjecture that tumors may escape from an immune response and be selected over time in vivo based on their high levels of inhibitory ligands (in our case CD155) [12, 15, 17, 24]. However, when using CSRs, overexpression of inhibitory ligands by tumors may actually be detrimental to the latter and thus, this could alternatively lead to a decrease in immunosuppression and ultimately, to a broader T-cell anti-tumor response.
In conclusion, we have demonstrated the efficacy of signal conversion by a novel CSR, namely the TIGIT-28 chimeric receptor that can work in concert with a TCR or CAR. TIGIT-28 can improve the anti-tumor activity of engineered T-cells in an antigen dependent setting leading to tumor regression. We are confident that this approach geared at the manipulation of costimulatory pathways bears important implications for the enhancement of T cell-based treatments using gene-transfer approaches.
We thank Ms. Riki Sabbag for her technical help in executing in vivo experiments work and Dr. Jennifer Benichou Israel Cohen, head of the statistical unit in the Faculty of Life Sciences, Bar-Ilan University for her advice on statistical data processing and presentation.
SH performed the experiments, analyzed the data, wrote the manuscript and prepared the figures. VE performed experiments, analyzed the data, wrote the manuscript and prepared the figures. SM performed the experiments, analyzed the data. AS performed the experiments. TB provided assistance with experiments and data analysis, CJC designed experiments, analyzed data and wrote the manuscript. All authors read and approved the final manuscript.
This work was supported by the Israel Science Foundation (1422/15).
Ethics approval and consent to participate
All the procedures were performed according to the guidelines of the Bar-Ilan university committee for animal welfare and the Israel Ministry of Health.
Consent for publication
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.