Background
Glioblastoma (GBM) is the most common primary malignant brain tumour in adults and is associated with an extremely aggressive clinical course and poor prognosis [
1]. The median progression-free survival in primary GBM is 6.9 months, and the median overall survival is 14.6 months with standard-of-care surgery, radiation therapy, and temozolomide [
2,
3]. Consequently, there is a high unmet clinical need for new treatment paradigms yielding more durable remissions.
The current neurosurgical management of GBM aims for maximal resection while avoiding additional neurological damage. Numerous methods have been developed to facilitate surgery, including 5-aminolevulinic acid (5-ALA) fluorescence-guided surgery, intraoperative neuro-navigation, and neurophysiological monitoring [
4,
5]. However, GBM recurrence is almost inevitable due to residual areas of diffuse microscopic infiltration of tumour cells into the surrounding brain parenchyma and intratumoural heterogeneity at the cellular and molecular levels.
Approximately 57% of GBMs contain a mutation, rearrangement, splicing alteration, and/or amplification of the epidermal growth factor receptor (EGFR). The most common EGFR variant is a deletion of exons 2–7, EGFRvIII, which often co-occurs with focal EGFR amplification, which together are associated with a more aggressive, immuno-evasive tumour phenotype and worse prognosis [
6]. Despite the well-known role of EGFR in GBM, the potential of targeting the receptor with tyrosine kinase inhibitors (TKIs) as well as monoclonal antibodies (mAbs) have been unfulfilled so far. Furthermore, a phase III study (ACT IV), for newly diagnosed patients with GBM treated with Rindopepimut, an EGFRvIII-targeted vaccine, also failed to demonstrate a survival benefit [
7].
Interestingly, recent studies have shown that inhibiting EGFR signalling may reduce tumour cell-intrinsic EGFR-induced programmed death-ligand 1 (PD-L1) upregulation, as well as extrinsic IFNγ-induced signals associated with CD8+ T cell infiltration into the tumour microenvironment (TME) [
1,
8]. However, attempts to incorporate immune checkpoint inhibitors (ICPIs) into GBM treatment regimens have demonstrated only modest and unpredictable responses [
9,
10]. This is most likely due to low burdens of somatic mutations and a relatively immune-depleted (“cold”) GBM microenvironment characterised by a high level of immunosuppressive cytokines (e.g. TGFβ, IL-10) which inhibit immune effector cell activity [
11]. Excitingly, several research groups have reported that high-level infiltration of immune effector cell populations, including CD8+ cytotoxic T-lymphocytes (CTLs), into the TME can improve response to ICPIs in GBM [
12,
13]. Therefore, in a clinical context, it would be desirable to restore intratumoural infiltration of CD8+ T cells to create an immunologically “hot” TME and, thus, promote the responsiveness of GBM to ICPIs.
One way to activate the TME immunologically would be through the use of photoimmunotherapy (PIT) and conventional photodynamic therapy (PDT).
PIT is a light-mediated therapeutic approach, where a photosensitiser (PS) is conjugated to a highly specific monoclonal antibody (mAb), antibody fragment, or affibody molecule that has the ability to engage the selected target of interest. Near-infrared (NIR) light irradiation of the conjugate lead to ligand release reaction of IR700 and under normoxic conditions to the production of heat and reactive oxygen species (ROS) that, consequently, initiate target-selective cell death and stimulate inflammation, followed by vascular shutdown and tissue ischaemia [
14‐
16]. For example, Nagaya et al. have shown that anti-CD44-IR700-mediated PIT can significantly delay tumour growth following a single treatment in three CD44-expressing syngeneic mouse models of oral squamous cell carcinoma [
17]. In addition, NIR-PIT targeting EGFR with anti-can225-IR700 resulted in rapid cell death in vitro and tumour growth inhibition in vivo, improving mouse survival [
18]. More importantly, EGFR-targeting IR700-cetuximab (ASP-1929, Akalux™, Rakuten Medical, Inc.) is currently being investigated in a global phase III clinical trial in head and neck cancer (
NCT03769506) [
19] and was registered for clinical use in Japan [
20]. Furthermore, it has been shown that both PIT and PDT can trigger immunogenic cell death (ICD), as exemplified by the release of damage-associated molecular patterns (DAMPs), including calreticulin (CRT), heat shock proteins HSP70/90, ATP, and high-mobility group box-1 (HMGB1) nuclear protein that subsequently activate immune cells upon binding to pattern recognition receptors [
21].
In view of the high expression rate and oncogenic nature of EGFR, we have postulated that PIT targeting this receptor could promote CD8+ T cell attraction and activation and overcome the immunologically “cold” status of GBM.
As an alternative to full-size antibodies, we have previously investigated the smaller, IR700-labelled EGFR-specific affibody molecule (Z
EGFR:03115-IR700), aiming for more effective tumour penetration, faster delivery, and clearance from non-targeted tissues [
22]. After demonstrating that Z
EGFR:03115-IR700 cell uptake enables imaging of EGFR expression in an orthotopic brain tumour model (U87-MGvIII), our proof-of-concept in vivo PIT study also showed the conjugate’s therapeutic efficacy in subcutaneous glioma xenografts [
22].
In the current study, we report that ZEGFR:03115-IR700-PIT promotes the production of DAMPs from cancer cells, also leading to dendritic cell (DC) maturation in vitro. In addition, when applied in a syngeneic mouse model, the treatment induces T cell responses that might overcome the “immunologically cold” status of GBM. Therefore, we believe that this therapeutic approach, following complete or cytoreductive resection of GBM, could lead to (i) elimination of residual or surgically inaccessible EGFR+ve cancer cells and (ii) subsequent stimulation of anti-tumour immunity.
Methods
Preparation of ZEGFR:03115-IR700
The conjugation of IRDye700DX-maleimide (IR700, ex. 689 nm, em. 700 nm; LI-COR® Bioscience, USA) to the Z
EGFR:03115-Cys affibody molecules (Affibody, Sweden) is described in detail in the supporting information (Additional File
1).
Cell lines and cell culture
Human GBM cell line DKMG and murine GBM cell line GL261 were purchased from the Celther Polska (Poland) and the German Collection of Microorganisms and Cell Cultures (DSMZ, Germany), respectively. U87-MG and U87-MGvIII were kindly provided by Dr. Frank Furnari (Ludwig Cancer Research, USA) [
23]. The primary, patient-derived cell lines WSz4, WSz50, and WSz57 have been recently established in our lab [
22]. The cells were grown as described in the supporting information. BL6-NPE-GFP-Luc murine GBM cell line was kindly provided by Dr. Steven Pollard (University of Edinburgh, UK) and cultured as previously reported [
24]. The genetic origin of all the cell lines was tested and authenticated by short tandem repeat (STR) DNA profiling analysis (Eurofins Medigenomix, Germany). The cells were also routinely tested and found to be negative for
Mycoplasma contamination (PCR detection kit, Surrey Diagnostics Ltd., UK).
Singlet oxygen production assay
Singlet oxygen (1O2) production was determined using the Singlet Oxygen Sensor Green reagent (SOSG, Thermo Fisher Scientific, UK) according to the protocol provided by the manufacturer. More details about the assay are described in the supporting information.
Cellular binding of ZEGFR:03115-IR700
Human and murine GBM cells were harvested and incubated in a medium with ZEGFR:03115-IR700 (30 nM) for 1 h at 4 °C, and samples were analysed using flow cytometry (BD™ LSRII). To test the targeting specificity and internalisation of the conjugate, cells were plated on confocal glass-bottomed dishes (Thermo Fisher Scientific, USA) in complete medium with ZEGFR:03115-IR700 (1 μM) for 1 h at 4 °C or 1, 3, and 6 h at 37 °C and analysed using a Zeiss LSM700 confocal microscope (Carl Zeiss Inc., Germany). A detailed description of the procedures is given in the supporting information.
In vitro PIT studies
U87-MGvIII cells were seeded on petri dishes 24 h before experiments. Afterwards, cells were incubated with ZEGFR:03115-IR700 (0.1 to 1 μM) for 1, 3, or 6 h at 37 °C. The media were then changed for phenol red-free DMEM medium and cells irradiated (8 or 16 J/cm2) using a LED light source (L690−66−60, Marubeni America Co., USA). Cell viability was determined using the CellTiter-Glo® (Promega, USA) luminescent assay 24 h post-light exposure. To assess ROS production, 5 μM 2′,7′-dichlorofluorescein diacetate (DCFDA; Sigma, UK) was added to phenol-red free medium during irradiation. The cell death at 1, 4, and 24 h post-irradiation was assessed using the Annexin V/Dead Cell Apoptosis Kit (Thermo Fisher Scientific, UK) according to the manufacturer’s instruction. To determine the post-PIT ATP and HMGB1 release, the ENLITEN® ATP assay (Promega, USA) and an HMGB1 ELISA kit (Tecan, IBL International, Germany) were used. Calreticulin exposure on the membrane was measured by flow cytometry (BD™ LSRII). All the methods are described in detail in the supporting information.
Co-culture with dendritic cells
The experimental details about co-culturing the immature dendritic cells (iDCs) with PIT-treated U87-MGvIII or DKMG cells are given in the supporting information.
Western blot
Western blotting was performed as previously described [
22]. Proteins released into the medium were extracted using an acetone precipitation protocol (Thermo Fisher Scientific, USA). The list of antibodies used and densitometric analysis are provided in the supporting information.
18F-AlF-NOTA-ZEGFR:03115 preparation
The preparation of NOTA-Z
EGFR:03115 and its radiolabelling with the
18F-Al complex was performed as previously described [
25].
In vivo studies
All experiments were performed in compliance with licences issued under the UK Animals (Scientific Procedures) Act of 1986 and following local ethical review. Studies were compliant with the UK National Cancer Research Institute Guidelines for Animal Welfare in Cancer Research [
26] and the ARRIVE (animal research: reporting in vivo experiments) guidelines [
27].
Mouse models
The detailed methods are described in the supporting information. Briefly, NCr athymic female mice (5–6 weeks) were bred in-house. C57BL/6J female mice (6–7 weeks) used for the syngeneic model, were purchased from Charles River, UK. The orthotopic GBM U87-MGvIII or BL6-NPE-GFP-Luc mouse models were established as previously described [
22,
24]. For the subcutaneous GBM xenografts, U87-MGvIII cells were injected over the right shoulder. Once tumours reached approximately 60 mm
3, mice were randomly distributed into the experimental groups.
PIT in vivo
For PIT treatment studies, subcutaneous and intracranial GBM U87-MGvIII xenografts were randomised into the following treatment groups: (i) light exposure only (100 J/cm2) and (ii) 18 μg ZEGFR:03115-IR700 with light exposure (100 J/cm2, 0.0886 W/cm2). For immunocompetent mice bearing intracranial tumours, 50 J/cm2 light dose was used. The tumours were irradiated with a LED light source (L690−66−60, peak 690 ± 20 nm) 1 h post-conjugate i.v. injection. More details are provided in the supporting information.
MR imaging
To monitor the orthotopic tumour growth, mice were imaged using the 1 T M3™ MRI system (Aspect Imaging, Israel) with a T2-weighted imaging sequence and a dedicated head coil. To perform high-resolution acquisitions, mice were scanned using the 7 T Biospec® horizontal micro-imaging system (Biospec®, Bruker, Germany). The imaging protocols are described in the supporting information.
PET imaging
Mice (n = 5) with MRI-confirmed brain tumours received an i.v. injection of 18F-AlF-NOTA-ZEGFR:03115 (12 μg; 2.4 ± 0.15 MBq/mouse), and PET/CT scans were acquired 1, 3, and 5 h post-injection of the radiotracer using an Albira PET/SPECT/CT imaging system. The detailed imaging and data analysis protocols are given in the supporting information.
Autoradiography
Dissected tumour and brain tissue samples were collected and immediately embedded in an optimal cutting temperature compound (Tissue-Tek® O.C.T, Netherlands) and snap-frozen in liquid nitrogen.
Further experimental details are given in the supporting information.
Fluorescent imaging
In vivo and ex vivo fluorescence images were acquired as stated in the supporting information using an IVIS Spectrum/CT system (Perkin Elmer, USA).
Immunohistochemistry
Formalin-fixed brain and tumour tissues were embedded in paraffin, sectioned (5-μm-thick slices), and mounted on microscope slides. Frozen embedded tissues were sectioned into 10-μm-thick slices and mounted on microscope slides before being fixed in ice-cold acetone. The detailed staining procedures with the various antibodies are described in the supporting information.
Tumour and T cell isolation
Tumour and surrounding brain tissue were harvested and dissociated via enzymatic digestion (Liberase TL, Roche, Switzerland). Single-cell suspension was prepared by straining the digested tissue through a 70-μm mesh. Further experimental details are given in the supporting information.
Serum cytokine analysis
The serum was separated from the whole blood collected from the mice at the 24 h endpoint, snap-frozen, and stored at − 80 °C until further analysis. Concentrations of various cytokines were analysed using a Mouse Cytokine Proinflammatory Focused 10-plex Array (Eve Technologies, Canada).
Statistical analysis
Unless otherwise stated, data were expressed as the mean ± SD. Statistical significance, sample size calculations, and correlation analysis are described in detail in the supporting information.
Discussion
Extensive GBM cell invasion into the normal brain parenchyma makes complete tumour removal practically impossible and disease recurrence inevitable. Besides, the GBM TME is recognised as highly immunosuppressive, posing a major hurdle for inducing immune-mediated destruction of remaining cancer cells. As a result, clinical trials evaluating checkpoint blockade in GBM patients have failed to demonstrate clear efficacy [
9,
10]. Recently, it became clear that some treatment approaches can alert and trigger the immune response within the immunosuppressive GBM TME. For example, studies in preclinical models have shown that the combination of ICPIs with a concurrent administration of focal radiation therapy, cancer cell-directed immunotoxins, and oncolytic viruses increase anti-GBM immunity [
29‐
31]. Moreover, EGFRvIII CAR-T cell therapy induced inflammatory responses in GBM patients turning “cold” GBM microenvironment into “hotter” without inducing neurotoxicity [
32].
In the present study, we demonstrate that NIR-PIT may induce direct GBM cell killing via ICD and attracts T effector cells locally in the GBM TME. So far, mAbs-based conjugates have been most frequently utilised for PIT purposes [
33,
34]. However, the large molecular size of mAbs and their extended blood circulation may slow penetration of the proteins into the tumour parenchyma. Consequently, it may hamper the response to PIT and result in long-lasting systemic photosensitivity [
35].
To overcome such limitations, van Driela et. al. have recently demonstrated that the use of small EGFR-targeted nanobody-IRDye700DX conjugates (15 or 30 kDa) leads to higher tumour:background contrast and enhanced tumour necrosis when compared with full-size mAb-based IRDye700DX conjugate [
36]. Along the same line, our previous studies suggested that affibody molecules (~ 7 kDa) conjugated to IR700 due to their rapid tumour accumulation and blood clearance are promising candidates for PIT purposes [
22,
37].
Herein, we further demonstrated that Z
EGFR:03115-IR700-PIT can trigger a local immune response in the brain tumour microenvironment. The conjugate binding to EGFR on the membrane of GBM cells induced receptor expression-dependent cell death upon NIR light exposure which was, in part, due to ROS production. Interestingly, Kato et al. have recently provided a theoretical mechanism by which photoactivated hydrolysis reaction following irradiation of mAb-based IR700 conjugates cause changes in the silicon-oxygen bond and silanol formation, which converts the dye from very hydrophilic to very hydrophobic [
16]. Whether similar effects occur in response to irradiation of Z
EGFR:03115-IR700 will need to be investigated. Additionally, the efficacy of Z
EGFR:03115-IR700 in vitro increased in a conjugate concentration-dependent manner, and significant phototoxicity was observed within 1 h post-light exposure of conjugate-treated cells.
Of importance, several studies, including ours, provide evidence that PIT can induce mobilisation of DAMPs involved in ICD [
28,
33,
37]. These molecules serve as an “eat-me” signal and mediate anti-tumour immune responses that are critical for the efficacy of the therapy and formation of long-term immunological memory [
38,
39]. Therefore, we investigated whether irradiation of Z
EGFR:03115-IR700 will result in the release of these danger signals. We observed high-level cell surface CRT exposure, rapid ATP secretion, and HMGB1 release only in PIT-treated cells, indicative of ICD. However, in cells treated with either Z
EGFR:03115-IR700 or light alone, these signals were not enhanced compared to controls. Furthermore, significant release of DAMPs by PIT-treated GBM cells subsequently activated and promoted maturation of antigen-presenting iDCs, as indicated by a marked expression of CD86 and HLA-DR.
Thereafter, in order to determine whether the conjugate is capable of inducing selective tumour cell death in vivo, we treated mice bearing subcutaneous U87-MGvIII xenografts with ZEGFR:03115-IR700-PIT.
Burley et al. have recently reported that EGFR targeting affibody molecule (Z
EGFR:03115) with high specificity recognise EGFR in vivo. For example, the U87-MGvIII-bearing mice injected with Z
EGFR:03115-IR700 displayed a strong fluorescent signal as compared to Z
TAQ-IR700 (a non-specific affibody molecule). The tumour fluorescent intensity of Z
EGFR:03115-IR700 was 6-fold higher than Z
TAQ-IR700 already 1 h post-injection [
22]. Furthermore, when Z
EGFR:03115 was radiolabelled with zirconium-89, only very low accumulation of the radioconjugate was found in tumours with low EGFR expression levels [
25].
Apart from a targeting vector, also the light dose delivered and the method by which it is delivered are crucial to the success of PIT. However, physical dosimetry during PIT is a complex process due to the nature of dynamic interactions between light, conjugate, oxygen, and biological response of different tissues, which clearly depends on the concentration of cytotoxic photoproducts and on the intrinsic photosensitivity. In the murine models of GBM, the explicit dosimetry to map the distribution of light delivery and direct measurement of the light fluence are technically challenging. Therefore, for the purposes of this manuscript, we individually selected the intensity of light for U87-MGvIII and BL6-NPE-GFP-Luc models based on the initial validation experiments. For the xenograft model, the therapeutic light fluence was chosen to be 100 J/cm2 in order to maximise treatment efficacy considering the penetration of the NIR light and inevitable photobleaching of IR700 during the illumination. Of note, this light dose was reduced to 50 J/cm2 in the syngeneic model to lessen oedema-related swelling caused by direct cytotoxic effects on tumour cells and subsequent inflammation post-PIT. The irradiation of ZEGFR:03115-IR700 restrained the growth of subcutaneous U87-MGvIII tumours in the PIT-treated mice in comparison with controls (light only), which validated the model and procedure we employed.
Encouraged by this potent anticancer activity in vitro and in vivo, we further evaluated this approach in the brain setting. It is well known that GBM progression leads to blood-brain barrier (BBB) structural changes including neuronal death, astrocyte endfeet displacement, and heterogeneous pericyte and astrocyte subpopulations, all of which can reduce the barrier functions through the formation of fenestrations and disruption of tight junctions [
40]. Even though it makes the BBB leaky and more permeable for small and large molecules, the barrier is still considered as one of the predominant restricting factors for the efficacy of therapies intended for the clinic. Given the limitations of planar optical imaging of brain tumours and quantification of fluorescence intensity, instead of Z
EGFR:03115-IR700, we initially used the radiolabelled conjugate
18F-AlF-NOTA-Z
EGFR:03115 to assess the efficacy of the affibody molecule in targeting EGFR-positive tumours in the brain setting. The acquired PET/CT images showed discrete focal accumulation of the radiotracer in the brain lesions already 1 h post-injection. Considering the small difference in size between the two conjugates, we expected Z
EGFR:03115-IR700 to exhibit similar in vivo behaviour to
18F-AlF-NOTA-Z
EGFR:03115.
Indeed, fluorescence images of the entire brain captured ex vivo post-Z
EGFR:03115-IR700 administration clearly indicated accumulation of the conjugate in the tumour and provided insights into its delivery. Despite a relatively equal distribution of Z
EGFR:03115-IR700 in the tumours, we observed some variability in the response to PIT between the mice. This could be linked to a non-uniform irradiation through the burr hole in the mouse skull resulting in uneven NIR-light delivery and light-induced photochemical production of ROS. In spite of these issues, hypointense signals were depicted on T
2*w images of U87-MGvIII tumours within 1 h post-PIT that corresponded to microhaemorrhagic lesions. Moreover, histopathological examination of the brain sections revealed high levels of necrosis induced by irradiation of Z
EGFR:03115-IR700 24 h post-treatment. Of importance, necrosis has been previously reported to be the characteristic form of cellular death post-PIT [
41,
42]. Furthermore, cytoplasmic HSP70, a stress-inducible chaperone protein, was released from the cells as early as 1 h after Z
EGFR:03115-IR700-PIT, as confirmed by IHC staining of tumour sections. As published earlier, the translocation of HSP70 depends on the NIR light dose and is related to either mitochondrial or direct surface stress disruption [
43,
44]. Moreover, accumulating evidence suggests that HSP70 plays a role in DC maturation and activation of other antigen-presenting cells [
45]. For example, it has been reported that HSP70 secreted from PDT-treated tumour cells promoted stimulation of DC and NK cells as well as the production of pro-inflammatory cytokines [
46]. In addition, Korbelik et al. showed that HSP70 secreted post-PDT was captured by macrophages that triggered toll-like receptor-based signal transduction and production of TNFα [
47]. Finally, we used the BL6-NPE-GFP-Luc syngeneic tumour model to look into the local immune response and activation of tumour-infiltrating lymphocytes post-Z
EGFR:03115-IR700 PIT. Excitingly, we identified enhanced immunological response after conjugate irradiation which resulted in the attraction and activation of CD4+ and CD8+ T cells in PIT-treated tumours compared to the control group. Furthermore, the expression of both IL-1β and IL-6, which have the ability to enhance the immune response against tumours by activating CD8+ T cells was also markedly increased. Interestingly, we also observed that Z
EGFR:03115-IR700-PIT reduced the level of compensatory immunosuppressive PD-L1 in U87-MGvIII and BL6-NPE-GFP-Luc cells in vitro. We speculate that the remaining PD-L1+ cells could still suppress the anti-tumour immune response and allow the tumour cells to survive immunologic cytotoxicity. Of note, Kleinovink et al. have recently shown in tumour models of colon carcinoma that the addition of CTLA-4 blockade prior to bremachlorin-PDT leads to a significant reduction in tumour burden compared to either treatment alone [
48].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.