Graphene nanoribbon-based supramolecular ensembles with dual-receptor targeting function for targeted photothermal tumor therapy

Triple negative breast cancer (TNBC) is one of the most malignant subtypes of breast cancer. Here, we report the construction of graphene nanoribbon (GNR)-based supramolecular ensembles with dual-receptor (mannose and αvβ3 integrin receptors) targeting function, denoted as GNR-Man/PRGD, for targeted photothermal treatment (PTT) of TNBC. The GNR-Man/PRGD ensembles were constructed through the solution-based self-assembly of mannose-grafted GNRs (GNR-Man) with a pyrene-tagged αvβ3 integrin ligand (PRGD). Enhanced PTT efficacies were achieved both in vitro and in vivo compared to that of the non-targeting equivalents. Tumor-bearing live mice were administered (tail vein) with GNR-Man/PRGD and then each mice group was subjected to PTT. Remarkably, GNR-Man/PRGD induced complete ablation of the solid tumors, and no tumor regrowth was observed over a period of 15 days. This study demonstrates a new and promising platform for the development of photothermal nanomaterials for targeted tumor therapy.


Introduction
Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer associated with poor patient prognosis. This is due to its high tumor heterogeneity, rapid proliferation and early metastasis. 1,2 TNBC lacks the expression of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2), 3,4 which eliminates the possibility of using targeted therapies. This necessitates the use of traditional chemotherapy, radiotherapy, and surgical intervention. [5][6][7][8] Therefore, new and effective therapeutic strategies are needed to overcome the limitations/challenges (non-invasive and non-toxic) for current treatments of TNBC. 9,10 In recent years, photothermal therapy (PTT) has emerged as an attractive light-based therapeutic approach for the treatment of diseases. [11][12][13][14][15] In brief, PTT involves the light irradiation of an optical absorber (PTT agent) that is localised at the diseased location. This results in the conversion of light energy to heat, which then induces cell death. [16][17][18] In comparison to traditional therapeutics, the use of light affords a non-invasive strategy with high specicity. [19][20][21][22][23] To date, several platforms have been developed for PTT and were demonstrated in a range of disease models. [24][25][26][27][28] Graphene nanoribbons (GNRs) have displayed signicant photothermal conversion efficiencies (PCE) for potential use as PTT. [29][30][31][32][33][34][35][36] However, targeted biocompatible platforms are needed for in vivo applications to limit off-target toxicities and to improve therapeutic outcomes.
Throughout nature, glycoconjugates are known to selectively bind to carbohydrate receptors and modulate a variety of physiological and pathological processes. Such examples include cell-cell recognition, 37 signal transmission, 38 virus invasion 39 and bacterial infection. 40 Previous studies have reported the overexpression of mannose receptors (MRs) on the cell surface of tumor-associated macrophages (TAMs). TAMs have been associated with the regulation of immunity, cancer metastasis and progression of ovarian and breast cancer. 41 In recent years, researchers have exploited the high affinity of mannose for MRs to develop targeted imaging and therapeutic agents. 42,43 These efforts inspired our group to develop glycoconjugate-based materials (abbreviated as glycomaterials) for targeted PTT of TNBC both in vitro and in vivo.
To the best of our knowledge, this study represents the rst example of structurally dened GNRs for targeted tumor therapy. Previously, GNRs have been shown to have a superior photothermal-conversion efficiency exceeding those of many photothermal materials commonly used for PTT. 29,30 However, these systems were not designed for biological application. In this work, to avoid off-target toxicity and realize tumor speci-city, we have aimed to modify structurally dened GNRs with both carbohydrate and peptide ligands to produce dualtargeting GNR-Man/PRGD supramolecular ensembles for the PTT treatment of tumors. Using GNR-Man/PRGD as an in vivo PTT agent, complete ablation of TNBC-xenogra solid tumors was achieved, with no tumor regrowth over a period of 15 days. In addition, immunohistochemical assays suggested good biocompatibility of the GNR-Man/PRGD ensembles.
The sonication of GNR-Gal, GNR-Man or GNR-PEO in a phosphate buffer saline (PBS) solution (0.01 M, pH 7.40, 0.5% Triton X-100 (TX-100), v/v) afforded their stable homogeneous dispersions without visible precipitation (Fig. 1a). Transmission electron microscopy (TEM) and atomic force microscopic (AFM) images indicated that GNR-Man and GNR-Gal formed ensembles with an average diameter of ca. 35 nm and 40 nm, respectively ( Fig. 1a and b). In contrast, GNR-PEO formed spring-like nanowires with an average diameter of 20 nm and a length up to micrometer scale ( Fig. 1a and b). 29 The hydrodynamic diameters of GNR-Man and GNR-Gal were determined to be 61 nm and 86 nm by dynamic light scattering (DLS), respectively ( Fig. S3 †). The larger particle diameters determined by DLS than those by TEM are attributed to stretching of the alkyl and oligo(ethylene oxide) chains in liquid phase. Typical Raman signals corresponding to the D band (disordered band) at 1325 cm À1 and the G band (graphite band) at 1603 cm À1 were consistently observed for GNR-PEO, GNR-Gal and GNR-Man. 47 This suggests that the introduction of the glycoligands did not affect the conjugated GNR backbone (Fig. 1c). 30 In addition, the modication with hydrophilic glycosyl and PEO groups significantly improved water solubility of the GNRs with respect to GNR-COOH, which is supported by zeta potential analysis (Fig. 1d). The good aqueous dispersibility of the GNRs is re-ected in a broad-band UV-vis-NIR adsorption ranging from 400 to 1000 nm (Fig. 1e).
The targeting ability of the glycosylated GNRs were subsequently evaluated using a cellular assay. Hep-G2 (human hepatoma) and MDA-MB-231 (TNBC) cell lines were used as they overexpress ASGPrs (galactose-selective receptor) and MRs (mannose-selective receptor), respectively. The corresponding receptor expression was conrmed by real-time quantitative polymerase chain reaction (Fig. S6 †). Cell viability study without light irradiation was performed for each GNR sample, in which minimal cytotoxicity for materials ranging from 0 to 40 mg mL À1 was observed. (Fig. S7 †). To illustrate the PTT efficacy, a concentration of 40 mg mL À1 was used for cellular experiments. The targeting ability of each GNR aer light irradiation (808 nm, 1.0 W cm À2 , 3 min) was evaluated by quantifying the number of dead cells using uorescence imaging with SYTOX Green, as shown in Fig. 3, GNR-PEO displayed a minimal phototoxicity for each cell line. In contrast, treatment of the appropriate cell lines with GNR-Gal and GNR-Man (GNR-Gal -Hep-G2; GNR-Man -MDA-MB-231) followed by light irradiation led to a 3.1-fold and 2.7-fold increase in dead cells, respectively, in comparison to that of GNR-PEO (Fig. 3a-d). Minimal dead cells were observed for GNR-Gal treated MDA-MB-231 cells and GNR-Man treated Hep-G2 cells aer light irradiation. These results illustrate the potential cellular targeting ability of the glycosylated GNRs.
The photothermal effect for both GNR-Gal and GNR-Man was dependent on the GNR concentration (Fig. S8 †) and the laser power (Fig. S9 †). The receptor-dependent targeting for the GNR-based glycomaterials was demonstrated via a competition assay. The pre-incubation of free galactose with Hep-G2 and the pre-incubation of free mannose with MDA-MB-231 reduced the photothermal therapeutic effect (decreased quantity of dead cells) of GNR-Gal and GNR-Man in a concentration-dependent manner, respectively (Fig. S10 †). This further suggests that the targeted PTT effect observed is dependent on the glycoligandreceptor recognition. These preliminary results illustrate the success of our glycosylation strategy for enhancing the selective internalization of GNRs in a target cancer cell line. Given our focus was on developing a therapeutic platform for TNBC, GNR-Man was the main focus for the remainder of this study.
To further enhance the targeting capability of the GNRs, we incorporated a peptide ligand onto GNR-Man (pyrenyl-KKKRGD, denoted as PRGD). This ligand is selective for integrin a v b 3 receptors, which are expressed on the surface of a range of cancer cells including breast cancer. 49 This chosen peptide ligand was modied with a pyrene moiety to enable its association with the GNRs through p-p stacking; the chemical structure of the pyrene-functionalized peptide PRGD is shown in Fig. 4a. The supramolecular assembly between GNR-Man and PRGD was achieved through their mixing in aqueous solution, followed by sonication for 15 min. The resultant mannose/ PRGD functionalized supramolecular assemblies (GNR-Man/ PRGD) were designed to target both MR and a v b 3 receptors on the surface of TNBC cells (Fig. 4a).
The supramolecular ensemble GNR-Man/PRGD was characterised by TEM, DLS and Raman spectroscopy. Particle-like structures were observed for GNR-Man/PRGD (Fig. S11a †), which are morphologically similar to that of GNR-Man.
However, an obvious increase in the sizes of the particles was observed in comparison to that of the GNR-Man counterpart (Fig. S11b †). In the Raman spectra, the I D /I G ratio of GNR-Man (1.2) increased to 1.3 aer self-assembly with PRGD, supporting the association of the GNR with the peptide ligand (Fig. S12 †). 50 The PCE of GNR-Man/PRGD did not change compared to that of GNR-Man only (Fig. 4d). DLS analysis revealed smaller PDI values for GNR-Man/PRGD and GNR-Gal/PRGD when compared to GNR-Man and GNR-Gal (Fig. S13a †), which is indicative of an enhanced aqueous dispersibility. The absolute zeta potential values for GNR-Man/PRGD and GNR-Gal/PRGD were larger than  that of GNR-Man and GNR-Gal (Fig. S13b †), which indicates an improved aqueous stability. Additional DLS analysis showed no changes to the hydrodynamic diameter of GNR-Man/PRGD and GNR-Gal/PRGD over the course of 7 days (Figs. S13c and d †). In contrast, the hydrodynamic diameters of GNR-Man and GNR-Gal were shown to increase gradually over time, suggesting their slow aggregation. Overall, this data suggests an improved aqueous stability and solubility for GNRs/PRGD assemblies.
With these ideal solution-based properties observed for GNR-Man/PRGD, we subsequently evaluated its PTT performance against MDA-MB-231 cells. Indeed, more dead cells were observed (determined via SYTOX green staining) for GNR-Man/ PRGD treatment when compared to cells treated with GNR-Man only ( Fig. 4b and c) (Fig. S14 †). While, no dark cytotoxicity was observed for GNR-Man/PRGD (Fig. S15 †).
To illustrate the importance of the RGD targeting unit, PTT experiments were performed using GNR-PEO/PRGD and GNR-PEO. These GNRs lacked the mannose functionality, which enabled the targeting ability of the RGD unit to be solely evaluated. The light irradiation of MDA-MB-321 cells using GNR-PEO/PRGD led to the death of more cells when compared to the use of GNR-PEO. The light irradiation of HeLa cells using GNR-PEO/PRGD and GNR-PEO led to no differences in the number of dead cells being observed (Fig. S16 †). This observation supports the targeting ability of PRGD on GNRs for MDA-MB-231 cells. No cytotoxicity was observed for the use of PRGD only (Fig. S16c †).
Antibody blockade assays were subsequently carried out to evaluate the importance of the targeting units on GNR-Man/ PRGD (mannose and PRGD) for their corresponding receptors (CD206 and CD61) on MDA-MB-231 cells. Annexin V-mCherry was used to visualise apoptotic cells. 54 As shown in Fig. 5a and b, blocking receptors CD206 and CD61 with their corresponding antibodies (MRC1 and ITGB3) resulted in the reduction of the number of apoptotic cells being observed (aer PTT treatment with GNR-Man/PRGD). This antibody-mediated reduction was shown in a concentration-dependent manner, which provides support that both MR and a v b 3 receptors are targets for GNR-Man/PRGD. Moreover, increasing the PRGD concentrations on GNR-Man led to an increased number of apoptotic cells being observed aer PTT. Finally, we noticed fewer apoptotic cells being produced when cells were incubated at 4 C with GNR-Man/PRGD prior to light irradiation (Fig. S18 †). This gives evidence that the cell internalization of GNR-Man/PRGD is kinetically controlled via a receptormediated endocytic process. 55 With the excellent dual-receptor targeting ability of GNR-Man/PRGD conrmed, the PTT performance of the GNR-based agents were evaluated using tumor bearing MDA-MB-231 subcutaneous xenogras of live mice. The tumor-bearing mice (n ¼ 3) were systematically administrated with PBS (0.01 M, pH 7.40), GNR-PEO (0.5 mg mL À1 ), GNR-Man (0.5 mg mL À1 ) and GNR-Man/PRGD (0.5 mg mL À1 /0.25 mM) using tail-vein injection, respectively. Aer 24 hours post-injection, tumors were irradiated (808 nm, 1.5 W cm À2 ) for 5 min. The corresponding photothermal images were taken ( Fig. 6a and S19 †), and the time-dependent local temperature increase was recorded (Fig. 6b). In addition, the tumor volume (Fig. 6c) and body weight (Fig. 6d) of the mice were measured every 3 days within 15 days. The administration of each GNR did not affect the body weight of the mice (Fig. 6d). Interestingly, minimal temperature increases (42.1 C) were observed for the control groups, PBS and GNR-PEO. In contrast, the local temperature at the tumor aer 2 min irradiation reached 54.4 C and 67.2 C for GNR-Man and GNR-Man/PRGD, respectively. The greater local temperature at the tumor site for the GNR-Man/PRGD-treated mice suggested an enhanced accumulation of the GNRs in the tumor. Gratifyingly, for the GNR-Man/PRGD treatment group the volume of the tumor decreased to zero in three days, and the tumor did not regrow even aer 15 days. In contrast, GNR-Man resulted in reduced shrinkage of tumor volume and regrowth of the tumor was observed aer 6 days (Fig. 7a). It is important to note, that GNR-Man displayed a statistically signicant therapeutic effect against the non-targeted GNR-PEO. This highlights the importance of mannose as a targeting unit.
To conrm the signicant therapeutic efficacy of GNR-Man/ PRGD, we removed the tumors from the mice aer 15 days. No solid tumor could be isolated from the mice treated using GNR-Man/PRGD due to the disappearance of the tumor, whereas  tumors of different sizes were obtained from the mice treated with GNR-PEO or GNR-Man (Fig. 7b). Hematoxylin-eosin (H&E) staining was performed on each isolated tumor and major organs. Typical TNBC cells were seen in the control, PBS, GNR-PEO and GNR-Man groups. In stark contrast, no TNBC cells were observed in the slice of the tissue removed from the tumor region of the mice treated with GNR-Man/PRGD on the eenth day (Fig. 7c). Subsequent immunochemical staining corroborated the absence of both integrin and MR receptors for GNR-Man/PRGD treated tissue slice group compared to those treated from the other groups (Fig. 7c). In addition, no obvious morphological changes were observed for the stained organ slices (Fig. S20 †), suggesting the biocompatibility of GNR-Man/ PRGD. However, before further development extensive pharmacological evaluations are required to determine the full biocompatibility of the GNRs for live subjects.

Conclusions
In summary, we have synthesized GNR-based glycomaterials, GNR-Gal and GNR-Man, by covalently graing glycoligands to structurally dened GNRs. GNR-Man/PRGD supramolecular ensembles with dual-receptor targeting function were further developed using the supramolecular self-assembly between GNR-Man and a pyrene-tagged RGD ligand (PRGD). The successful targeting of GNR-Man/PRGD towards both MR and a v b 3 integrin receptors resulted in a highly effective PTT for TNBC in vitro and in vivo, evidenced through the nearly complete tumor ablation and no tumor regrowth being observed 15 days aer PTT. These results clearly demonstrate the potential of dual-receptor targeting for effective cancer therapy. In addition, this study illustrates the potential application of structurally well-dened GNRs as photothermal agents for tumor therapy.

Ethical statement
All animal experiments were conducted under the principles of the Regulations of Experimental Animal Administration issued by the State Committee of Science and Technology of the People's Republic of China and followed the guidelines of the Animal Care and Use Committee of Shanghai Institute of Materia Medica, Chinese Academy of Sciences.

Data availability
All experimental data, procedures for data analysis and pertinent data including NMR of compounds used are provided in the ESI.

Conflicts of interest
There are no conicts to declare.