Multifunctional PD-L1 Targeted Fe 3 O 4 @DOX Nanoparticles Under MRI Monitoring For Combination Therapy of Triple-Negative Breast Cancer

Background: The triple negative breast cancer (TNBC) is the most dicult subtype of breast cancer to treat, with currently restricted treatment of chemotherapy. However, evidence suggests that immunomotropy is only effective for a small portion of patients. The combination therapy of both immunotherapy and chemotherapy has demonstrated great promise in its treatment eciency. However, this combination therapy is limited by both the organ toxicity of the chemotherapeutic drugs and the inaccessibility of the in vivo monitoring of individual tumor response to therapies. With the development of the nanomedicine, synchronous targeting therapy, imaging diagnosis and monitoring can be achieved convenientlly using the multifunctional nano-platform for TNBC. Results: In our study, we prepared the DOX loaded Fe 3 O 4 nanoparticles targeting the PD-L1 (PD-L1@Fe 3 O 4 -DOX, FPD) to obtain the real-time magnetic resonance imaging (MRI) monitoring and combination therapy of both chemotherapy and immunotherapy for TNBC. The results showed that FPD inhibited tumor growth more effectively than either DOX chemotherapy or PD-L1 immunotherapy alone. Conclusion: Our study demonstrated that FPD has shown a great potential for theranostics and clinical translation in synchronous MRI imaging-guided monitoring and combined treatment of both DOX and immunotherapy of TNBC.


Introduction
Breast cancer is a severe challenge in women health. The incidence rate of breast cancer has generally been rising over the past 50 years with rapid increases observed particularly in developing countries [1,2]. Different subtypes of breast cancers arise from different gene mutations occurring, causing di culty in breast cancer diagnosis and treatment [3]. Among these subtypes, triple-negative breast cancer (TNBC) is highly heterogeneous and is one of the most severe forms of breast cancer [4].
Surgical removal is generally the main treatment for early breast cancer. However, compared with other subtypes, TNBC patients have shown higher risk of postoperative recurrence and distal metastasis, and lower survival rate. Furthermore, due to the lack of the expression of progesterone receptor (PR), estrogen receptor (ER), and human epidermal growth factor receptor (HER2) in TNBC, current molecular targeted therapy and hormone therapy are ineffective in treating TNBC patients [5]. Therefore, the traditional chemotherapy remains the primary treatment option for patients with TNBC at intermediate and advanced levels [6].
Doxorubicin (DOX) is used in many chemotherapy regimens due to its prominent and broad-spectrum antineoplastic characteristics [7]. The main mechanism of action of DOX is that the product embeds into DNA and inhibits the synthesis of nucleic acid, showing the strong cytotoxic effect. As a type of periodic non-speci c drug, DOX shows a wide anti-tumor spectrum and is commonly used in the treatment of breast cancer [8]. However, the application of DOX has been hindered by toxicities such as hematopoietic suppression, nausea, extravasation, and alopecia, with cardiotoxicity as the most severe side-effect, leading to cardiomyopathy and/or congestive heart failure. Till now, there still exists an urgent medical demand to explore pharmacotherapeutic strategies and reduces the side effects, that can improve the treatment outcome of TNBC [9].
Immunotherapy provides a new option and direction for TNBC treatment. The host immune responses to tumor cells can be induced by tumor-in ltrating lymphocytes (TILs) present within the tumor microenvironment, while the TNBC is more likely to harbor TILs than other subtypes of breast cancer.
Tumor cells can overexpress immune checkpoint-related molecules to inhibit T cell activity and escape the body's immune response to survive [10]. The immune checkpoint inhibitors (ICBs) can inhibit the activity of immune checkpoint molecules to release the activity of T cells and restore the killing effect on tumor cells to ultimately achieve the anti-tumor effect [11]. At present, there are mainly two types of ICBs, including CTLA-4 receptor inhibitors and PD-1/PD-L1 inhibitors [12]. However, a large number of studies have shown that the single drug treatment of immune checkpoint inhibitors has limited bene ts for TNBC patients with the objective response rate (ORR) ranging from 5% in PD-1/PD-L1 positive population to 23% in rst-line PD-L1 positive population [13].
Studies have shown that several chemotherapeutic agents commonly used in breast cancer treatments can promote immunogenic cell death, thereby releasing tumor cell antigens and promoting immune response [14]. Therefore, chemotherapeutic agents combined with immune checkpoint inhibitors may achieve enhanced therapeutic effects [15]. However, this combination therapy is limited by the organ toxicity of the chemotherapeutic drugs and the inaccessibility of in vivo monitoring of individual tumor response to therapies. Therefore, the current focus of TNBC treatment is to improve the therapeutic effect and to reduce drug toxicity and side effects. It has been suggested that effective combination therapy and improvement on the targeting of drugs for diagnosis and treatment are the main potential solutions to these problems [16,17].
Furthermore, various imaging techniques, including infrared scanning, mammography, computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, and positron emission tomography (PET), may provide the diagnosis, assessment, and monitoring for breast cancer [18]. Among these imaging technologies, MRI has become one of the most important screening methods for breast diseases due to its signi cant resolution on soft tissues and high detection sensitivity (~ 96%), which can reduce the number of false positives in mammography examination and avoid unnecessary biopsy. However, the limitation of MRI in the actual clinical practice is that MRI is not capable of achieving targeted imaging of lesions and real-time monitoring of therapeutic drugs [19][20][21].
In recent years, with the rapid development of nanomedicine, multifunctional nanosystems with multiple diagnostic and therapeutic effects have shown signi cant potential in the real-time targeted and accurate diagnosis and treatment of malignant tumors, providing the potential solutions to the problems of TNBC [22]. Studies have demonstrated that as an ideal targeted contrast agent for MRI, the iron oxide nanoparticles (Fe 3 O 4 -NP S ) have shown higher biological safety, relaxation, and adjustable surface modi cation than both the manganese oxide nanoparticles and clinical chelated gadolinium contrast agents. Furthermore, due to its sound magnetic characteristics and biocompatibility, the surface of Fe 3 O 4 -NP S is easy to be modi ed, making Fe 3 O 4 -NP S the suitable materials to construct targeted drug delivery carrier system [23,24].
In our study, we developed and characterized a MRI-based Fe 3 O 4 -NP S with targeted PD-L1 antibody and linked DOX (PD-L1@Fe3O4-DOX, FPD). This nanohybrid technology aims to establish and enhance the synchronous MRI monitoring and combined chemotherapy and immunotherapy for TNBC, ultimately to improve the therapy e cacy in the treatment of TNBC.

Synthesis and modi cation of Fe 3 O 4 nanoparticles
The Fe 3 O 4 nanoparticles were synthesized based on a previous study with some modi cation [25]. Brie y, 10.8 g FeCl 3 ·6H2O and 36.5 g of sodium oleate were dissolved in a mixed solvent composed of 60 mL distilled water, 80 mL ethanol, and 140 mL hexane. The mixed solution was stirred and heated to 70°C and kept for 4 h. The upper organic layer was then collected, washed with distilled water, and dried by evaporation to obtain the waxy material, i.e., the iron-oleate complex. The iron-oleate complex was mixed well with 5.7 g oleic acid and 200 g of 1-octadecene, then heated to 320°C and kept for 30 min. The resulting mixture was cooled to room temperature and added with an excessive amount of ethanol to precipitate the nanoparticles. The solution was centrifuged in an Eppendorf Centrifuge (Model: Male BALB/c mice (4 weeks old) were purchased from the Animal Center at Shandong University. These mice were 4T1 injected with 1 × 10 7 4T1 cells in their right mammary glands. The mice were treated when the tumors grew to the approximately 100 mm 3 . Experiments were approved by the Ethics Committee of Qilu Hospital of Shandong University and conducted according to institutional guidelines for animal care.

Cytotoxicity of nanomaterials
The cytotoxicity of Fe 3 O 4 nanospheres on cell viability was evaluated by the Cell Counting Kit-8 (CCK-8) assay (Beyotime, Jiangsu, China). The 4T1 and MCF-10A cells were seeded on the 96-well plates at a density of 8.0 × 10 3 cells/well to complete the adhesion of 24 h. The cells were added with various concentrations of Fe 3 O 4 (25, 50, 100, and 200 µg/ml) and were incubated for 6, 12, 24, and 48 h, resepctively. The control cells were cultured in the regular medium without any stimuli. A total of 10 µl of CCK8 solution was added to each well, and the plates were incubated at 37°C for 2 h. Absorbance was measured at 450 nm (Molecular Devices, VT, USA). Cell viability of controal cells was considered 100%.

Cellular uptake of nanoparticles
The cellular uptake of nanoparticles was evaluated by prussian blue staining, which was used to detect the presence of iron element, following the procedures reported previously [27]. The cells were observed under a light microscope to determine the intracellular distribution of nanoparticles.
To further verify in vitro the targeting and cellular uptake of the nanoparticles, the concentration of Fe was measured by inductively coupled plasma-mass spectrometry (ICP-MS; Agilent Technologies, UK) following the procedures reported previously [28]. The number of Fe nanoparticles in a single cell was calculated by the equation: n = (N/D)/M, where N was the total number of atoms, D was the cell density in the ICP-MS sample, and M was the mass of a single Fe nanoparticle.
Following incubation, cells were collected with a scraper, washed with PBS, and centrifuged at 500 rpm, with the pellet resuspended in PBS for MRI study. The nanocomposite was tested for MRI signal intensity To further quantitatively evaluate the local effect of DOX, ROS staining was performed using Dichlorodihydro uorescein diacetate assay (DCFH-DA) following the manufacturer's protocol. After each group was cultured based on the treatment methods described above, the cells were washed thrice with PBS and incubated in DCFH-DA at 37°C for 20 min. The cells were observed under a uorescence microscope. The percentage of ROS-positive cells were calculated by dividing the total number of ROSpositive cells by the total number of cells in the same eld.
Material relaxation rate and in vivo MRI imaging MRI were performed using a clinical 3.0-T MRI scanner (GE compony Discovery MR750) with an 8 circular head coil at room temperature. Fast spin echo (FSE) T2-weighted images (T2WI) and T2-mapping were acquired, and T2-mapping was also acquired using single section multi-spin-echo sequences. The detailed acquisition parameters of T2-weighted images were as follows: TR/TE = 5000/111 ms, FOV = 100 mm, matrix of 256×256, section thickness of 2 mm, a circular region of interest was selected in each sample, and the values of T2 relaxation times were obtained. The T2 relaxivities of FPD were calculated from the slope of the linear plots of the r2 relaxation rates (1/T2) versus iron concentration. The increase in the r2 relaxation rates (1/T2) with increasing iron concentration was analyzed by the linear least squares regression analysis. At least four mice per group were included in this analysis. The carrier-tonoise ratio (CNR) in tumor was calculated as follows: CNR = [Tumor SNR (Signal-to-noise ratio) -Muscle SNR] / (standard deviation of background).

E cacy of FPD in vivo
The mice with 4T1 tumor were divided into 5 treatment groups (n = 5): Starting from Day 0, the mice were intravenously injected with different groups of nanoparticles or PBS and the tumor diameters were measured by a digital caliper every three days for a total of 15 days. The tumor volume in mm 3 was calculated using the following formula: Volume = 0.5 × a × b 2 , where a and b were the largest and the smallest diameters, respectively. The body weight of mice was measured every 3 days with the survival of the mice also determined.

Tissue preparation
The mice were anesthetized and killed 15 days after the experiments with the tumors removed, weighed, and sectioned. Tumor tissue and major organs were removed and xed in paraformaldehyde for over 2 days, then embedded in para n and sectioned at a thickness of 3 µm.

Hematoxylin and Eosin (HE) staining
The morphological characteristics of tumor tissue and major organs were observed based on HE staining of the para n sections, which were dewaxed and dehydrated with a gradient of alcohol of a series of concentrations. The HE staining was performed using an HE staining kit (Beyotime, Jiangsu, China) according to the manufacturer's instructions. The sections were observed under a microscope (Olympus, BX60, Tokyo) Tunel staining Sections of tumor tissue and major organs were incubated with proteinase K solution for 15 min followed by incubation with TUNEL reaction mixture (Beyotime, Beijing, China). After washing thrice with PBS, the sections were mounted with a mounting solution containing the nuclear dye DAPI and observed using either a conventional uorescence microscope or a confocal laser uorescence microscope.

Flow cytometry
The number of CD3 + CD8 + T cells in the tumor was studied by ow cytometry based on the gating strategies as reported previously [29]. The T cells were labeled with both anti-CD3 (Abcam, ab16669) and anti-CD8 antibodies (Abcam, ab203035).

Statistical analysis
All data were analyzed with SPSS (version 17.0; SPSS, USA). The data were presented as the mean ± standard error of the mean (SE) with the statistical signi cance (p < 0.05) determined using Student's ttest.

Characterization of nanomaterials
The morphological characteristics and crystal structural features of synthesized nanoparticles were investigated by TEM and HRTEM, respectively. As shown in Fig. 2A, the nanoparticles were of uniform shape with the size of ~ 20 nm. The HRTEM image (Fig. 1B) Fig. S1C, Fig. S1D, and Fig. 1C, respectively. The conjugation of DOX molecule showed no evident effect on the morphology of the Fe 3 O 4 nanoparticles (Fig. S1C). However, after the conjugation of PD-L1 antibody, the surface of Fe 3 O 4 was coated by a layer of organic component, probably due to the PD-L1 antibody with high molecular weight (Fig. S1D). Similarly, after the conjugation of both DOX molecule and PD-L1 antibody, the Fe 3 O 4 nanoparticles were also coated by an organic layer (Fig. 1C). The XRD pattern showed that the synthesized nanoparticles corresponded to the standard PDF card (19-0629) ( Fig. 1D; Fig. S1A), indicating that the nanoparticles were the pure Fe 3 O 4 phase with high crystallinity. These results were consistent with those of the HRTEM observations. The superparamagnetic property of the synthesized Fe 3 O 4 nanoparticles was measured through the saturation magnetization curve based on a vibrating sample magnetometer with the saturation magnetization at room temperature con rmed as large as 53 emu/g (Fig. 1E). To con rm the modi cation of DMSA molecule on the surface of Fe 3 O 4 nanoparticles, the FTIR spectra were analyzed  (Fig. 1I). These results were consistent with those based on TEM observations.

In vitro imaging
We investigated the feasibility of imaging protein targets PD-L1 in 4T1 cells (Fig. 2A). The results showed that PD-L1 was widely expressed on the surface of 4T1 cells, con rming its feasibility as a target.
To further verify the targeting of nanoparticles and the effectiveness of MRI in 4T1 cells, post-uptake MRI was performed on non-targeted (Fe 3 O 4 ) and targeted (FP) groups of nanoparticles (Fig. 2B). The results showed that T2WI of the targeted group was signi cantly lower than that of the non-targeted group, and the signal intensity in the targeted group decreased with the increase of concentration, indicating that the uptake of nanoparticles in the targeted group was more than that in the non-targeted group (P < 0.001) in a concentration-dependent manner (Fig. 2C).
To see the in vitro uptake properties of these nanoparticles, we carried out Prussian blue staining and iron assay on 4T1 cells (Fig. 2D, E). Under the same concentrations (10-40 µg/mL), FPD was stained stronger than Fe 3 O 4 group. ICP-MS was performed to detect the iron content (Fig. 2E). Compared with Fe 3 O 4 group, the FPD group showed signi cantly higher uptake for iron content (P < 0.001).
The cellular uptake of nanoparticles and ultrastructural features in 4T1 cells were observed using TEM (Fig. 2F). More FP/FPD nanoparticles were internalized compared to Fe 3 O 4 nanoparticles with the nanoparticles gradually peaked at 4 h and dissipated at 24 h.

Therapeutic e cacy of nanoparticles in vitro
The CCK-8 method was used to investigate the toxicity of Fe 3 O 4 nanoparticles to cells. In 6, 12, 24, and 48 h, these nanoparticles did not show noticeable cytotoxicity toward 4T1 and MCF-10A cells (Fig. 3A).
Double staining of Hoechst 33342 and PI was used to evaluate cell apoptosis (Fig. 4B). The results showed that the nuclei of apoptotic cells in FD and FPD groups emitting a shrunken, fragmented with strong blue and red uorescence. The results of the quantitative assessment showed that the survival rates of normal cells in FPD group were lower than those in FD group ( Fig. 3C; P < 0.05).
The ROS level in the FD/FPD group was signi cantly higher (P < 0.05) than that in the Fe 3 O 4 /FP group, indicating that DOX led to elevated ROS levels in the cells, while the ROS level of FPD group was signi cantly higher than that of FD group (Fig. 3D, E).

MRI imaging in vivo
We measured the relaxation rate of FPD to evaluate the performance of FPD as a contrast agent in MRI. T 2 WI images and pseudo color map of FPD at 0.025325, 0.05065, 0.1013, 0.2026, and 0.4052 mM were shown in Fig. 5A. The tting curve (Fig. 4B) obtained showed that the relaxation rate of FPD was 146.43 mM − 1 s − 1 , which is far high the relaxivities of usual MRI contrast agents, indicating that FPD can be used as a negative contrast agent for T 2 WI.
Next, T2-weighted MRI scans were acquired before and after the intravenous injection of FPD at different time points (1, 4, 8, and 24 h) for the mice with 4T1 tumor. As shown in Fig. 4C-D, MRI showed that the lowest signal value was achieved in the tumor site at 4 h after probe injection both in Fe 3 O 4 and FPD group, but a more signi cant decreased MRI signal value in the tumor region was detected in FPD group (P < 0.001). And the signal value at tumor site gradually recovered to its initial state 24 hours after nanoparticles injection. These results were consistent with the results of cell imaging, supporting the speci c targeting of FPD to tumors. FPD aggregates much more in tumor sites than Fe 3 O 4 , suggesting that FPD can be used as an optimal probe for in vivo MRI to improve the accuracy of biomedical imaging.

E cacy of combination therapy in vivo
To determine thesynergistic bene t of DOX combined with PD-L1 immunotherapy, we performed the in vivo antitumor experiments.
The mice with 4T1 tumor were divided into 5 groups and treated with PBS, Fe 3 O 4 , FP, FD, or FPD, respectively. The body weight and tumor volumes in mice were continuously monitored to assess the antitumor e cacy for a total of 15 days (Fig. 5A). No signi cant differences in tumor volume were observed between the PBS and the Fe 3 O 4 groups (Fig. 5B, C). The volume of tumor in FD group were signi cantly smaller than that in the FP group, while the FPD group showed the smallest tumor volume, indicating the highest antitumor e cacy. The survival rate of mice with different treatments was monitored to the end of the experiment. The data suggested that the treatments of DOX and PD-L1 group increased the survival rate compared to other two groups, the increased potential tumor therapeutic effects due to the combination therapy of DOX and PD-L1 immunotherapy (Fig. 5D). Besides, the mouse body weight had no signi cant difference among the ve groups of mice until the end of the trial (Fig. 5E).

E cacy of in vivo chemotherapy
To further verify the therapeutic effect of the nanomaterials on tumor tissue, the TUNEL apoptotic assay was performed on the tumor tissues (Fig. 6A, B). Results showed that the therapeutic effect of FPD on tumor tissues was signi cantly increased in comparison to that of other groups (P < 0.01). Furthermore, in order to further verify the biosafety of the nanomaterials, we also performed TUNEL staining on the major organs of the mice to mark apoptotic cells. Results showed that few apoptotic cells were observed on the major organs, including the heart, liver, spleen, lung, and kidney (Fig. 6C, D). The qRT-PCR results of three apoptosis-associated genes (Caspase-3, bax and bcl-2) revealed that FPD was not only lethal to the tumor tissue, but also safe compared with the control group (Fig. 6E). Moreover, the results of HE staining also revealed that the nanoparticles showed low organotoxicity (Fig. S2).

E cacy of in vivo immunotherapy
We performed immuno uorescence on PD-L1 in tumor tissues to further verify the feasibility of PD-L1 as a target of the nanoparticles (Fig. 7A). The results of Prussian blue staining on tumor tissue sections showed that the tumor absorption capability in the FPD group was signi cantly higher than that in the control group (Fig. 7B).
In order to verify the e cacy of immunotherapy in vivo, the contents of CD3 + CD8 + T cells in tumor tissues were measured. Flow cytometry test results showed that the content of CD3 + CD8 + T cells in the FPD group was signi cantly higher than that in the control group and FD group (Fig. 7C, D), while there were more CD3 + CD8 + T cells in the FD group than that in the control group, indicating the synergistic effect of chemotherapy and immunotherapy on the tumor.

Discussion
Fe 3 O 4 is a type of ferromagnetic material, while the nano-spheres based on Fe 3 O 4 have a wide range of applications in biomedical and bioengineering elds due to their special physical, chemical, and biological properties [30]. Therapeutic drugs based on Fe 3 O 4 nanoparticles are gradually attracting increasing attention in medical areas due to its integration of both imaging and therapeutic functions into a single nanoagent, enabling accurate cancer diagnosis and personalized medical treatment as well as real-time monitoring of drug distribution and evaluation of treatment outcomes [31,32]. In recent years, with the rapid development of nanotechnology, the diagnosis and treatment of multifunctional and multimodal nanocarriers in the eld of biomedical applications have shown a signi cant potential.
Several types of iron oxide nanoparticles have been evaluated in preclinical and clinical trials with some already available as commercial products on the market. It is reasonably expected that a variety of Fe 3 O 4based diagnosis-to-drug complexes will be routinely used in clinical practice in the coming decades [33].
In our study, we developed a PD-L1@Fe 3 O 4 -DOX (FPD) nano-drug delivery system. Our results provide a novel integrated multifunctional nanosystem for medical diagnosis and treatment.
With increasing incidence and survival of TNBC, the discovery of novel therapeutic approaches is required to advance the treatment outcomes of patients [34]. Recently, nanoparticles are regarded as a novel solution to improving the therapeutic e cacy and reducing the systemic toxicity in antitumor treatment owing to their smaller size, shorter circulation time, and larger modi ed surface area in comparison to the traditional materials. Different from traditional drug-delivery system, the nanoparticles with modi ed surface can bind to receptors expressed at the target site and can be incorporated into cancer cells through its pharmacological action [35]. In routine clinical practice, the combination chemotherapy for breast cancer is often the preferred treatment plan. The bene ts of multidrug combination therapy include increasing or maintaining e cacy by reducing dose, reducing toxicity, and delaying drug resistance. Due to these advantages, combination therapy has now become a routine application strategy in clinical practice [36,37]. In our study, we synthesized FPD through modi cation to achieve the effect of combination therapy of tumor in mice. We further performed the ICP-MS assay, the Prussian blue staining assay, the Hoechst 33342/PI uorescence double staining assay, and the ROS assay to investigate the cellular uptake, targeting, and killing ability of FPD cells. The results showed that, on the one hand, Fe 3 O 4 (before modi cation) and FP only attached to PD-L1 antibody showing no evident damage to cells with sound biosafety; on the other hand, FPD combined with PD-L1 antibody and chemotherapy drug DOX showed an increased killing effect on tumor cells, with FPD more e cient than FD.
We performed MRI imaging to evaluate the PD-L1 as a biomarker for targeted TNBC imaging and therapy. Our results revealed a relatively lower signal for FPD in the tumor regions compared to Fe 3 O 4 , indicating that FPD can speci cally target the 4T1 tumors and accumulate at the tumor site for a longer time than the non-targeted Fe 3 O 4 control. For MRI, the Fe chelates are still commonly used as the agents for tumor diagnosis [38,39]. In contrast, FPD signi cantly enhanced the e cacy of nanoparticles as MRI contrast agents.
Finally, we used Fe 3 O 4 to load DOX to signi cantly improve the biosafety. On the one hand, the nanospheres may protect the DOX from the degradation by relevant immune cells or enzymes in the blood, though we note that more studies are needed to verify these speculations. On the other hand, nanospheres enhnace targeting, effectively prevent DOX deposition in other organs, and improve biosafety, suggesting that the doses of DOX can be increased in the treatments.
In conclusion, the FPD nano-drug loading and monitoring system synthesized in this study can reduce the   Hoechst 33342/PI double staining. ROS was quanti ed (D) by DCFH-DA staining (E). Data are presented as mean ± SD. Symbols *, **, and *** indicate the signi cant differences based on Student's t test set at P < 0.05, P < 0.01, and P < 0.001, respectively. Black asterisks indicate the comparison with Fe3O4treatment group. Red asterisk indicates the comparison with the FD-treatment group. circles marked the tumor areas. Symbols *, **, and *** indicate the signi cant differences based on Student's t test set at P < 0.05, P < 0.01, and P < 0.0001, respectively. Data are presented the mean ± SD of at least ve animals for each group.   indicate the signi cant differences based on Student's t test set at P < 0.05, P < 0.01, and P < 0.0001, respectively. Data are presented as the mean ± SD of at least ve animals for each group.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.