Nanoparticle formulation of mycophenolate mofetil achieves enhanced efficacy against hepatocellular carcinoma by targeting tumour‐associated fibroblast

Abstract Hepatocellular carcinoma (HCC) is one of the most aggressive tumours with marked fibrosis. Mycophenolate mofetil (MMF) was well‐established to have antitumour and anti‐fibrotic properties. To overcome the poor bioavailability of MMF, this study constructed two MMF nanosystems, MMF‐LA@DSPE‐PEG and MMF‐LA@PEG‐PLA, by covalently conjugating linoleic acid (LA) to MMF and then loading the conjugate into polymer materials, PEG5k‐PLA8k and DSPE‐ PEG2k, respectively. Hepatocellular carcinoma cell lines and C57BL/6 xenograft model were used to examine the anti‐HCC efficacy of nanoparticles (NPs), whereas NIH‐3T3 fibroblasts and highly‐fibrotic HCC models were used to explore the anti‐fibrotic efficacy. Administration of NPs dramatically inhibited the proliferation of HCC cells and fibroblasts in vitro. Animal experiments revealed that MMF‐LA@DSPE‐PEG achieved significantly higher anti‐HCC efficacy than free MMF and MMF‐LA@PEG‐PLA both in C57BL/6 HCC model and highly‐fibrotic HCC models. Immunohistochemistry further confirmed that MMF‐LA@DSPE‐PEG dramatically reduced cancer‐associated fibroblast (CAF) density in tumours, as the expression levels of alpha‐smooth muscle actin (α‐SMA), fibroblast activation protein (FAP) and collagen IV were significantly downregulated. In addition, we found the presence of CAF strongly correlated with increased HCC recurrence risk after liver transplantation. MMF‐LA@DSPE‐PEG might act as a rational therapeutic strategy in treating HCC and preventing post‐transplant HCC recurrence.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer-related death worldwide, with a five-year survival rate of only 18%. 1 Despite advances in HCC administration, medical treatment still exhibits minimal benefit on patient survival. 2 However, increasing evidence has documented that therapeutic agents targeting tumour microenvironment (TME) may provide a promising strategy for HCC treatment. [3][4][5] Mycophenolate mofetil (MMF) is an extensively used immunosuppressive agent in liver transplantation (LT). 6 In patients, MMF is quickly converted to an active metabolite, mycophenolic acid (MPA), which selectively inhibits inosine monophosphate dehydrogenase (IMPDH) and blocks the de novo biosynthesis of guanosine nucleotides. 7 In addition to the anti-rejection effect, increasing evidence has documented that MMF possesses antitumour potency and decreases the risk of developing malignancies following LT. [8][9][10][11] More importantly, recent studies have demonstrated that MPA exhibits a potent antifibrotic activity both in vitro and vivo. 12 Treatment with MPA significantly inhibited hepatic stellate cells activation and reduced matrix accumulation in patients with chronic graft rejection. 13,14 It is well known that liver fibrosis or cirrhosis is closely related with HCC development as 80%-90% HCC developed in livers with underlying fibrotic or cirrhotic background. 15 Moreover, a group of activated fibroblasts called cancer-associated fibroblast (CAF) have been proved to participate in the regulation of tumour microenvironment and greatly enhance the proliferation and metastasis of HCC, indicating that MMF may inhibit HCC growth via targeting these cells. 16 In addition, LT has been widely accepted as an optimal treatment for HCC. 17 Nevertheless, post-transplant tumour recurrence greatly limits the efficacy of LT. Given the antitumour property of MMF, it may well be feasible to prevent HCC recurrence after LT by employing MMF. However, due to its poor watersolubility, oral administration remains the only route for this agent in clinics. And it was reported that MMF exhibited variable antitumour effects in vivo, which probably due to its poor bioavailability. 18 Hence, there is considerable motivation for the development of efficacious and safe approaches to deliver MMF in vivo, which could address the practical need of treating HCC and preventing post-transplant HCC recurrence simultaneously. Nanoparticle (NP)-mediated drug delivery system offers great promise for cancer therapy by improving pharmacokinetic properties, facilitating accumulation within solid tumours and minimizing the off-target and adverse effects. [19][20][21] Numerous studies have been attempted to deliver therapeutic agents through NPs to achieve better anti-CAF and anti-HCC efficiency. 22,23 Amphiphilic polymer materials such as poly (ethylene glycol)-block-poly (D, L-lactic acid) (PEG-PLA) and 1, 2-distearoyl-sn-glycero-3-phosphoethanolamine-N-poly (ethylene glycol) (DSPE-PEG) are widely used as biodegradable nano-carriers to construct favourable delivery systems. 24,25 Upon amphiphilic assembly of mixed drugs and matrices, a core-shell nanostructure can be obtained by packaging the hydrophobic drugs into the hydrophobic core whereas the hydrophilic components constitute the outer shell. 26 Unfortunately, some therapeutic drugs have shown limited entrapment efficiency in these delivery materials due to their unfavourable physicochemical properties. In a preliminary experiment, we attempted to encapsulate the MMF agent into both mentioned matrices but failed to obtain stable nanoformulations.
In the present study, a structurally tailored MMF prodrug was constructed through chemical derivatization of the MMF compound with an unsaturated fatty acid, linoleic acid (LA). Compared with the parent MMF, the prodrug (MMF-LA) showed enhanced lipophilicity which enabled its incorporation into amphiphilic copolymer materials to form stable NPs. For this purpose, PEG 5k -PL A 8k and DSPE-PEG 2k were employed to establish the MMF-LA@PEG-PLA and MMF-LA@DSPE-PEG nanosystems, respectively. We

| Characterization of NPs
The morphology of MMF-LA NPs was characterized by transmission electron microscopy (TEM, Olympus, Japan). The size distribution and zeta potential of NPs were evaluated using dynamic light scattering (DLS, Malvner, UK).

| Colony formation assay
See online supplementary Methods S1.

| In vitro cytotoxicity assay
Cytotoxicity assay was performed using CCK-8 (MCE, USA). Cells were seeded in 96-well plates with a density of 2500-3000 cells/well. After an overnight incubation, cells were treated with free MMF or MMF-LA NPs in different concentrations. After 48 hours, CCK-8 was carried out and the absorbance of each well at 450 nm was determined by spectrophotometer (Bio-Rad, USA). IC50 was calculated according to the following method: Y = 100/(1 + [X/IC50] P ). Y represents cell viability whereas X represents drug concentration, and the P value was .05.

| Cell cycle analysis
Cells were seeded in 6-well plates with a density of 1.5 × 10 4 cells/ well. After an overnight incubation, cells were treated with free MMF (2 µg/mL) or MMF-LA NPs (at 2 µg/mL MMF-equivalent dose) for 48 hours. Cells were harvested, washed with PBS and then fixed in 70% cold ethanol. One night later, cells were washed and incubated with propidium iodide (PI) for 30 minutes at room temperature. The relative proportion of cells in each cell cycle phase was analysed using flow cytometry.

| Western blot analysis
NIH-3T3 Cells were treated with free MMF (2 µg/mL) or MMF-LA NPs (at 2 µg/mL MMF-equivalent dose) for 48 hours. Cells were then collected and lysed using RIPA buffer (Thermo, USA). See online supplementary Methods S1.

| Immunohistochemistry (IHC)
Tumours were harvested and fixed in 4% formaldehyde for one week prior the paraffin embedding. Immunohistochemistry was carried out as described previously. 27 Primary antibodies used in this study were listed in the online supplementary Methods S1.

| Immunofluorescence (IF)
NIH-3T3 cells and LX2 cells were plated in confocal petri dishes.

| Boyden chamber migration assay
See online supplementary Methods S1. To confirm the tumour-facilitating effect of CAF on HCC progression, a highly fibrotic HCC model (Model I) was established by co-inoculating of Hep1-6 HCC cells with NIH-3T3 fibroblasts at a ratio of 2:1 into the right flank of nude mice. Before inoculation, NIH-3T3 fibroblasts were activated by incubating with TGF-β1 for 96 hours at a dose of 10 ng/mL. 28 The anti-CAF capacity of MMF-LA NPs was then explored base on this model.

| Animal experiments
To confirm the anti-CAF capacity of MMF-LA NPs, another highly HCC model (Model II) was constructed by co-inoculating LM3 HCC cells with LX2 cells, an activated human hepatic stellate cell line, at a ratio of 2:1 into the right flank of nude mice. Moreover, PDX models were established by using tumour samples collected from a patient who was diagnosed as HCC with liver cirrhosis.

| CAF-specific drug delivery
To examine the CAF targeting capability of MMF-LA NPs, we specifically labelled MMF-LA NPs by co-assembling a Near-infrared (NIR) dye, DiR, with MMF-LA and DSPE-PEG copolymer to generate MMF-LA@DSPE-PEG-DiR nanosystem. The highly fibrotic HCC model (Model I) and non-highly fibrotic HCC model (without NIH-3T3 cells) were constructed. When the tumour length reached 10 mm, free DiR and MMF-LA@DSPE-PEG-DiR were injected respectively. After 24 hours, all mice were anesthetized and the fluorescence signal was detected using Clairvivo OPT (SHIMADZU Corporation, Japan).

| Statistical analysis
Data were analysed using Prism 6 software and SPSS17.0 software. Shapiro-Wilk test was used to determine the normality of variables.
For normally distributed variables, Student's t test and one-way ANOVA test followed with Tukey's HSD were applied for the comparison between two groups and among multiple groups, respectively. For variables non-normally distributed, the comparison was performed using Mann-Whitney U test. Tumour volume (V) was calculated as following: V = (L × W 2 )/2, L represents length whereas W represents width. Chi-square test was performed to determine the associations between CAF density and clinicopathological parameters. Hepatocellular carcinoma recurrence-free survival (RFS) was analysed by Kaplan-Meier analysis combined with log-rank test. P value < .05 was considered statistically significant (*P < .05, **P < .01, ***P < .001, ****P < .0001).

| Preparation and Characterization of MMF-LA NPs
To overcome the unfavourable hydrophobicity of MMF, we designed and synthesized a biodegradable prodrug of MMF by covalently conjugating LA to MMF via esterification. Compared with MMF, encapsulation of MMF-LA within DSPE-PEG 2k and PEG 5k -PLA 8k yielded stable NPs because of the physicochemical similarity between the prodrug and NP core. Schematic illustration of the construction of MMF-LA@DSPE-PEG nanosystem is shown in Figure 1A. The morphology and particle sizes of MMF-LA@DSPE-PEG and MMF-LA@ PEG-PLA were characterized using TEM and DLS. As presented in Figure 1B Figure 1C). In addition, the zeta potentials of MMF-LA@DSPE-PEG and MMF-LA@PEG-PLA were slightly negative and positive, respectively ( Figure 1D). and Huh7 cells ( Figure 2E). Together, these results demonstrated that both MMF-LA NPs showed comparable capacity in inducing cytotoxicity and preventing cell proliferation with free MMF against HCC cell lines, suggesting that no obvious reduction of toxicity was observed after constructing the MMF into nanoformulations.

| MMF-LA@DSPE-PEG inhibited HCC growth in vivo
As a commonly used immunosuppressive agent, MMF is known to effectively inhibit the proliferation of lymphocytes, which may af-  Figure 3D). We further examined the proliferation activity of tumours by examining the expression status of PCNA using IHC. As shown in Figure 3E

| In vitro anti-fibrotic assays
Considering the prominent role of fibrosis in HCC tumorigenesis, we explored the anti-fibrotic activity of MMF-LA NPs using NIH-3T3 fibroblasts. As shown in Figure 4A

| CAF promote HCC growth and increase HCC recurrence risk after liver transplantation
To confirm the HCC-promoting efficacy of CAF, we constructed a highly fibrotic HCC model by subcutaneously injecting Hep1-6 HCC cells with NIH-3T3 fibroblasts into Balb/c nude mice.
To acquire similar function of CAF, NIH-3T3 fibroblasts were pre-activated by incubating with TGF-β1 for 96 hours and were proved to express alpha-smooth muscle actin (α-SMA), a widely accepted marker of activated fibroblasts and CAF before injection ( Figure 5A). As expected, HCC cells injected with activated fibroblasts exhibited greater tumorigenesis potential compared with the HCC cells injected alone ( Figure 5B,C). Immunohistochemistry analysis further confirmed the increased expression levels of α-SMA, fibroblast activated protein (FAP) and collagen IV within tumours ( Figure 5D,E). We then explored the relationship between CAF and post-transplant HCC recurrence. Clinical data of 68 HCC patients who received LT at our hospital were retrospectively collected and the CAF density in HCC samples obtained from these patients was determined using IHC for α-SMA. According to the proportion of α-SMA-positive cells accumulated in tumours, the CAF density was scored as 1, 2 or 3, and the representative images are shown in Figure 5F. Kaplan-Meier analysis was performed and the result showed that patients with moderate CAF density (score 1 + 2) had significantly superior RFS rates than those abundant with CAF (score 3) ( Figure 5G). The 1-and 3-year RFS rates were 90.2% and 82.4%, respectively, in score 1 + 2 group, and the corresponding values were 76.3% and 53.6%, respectively, in score 3 group. Moreover, the associations between CAF density  Table S2, high CAF density was associated with significantly increased proportion of microvascular invasion (13 vs 3, P = .009). And patients with low CAF density tended to have better differentiated tumours, though the difference was not statistically significant (4 vs 0, P = .089).

| MF-LA@DSPE-PEG inhibited HCC growth by suppressing CAF
We next investigated the anti-HCC and anti-CAF activity of nano-MMF based on the established highly-fibrotic HCC model (Model I). As shown in Figure 6A Figure 7C). Schematic illustration of the working mechanism of MMF-LA@DSPE-PEG is shown in Figure 7D.

| D ISCUSS I ON
Mycophenolate mofetil can be rapidly hydrolysed into MPA by esterase in circulation. 29 The generated MPA selectively inhibits IMPDH which is absolutely required in the de novo pathway of purine generation, inducing multiple cell effects. 30 Among these effects, the antitumour property of MPA has aroused the interest of scientists since the late 1960s. 31 However, previous studies have reported that MMF only marginally inhibited tumour progression in vivo due to its poor drug availability. 18 In this study, we established a combinatorial strategy of facilely constructing MMF with LA, one of the polyunsaturated fatty acids (PUFAs), to obtain a biodegradable prodrug MMF-LA, and sequentially co-assembled with amphiphilic polymer materials PEG 5k -PLA 8k and DSPE-PEG 2000 , respectively.
'PUFAylation' endowed the MMF-based nanotherapeutics improved and reducing immune surveillance. 16 For instance, CAF-secreted chemokines like CCL7 and CXCL16 strongly induced the activation of TGFβ pathway in HCC, which significantly accelerated tumour metastasis. 34 In addition, it was demonstrated that CAF promoted the tumour-initiating cell plasticity of HCC through the activation of HGF/c-Met cascade thus enhanced HCC tolerance to chemotherapeutic agents. 35 To confirm the high HCC-promoting efficacy of CAF, the present study established a highly-fibrotic HCC model by co-inoculating of Hep1-6 HCC cells with NIH-3T3 fibroblasts at a ratio of 2:1 into nude mice. Before co-inoculation, we noticed that NIH-3T3 fibroblasts did not normally express α-SMA, a widely F I G U R E 5 CAF significantly enhanced HCC growth in vivo. A, Expression levels of α-SMA determined by Immunofluorescence. NIH-3T3 fibroblasts were co-cultured with TGF-β1 or cultured alone for 96 h. B, Tumour images of different groups. Hep1-6 cells were injected with activated fibroblasts or injected alone into the nude mice, (n = 6). C, Tumour growth curves of different groups, ***P < .001. D, Expression levels of α-SMA, FAP and collagen IV determined by Immunohistochemistry. The scale bars: 50, 100 or 200 µm. E, Quantitative analysis of panel D (Image J software), data are shown as the mean ± SD, (n = 3), ***P < .001. F, Representative images showing low α-SMA expression (CAF density = 1), median α-SMA expression (CAF density = 2) and high α-SMA expression (CAF density = 3) in HCC samples obtained from HCC patients underwent liver transplantation. G, Kaplan-Meier analysis of patients with moderate CAF density (group 1 + 2) and high CAF density (group 3) accepted marker of activated fibroblast and CAF, indicating that NIH-3T3 fibroblasts remained inactive status. To better mimic the CAF function, we co-cultured NIH-3T3 fibroblasts with TGF-β1 for 96 hours before injection as previous studies had reported that TGF-β1 was a dominant profibrotic cytokine in the activation of fibroblasts and the progression of fibrosis. 36 The expression status of α-SMA was then verified by IF. As expected, NIH-3T3 fibroblasts started to express α-SMA after stimulation of TGF-β1, underlying these cells were successfully activated. Based on the highly fibrotic HCC model, we confirmed that CAF dramatically enhanced HCC growth as HCC cells injected with activated fibroblasts showed greater tumorigenesis potential compared with the HCC cells injected alone.
Given the high HCC-promoting efficacy of CAF, we suggested that CAF might influence the LT outcome by increasing HCC recurrence risk. To verify this hypothesis, clinical data of 68 HCC patients who had undergone LT at our hospital were retrospectively collected and analysed. As expected, the results revealed that patients with low CAF density had significantly better RFS rates. Moreover, high CAF density was associated with significantly increased proportion of microvascular invasion, which was consistent with the previous studies that CAF greatly promoted HCC migration and invasion. 16,37 Our results suggested that CAF might increase HCC recurrence risk after LT by facilitating the incidence of microvascular invasion, an independent risk factor for recurrence. 17 Therefore, depleting CAF might be a rational strategy to suppress HCC growth and metastasis, and decrease post-LT HCC recurrence risk.
Interestingly, accumulating evidence suggests that MMF possesses remarkable anti-fibrotic property. For instance, co-culture human mesangial cells with MPA not only inhibited cell proliferation but also decreased the fibronectin and Collagen I deposition induced by profibrotic cytokines. 38 Using fibroblasts isolated from rejected rat cardiac allografts, Johnsson et al found that MPA might be used to delay allograft fibrosis as it almost totally inhibited fibroblasts proliferation. 39 Given these, we explored the anti-fibrotic efficacy of nano-MMF using NIH-3T3 fibroblasts and the established highly fibrotic HCC models. The results revealed that all forms of MMF dramatically inhibited fibroblasts proliferation and tubulin expression in vitro. Moreover, MMF-LA@DSPE-PEG significantly suppressed tumour growth and dramatically reduced CAF density in tumours when compared with free MMF and MMF-LA@PEG-PLA.
In addition, MMF-LA@DSPE-PEG significantly inhibited the process of angiogenesis which was proved to promote tumour metastasis. 40 We next investigated the CAF-targeting capacity of MMF-LA@ DSPE-PEG to further confirm its anti-fibrotic efficacy. As expected, MMF-LA@DSPE-PEG majorly accumulated in the α-SMA positive area, underlining that most of MMF-LA@DSPE-PEG NPs were uptaken by CAF. Given the pivotal role of microenvironment in HCC development, it is a strong rationale to modulate the crosstalk between HCC cells and stroma cells. Interestingly, the established MMF-LA@ DSPE-PEG NPs not only inhibited HCC cells directly but also modulated HCC-promoting microenvironment by suppressing CAF and decreasing vascular density. In addition, MMF is widely used as a therapeutic agent in various of immune-disordered diseases. 42 And it is well known that HCC is closely associated with immune disregulation especially the chronic inflammation, suggesting the possible application of MMF in treating HCC. In this study, we also demonstrated that high CAF density was strongly associated with increased proportions of microvascular invasion and patients with high CAF density had significantly inferior RFS rates. Collectively, we believe that MMF-LA@DSPE-PEG could achieve great anti-HCC efficacy by effectively depleting CAF and also be utilized as an ideal strategy to prevent HCC recurrence after LT.
In conclusion, we described a simple approach for constructing a systemically injectable MMF-based nanoplatform by covalently conjugating the MMF with LA followed by co-assembling with amphiphilic copolymer. We demonstrated that the optimized nanotherapeutic MMF-LA@DSPE-PEG NPs exhibited enhanced cytotoxicity against a panel of human HCC cell lines and achieved improved anti-HCC efficacy in vivo by increasing tumour accumulation, killing HCC cells, suppressing CAF and inhibiting tumour angiogenesis, as F I G U R E 7 CAF-targeting capacity of MMF-LA@DSPE-PEG. A, Accumulation of Free-DIR and MMF-LA@DSPE-PEG-DIR within tumours. B, Quantitative analysis of fluorescence intensity, n = 3, ***P < .001. C, Colocalization of MMF-LA@DSPE-PEG-DIR and CAF. Green: α-SMA; red: MMF-LA@DSPE-PEG-DIR, blue: nuclei. The scale bars: 50 µm. D, Schematic illustration of the working mechanism of MMF-LA@DSPE-PEG. Cancer-associated fibroblast (CAF) greatly enhance tumour growth. The established MMF-LA@DSPE-PEG nanoparticles effectively target CAF and then are internalized by these cells. As a consequence, CAF are dramatically suppressed and CAF-related tumour growth is inhibited compared to free MMF. We envision that the MMF-LA@DSPE-PEG NPs could find a practical application in treating HCC and preventing post-transplant HCC recurrence.

ACK N OWLED G EM ENTS
We sincerely appreciated the financial supports by the Grants from the National Natural Science Foundation of China (No. 31671019) and National S&T Major Project (No. 2017ZX10203205). We also sincerely thank the support from the Research Unit Project of Chinese Academy of Medical Sciences (No. 2019-I2M-5-030) and Zhejiang Chinese Medical Project (No. 2018ZB072). We also thank Dr EDOO