Modifying the tumour microenvironment and reverting tumour cells: New strategies for treating malignant tumours

Abstract The tumour microenvironment (TME) plays a pivotal role in tumour fate determination. The TME acts together with the genetic material of tumour cells to determine their initiation, metastasis and drug resistance. Stromal cells in the TME promote the growth and metastasis of tumour cells by secreting soluble molecules or exosomes. The abnormal microenvironment reduces immune surveillance and tumour killing. The TME causes low anti‐tumour drug penetration and reactivity and high drug resistance. Tumour angiogenesis and microenvironmental hypoxia limit the drug concentration within the TME and enhance the stemness of tumour cells. Therefore, modifying the TME to effectively attack tumour cells could represent a comprehensive and effective anti‐tumour strategy. Normal cells, such as stem cells and immune cells, can penetrate and disrupt the abnormal TME. Reconstruction of the TME with healthy cells is an exciting new direction for tumour treatment. We will elaborate on the mechanism of the TME to support tumours and the current cell therapies for targeting tumours and the TME—such as immune cell therapies, haematopoietic stem cell (HSC) transplantation therapies, mesenchymal stem cell (MSC) transfer and embryonic stem cell‐based microenvironment therapies—to provide novel ideas for producing breakthroughs in tumour therapy strategies.


| INTRODUC TI ON
Tumour incidence and mortality are increasing yearly, with particularly rising trends in younger populations. 1 In 2018, 18.1 million new tumour cases were reported worldwide, and 9.6 million people died from tumours, making them one of the greatest threats to human health. 2 The generation and development of tumours were previously believed to depend on only tumour suppressor or oncogene mutations, the basis of the "tumour-centric" view. 3 Therapies derived from this theory, whether drugs, surgeries or radiation therapies, are all based on killing tumour cells with inevitable secondary damage and increasing treatment resistance. Researchers have found that the tumour microenvironment (TME) plays a pivotal role in the generation, progression and metastasis of tumours. A century ago, Stephen Paget found that breast cancer metastasis displayed organ (tissue) preference, which related to the cell environment of the targeted organ (tissue). He boldly assumed that tumour progression is controlled by the interaction of tumour cells and the external environment, and first proposed the concept of the TME. 4 Various components of the TME constitute an intricate network that precisely regulates tumour fate and the interactions of tumour cells with other components. This enables tumour cells to steadily proliferate, resist apoptosis, escape from immune elimination, maintain stemness and metastasize to distant sites. The TME theory superseded the theory that the fate of tumour cells is determined only by their genetic material and provided a new perspective for comprehensively understanding tumour metastasis and drug resistance mechanisms.
Traditional anti-tumour chemoradiotherapy is strongly cytotoxic because it denatures nucleic acids and proteins in tumour cells; however, this also results in damage to normal cells and causes serious adverse reactions, even secondary tumour formation. [5][6][7][8] Tumour cells escape apoptosis by constantly generating new gene mutations that mediate tumour drug resistance. To solve the problem of the poor specificity of chemoradiotherapy, targeted therapies and immune therapies have been developed. 9 Although immune therapies, such as anti-programmed death 1(PD-1)/PD-L1 treatment, show considerable efficacy in several tumours, they still have individual specificity. Meanwhile, the high incidence of severe autoimmune adverse reactions after immune therapy poses a new threat to patients' lives. [10][11][12][13] With the gradual deepening of understanding of TME, targeting TME compounds to undermine protecting hotbed of tumours have become an effective means of cancer treatment. Large amount of pre-clinic and clinic study proved the quietly success in targeting angiogenesis, extracellular matrix (ECM) and cells components within TME. 14 In recent years, cell therapies are fast rising and have been proven to have powerful functions and ensured safety. Compared with the single role of drug, cells may act on TME from multi-angle and through many ways at one time due to its better plasticity.
It is manifested that cell therapies can inhibit or reverse tumours for which there is currently no effective therapy. We suggest that utilizing a therapeutic cell's own microenvironment to regulate and modify the TME, thereby destroying the tumour nests that tumour cells depend on for survival, constitutes a new direction for tumour treatment. We will elaborate on the current therapies, especially cell therapies, for targeting tumours and the TME-such as immune cell therapies, stem cell replacement therapies (mainly used for bone marrow-derived tumours), mesenchymal stem cell (MSC) transfer and embryonic stem cell-based microenvironment therapies-to provide novel ideas for the optimization of tumour therapy strategies.

| COMP OS ITI ON AND FUN C TI ON OF THE TME
The components surrounding the tumour cells constitute the functional TME in which tumour cells initiate and grow and from which they invade and metastasize. The TME is a sophisticated network that includes various tumour-associated cells, such as cancer-associated fibroblasts (CAFs), cancer stem cells (CSCs), MSCs, tumourassociated immune and inflammatory cells, pluripotent stromal cells, cancer-associated adipocytes, pericytes and endothelial cells (ECs).
The tumour-associated cells secrete tumour-associated exosomes (TEXs) and soluble molecules, including cytokines, kinases, protein, transcription factors, growth factors, hormones and free radicals.
They regulate and respond to each other, construct the scaffold of the tumour ECM, and are nourished by tumour angiogenesis, manifesting as a special niche for innutrition, acidity, hypoxia and ischaemia. 15,16 The TME has a degree of tumour-derived organ and tissue specificity, reviews of Schumacher et al 17 also summarized that human tumours vary substantially in the composition of their microenvironment, and this is likely to regulate cancer cell morphology and influence the ability of the individual T-cell immunotherapy.

| The ECM forms the 3-dimensional tumour protective nest
The tumour ECM, comprising collagen, fibronectin, laminin, vitronectin and tenascin, is secreted by cells in the TME. Via autocrine and paracrine signalling, tumour cells and microenvironment cells, especially CAFs, transform the ECM to an advantageous phenotype for tumour progression. 18 Collagen and fibronectin in the ECM provide physical support for tumour cells, and proteoglycans act as binding factors for growth factors and cytokines. 16 Thus, the ECM not only provides a 3-dimensional structure for tumour growth, but, more importantly, it provides precise biomechanical and biochemical regulation for tumour cells. 19 The alteration of the orderly isotropic arrangement of the collagen, called matrix remodelling, results in tumour-associated collagen signatures (TACS). 20 Different TACS stages correspond to different degrees of tumour progression. Over activated and transformed CAFs induce type I collagen aberrations, transforming type I TACS to type III, which manifests as fibre thickening, cross-linking, deposition, distribution perpendicular to the tumour boundary, developing tensile stresses and formation of a tumour invasion track. 21 This growing fibrosis and stiffness of ECM results in tissue "desmoplasia," which stimulate tumorigenesis, metastasis and immune surveillance. 22,23 Furthermore, the ECM stiffness is obstructive for effective permeation of drugs to the intratumoral area and promotes transformation of cancer cells to CSCs, and these two aspects can induce resistance to anti-cancer therapies. 24 Whatcott et al 25 demonstrated that high desmoplasia is related to low survival rates of pancreatic ductal adenocarcinoma patient, indicating the reprogramming in the tumour stroma is most effective for targeting desmoplastic tumours, such as pancreas. 26

| Angiogenesis causes hypoxia and low drug penetration while promoting tumour immune escape and metastasis
The infinite proliferation and malignant metastasis of solid tumours are inseparable from angiogenesis in the TME. Tumour angiogenesis is the formation of abnormal vascular tissue with an immature structure, abnormal function and high permeability; it contributes to abnormal haemodynamics, poor nutrient, and oxygen supply, and TME hypoxia and hypoperfusion. Hypoxia is a common condition in many solid tumours. The hypoxic TME harbours cells with a burst of

| CAFs
Normal fibroblasts and adipose-derived stem cells in the TME are transformed into CAFs that overexpress α-smooth muscle actin (α-SMA) and fibroblast activation protein. 34 CAFs occupies approximately 80% of pancreatic and breast tumour volume, and from cellular view, CAFs take about 40%-50% of the whole population of cells within the tumour microenvironment. 26 The secretory capacity of CAFs is much greater than that of normal stromal cells, and CAFs are the main source of tumour cell ECM. 35 CAFs and tumour cells reciprocally regulate each other through a feedback mechanism. CAFs represent metabolic symbiosis with cancer cells, which is important therapeutically. 36,37 TGFβ1, PDGF, FGFβ and connective tissue growth factor secreted by tumour cells promote the transformation of normal fibroblasts into CAFs. TGFβ, FGF and hepatocyte growth factor (HGF) secreted by CAFs act on tumour cells, inducing epithelial-mesenchymal transition (EMT) and a pre-metastatic state. [38][39][40] The matrix proteins secreted by CAFs differ in composition and content from those produced by normal tissues, forming the TME-specific ECM stiffness. However, CAFs also play an important role in matrix degradation by secreting matrix enzymes such as MMPs, which engages cancer cells to reshape their morphology and breakthrough basement membrane. Both deposition and degradation of ECM induced by CAFs are cancer promoting and CAFs could be essential target for cancer therapies. 41 In addition, CAFs interplay with other stromal cells to regulate TME ecology. CAFs' secretomes reprogramme cancer cells, immune cells and ECs to facilitate cancer cell invasion and metastasis. Increased glycolysis of CAFs creates acidity in TME to increase generation of myeloid-derived suppressor cells (MDSCs), inhibit maturation of tumour-associated dendritic cells (TADCs) and impair the activity of NK and effector T cells. 15 As the most population of stomal cells, CAFs accumulation in the TME is often correlated with poor prognosis in many tumours. 42

| Immune cells
Both innate and adaptive immune cells in the TME show low tumourrecognition and killing ability, which is a key factor contributing to tumorigenesis and progression ( Figure 1).
The TME affects innate immune cells in various ways. T cells, thus causing CTLs dysfunction and exhaustion. Therefore, the T cells lose their normal functions and even mediate some of the major tumour protective effects in the TME. 54 Recalling immunity cycle in tumour through priming and activation of effector T cells is the key to exert durable and fruitful effects against tumour. 55

| Tumour-associated ECs
Tumour-associated ECs include blood endothelial cells (BECs) and lymphatic endothelial cells (LECs). VEGFA in the microenvironment promotes the formation of new blood vessels from BECs, and VEGFC and VEGFD activate LECs to form new lymphatic vessels in the TME. 56  and B7-H4. 57 The leaky vasculature also causes precipitation and accumulation of waste products, and promotes TME acidity that influences tumour metabolism. 58 The acidified TME drives tumour local invasion. 59

| Pericytes
Pericytes differentiate from mesenchymal precursors and are recruited to tumours by platelet-derived growth factor-β (PDGFβ) gradients. 60 Pericytes are heterogenous in their function and are coming F I G U R E 1 Abnormal and inactivated immune cell populations in the TME. Monocytes in the TME are prompted to differentiate into tumour-supporting M2 macrophages and MDSCs, whereas their differentiation into M1 macrophages and DCs is impaired. The presence of fewer DCs and more MDSCs results in the inhibition of effector T-cell responses through the down-regulation of the TCR and IL-2R. NK cells in the TME are inhibited through the down-regulation of their expression of NK activating receptors, such as NKG2D, NKp30, NKp46 and NKG2C. Traditional effector T cells transform into Tregs that have tumour protective effects. DC, dendritic cell; IL-2R, interleukin-2 receptor; MDSC, myeloid-derived suppressor cell; NFκB, nuclear factor kappa-B; NK, natural killer; TCR, T-cell receptor; TME, tumour microenvironment; Tregs, regulatory T cells into focus these years. Nestin+/NG2+ "Type-2" pericytes contribute to tumour angiogenesis by promoting ECs survival through their secretomes including VEGFA, ANGPT1 and ECM components. They also express neural cell adhesion molecule 1 (NCAM1) and the NG2 proteoglycan, which contribute to vascular maturation by increasing pericyte recruitment. 61 Furthermore, pericytes are involved in CSCs maintenance, tumour metastasis and immune microenvironment. Upregulation of PD-L1, CD90, PDGFRβ, CD248 and Rgs5 which inhibit CD4+ and CD8+ cytotoxic T-cell activity was reported in pericytes derived from within tumour microenvironments, which facilitates immunosuppression and eventual immune evasion of tumour cells. 62

| TEXs are TME messengers
Exosomes are secreted vesicle-like membrane structures with a diameter between 30 and 100 nm. They carry a variety of proteins, lipids and nucleic acids, and contribute to intercellular communication. 67 TEXs, as the microcosms of tumour cells, carry a large amount of tumour-derived materials with source-cell specificity; TEXs are one of the main methods of TME signal interaction. TEXs participate in almost all tumour processes, including angiogenesis, matrix remodelling, tumour metastasis and immune evasion.
TEXs regulate tumour angiogenesis by activating ECs. 68 Pancreatic cancer-derived CD106 + CD49d + TEXs are recruited, recognized and internalized by tumour-associated ECs. They induce the expression of VEGF and other angiogenic proteins, such as CXCL5, MIF, and CCR1, by ECs and promote angiogenesis in the TME. 69 In addition, many non-coding RNAs in TEXs have been shown to play a key role in angiogenesis. Non-coding RNAs (including miR-9, miR-21 and miR-210) in TEXs can promote angiogenesis in lung cancer by activating STAT3, ephrin A3 and MMP2/9. 70-73 Conigliaro et al found that exosomes secreted by CD90 + liver cancer cells are rich in long non-coding RNA (lncRNA) H19, which can up-regulate the expression of VEGF, VEGF receptor and ICAM in vascular ECs, thereby promoting tumour angiogenesis and tumour cell adhesion and migration to the site of neovascularization. This finding suggests that lncRNA H19 may also be related to haematological metastasis of tumours. 74 Lang et al 75,76 confirmed that lncRNA CCAT2 and ln-cRNA POU3F3 in glioma-derived exosomes are related to angiogenesis in vivo. Meanwhile, TEXs derived from various tumours-such as melanoma, 77 chronic myeloid leukaemia, 78 glioma, 79 and breast, 80,81 colon, 82 and ovarian cancer 83 have been shown to promote angiogenesis in the TME.
Matrix enzymes in TEXs degrade normal ECM, promote matrix remodelling and create a pre-metastasis microenvironment. 84 Manipulation of extracellular vesicles (EVs) to carry a desired cargo is a novel strategy for tumour therapy. EVs can be modified with specific receptors so as to target the cell/s of interest; this will pursue a long-term content storage, virtue by no phenotypical alteration inside the TME. This approach is impressive and covers the current limitation for application of stem cells for cancer therapy due to encountering phenotypical alterations in the TME. Another impressive feature with EVs therapy is their stability and their capacity to cross biological barriers efficiently. 89

| TME-MED IATED TUMOUR PROTEC TION AND PROMOTION MECHANIS MS
Various parts of the TME act as tumour hotbeds to promote their malignancy ( Figure 2). To aid the search for effective anti-tumour targets, we will consider the role of the TME in maintaining the tumour reserve, promoting tumour metastasis and resisting killing from 2 angles: (a) promotion of CSCs generation and (b) initiation of EMT.

| The TME protects CSCs
CSCs are a relatively quiescent tumour cell subpopulation that undergoes active DNA repair, similar to normal adult stem cells. As the "foundation" of tumours, CSCs mediate therapy resistance and metastasis of tumours; they make a tumour "the endless weed under wildfire.

| The TME promotes EMT to initiate metastasis
Metastasis is a sign of tumour deterioration and progression, and a leading cause of cancer death. Tumour metastasis is multistep process regulated by many factors, but EMT is considered the initial step.
EMT occurs when epithelial cells lose their unique polarity, adhesion ability, and surface expression of E-cadherin and β-catenin, then assume the morphology of mesenchymal cells and express mesenchymal cell markers such as α-SMA, vimentin, N-cadherin and SNAIL. 96 EMT is involved in normal physiological processes such as embryonic development and wound repair, but also plays an important role in tumour metastasis and CSCs generation.  with tumour ECM fibronectin also readily undergo EMT. 102 High expression of HIF1α in the hypoxic TME inhibits caveolin-1 expression F I G U R E 2 Composition of the TME and its effects on tumour development. The TME is a complex network of tumour-associated cells (such as CAFs, tumour-associated immune cells, and ECs), TEXs, soluble molecules (such as cytokines, growth factors and hormones) and tumour-specific ECM, which is nourished by tumour angiogenesis. The components act as a tumour nest that maintains CSCs, promotes EMT, enables escape from immune surveillance and provides resistance to therapies through specific signal pathways. CAF, cancerassociated fibroblast; CSCs, cancer stem cells; ECM, extracellular matrix; ECs, endothelial cells; EMT, epithelial-mesenchymal transition; TEX, tumour-associated exosome; TME, tumour microenvironment in tumour cells; a negative feedback mechanism up-regulates the expression of the caveolin-1-related protein epidermal growth factor receptor, thereby activating the STAT3 signalling pathway and reducing the expression of epithelium-specific markers. 103 At the same time, mesenchymal transition of tumour-associated vascular ECs reduces tight junctions, thereby weakening the barrier effects of blood vessels.
However, the current understanding of cellular phenotypical modification during cancer metastasis is acquiring a partial EMT, namely a hybrid E/M phenotype. This indicates that tumours cells can represent mesenchymal phenotype while simultaneously exhibiting epithelial potentials. This hybrid phenotype can be seen in collective invasion of tumour cells, and it accounts for the greatest capacity to pursue tumour metastasis. 104 These all indicate that targeting EMT activity and weakening the EMT-promoting link can play a key role in inhibiting tumour metastasis.

| TARG E TING THE TME TO B RE AK THROUG H THE FORTRE SS OF THE TUMOUR
Isolation of tumour cells by the TME results in low treatment efficiencies, low clinical response rates and drug resistance. 13,105 It has been suggested that highly specific therapeutic strategies that break the barrier of the TME and disrupt its strong protective effects may be safer and more effective. 10

| Targeting drugs attack the TME separately
Understanding the major events occurring in the TME that support primary tumour growth and how these events impact the modulation of the environment is of utmost relevance to assist the definition of efficient therapy strategies. A good deal of current strategies was used to target TME components.

| Cell therapy to modify the TME is a new strategy for tumour treatment
Even though targeting molecules inhibit tumour to a great extent, there is still some insurmountable difficulties. For example, agents that degrade and/or deconstruct ECM must be used carefully, since they may induce metastasis instead of avoiding tumour progression. We boldly assume that cell therapies may solve some of the problems of targeting drugs attributing to cell's natural advantages. (a) The cell itself has good plasticity, which provides the possibility of acting on multiple TME components at the same time to fully degrade TME, which can effectively prevent the compensatory regeneration and protection of other components when simply targeting a certain part; (b) the cell has homing properties. The "localization system" of the treatment cells has the tendency of tumour tissue and homing under the action of cytokines and chemokines, which greatly avoids the systemic side effects brought by non-selective drugs; (c) cell has high permeability. Due to tumour stiffness and abnormal angiogenesis, the drug's permeability to TME is poor, but cells can effectively enter TME through deformation movements, secretion of matrix-regulating enzymes, etc; (d) many cells have the ability to regulate normal tissue regeneration and repair, which can correctly repair tissue damage caused by chemotherapy drugs and avoid the adverse impact of bystander effects on normal cells (Figure 3).

| Immune cell therapy for precisely targeting tumour cells
The TME regulates the differentiation and activation of immune cells to reduce their ability to recognize and eliminate tumour cells.
Immune cell therapies that more accurately target tumour cells while as chronic lymphocytic leukaemia, lymphoma and multiple myeloma (NCT02135406); however, the response rates of these diseases are lower than that of B-ALL. 115,116 CAR-T cells targeting other molecules, such as CD20, CD22, CD30 and ROR1, have been used successfully to target different tumours (NCT02315612 and NCT00621452). 117 In addition to blood tumours, many studies have tried to target solid tumours using CAR-T cells, such as GD2 CAR-T for neuroblastoma and GPC3-targeted CAR-T cells for hepatocellular carcinoma, which have prolonged patient survival. 118,119 However, CAR-T is highly effective for malignancies of haematologic system; for solid tumours, the application of this strategy is elusive mainly because of their suppressive TME, weak TIL trafficking, 120,121 and has a high probability of side effects. In clinical trials of CD19-specific CAR-T-cell treatment, all subjects without exception showed cytokine release syndrome.
In addition, complications such as neurotoxicity, B-cell depletion and off-target effects also seriously threaten patients' lives. 122 In addition to CAR-modified T cells, CAR-NK cells have achieved a high response rate in the treatment of myelodysplastic syndrome and acute myeloid leukaemia (AML). CAR-NK cells for the treatment of B-cell lymphomas (NCT01974479 and NCT03056339) and F I G U R E 3 Cell therapy to modify the TME. Traditional radiotherapy, chemotherapy and targeted drugs all rely of tumour cell killing as their main mechanism of action, with the attendant serious adverse reactions and rapid resistance. Current cell therapies for targeting tumours and the TME-modified immune cell therapies, haematopoietic stem cell transplantation, mesenchymal stem cell transfer, and embryonic stem cell-based microenvironment therapies-provide novel ideas for exploring breakthrough of tumour therapy strategies metastatic solid tumours (NCT03415100) are in clinical trials; however, they face problems such as low CAR transfection efficiency and short in vivo survival time. 123 To resolve the issues of limited sources of NK cells and short survival time, NK cells have been generated from induced pluripotent stem cells (iPSCs) and cord blood, and their therapeutic effects on recurrent or refractory blood and solid tumours have been assessed (NCT01729091, NCT03019640, NCT02280525 and NCT03539406). However, the low induction and differentiation efficiencies of iPSCs and the relatively high risk of tumour formation still limit their application.
Exosomes secreted by CAR cells have been confirmed to carry the CAR structure. They maintain the original targeting characteristics of the parent cells in the context of lower off-target cytotoxicity, with high efficacy and safety in preclinical experiments. 124 CAR exosomes may be used to optimize or even replace existing cell-based CAR immunotherapies.

| Haematopoietic stem cell transplantation to reconstruct haematopoietic function
The origins of many tumours are associated with the dysplasia, depletion, and dysregulated proliferation and differentiation of normal stem cells. The existence of a normal stem cell microenvironment has become a prognostic indicator for tumour treatment. 125 Eliminating malignant CSCs and supplementing with normal stem cells has become a reliable treatment for many tumours, especially haematologic tumours. 126 Haematopoietic stem cell transplantation (HSCT) is to completely remove the abnormal bone marrow haematopoietic stem cell (HSC) microenvironment by means of radiotherapy and chemotherapy, and then rebuild the haematopoietic microenvironment and restore the normal haematopoietic function by transplanting normal stem cells. As the only curative treatment for malignant and non-malignant diseases of the haematopoietic system, HSCT has developed rapidly in the past 25 years. [127][128][129][130] The first step of HSCT is the pretreatment of patients with radio-chemotherapy to destroy the dysfunctional haematopoietic and immune systems and to ensure low immune responsiveness by the patients, thus laying the foundation for long-term survival of the transplanted cells. Then, HSCs-mobilized from the donor by granulocyte-colony-stimulating factor-are purified, expanded in vitro, and transplanted into the recipient, where they typically home to the bone marrow and gradually reconstitute the recipient's haematopoietic system. HSCT brings graft-versus-leukaemia (GVL) effect in recipients, which helps eradicate tumours and is considered to be multifactorial. 131 The mechanism of GVL, though poorly understood, might be similar to graft-versus-host disease (GVHD) phenomenon that is mediated by donor's immune compounds. The principal cytotoxicity in GVL is mediated by donor T cells and NK cells with ancillary roles played by dendritic cells, B cells and minor histocompatibility antigens. 132 Although HSCT can fundamentally eradicate tumours, several side effects, including acute infection, graft rejection, chronic GVHD and secondary tumour development after transplantation, still impair the long-term survival of patients. 133,134 Approximately 1.2%-1.6% of patients have secondary tumours 5 years after HSCT, and the cumulative incidence of secondary tumours increases to 2.2%-6.1% after 10 years, and 3.8%-14.9% after 15 years. 135 To reduce the side effects and improve the survival rates of patients, HSCT programmes have continuously innovated. Peripheral blood-and umbilical cord blood-derived HSCs are gradually replacing bone marrow HSCs for transplantation, which solves the problem of insufficient cells. 136 HSCT performed under non-myelosuppressive conditions by reducing the doses of chemotherapy and radiation, or even without pretreatment, reduces the incidence of the GVHD response and improves long-term transplantation efficacy. 137 It is clear that maximizing the GVL effect while minimizing GVHD is the holy grail of transplant immunology. Given that GVHD and GVL have similar mechanisms, prophylaxis for GVHD might affect GVL intensity, which should be taken into account when formulating a transplantation strategy.
HSCT is becoming increasingly prominent in the treatment of some solid tumours, especially refractory tumours. 138 As early as 1997, researchers found that breast tumours in mice shrank after allogeneic HSCT. 139 In recent years, HSCT has been studied for the treatment of relapsed and refractory glioma, and as a consolidation therapy for the remission stage, resulting in a 1-year tumour-free survival rate of up to 90%-93% and a 3-year survival rate of 63.7%. 140,141 After receiving allogeneic HSCT, especially non-myelosuppressive transplantation, patients with metastatic renal cell carcinoma showed a relatively high response rate, likely due to the restoration of T cells and other immune cells capable of disrupting the immunosuppressive TME. 142,143

| MSC therapy
MSCs are non-haematopoietic stem cells primarily found in the bone marrow. They can differentiate into adipose tissue, bone and cartilage under the appropriate conditions. 144 In addition to the bone marrow, MSCs also exist in cord blood, peripheral blood and adipose tissue. 145 Due to their relative abundance, shared features with tumours and homing characteristics, MSCs can effectively enter the TME, thereby overcoming low drug delivery efficiency and limitations on TME penetration. 146 [160][161][162] However, there also have been reports that MSC exosomes promote tumours. 163 The best method to make use of the advantages of MSCs while avoiding their tumour-promoting activity has become the focus of MSC therapy research.
Researchers are increasingly modifying MSCs to optimize their anti-tumour function in 2 ways. First, they are enhancing the synthesis and release of endogenous and exogenous anti-tumour factors from MSCs. Second, they are strengthening their homing to the TME and prolonging their activity there to achieve more efficient TME penetration and anti-tumour effects.
In order to ensure the tumour-killing activity of MSCs and prevent them from promoting tumours, many researchers have enhanced MSC synthesis and release of IFN-γ, which inhibits tumour cell proliferation, and tumour necrosis factor-related apoptosis-inducing ligand, which promotes apoptosis. 164,165 These approaches have enhanced MSC anti-tumour effectiveness in preclinical and clinical experiments. MSCs transfected with suicide genes and carrying biologically active anti-tumour substances reach the TME via recruitment by chemotactic signals such VEGF and TGFβ1 secreted by tumour cells and CAFs. Once in the TME, they initiate the suicide programme to release the drugs with greater specificity and in greater concentration in the TME than achievable by conventional methods. [166][167][168] These findings suggest the feasibility of using modified MSCs as efficient drug carriers. MSCs transfected with only a CCL5 promoter-driven or ganciclovir-induced suicide gene can also cause a degree of tumour cytotoxicity in animal models and clinical trials of hepatocellular carcinoma, pancreatic cancer, and breast cancer. 169,170 In order to enhance the in vivo activity of MSCs, researchers have encapsulated MSCs with synthetic biodegradable ECM and implanted them in the brains of mice with glioblastoma. This approach resulted in dramatic shrinkage of the tumour and ensured the long-term biological activity of the MSCs. 171 Other studies have enhanced the homing of MSCs through gene editing. HIF1α induced by the hypoxic TME up-regulates downstream SDF-1α to promote tumour proliferation and metastasis. Overexpression of the SDF-1α receptor CXCR4 in MSCs enhances the penetration of MSCs into the hypoxic glioma microenvironment. 172 However, MSCs remain a double-edged sword. The stronger the effects of MSCs as drug carriers that penetrate the TME to kill tumours, the more serious the side effects caused by the drugs' non-selective destruction of normal cells. The current drawbacks associated with MSC therapies have prompted researchers to seek safer, more accurate anti-tumour treatment methods.

| Embryonic stem cell microenvironmental therapy
The early embryonic microenvironment has the powerful ability to repair erroneous genetic material and inhibit oncogene expression; thus, embryonic cells have innate tumour immunity characteristics.
An increasing amount of research is being devoted to applying this tumour-hostile microenvironment to tumour therapy. 173,174 The embryonic microenvironment can reverse tumour fate. Lee et al 175  to the aorta-gonad-mesonephros region (embryonic haematopoietic tissue), the yolk sac, and the peripheral blood. 177 The ability of the embryonic microenvironment to revert tumour cells gradually weakens as the embryo develops and differentiates. 178 (Table 1).
In order to apply the powerful tumour-reverting effects of the  173 Zebrafish embryo extracts can inhibit breast cancer by down-regulating the expression of translationally controlled tumour protein and promoting E-cadherin/β-catenin redistribution to reshape the cytoskeleton. 182 Mouse ESC-conditioned medium was found to inhibit breast cancer cell proliferation, promote apoptosis and inhibit malignant behaviour. 183 The mechanisms by which the ESCM reverts tumours are still not completely clear. Current studies suggest that tumour reversion may be related to the PI3K/AKT, STAT3, Notch1, and other pathways, and that exosomes derived from ESCs may play an important role in their signal transduction. 178,183,184 In addition, transmembrane communication and cell contact-mediated signal transduction may play key roles in ESCM regulation of tumour cell fate. The totipotency of ESCs enables them to regulate the TME in many ways, not only by affecting tumour cells, but also by reshaping other TME components, enhancing normal cell functions, and supporting the regeneration and repair of the body. ESC treatment raises new hope for radically eliminating the side effects caused by killing normal cells during tumour treatment.

| SUMMARY AND PROS PEC TS
The TME can mediate tumour cell immune escape, promote CSCs formation and enhance tumour metastasis ability, thereby promoting tumorigenesis and development. The distinctive features of the TME, including the abnormal haemodynamics due to neovascularization, the ECM that is difficult to penetrate, and the hypoxic state that confers resistance to oxidative damage, play important roles in the drug resistance of tumour cells. Cell therapies utilizing engineered immune cells have shown good anti-tumour effects, but poor curative effects for solid tumours. Furthermore, the main aim of this type of cell therapy is still to kill tumour cells.  optimizing the favourable characteristics of cells with engineered modifications to compensate for their shortcomings, or replacing source cells with cell-specific microvesicles.

ACK N OWLED G EM ENTS
This work was supported by The National Key R&D programme of China (2018YFC1106000).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request.