Introduction Cancer metastasis is the most devastating aspect of cancer, since the 90% of cancer mortality is associated with cancer metas-tasis, and therefore represents one of the main challenges for biomedical research [1]. Although some great advantages in treatments related to primary tumour have been reached, the mechanism involved in the metastatic process has not been com-pletely understood yet. Metastasis is formed after a sequence of events called the metastasis cascade, which is divided in several and interre-lated steps: local invasion, intravasation, survival in the circulation, extravasation and colonization [2]. Before the coloniza-tion of the secondary organ, cancer cells have to create a favourable environment for their growth, secreting factors from the primary tumour able to modify the struc-ture of the secondary organ. This process is called “pre-metastatic niche” formation. After this step, the invasion of the second-ary tissue occurs [1]. In accordance with the Paget’s “seed and soil hypothesis”, breast cancer has specific targets for the formation of metastasis, es-pecially bone metastasis, which occurs in 70% of patients affected by breast cancer metastases [3, 4]. Breast cancer metastases typically cause osteolytic lesions, which are characterized by the development of a vicious cycle in-volving osteoclasts and their growth fac-tors (Figure 1). In particular breast cancer cells secrete parathyroid hormone related peptide, which stimulates osteoblasts to produce RANKL (receptor activator of nuclear factor-κB ligand). This in turn ac-tivates osteoclasts, increasing their num-ber and so bone resorption, and conse-quently the release of growth factors from bone matrix, which stimulate the growth of tumour cells [1].Cancer research on bone metastases is fo-cused on identifying the main molecules that cause the unbalance between bone formation and resorption, in order to find an effective cure for bone metastases. A promising target for cancer therapy is the mammalian target of rapamycin (mTOR) pathway, which is an important regulator of cell growth and survival. An abnormal signaling of mTOR has been as-sociated with pathological states, like bone cancer metastases. The mTOR inhi-bition can control bone metastases growth by blocking the abnormal production of osteoclasts and promoting the production of OPG, a competitor of RANKL [5, 6]. Previous in vitro studies also reported that adenosine represents a promising mole-cule for the treatment of cancer metasta-ses. Indeed, adenosine is one of the main factors that contribute to the protection of skeletal muscle from the formation of can-cer metastases. In the work by Jeon et al. [7] is reported that the presence of adeno-sine in a bone mimicking microenviron-ment produced a lower extravasation rate of tumour cells in the microenvironment. This suggests that adenosine could be used within in vitro bone models as a target for breast cancer cells inhibiting their action, as described for the skeletal muscle. Significant efforts were made in order to develop in vivo and in vitro models to in-vestigate the mechanisms involved in bone cancer metastases. These systems in-clude a variety of bi-dimensional (2D) cell cultures and animal models, although there is no ideal model able to fully mimic all the biological events characterizing the formation of bone metastases. Even if 2D in vitro models provided remarkable out-comes in the study of tumour growth, they do not take into account the three-dimen-sional (3D) interactions among the several types of cells present in the bone microen-vironment, such as osteoblasts, osteo-clasts, bone tissue resident macrophages and endothelial cells [8]. In addition, 2D in vitro models do not allow to recapitulate the complex cell distribution of real tu-mours, thus limiting their use for drug test-ing. Indeed, cancer cells in 3D microenvi-ronment have a drug resistance more sim-ilar to the one found in patients than can-cer cells cultured in 2D models. Moreover, more complex systems like in vivo animal models do not faithfully reproduce human biological processes due to species-spe-cific differences. The use of human cells is fundamental to avoid species-related bi-ases and obtain more physiological re-sults. An ideal compromise between these two systems is represented by 3D in vitro models, that couple the easiness handling of 2D models and offer a more reliable hu-man mimicking microenvironment, com-pared to animal studies [8]. In the lasts decades, several promising 3D bone models were created to more closely mimic the metastatic microenvironment, nevertheless no 3D in vitro model still able to completely reproduce the complexity of bone metastases. Aim of the work The aim of this work is to develop 3D hu-man bone in vitro models able to repro-duce the early breast cancer metastasis in bone, in order to study the different bone cell-cell interactions present in this phase of the metastatic cascade and analyse the effect of specific anti-metastatic drugs. To reach this goal, we based our 3D in vitro models on one previously developed by our group [9]. We firstly characterized the cells present in the models, then we optimized the seed-ing density of the different types of cells. Subsequently, we developed a Healthy Bone Model and a Metastatic Bone Model, including osteoblasts, osteoclasts, endothelial cells and cancer cells. We tested their reliability comparing them with a Healthy Muscle Model and a Meta-static Muscle Model, including endothe-lial cells muscle cells, muscle fibroblasts and cancer cells. We considered the bone tropism of breast cancer and compared the metastatic invasion in the Metastatic Mus-cle Model and in the Metastatic Bone Model. After, we included macrophages,another key component of the bone micro-environment, to complete our bone mod-els. We investigated the behaviour of cancer cells in the Metastatic Bone Models, in or-der to evaluate their interaction with each cell type and eventually the presence of osteomimicry, namely the ability of can-cer cells to mimic a peculiar behaviour of bone cells, like the expression of specific osteoblast or osteoclast markers [10]. Lastly we performed two drug screening tests using rapamycin or adenosine within the Metastatic Bone Model enriched with macrophages, and one using PSB-10, an inhibitor of the A3 adenosine receptor, on the Metastatic Muscle Model. We per-formed these drug screening tests in order to evaluate if the obtained results are sim-ilar to the ones found in literature and if the developed models could be used in the future to test novel anticancer therapies [6, 7]. Material and methods Cell constructs were embedded within poly-methylmethacrylate (PMMA) cages formed by an arch-shaped structure with two open windows to allow an easy hydro-gel filling and recovery. PMMA 3D masks were bonded to glass coverslips to close the structure (Figure 2).Cells were suspended in culture medium and mixed with fibrin gel, then the suspen-sion was injected in 3D masks and left pol-ymerizing in humid chambers. Cell con-structs were cultured for 7 days. The im-ages taken at the confocal microscope were analysed through the image pro-cessing software ImageJ to investigate the interactions among cells, in particular cell spatial disposition in the culture space, cell co-localization with different cell types and cell morphology. The presence of osteomimicry by cancer cells was evaluated performing a bone re-sorption assay on dentine slices, and ana-lysing stereo-microscopical images. In particular, the number of resorption pits on each slice was counted. The outcomes of the drug tests were quan-tified through calculating morphology and proliferation of cancer cells, which allow to statistically quantify the effects of drugs on the constructs. We estimated proliferation and localiza-tion of cells in terms of area fraction and superimposition of z-projection of the confocal images. The morphology of can-cer cells was analysed in terms of changes of major/minor axis index, while the re-sorption capacity of cancer cells was cal-culated counting the number of cells in a defined number of region of interests (ROI). Results and Discussion We developed four different 3D in vitro models that allow the study of organ-spe-cific human breast cancer early metastasis in a bone microenvironment. The models are Healthy Bone Model, Metastatic Bone Model, Healthy Bone Model + Macro-phages and Metastatic Bone Model + Macrophages, made by a suspension of human bone cells and breast cancer cells in a fibrin gels which were embedded in 3D masks (Figure 3). We started choosing which were the most important cells in the physiology of the or-gan we wanted to recreate and we opti-mized the seeding density. We evaluated the identity of osteoblasts through the im-munofluorescent staining of collagen I, osteocalcin and osteopontin, which are characteristic osteoblast markers. We in-vestigated the correct differentiation of os-teoclasts verifying RANK and TRAP (tar-trate resistant acid phosphatase) staining. Based on the study by Bersini et al. [9], we evaluated the formation of a complex vas-cular network using endothelial cell seed-ing density of 3 McellsmL⁄ and an endo-thelial cell/osteoblast ratio of 2:1. Finally, we optimized cancer cell, osteoclast and macrophages seeding density, which was set to 0.15 McellsmL⁄.Both the Healthy Bone Model and the Metastatic Bone Model were compared with the Healthy Muscle Model and the Metastatic Muscle Model. As known from the literature, skeletal muscle is not a tar-get for breast cancer metastasis and so muscle tissues are less affected by the presence of cancer cells than bone [11]. The outcome of the comparison between the Metastatic Bone Model and the Meta-static Muscle Model confirmed the lower proliferation of cancer cells in the Meta-static Muscle Model and indirectly proved the reliability of the Bone Models (Figure 4).We analysed the spatial disposition of cells in all bone models and discovered that osteoblasts and cancer cells tend to dispose close to vessels, while osteoclasts spread in a homogeneous arrangement both in Healthy Bone Model and in Meta-static Bone Model. The behaviour of can-cer cells can be explained by their relation with endothelial cells in the metastatic cascade and by the affinity of breast can-cer cells with bone marrow endothelium [1]. It is important to highlight also the differ-ent morphology of the vascular network in Healthy Bone Models and in Metastatic Bone Models (Figure 5). Comparing the normal and the metastatic condition, it is clear that the vascular network in the Met-astatic Model had a slower development and vessels were less interconnected than the ones in the Healthy Models. Instead, it is clear that the enrichment of the models with macrophages led to a homogeneous and interconnected vascular network com-paring the first two Bone Models with Healthy Bone Model + Macrophages and in Metastatic Bone Model + Macrophages. Indeed, it is well known that macrophagesare able to promote angiogenesis in vitro [12].We focused also on the interaction be-tween cancer cells and osteoclasts. Cancer cells co-localized with osteoclasts in the Metastatic Bone Model. In order to inves-tigate this interaction, we decided to make an immunostaining for TRAP in the 3D co-culture and resulted that cancer cells expressed this specific osteoclast marker in the Metastatic Bone Model. We thought that this behaviour could be induced by the presence of free RANKL secreted by oste-oblasts in the culture environment, which could induce an osteoclast-like behaviour in cancer cells, known as osteomimicry[10]. This theory is supported by the out-come of the bone resorption assay that re-vealed that cancer cells treated with RANKL were able to create more resorp-tion pits than the untreated cancer cells (Figure 6).The results of the rapamycin and adeno-sine tests on Metastatic Bone Model + Macrophages are shown in Figure 7. Ra-pamycin acts both on the ability of endo-thelial cells to create a vascular network and reducing the proliferation of cancer cells, when administered at high concen-trations (20nM and 100nM). On the other hand, if it is administered at the third day of culture with a 2nM dose, it allows en-dothelial cells to develop microvessels and also reduces the proliferation of cancercells. Vessels appeared more homogene-ous and cancer cells proliferation seemed slowed down, even if in in a lower velocity than at higher concentration, but this al-lowed to maintain a higher stability of the microenvironment. The drug test that we performed on Metastatic Bone Model + Macrophages for adenosine had a positive outcome, indeed the presence of adeno-sine in the bone microenvironment im-paired the proliferation of cancer cells keeping the vascular network similar to the one of the control. Even the test with the antagonist of the A3 adenosine recep-tor (PSB-10) on Metastatic Muscle Model gave positive results. PSB-10 inhibits the ability of muscle cells to overcome the in-vasiveness of cancer cell, if it is given at a concentration of 1μM. The positive out-comes of these two drug screening tests plus the additional test of PBS-10 on Met-astatic Muscle Model allow the validation of Metastatic Bone Model + Macrophages as platform for further screening for other anticancer therapies.Considering previous models, at least two types of bone cells were used in co-culture with cancer cells, so they did not allow to consider the relations happening between all the types of cells involved in the phys-iology of bone [8]. For this reason, we want to highlight that our models allowed to investigate some relevant aspects in the biological mechanisms of the early metas-tasis microenvironment, like cell-cell in-teractions and osteomimicry of cancer cells. This has been accomplished includ-ing in our models four types of human bone cells, namely osteoblasts, osteo-clasts, endothelial cells and macrophages, and analysing cells in a controlled 3D in vitro model closely mimicking the early bone metastasis.Conclusions We have presented reliable and reproduc-ible 3D bone models able to mimic the early breast cancer metastatic microenvi-ronment. The bone models are valuable to study biological processes occurring in this tissue, like the specific interactions of osteoclasts and cancer cells with osteo-genic growth factors using rapid quantifi-cations We demonstrated that these models could be employed to perform drug screening tests. The reliability and versatility of the models and the employment of only hu-man cells offer the possibility to apply these systems for the identification of more effective anticancer therapies through high-throughput screenings with patient-derived cells. These models could be used to study dif-ferent pathologies and their versatility could allow the engineering of multi-or-gan systems, as proved by the develop-ments of both bone and muscle in vitro models, using the same experimental con-ditions. Moreover, the proposed models offer higher efficiency and versatility and lower costs compared to the traditional in vitro models and in vivo models.
Introduzione La formazione di metastasi rappresenta l’aspetto più devastante delle patologie tu-morali, dato che il 90% della mortalità as-sociata al cancro è dovuta alla loro pre-senza [1], di conseguenza rappresenta una delle sfide più importanti per la ricerca biomedica. Nonostante siano stati fatti nu-merosi progressi nella cura dei tumori pri-mari, non sono ancora stati compresi tutti i meccanismi biologici alla base del pro-cesso metastatico. La metastasi si forma in seguito a una se-quenza di eventi chiamata “cascata meta-statica”, questa si divide in una sequenza di step correlati tra di loro: invasione lo-cale, ingresso nel circolo ematico, soprav-vivenza all’interno della circolazione san-guigna, extravasazione e colonizzazione [2]. Prima di poter colonizzare il tessuto secondario, le cellule tumorali devono riu-scire a creare un ambiente favorevole per la loro crescita, secernendo fattori dal tu-more primario che siano in grado di modi-ficare la struttura dell’organo secondario, questo processo viene chiamato forma-zione della “nicchia pre-metastatica”. Dopo questo step può avvenire la coloniz-zazione dell’organo secondario [1]. In accordo con la teoria del “seed and soil” postulata da Paget, il tumore alla mam-mella ha dei target specifici per la forma-zione di metastasi, in particolare il 70% dei pazienti affetti da tumore alla mam-mella soffrono di metastasi all’osso [3, 4]. Solitamente le metastasi ossee derivate da tumore alla mammella portano alla forma-zione di lesioni di tipo osteolitico, caratte-rizzate dallo sviluppo di un circolo vizioso che coinvolge gli osteoclasti e i fattori di crescita ad essi associati (Figura 1). In breve, le cellule tumorali secernono PTHrP (parathyroid hormone-related pep-tide), che stimola gli osteoblasti a produrre RANKL (receptor activator of nuclear fac-tor-κB ligand), che a sua volta attiva gli osteoclasti, aumentando il loro numero e ilriassorbimento osseo. Di conseguenza i fattori, rilasciati dalla matrice ossea, indu-cono la crescita delle cellule tumorali [1]. La ricerca sui tumori è focalizzata sull’identificazione delle molecole più im-portanti coinvolte nelle metastasi ossee che portano allo squilibrio tra deposizione e riassorbimento dell’osso, in modo da tro-vare una cura efficace per questa patolo-gia. Un target promettente per la terapia tumo-rale è il pathway dell’mTOR (mammalian target of rapamycin), un importante rego-latore della crescita e sopravvivenza cellu-lare. L’anormale produzione di mTOR è stata associata a stati patologici, come la metastasi all’osso. L’inibizione dell’mTOR può quindi controllare la cre-scita metastatica e bloccare l’elevata pro-duzione di osteoclasti, promuovendo la produzione di OPG, un competitore di RANKL [5, 6]. Precedenti studi su modelli in vitro hanno proposto l’adenosina anch’essa come ini-bitore delle metastasi tumorali, infatti l’adenosina è uno dei maggiori fattori che contribuiscono a proteggere il muscolo scheletrico dalla formazione di metastasi. Nello studio di Jeon et al. [7] è stato ripor-tato che la presenza di adenosina in un mo-dello di osso vascolarizzato ha portato a una diminuzione della portata di extrava-sazione delle cellule tumorali. Questo sug-gerisce che l’adenosina può essere utiliz-zata come inibitore della proliferazione delle cellule tumorali nei modelli in vitro di osso.È stata posta molta attenzione allo studio dei meccanismi coinvolti nella formazione delle metastasi ossee, tra cui lo sviluppo di varie culture cellulari bidimensionali (2D) e modelli animali in vivo, tuttavia non esi-ste ancora un modello in grado di mimare completamente tutti gli eventi biologici coinvolti nella formazione di metastasi all’osso. Anche se con i tradizionali mo-delli 2D in vitro si è riusciti a ottenere pro-gressi importanti, questi non tengono conto delle interazioni tridimensionali (3D) tra i diversi tipi di cellule presenti nell’osso, come osteoblasti, osteoclasti, macrofagi residenti nel tessuto osseo e cel-lule endoteliali [8]. Inoltre, le limitazioni spaziali di questi modelli riducono il loro utilizzo nei test farmacologici, infatti nei microambienti 3D le cellule tumorali hanno una resistenza ai farmaci più simile a quella trovata nei pazienti rispetto ai mo-delli 2D. Inoltre, modelli più complessi, come i modelli animali in vivo, hanno delle limitazioni in quanto non riescono a riprodurre fedelmente processi biologici propri dei tessuti umani a causa delle dif-ferenze specie-specifiche. Per questo mo-tivo è fondamentale l’utilizzo di cellule umane per evitare l’alterazione dei risul-tati dovuti all’utilizzo di cellule prove-nienti da altri organismi e per ottenere ri-sultati più affidabili. I modelli 3D sono in grado di fornire un giusto compromesso tra questi due sistemi, combinando la faci-lità di analisi dei sistemi tradizionali 2D in vitro e offrendo la possibilità di riprodurre in maniera più affidabile i processi biolo-gici del microambiente 3D in vivo. Negli ultimi anni sono stati sviluppati pro-mettenti modelli di osso 3D, in grado di rappresentare più attentamente l’ambiente metastatico, tuttavia nessuno di questi mo-delli può mimare completamente la com-plessità dei processi biologici coinvolti nella formazione delle metastasi ossee. Obiettivo Lo scopo di questa tesi è lo sviluppo di modelli 3D di osso in grado di riprodurre l’ambiente iniziale della formazione delle metastasi ossee, in modo da studiare le di-verse interazioni tra le cellule dell’osso coinvolte in questa fase della cascata me-tastatica e di analizzare l’effetto di terapie anti-metastatiche. Per raggiungere questo obiettivo abbiamo basato i nostri modelli su un modello di osso vascolarizzato sviluppato preceden-temente dal nostro gruppo di ricerca [9]. Per prima cosa abbiamo caratterizzato le cellule presenti nei modelli, poi abbiamo ottimizzato la densità di semina dei tipi cellulari. In seguito abbiamo sviluppato unModello di Osso Sano ed un Modello di Osso Metastatico, introducendo osteobla-sti, osteoclasti, cellule endoteliali e cellule tumorali. L’affidabilità di questi modelli è stata testata sfruttando il tropismo osseo delle cellule da tumore alla mammella e quindi confrontando i modelli di osso con un Modello di Muscolo Sano ed un Mo-dello di Muscolo Metastatico, parago-nando l’invasività delle cellule tumorali nei due microambienti. I modelli di mu-scolo sono formati da cellule endoteliali, cellule muscolari, fibroblasti muscolari e cellule tumorali. In seguito i modelli di osso sono stati completati andando ad in-serire i macrofagi in quanto componenti chiave per il controllo dell’omeostasi di questo tessuto. Abbiamo studiato il comportamento delle cellule tumorali nei Modelli di Osso Me-tastatico in modo da valutare la loro inte-razione con gli altri tipi cellulari presenti nei modelli ed eventualmente la presenza di “osteomimicry”, ovvero la capacità delle cellule tumorali di copiare dei com-portamenti propri delle cellule dell’osso, ad esempio l’espressione di marker speci-fici di osteoblasti e osteoclasti [10]. Infine abbiamo eseguito due test per lo screening di due farmaci, ovvero rapami-cina e adenosina, sul Modello di Osso Me-tastatico arricchito con i macrofagi. Inol-tre abbiamo effettuato un altro test utiliz-zando il PSB-10, un inibitore del recettore A3 dell’adenosina, nel Modello di Mu-scolo Metastatico in modo da poter con-frontare il suo esito con i risultati trovati in letteratura e poter confermare così l’affi-dabilità dei nostri modelli [7]. Questi test farmacologici sono stati effettuati per ve-rificare la possibilità di utilizzare in futuro i nostri modelli per testare nuove terapie anti-tumorali. Materiali e metodi I costrutti cellulari sono stati incorporati in mascherine di poli-metilmetacrilato (PMMA), formate da una struttura a forma d’arco con due finestre aperte per permet-tere l’inserimento degli idrogeli e il loro recupero. Le mascherine 3D sono state unite a dei vetrini per poter chiudere la struttura (Figura 2).Le cellule sono state sospese nel mezzo di coltura e miscelate con il gel di fibrina, dopo di che la sospensione è stata iniettata nelle mascherine 3D e lasciata a polime-rizzare nelle camere umide. I costrutti cel-lulari sono stati mantenuti in coltura per 7 giorni. Le immagini prese al microscopio confocale sono state analizzate con l’aiuto di un software per il trattamento delle im-magini, ImageJ, per studiare le interazioni tra i diversi tipi cellulari, in particolare la disposizione spaziale delle cellule, la loro co-localizzazione e la loro morfologia. La presenza di osteomimicry da parte delle cellule tumorali è stata valutata ese-guendo un assay di riassorbimento osseo su dischi di dentina e analizzando le im-magini prese allo stereo microscopio. In particolare, è stato contato il numero dei punti di riassorbimento su ogni campione. I risultati dei test farmacologici sono stati calcolati andando a quantificare la morfo-logia e la proliferazione delle cellule tu-morali, che ha permesso poi di valutare statisticamente l’effetto dei farmaci sui co-strutti cellulari. Abbiamo valutato la proliferazione e loca-lizzazione delle cellule in termini di fra-zione di area e la super-imposizione delle proiezioni lungo l’asse z delle immagini prese al microscopio confocale. La morfo-logia delle cellule tumorali è stata analiz-zata calcolando l’indice del rapporto tra asse maggiore e minore, mentre la capa-cità di riassorbimento osseo contando il numero dei buchi di riassorbimento osseo delle cellule tumorali in definite regioni di interesse (ROI). Risultati e Discussione Abbiamo sviluppato quattro diversi mo-delli 3D in vitro che hanno permesso di mimare l’organo-specificità della fase ini-ziale della formazione di metastasi da tu-more alla mammella in microambienti os-sei. I modelli presentati sono il Modello di Osso Sano, Modello di Osso Metastatico, Modello di Osso Sano + Macrofagi, Mo-dello di Osso Metastatico + Macrofagi,composti da una sospensione di cellule os-see umane e di cellule da tumore alla mammella in un gel di fibrina inserito in mascherine 3D (Figura 3). Inizialmente abbiamo scelto di inserire i tipi di cellule più importanti per la fisiolo-gia dell’organo che volevamo ricreare e poi abbiamo ottimizzato le densità di se-mina. Abbiamo caratterizzato gli osteobla-sti con delle colorazioni immunofluore-scenti di collagene di tipo I, osteocalcina e osteopontina, che sono maker caratteristici degli osteoblasti. Abbiamo studiato il cor-retto differenziamento degli osteoclasti con delle colorazioni di RANK e TRAP (tartrate resistant acid phosphatase). Ba-sandoci sullo studio di Bersini et al. [9], abbiamo verificato la formazione di una rete vascolare complessa utilizzando una densità di semina di 3 Mcellsml⁄ per le cellule endoteliali e un rapporto tra cellule endoteliali/osteoblasti di 2:1. Infine ab-biamo ottimizzato la densità di semina di cellule tumorali, osteoclasti e macrofagi a 0,15 Mcellsml⁄.Il Modello di Osso Sano e il Modello di Osso Metastatico sono stati confrontati con i Modelli di Muscolo. Come è noto in letteratura, il muscolo scheletrico non è un target per le metastasi da tumore alla mammella e quindi i tessuti muscolari non sono colpiti come l’osso dalla presenza delle cellule tumorali [11]. Il risultato del confronto tra il Modello di Osso Metasta-tico ed il Modello di Muscolo Metastatico ha confermato una minore proliferazione delle cellule tumorali nel Modello di Mu-scolo e indirettamente ha provato l’affida-bilità dei Modelli di Osso (Figura 4).Il Modello di Osso Sano e il Modello di Osso Metastatico sono stati confrontati con i Modelli di Muscolo. Come è noto in letteratura, il muscolo scheletrico non è un target per le metastasi da tumore alla mammella e quindi i tessuti muscolari non sono colpiti come l’osso dalla presenza delle cellule tumorali [11]. Il risultato del confronto tra il Modello di Osso Metasta-tico ed il Modello di Muscolo Metastatico ha confermato una minore proliferazione delle cellule tumorali nel Modello di Mu-scolo e indirettamente ha provato l’affida-bilità dei Modelli di Osso (Figura 4).Ci siamo concentrati anche sulle intera-zioni delle cellule tumorali con gli osteo-clasti. Le cellule tumorali co-localizzano con gli osteoclasti nel Modello di Osso Metastatico. Al fine di studiare questa in-terazione abbiamo deciso di effettuare una colorazione immunologica per la TRAP nella co-cultura 3D e ne è risultato che le cellule tumorali sono in grado di espri-mere questo specifico marker di attiva-zione per gli osteoclasti nel Modello di Osso Metastatico. Abbiamo pensato che questo potesse essere dovuto alla presenza di RANKL libero secreto dagli osteoblasti nello spazio di cultura e questo potrebbe indurre le cellule tumorali ad assumere un comportamento simile a quello degli osteoclasti, noto come “osteomimicry” [10]. Questa teoria è sostenuta dall’esperi-mento di riassorbimento osseo in cui le cellule tumorali trattate con RANKL sono state in grado di formare più buchi di rias-sorbimento delle cellule tumorali non trat-tate (Figura 6).I risultati dei test con la rapamicina e l’adenosina sul Modello di Osso Metasta-tico + Macrofagi sono visibili in Figura 7. La rapamicina, somministrata ad elevate concentrazioni (20nM and 100nM), ha sia agito sull’abilità selle cellule endoteliali di formare una rete vascolare sia ha ridotto la proliferazione delle cellule tumorali. D’al-tra parte, se somministrata al terzo giorno di coltura a una dose di 2Nm, permette alle cellule endoteliali di formare microvasi e riduce la proliferazione delle cellule tumo-rali, anche se in minore velocità rispetto a concentrazioni più alte. Inoltre permette al microambiente di mantenere una mag-giore stabilità. Il test per l’adenosina effet-tuato sul Modello di Osso Metastatico +Macrofagi ha dato un risultato positivo, in-fatti la presenza di adenosina nel mi-croambiente osseo ha permesso lo svi-luppo di una rete vascolare simile a quella del controllo. Anche il test effettuato uti-lizzando l’antagonista del recettore A3 dell’adenosina (PSB-10) sul Modello di Muscolo Metastatico ha dato un risultato positivo. Il PSB-10 inibisce la capacità delle cellule muscolari di controllare l’in-vasività delle cellule tumorali se sommini-strato a una concentrazione di 1μM. Il ri-sultato positivo dei precedenti due test far-macologici più il test aggiuntivo del PSB-10 sul Modello di Muscolo Metastatico ha permesso la validazione del Modello di Osso Metastatico + Macrofagi per il suo utilizzo come piattaforma per futuri scree-ning di nuove terapie anti-metastatiche.Nei modelli presenti in letteratura sono stati utilizzati al massimo due tipi di cel-lule ossee in co-coltura con le cellule tu-morali, questo non permette di considerare le diverse relazioni tra tutti i tipi di cellule coinvolte nei processi biologici del tessuto osseo [8]. Per questa ragione ci teniamo ad evidenziare che i nostri modelli hanno per-messo di studiare degli aspetti rilevanti dei meccanismi biologici coinvolti nell’am-biente iniziale della formazione delle me-tastasi ossee, come le interazioni cellula-cellula e l’osteomimicry delle cellule tu-morali. Questi risultati sono stati ottenuti introducendo nei nostri modelli quattro tipi di cellule umane, ovvero osteoblasti, osteoclasti, cellule endoteliali e macro-fagi, e analizzando le co-colture in un mo-dello 3D in vitro altamente controllato che è in grado di simulare l’ambiente pre-me-tastatico nell’osso. Conclusioni Abbiamo presentato dei modelli 3D di osso affidabili e facilmente riproducibili in grado di simulare l’ambiente pre-metasta-tico di tumori alla mammella. Questi mo-delli di osso sono applicabili per lo studio dei processi biologici propri di questo tes-suto, come le interazioni tra osteoclasti e cellule tumorali con i fattori di crescita osteogenici.Abbiamo dimostrato che questi modelli possono essere utilizzati per effettuare test farmacologici. Inoltre l’affidabilità e la versatilità di questi modelli e l’utilizzo di sole cellule umane offre la possibilità di poter applicare questi sistemi per identifi-care terapie più efficaci contro il cancro utilizzando high-throughput screening test con cellule da paziente. I modelli proposti offrono maggiore effi-cienza e versatilità e minor costo in con-fronto ai modelli tradizionali in vitro e in vivo e potrebbero essere utilizzati in fu-turo per studiare diverse patologie. In futuro, la loro versatilità potrebbe per-mettere l’ingegnerizzazione di sistemi multiorgano, come dimostrato dallo svi-luppo di modelli di osso e di muscolo usando le stesse condizioni sperimentali.
Engineering 3D functional human vascularized bone models for the study of early breast cancer metastases
COLOMBO, MARIA VITTORIA
2016/2017
Abstract
Introduction Cancer metastasis is the most devastating aspect of cancer, since the 90% of cancer mortality is associated with cancer metas-tasis, and therefore represents one of the main challenges for biomedical research [1]. Although some great advantages in treatments related to primary tumour have been reached, the mechanism involved in the metastatic process has not been com-pletely understood yet. Metastasis is formed after a sequence of events called the metastasis cascade, which is divided in several and interre-lated steps: local invasion, intravasation, survival in the circulation, extravasation and colonization [2]. Before the coloniza-tion of the secondary organ, cancer cells have to create a favourable environment for their growth, secreting factors from the primary tumour able to modify the struc-ture of the secondary organ. This process is called “pre-metastatic niche” formation. After this step, the invasion of the second-ary tissue occurs [1]. In accordance with the Paget’s “seed and soil hypothesis”, breast cancer has specific targets for the formation of metastasis, es-pecially bone metastasis, which occurs in 70% of patients affected by breast cancer metastases [3, 4]. Breast cancer metastases typically cause osteolytic lesions, which are characterized by the development of a vicious cycle in-volving osteoclasts and their growth fac-tors (Figure 1). In particular breast cancer cells secrete parathyroid hormone related peptide, which stimulates osteoblasts to produce RANKL (receptor activator of nuclear factor-κB ligand). This in turn ac-tivates osteoclasts, increasing their num-ber and so bone resorption, and conse-quently the release of growth factors from bone matrix, which stimulate the growth of tumour cells [1].Cancer research on bone metastases is fo-cused on identifying the main molecules that cause the unbalance between bone formation and resorption, in order to find an effective cure for bone metastases. A promising target for cancer therapy is the mammalian target of rapamycin (mTOR) pathway, which is an important regulator of cell growth and survival. An abnormal signaling of mTOR has been as-sociated with pathological states, like bone cancer metastases. The mTOR inhi-bition can control bone metastases growth by blocking the abnormal production of osteoclasts and promoting the production of OPG, a competitor of RANKL [5, 6]. Previous in vitro studies also reported that adenosine represents a promising mole-cule for the treatment of cancer metasta-ses. Indeed, adenosine is one of the main factors that contribute to the protection of skeletal muscle from the formation of can-cer metastases. In the work by Jeon et al. [7] is reported that the presence of adeno-sine in a bone mimicking microenviron-ment produced a lower extravasation rate of tumour cells in the microenvironment. This suggests that adenosine could be used within in vitro bone models as a target for breast cancer cells inhibiting their action, as described for the skeletal muscle. Significant efforts were made in order to develop in vivo and in vitro models to in-vestigate the mechanisms involved in bone cancer metastases. These systems in-clude a variety of bi-dimensional (2D) cell cultures and animal models, although there is no ideal model able to fully mimic all the biological events characterizing the formation of bone metastases. Even if 2D in vitro models provided remarkable out-comes in the study of tumour growth, they do not take into account the three-dimen-sional (3D) interactions among the several types of cells present in the bone microen-vironment, such as osteoblasts, osteo-clasts, bone tissue resident macrophages and endothelial cells [8]. In addition, 2D in vitro models do not allow to recapitulate the complex cell distribution of real tu-mours, thus limiting their use for drug test-ing. Indeed, cancer cells in 3D microenvi-ronment have a drug resistance more sim-ilar to the one found in patients than can-cer cells cultured in 2D models. Moreover, more complex systems like in vivo animal models do not faithfully reproduce human biological processes due to species-spe-cific differences. The use of human cells is fundamental to avoid species-related bi-ases and obtain more physiological re-sults. An ideal compromise between these two systems is represented by 3D in vitro models, that couple the easiness handling of 2D models and offer a more reliable hu-man mimicking microenvironment, com-pared to animal studies [8]. In the lasts decades, several promising 3D bone models were created to more closely mimic the metastatic microenvironment, nevertheless no 3D in vitro model still able to completely reproduce the complexity of bone metastases. Aim of the work The aim of this work is to develop 3D hu-man bone in vitro models able to repro-duce the early breast cancer metastasis in bone, in order to study the different bone cell-cell interactions present in this phase of the metastatic cascade and analyse the effect of specific anti-metastatic drugs. To reach this goal, we based our 3D in vitro models on one previously developed by our group [9]. We firstly characterized the cells present in the models, then we optimized the seed-ing density of the different types of cells. Subsequently, we developed a Healthy Bone Model and a Metastatic Bone Model, including osteoblasts, osteoclasts, endothelial cells and cancer cells. We tested their reliability comparing them with a Healthy Muscle Model and a Meta-static Muscle Model, including endothe-lial cells muscle cells, muscle fibroblasts and cancer cells. We considered the bone tropism of breast cancer and compared the metastatic invasion in the Metastatic Mus-cle Model and in the Metastatic Bone Model. After, we included macrophages,another key component of the bone micro-environment, to complete our bone mod-els. We investigated the behaviour of cancer cells in the Metastatic Bone Models, in or-der to evaluate their interaction with each cell type and eventually the presence of osteomimicry, namely the ability of can-cer cells to mimic a peculiar behaviour of bone cells, like the expression of specific osteoblast or osteoclast markers [10]. Lastly we performed two drug screening tests using rapamycin or adenosine within the Metastatic Bone Model enriched with macrophages, and one using PSB-10, an inhibitor of the A3 adenosine receptor, on the Metastatic Muscle Model. We per-formed these drug screening tests in order to evaluate if the obtained results are sim-ilar to the ones found in literature and if the developed models could be used in the future to test novel anticancer therapies [6, 7]. Material and methods Cell constructs were embedded within poly-methylmethacrylate (PMMA) cages formed by an arch-shaped structure with two open windows to allow an easy hydro-gel filling and recovery. PMMA 3D masks were bonded to glass coverslips to close the structure (Figure 2).Cells were suspended in culture medium and mixed with fibrin gel, then the suspen-sion was injected in 3D masks and left pol-ymerizing in humid chambers. Cell con-structs were cultured for 7 days. The im-ages taken at the confocal microscope were analysed through the image pro-cessing software ImageJ to investigate the interactions among cells, in particular cell spatial disposition in the culture space, cell co-localization with different cell types and cell morphology. The presence of osteomimicry by cancer cells was evaluated performing a bone re-sorption assay on dentine slices, and ana-lysing stereo-microscopical images. In particular, the number of resorption pits on each slice was counted. The outcomes of the drug tests were quan-tified through calculating morphology and proliferation of cancer cells, which allow to statistically quantify the effects of drugs on the constructs. We estimated proliferation and localiza-tion of cells in terms of area fraction and superimposition of z-projection of the confocal images. The morphology of can-cer cells was analysed in terms of changes of major/minor axis index, while the re-sorption capacity of cancer cells was cal-culated counting the number of cells in a defined number of region of interests (ROI). Results and Discussion We developed four different 3D in vitro models that allow the study of organ-spe-cific human breast cancer early metastasis in a bone microenvironment. The models are Healthy Bone Model, Metastatic Bone Model, Healthy Bone Model + Macro-phages and Metastatic Bone Model + Macrophages, made by a suspension of human bone cells and breast cancer cells in a fibrin gels which were embedded in 3D masks (Figure 3). We started choosing which were the most important cells in the physiology of the or-gan we wanted to recreate and we opti-mized the seeding density. We evaluated the identity of osteoblasts through the im-munofluorescent staining of collagen I, osteocalcin and osteopontin, which are characteristic osteoblast markers. We in-vestigated the correct differentiation of os-teoclasts verifying RANK and TRAP (tar-trate resistant acid phosphatase) staining. Based on the study by Bersini et al. [9], we evaluated the formation of a complex vas-cular network using endothelial cell seed-ing density of 3 McellsmL⁄ and an endo-thelial cell/osteoblast ratio of 2:1. Finally, we optimized cancer cell, osteoclast and macrophages seeding density, which was set to 0.15 McellsmL⁄.Both the Healthy Bone Model and the Metastatic Bone Model were compared with the Healthy Muscle Model and the Metastatic Muscle Model. As known from the literature, skeletal muscle is not a tar-get for breast cancer metastasis and so muscle tissues are less affected by the presence of cancer cells than bone [11]. The outcome of the comparison between the Metastatic Bone Model and the Meta-static Muscle Model confirmed the lower proliferation of cancer cells in the Meta-static Muscle Model and indirectly proved the reliability of the Bone Models (Figure 4).We analysed the spatial disposition of cells in all bone models and discovered that osteoblasts and cancer cells tend to dispose close to vessels, while osteoclasts spread in a homogeneous arrangement both in Healthy Bone Model and in Meta-static Bone Model. The behaviour of can-cer cells can be explained by their relation with endothelial cells in the metastatic cascade and by the affinity of breast can-cer cells with bone marrow endothelium [1]. It is important to highlight also the differ-ent morphology of the vascular network in Healthy Bone Models and in Metastatic Bone Models (Figure 5). Comparing the normal and the metastatic condition, it is clear that the vascular network in the Met-astatic Model had a slower development and vessels were less interconnected than the ones in the Healthy Models. Instead, it is clear that the enrichment of the models with macrophages led to a homogeneous and interconnected vascular network com-paring the first two Bone Models with Healthy Bone Model + Macrophages and in Metastatic Bone Model + Macrophages. Indeed, it is well known that macrophagesare able to promote angiogenesis in vitro [12].We focused also on the interaction be-tween cancer cells and osteoclasts. Cancer cells co-localized with osteoclasts in the Metastatic Bone Model. In order to inves-tigate this interaction, we decided to make an immunostaining for TRAP in the 3D co-culture and resulted that cancer cells expressed this specific osteoclast marker in the Metastatic Bone Model. We thought that this behaviour could be induced by the presence of free RANKL secreted by oste-oblasts in the culture environment, which could induce an osteoclast-like behaviour in cancer cells, known as osteomimicry[10]. This theory is supported by the out-come of the bone resorption assay that re-vealed that cancer cells treated with RANKL were able to create more resorp-tion pits than the untreated cancer cells (Figure 6).The results of the rapamycin and adeno-sine tests on Metastatic Bone Model + Macrophages are shown in Figure 7. Ra-pamycin acts both on the ability of endo-thelial cells to create a vascular network and reducing the proliferation of cancer cells, when administered at high concen-trations (20nM and 100nM). On the other hand, if it is administered at the third day of culture with a 2nM dose, it allows en-dothelial cells to develop microvessels and also reduces the proliferation of cancercells. Vessels appeared more homogene-ous and cancer cells proliferation seemed slowed down, even if in in a lower velocity than at higher concentration, but this al-lowed to maintain a higher stability of the microenvironment. The drug test that we performed on Metastatic Bone Model + Macrophages for adenosine had a positive outcome, indeed the presence of adeno-sine in the bone microenvironment im-paired the proliferation of cancer cells keeping the vascular network similar to the one of the control. Even the test with the antagonist of the A3 adenosine recep-tor (PSB-10) on Metastatic Muscle Model gave positive results. PSB-10 inhibits the ability of muscle cells to overcome the in-vasiveness of cancer cell, if it is given at a concentration of 1μM. The positive out-comes of these two drug screening tests plus the additional test of PBS-10 on Met-astatic Muscle Model allow the validation of Metastatic Bone Model + Macrophages as platform for further screening for other anticancer therapies.Considering previous models, at least two types of bone cells were used in co-culture with cancer cells, so they did not allow to consider the relations happening between all the types of cells involved in the phys-iology of bone [8]. For this reason, we want to highlight that our models allowed to investigate some relevant aspects in the biological mechanisms of the early metas-tasis microenvironment, like cell-cell in-teractions and osteomimicry of cancer cells. This has been accomplished includ-ing in our models four types of human bone cells, namely osteoblasts, osteo-clasts, endothelial cells and macrophages, and analysing cells in a controlled 3D in vitro model closely mimicking the early bone metastasis.Conclusions We have presented reliable and reproduc-ible 3D bone models able to mimic the early breast cancer metastatic microenvi-ronment. The bone models are valuable to study biological processes occurring in this tissue, like the specific interactions of osteoclasts and cancer cells with osteo-genic growth factors using rapid quantifi-cations We demonstrated that these models could be employed to perform drug screening tests. The reliability and versatility of the models and the employment of only hu-man cells offer the possibility to apply these systems for the identification of more effective anticancer therapies through high-throughput screenings with patient-derived cells. These models could be used to study dif-ferent pathologies and their versatility could allow the engineering of multi-or-gan systems, as proved by the develop-ments of both bone and muscle in vitro models, using the same experimental con-ditions. Moreover, the proposed models offer higher efficiency and versatility and lower costs compared to the traditional in vitro models and in vivo models.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Magistrale_Maria_Vittoria_Colombo.pdf
solo utenti autorizzati dal 12/09/2020
Descrizione: testo della tesi
Dimensione
3.14 MB
Formato
Adobe PDF
|
3.14 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/135915