Transforming growth factor β‐mediated micromechanics modulates disease progression in primary myelofibrosis

Abstract Primary myelofibrosis (PMF) is a Ph‐negative myeloproliferative neoplasm (MPN), characterized by advanced bone marrow fibrosis and extramedullary haematopoiesis. The bone marrow fibrosis results from excessive proliferation of fibroblasts that are influenced by several cytokines in the microenvironment, of which transforming growth factor‐β (TGF‐β) is the most important. Micromechanics related to the niche has not yet been elucidated. In this study, we hypothesized that mechanical stress modulates TGF‐β signalling leading to further activation and subsequent proliferation and invasion of bone marrow fibroblasts, thus showing the important role of micromechanics in the development and progression of PMF, both in the bone marrow and in extramedullary sites. Using three PMF‐derived fibroblast cell lines and transforming growth factor‐β receptor (TGFBR) 1 and 2 knock‐down PMF‐derived fibroblasts, we showed that mechanical stress does stimulate the collagen synthesis by the fibroblasts in patients with myelofibrosis, through the TGFBR1, which however seems to be activated through alternative pathways, other than TGFBR2.


| INTRODUC TI ON
Primary myelofibrosis (PMF) is a Ph-negative myeloproliferative neoplasm (MPN), characterized by advanced bone marrow fibrosis and extramedullary haematopoiesis that is most pronounced in the liver and spleen. [1] Among MPNs, approximately 50% of patients with PMF and essential thrombocythemia (ET) are JAK2 V617F positive, whereas >90% polycythemia vera (PV) patients are positive for this mutation. [2][3][4] To date, several 'driver' mutations have been identified in the pathogenesis of PMF, involving target genes such as JAK2, CALR and MPL, whereas less commonly involved genes include ASXL1, SRSF1 and U2AF1. [5][6][7][8] The bone marrow fibrosis results from proliferation of fibroblasts that are influenced by several cytokines in the microenvironment including transforming growth factor-β (TGF β), basic fibroblast growth factor (FGF), platelet-derived growth factor (PDGF) and calmodulin. [9][10][11][12][13] A likely source of these factors is the underlying malignant clone that may acquire different phenotypic states including megakaryocyte or monocyte differentiation. [12,14,15] In support of these data, we have recently shown that the fibroblasts involved in this process do not derive from the malignant clone but seem to represent over-stimulated normal cells. [16] The main cytokine believed to be involved in generation of PMFassociated fibrosis is TGF β. [17] TGF-β acts on fibroblast physiology by increasing synthesis of collagen, type I, III, IV and V, as well as production of fibronectin, proteoglycans and tenascin. [18] Although the importance of TGF β in fibroblast proliferation was previously underlined in studies of PMF, micromechanics related to the niche has yet been elucidated, as progression in PMF occurs in a tense niche, with a relatively high level of mechanical stress. [18,19] The role of bone marrow mechanical stress via TGF-β modulation was assessed by Balooch et al, with potential clinical implications. [20] Data were confirmed by Zhao et al [21] that have shown that bone marrow-derived mesenchymal stem cells have resistance to flow shear stress through AMP-activated protein kinase signalling.
In the current study, we hypothesized that mechanical stress produced during PMF progression modulates TGF-β-dependent signalling. TGF-β is secreted by the malignant megakaryocytes, leading to further activation and subsequent proliferation and invasion of bone marrow fibroblasts, both in the bone marrow and in distant organs.
Thus, we demonstrate the important role of micromechanics in the microenvironment and the development and progression of PMF.

| Cell cultures
Primary myelofibrosis-derived fibroblast cell lines used were isolated and characterized as previously published [16] (Figure 1). All cells were grown in a humidified atmosphere at 37°C air, 95%; carbon dioxide (CO 2 ), 5%. Cell passage and cultures were carried out as previously described. [22][23][24] All cells were cultured in Dulbecco's modified Eagle's medium (DMEM), supplemented with penicillin/ streptomycin and non-essential amino acids. Hanging drops were used to isolate the cells from any external stimuli, and cells were further treated in hanging drops for 4 days until spheroids were formed and the fibroblasts were inactivated. Re-activation of the fibroblasts was carried out by digesting the spheroids with trypsin and culture on a plastic dish. Once re-activated, proliferation and collagen type I and III increased. For the hanging drop experiments, classic Petri dishes were used, making this experiment feasible and reproducible in any laboratory setting. Images of the cells were obtained by inverted phase microscopy from activated, re-activated as well as organoids in a 30-μ dish (50 mm) using an inverted Zeiss Axio Observer Z1 microscope, as previously described. [25][26][27]

| shRNA TGFBR1 and TGFBR2 knock-down
Transforming growth factor-β receptor (TGFBR) 1  in extramedullary sites. Using three PMF-derived fibroblast cell lines and transforming growth factor-β receptor (TGFBR) 1 and 2 knock-down PMF-derived fibroblasts, we showed that mechanical stress does stimulate the collagen synthesis by the fibroblasts in patients with myelofibrosis, through the TGFBR1, which however seems to be activated through alternative pathways, other than TGFBR2.

K E Y W O R D S
fibroblast activation, invasion, micromechanics, myelofibrosis, proliferation, TGF-β (9ſg) + VSVG (3ſg) + 800 µL Opti-MEM. Medium was changed the following day. Virus titre was at its highest approximately 24 hours after transfection. The medium was passed through a 0.45 µm filter unit attached to a syringe. Transduction of cells with lentivirus supernatant and selection with puromycin was carried out at Day 5.
The puromycin selection process was done using 4-4.25 µg/mL for PMF-derived fibroblasts and continued until all non-transduced control cells were dead. Later, living cells were transferred to T25 flasks and expanded before assessment of the knock-down via qPCR.

| Quantitative RT-PCR (q RT-PCR) for mRNA expression
Quantitative RP-PCR was performed to confirm the expression of the collagen type I and type III expression, using TaqMan PCR technology. [36][37][38] For the PCR, following first-strand synthesis, 10 μL The primers used for collagen III: the forward primer is 5'-GGATCAGGCCAGTGGAAATGTAAAGA-3', and reverse primer is 5'-CTTGCGTGTTCGATATTCAAAGACTGTT. Cycle passing threshold (Ct) was recorded and normalized to human GAPDH (hGAPDH) expression. Relative expression was calculated as 2Ct_FLT3-Ct_ GAPDH. PCR reactions were carried out in triplicate. All mRNA q RT-PCR values were normalized to β-actin, and the relative expression was calculated as 2Ct_target gene-Ct_GAPDH. PCR reactions were carried out in triplicate. As internal controls, we used hGAPDH F I G U R E 1 Isolation and characterization of PMF-derived fibroblasts forward, 5′-GTGGTCTCCCTGACTTTCAACAGC-3′, and hGAPDH reverse, 5′-ATGAGGTCCACCACCTGCTTGCTG-3′ (149-bp amplicon).
The membranes were blocked with TBS containing 5% skim milk and 0.1% Tween-20, then incubated with the primary antibody overnight, at 4°C. Antibodies for collagen I and collagen III were purchased from Abcam (Cambridge, UK). The membranes were incubated after washing with HRP-conjugated goat anti-rabbit IgG (Calbiochem, Gibbstown, NJ, USA) and analysed using enhanced chemiluminescence plus reagent (GE Healthcare, Buckinghamshire, UK). Group A (experimental group) mice were injected with GFPpositive PMF-derived fibroblasts in the central venous system, whereas group B (control group) were injected with normal bone marrow-derived fibroblasts. The in vivo experiments allowed us to investigate whether PMF-derived fibroblasts had the potential to invade hematopoietic organs following injection in the systemic blood flow of immunocompromised animals. One month following injection of 1 × 10 6 cells in the central venous system, all animals were sacrificed, and pathology slides were used to assess both the bone marrow and lung tissue using dark-field microscopy.

| Confocal microscopy
Confocal microscopy was performed on an Olympus FLUOVIEW FV1200 laser scanning confocal microscope (Olympus, Tokyo, Japan). Image acquisition was performed using the UPLSAPO 10X2 (0.4 NA) objective. The images were obtained using channel mode (488 nm excitation and bright field). Other settings for the image acquisition were determined using the FV10-ASW 4.2 software.
Images were processed using ImageJ.

| Transthoracic echocardiography in PMF patients
Routine transthoracic echocardiography (TTE) (Phillips Affinity 50G echograph) was used to investigate pulmonary hypertension in PMF

| Statistical analysis
The data were analysed using two statistical software packages (GraphPad Prism 5.0 and R; GraphPad software, Inc, La Jolla, CA, USA). Initial results were assessed by using the Shapiro-Wilk test to determine whether the distribution was normal. The distribution of all the obtained data was Gaussian; thus, it was analysed using a parametric test (two-way ANOVA with Tukey post-test). The differences were considered significant when P value was <0.05. show that we successfully integrated the shRNA and pGIPZ constructs in PMF-derived fibroblasts, and from a transcriptomic point, the knock-down for TGFBR2 was successful. Higher passage number can negatively impact transduction efficacy (data not shown), and the phenomenon observed for the TGFBR2 knock-down might be a compensatory mechanism or the heterodimer nature of the receptor. [39]

| Micromechanics of PMF-derived fibroblasts
We compared fibroblasts obtained from myelofibrosis patients with a healthy bone marrow stroma. A dashed line is used to illustrate a fold change of 1. Up-regulation of TGFBR1 and down-regulation of TGFBR2 can be observed in both primary lines compared with normal stroma. This can lead to the hypothesis that myelofibrosisderived fibroblasts, acting as cancer-associated fibroblasts, are highly dependent on TGFBR1, with a possible compensatory downregulation of TGFBR2 ( Figure 3A). As shown in Figure 3B, Western blotting confirmed the RT-PCR data.
The influence of mechanical stress is shown in Figure 4A (RT-PCR data), B (Western blotting data at 6 hours) and C (Western blotting data at 24 hours). Fold change was determined using the delta CT method. β2-microglobulin was used as housekeeping gene. We because fold changes vary widely between them, so we considered that a free y-axis would be a more appropriate representation.

| Invasive potential of PMF-derived fibroblasts
After proving that micromechanics plays a role in PMF progression following PMF-derived fibroblast activation and increased proliferation, we aimed at assessing the potential of these cells, acting as cancer-associated fibroblasts, to invade distant organs. Thus, we used nude immunocompromised mice for in vivo experiments.

| Clinical evaluation of pulmonary hypertension in PMF patients
After showing that PMF-derived fibroblasts have homing potential in the lung tissue, with possibly subsequent altering of the local mechanics, we assessed pulmonary hypertension in PMF patients by echocardiography ( Figure 6A-I). When assessing pulmonary hypertension in PMF patients, we used a small cohort of patients (n = 5) for a preliminary proof-of-concept study. The link between patient characteristics, data on age, disease status or comorbidities F I G U R E 3 A, RT-PCR shows that myelofibrosis-derived fibroblasts, acting as cancer-associated fibroblasts, are highly dependent on TGFBR1, with a possible compensatory down-regulation of TGFBR2. B, Western blotting shows that myelofibrosis-derived fibroblasts, acting as cancer-associated fibroblasts, are highly dependent on TGFBR1, with a possible compensatory down-regulation of TGFBR2  Figure 6E. This is measured from the standard A4C view without foreshortening. Measurement is taken at end-diastole. Ratio of >1 measured at end-diastole suggests RV dilatation (47.9/44.9).  Figure 6G presents the RV pulsed tissue Doppler S wave (S′) velocity. PW tissue Doppler S wave measurement is taken at the lateral tricuspid annulus in systole. S′ wave velocity <9.5 cm/s indicates RV systolic dysfunction. Figure 6H presents the tricuspid annular plane systolic excursion (TAPSE). The excursion of the lateral tricuspid annulus is measured by M-mode between end-diastole and peak systole for a measure of longitudinal RV systolic function. TAPSE < 1.7 cm is highly suggestive of RV systolic dysfunction. Finally, Figure 6I shows the inferior vena cava diameter (IVC) Subcostal (2D M-mode). The diameter is measured perpendicular to the IVC long axis, 1 cm from the RA junction at end expiration. The IVC diameter >21 mm with decreased inspiratory collapse (<50% with a sniff or <20% with quiet respiration) is considered abnormal.

| DISCUSS IONS
TGF-β plays a major role in the biology of primary myelofibrosis, is being secreted by the malignant megakaryocytes and is likely further affecting the bone marrow microenvironment. [12,40] TGF-β is a cytokine secreted in 3 isoforms: TGF-β1, TGF-β2 and TGF-β3, out of which TGF-β1 is the most abundant. [41,42] TGF-β1 is initially secreted in a latent form and further activated by reactive oxygen species, as well as various proteases such as plasmin, integrins and trombospondin-1, and binds to two types of receptors: type I (TGFβR1) and type II (TGFβR2). [43] Initially, TGF-β1 binds to TGFβR2; then, TGFβR2 stimulates the protein kinase activity of TGFβR1 by recruiting, binding and transphosphorylating it [44] (Figure 7). Further on, the transcription factors Smad2/3 are recruited and phosphorylated, which bind next to Smad4 and translocate in the nucleus where they interact with coactivators (as are CBP and p300), co-repressors (as are c-Ski, SnoN or TGIF) and transcription factors (as are Runx1 or E2F), thus regulating the transcription of TGF-β-responsive genes.
[44] Smad phosphorylation was linked to bone marrow physiology by Lee et al, [45] who showed that BMP signalling through both Smad and p38 mitogen-activated protein kinase (MAPK) modulates the differentiation of mesenchymal stem cells in the bone marrow microenvironment. Cell shape and cytoskeleton micromechanics are altered by Smad protein phosphorylation in the bone marrow embryology via RhoA/ROCK-mediated tension generated by BMPinduced signalling. [46] Other groups have looked at the essential interplay between biochemical and mechanical cues in cell differentiation and as micromechanics plays key roles in both the normal physiology and embryology. [47] Our hypothesis was that mechanical stress might modulate the progression of primary myelofibrosis by targeting TGF-β pathways in the bone marrow microenvironment.
Our results show an increase in collagen III, collagen I and al- We also assessed collagen I and III secretion for the TGFβR1 knock-down cells and found an important decrease in synthesis, both at the 6 and 24 hours following the exposure to micromechanical stress. Interestingly, the secretion of both collagen I and III at 24 hours was lower than the one at 6 hours, which suggests that the mechanism of synthesis was exhausted over time. In addition, the secretion of alpha-SMA was higher in TGF-β. The secretion of alpha-SMA was close to undetectable at 6 hours but increased at 24 hours.
Cancer-associated fibroblasts are known and described to promote tumour progression and invasion but are reported to lack any potential to invade surrounding tissues themselves. [51][52][53][54]  Even if further studies, both in the pre-clinical setting and in the clinical data, must validate or invalidate our statement, we hypothesize that PMF-derived fibroblasts have the potential to migrate in the lung parenchyma and change the local microenvironment, altering with the local micromechanics in order to promote disease progression and promote extramedullary haematopoiesis.
A major weakness of our study is that we used only a single-cell in vitro analysis system-the PMF-derived fibroblasts. These fibroblasts are not part of the neoplastic clone and are stimulated by malignant myeloid progenitors to proliferate and produce extracellular matrix, proven by Fialkow et al, [61] by Ciurea et al, [12] and by our group recently. [16] Thus, one might consider that the influence of mechanical stress on collagen synthesis might be linked to general fibroblast biology, rather than PMF. The same micromechanical stress might also act on malignant myeloid progenitors and their cytoplasmic protrusions that extend to the bone marrow. PMF basic biology should thus be investigated by using at least a co-culture system of fibroblasts and their interaction with hematopoietic progenitors.
The co-culture is the next step in our research, as we aimed to understand each individual cell in the bone marrow microenvironment, this manuscript being of interest especially from a basic fibroblast biology standpoint.

| CON CLUS ION
In conclusion, our data suggest that mechanical stress does indeed stimulate the collagen synthesis by the fibroblasts in patients with myelofibrosis. Further studies are nevertheless needed to elucidate the exact pathway or pathways through which it acts, as our studies suggest that it is the mechanism does not follow the TGF-β pathway completely and other downstream pathways should be investigated in the future, such as MAPK, ROCK or retinoic acid-mediated pathways.
F I G U R E 7 Molecular mechanism of TGF-beta receptor action

CO N FLI C T O F I NTE R E S T
There is no conflict of interest to be declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of the study are available from the corresponding author upon reasonable request.