The human amniotic fluid stem cell secretome effectively counteracts doxorubicin-induced cardiotoxicity

The anthracycline doxorubicin (Dox) is widely used in oncology, but it may cause a cardiomyopathy with bleak prognosis that cannot be effectively prevented. The secretome of human amniotic fluid-derived stem cells (hAFS) has previously been demonstrated to significantly reduce ischemic cardiac damage. Here it is shown that, following hypoxic preconditioning, hAFS conditioned medium (hAFS-CM) antagonizes senescence and apoptosis of cardiomyocytes and cardiac progenitor cells, two major features of Dox cardiotoxicity. Mechanistic studies with mouse neonatal ventricular cardiomyocytes (mNVCM) reveal that hAFS-CM inhibition of Dox-elicited senescence and apoptosis is associated with decreased DNA damage, nuclear translocation of NF-kB, and upregulation of the NF-kB controlled genes, Il6 and Cxcl1, promoting mNVCM survival. Furthermore, hAFS-CM induces expression of the efflux transporter, Abcb1b, and Dox extrusion from mNVCM. The PI3K/Akt signaling cascade, upstream of NF-kB, is potently activated by hAFS-CM and pre-treatment with a PI3K inhibitor abrogates NF-kB accumulation into the nucleus, modulation of Il6, Cxcl1 and Abcb1b, and prevention of Dox-initiated senescence and apoptosis in response to hAFS-CM. These results support the concept that hAFS are a valuable source of cardioprotective factors and lay the foundations for the development of a stem cell-based paracrine treatment of chemotherapy-related cardiotoxicity.

By contrast, Dox-elicited senescence and apoptosis were not reduced by pre-incubation with the conditioned medium from the human keratinocyte cell line, NCTC 2544 (hNCTC-CM Normo and hNCTC-CM Hypo ), used as an internal control (Fig. 1b,c). Compared with untreated cells, no significant change in SA β -galactosidase or cleaved caspase-3 expression was seen when incubating H9c2 cardiomyoblasts with the hAFS-CM Normo or hAFS-CM Hypo alone (see Supplementary Fig. S2).
hAFS-CM Hypo effectively protects mouse neonatal cardiomyocytes against Dox. Next, the cardioprotective potential of hAFS-CM Hypo -which had proved to be the most effective form of hAFS-CM to prevent Dox toxicity in the experiments with H9c2 cardiomyoblasts -was confirmed on primary mouse neonatal ventricular cardiomyocytes (mNVCM). Consistent with earlier studies 16 , SA β -galactosidase positive mNVCM were about 45% after Dox treatment. Pre-incubation with hAFS-CM Hypo diminished the frequency of senescent cells by 47% (Fig. 2a,b: upper panel; 44.7 ± 2.9% vs 23.6 ± 2.4% of SA β -galactosidase positive cells). Similar results were obtained by immunostaining for p16 INK4a , with the hAFS-CM Hypo decreasing the pro-senescent effect of Dox by 35% (Fig. 2b lower panel; 48.8 ± 1.8% vs 31.8 ± 1.5% of p16 INK4a positive cells). Moreover, there were twice as many cleaved caspase-3 positive cells with Dox as without treatment, and this increase in apoptosis was also reduced by the hAFS-CM Hypo by 41% (Fig. 2c,e: 41.6 ± 2.1% vs 24.7 ± 1.1% cleaved caspase-3 positive cells). Correspondingly, cell viability, as measured by MTT assay, was increased by 10.5% by pre-incubation with hAFS-CM Hypo (Fig. 2d,f). Importantly, hNCTC-CM did not prevent Dox-initiated senescence and apoptosis of mNVCM (Fig. 2b,e,f), nor were these cell responses significantly modified by hAFS-CM Hypo or hAFS-CM Normo alone (see Supplementary Fig. S2).
Immunoreactivity for phosphorylated H2AX (γ H2AX), a sensitive indicator of DNA double-strand breaks and damage 17 , was much more widespread and intense following exposure to Dox, than in untreated mNVCM (Fig. 2g). Mean fluorescence intensity for γ H2AX was increased by about 4-fold by Dox compared to untreated cells, but limited to a 2.4 fold change, corresponding to a 37% reduction, by pre-incubation with the hAFS-CM (Fig. 2g, right panel).
hAFS-CM Hypo acts through the NFkB-controlled pro-survival genes Il6 and Cxcl1 and the PI3K/ Akt pathway. To identify the pathways responsible for hAFS-CM Hypo antagonism of Dox cardiotoxicity, gene expression profiles were compared between mNVCM treated with Dox or incubated with hAFS-CM Hypo followed by Dox exposure. Microarray analysis using Affymetrix technology revealed that pre-treatment with Scientific RepoRts | 6:29994 | DOI: 10.1038/srep29994 hAFS-CM Hypo led to the up-regulation of genes coding for cytokines, chemokines, growth factors, and chemokine and cytokine receptor binding proteins ( Fig. 3a and Supplementary Table S1). In particular, Il6 and Cxcl1 were highly expressed. Confirmatory real time qRT-PCR at 6 hours after Dox treatment demonstrated that these two genes were up-regulated by 7 and 29 fold compared to untreated cells and by 17-and 59 fold compared to Dox treatment, respectively, when the hAFS-CM Hypo preceded Dox exposure (Fig. 3b). Since IL-6 and CXCL-1 have been shown to be involved in mNVCM survival 18,19 , we investigated whether blockade of these two cytokines might influence the protective effects of hAFS-CM Hypo . The decrease in apoptosis attained with hAFS-CM Hypo was significantly blunted when pre-incubation with hAFS-CM Hypo was followed by treatment with Dox, along with antibodies against IL-6 or IL-6 as well as CXCL-1 (either soon after or after 6 hours post Dox administration, Fig. 3c), thus indicating that hAFS-CM Hypo inhibits Dox-induced apoptosis via IL-6 and, to a lesser extent, CXCL-1.

hAFS-CM Hypo antagonizes Dox toxicity on human cardiac progenitor cells. Damage, senes-
cence and depletion of human endogenous cardiac progenitor cells (hCPC) are thought to be critical for the pathogenesis of Dox cardiomyopathy 4,22 . Therefore, we assessed whether hAFS-CM Hypo may also protect hCPC against Dox. hCPC were isolated from human atrial samples as previously described 23 and treated with Dox with or without prior incubation with hAFS-CM Hypo or hNCTC-CM Hypo . The frequency of SA β -galactosidase positive cells was significantly increased by Dox, as expected, and reduced by 51% by pre-incubation with hAFS-CM Hypo (50.1 ± 6.4% vs 24.5 ± 1.9% of SA β -galactosidase positive cells), whilst hNCTC-CM Hypo was not effective (Fig. 6a,b). The same trend was observed when senescence was evaluated by analysing the expression of p16 INK4a , with hAFS-CM Hypo decreasing p16 INK4a levels by 30% (2.3-fold increase vs 1.6-fold increase compared to untreated cells, Fig. 6c,d). In agreement with the results of previous, similar in vitro studies 4,24 , Dox induced hCPC apoptosis, as assessed by staining for activated caspase-3, to a limited extent (4.5 ± 0.6% vs 0.9 ± 0.1% cleaved caspase-3 positive cells, Fig. 6e,f). The rate of apoptosis was even lower when cells were pre-incubated with hAFS-CM Hypo before exposure to Dox, although not significantly (4.5 ± 0.6% vs 3.8 ± 0.5% cleaved caspase-3 positive cells with Dox alone vs hAFS-CM Hypo + Dox, respectively; Fig. 6e,f). No significant effects on either senescence or apoptosis markers were exerted by hAFS-CM or hNCTC-CM alone (see Supplementary Fig. S2).

Discussion
This work provides robust evidence, gathered from the study of different cellular models, that the hAFS secretome (hAFS-CM) antagonizes Dox cardiotoxicity. As Dox-induced left ventricular dysfunction and heart failure are relatively common and by themselves have clinical burden, cause hospitalization, and carry a risk of mortality 1,5 , our results hold high translational potential value and great promise for future therapeutic strategies.
DNA damage has been pinpointed as the driver of the noxious effects of Dox on cardiomyocytes and CPC 10 , which eventually leads to their senescence and apoptosis 6,8,9,17 . These cellular alterations are the basis for a cardiomyopathy that typically has delayed clinical presentation. Although the cardiac cytotoxicity of Dox can be observed shortly after first infusion of the drug 25 , myocardial tissue derangement, cardiac dilatation and decline in left ventricular ejection fraction tend to occur towards the end of chemotherapy or thereafter -even years later 26,27 . Whereas depletion of terminally-differentiated cardiomyocytes is intuitively associated with decreased ventricular mass and, thereby, contractile force, senescence of CPC may worsen Dox cardiomyopathy by hindering the intrinsic ability of the myocardium to self-repair and regenerate via the activation of endogenous cardiac progenitors 22 . Previously published work showed that senescence is more pronounced than apoptosis in CPC exposed to Dox and may be primarily responsible for deficient CPC activity, conferring greater susceptibility to myocardial injury and, eventually, development of anthracycline-related cardiomyopathy 4,24 . We confirmed that Dox strongly triggered CPC senescence and found that hAFS-CM counteracted it; conversely, Dox-induced apoptosis of CPC survival was limited and, possibly because of such a small magnitude of effect, it was non-significantly decreased by hAFS-CM. Furthermore, hAFS-CM antagonized Dox-initiated DNA damage, senescence, and apoptosis of cardiomyocytes. Thus, we conclude that hAFS-CM offsets fundamental and early aspects of Dox cardiotoxicity.
In the last decade, many authors have demonstrated that structural and functional improvements obtained after engraftment of stem cells into the heart can largely be attributable to secreted factors, collectively indicated as secretome, rather than to the direct trans-differentiation of the exogenous stem cells into cardiovascular cells forming new viable tissue [28][29][30] . As a consequence, a current paradigm in cardiac regenerative medicine is to exploit the stem cell secretome for therapy. In a rat model of ischemia/reperfusion injury, we previously showed that systemic injection of hAFS-CM remarkably reduced infarct size in the short term to the same extent as administration of hAFS 14 . The experiments presented here integrate these earlier findings and confirm that, in general, the hAFS secretome is a valuable source of cardioprotective factors.
Other authors have reported that both transplantation of embryonic stem cells (ES) and the ES-CM oppose Dox-induced cardiomyopathy in mice [31][32][33][34] . Protection against Dox cardiotoxicity via potent paracrine effects has also been described for the secretome of mesenchymal stem cells (MSC) derived from either human induced pluripotent stem cells (iPS) or from bone marrow 35 . Overall, these results are consistent with ours and suggest that, in principle, the stem cell secretome may be employed to treat Dox cardiac side effects, while avoiding the problems related to standard cell therapy, such as immune rejection and teratogenicity. Yet, important limitations with ES and iPS or bone marrow MSC should not be overlooked: ES are not free of ethical constraints, and culture of iPS is technically challenging and rather time-consuming. These drawbacks are not encountered with adult bone marrow MSC, which, however, are obtained by invasive sampling with low yield and present limited self-renewal potential. By contrast, there are no ethical concerns related to hAFS, as they are isolated from remaining samples of amniotic fluid collected by amniocentesis for prenatal screening. Moreover, hAFS are endowed with remarkable self-renewal capability, ES-like properties -such as proliferative ability and expression of pluripotent markers -and a higher paracrine potential than that of adult stem cells. That hAFS are immature foetal cells and developmentally very "young", may go some way to explain their more powerful paracrine potential than adult stem cells. Finally, they withstand cryopreservation for a long time while maintaining a stable karyotype 11 , which makes their banking and scale up expansion very feasible.
Multiple pathways may mediate the antagonism of Dox toxicity by the hAFS-CM. Here it is suggested a PI3K/ Akt-dependent role for NF-κ B and its target genes, Il6 and Cxcl1. Direct evidence of the involvement of IL-6 and, secondarily, CXCL-1 was obtained by using specific blocking antibodies. IL-6 is released by stressed cardiomyocytes and promotes cardiac cell survival in an autocrine/paracrine fashion 18 . It acts through several intracellular mediators, including PI3K/Akt 36,37 , which raises the possibility of a feed-forward signalling loop, whereby PI3K/Akt promote the transcription of Il6 and IL-6 activates PI3K/Akt. Furthermore, CXCL1, the murine homologue of IL-8, is an established angiogeneic factor 38,39 and has been reported to exert anti-inflammatory and pro-survival effects in a mouse model of autoimmune myocarditis 19 . Interestingly, PI3K/Akt activity has repeatedly been linked to protection against Dox cardiotoxicity 16,40 , including by the ES-CM 32 . In addition, NF-κ B is the nexus of one of the functional networks associated with the top 20 factors preferentially overexpressed in the iPS-derived MSC-CM, which has also been found to reduce Dox cardiotoxicity 35 . Hence, the secretomes of different stem cell types may have in common at least part of the cellular mechanisms by which they counteract Dox cardiac damage. As discussed above, however, compared with the conditioned medium of other stem cells, the hAFS-CM has unique characteristics, such as the ease of isolation from discarded clinical samples and the high-self renewal capacity in culture, making this cells an ideal source to exploit for future paracrine therapy.
The sensitivity of cardiac cells to Dox is dictated by the balance between drug retention and efflux, the latter being largely function of the activity of the transporter, ABCB1B (also known as Multidrug Resistance Protein 1 or P-glycoprotein). In cardiomyocytes, both induction of this gene by exogenous stimuli 41 , and overexpression via genetic construct 42 , confer resistance to Dox injury. Moreover, in patients there is a correlation between polymorphisms in ABCB1B and susceptibility to anthracycline cardiotoxicity 43 . We now report that the hAFS-CM also promotes Abcb1b transcription via PI3K/Akt and, as a consequence, Dox extrusion from cardiomyocytes. Remarkably, it has recently been demonstrated that NF-κ B, which is recruited by the hAFS-CM, is a positive regulator of Abcb1b expression 20 . This work did not aim to evaluate the effects of hAFS-CM on cancer cells. Nonetheless, it is worth noting that in a human breast cancer cell line the hAFS-CM did not modulate ABCB1 levels or Dox efflux, indicating that, at least in some tumour cell types, Dox metabolism is not enhanced by hAFS-CM. In terms of the prospect of a future therapeutic use of the hAFS secretome, this finding is reassuring.
We acknowledge that the present study has limitations. The data obtained with the embryonic H9c2 cardiomyoblasts and neonatal cardiomyocytes (mNVCM) were not replicated with mature adult cardiomyocytes. Since these cell types correspond to different stages of development and, thereby, display substantial phenotypic differences, this aspect may be especially relevant from a translational point of view and needs further evaluation. Furthermore, the effects of sustained incubation of cardiomyocytes and CPC with hAFS-CM need to be investigated, in order to define the optimal duration of treatment for possible therapeutic application. In addition, in vivo experiments are needed to investigate the effects of the hAFS-CM on mature adult cardiomyocytes and CPC in vivo while within their own microenvironment, which comprises other cell types possibly influenced by the hAFS-CM.
As far as the mechanisms of the hAFS-CM paracrine cardioprotection are concerned, it is likely that other signalling pathways, besides PI3K/Akt and a set of NF-κ B responsive genes, contribute to the hAFS-CM prevention Scientific RepoRts | 6:29994 | DOI: 10.1038/srep29994 of Dox toxicity. Furthermore, detailed characterization of hAFS-CM composition is essential to pinpoint the soluble factors that specifically mediate the inhibition of Dox DNA damage, senescence, and apoptosis. Finally, a more detailed analysis of the possible influence of donor age on the hAFS secretome is required to understand whether there may be significant variability between cell batches obtained from different women. Indeed, in our study, we used the secretome of different hAFS obtained by pooling together amniotic fluid samples from 3 different donors in order to limit possible variations of donor age on cell proliferation.
In conclusion, we prove, in principle, that the hAFS secretome protects cardiomyocytes and CPC against Dox toxicity. These results substantiate the concept of a stem cell based paracrine approach via cell-free delivery of bioactive factors to prevent cardiotoxicity of Dox and possibly other anticancer agents. Such a treatment might be especially important for survivors of childhood cancer, in whom cardiac complications of chemotherapy can significantly curtail life expectancy. In addition, the data presented herein may lay the foundations for future research work, using the hAFS secretome as an easily obtainable and appealing source of paracrine cardioactive molecules that may become an advanced medicinal product for new cardiac regeneration strategies, without some of the ethical and technical limitations associated with embryonic and adult stem cell use.

Methods
Cell culture. hAFS were sorted for c-kit expression (CD117 MicroBead Kit, Miltenyi Biotechnology) within the cells isolated from left over samples of II trimester amniotic fluid, collected via prenatal screening amniocentesis, with informed consent from donors and proved negative for disease 11,44,45 . The protocol complied with the Helsinki Declaration and was approved by the local ethical committee Comitato Etico Regionale IRCCS AOU San Martino -IST (protocol 0036463/15 P.R. 428REG2015). Cells were cultured in Minimal Essential Medium (MEM) alpha with 15% fetal bovine serum (FBS), 1% L-glutamine, 1% penicillin/streptomycin, (Gibco-Thermo Fisher Scientific), 18% Chang B, and 2% Chang C (Irvine Scientific) at sub-confluence. To limit the intrinsic variability of primary cultures, which might be influenced by the donors different age (samples were obtained by women from 25 up to 46 years old, mean: 38.4 ± 3.3 years old), 3 different human amniotic fluid samples were pooled together to isolate hAFS for each experiment.
The H9c2 cell line was bought from the European Collection of Authenticated Cell Cultures (Salisbury, UK) and cultured as already described 15,16 . mNVCM were isolated via enzymatic digestion from 2-day-old C57/Bl6 mouse heart by multiple digestions in a collagenase II solution (300 U/ml, Worthington Biochemicals), according to 46 , and seeded on gelatin (1% solution, Sigma-Aldrich) coating at 10 5 cells/cm 2 in culture medium (69% Dulbecco′ s Modified Eagle Medium, DMEM, 15% M199, 10% horse serum, 5% FBS, 1% penicillin/streptomycin and 1% L-glutammine, Gibco-Thermo Fisher Scientific). All animal procedures were carried out in compliance with national and international laws and specific authorisation (protocol 792/2015-PR from Animal Facility of IRCCS AOU San Martino-IST).
hCPC were isolated as previously described 23 from human auricolae fragments, obtained following written informed consent from patients, in compliance with the Helsinki Declaration and upon approval of the local ethical committee IRCCS Istituto Europeo di Oncologia and Centro Cardiologico Monzino (protocol CCFM C9/607). Briefly, the myocardial tissue was repeatedly digested at 37 °C in a 3 mg/ml collagenase solution (Serva), cells were FACS-sorted (FACSAria, Beckton-Dickinson) for c-kit expression using an APC-conjugated antibody (anti-CD117, clone YB5.B8; BD Biosciences) and cultured in Ham's F12 medium (Lonza) with 10% FBS (Thermo Fisher Scientific), 2 mM L-glutathione and 5 × 10 −3 U/mL human erythropoietin (both Sigma-Aldrich), 10 ng/mL bFGF (Peprotech), and antibiotics (Lonza). Supplementary   Fig. S1. hAFS and NCTC 2544 cells were cultured for 24 hours in serum-free medium (SF: MEM alpha medium for hAFS and MEM/EBSS for NCTC 2544, both with 1% L-glutamine and 1% penicillin/streptomycin) in normoxia (20% O 2 and 5% CO 2 at 37 °C) or hypoxia (1% O 2 and 5% CO 2 at 37 °C in an hypoxic workstation, Baker Ruskinn, Carlibiotec s.r.l.). The hAFS-CM Normo , hAFS-CM Hypo , hNCTC-CM Normo and hNCTC-CM Hypo were concentrated 20 times using ultrafiltration membranes with a 3 kDa selective cut-off (Amicon Ultra-15, Millipore). Protein concentration was measured by Bradford assay. Samples were stored at − 80 °C until use. At least 3 different batches of hAFS-CM and hNCTC-CM were used for each experiment. Experiment outline. hAFS-CM was used at 40 μ g/ml, this concentration being the most protective in preliminary experiments with H9c2 cardiomyoblasts. The hNCTC-CM was employed at the same concentration. Cells were pre-incubated with hAFS-CM or hNCTC-CM for 3 hours prior to exposure to Dox. Apoptosis and cell viability were measured after Dox treatment (1 μ M) for 21 hours. Blocking experiments were performed by adding 0.025 μ g/ml anti-IL-6 and/or 0.25 μ g/ml anti-CXCL1 antibodies (clone MP5-20F3 and 48415, respectively, R&D Systems) at the end of the 3 hour-incubation with the hAFS-CM when Dox treatment began, or 9 hours after hAFS-CM incubation, i.e. 6 hours after starting Dox treatment. Apoptosis was then evaluated 21 hours after Dox exposure. Senescence was examined after 3 hours of Dox exposure (H9c2: 0.1 μ M, mNVCM and hCPC: 0.2 μ M) followed by 42 hours in complete medium. DNA damage and gene expression profile were evaluated 6 hours after treatment with Dox (1 μ M), whereas nuclear translocation of NF-κ B was assessed after Dox (1 μ M) incubation for 1.5 hours. To determine the acute effect on Akt phosphorylation, mNVCM were treated for 10 minutes with the hAFS-CM Hypo , with or without pre-treatment with LY-294002 (20 μ M, Sigma-Aldrich) for 1 hour. Before every experiment, cells were incubated in low-serum culture medium (DMEM low-glucose with 0.5% FBS, 1% L-glutamine, and 1% penicillin-streptomycin) for 1 hour. Treatment outline is shown in Supplementary Fig. S1.
Cell viability assay. Following Dox exposure with or withour hAFS-CM or hNCTC-CM pre-incubation, MTT solution (250 μ g/ml; Sigma-Aldrich) was added to the cells for 4 hours and absorbance read at 570 nm on a VersaMax (GE Intelligent Platforms) plate reader.
Assessment of DNA damage. After exposure to Dox (1 μ M) with or without pre-treatment with the hAFS-CM Hypo , mNVCM were fixed in paraformaldehyde solution (4%, Sigma Aldrich) and incubated with a rabbit monoclonal antibody against γ H2aX (Ser139, clone 20E3, Cell Signalling Technologies) followed by Alexa Fluor488-conjugated secondary antibody (Thermo Fisher Scientific). Fluorescence emission was quantified using an Infinite 200 PRO plate reader (Tecan). Immunocytochemistry, immunofluorescence and cell culture images were acquired with an Axiovert microscope equipped with Axiovision software (Carl Zeiss).
Gene expression profiling. The gene expression profile of mNVCM treated with Dox (1 μ M) with or without pre-incubation with the hAFS-CM Hypo , was analysed by Affymetrix MG 430.2 microarray technology. Total RNA was extracted using the RNeasy Micro Kit (Qiagen). cRNA for hybridization was prepared using the GeneChip ® 3′ IVT PLUS Reagent Kit following the instructions provided by Affymetrix. Hybridisation and scanning were performed by standard procedures, as recommended by Affymetrix. Data were deposited in the Gene Expression Omnibus repository (www.ncbi.nlm.nih.gov/geo/, accession number: GSE74513) and pre-processed using the RMA algorithm with quantile normalization 48 . Differentially expressed genes were identified applying a threshold of two-fold increase or decrease of the expression values when comparing the two conditions. Enrichment of functional categories of genes was analyzed by EnrichR Gene Set Enrichment Analysis (http:// amp.pharm.mssm.edu/Enrichr/) 49 .
Real-time quantitative RT-PCR analysis. Total RNA from mNVCM and MDA-MB-231 cells was extracted using Qiazol Lysis Reagent (Qiagen) and cDNA obtained using the iScriptTM cDNA Syntesis Kit (Bio-Rad). Real-time qRT-PCR was carried out on a 7500 Fast Real-Time PCR System (Thermo Fisher Scientific) using Syber Green Master Mix (Thermo Fisher Scientific). Primer sequences for mouse Il6, Cxcl1, Abcb1b, Hprt, and human ABCB1 and GAPDH were designed using the NCBI Primer-Blast tool (http://www.ncbi.nlm.nih.gov/ tools/primer-blast/; see Supplementary Table S2 for sequences). Gene expression levels were normalized using Hprt or GAPDH as endogenous control by applying the 2 −ddCt method.