Irradiated mesenchymal stromal cells induce genetic instability in human CD34+ cells

Radiation-induced bystander effects (RIBE) in human hematopoietic stem and progenitor cells may initiate myeloid neoplasms (MN). Here, the occurrence of RIBE caused by genotoxic signaling from irradiated human mesenchymal stromal cells (MSC) on human bone marrow CD34+ cells was investigated. For this purpose, healthy MSC were irradiated in order to generate conditioned medium containing potential genotoxic signaling factors. Afterwards, healthy CD34+ cells from the same donors were grown in conditioned medium and RIBE were analyzed. Increased DNA damage and chromosomal instability were detected in CD34+ cells grown in MSC conditioned medium when compared to CD34+ cells grown in control medium. Furthermore, reactive oxygen species and distinct proteome alterations, e.g., heat-shock protein GRP78, that might be secreted into the extracellular medium, were identified as potential RIBE mediators. In summary, our data provide evidence that irradiated MSC induce genetic instability in human CD34+ cells potentially resulting in the initiation of MN. Furthermore, the identification of key bystander signals, such as GRP78, may lay the framework for the development of next-generation anti-leukemic drugs.


Introduction
Radiation therapy for neoplastic or non-neoplastic disorders may induce myeloid neoplasms (MN) in humans which are referred to as therapy-related myeloid neoplasms (t-MN) [1]. t-MN comprise the therapy-related cases of acute myeloid leukemia (t-AML), myelodysplastic syndromes (t-MDS) and myelodysplastic/myeloproliferative neoplasms (t-MDS/MPN) [1]. t-MN show genetic similarity to other high-risk MN [2,3] and are associated with poor prognosis [4,5]. Risk factors for the development of t-MN may include (a) inherited germline mutations in cancer susceptibility genes (e.g., BRCA1 wt/mut ), (b) acquired DNA damage in hematopoietic stem and progenitor cells (HSPC), (c) selection of pre-existing mutated hematopoietic clones (e.g., TP53 wt/mut ) and (d) alterations in the bone marrow stromal niche [6].
In summary, RIBE might be mediated in HSPC by genotoxic signaling from irradiated MSC and may account for a major pathomechanism in the initiation of certain MN. In contrast, the occurrence of RIBE in human HSPC has never been verified and genotoxic signaling factors are unknown yet. Therefore, our study was designed to analyze RIBE in CD34+ myeloid progenitor cells by immunofluorescence microscopy of γH2AX (as a readout of DNA damage), by analysis of G-banded chromosomes (for detection of chromosomal instability (CIN)) and by luminescence plate reading of cell viability. Furthermore, ROS and proteome alterations were assessed in irradiated MSC, MSC conditioned medium and CD34+ cells grown in MSC conditioned medium for the identification of potential genotoxic signaling factors.

Femoral head preparation
This study was approved by the Ethics Committee II, Medical Faculty Mannheim, Heidelberg University. Procedures were performed in accordance with the local ethical standards and the principles of the 1964 Helsinki Declaration and its later amendments. Written informed consent was obtained from all study participants. Femoral heads of 12 patients with coxarthrosis (7 females, 5 males, mean age: 69 years) undergoing endoprothetic surgery were collected ( Table 1). The bones were broken into fragments and incubated for 1 hour at 37 °C in phosphate-buffered saline (PBS) supplemented with 1 mg/ml collagenase type I (Thermo Fisher, Waltham, US). The supernatants were filtered through 100 µm pores of a cell strainer (Greiner Bio-One, Kremsmünster, Austria). MSC were grown from the fragments retained in the cell strainers in serum-free StemMACS MSC Expansion Media XF (Miltenyi Biotec, Bergisch Gladbach, Germany) supplemented with 1% penicillin/streptomycin. Adherent MSC were expanded in T175 flasks in a humidified 5% CO 2 atmosphere at 37 °C and passaged at 80% confluency. Furthermore, CD34+ cells were enriched from the filtrates by Ficoll density gradient centrifugation and magnetic-activated cell sorting using CD34 antibody-conjugated microbeads (Miltenyi Biotec). CD34+ cells were grown in serum-free StemSpan SFEMII (Stemcell Technologies, Vancouver, Canada) supplemented with StemSpan Myeloid Expansion supplement (SCF, TPO, G-CSF, GM-CSF) (Stemcell Technologies) and 1% penicillin/streptomycin in a humidified 5% CO 2 atmosphere at 37 °C.

Preparation of MSC conditioned medium
MSC were grown in T175 flasks until reaching 80% confluency. MSC were rinsed in PBS before fresh serum-free StemSpan SFEMII was added. Afterwards, MSC were irradiated with 2 Gy of 6 MV x-rays in a Versa HD linear accelerator (Elekta, Stockholm, Sweden), while control MSC were not irradiated. Medium was conditioned by the irradiated and nonirradiated MSC for a period of 4 h incubation at 37 °C to generate MSC conditioned medium and control medium, respectively. The media were centrifuged at 1,200 rpm for 10 min, and supernatants were stored at -20 °C.

RIBE analyses
RIBE were analyzed in CD34+ cells at day 6 after culture for 3 days in untreated medium followed by culture for 3 days in conditioned medium or control medium, respectively.
Immunofluorescence staining of the DNA double-strand-break marker γH2AX [32] was performed using a JBW301 mouse monoclonal anti-γH2AX antibody (Merck, Darmstadt, Germany) and an Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (Thermo Fisher) [33,34]. At least 50 nuclei were evaluated in each analysis. Cytogenetic analysis of G-banded chromosomes was performed according to standard procedures [35]. At least 25 metaphases were analyzed in each sample according to ISCN 2016 [36]. Cell viability was assessed using the CellTiter-Glo luminescent cell viability assay (Promega, Fitchburg, US) according to the manufacturer´s instructions. Luminescence was measured using a microplate reader (Tecan, Männedorf, Switzerland). ROS were analyzed using the ROS Detection Kit (PromoCell, Heidelberg, Germany) according to the manufacturer´s instructions. Luminescence was measured using a microplate reader (Tecan).

Protein quantitation using mass spectrometry
A proteomics approach for label-free quantitation using nanoscale liquid chromatography coupled to tandem mass spectrometry (nano LC-MS/MS) was applied for comparison of proteome differences.

Sample preparation for proteome analysis
Samples were prepared from 2 Gy irradiated MSC 4 h after irradiation and from nonirradiated control MSC. All MSC of 80% confluent T175 flasks were collected and washed three times in PBS. Afterwards, MSC were lysed in 200 µl RIPA buffer supplemented with Halt Protease Inhibitor Cocktail (100X) (Thermo Fisher) on ice for 30 min. Further, MSC conditioned medium and control medium were prepared using serum-free StemMACS MSC Expansion Media XF as stated before. Finally, samples from CD34 + cells were prepared at day 6 after culture for 3 days in untreated medium followed by culture for 3 days in conditioned medium or control medium, respectively. After washing the samples three times in PBS, 1x10 6 CD34 + cells of each sample were lysed in 200 µl RIPA buffer supplemented with Halt Protease Inhibitor Cocktail (100X) on ice for 30 min. Lysates were stored at -20 °C.

Sample fractionation by SDS-PAGE and in-gel digestion
Cell culture supernatants were concentrated tenfold before SDS polyacrylamide gel electrophoresis (SDS-PAGE) by ultrafiltration (MWCO 5 kDa). Samples were heated to 95 °C for 5 min and cooled on ice prior to loading on NuPAGE 4-12% Bis-Tris gels (Thermo Fisher). SDS-PAGE was performed of all compared samples in parallel according to the manufacturer's specification. Proteins were fixed within the polyacrylamide matrix by incubating the entire gel in 5% acetic acid in 1:1 (vol/vol) water:methanol for 30 min. After Coomassie staining (60 min) the gel slab was rinsed with water for 60 min. Each lane was excised and subdivided in three fractions according to protein complexity over standardized molecular weight ranges. Gel fractions were cut into small pieces. Subsequently, proteins were destained by 100 mM ammonium bicarbonate/acetonitrile 1:1 (vol/vol) before reduction for 30 min in 10 mM DTT and alkylation for 30 min in 50 mM iodoacetamide. Finally, proteins were digested by trypsin overnight at 37 °C. Peptides were collected from supernatant and extracted additionally from gel pieces by 1.5% formic acid in 66% acetonitrile for 15 min.
Peptides from both steps were combined and dried down in a vacuum centrifuge.

Mass spectrometry
Fractions of dried peptides were re-dissolved in 35 µl 0.1% trifluoroacetic acid and analyzed individually. For this, peptides were loaded on a 75 µm x 2 cm Acclaim C18 precolumn (Thermo Fisher) using an RSLCnano HPLC system (Thermo Fisher). Then, peptides were eluted with an aqueous-organic gradient (4-44% acetonitrile, 0.1% formic acid) for 130 min and separated on a 75 µm x 15 cm Acclaim C18 column (Thermo Fisher) with a flow rate of 300 nl/min. A Triversa Automate (Advion, Ithaca, US) was used as ion source to produce a stable electrospray, which was analyzed on a LTQ Orbitrap XL mass spectrometer (Thermo Fisher). Each scan cycle consisted of one FTMS full scan and up to 10 ITMS dependent MS/MS scans of the ten most intense ions with dynamic exclusion set to 30 sec. Mass width was set to 10 ppm and monoisotopic precursor selection was enabled. All analyses were performed in positive ion mode.

Comparative proteome analysis
Differences in proteomes between treatment groups were analyzed by Proteome Discoverer version 2.4 (Thermo Fisher). Comparisons were made between matching sample types and fractions. CD34+ cells and MSC analyses were based on 5 replicates. For the comparison of protein supplement-free cell culture supernatants, 4 replicates were utilized. The analyses were based on at least 10 ppm mass accuracy and 1% false discovery rate. Peptides were identified using the SEQUEST algorithm and a human proteome database retrieved from UniProt (Aug. 2019, https://www.uniprot.org). Protein abundance was calculated based on intensities of unique precursor ions and limited to unmodified peptides with high confidence.
Precursor ion intensities were normalized to the total peptide amount in each sample. Protein abundance ratios derived from irradiated vs. non-irradiated cell samples were calculated as median of pairwise precursor comparison of replicates to reflect the pairwise experimental design of treatments. Missing intensities were imputed based on replicates, and statistics were calculated by background based ANOVA. In cell culture supernatants, the number of required background elements was insufficient for background based ANOVA. Therefore, ttests were calculated for individual proteins. Furthermore, differences in cell culture supernatant were based on the top three scored unique peptides to account for protein processing, such as signal peptide truncation, etc.. All protein identifications were filtered for a required minimum of at least two unique peptides. A minimum of two distinct peptides with similar regulation was utilized as a requirement for calculated ratios during manual inspection. In addition, a minimum detection in at least three replicates was an essential inclusion criterion for calculated ratios during manual inspection. Tables summarizing differences in proteomes between treatment groups meet all criteria described above and include corresponding p values.

Statistical analysis
Proteomic data were analyzed as outlined in the section above. All other statistical calculations were done with SAS software, release 9.4 (SAS Institute, Cary, US). For comparisons between treated groups and controls, Wilcoxon two-sample tests were used.
One sample t-tests were used in order to investigate if mean fold changes (fc) were different from 1.

Validation of cell-free MSC conditioned medium
To ensure that MSC conditioned medium was cell-free in our experiments only centrifuged

ROS in MSC and CD34+ cells
ROS were analyzed in 2 Gy irradiated MSC samples (n = 8) at 4 h after irradiation and in non-irradiated control MSC. Increased ROS levels (p = 0.0105) were detected in irradiated MSC (fc = 1.8 ± 0.2; mean ± standard error of mean (SEM)) when compared to nonirradiated MSC (fc = 1) (Fig. 1a). Furthermore, ROS were analyzed in CD34+ cell samples (n = 5) expanded for 3 days in untreated medium followed by culture for 3 days in conditioned medium or control medium, respectively. ROS levels tended to be increased (p = 0.2206) in CD34+ cells grown in conditioned medium (fc = 1.2 ± 0.2) when compared to ROS levels in CD34+ cells grown in control medium (fc = 1) (Fig. 1b).

Viability of CD34+ cells
Viability was assessed in CD34+ cell samples (n = 10) grown for 3 days in untreated medium followed by culture for 3 days in conditioned medium or control medium, respectively.
Viability of CD34+ cells grown in conditioned medium (fc = 1.1 ± 0.1; mean ± SEM) was similar when compared to viability of CD34+ cells grown in control medium (fc = 1; data not shown).  Table 2). In MSC, 31 of 1924 identified proteins (1.6%) were regulated at least twofold within 4 hours upon a single irradiation dose of 2 Gy compared to controls. The majority was upregulated (94%) and about half participate in translation, protein folding as well as protein degradation. Six altered proteins are part of the cytoskeleton and participate in its dynamic regulation. Four are members of nuclear transport mechanisms and the nuclear pore complex. The remaining participate in energy metabolism (e.g., glycolysis), oxidative stress detoxification, cellcell/matrix interaction and transmembrane signaling.

Proteome analysis in MSC, MSC conditioned medium and CD34+ cells
In the corresponding conditioned medium of MSC, 4 of 265 identified proteins (1.5%) were found increased in their abundance by factor 2 or higher 4 h after irradiation vs. controls.
Remarkably, three of these proteins are key proteins in the endoplasmatic reticulum (ER) and known for their role in protein folding as well as protein quality control. Besides one upregulated member of the glycolysis, no other proteins were differentially abundant in conditioned medium.
Exposure of CD34+ cells to the conditioned medium of irradiated MSC for three days induced quantitative changes of a minimum factor 2 in 5 of 2003 identified proteins (0.25%).
Similar to MSC, affected proteins participate in translation, protein degradation and cytoskeleton dynamics. Notably, eIF3f was lower abundant in CD34+ cells, whereas it was higher abundant in MSC in conjunction with irradiation. Unique to the response in CD34+ cells were changes in proteins participating in transcriptional regulation/chromatin remodeling and ERBB3 regulation. Overall, the response in CD34+ cells to conditioned medium affected much less proteins than in MSC, which were directly exposed to irradiation.

Discussion
The aim of our study was to analyze genetic alterations induced by DNA damage signaling from irradiated MSC to human CD34+ cells as a potential mechanism of MN initiation. For this purpose, RIBE were analyzed in human CD34+ cells grown in medium conditioned by 2 Gy irradiated human MSC. Notably, increased numbers of γH2AX foci as well as structural and numerical chromosomal aberrations were detected in CD34+ cells grown in MSC conditioned medium when compared to CD34+ cells grown in control medium. The increased numbers of γH2AX foci in CD34+ cells grown in MSC conditioned medium may not only indicate critical DNA damage potentially contributing to MN initiation, e.g., by activation of oncogenes or inactivation of tumor suppressor genes [37]. In addition, γH2AX foci may indicate DSB [32] involved in chromosomal rearrangements such as deletions, inversions and translocations. Indeed, t-MN related chromosomal aberrations were found in CD34+ cells grown in MSC conditioned medium when compared to whole chromosomes in CD34+ cells grown in control medium. Particularly, chtb(5q), chtb(7q), chtb(11q) and chtb(13q), that were found in CD34+ cells grown in MSC conditioned medium, coincided well with del(5q), del(7q), t(11q23.3) and del(13q), that are present in about 42%, 49%, 3% and < 5% of t-MN, respectively [1,6]. In addition, t-MN related aneuploidies, e.g., tetraploidies and octoploidies, were detected in CD34+ cells grown in MSC conditioned medium. Numerical chromosomal aberrations are caused by defects in mitosis, e.g., chromosomal non-disjunction and cytokinesis failure [38]. Moreover, tetraploidies are hallmark precursor lesions in diverse cancers, e.g., cervical cancer and neuroblastoma and occur in about 1% of AML but 13% of t-AML cases [38,39]. As tetraploid cells harbor 4n centrosomes, multipolar spindles may form potentially driving a CIN phenotype. With ongoing dedifferentiation of CD34+ cells CIN may further aggravate in the course of disease evolution, e.g., by frequent inactivation of TP53, which may result in rapid t-MN development [38]. Finally, the increased numbers of γH2AX foci and chromosomal aberrations did not seem to affect overall viability of CD34+ cells within the observation period as viability was similar in CD34+ cells grown in conditioned medium and in CD34+ cells grown in control medium.
ROS were analyzed in irradiated MSC and CD34+ cells grown in MSC conditioned medium for their potential contribution to bystander signaling from irradiated MSC to CD34+ cells.
Increased ROS levels were detected in irradiated MSC and in CD34+ cells grown in MSC conditioned medium. While ROS are known genotoxic molecules generated by endogenous and exogenous sources in each cell, ROS may also function as important regulators of intracellular signaling pathways, e.g., by covalent modification of specific cysteine residues in redox-sensitive target proteins [40]. Oxidation of specific cysteine residues in turn can lead to reversible modification of enzyme activity [40] with effects on diverse pathways including metabolism, differentiation and proliferation [41]. Hence, ROS may not only induce DNA damage but also dysregulate cellular pathways, thereby contributing to oncogenic transformation of CD34+ cells.
In order to identify potential mediators for the observed oncogenic transformation in CD34+ cells as well as mechanisms leading to their release in MSC and transduction in CD34+ cells, comparative proteome analyses were performed in three tiers of (a) irradiated MSC, (b) MSC conditioned medium and (c) CD34+ cells grown in MSC conditioned medium. Among these three comparisons, irradiated MSC showed the largest change in proteome, which is in accordance with the impact of the primary stimulus. Still, the response can be regarded as rather moderate, because only 1.6% of the analyzed proteome was altered by a factor 2 or higher. An underlying mechanism might be the relative radioresistance of MSC [42]. The majority of altered proteins in response to irradiation take part in translation, protein folding as well as protein degradation, indicating disturbed protein homeostasis and required replacement, repair and degradation of proteins. Interestingly, three of the few quantitatively altered proteins in MSC conditioned medium upon irradiation were key ER proteins involved in protein folding and their quality control. The highest increase was observed for GRP78, an ER chaperone, which dissociates from luminal domains of IRE1, PERK and ATF6 in consequence of ER stress resulting in activation of the unfolded protein response (UPR) [43] and promotion of the ER-associated protein degradation pathway (ERAD) [44]. In turn, ERAD relies on substrate degradation via the ubiquitin-proteasome system. Notably, two proteasome activator proteins (ECM29 and PA28-gamma) as well as a key assembly factor of SCF E3 ubiquitin ligase complexes (p120 CAND1) were all increased in irradiated MSC.
Altogether the results indicate that irradiation resulted in ER stress. The stress response may be induced in part by associated ROS. At proteome level, MSC responded to increased oxidative stress by elevating levels of peroxiredoxin-2 and GSTP1-1.
The perception about GRP78 has changed over the past decade, as a growing number of signaling processes become apparent, which are not related to its canonical role in the ER [45,46]. It appears that GRP78 is not exclusively present in the ER but can be relocated to the cell surface (csGRP78) or even secreted into the extracellular medium (sGRP78). Both have been described to confer critical roles in the context of cancer development and cell survival [45,46]. For example, sGRP78 can act as a pro-apoptotic ligand of csGRP78 on pancreatic β-cells [47], but as a mediator of pro-survival kinase signaling in endothelial cells [48]. In addition, csGRP78 plays a mechanistic role in PI3K/AKT driven leukemogenesis [49] and in Cripto/csGRP78 regulated hematopoietic stem cell survival [50]. Therefore, monitoring of sGRP78 and targeting of csGRP78 is evaluated in anti-cancer therapy [45,46].
Considering these emerging roles of GRP78, non-canonical csGRP78 signaling may impact the survival of CD34+ cells harboring genetic aberrations and contribute to oncogenic bystander signaling. The remaining two ER proteins with increased abundance upon irradiation in MSC conditioned medium were PDIA3 and calreticulin. PDIA3 catalyzes the rearrangement of disulfide bonds [51] and thereby enables correct folding of newlysynthesized glyco-proteins [52]. In addition, it interacts with the ER resident calcium binding lectins calreticulin and calnexin. Similar to the function of proteins altered in response to irradiation in MSC, calreticulin and calnexin participate in protein quality control and folding, more specifically, in a process known as the calreticulin/calnexin cycle [53]. The fact that three ER proteins with related function were specifically increased in the conditioned medium upon MSC irradiation, while the vast majority of other cytosolic and ER proteins were unaffected, suggests a specific release rather than uncontrolled cell lysis or unspecific cellular loss of the ER.
In CD34+ cells, the conditioned medium from irradiated MSC induced only minute detectable changes at proteome level after three days of exposure. Individual proteins participating in degradation, translation and cytoskeleton dynamics represent similar processes affected in MSC. Unique to CD34+ cells were proteins participating in chromatin remodeling (HMGB1) and ERBB3 signaling (EBP1). In particular EBP1 has oncogenic potential [54] and is highly expressed in AML cells [55], but HMGB1 assumes a number of roles in cancer development as well [56]. In addition, IQGAP1 can promote malign development [57]. As a consequence, several modes of action, which work individually or in conjunction, may transduce radiationinduced bystander signaling in effector cells.
Our data describe a sequence of cellular events from the primary response of irradiated MSC, over transmission of genotoxic signals in conditioned medium to the induction of mechanisms leading to critical DNA damage and CIN in CD34+ cells (Fig. 5). Ultimately, such genetic aberrations in effector cells have the potential for MN development. The results provide a fundamental basis for in-depth mechanistic research and novel therapeutic interventions to reduce the risk of t-MN development. Accordingly, antioxidants, such as Nacetylcysteine, might be able to counteract ROS in MSC and HSPC, thereby diminishing the risk for t-MN after irradiation. In addition, monoclonal antibodies (e.g., MAb159) [58] and peptidomimetics (e.g., BMTP-78) [59] targeting non-canonical csGRP78 signaling hold the potential to reduce the risk of t-MN.
In conclusion, genotoxic signaling by irradiated MSC emerges as a major pathomechanism in the initiation of certain MN offering the opportunity to take advantage of targeted therapeutic interventions. Specifically, our data suggest that bystander signals released by irradiated MSC, such as GRP78, are potential mediators of DNA damage and CIN in CD34+ cells, thereby providing a strong mechanistic link to the initiation of MN. More work is necessary to dissect the signaling pathways behind such oncogenic mediators which may define the targets of next-generation anti-leukemic drugs.

209
tetrapolidies are found in about 1% of AML but 13% of t-AML cases [29]. Hence, our finding of

232
Ultimately, it has the potential to uncover the identity of key bystander signals, which is 233 fundamental for the development of next-generation anti-leukemic drugs.