Prostaglandin E2 Enhances Aged Hematopoietic Stem Cell Function

Aging of hematopoiesis is associated with increased frequency and clonality of hematopoietic stem cells (HSCs), along with functional compromise and myeloid bias, with donor age being a significant variable in survival after HSC transplantation. No clinical methods currently exist to enhance aged HSC function, and little is known regarding how aging affects molecular responses of HSCs to biological stimuli. Exposure of HSCs from young fish, mice, nonhuman primates, and humans to 16,16-dimethyl prostaglandin E2 (dmPGE2) enhances transplantation, but the effect of dmPGE2 on aged HSCs is unknown. Here we show that ex vivo pulse of bone marrow cells from young adult (3 mo) and aged (25 mo) mice with dmPGE2 prior to serial competitive transplantation significantly enhanced long-term repopulation from aged grafts in primary and secondary transplantation (27 % increase in chimerism) to a similar degree as young grafts (21 % increase in chimerism; both p < 0.05). RNA sequencing of phenotypically-isolated HSCs indicated that the molecular responses to dmPGE2 are similar in young and old, including CREB1 activation and increased cell survival and homeostasis. Common genes within these pathways identified likely key mediators of HSC enhancement by dmPGE2 and age-related signaling differences. HSC expression of the PGE2 receptor EP4, implicated in HSC function, increased with age in both mRNA and surface protein. This work suggests that aging does not alter the major dmPGE2 response pathways in HSCs which mediate enhancement of both young and old HSC function, with significant implications for expanding the therapeutic potential of elderly HSC transplantation.


Introduction
Hematopoietic stem cells (HSCs) are responsible for continual replacement of all blood cell types, and HSC transplantation (HSCT) is a life-saving option for many patients with hematologic diseases. However, HSC function and engraftment capacity drastically decrease with age [1][2][3][4][5][6][7]. Loss of immune function, anemia, myeloid skewing, and increased incidence of myeloproliferative diseases and leukemia are observed in the elderly, and aged bone marrow (BM) cells are unfavorable for transplantation [4,[8][9][10][11]. While many factors are thought to contribute to HSC aging, including cell-intrinsic and niche-mediated [2,12,13], specific molecular pathways functionally linked to aged HSC defects are not well characterized, and no treatment exists to enhance aging hematopoiesis or augment the transplantation potential of these cells. The number of people > 65 years of age is projected to almost double between 2012 and 2050 [14], intensifying the problem of aged HSC dysfunction and the critical need for novel therapeutic approaches for those in need of HSC support.
Prostaglandin E2 (PGE 2 ) is a bioactive lipid with hematopoietic roles described since the 1970s [15][16][17]. More recently, the stable derivative 16,16-dimethyl PGE 2 (dmPGE 2 ) was found to enhance HSC frequency and transplantation efficiency in both zebrafish and murine models [18,19]. We have previously shown that pulse exposure to dmPGE 2 enhances homing, survival, and proliferation for both young (2-3 mo) mouse BM and human cord blood derived CD34 + cells, which was associated with increased CXCR4 and Survivin expression, decreased apoptosis, and increased HSC proliferation [19,20]. Further, dmPGE 2 pulse exposure augmented competitiveness of human cord blood grafts in a Phase I clinical trial with double cord blood transplantation [21], and has been used clinically to enhance engraftment of gene modified HSCs [22]. Thus, dmPGE 2 is known to enhance HSC function and has potential for clinical translation.
The objective of the current work was to assess the effects of dmPGE 2 pulse exposure on aged HSCs at both the functional and molecular level in comparison to young HSCs. Functionally, dmPGE 2 pulse prior to competitive serial transplantation significantly enhanced the long-term repopulating capacity of aged HSCs similarly to young. A high-throughput transcriptome comparison of old and young HSC responses to dmPGE 2 highlighted common transcriptional pathways in old and young mediating HSC enhancement, and identified novel signaling alterations in HSCs with age.

Materials and Methods
Mice All studies were approved by the Indiana University School of Medicine Institutional Animal Care and Use Committee. Aged mice: C57BL/6J male mice were purchased at 10 weeks old from Jackson Laboratories (Bar Harbor, ME) and aged in our facility until used at 25 mo of age. Young mice: C57BL/6J male mice were purchased at 10 weeks old from the Indiana University In Vivo Therapeutics Core (IVTC) and used at 3 mo of age. Transplant recipient mice: B6.BoyJ congenic (CD45.1) male and female mice were purchased at 6-8 weeks of age from the IVTC and used as transplant recipients at 8-10 weeks of age. Transplants were randomized by recipient sex and age such that young and old donor samples were equally distributed between recipients that were male or female, and younger (closer to 8 weeks) or older (closer to 10 weeks). Mice were randomized within cages so that each cage contained recipients of both vehicle and dmPGE 2 -pulsed cells, but from the same donor when possible for rigor of comparison between matched samples.

BM Collection
Femurs, tibiae, pelvic bones, and humeri from young and old C57BL/6J mice were flushed with cell buffer (PBS containing 2 % FBS and 2 mM EDTA). Cells were passed through a 40 μm filter, and total nucleated cells (TNC) enumerated using an Element HT5 Hematology Analyzer (Heska Corporation, Loveland, CO). In the transplantation experiment, an aliquot of cells from each mouse was removed for flow cytometric assessment of young vs. old BM prior to dmPGE 2 pulse.

DmPGE 2 Pulse Exposure
DmPGE 2 in methyl acetate from Cayman Chemicals (Ann Arbor, MI) was stored at -20 o C. Prior to use, dmPGE 2 was evaporated to dryness on ice under N 2 and reconstituted in 100 % EtOH at a stock concentration of 10 mg/ml (26.28 mM). Whole BM (WBM) cells (transplantation experiments) or lineage-depleted WBM cells (RNA-seq experiments) from individual C57BL/6J (CD45.2) young or old mice were split in two portions. One portion was pulsed in a concentration of 10 µM dmPGE 2 in cell buffer, and the other in an equivalent volume of vehicle (100 % EtOH) in cell buffer, at 2.5 × 10 6 TNC/mL for 1 h in a humidified CO 2 incubator at 37°C, with vortexing every 15 min. Cells were then centrifuged at 500 x g for 10 min to remove dmPGE 2 or vehicle and washed with cell buffer.

Competitive Serial Transplantation
Pulsed WBM cells from 4 young and 4 old donor C57BL/6J mice (CD45.2) were transplanted into 6 recipients per donor, where 3 received cells pulsed with dmPGE 2 and 3 received cells from the same donor pulsed with vehicle for matched analysis. To that end, 8-10 week-old congenic recipient mice (CD45.1) were exposed to 137 Cs irradiation (11 Gy split dose, 4 h apart) using a Mark 1 Irradiator (JL Shepherd, San Fernando, California), as previously described [23]. DmPGE 2 -or vehicle-pulsed donor WBM cells were resuspended in PBS and combined with competitor CD45.1 WBM cells not pulsed with dmPGE 2 or vehicle in a 3:2 ratio, for a final retro-orbital injection of 100 uL containing 3 × 10 5 donor and 2 × 10 5 competitor TNCs per recipient mouse. Recipients were given autoclaved acidified water (pH 2.0-3.0) and irradiated Uniprim diet (Envigo, Madison, WI) for 1 week prior to, and 4 weeks after irradiation/transplantation. Peripheral blood (PB) was analyzed monthly for donor chimerism and for multilineage reconstitution at month 6.
After 7 mo, WBM was collected from all primary recipients. Cells from each set of 3 replicate primary recipients (or 2 in two cases where one recipient died in primary phase) were combined equally and 1.5 × 10 6 TNC transplanted into 3 secondary CD45.1 recipients conditioned with radiation as above. PB was analyzed monthly for donor chimerism, except for months 4-5 due to pandemic-related laboratory restrictions, and for multilineage reconstitution at month 6.
WBM Processing, HSC Sorting, and RNA Extraction for Sequencing WBM cells from 8 young and 4 old mice were enriched for immature cells by magnetic lineage depletion (EasySep Mouse Hematopoietic Progenitor Cell Isolation kit, STEMCELL Technologies), then pairs of young samples were combined to increase the number of pHSCs available for sorting from 4 individual BM samples from young mice alongside 4 individual BM samples from old mice. Cells were then pulsed with dmPGE 2 or vehicle for 1 h as described above, immediately stained, and viable pHSCs isolated by fluorescence associated cell sorting (FACS) using a SORP Aria flow cytometer (BD Biosciences). Cells were sorted directly into lysis buffer for RNA extraction using the RNeasy Plus Micro Kit (Qiagen, Hilden, Germany). A 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) was used for quality control of all RNA preparations, giving a median RNA integrity of 9.6 RIN (range 6.2-10.0).

RNA Sequencing and Statistics/Bioinformatics
Library preparation was performed using the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech, Mountain View, CA) and the Nextera XT DNA Lib Kit (Illumina, San Diego, CA). Two hundred picomolar pooled libraries were utilized per flow cell for clustering amplification on cBot using HiSeq 3000/4000 PE Cluster Kit and sequenced with 2 × 75 bp paired-end configuration on HiSeq4000 (Illumina) using a HiSeq 3000/4000 PE SBS Kit. Sequencing data were assessed for quality using FastQC version 0.11.5 (Babraham Bioinformatics, Cambridge, UK). The sequence reads were mapped to the mouse genome (UCSC mm10) using STAR (Spliced Transcripts Alignment to a Reference) [24] version 2.5 using parameter "-outSAMmapqUnique 60". To evaluate the quality of the RNA-seq data, number of reads that fall into different annotated regions (exonic, intronic, splicing junction, intergenic, promoter, UTR, etc.) of the reference genes were determined with bamUtils [25] version 0.5.9.
Uniquely mapped sequencing reads were assigned to mm10 refGene genes and quantified using featureCounts (from subread) [26] version 1.5.1 using parameters "-s 2 -p -Q 10". Low quality mapped reads (including reads mapped to multiple positions) were excluded. Differential expression (DE) analysis was performed with edgeR [27]. In this workflow, the statistical methodology applied uses negative binomial generalized linear models with likelihood ratio tests. A paired design was used to compare samples across conditions such that the relationships between samples from the same animals were retained ("blocking" in edgeR). False discovery rate (FDR) calculations therefore reflect the collective significance of the treatment on gene expression while accounting for baseline differences between the animals.
DE data was analyzed for biological insights using Ingenuity Pathway Analysis (IPA, Qiagen, Hilden, Germany) [28]. Two IPA Core Analyses were run, one for the young and one for the old dmPGE 2 vs. vehicle comparisons, with cutoffs set to FDR < 0.05 and log 2 FC>|0.5| (FC, fold change). The young and old analyses were compared using the IPA Comparison Analysis function. Upstream Regulators were filtered on Genes, RNAs, and Proteins, zscore cutoff |2|, p-value cutoff 1.3 (log 10 ). Diseases and Functions were filtered on Cellular and Molecular Functions, with additional removal of cancer cell-specific functions, z-score cutoff |2|, p-value cutoff 1.3 (log 10 ). The set of 70 genes increased by dmPGE 2 in young (FDR < 0.05) but not in old (FDR > 0.6) was submitted to the DAVID 6.8 Functional Annotation tool using default settings and output from the Functional Annotation Chart filtered on UP_KEYWORDS and cutoff at Benjamini-adjusted p-value < 0.01 (david.ncifcrf.gov) [29].
Heatmaps were generated in Microsoft Excel using conditional formatting for cell color based on FPKM values (fragments per kilobase per million mapped reads). Heatmaps depicting relative gene expression among all samples/groups (blue-red) used the following formula for each cell: (FPKMaverage FPKM across all samples all groups)/standard deviation of FPKM across all samples all groups. Heatmaps depicting change in (Δ) gene expression between paired samples (gray-orange) used the formula: (FPKM dmPGE2 -FPKM vehicle )/standard deviation of FPKM across all samples in age group, where superscripts denote different treatments of the same mouse BM sample.

Statistics
Other statistical analyses were performed using Microsoft Excel and GraphPad Prism 8. All data with error bars represent mean ± SEM. Paired t-tests were performed between matched vehicle-and dmPGE 2 -treated samples from the same donor in chimerism analyses, 1-tailed for expected increase with dmPGE 2 . Unpaired t-tests were performed between young and old flow cytometry data points (2-tailed).

DmPGE 2 Enhances Long-term Serial Repopulation Capacity of Aged HSCs
Since dmPGE 2 enhances homing, survival and proliferation of long-term repopulating HSCs in young mice [19,20], we tested whether a similar effect could be demonstrated on HSCs from old mice, which are known to demonstrate reduced regenerative potential and myeloid skewed differentiation [1,30]. WBM aliquots from 3 mo (young) or 25 mo (old) mice were pulsed with either dmPGE 2 or vehicle prior to competitive serial transplantation, allowing for matched analysis of dmPGE 2 effects on long-term repopulating HSC potential ( Fig. 1a). PB chimerism in recipients was assessed monthly by flow cytometry for CD45.2 donor cell frequency and for trilineage distribution at 6 mo post-transplant (Fig. 1b). In primary transplants, dmPGE 2 pulse increased long-term donor-derived chimerism for all 4 young donors and 3 of 4 old donors (Fig. 1c). After secondary transplantation, longterm repopulation was increased by dmPGE 2 pulse for all donors of both age groups (Fig. 1d). Chimerism of aged grafts at 6 mo post-secondary transplant was increased an average of 27 %, similar to the average increase of 21 % for young ( Fig. 1e), indicating enhanced HSC self-renewal capacity for both young and aged BM grafts by dmPGE 2 .

Lineage Reconstitution
Analysis of PB lineage reconstitution showed that pulse with dmPGE 2 increased myeloid frequency in all four young grafts, observed in both primary and secondary transplantations ( Fig. 1c, d, dark green). However, this was not at the expense of lymphoid production; total lymphoid and myeloid production were both increased by dmPGE 2 pulse exposure in comparison to competitor cells as a frequency of total peripheral CD45 cells (CD45.1 + CD45.2) (Fig. 1f). Transplantation of grafts from old mice resulted in myeloid-skewed PB reconstitution as expected, regardless of vehicle or dmPGE 2 treatment (Fig. 1c, d, dark green). However, similar to young, both lymphoid and myeloid production were increased by dmPGE 2 in comparison to competitor cells (Fig. 1f). One old graft lost long-term myeloid production after secondary transplantation, with poor overall chimerism for both dmPGE 2 and vehicle (Old 4, Fig. 1d), but still showed increased lymphoid potential with dmPGE 2 (Fig. 1f).

Donor BM Phenotypic HSC (pHSC) Frequencies and EP4 Expression Pre-transplant
Prior to pulse and transplantation, aliquots of WBM cells from young and old donors were analyzed for baseline pHSC frequency and dmPGE 2 receptor expression (Fig. 2a). The pHSC population in old mice was greatly expanded compared to that in young mice, as has been extensively described [1,2,7], exhibiting 24-fold higher frequency (Fig. 2b) similar to our previously published data [23]. EP4, the receptor primarily implicated in HSC functional responses to dmPGE 2 [31][32][33], was strongly expressed on the surface of aged HSCs and was higher compared to young (Fig. 2c).

BM Chimerism and Stem/progenitor Frequencies Post-transplant
Donor-derived BM chimerism was assessed at the time of secondary transplantation (7 mo post-primary transplant). CD45.2 BM chimerism was significantly higher after dmPGE 2 pulse for all grafts from young donors, and was variably increased after dmPGE 2 pulse for 3 of 4 grafts from old donors (Fig. 2d), mirroring the effect on primary PB chimerism. The single old donor without increased BM chimerism also did not show increased PB chimerism after primary transplantation ("Old 1", Fig. 1c). However, the subsequent superiority of the dmPGE 2 -treated graft in secondary transplantation ("Old 1", Fig. 1d) suggests it may retain a dmPGE 2 -mediated qualitative advantage in HSC selfrenewal revealed by the stress of serial transplantation.
While the aged grafts contained 24-fold higher pHSC frequency compared to young grafts prior to pulse and transplantation (Fig. 2b), similar pHSC frequencies were found among

Shared and Divergent Transcriptional Responses to DmPGE 2 in Young and Old HSCs
To identify molecular correlates of enhanced long-term HSC function by dmPGE 2 in both young and aged HSCs, and to identify signaling differences in HSCs with age, pHSCs from pulsed BM samples were immediately isolated by FACS and analyzed using RNA-seq (Fig. 3a). Differentially expressed genes were defined as FDR < 0.05 by pair-wise analysis between dmPGE 2 -and vehicle-pulsed samples from the same mouse, across 4 young samples or 4 old samples to determine age-specific transcriptional effects. The paired analyses allow for robust detection of dmPGE 2 effects despite baseline gene expression differences between individual mice, which can become particularly variable with advanced age. DmPGE 2 significantly affected 230 genes in young HSCs (184 increased, 46 decreased), and 112 genes in old HSCs (85 increased, 27 decreased). Of these, 53 common genes reached significance in both age groups (49 increased, 4 decreased; Fig. S1). Substantially fewer genes reached FDR < 0.05 in old HSCs, which may be due in part to greater variability in responsiveness among old mice, but may also reflect an overall decrease in HSC responses to dmPGE 2 with age. All differentially expressed genes reaching FDR < 0.05 in both age groups combined were clustered based on similar (ageindependent) or unique (age-dependent) dmPGE 2 effects between young and old, and were visualized both for relative expression across all samples (Fig. 3b) and for individually paired Δ gene expression induced by dmPGE 2 (Fig. 3c). The top two clusters in Fig. 3b group and < 0.6 in the other), and comprised a majority of the genes. Since dmPGE 2 pulse effectively enhanced HSC function in serial transplantation for both age groups, we first focused analysis on shared gene effects to narrow the pathways likely involved in the mechanism of enhancement. The lower clusters represent genes affected differently by dmPGE 2 in old and young HSCs; these are not likely involved in the mechanism of enhancement but can shed light on changes in HSC signaling pathways with age.

Shared DmPGE 2 Signaling in Young and Old HSCs
IPA was used to identify the most likely upstream regulators activated by dmPGE 2 in HSCs based on all downstream gene expression changes (FDR < 0.05 and log 2 FC>|0.5|) in young and old mice, and a comparison analysis of results from each age group revealed the same top regulators predicted in young and old (Fig. 4a). The regulator with the highest activation zscore in both age groups was CREB1, a known mediator of PGE 2 signaling through receptors EP2 and EP4 [34], supporting similar HSC-intrinsic signaling in old and young. Among the CREB1-regulated genes contributing to each zscore, 16 were shared between young and old (Fig. 4b).
Cellular functions activated by dmPGE 2 were also predicted, with 'Cell survival' and 'Cellular homeostasis' reaching significant activation z-scores in both young and old HSCs (Fig. 4c). Overlapping gene sets contributed to both functions, and 18 of these genes were shared between young and old (Fig. 4d). Seven of these genes (bold) are also known to be regulated by CREB1 (Fig. 4b). Thus, CREB1 activation by dmPGE 2 may be increasing HSC survival and homeostasis through Bhlhe40, Cdkn1a, Cebpb, Gadd45b, Nr4a2, Pim3, and Vegfa, among others in these heatmaps potentially not yet functionally linked. Overall, these genes and predicted regulators (Fig. 4a)  Since dmPGE 2 activates CREB1 through either EP2 or EP4, baseline mRNA expression of each receptor was compared (Fig. 4e). EP2 was not detectable above background in young or old HSCs while EP4 was highly expressed and increased with age, confirming the flow cytometry findings for EP4 (Fig. 2c). Also, dmPGE 2 pulse exposure decreased EP4 expression in each of the old HSC samples and 3 of 4 young samples (Fig. 4f). Desensitization through EP4 has been noted [35], and this negative feedback effect on EP4 expression has been reported in murine HSCs after in vivo dmPGE 2 treatment [36]. Together these findings further support EP4 as the relevant receptor for enhancement of both young and old HSC long-term function.

Divergent DmPGE 2 Signaling in Young and Old HSCs
The third cluster in Fig. 3b/c, and the second largest cluster overall, is comprised of 70 genes significantly increased by dmPGE 2 in young HSCs (FDR < 0.05) but not affected at all in old HSCs (FDR > 0.6). Visualizing relative expression in Fig. 3b, many of these genes appear already elevated at baseline in vehicle-treated old HSCs and are not further increased with dmPGE 2 . These genes were classified for functional annotation enrichments using DAVID bioinformatics analysis ( Table 1). The top enrichment category was 'Phosphoprotein', comprising 44 of the 70 genes. These phosphoproteins also made up 23/29 genes from 'Alternative splicing', and 14/17 genes from 'Transcription', the next two most enriched categories. The category of 'Alternative Splicing' is defined by UniProt as "Protein for which at least two isoforms exist due to distinct pre-mRNA splicing events" (uniprot.org). Since alternative splicing has been implicated in the development  of myeloproliferative disorders which increase in prevalence with age [37,38], these genes are reported in Table 2. Thus, a sizeable subset of dmPGE 2 -induced genes observed in young HSCs become less responsive with age and are enriched for phosphoproteins with alternative splice variants and those involved in transcriptional regulation.
An additional "difference of difference" FDR calculation was utilized to compare the treatment effect in old versus young HSCs and identify genes affected significantly differently by dmPGE 2 in each age group (Fig. 5a). Each of these genes was affected in the opposite direction by dmPGE 2 in old and young HSCs (FDR < 0.05). Several had higher average expression with age in the vehicle-treated samples and were decreased by dmPGE 2 in old HSCs, as opposed to being increased by dmPGE 2 in young HSCs, including Ccbe1, Evc, Mlk2, Mycbp2, Rorb, and Ubr4. Since Rorb was so strongly upregulated with age and differentially affected by dmPGE 2 , its close family members Rora and Rorc were also examined (Fig. 5b). Rora was slightly increased with age and strongly upregulated by dmPGE 2 in both young and old HSC, while Rorc was strongly increased with age and slightly elevated by dmPGE 2 in old but not in young. In addition, two genes (H2afx and Kcnj5) had lower average expression with age and were increased by dmPGE 2 in old HSCs, opposite to the dmPGE 2 effect in young, and four others had similar baseline expression in young and old HSCs but opposite treatment effects (Elf2, Max, Psap, and Ncoa2; Fig. 5a). These RNAseq analyses identify potential molecular targets for agerelated HSC dysfunction.

Discussion
After HLA matching, donor age is generally the most critical factor determining survival after HSCT [11]. Grafts from older donors have been associated with increased graft failure even when controlling for cell number [39,40], which may relate to declining inherent stem cell quality [1][2][3][4][5]41]. Increased incidence of graft versus host disease (GVHD) in recipients of grafts from older donors has also been observed [10,42], potentially related to an increase in antigenexperienced lymphocytes with age [43,44] or the general increase in low-level inflammation that characterizes aging [45]. However, neither of these associations show consistent relationships with HSCT outcome [10,46], and a combination of several age-related factors likely contribute. For these reasons, transplant physicians tend to favor younger donors, but finding the appropriately matched donor can sometimes be difficult or impossible. The probability of finding an unrelated matched donor varies with ethnicity, and can be especially problematic for some ethnic groups, e.g. African-Americans [47]. Elderly family members would have strong motivation to donate if suitably matched, and strategies to enhance aged grafts could help increase the possibility of success.
While hematopoietic malignancies treatable by autologous transplant increase in prevalence with age, elderly patients are often not eligible due to the rigors of the HSCT process as well as their own declining HSC quality. However, recent strategies employing reduced-intensity conditioning for elderly patients, as well as advances in transplant technique and supportive care, increasingly enable allogeneic and autologous HSCT in this population [48][49][50]. The ability to augment the function of aged grafts prior to infusion could facilitate successful, lifesaving autologous transplants for older patients. Here we found that ex vivo pulse exposure to dmPGE 2 can enhance the transplantation capacity of murine HSCs of advanced age. Previous work in young grafts has shown this effect to translate from zebrafish and mice [18,19] to non-human primates [51,52] and ultimately to enhancement of human cord blood transplantation [21]. Thus, the current findings strongly support further clinical evaluation. PGE 2 is an eicosanoid synthesized within most body tissues by many different cell types, acting in autocrine or paracrine fashion [53]. Activities mediated by PGE 2 are highly pleiotropic, depending on the tissue/cell type and expression of its four G-protein coupled receptors EP1-4 [35,54]. EP1 signals primarily through PKC and Ca 2+ mobilization, EP2 and EP4 induce cAMP production and subsequent cAMP response element-binding protein (CREB) activation as observed here, and EP3 inhibits cAMP production [35,54]. EP4 has been recognized as a key functional regulator for HSCs, including transplantation studies [32,33]. In the setting of radiation exposure, where dmPGE 2 protects and enhances HSC function [36], only dmPGE 2 or EP4 agonism conferred survival from lethal irradiation [55]. In the current transcriptomic analysis, the predominance of CREB1induced gene expression following dmPGE 2 pulse in both young and old HSCs, along with strong EP4 expression but undetectable EP2 mRNA levels as observed here and previously by RNA-seq of purified murine pHSCs [36], strongly supports EP4 as the relevant receptor mediating HSC enhancement regardless of age.
An interesting finding in this study was increased EP4 expression in HSCs with age. PGE 2 production is known to increase with age in macrophages [56] and decrease with age in gastrointestinal tissues [57]. In skeletal muscle, the capacity for PGE 2 synthesis increases with age while receptor levels are downregulated [58]. It remains unclear if basal PGE 2 levels change with age within the BM, and which factors would drive the increase in EP4 expression on HSCs. The intensity of CREB1-regulated genes was noticeably higher in old HSCs after dmPGE 2 pulse and may be related to the number of EP4 receptors, though higher basal expression of these genes was also seen in control HSCs in a variable manner between aged mice (Fig. 4b). Thus, the change in CREB1regulated genes was greater in some old mice but not in others, and the relevance of increased EP4 expression on aged HSCs remains uncertain. Ultimately, this investigation established that HSCs of advanced age do not lose expression or signaling through the pivotal EP4 receptor. While the primary objective of these studies was not to compare old versus young HSC function, but rather to evaluate the effect of dmPGE 2 on old grafts in parallel with young grafts, the transplantation experiments were performed simultaneously with the same cohorts of recipients and competitor cells. The studies indicated that the old and young grafts functioned similarly in regard to overall long-term and serial chimerism capacity. Since the old grafts contained approximately 24-fold higher pHSC frequency, and equivalent numbers of WBM cells were transplanted, these observations are in line with the reported substantial decrease in function of pHSCs with age [2,3,6,7]. We also observed myeloid-skewed reconstitution from aged HSCs as described [2-4, 6, 30]. Interestingly, dmPGE 2 pulse consistently increased the relative myeloid contribution of the young donor cells, bringing their lineage ratios closer to those of the aged. However, dmPGE 2 also increased the overall frequencies of donor lymphoid cells in comparison to competitors, suggesting dmPGE 2 has a positive effect on both major immune cell branches but augments myeloid reconstitution to a greater degree. DmPGE 2 also augmented both branches for old HSCs compared to competitors without further affecting the inherent myeloid skew with age. Of interest, we have previously reported an increase in the proportion of myeloid cells in PB of mice transplanted with young HSCs pulsed with dmPGE 2 following primary and secondary transplant, however this was not consistent across tertiary and quaternary transplants and was without overall effect on the enhancement of induced stem cell competitiveness [59].
Bioinformatic analysis of dmPGE 2 signaling in young versus old HSCs revealed that the core response pathways remained largely unchanged with age. Several ageindependent genes were identified as both increased by CREB1 signaling and involved in the significantly predicted functions of 'Cell survival' and 'Cellular homeostasis'. Many of these genes additionally have described roles in hematopoiesis, supporting their involvement in HSC modulation by dmPGE 2 . Vegfa encodes for vascular endothelial growth factor A (VEGF-A) which, while first discovered for its primary role in angiogenesis [60], enhances human HPC formation [61], promotes hematopoietic cell generation from embryonic stem cells of both mouse [62] and human [63], and regulates HSC survival [64]. Cebpb is the gene for CCAAT enhancer binding protein beta (C/EBPβ), a transcription factor that promotes lymphopoiesis [65] and emergency myelopoiesis [66,67] at the level of stem and progenitor cell regulation [68,69]. Gadd45b, encoding growth arrest and DNA-damageinducible beta (GADD45β), appears essential for DNA damage protection and survival of HSCs/HPCs and induced pluripotent stem cells (iPSCs) under stress [70]. Cdkn1a encodes the cyclin-dependent kinase inhibitor P21, which preserves HSC quiescence under stress and promotes HSC selfrenewal in serial transplantation [71], while Nr4a2 encoding the transcription factor nuclear receptor subfamily 4 group A member 2, also known as NURR1, also attenuates HSC cycling [72] and may contribute to the maintenance of stem cell quiescence during the stress of transplantation.
In addition to transplantation, steady-state hematopoiesis in older humans is subject to increased bone marrow failure and decreased hematologic tolerance of cytotoxic injury, as well as the increased propensity for myeloproliferative disorders and cancerous transformation [8,9]. A broader understanding of aged HSC function is essential to development of novel treatments for hematopoietic compromise in the elderly. The highthroughput genomic comparison of old and young HSC responses to dmPGE 2 provided a unique modality for investigating changes in HSC stimulation response pathways with age. Several signaling alterations identified here may have relevance for targeting in treatment of age-induced HSC defects.
Genes differentially affected by dmPGE 2 in young and old HSCs included numerous phosphoproteins induced in young but not in old, many of which were already elevated with age. IPA did not return any significant predictions for a common upstream regulator controlling these genes (no more than 5 had shared association with any given regulator), but functional categories including 'Alternative splicing' exhibited significant enrichment (Table 1). Abnormalities in alternative splicing have been implicated in the development of myeloproliferative disorders that increase in prevalence with age, with over 50 % of myelodysplastic syndromes harboring spliceosome factor mutations in the dominant clone [37,38]. The current analysis suggests dmPGE 2 -responsive genes that become less responsive with age tend to be those with alternative splice variants, and tend to have higher mRNA levels detectable in old HSCs pre-stimulation (Table 2). However, the current experimental design did not distinguish between splice variants, and further investigation is needed to determine whether these transcripts could be affected by dysregulated splicing in HSCs of advanced age.
Several specific genes were identified as significantly oppositely affected by dmPGE 2 stimulation in old versus young HSCs, revealing divergent molecular responses potentially related to aging defects. Ccbe1 and Rorb were particularly elevated with age and strongly decreased by dmPGE 2 only in old HSCs. Ccbe1, encoding for collagen and calcium binding EGF domains 1 (CCBE1), is a secreted protein thought to function in remodeling of extracellular matrix and c e l l m i g r a t i o n , a n d i s a n i m p o r t a n t f a c t o r i n lymphangiogenesis [73][74][75]. It is also an essential mediator of erythroblastic island formation for erythropoiesis in fetal liver, though it is not required for postnatal erythropoiesis [76]. This gene has otherwise not been associated with hematopoiesis, and gene expression levels found here in young HSCs were near-zero at baseline with a very slight elevation by dmPGE 2 . However, the Ccbe1 transcript was much more detectable in aged HSCs and was strongly and consistently downregulated by dmPGE 2 stimulation. Thus, transcription of Ccbe1 appears to be 'turned on' in HSCs by an unknown aging factor that may be sensitive to 'turning back off' by dmPGE 2 signaling. However, a potential role for this protein in aged HSCs remains to be explored.
A more substantial target may be Rorb, which encodes for RAR related orphan receptor B (RORβ), a member of the highly conserved ROR family of receptor tyrosine kinases [77]. These kinases, including RORβ, are known to negatively regulate WNT/B-catenin signaling [78,79], an important facilitator of HSC fate decisions [80]. In the context of dmPGE 2 stimulation, dmPGE 2 enhanced WNT signaling during zebrafish embryogenesis and was required for WNTmediated regulation of HSC development [81]. In addition, R O R β is el ev ate d with a ge i n m a rro w-d er i ve d osteoprogenitor cells, contributing to development of osteoporosis [79,82]. Our study reveals that RORβ is elevated with age in HSCs, and decreased in response to dmPGE 2 in an agedependent manner. Rora and Rorc also exhibited unique expression patterns affected by both age and dmPGE 2 treatment (Fig. 5b). Together, these findings may indicate a novel mechanism of age-associated dysregulation of HSC fate decisions through increased ROR expression, and reveal an intriguing avenue for downregulation of RORβ in aged HSCs through the PGE 2 signaling pathway.
We previously reported that dmPGE 2 pulse of bone marrow cells increases HSC Cxcr4 expression, homing, and engraftment, and the proportion of homed HSCs in S/G 2 /M phase post-transplant [19,20,83]. Enhanced homing and engraftment required dmPGE 2 stabilization of HIF1a without affecting mRNA and was associated with increased HIF1a transcriptional activity, with an increase in the downstream genes adrenomedullin (Adr) and glucose transporter-1 (Slc2a1) [84]. In HIF1a deficient mice, HSCs have been reported to lose cell cycle quiescence with impaired transplant capacity [85] and administration of HIF1a stabilizers in vivo induces HSC quiescence and enhances recovery from sublethal irradiation [86]. We recently showed that administration of dmPGE 2 prior to lethal irradiation results in inhibition of HSC cell cycle [36] in agreement with these reports. These in vivo studies are in contrast however to the dmPGE 2 in vitro pulse studies that show enhanced HSC proliferation, and caution against extrapolation between effects of dmPGE 2 manifested on isolated HSCs compared to those manifested within the hematopoietic niches in intact animals.
Cell cycle and homing markers were not evaluated in this report; rather the focus was on evaluating whether dmPGE 2 pulse enhances long-term repopulation by aged HSCs, and whether HSC transcriptional signaling early after dmPGE 2 pulse is altered with age. We did not see increased Cxcr4 expression in the current RNA-seq analysis, likely due to timing; our previous reports showed increased CXCR4 on HSCs when measured 24 h post dmPGE 2 pulse, while the current RNA-seq data was derived 1 h after exposure. Moreover, in bone marrow cells from young mice pulsed with dmPGE 2 , upregulation of Cxcr4 mRNA is not observed before 2-3 h after a 2 h pulse (unpublished). HIF1a stabilization was also not directly evaluated in aged HSCs, however a trend in increased Adr and Slc2a1 genes was noted in 3 of 4 young and 3 of 4 aged HSCs (not shown), similar to our findings in young HSCs [84]. Furthermore, administration of dmPGE 2 to aged mice in vivo results in increased expression of CXCR4 on marrow cells and increased homing and engraftment compared to vehicle controls (unpublished). The similarity in some key parameters lends confidence that some functional mechanisms may be shared. Future comparison of the responses of young versus aged HSCs are warranted, including biological evaluation as well as genetic studies based upon the new RNA-seq findings.
In conclusion, this study has identified that aged HSCs primarily retain the molecular capacity to respond to dmPGE 2 pulse exposure and initiate transcriptional programs enhancing survival and long-term repopulating function, which has potential importance toward the goal of enhancing aged human grafts for transplantation. Moreover, age-related alterations in HSC signaling in response to PGE 2 were identified as potential targets for treatment of age-related defects.