Pluripotency and immunomodulatory signatures of canine induced pluripotent stem cell-derived mesenchymal stromal cells are similar to harvested mesenchymal stromal cells

With a view towards harnessing the therapeutic potential of canine mesenchymal stromal cells (cMSCs) as modulators of inflammation and the immune response, and to avoid the issues of the variable quality and quantity of harvested cMSCs, we examined the immunomodulatory properties of cMSCs derived from canine induced pluripotent stem cells (ciMSCs), and compared them to cMSCs harvested from adipose tissue (cAT-MSC) and bone marrow (cBM-MSC). A combination of deep sequencing and quantitative RT-PCR of the ciMSC transcriptome confirmed that ciMSCs express more genes in common with cBM-MSCs and cAT-MSCs than with the ciPSCs from which they were derived. Both ciMSCs and harvested cMSCs express a range of pluripotency factors in common with the ciPSCs including NANOG, POU5F1 (OCT-4), SOX-2, KLF-4, LIN-28A, MYC, LIF, LIFR, and TERT. However, ESRRB and PRDM-14, both factors associated with naïve, rather than primed, pluripotency were expressed only in the ciPSCs. CXCR-4, which is essential for the homing of MSCs to sites of inflammation, is also detectable in ciMSCs, cAT- and cBM-MSCs, but not ciPSCs. ciMSCs constitutively express the immunomodulatory factors iNOS, GAL-9, TGF-β1, PTGER-2α and VEGF, and the pro-inflammatory mediators COX-2, IL-1β and IL-8. When stimulated with the canine pro-inflammatory cytokines tumor necrosis factor-α (cTNF-α), interferon-γ (cIFN-γ), or a combination of both, ciMSCs upregulated their expression of IDO, iNOS, GAL-9, HGF, TGF-β1, PTGER-2α, VEGF, COX-2, IL-1β and IL-8. When co-cultured with mitogen-stimulated lymphocytes, ciMSCs downregulated their expression of iNOS, HGF, TGF-β1 and PTGER-2α, while increasing their expression of COX-2, IDO and IL-1β. Taken together, these findings suggest that ciMSCs possess similar immunomodulatory capabilities as harvested cMSCs and support further investigation into their potential use for the management of canine immune-mediated and inflammatory disorders.


Materials and methods
All methods involving the use of animals and/or animal tissues were carried out in accordance with relevant guidelines and regulations. The collection and use of animal tissues was approved by the Animal Ethics www.nature.com/scientificreports/ increments for 0.05 s. The relative expression ratios of genes were calculated by the Delta Ct method. Dissociation curve analysis was implemented to confirm the specificity of the PCR products.
Deep sequencing of ciMSC, cBM-MSC and ciPSC transcriptomes. RNA was extracted from one line of each of the ciPSCs (Clone A), ciMSCs (derived from Clone A ciPSCs), and cBM-MSCs as described above. 100 base-pair paired-end mRNA sequencing was performed by the Australian Genome Research Facility Ltd (www.agrf.org.au) on an Illumina HiSeq 4000 platform. Primary sequence data underwent demultiplexing, quality control, alignment, transcript assembly, quantification and normalisation, followed by differential expression analysis, as performed by the AGRF. Sequence reads were screened for the presence of any crossspecies contamination and mapped against the canine reference genome CanFam3.1 (GCA_000002285.2) (https ://asia.ensem bl.org/Canis _famil iaris ). Genes were defined as expressed if the CPM ≥ 1. EdgeR was used to generate multidimensional scaling (MDS) plots using both raw gene counts and after normalisation by EdgeR's TMM algorithm to account for the different library sizes for each sample. Both the raw gene count and normalised gene count MDS plots were generated from the data of the 500 most variably expressed genes across all samples. Venn analysis was performed using the Venny tool at http://bioin fogp.cnb.csic.es/tools /venny . Due to financial and logistical constraints only one sample of the ciPSCs and ciMSCs, and cBM-MSCs from one individual, were used for RNA sequencing; therefore, the RNAseq data is indicative of genes that are expressed, but without the number of samples required to perform statistical analyses no comment can be made regarding differential expression between the cell types.
Isolation of leukocytes from canine blood. 40 ml of whole blood was aseptically collected in Vacuette blood collection tubes (InterPath Services, Australia) from two healthy adult mixed-breed dogs at the School of Veterinary Science, University of Queensland. Leukocytes were isolated using the ACCUSPIN System-Histopaque-1077 (Sigma-Aldrich, Australia) according to the manufacturer's instructions.
Enzyme-linked immunoassays. Culture supernatants were used to determine the concentration of TGF-β1, VEGF, IL-8 and IL-1β in the different co-culture groups. Canine-specific Quantikine ELISA kits for TGF-β1 (R&D Systems, USA), IL-8 (R&D Systems), VEGF (R&D Systems) and IL-1β (R&D Systems) were used according to the manufacturer's instructions. Each sample was assayed in triplicate. Plates were analysed with an Infinite M200 (Tecan, Switzerland) microplate reader at the Australian National Fabrication Facility (ANFF, The University of Queensland, Brisbane, Australia).

Statistical analysis.
Results are presented as the mean ± standard error of the mean (SEM). The comparative analysis between treatment groups was conducted using one-way ANOVA and the means were compared with Student's t-test using the GraphPad7 Prism software (San Diego, CA, USA). Significance is defined as: ns = not significant p > 0.05; *p ≤ 0.05; **p ≤ 0.005; ***p ≤ 0.0002; ****p ≤ 0.0001.
Due to financial constraints we were only able to perform RNAseq on one sample from each of the ciMSCs, cBM-MSCs, and ciPSCs; thus, we used quantitative RT-PCR to examine the expression of 10 key genes, identified from the RNAseq as being differentially expressed between the cMSCs and ciPSCs, in 2 independent biological samples from each of the ciMSCs, cAT-MSCs, cBM-MSCs and ciPSCs. Our previous work 42 identified that ciMSCs and cBM-MSCs express the core pluripotency factors POU5F1/OCT-4, SOX-2 and NANOG, which was confirmed by the RNAseq analysis in this study, and so we focused the qRT-PCR analysis on the two genes that are specific for naïve pluripotency: ESRRB and PRDM-14. We also examined the expression of 8 genes that were differentially expressed between the cMSCs and ciPSCs in the RNAseq analysis and which are associated with important pathways in MSCs, specifically: FGF-2, FGF-5, TNFSF-18, TLR-2, TLR-4, TLR-9, CXCR-4 and LOXL-2. www.nature.com/scientificreports/ Supporting the RNAseq data, ESRRB and PRDM-14 are expressed at appreciable levels only by the ciPSCs (Fig. 1e). Similarly, FGF2 expression is significant only within the ciMSCs, cAT-MSCs and cBM-MSCs with barely detectable levels of expression in the ciPSCs (Fig. 1e). Expression of FGF5 is similarly restricted to the harvested cMSCs and ciMSCs, with no detectable expression in the ciPSCs which is in keeping with the results from the RNAseq; however, the expression level in one of the two lines of iMSCs is also very low (Fig. 1e). TNFSF-18 is robustly expressed by the cAT-, cBM-and ciMSCs and not by the ciPSCs, but again, the expression level in one of the two lines of iMSCs is very low (Fig. 1e). Both the ciMSCs and cBM-MSCs express TLR-2 and TLR-9, as seen in the RNAseq analysis (Fig. 1e). In contrast, the cAT-MSCs express barely detectable levels of TLR-2 (Fig. 1e). A surprising finding from the RNAseq data was that neither the ciMSCs nor the cBM-MSCs expressed TLR-4. This observation is supported by the qRT-PCR data, which also demonstrates a lack of TLR-4 expression in the AT-MSCs (Fig. 1e). Significant expression of CXCR-4 is found in ciMSCs, cAT-MSCs and cBM-MSCs but not in ciPSCs (Fig. 1e). Levels of LOXL-2 expression are higher in ciMSCs and cBM-MSCs than in ciPSCs, with the highest levels in the cAT-MSCs (Fig. 1e). Thus, both the RNAseq and qRT-PCR data point to the ciMSCs as being more similar in their transcriptional profiles to AT-and BM-MSCs than to ciPSCs. ciMSCs constitutively express immunomodulatory and anti-inflammatory factors and respond to priming with pro-inflammatory cytokines. ciMSCs constitutively expressed the immunomodulatory factors inducible nitric oxide synthase (iNOS), galectin-9 (GAL-9), transforming growth factor-β1 (TGF-β1), prostaglandin receptor-2α (PTGER-2α) and vascular endothelial growth factor (VEGF), and the pro-inflammatory factors cyclooxygenase-2 (COX-2), interleukin-1β (IL-1β) and interleukin-8 (IL-8) (Fig. 2). cAT-MSCs had a similar constitutive expression profile, although they expressed iNOS and HGF at significantly lower levels (Supplementary Table 2), and VEGF at significantly higher levels (Supplementary Table 2), than ciMSCs (Fig. 2).
Expression of HGF was restricted predominantly to cAT-MSCs across all three treatment groups ( Fig. 3g; Supplementary Table 2). Similarly, VEGF was also most strongly expressed by cAT-MSCs, with significantly lower levels of expression detected in cBM-MSCs and ciMSCs ( Fig. 3h; Supplementary Table 2). While cAT-MSCs and ciMSCs expressed similar levels of IL-8 constitutively, the ciMSCs showed the most increased response to all three treatments, with the strongest response to cTNF-α ( Fig. 3i; Supplementary Table 2). Expression of IL-1β remained relatively unchanged in ciMSCs cultured with cTNF-α, cIFN-γ and cTNF-α/cIFN-γ; similarly, cBM-MSCs maintained consistent levels of expression across all three treatment groups ( Fig. 3j; Supplementary  Table 2). In contrast, cAT-MSCs significantly downregulated their expression compared to constitutive levels ( Fig. 3j; Supplementary Table 2).

Effect of mitogen-stimulated canine lymphocytes on inflammatory cytokine expression of MSCs.
When co-cultured with mitogen-stimulated lymphocytes, ciMSCs significantly downregulated their expression of iNOS, TGF-β1, HGF and PTGER-2α ( Fig. 4; Supplementary Table 3). Although HGF expression levels also significantly decreased, transcription levels in the control cultures were so low that they are likely not indicative of expression ( Fig. 4; Supplementary Table 3). cAT-MSCs downregulated their expression of TGF-β1 and VEGF, while iNOS and PTGER-2α remained unchanged ( Fig. 4; Supplementary Table 3). In response to coculture, ciMSCs upregulated their expression of COX-2 and IDO, and both ciMSCs and cAT-MSCs increased their expression of IL-1β ( Fig. 4; Supplementary Table 3). Expression of GAL-9 and IL-8 increased in cAT-MSCs but remained unchanged in ciMSCs, while expression of VEGF decreased in cAT-MSCs and was unchanged in ciMSCs ( Fig. 4; Supplementary Table 3).  Table 4). In contrast, lymphocytes co-cultured with cAT-MSCs increased their expression of COX-2, www.nature.com/scientificreports/ TGF-β1 and possibly HGF, although expression levels are so low as to be near the detection threshold ( Fig. 5; Supplementary Table 4).

Effects of co-culture on the secretion of factors by lymphocytes and MSCs. The concentrations
of canine IL-1β, IL-8, TGF-β1 and VEGF were measured in the supernatant collected from cultures of lymphocytes, ciMSCs and cAT-MSCs, and co-cultures of lymphocytes with each of ciMSCs and cAT-MSCs. In agreement with the qRT-PCR data, both ciMSCs and cAT-MSCs produce IL-1β, IL-8, TGF-β1 and VEGF ( Fig. 6; Supplementary Table 5). Furthermore, the relative expression levels of the genes between the two types of MSCs is reflected at the protein level with VEGF RNA and protein expression significantly higher in cAT-MSCs as   Table 5). Lymphocytes similarly produce all four factors ( Fig. 6; Supplementary Table 5). Based on the qRT-PCR data that showed lymphocytes did not alter their transcription of IL-1β in response to co-culture with either ciMSCs or cAT-MSCs, but both types of MSCs increased their transcription of IL-1β when co-cultured, the increase in IL-1β measured in the medium from co-cultures is likely produced by the ciMSCs and cAT-MSCs rather than the lymphocytes ( Fig. 6; Supplementary Table 5). In contrast, based on the qRT-PCR data, the increase in IL-8 in co-cultures is more likely from the cAT-MSCs and lymphocytes than from the ciMSCs ( Fig. 6; Supplementary Table 5).
Both ciMSCs and cAT-MSCs downregulated their expression of TGF-β1 when co-cultured, while lymphocyte expression, which was lower than that observed in the MSCs, increased or remained unchanged, when cocultured with cAT-MSCs and ciMSCs, respectively. Thus, lower levels of TGF-β1 were measured in the medium of co-cultured ciMSCs and cAT-MSCs than when the cells were cultured alone, and are similar to the levels detected in medium from lymphocyte cultures ( Fig. 6; Supplementary Table 5).
Both cAT-MSCs and lymphocytes expressed significantly higher levels of VEGF than ciMSCs and downregulated their expression in co-culture. This dynamic is reflected at the protein level where cAT-MSC/lymphocyte co-cultures have VEGF levels in between the levels for each when cultured separately, and the measurement for ciMSC/lymphocyte co-cultures are higher than the ciMSCs cultured alone but lower than the levels measured for lymphocytes or cAT-MSC/lymphocyte co-cultures ( Fig. 6; Supplementary Table 5).

Discussion
In this study we compared the transcriptome of ciMSCs with cAT-MSCs, cBM-MSCs and ciPSCs and show expression of key pluripotency factors by all cell types. Previous studies have similarly demonstrated the expression of pluripotency factors by canine MSCs isolated from adipose tissue 45 , bone marrow 42,45 and amniotic fluid 46 . In contrast, ESRRB and PRDM-14, bother than primed, pluripotency [47][48][49][50] are expressed only in the ciPSCs and not the ciMSCs, cAT-MSCs or cBM-MSCs, which is not surprising since the ciPSCs are pluripotent 42,51 while all three types of MSCs are multipotent 42 . Also unique to the ciPSCs is the expression of FGF-4 which, in the mouse embryo, is secreted by the epiblast cells of the inner cell mass (ICM) under transcriptional regulation by Oct-4 and Sox-2 52 where it is thought to play a role in the development of the embryo through the conversion of the ICM into primitive endoderm 53,54 .
Endogenous and exogenously administered MSCs migrate towards tumours and sites of ischaemia and inflammation in response to a range of signalling molecules including the chemokine stromal cell-derived factor-1 (SDF-1), through interaction with its cognate receptor CXC chemokine receptor 4 (CXCR-4), which is expressed on the surface of MSCs [55][56][57][58] . Importantly, when considering future therapeutic applications, our ciMSCs express CXCR-4, as do the cAT-MSCs and cBM-MSCs, while it is not expressed by the ciPSCs. This lack of TLR-4 expression is very surprising since TLR-4 signalling is responsible for priming human MSCs towards a proinflammatory phenotype, while TLR-3 priming induces an anti-inflammatory response 59,64 . Based on limited studies of various canine cell types (not including MSCs) the expression of TLR-4 in the dog appears to follow a similar profile to that described for other species 68 and so we could reasonably expect canine MSCs to similarly express high levels of TLR-4. A search of the literature did not yield any insight as to a possible explanation for the lack of TLR-4 expression in our canine MSCs, except to note that the expression of TLR-4 by human Wharton's jelly-derived MSCs appears to be variable 59,69 and so the lack of TLR-4 expression in our canine MSCs may reflect a species difference or perhaps an effect of culture conditions. The transcriptome of our ciMSCs is more similar to that of the cBM-MSCs and cAT-MSCs than that of the ciPSCs. This is in contrast to the data of Chow et al. 70 whose ciPSC-derived MSCs showed a gene expression profile that was markedly different from that of cAT-MSCs and cBM-MSCs, and much more closely resembled that of the ciPSCs from which they were generated. It is possible that the ciPSCs generated by Chow and colleagues 70 were in a more primed, rather than naïve, state of pluripotency and that this has affected the nature of the resultant ciPSC-derived MSCs. It is perhaps significant that the ciPSCs that we used to generate our ciMSCs show many of the hallmarks of naïve pluripotency including expression of ESRRB and PRDM-14.
MSC secretion of either IDO or iNOS, depending on the species, has been shown to suppress T cell proliferation 31,[71][72][73][74] . In human, IDO is the key mediator of T cell suppression 31,75-79 while in mouse 78 and horse 80 iNOS is the major inhibitor of T cell activation. However, recent reports suggest that IDO, in addition to iNOS, may be involved in the immunomodulatory roles of equine MSCs 35,37,38 . In this study, both ciMSCs and cAT-MSCs constitutively express iNOS and when co-stimulated with cTNF-α and cIFN-γ, ciMSCs upregulated their expression of iNOS by tenfold. That cAT-MSCs did not show an increase in iNOS expression beyond constitutive levels, and cBM-MSCs expressed very low levels in response to cTNF-α/cIFN-γ, is in keeping with the observations by Chow et al. 70 that cAT-MSCs and cBM-MSCs do not employ the iNOS/NO-mediated pathway for immunosuppression. In contrast, the strong upregulation of iNOS expression in ciMSCs is similar to observations in the horse where priming of equine bone marrow-derived MSCs with IFN-γ or TNF-α/IFN-γ similarly induced an upregulation of iNOS 81 . Expression of iNOS significantly decreased in ciMSCs co-cultured with mitogen-stimulated lymphocytes. This would appear to be at odds with our observation of an upregulation of iNOS in ciMSCs exposed to cIFN-γ/cTNF-α. However, previous studies have demonstrated that the production of TNF-α by canine lymphocytes is reduced upon co-culture with cAT-MSCs 34 , and the secretion of IFN-γ by canine lymphocytes is similarly suppressed when co-cultured with cAT-MSCs and cBM-MSCs 82 . Thus, the decrease in iNOS expression by ciMSCs co-cultured with lymphocytes may be due to low levels of TNF-α and IFN-γ being produced by the canine lymphocytes, possibly as a consequence of suppression by the ciMSCs.
All three types of MSCs responded to stimulation with cTNF-α/cIFN-γ by upregulating their expression of IDO. Kang et al. 34 similarly observed increased expression of IDO in canine AT.MSCs co-cultured with Figure 6. Effects of co-culture on the secretion of factors by lymphocytes, ciMSCs and cAT-MSCs. Lymphocytes, ciMSCs and cAT-MSCs produce IL-1β, IL-8, TGF-β1 and VEGF. Based on mRNA levels (see Figs. 5 and 6), the increase in IL-1β measured in the medium from co-cultures of cMSCs and lymphocytes is likely produced by the ciMSCs and cAT-MSCs rather than the lymphocytes. When similarly referenced to mRNA levels, cAT-MSCs and lymphocytes in co-culture upregulate their expression of IL-8 while ciMSCs do not. Significance is defined as: *p ≤ 0.05; **p ≤ 0.005; ***p ≤ 0.0002; ****p ≤ 0.0001. www.nature.com/scientificreports/ concanavalin-stimulated lymphocytes shown to be secreting cTNF-α and cIFN-γ. In our study, while ciMSCs significantly upregulated their expression of IDO when co-cultured with concanavalin-stimulated lymphocytes, the transcript levels of IDO decreased in co-cultured cAT-MSCs. This discrepancy between our cAT-MSC data and that of Kang et al. 34 might reflect insufficient levels of IFN-γ and TNF-α being produced by the lymphocytes to stimulate the AT.MSCs, as discussed in the preceding paragraph. Following TLR-3 priming, the release of TGF-β1 by activated anti-inflammatory MSCs suppresses the proliferation and secretion of cytokines by T lymphocytes and natural killer cells and also inhibits the stimulatory effect of dendritic cells on T lymphocytes 22,25,[82][83][84][85][86][87][88] . Constitutive expression of TGF-β1 by ciMSCs, cAT-MSCs and cBM-MSCs (RNAseq data) is in keeping with the data of other studies 34,82,89 that have similarly demonstrated the constitutive transcription of TGF-β1 in cBM-MSCs, cAT-MSCs and ciMSCs, respectively. While cAT-MSCs showed a stronger transcriptional response to cTNF-α and cIFN-γ than ciMSCs, both types of MSCs expressed similar levels of TGF-β1 mRNA when cultured with combined cTNF-α/cIFN-γ.
IL-8 is an MSC-derived chemokine released at the site of injury to enhance the migration and activation of neutrophils 90,91 . In this study, cAT-MSCs and ciMSCs expressed similar levels of IL-8 constitutively. The constitutive transcription of IL-8 has previously been described in canine AT.MSCs and human BM.MSCs 92 . ciMSCs showed the strongest response to all three treatments, particularly to cTNF-α. The induced upregulation of IL-8 by inflammatory stimuli has also been reported in human 93 and equine MSCs 81,94 .

Conclusion
In both their transcriptome and in their functional responses to inflammatory cytokines and mitogen-stimulated lymphocytes, our ciMSCs are highly similar to harvested MSCs, supporting further investigation into their potential therapeutic applications for immune-mediated and inflammatory conditions in the dog.

Data availability
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.