Co-Culture of Primary Human Coronary Artery and Internal Thoracic Artery Endothelial Cells Results in Mutually Beneficial Paracrine Interactions

Although saphenous veins (SVs) are commonly used as conduits for coronary artery bypass grafting (CABG), internal thoracic artery (ITA) grafts have significantly higher long-term patency. As SVs and ITA endothelial cells (ECs) have a considerable level of heterogeneity, we suggested that synergistic paracrine interactions between CA and ITA ECs (HCAECs and HITAECs, respectively) may explain the increased resistance of ITA grafts and adjacent CAs to atherosclerosis and restenosis. In this study, we measured the gene and protein expression of the molecules responsible for endothelial homeostasis, pro-inflammatory response, and endothelial-to-mesenchymal transition in HCAECs co-cultured with either HITAECs or SV ECs (HSaVECs) for an ascending duration. Upon the co-culture, HCAECs and HITAECs showed augmented expression of endothelial nitric oxide synthase (eNOS) and reduced expression of endothelial-to-mesenchymal transition transcription factors Snail and Slug when compared to the HCAEC–HSaVEC model. HCAECs co-cultured with HITAECs demonstrated an upregulation of HES1, a master regulator of arterial specification, of which the expression was also exclusively induced in HSaVECs co-cultured with HCAECs, suggestive of their arterialisation. In addition, co-culture of HCAECs and HITAECs promoted the release of pro-angiogenic molecules. To conclude, co-culture of HCAECs and HITAECs results in reciprocal and beneficial paracrine interactions that might contribute to the better performance of ITA grafts upon CABG.


Introduction
Coronary artery bypass grafting (CABG) represents the most frequent surgical intervention for the treatment of coronary artery disease [1][2][3][4][5]. Conduits for CABG include autologous veins (e.g., saphenous vein, SV) and arteries (e.g., internal thoracic artery, ITA) [2,3]. Currently, the left ITA is considered the most appropriate vessel to be used first, yet patients often require multiple or composite grafts for complete revascularisation [2,3,6,7]. The selection of the second and further conduits for multiple grafting, as well as those to be included into the composite graft, is a matter of debate [2,3,6,7].

Results
To investigate whether HITAECs, but not HSaVECs, induce advantageous paracrine effects for HCAECs, and whether each of these intercellular interactions has a distinct pattern of detrimental and beneficial effects, we applied a co-culture model where HCAECs were settled at the bottom of a cell culture dish while conduit ECs (HITAECs or HSaVECs) were located at the surface of a 10-µm-thick translucent polycarbonate membrane insert. The pore diameter of this insert (0.4 µm) permitted the communication of co-cultured cells by means of soluble factors and extracellular vesicles while excluding cell migration. Therefore, cell lines were reliably separated, yet still constantly affecting the behaviour of each other through paracrine signals. Sequential time points of the supernatant collection with subsequent RNA/protein extraction after the cell lysis allowed for the assessment of the temporal expression patterns in HCAECs from both of the co-culture models and also in HITAECs/HSaVECs. Monocultures of HCAECs, HITAECs, and HSaVECs were used as a kind of control.
We first investigated whether co-culturing with HITAECs or HSaVECs prevents or contributes to the development of endothelial dysfunction in HCAECs by measuring the relative levels of eNOS (NOS3 gene), pro-inflammatory cytokines IL-6 (IL6 gene), and IL-8 (CXCL8 gene), molecules which are responsible for the monocyte/lymphocyte attachment to ECs (VCAM1/VCAM1/, ICAM1/ICAM1, and E-selectin (SELE/SELE)), EndoMT transcription factors (Snail/SNAI1, Slug/SNAI2, Twist1/TWIST1, and Zeb1/ZEB1), EndoMT markers (endothelial marker VE-cadherin/CDH5 and mesenchymal marker N-cadherin/CDH2), and arterial specification transcription factors (HEY1 and HES1). Selective gene expression profiling by means of reverse transcription-quantitative polymerase chain reaction (RT-qPCR) found that pro-inflammatory and EndoMT transcripts were underrepresented in HCAECs upon co-culturing with HITAECs at different time points as compared to the HCAEC-HSaVEC co-culture model ( Figure 1A). For instance, IL6 and CXCL8 genes were downregulated in HCAECs at both 24 and 48 h of co-culture with HITAECs, whereas VCAM1, ICAM1, SELE, SNAI1, and SNAI2 genes were differentially expressed at these time points ( Figure 1A). Downregulation of SNAI1 and SNAI2 genes was noted at 6 and 24 h, followed by a decrease in the expression of VCAM1, ICAM1, and SELE genes, which encode the respective cell adhesion molecules ( Figure 1A). No beneficial or detrimental pattern was identified when comparing the abundance of abovementioned transcripts in HCAECs from both co-culture models and HCAECs cultured separately ( Figure 1A).
Measurement of the corresponding proteins by Western blotting ( Figure 1B) found an increase in the eNOS level in HCAECs co-cultured with HITAECs when compared to the HCAEC-HSaVEC model upon 48 h of co-culture ( Figure 1B,C). Unexpectedly, the level of VCAM1 was also augmented, yet still being lower in both co-cultures than in the monocultures ( Figure 1B,C). The level of Snail and Slug, however, was steadily reduced in HCAECs co-cultured with HITAECs, corroborating the results of the gene expression analysis ( Figure 1B,C). Together with the upregulated transcription factor of arterial specification HES1, this might indicate that HITAECs supply HCAECs with supportive arterial differentiation cues, whereas HSaVECs deregulate the maintenance of arterial differentiation in HCAECs through paracrine signaling. Notably, both CD31 and vascular endothelial (VE) cadherin expression remained unaffected, suggesting that the alterations of endothelial identity occur far beyond the investigated time points.
We next assessed whether HCAECs differentially affect HITAECs and HSaVECs during the co-culture. Expression of pro-inflammatory genes (CXCL8, ICAM1, and SELE) was higher in HITAECs than in HSaVECs, whereas EndoMT transcription factors SNAI2 and TWIST1 were downregulated, along with the CDH2 gene encoding N-cadherin ( Figure 2A). In contrast, the CDH5 gene encoding VE-cadherin was overexpressed in HITAECs ( Figure 2A). Having compared HITAECs and HSaVECs co-cultured with HCAECs with the respective monocultures, we found that co-culture with HCAECs repressed the expression of the VCAM1, ICAM1, and SELE genes in HITAECs and HSaVECs, although simultaneously promoting IL6 and CXCL8 gene expression in HITAECs (Figure 2A). Interestingly, the monoculture of HITAECs showed a higher expression of cell adhesion molecule genes (VCAM1, ICAM1, and SELE) and a lower expression of EndoMT markers (SNAI2, ZEB1, and CDH2 genes) than HSaVECs (Figure 2A). RT-qPCR profiling, HIT signifies the ratio of transcript levels (measured as ∆Ct) in HCAECs co-cultured with HITAECs to those in HCAECs co-cultured with HSaVECs. HIT (Ctrl) and HSV (Ctrl) represent the ratios of transcript levels (measured as ∆Ct) in HCAECs co-cultured with HITAECs or HSaVECs to those in HCAEC monoculture. Results are represented as the heat map; green, gray, and red colours indicate fold change ≤0.50, 0.51-1.99, and ≥2.00, respectively; (B) Western blotting measurements. HCA-HIT represents HCAECs co-cultured with HITAECs, HCA-HSV signifies HCAECs co-cultured with HSaVECs. HCA and HSV signify HCAEC and HSaVEC monocultures, respectively; (C) Semi-quantitative analysis of Western blotting results by densitometry. HIT represents the ratio of band density in HCAECs co-cultured with HITAECs to that in HCAECs co-cultured with HSaVECs. HIT (Ctrl) and HSV (Ctrl) abbreviate the ratios of band density in HCAECs co-cultured with HITAECs or HSaVECs to that in HCAEC monoculture. HCA (Ctrl) abbreviates the ratio of band density in HCAEC monoculture to that in HSaVEC monoculture. Snail + Slug are shown at both 56 and 29 kDa values. Results are represented as green, gray, and red colours on the the heat map, indicating fold changes of ≤0.75, 0.76-1.24, and ≥1.25, respectively. Similar to HCAECs, HITAECs from the arterial endothelial co-culture model were characterised by an upregulation of eNOS and VCAM1 proteins ( Figure 2B,C). Strikingly, co-culture with HCAECs abrogated the expression of Snail and Slug in HITAECs and endowed HSaVECs with HES1 expression, suggestive of HSaVEC arterialisation during the co-culture with HCAECs ( Figure 2B,C). In addition, the expression of VE-cadherin in HSaVECs also increased upon the co-culture with HCAECs ( Figure 2B,C). In accord with earlier reports [24], HITAECs demonstrated higher eNOS expression than HSaVECs and also had a higher VE-cadherin/N-cadherin ratio ( Figure 2B,C).
Taken together, patterns of molecular response in both co-culture models significantly changed over time but indicated mutual benefits of the paracrine interactions between HCAECs and HITAECs. Notable discrepancies between the expression of the measured molecules at the gene and protein level suggested significant involvement of post-transcriptional regulation. Finally, we evaluated the changes in the EC secretome in different co-culture setups. Pro-angiogenic proteins tended to be upregulated at 48 h in the HCAEC/HITAEC supernatant, in comparison with HCAEC/HSaVEC supernatant, although this was not the case at the 6-h time point ( Figure 3A). Notably, the level of the major pro-inflammatory cytokine interleukin-8 in the supernatant was comparable across the co-culture models, although being slightly higher in the HCAEC/HSaVEC supernatant at the 6-h time point ( Figure 3B). However, a more sensitive enzyme-linked immunosorbent assay identified increased IL-6 and IL-8 content in the supernatant from the HCAEC-HITAEC co-culture model ( Figure 3C). Collectively, our results support the hypothesis regarding the beneficial, although time-dependent, effects of HCAEC and HITAEC co-culturing. In other words, HCAECs and HITAECs reciprocally induce a favourable expression pattern.

Discussion
Despite numerous clinical studies [7][8][9][14][15][16] having been performed and many narrative [2,3,12] and systematic reviews [6,13] having been published, there is no clear consensus regarding the choice of the second-best graft for CABG surgery. Among all vessels, right ITAs and SVs are by far the most frequently used in this regard. Although the majority of CABG interventions rely on the SV because of its relatively easy harvest, resistance to manipulation and vasospasm, shorter operative time, and lower risk of iatrogenic complications, which collectively lead to better short-term results, an increasing number of reports indicates the advantages of bilateral ITA grafting, which is associated with significantly higher long-term patency [13][14][15][16]. The advantages of ITA include its superior mechanical properties, as it shares most of the histoarchitecture features of CAs, including multiple layers of vascular smooth muscle cells enclosed by two elastic membranes [11]. However, the diameter of the SV (4-5 mm) [42][43][44] is generally similar to the left main CA (4-5 mm) or proximal left anterior descending (LAD) CA (4 mm) [45,46] in contrast to ITA (2-3 mm) [44,[47][48][49], although the diameter of the latter is identical to the distal LAD artery (2 mm) [45]. The diameter of the CA, however, may vary from 2 to 5 mm at all segments [46] and therefore individual anatomical features of both the CA and candidate conduits should be taken into account to select an optimal graft for CABG surgery.
Although the short-term benefits of the SV conduits are clearly associated with their convenience for surgical handling, the long-term advantages of ITA grafts are determined by their physiological features, which might include better integration of the HITAEC and HCAEC endothelial layers. In other words, the interactions between HCAECs and HITAECs may sustain coronary homeostasis, whereas those between HCAECs and HSaVECs may fail to endow the corresponding vessels with resistance to restenosis and atherosclerosis. Another scenario implies differential patterns of beneficial and deleterious effects exhibited by each of the EC combinations (HCAEC-HITAEC and HCAEC-HSaVEC). In this study, we attempted to simulate the interactions between HCAECs and conduit ECs (HITAECs and HSaVECs), employing a co-culture model and measuring the expression of key endothelial genes and proteins, as well as secreted factors.
Co-culture of HCAECs and HITAECs led to the increased production of eNOS, reduced expression of EndoMT transcription factors Snail and Slug, and upregulation of arterial specification transcription factor HES1 in HCAECs in pairwise comparison with the HCAEC-HSaVEC co-culture model. Importantly, co-culture with HCAECs augmented eNOS expression and concurrently abrogated the synthesis of Snail and Slug in HITAECs, while bestowing HSaVECs with an arterial signature protein, HES1. Taken together, these observations testify to the synergistic and mutually favourable interactions between HCAECs and their conduit ECs, in particular HITAECs. It also suggests an additional mechanistic explanation for the arterialisation of SV conduits in the heterotopic position upon CABG in addition to pulsatile flow, increased blood pressure, higher shear stress, and oxygen-rich blood [12].
To date, the exact mechanisms of these paracrine effects are obscure and remain to be investigated in detail. Nevertheless, eNOS acts as a potent stimulator of angiogenesis in animal models of experimental ischaemia [81][82][83][84], being reciprocally enhanced by VEGF [85][86][87]. Another critical angiogenic regulator is the Notch pathway [88,89] which also contributes to the production of eNOS [90] and interacts with VEGF in EC sprouting [91][92][93]. We suggest that the relative increase in pro-angiogenic molecules at 48 h in the HCAEC-HITAEC co-culture setup might be related to the upregulation of eNOS and HES1, a master regulator of the Notch pathway (as compared to the HCAEC-HSaVEC model). In addition to the pro-angiogenic switch, an increased expression of eNOS in a HCAEC-HITAEC co-culture setup might explain both the reduction in EndoMT transcription factors and the paradoxical increase in IL-6 and IL-8 secretion, as the loss of eNOS promotes [94] and accompanies [95,96] EndoMT, whereas its upregulation potentiates inflammation [97,98].
Communication of HCAECs and HITAECs (or HSaVECs) upon CABG surgery may be carried out by extracellular vesicles (microvesicles, exosomes, and exomeres), which transfer cargo between the cells through the circulatory system [99][100][101][102][103], or by secreted molecules such as IL-8, a major pro-inflammatory cytokine [104]. As such, endothelial extracellular vesicles transport angiogenic mRNA (VEGF, basic fibroblast growth factor, stromal cell-derived factor/CXCL12), supporting endothelial homeostasis [105][106][107][108]. In endothelial dysfunction, the profile of extracellular vesicles shifts from anti-apoptotic, anti-inflammatory and pro-angiogenic to pro-apoptotic, pro-inflammatory, and anti-angiogenic, promoting the development of the pathological microenvironment [109,110]. Mechanical and biochemical alterations induced by the artificial anastomosis of the CA with SVs may negatively affect the endothelium, probably triggering adverse changes in the profile of extracellular vesicles secreted by HCAECs and conduit ECs. In contrast, total arterial revascularisation may support endothelial homeostasis and preserve physiological extracellular vesicle profile. Further studies are required to explore whether the extracellular vesicles are behind the beneficial interactions between HCAECs and HITAECs revealed in this investigation.
The drawbacks of our model include the absence of other cell populations which could contribute to paracrine signaling (e.g., vascular smooth muscle cells) and the limited duration of the time-lapse analysis (from 6 to 48 h), which could affect the results due to the specific temporal expression patterns we have shown. The use of blood vessel explants could be an appropriate solution, as they are widely established in atherosclerosis research [111][112][113] and CABG conduits are also investigated ex vivo [114,115]. However, this is barely possible for the CA. Regarding the time points, confluent EC cultures undergo active metabolism and therefore the medium should be collected within the 48 h maximum to ensure the proper maintenance of endothelial homeostasis in serum-free conditions. Hence, we believe the co-culture model is suitable for the in vitro analysis of possible interactions between HCAECs and conduit ECs. Nonetheless, co-culture of arterial ECs with the respective (for instance, coronary and internal thoracic artery) vascular smooth muscle cells and subsequent profiling of their transcriptome, proteome, and secretome would be useful to interrogate how the interactions of ECs with other vascular populations may modulate blood vessel homeostasis upon CABG. Another intriguing issue is that the results of RT-qPCR and Western blotting profiling considerably differed across the time points. However, this may be explained by the involvement of post-transcriptional regulation (e.g., by miRNA or translation initiation factors) and has been observed previously during the comparison of endothelial differentiation signatures defined by RNA-seq and different Western blotting modalities [116].
In conclusion, we suggest that the paracrine interactions between the CA and its bypass graft (in particular ITA) ECs after CABG might be reciprocal and beneficial, probably contributing to the better long-term patency of the ITA grafts following CABG. Further studies in this direction would identify the patterns and mechanisms of this synergy.

Co-Culture Model
Confluent (85%-90%) HCAECs (300K-05a, Cell Applications, San Diego, CA, USA) were co-cultured with either HITAECs (308K-05a, Cell Applications, San Diego, CA, USA) or HSaVECs (C-12231, PromoCell, Heidelberg, Germany) in co-culture chambers (CLS3419-12EA, Sigma-Aldrich, St. Louis, MO, USA) filled with MesoEndo Basal Medium (210-500, Cell Applications, San Diego, CA, USA). As a kind of control, we used HCAEC, HITAEC, and HSaVEC monocultures grown separately in the same type of the dish. All cell culture procedures were performed strictly under sterile conditions. For the whole time of the experiment, cell cultures were incubated at physiological temperature (37 • C), 95% air:5% CO 2 atmosphere, and high humidity. Upon 6, 24, or 48 h of co-culture, we collected the conditioned medium to measure the levels of cytokines and angiogenic molecules.
After the supernatant collection, cells were washed with ice-cold phosphate-buffered saline (PBS) and lysed with TRIzol Reagent (15596018, Thermo Fisher Scientific, Waltham, MA, USA) for RNA extraction or with a radioimmunoprecipitation (RIPA) assay buffer (89901, Thermo Fisher Scientific, Waltham, MA, USA) supplied with Halt protease and phosphatase inhibitor cocktail (78444, Thermo Fisher Scientific, Waltham, MA, USA) for the total protein extraction, according to the manufacturer's protocols. Quantification and quality control of the isolated RNA was performed employing a Qubit 4 fluorometer (Q33238, Thermo Fisher Scientific, Waltham, MA, USA), Qubit RNA BR assay kit (Q10210, Thermo Fisher Scientific, Waltham, MA, USA), Qubit RNA IQ assay kit (Q33222, Thermo Fisher Scientific, Waltham, MA, USA), Qubit RNA IQ standards for calibration (Q33235, Thermo Fisher Scientific, Waltham, MA, USA), and Qubit assay tubes (Q32856, Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's protocols. Quantification of total protein was conducted using a BCA Protein Assay Kit (23227, Thermo Fisher Scientific, Waltham, MA, USA) and a Multiskan Sky microplate spectrophotometer (51119700DP, Thermo Fisher Scientific, Waltham, MA, USA) in accordance with the manufacturer's protocols.

Gene
Forward Primer Sequence Reverse Primer Sequence Incubation with the antibodies was performed using iBind Flex Solution Kit (SLF2020, Thermo Fisher Scientific, Waltham, MA, USA), iBind Flex Cards (SLF2010, Thermo Fisher Scientific, Waltham, MA, USA), and an iBind Flex Western Device (SLF2000, Thermo Fisher Scientific, Waltham, MA, USA) for 3 h, according to the manufacturer's protocols. Chemiluminescent detection was performed using SuperSignal West Pico PLUS chemiluminescent substrate (34580, Thermo Fisher Scientific, Waltham, MA, USA) and a C-DiGit blot scanner (LI-COR Biosciences, Linkoln, NE, USA) in a high-sensitivity mode (12-min scanning). Densitometry was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA) using the standard algorithm (consecutive selection and plotting of the lanes with the measurement of the peak area and subsequent adjustment for the loading control (CD31) and reference group). The adjusted densitometry values were finally represented as a heat map (green, gray, and red colours reflected fold changes of ≤0.75, 0.76-1.24, and ≥1.25, respectively).

Secretome Profiling
Conditioned medium from both co-culture models collected at 6-and 48-h time points was profiled for angiogenic factors and cytokines, employing the respective dot blotting kits (ARY007 and ARY005B, R&D Systems, Minneapolis, MN, USA) according to the manufacturer's protocols. Chemiluminescent detection was performed using a C-DiGit blot scanner (LI-COR Biosciences, Linkoln, NE, USA) in a high-sensitivity mode (12-min scanning). Additionally, the concentrations of IL-6 and IL-8 in the cell culture supernatant were analysed utilising respective enzyme-linked immunosorbent assay kits (ab178013 and ab46032, Abcam, Cambridge, UK), according to the manufacturer's protocols. The colorimetric detection was carried out utilising a Multiskan Sky microplate spectrophotometer (51119700DP, Thermo Fisher Scientific, Waltham, MA, USA) at a 450-nm wavelength.

Statistical Analysis
Statistical analysis was performed in GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA). For descriptive statistics, data were represented by the median, 25th, and 75th percentiles, and range. Groups were compared by means of the Mann-Whitney U-test. P values ≤ 0.05 were regarded as statistically significant.

Conclusions
Paracrine interactions between the CA and its bypass graft (in particular ITA) ECs after CABG might be reciprocal and beneficial, probably contributing to the better long-term patency of the ITA grafts following CABG. Funding: This research and the related Article Processing Charge (APC) was funded by Russian Foundation for Basic Research, grant number 19-015-00198 "Interactions between coronary artery and internal mammary artery endothelial cells as a basis for application of autologous arteries as conduits for coronary artery bypass graft surgery".

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.