Preliminary in vitro study of the heme uptake by the HAECsAfter treatment of HAECs with Hb (200 and 800 µM), shape and pattern of the cells was slightly altered (Fig. 1). In contrast, we did not observe a change in cell morphology between 200 and 800 µM. The endothelial monolayer was not disrupted even with the highest Hb concentration used. The staining with DAPI after 24 h of incubation with Hb showed no difference in the cell viability, meaning that in this time-frame, the used Hb concentration was not toxic to the HAECs. There was only a slight increase in PE-Texas red autofluorescence of the HAECs treated with Hb in comparison to the control ones. This increase was not dependent on the concentration of the Hb used.Fig. 1Evaluation of Hb effect on HAECs after treatment for 24 h with 200 or 800 µM Hb or a vehicle (control). (A) Representative images of HAECs; (B) viability of HAECs assessed by flow cytometry and presented as percentage of DAPI negative cells (live cells); (C) autofluorescence of the HAEC cells in PE-Texas Red channel, shown as median fluorescence intensity (MFI).In order to determine which excitation wavelength is the most suitable to track heme uptake by HAECs with Raman spectroscopy, we first measured UV–Vis absorption spectra of Hb and oxidized cytochrome C, which is dominant form of CytC in fixed cells21. As presented in Fig. 2, the application of 532 nm excitation provided a similar level of resonance enhancement in the case of both molecules. Moreover, the same heme-related modes connected with the Q0 band are enhanced in case of both molecules. Therefore, differentiation between Hb and oxidized CytC using 532 nm is challenging. The examples of Raman imaging with this excitation were presented in the Supplementary Materials as Figure S1. Even though we applied a high concentration of Hb (800 µM), the tracking of heme uptake still remained ambiguous due to the similarity of its spectral pattern with cytochrome C. In the case of Hb, v4 mode located around 1585 cm−1 is usually more intense, contrary to v21 mode at around 1310 cm−1, which is more intense for cytochrome C compared to Hb. Therefore, their ratio can be a hint in Hb and CytC differentiation (Fig. S1A), but this change is relatively too small to provide clear-cut definition of Hb- and CytC-rich regions.Fig. 2UV–Vis absorption spectra of hemoglobin and oxidized cytochrome C presented in the 300–700 nm spectral range with marked excitation wavelengths used in Raman imaging.In turn, application of 488 nm excitation enabled stronger enhancement of Hb bands in Raman spectra compared to CytC, due to the proximity of the Soret band to 488 nm, what is connected with a strong pre-resonance effect16,22,23. It indicated, that heme encapsulated within cells should be characterized by more intense symmetric modes, e.g., ν4, in the Raman spectra in comparison with CytC. Therefore, this excitation was chosen to reveal cells’ heme uptake and differentiate Hb from CytC in in vitro studies. On the other hand, 532 nm excitation was used to capture this process in aorta, where cytochromes did not impede heme identification.Heme uptake by the HAECs—Raman imagingIn the Fig. 3 there is presented the scheme of the experiments with heme uptake using Raman imaging.Fig. 3The scheme of the experiment with heme uptake process.Figure 4 contains examples of Raman images of representative HAECs with heme accumulated during the uptake process. Next to the visual image, each panel comprises Raman integration and K-means clustering (KMC) images obtained with 488 nm excitation line. The Raman integration images present distribution of organic matter, heme, nucleus and lipids. Similarly, KMC images were constructed with division of the consistent dominant classes which were presented with the corresponding Raman spectra averaged from each class.Fig. 4Representative examples of heme uptaken by HAECs after treatment with 200 µM Hb concentration (A, B and C panels). Each panel comprises visual and Raman images (×60) of the HAECs showing the distribution of organic matter (integration in the 2820–3050 cm−1 range), heme (integration in the 1360–1390 cm−1 range), nucleus (integration in the 780–800 cm−1 range) and lipids (integration in the 2820–2880 cm−1 range), followed by KMC image with color coding as follows: cytoplasm—grey, heme—red, lipids—orange, nucleus—light green and protein rich cytoplasm—dark green. Raman images were recorded with the use of 488 nm excitation. To each cell component there is presented corresponding average Raman spectrum acquired from KMC.In agreement with the literature, cellular organic matter was distinguished by the integration of broadband in 2820–3050 cm−1 originating from CH2 and CH3 stretching vibrations, typical for biological samples15,24,25. Heme was recognized by the characteristic band located at around 1375–1378 cm−1, which is assigned to v4 mode—pyrrole half-ring stretching vibration15,26,27. As can be noticed in the KMC images, the red class was visible solely in the cellular interior. Close to the cellular membranes were localized lipid droplets, which were characterized by 2850 cm−1 band in the Raman spectra24,28. Their appearance could indicated cellular senescence induced by heme presence, which lead to oxidative cell injury and decreased cell viability in consequence15. Additionally, in the panel C were differentiated regions rich in proteins, associated with other cellular compartments.More detailed careful analysis of the spectral pattern of the red classes assigned to heme, revealed bands at 1567 and 750 cm−1, which are ascribed to the labile heme—hematin (Fig. S2). Tracking the position of the v4 mode allows to determine the iron ion oxidation state. In case of a ferrous hemes (Fe2+), the v4 is located at around 1360 cm−1, while for a ferric hemes (Fe3+) it blue-shifted to around 1375–1380 cm−116,26,27. In all recorded spectra assigned to the heme class, the v4 mode was located between 1375–1378 cm−1, what taken together indicated free ferric heme (Fe3+). This is also highlighted by the low intensity of the bands at around 675 and 750 cm−1, what is characteristic for the heme without the globin part, in contrast to ferric Hb where these bands are more prominent. Especially, the lack of the band at 675 cm−1 excludes the assignment of the red class to ferric Hb. (Fig. S2). It is worth to mention, that observation of the heme uptake process in HAECs using Raman imaging and 488 nm excitation was possible even in relatively low Hb concentration (200 µM).Heme uptake by the ECs in the aorta cross-sections and en faceThe search for the heme uptake by the ECs in the aorta was started from representative cross-section through the aorta isolated from wild-type C57Bl/6J and FVB mouse strains (Figs. 5, 6, 7). The Raman integration images presented in Figs. 5, 6, 7 and 8 were constructed by integration of certain spectral ranges and represent organic matter (2800–3050 cm−1, stretching vibrations of the CH2 and CH3 groups), elastin (510–550 cm−1, stretching vibrations of the S–S), and heme (ν37 mode originated from ν(CαCm)as in-plane vibrations)15,29.Fig. 5Representative example of heme uptaken by ECs in the cross-section aorta isolated from wild type mice. Each panel comprises visual and Raman images (×100) of the aorta cross-sections showing the distribution of organic matter (integration in the 2800–3050 cm−1 range), heme (integration in the 1565–1595 cm−1 range), nucleus (integration in the 780–800 cm−1 range), and elastin (integration in the 510–550 cm−1 range) followed by KMC image with color coding as follows: cytoplasm—grey, heme—red, nucleus –green and elastin—blue. Raman images were recorded with the use of 532 nm excitation. To each cell component there is presented corresponding average Raman spectrum acquired from KMC.Fig. 6Representative examples of heme uptaken by ECs in the aorta acquired from FVB mice. Each panel comprises visual and Raman images (×100) of the aorta cross-sections showing the distribution of organic matter (integration in the 2800–3050 cm−1 range), elastin (integration in the 510–550 cm−1 range) and heme (integration in the 1565–1595 cm−1 range). Raman images were recorded with the use of 532 nm excitation.Fig. 7Representative example of heme uptaken by ECs in the en face aorta isolated from wild type mice. The next panels show images Raman images (×100) of the en face aorta showing the distribution of organic matter (integration in the 2800–3050 cm−1 range), heme (integration in the 1565–1595 cm−1 range), nucleus (integration in the 780–800 cm−1 range), and elastin (integration in the 510–550 cm−1 range) Raman images were recorded with the use of 532 nm excitation.Fig. 8Representative examples of heme uptaken by ECs in the aorta acquired from ApoE/LDLR−/− mice. Each panel comprises visual and Raman images (×100) of the aorta cross-sections showing the distribution of organic matter (integration in the 2800–3050 cm−1 range), elastin (integration in the 510–550 cm−1 range) and heme (integration in the 1565–1595 cm−1 range). Raman images were recorded with the use of 532 nm excitation.Figure 5 shows an example of heme uptake, which is not limited to the endothelial layer itself, but represents the phenomenon of heme entering the subendothelial layers of the aorta. This is evidenced by the location of the Raman signal coming from heme (Fig. 5. Red class) in relation to the location of subsequent elastin layers (Fig. 5. Blue class). The heme uptake was also evidenced in the aorta cross-section from FVB mice (Fig. 6). This stays in agreement with data from the literature, which describes the phenomenon of heme uptake into smooth muscle cells (particularly evident in Figs. 5 and 7). Excess of labile heme induces a concentration-dependent migration and proliferation of vascular smooth muscle cells (VSMCs), which depends on the production of reactive oxygen species (ROS) derived from NADPH oxidase (NADPHox) activity30. Also, heme causes TNF- and ROS-dependent macrophage cell death with characteristics of programmed necrosis31.Therefore, it demonstrates that heme acts as a double-edged sword. While under physiological conditions heme is crucial for various biological processes, an excess of labile heme acts as a pro-oxidant, enhancing oxidative stress and inflammation32.To better document that the presence of the heme signal inside the blood vessel was not a contaminant, Raman-based depth-profiling was performed into the blood vessel (Fig. 7). For this purpose, the blood vessel rings were split-open and measured from the depth to the endothelial side. The signal from heme is visible not only on the surface of the blood vessel, but also deep inside, which indicates active incorporation of heme by the endothelium.In search of the heme uptake by the ECs in the aorta cross-sectionsHeme uptake by the endothelium is not a characteristic of solely healthy endothelium, but also occurs in altered blood vessels. We examined representative cross-sections obtained from the murine model of atherosclerosis (ApoE/LDLR−/− mice; Fig. 8A, B). Accumulation of the heme was especially prominent between the elastin fibers. Amount of the heme accumulated within the elastin layers in the aorta cross-sections obtained from ApoE/LDLR−/− mice (Fig. 8A, B) was relatively higher compared to the C57Bl/6J and FVB controls. However, in order to carry out robust quantitative analyses, it is imperative to expand the measurements to encompass a more extensive cohort of atherosclerotic animals. The current study aims to elucidate the measurement methodology and potential, rather than presenting comparative results between control and atherosclerotic mice models.
