Label-free chemical imaging of cytochrome P450 activity by Raman microscopy | Panda Anku

Visualization of CYP induction by Raman microscopy

Rifampicin (RIF), a well-known CYP inducer, especially for CYP3A4, was applied to HepaRG cell culture to induce the expression of CYPs33, and hyperspectral Raman images were captured. HepaRG cells without RIF treatment were taken as controls. After comparing the Raman spectrum of the hepatocytes with RIF treatment (Fig. 1a), we found the most significant difference in the appearance of Raman bands at 1370 cm−1 and 1636 cm−1 (inserts, in Fig. 1a). From the previous resonance Raman studies of the CYP protein, substrate binds to CYPs at the ferric and low-spin (LS) resting state and triggers the CYP catalytic cycle. 1370 cm−1 and 1636 cm−1 were generally used as oxidation and LS state marker of CYP, respectively28,30,34. The appearance of 1370 cm−1 and 1636 cm−1 peaks after RIF treatment may suggest that the induced CYP is at an oxidized and LS state.

Fig. 1: Measurement of CYP induction by Raman microscopy.
figure 1

a Average Raman spectra of HepaRG cells treated with (red) or without (gray) rifampicin (RIF). The Raman shifts of 1370 cm−1 and 1636 cm−1 increased after treatment with RIF for 48 h; enlarged in the insets. b Reconstructed Raman images at 600 cm−1, 675 cm−1, 1000 cm−1, 1370 cm−1, and 1636 cm−1, which can be assigned to cyt c, cyt b5, phenylalanine, oxidized CYPs, and LS CYPs, respectively. The comparison of these cellular components demonstrated that 1370 cm−1 and 1636 cm−1 increased after CYP induction. Scale bars, 20 µm. c Immunofluorescence staining of cyt b5 and CYP3A4 at the same positions as Raman measurement. Scale bars, 20 µm. d Western blots of CYP3A4, cyt c, cyt b5, and β-action. e CYP3A4 activity assay indicates CYP induction after RIF treatment. Error bars indicate SD between triplicates. ***P < 0.001.

Raman imaging enables visualization of the spatial distribution of molecules at the target wavenumbers. The cytosolic distribution of 1370 cm−1 and 1636 cm−1 after CYP induction is shown in Fig. 1b, compared with peaks at 600 cm−1 (reduced heme c, mainly cytochrome c or cyt c), 675 cm−1 (reduced heme b), and 1000 cm−1 (phenylalanine). A clear cytoplasm contrast appeared at 1370 cm−1 and 1636 cm−1 after RIF treatment. Since cyt c localizes in the mitochondria, the increasing cytoplasm contrast after RIF treatment would not come from c-type heme but rather b-type heme.

In hepatocytes, CYP3A4 and cytochrome b5 (cyt b5) are the most abundant heme b proteins associated with drug metabolism and electron transfer, respectively35,36,37. Therefore, we performed immunofluorescence staining for CYP3A4 and cyt b5 after Raman observation. Figure 1c shows the immunofluorescence staining of CYP3A4 and cyt b5 at the same position as the Raman observation. The fluorescent signals of CYP3A4 increased after RIF induction, while the signals of cyt b5 remained constant, consistent with the western blotting results (Fig. 1d). After RIF induction, the Raman signal of 1370 cm−1 and 1636 cm−1 increased, similar to the changes observed in the fluorescent signal of CYP3A4 staining (Fig. 1c) and CYP3A4 activity assay (Fig. 1e). These results suggest that the increase of Raman shifts at 1370 cm−1 and 1636 cm−1 can be associated with the induction of CYP3A4 by RIF. Although RIF is a strong CYP3A4 inducer, other CYPs, like 2B6, were also elevated in HepaRG cells when treated with RIF (Supplementary Fig. 1). Since the heme b core is the same in all CYP subtypes, other CYP types should also contribute to the arise of Raman peaks at 1370 cm−1 and 1636 cm−1. Therefore, we treated HepaRG cells with DMSO and omeprazole, to induce all CYPs and CYP1A2, respectively. The significant increase in 1370 cm−1 and 1636 cm−1 was revealed in DMSO-treated cells but not omeprazole (Supplementary Fig. 2). Since the same Raman shifts were detected in DMSO-treated cells, the Raman shifts at 1370 cm−1 and 1636 cm−1 can be assigned to several CYPs including CYP3A4. Although the treatment with OME elevated the expression of 1A2 for several folds, this amount is still undetectable by Raman microscopy possibly due to the low basal expression of CYP1A2 in HepaRG cells (Supplementary Fig. 2)32,38.

Primary human hepatocytes (PHHs) are widely used in drug metabolism and screening study due to high basal expression of CYPs39. Therefore, we measured PHHs by Raman microscopy and compared with HepaRG cells. PHHs and HepaRG cells were observed right after cells have fully adhered to the collagen-coated surface. The Raman shifts at 1370 cm−1 and 1636 cm−1 were detected both in PHHs and HepaRG cells (Supplementary Fig. 3a). The time-course observation demonstrated the decreasing tendency of 1370 cm−1 and 1636 cm−1 after plating (Supplementary Fig. 3b), which was consistent with the immunostaining of CYP3A4 (Supplementary Fig. 3c). The expression of CYP in PHHs rapidly decreased once after plating, this known feature of primary hepatocytes was monitored by Raman microscopy40. Besides 1370 cm−1 and 1636 cm−1, we found dominant peaks at 1157 cm−1 and 1512 cm−1 (Supplementary Fig. 3a, indicated by asterisks), which can be assigned to carotenoids stored in the liver41. Using Raman microscopy, we confirmed the same vibrational modes coming from CYPs in both PHHs and HepaRG cells. On the other hand, the difference between these cell lines was also clarified, demonstrating the potential of label-free Raman in identifying features of different cell lines based on their intrinsic cellular components.

Our results indicate that the oxidized and LS form of CYPs is directly related to the induction in living hepatocytes, which supports the conventional belief that the ferric and LS state is the starting point of the CYP catalytic cycle42,43. The redox state of CYP has been studied intensively in a reconstituted protein system but not in vitro cells by Raman spectroscopy. Johnston et al. investigated the redox state of human recombinant CYPs expressed in E. coli and living rat hepatocytes using Fe(II)-CO versus Fe(II) difference spectroscopy and reported that both ferrous and ferric CYPs existed in vitro44. However, the CO difference spectrum measurement remains a concern for interrupting the natural redox state of CYP45. Using Raman microscopy, we directly visualized cellular CYPs without external labels and demonstrated that ferric CYPs with LS were the major form after induction in living HepaRG cells.

Non-destructive detection of hepatocyte-specific response to CYP induction

Non-invasive characterization of different cell populations in an in vitro model could pave the way for the evaluation of cell type-specific and subsequent intercellular responses after drug administration. Differentiated HepaRG cells are known to present two morphologically and functionally distinct cell populations, including hepatocytes and biliary cells46. Hepatocytes form colonies and are surrounded by flat biliary cells. Raman spectra of hepatocyte colonies and biliary monolayer were collected in HepaRG cell without induction. Clear differences were found in the Raman spectra of the two cell populations (Fig. 2a). The Raman spectra of biliary cells were dominated by resonance peaks from reduced cyt c (600 cm−1, 640 cm−1, 750 cm−1, 1130 cm−1, 1313 cm−1, and 1585 cm−1), protein (1000 cm−1, phenylalanine), and lipids (1450 cm−1, CH2 deformation) bands. In contrast, hepatocytes (without induction) preserved all the above-mentioned Raman shifts with higher intensities, indicating that these biomolecules are more abundant in hepatocytes. Furthermore, Raman shifts corresponding to reduced b-type cytochrome (675 cm−1 and 1304 cm−1) and glycogen (940 cm−1 and 1084 cm−1) appeared in hepatocytes. A comparison of representative 675 cm−1 and 940 cm−1 peaks is shown in the insets of Fig. 2a. The reconstructed Raman images comparing hepatocytes and biliary cells are shown in Supplementary Fig. 4a. The immunostaining images of cyt b5 at the same position of Raman observation indicate that more cyt b5 is expressed in hepatocytes than in biliary cells (Supplementary Fig. 4b).

Fig. 2: Hepatocyte-specific response to CYP induction was monitored by Raman microscopy.
figure 2

a Raman spectra of hepatocytes (red) and biliary cells (gray) without CYP induction. b Immunofluorescence staining of CYP3A4 (green) and CK-19 (red) with nuclear counterstain by DAPI. Hepatocyte cells (HC) are CYP3A4 positive; while biliary cells (BC) are CK-19 positive. Scale bar, 50 µm. Reconstructed Raman images of control hepatocytes (c Ctrl-H), control biliary cells (d Ctrl-B), RIF-treated hepatocytes (e RIF-H), and RIF-treated biliary cells (f RIF-B) show the specific response of hepatocytes to RIF induction. Scale bars, 20 µm. g Average spectra of Ctrl-H (gray), Ctrl-B (light gray), RIF-H (red), and RIF-B (light red).

After RIF treatment, expression of CYP3A4 was found in hepatocyte region but not in biliary cells by immunostaining  of CYP3A4 (Fig. 2b, green). The staining of CK-19, a biliary marker, was performed together to identify biliary cells (Fig. 2b, red). The reconstructed images at 600 cm−1, 1370 cm−1, and 1636 cm−1 are shown to reveal the localization of cyt c, oxidized CYPs, and LS CYPs, respectively (Fig. 2c–f). Hepatocytes showed a clear cytoplasm contrast of 1370 cm−1 and 1636 cm−1 after RIF induction (Fig. 2e), whereas only mitochondrial contrasts at 600 cm−1 were observed in biliary cells (Fig. 2f), indicating that the occurrences of 1370 cm−1 and 1636 cm−1 were more specific to hepatocytes. This change in biochemical components was clearly observed in the comparison of Raman spectra (Fig. 2g). It indicates that Raman microscopy is capable to monitor hepatocyte-specific CYP induction without labeling.

Raman microscopy enables quantitative analysis of CYP activity

To further investigate the correlation between Raman shifts assigned to CYPs (1370 cm−1 and 1636 cm−1) and CYP activity, we examined the inducer concentration-dependent, inducer type-dependent, down-regulatory, and time-dependent effects on CYP activity by Raman microscopy at the single-cell level. As the shape of the peak at 1370 cm−1 was strongly affected by the adjacent 1360 cm−1 peak assigned to reduced hemes, we chose the Raman shift at 1636 cm−1 for the quantitative analysis of CYP activity. The signal processing method used for the analysis is shown in Supplementary Fig. 5.

First, we applied different concentrations of RIF in a gradient to the cell culture. The high spatial resolution of Raman imaging allowed us to investigate the cell-cell differences in CYP3A4 activity, as shown in the Raman images (Fig. 3a–d). Both the Raman intensity at 1636 cm−1 and the luminescence intensities increased with increasing RIF concentration (Fig. 3e). The higher background in the Raman results could be a signal from oxidized cyt c or other types of CYPs. The Raman intensity at 600 cm−1 indicated that cyt c was constant regardless of CYP induction (Supplementary Fig. 6a), which was supported by the western blotting results (Fig. 1d), indicating that the increase in the 1636 cm−1 was mainly caused by the induction of CYPs. The analysis of the Raman intensity at the single-cell level is shown in Fig. 3i (boxplot). The ratio above the third quantile of the control condition gradually increased from 25 to 76% when the RIF concentration reached 4 µM, and the ratio below the first quantile compared with control conditions decreased from 25 to 8% (Fig. 3i), demonstrating an increase in higher CYP-expressing populations. We also investigated the universal usage of Raman measurements on other CYP inducers (Fig. 3f, j), including phenytoin (PHY), dexamethasone (DEX), and carbamazepine (CBZ), all of which showed the similar tendency as that seen in the luminescence assay of CYP3A4 (Fig. 3f). The induction of CYP1A2, CYP2B6, and CYP2C9 is shown in Supplementary Fig. 7, indicating that the change in CYP3A4 is dominant.

Fig. 3: CYP activity at various conditions measured by Raman microscopy.
figure 3

ad Reconstructed Raman images at different RIF concentrations from 0, 0.04, and 0.4 to 4 µM. Images constructed at 1636 cm−1 are labeled in green, and images constructed at 600 cm−1 are labeled in red. The number of 1636 cm−1 positive cells increased with increasing concentration of RIF. Scale bars, 20 µm. eh Comparison of CYP activity measured with luminescence assay and average Raman intensity at 1636 cm−1 with RIF in a gradient (e), various CYP3A4 inducers (f), IL-6 (g), and at the different culture time (h), of which the Raman intensity of individual cells is shown in (il), respectively. Error bars eh indicate SD between triplicates. Box plots illustrate the spread within cell populations (center line, median; box, quartiles; whiskers, 1.5× interquartile range). Statistical significance was evaluated by one-way ANOVA followed by post hoc Tukey-Kramer’s range tests (compared with the left column/box). ***P < 0.001; **P < 0.01; *P < 0.5.

Inflammation and infection downregulate the expression and activity of CYPs, altering the pharmacokinetics and clearance of drugs47. The inhibitory effect varies among patients due to pharmacogenetic variations in the inflammatory pathway48. Therefore, we investigated the possibility of using Raman microscopy to analyze the downregulation of CYP by interleukin-6 (IL-6), a principal regulator of the hepatic acute-phase response49. 2 ng/mL IL-6 was added to HepaRG cells and treated for 48 h to suppress CYP3A4 expression. Downregulation of CYP3A4 expression was confirmed by measuring CYP3A4 activity (Fig. 3g). The Raman intensity at 1636 cm−1 decreased, consistent with the results of CYP3A4 activity assay (Fig. 3g, k).

Next, we evaluated the dynamics of CYP3A4 expression without any external interference. HepaRG cells expressed higher levels of CYP3A4 in the first several hours after thawing, which decreased gradually until 48 h. This natural decrease was confirmed by analyzing the intensity of the peak at 1636 cm−1 (Fig. 3h, l).

To eliminate the possible contamination of Raman scattering from inducers or down-regulators. The Raman spectra of the stock solution of each chemical are shown in Supplementary Fig. 8. No overlapping at 1370 cm−1 and 1636 cm−1, as well as no characteristic peaks of these inducers and down-regulators, were found in the cell spectrum.

Taken together, the analysis of the inducer concentration-dependent, inducer-type-dependent, down-regulatory, and time-dependent effects demonstrated that 1636 cm−1 can be utilized as a Raman indicator to detect CYP activity under various conditions. This suggests that Raman microscopy can be applied as a screening method to evaluate the modulation of CYP activity by drugs and inflammatory stimuli.

Mechanism-based inhibitor binds to active CYP

To prove the feasibility of using the 1636 cm−1 Raman shift to indicate induced CYP, a mechanism-based inhibitor was applied to the cell culture. Mechanism-based inhibitors bind to CYP as a substrate analog and trigger the catalysis reaction50. The catalysis intermediate binds to the heme core or apoprotein and blocks the entrance of other substrates to irreversibly inactivate the enzyme. Azamulin (AZA), a highly specific and mechanism-based inhibitor of CYP3A451, can bind to the heme core of CYP3A4 and cause ferric LS to high-spin (HS) transition52 (Fig. 4a). The Raman spectra of HepaRG cells with or without AZA treatment were collected (Fig. 4b), and the CYP3A4 activity assay was performed to confirm the inhibition effect, as shown in Fig. 4c. Both the Raman signal at 1636 cm−1 and 1370 cm−1 decreased after 10 min of treatment with 10 µM AZA (Fig. 4c). The decrease in the Raman shift was smaller than that in the activity assay, presumably due to the signal of oxidized cyt c or other CYP subtypes at the same wavelength, for reasons discussed in the following paragraph.

Fig. 4: Detection of CYP inhibition after azamulin (AZA) treatment.
figure 4

a Schematic diagram of a possible mechanism of AZA binding with CYP. Active CYPs maintain a ferric low-spin (LS) state by using water as a ligand. After AZA enters the enzymatic pocket, water is replaced by AZA, which causes LS to high-spin (HS) transition. Although metabolic intermediates of AZA were not identified, the crystal structure indicated the pleuromutilin head is required for mechanism-based inhibition. The possible ligation was depicted as a black dotted line. b The Raman shifts at 1370 cm−1 and 1636 cm−1 decreased after treatment with AZA. c Comparison of average Raman signal at 1636 cm−1 and CYP3A4 activity assay. Error bars indicate SD between triplicates. ***P < 0.001; **P < 0.01. d Reconstructed Raman images at 600 cm−1, 675 cm−1, 1370 cm−1, and 1636 cm−1. Scale bars, 20 µm. e Immunostaining of cyt b5 and CYP3A4 at the same positions of Raman measurement. Scale bars, 20 µm.

The sensitivity of Raman spectroscopy to the spin-state transition allowed us to evaluate the inhibition of CYP activity in living cells using the Raman peak at 1636 cm−1, as shown in Fig. 3d. A decrease of cytoplasm contrast was found in the Raman images of 1370 cm−1 and 1636 cm−1. Furthermore, immunofluorescence staining (Fig. 4e) and western blotting (Supplementary Fig. 9) of CYP3A4 demonstrated that AZA treatment did not alter the amount of CYP3A4, suggesting that the decrease in Raman intensity at 1636 cm−1 was associated with the change in the activity of CYP but not the amount. In the previous studies using recombinant protein, AZA bound to ferric CYPs and caused a ferric LS to HS transition52. The spin-state transition of CYP protein upon substrate binding has been investigated using resonance Raman spectroscopy due to its high sensitivity and possibility to measure under physiological conditions. Three spin-state marker bands are sensitive to the changes in porphyrin core size caused by the transition of spin state. These bands were found around 1500 cm−1 (ν3), 1590 cm−1 (ν2), and 1640 cm−1 (ν10) for the LS state and around 1490 cm−1, 1570 cm−1, and 1620 cm−1 for the HS state28,30,34. In our measurement, 1500 cm−1 and 1590 cm−1 were not detectable due to the interference of other cellular components or the possible lower efficiency of Raman scattering at 532 nm. A decrease of 1636 cm−1 was observed which allowed us to evaluate inhibition in living cells using this peak. Although 1370 cm−1 is not a so-called spin-state marker, the decrease in intensity was observed in our measurement (Fig. 4b), which might be associated with the reduction of CYP after binding to AZA.

We also performed the inhibition experiment using CYP3A4 overexpressed microsome, which is a simplified subcellular system preserving CYP activity. Microsomes were treated with 10 µM AZA for 10 min, and the Raman spectra were collected (Supplementary Fig. 10). We detected the transition from 1636 cm−1 to 1621 cm−1 and even shift from 1370 cm−1 to 1367 cm−1 as well as the decrease in intensity of 1636 cm−1 and 1370 cm−1. In living cell condition, Raman shifts ~ 1621 cm−1 were overwhelmed by other cellular peaks, and 1370 cm−1 was affected by the adjacent peak at 1360 cm−1. Therefore, the decrease in 1636 cm−1 was used to evaluate CYP inhibition in living cells.

From the spectroscopic view, the decrease in 1636 cm−1 without any change in the protein amount corresponds to the fact that the inhibition by AZA is due to the spin-state transition. After treatment with AZA, we found that the Raman image of 1636 cm−1 provided a mitochondrial contrast, which was also observed with reduced cyt c (600 cm−1). Since AZA does not affect the redox state of cyt c in HeLa cells (Supplementary Fig. 11) and the amount of cyt c in HepaRG cells (Supplementary Fig. 6), these results indicate that the 1636 cm−1 Raman signal also contains oxidized cyt c, contributing as a constant background in the measurements. The alternations of oxidized CYP signals induced the change in the Raman signal shown in Fig. 3 and Fig. 4.

CYP induction with glycogen depletion

The liver is an essential organ associated with not only drug metabolism but also glucose metabolism. The crosstalk between drug and glucose metabolism has been investigated to understand drug-induced metabolic disorders53. Nuclear receptors for drugs54, including pregnane X receptor (PXR), constitutive androstane receptor, retinoid X receptor, have been shown to modulate enzyme expression in glucose metabolism55. In particular, the induction of CYP3A4 by RIF is mediated by the activation of PXR33.

We investigated the intracellular glycogen storage and RIF-induced CYP activity in situ using Raman microscopy. HepaRG cells resemble the features of adult hepatocytes with the ability to express CYP3A4 (1636 cm−1), cyt b5 (675 cm−1), and store glycogen (940 cm−1), and lipids (1450 cm−1). As each of these hepatic functional biomolecules features in specific Raman shifts, the time-dependent changes of these molecules were all visualized by the Raman microscopy. The time-course cell Raman images at different culture days and conditions are shown in Fig. 5a. The increase of cellular glycogen with the culture time was imaged using the reconstructed Raman images at 940 cm−1.

Fig. 5: CYP induction with glycogen depletion.
figure 5

a Hyperspectral Raman images reveal dynamic changes in hepatic functional molecules (cyt b5, glycogen, and CYPs) and general cellular components (cyt c, phenylalanine, and lipids) from D0 to D7. An accumulation of glycogen is observed, whereas treatment of CYP inducer deceases the glycogen amount. Scale bars, 20 µm. b Quantitative analysis of the Raman intensity at 940 cm−1 and 1636 cm−1 shows decreased glycogen with increased CYP after CYP inducer treatment. Box plots illustrate the spread within cell populations (center line, median; box, quartiles; whiskers, 1.5× interquartile range). ***P < 0.001. c Single-cell-based dot plot with normalized Raman intensity at 940 cm−1 (x axis) and 1636 cm−1 (y axis). Each dot indicates one cell. d PAS staining demonstrates the accumulation of glycogen during the 7-day culture and consumption with RIF treatment. Scale bars, 100 µm.

Interestingly, we found that the amount of glycogen decreased as CYP expression increased (Fig. 5b). Figure 5c shows the negative correlation between CYP induction and glycogen storage, where each spot demonstrates a single cell. More high-CYP and low-glycogen cell populations were found after treatment with RIF (Fig. 5c). The inhibitory effect of CYP inducers on glycogen storage was detected when treated with other inducers (Supplementary Fig. 12). Following this, we confirmed the inhibitory effect of CYP inducers on glycogen storage by PAS staining (Fig. 5d). Our findings are in accordance with in vivo studies in which RIF treatment was found to reduce the glycogen content in rat livers56. The possible mechanism is through the activation of PXR which promotes the expression of PECPK1 and G6P30, two important enzymes in glucogenesis and glycogenolysis57.

Meanwhile, no significant difference was found in the Raman intensity at 600 cm−1 (cyt c), 675 cm−1 (cyt b5), and 1450 cm−1 (lipid) (Supplementary Fig. 13) after CYP induction, indicating a potential lack of the side effects of RIF on mitochondrial function and lipid metabolism. With a single scan of the cell sample, the negative regulation of glycogen storage was observed simultaneously with CYP induction by Raman microscopy, demonstrating that the multiplexity of drug response can be assessed in a non-destructive manner.

Leave a Comment