1.

Introduction

Human tissue autofluorescence spectroscopy and imaging are being explored as potential methods of noninvasive disease diagnosis. ^{1 2 3 4} Autofluorescence studies on in vivo and ex vivo human tissues include the cervix, ^{5 6 7 8 9} bladder, ^{10 11 12} lung, ^{13 14 15 16 17 18} breast, ^{18 19 20 21} esophagus, ^{22 23 24} aorta, ²⁵ stomach, ²⁶ uterus, ²⁷ skin, ^{28 29 30 31 32} and colon. ^{33 34 35 36 37 38 39 40 41} Tissue autofluorescence contains contributions from endogenous cellular and extracellular biomolecular species, and is influenced by light absorption and scattering, making detailed physical interpretations difficult. Studies on isolated cells in culture should help determine what portion of the tissue fluorescence arises from intrinsic cellular components. This understanding may, in turn, advance the application of autofluorescence methods to noninvasive diagnosis of human neoplastic disease.

Several endogenous cellular fluorophores have been reported in the literature. ¹ One of the most studied is the reduced form of nicotinamide adenine dinucleotide (NADH). ^{42 43 44} This reduced coenzyme typically has an excitation maximum near 380 nm and an emission maximum near 440 nm. Contributions from flavins (FAD, FMN, and riboflavin) are often observed in cellular autofluorescence. ^{42 43 44 45} Their steady-state fluorescence characteristics show an excitation maximum near 450 nm and an emission maximum near 515 nm. Lipofuscin granules are age-related lipophylic materials composed primarily of malonaldehyde. ^{46 47 48} These pigments absorb between 340 and 395 nm and emit in the red region of the electromagnetic spectrum (up to 640 nm). ^{1 2 49} Another known endogenous cellular fluorophore is the amino acid tryptophan. Tryptophan has two absorption bands observed at 220 and 287 nm, which emit coincidently at 350 nm. ^{50 51 52 53}

Over the past decade, several studies have characterized the autofluorescence properties of human epithelial cells. These cells are of interest, because carcinoma, the most common form of human cancer, derive from epithelial cells. In one report, both steady-state and time-resolved fluorescence spectroscopies were employed to measure unspecified epithelial cell autofluorescence. ⁴³ Under laser excitation, fluorescence emission and discrete lifetimes attributed to endogenous NADH and flavins were detected. Using confocal microscopy, the study detected increases in cellular autofluorescence due to formalin and aminolevulinic acid treatment. It was the conclusion of the study that the fluorescence properties monitored in these cells might be useful as a diagnostic tool to detect neoplastic changes. In another study, human oral epithelial cells (explanted from the normal mucosa of the tonsillar pillar) were monitored to determine if autofluorescence could differentiate between slowly growing (treated with TGF-β) and rapidly growing (untreated) cells. ⁵⁴ It was found that changes in endogenous NADH fluorescence discriminated between different proliferation rates. A subsequent report indicated that autofluorescence signatures might be useful in distinguishing between benign (normal) and malignant (squamous carcinoma line, 1483) oral epithelial cells. ⁵⁵ This study detected spectral shifts, as well as changes in autofluorescence intensity between cell types. The observed differences between normal and malignant cell emission spectra were attributed to changes in the optical absorption of residues such as tyrosine, tryptophan, phenylalanine, and others. Autofluorescence of epithelial cells from the human larynx, both squamous cell carcinoma (Hep ₂ ) and nontumorigenic (Vero) lines, have also been reported. ⁵⁶ This investigation observed autofluorescence emission at 340 and 440 nm that might be useful as a marker of malignancy or for uncovering mechanisms linked to transformation of epithelial cells. Recent studies monitored the autofluorescence from human epithelial cervical cancer lines (HeLa and SiHa), and found tryptophan, NADH, and FAD to be the primary fluorophores. ^{57 58 59}

Autofluorescence from bronchial tissue has been investigated as a means of lung cancer detection. ^{13 14 15 16 17 18 60 61 62 63 64 65 66 67 68 69} Currently, a FDA approved fluorescence imaging device for the lung relies upon autofluorescence properties to detect intraepithelial neoplastic lesions. ¹⁶ While there are prior reports of tissue autofluorescence, there are few prior studies of isolated bronchial epithelial cell autofluorescence. One such study recently reported in the literature ⁷⁰ measures human bronchial cell autofluorescence from a single sample, and will be discussed later. An important motivation for the work presented here, is to thoroughly investigate well-characterized cellular models, which may help to interpret the results of tissue studies. In this report, we describe the first study on the autofluorescence characteristics of both an immortalized and a distinct tobacco-carcinogen-transformed human bronchial epithelial cell line. These cell lines are useful models for studies seeking to distinguish between normal and transformed human bronchial epithelial cells. The goal of this work is to characterize the autofluorescence spectra of these cellular models. Spectrofluorimetry and confocal fluorescence microscopic imaging were employed to determine the predominant contributors to the observed cellular autofluorescence patterns and to compare differences in autofluorescence between the immortalized and carcinogen-transformed phenotypes. To date, such an investigation on these cell lines has not been reported.

2.

Experiment

2.2.

Spectrofluorimetry

Spectrofluorimetry measurements were performed on the model systems described above to characterize their autofluorescence properties. A Perkin–Elmer LS-50 dual monochromator luminescence spectrometer was used for fluorescence measurements throughout the ultraviolet, visible, and near-infrared regions of the electromagnetic spectrum. The instrument contained a xenon arc-discharge lamp as an excitation light source and a red-enhanced photomultiplier tube (Hamamatsu R928), with an extended range of 200–900 nm, as an emission detector. Emission was collected at a 90° angle relative to the excitation light path. The wavelength accuracy of the instrument was ±1.0 nm, and the reproducibility was ±0.5 nm. All spectra were corrected for solvent contributions, by background subtraction of a prerecorded solvent emission, and for the nonuniformity of the instrument response, by division of a wavelength-dependent sensitivity factor. ⁵⁰ These sensitivity factors were previously obtained, for this spectrometer, by recording the fluorescence emission of quinine sulfate (0.1 N) and forming the ratio with the known standard spectrum. Equation (1) describes the spectral correction procedure:

Eq. (1)

$$\text{corrected emission}=\frac{M(\text{λ})-B(\text{λ})}{S(\text{λ})}\text{,}$$

where M(λ) is the measured sample emission and B(λ) and S(λ) are the background and sensitivity factors. Photobleaching of the cell suspension was not observed, as determined by rescanning several spectra at the completion of each experiment. All measurements were made a minimum of three times, using new cell cultures each time.

Importantly, we found that high cell counts can lead to appreciable spectral artifacts, including intrusive harmonic peaks and intensity fluctuations. The origin of these artifacts is presumably due to increased scattering of the excitation light as well as increased reabsorption of the fluorescence emission. ⁵⁰ Therefore, to determine an acceptable cell count range for recording fluorescence spectra, we measured the BEAS-2B emission intensity (of a band consistent with the amino acid tryptophan) as a function of the cell number density. The results of this study are shown in Figure 1. For cell counts ranging from 60 000 to approximately 500 000 per milliliter, the emission intensity increased nearly linearly. Beyond this range, the curve deviated from this trend, presumably due to reabsorption and scattering effects. While the observed linearity in the 60 000–500 000 cells per milliliter region was not perfect, the spectra recorded in this range were sufficiently free from the spectral artifacts noted at higher cell counts. The source of deviation from linearity arises from both the above-mentioned effects and error associated with cell counting. For the experiments reported here, cell counts were consistently held at a level of approximately 300 000 cells/ml, as determined by hemocytometry.

Figure 1

Maximum emission intensity of a band consistent with tryptophan (excitation at 287 nm), collected on the LS-50 spectrofluorimeter, as a function of BEAS-2B cell number. A nearly linear emission intensity trend was observed for counts <500 000 cells/ml. Beyond this range, reabsorption and scattering artifacts affect emission intensities. Error associated with the data points arose primarily from cell-counting techniques and was estimated to be as much as 10–20 of the cell number.

3.

Results

3.1.

Spectrofluorimetry

As discussed in the experimental spectrofluorimetry section, cell counts were held at approximately 300 000 cells per milliliter. This had the result of minimizing scattering and reabsorption artifacts in the fluorescence measurements, see Figure 1. The spectrofluorimeter measurements described above were used to record the excitation/emission matrices (EEMs) of BEAS-2B and BEAS-2B _NNK cell suspensions, as shown in Figure 2. Excitation wavelengths were chosen approximately every 20 nm and ranged from 220 to 675 nm. Emission wavelengths ranged from 10 nm beyond the excitation wavelength to 800 nm. Artifacts from grating harmonics and/or ghosts were removed from the recorded data. ⁷⁶ The maximum intensities of the EEMs have been normalized for comparison and the wavelengths limited for presentation clarity. Two dominant peaks were observed in both the BEAS-2B [Figure 2(a)] and BEAS-2B _NNK [Figure 2(b)] cells. These two peaks occurred at 220/330 and 287/330 nm (excitation/emission wavelengths). These peaks corresponded closely with the peaks observed in the EEM of pure tryptophan (220/350 and 287/350 nm), shown in Figure 2(c), and to those presented in the literature. ⁵⁰ The peak at longer-wavelength excitation is reported to be the result of two closely spaced levels, the ¹ L _a and ¹ L _b of the indole moiety, while the shorter-wavelength peak is the result of a higher electronic transition. ⁵⁰

Figure 2

EEMs of (a) BEAS-2B cells, (b) BEAS-2B _NNK cells, and (c) pure tryptophan in PBS. Spectra were collected for excitation wavelengths from 220 to 675 nm and emission wavelengths to 800 nm. The dominant feature in both immortalized and tobacco-carcinogen-transformed cells EEMs corresponded closely to the tryptophan EEM.

Figure 3 shows excitation and emission spectra for pure tryptophan in PBS [Figures 3(a) and 3(b)] and BEAS-2B (solid) and BEAS-2B _NNK (dashed) cells in PBS [Figures 3(c) and 3(d)]. The spectra were recorded at the maxima of emission or excitation, as determined by the corresponding EEM. The autofluorescence spectra recorded from both bronchial epithelial cell lines closely resembled the characteristic fluorescence of the amino acid tryptophan.

Figure 3

(a) Excitation spectrum of pure tryptophan in PBS, emission wavelength at 350 nm. (b) Emission spectrum of pure tryptophan in PBS, excitation wavelength at 287 nm. (c) Excitation spectra of BEAS-2B cells in PBS (solid) and BEAS-2B _NNK cells in PBS (dashed), emission wavelength at 330 nm. (d) Emission spectra of BEAS-2B cells in PBS (solid) and BEAS-2B _NNK cells in PBS (dashed), excitation at 287 nm.

A minor autofluorescence feature observed in both the BEAS-2B and BEAS-2B _NNK cell suspensions was observed with an emission maximum at 655 nm, upon excitation at 460 nm. This broad featureless emission (full width at half maximum of 150 nm) was approximately 40 times weaker than that from the emission band at 330 nm [see the peaks in Figures 2(a) and 2(b)]. To investigate the origin of this emission, the excitation spectrum was recorded (610 nm emission) between 325 and 600 nm. This range was chosen based on values expected for porphyrins and lipofuscin, two possible endogenous fluorophores roughly corresponding to the observed emission. ² No distinct excitation feature corresponding to reported endogenous fluorophores was found in this range. Further studies on these very small signals are needed to assign the emission observed at 655 nm.

No appreciable difference in autofluorescence emission intensity was noted between the BEAS-2B and BEAS-2B _NNK human bronchial epithelial cells, for either the tryptophan or the 655 nm band. In addition, spectrofluorimetry revealed no emission from other common endogenous fluorophores, such as NADH and flavins. This was presumably due to the low sensitivity of the LS-50 spectrofluorimeter. In recent studies on human cervical epithelial cell suspensions, tryptophan emission was found to be at least two orders of magnitude higher than the fluorescence emission from either NADH or FAD. ⁵⁹ To measure the weaker emission of other endogenous cellular fluorophores, we employed sensitive confocal fluorescence imaging methods, described next.

3.2.

Confocal Fluorescence Microscopy

Using the Leica confocal microscope described above, we measured the autofluorescence of individual BEAS-2B and BEAS-2B _NNK cells. This highly sensitive spectrophotometer-coupled instrument allowed for the detection and spatial localization of distinct fluorophore species and excluded any appreciable contribution from other species, via tunable wavelength discrimination. Figure 4 shows two differential interference contrast (DIC) images of BEAS-2B cells, with corresponding overlays of pseudocolor confocal fluorescence images. The images in Figure 4(a) were collected using an excitation wavelength of 351 nm and emission intensity integration between 426 and 454 nm. The images revealed granular cytoplasmic autofluorescent structures with excitation/emission characteristics that corresponded closely with known NADH values. Figure 4(b) shows the same cells excited at 488 nm, with the resulting emission integrated between 530 and 555 nm. The emission in this wavelength range was consistent with the presence of flavins in the cells. From these images, it is apparent that the instrument was capable of detecting emission from these weakly fluorescing endogenous species present in human bronchial epithelial cells. Attempts at recording emission at wavelengths greater than 610 nm (488 nm excitation wavelength) were unsuccessful.

Figure 4

Differential interference contrast images of immortalized human bronchial epithelial cells (BEAS-2B) with pseudocolor overlays of autofluorescence detected by confocal microscopy. (a) Cells were excited at 351 nm and the fluorescence emission was monitored between 426 and 454 nm. The granular, cytoplasmic autofluorescence detected (shown in red) corresponded closely to endogenous NADH. (b) The same cells were excited at 488 nm and the fluorescence emission was monitored between 530 and 555 nm. The granular, cytoplasmic autofluorescence detected (shown in green) corresponded closely to endogenous flavin emission. The cell on the left is in prophase of mitosis, as evidenced by the condensed chromosomes that are clearly visible.

Confocal autofluorescence images comparing BEAS-2B and BEAS-2B _NNK cells are shown in Figure 5. Figure 5 shows emission consistent with both NADH (excitation wavelength at 351 nm, emission integrated between 426 and 454 nm) and flavins (excitation wavelength at 488 nm, emission integrated between 530 and 555 nm) for the (a) BEAS-2B and (b) BEAS-2B _NNK cells. DIC images of the corresponding fields are shown at the bottom of Figure 5, for reference. All microscope and laser parameters were held constant between corresponding images, which were acquired within minutes of one another. New areas were imaged on each sample to avoid photobleaching, and optical sections were chosen to be similar by focusing on nucleoli. These images indicate that the tobacco-carcinogen-transformed BEAS-2B _NNK cells exhibited reduced NADH and flavin emission as compared to the immortalized BEAS-2B human bronchial epithelial cells. In addition, the punctate cytoplasmic fluorescent granules present in the immortalized cells appeared to be more dispersed in the carcinogen-transformed cells than in parental BEAS-2B cells. The microscopy system’s configuration did not permit direct measurements of the excitation laser intensity or fluorescence emission intensity at the sample. This prevented any comparison between the relative NADH and flavin fluorescence intensity in situ.

Figure 5

Confocal autofluorescence microscopy images of (a) BEAS-2B and (b) BEAS-2B _NNK cells. The top images show emission consistent with endogenous NADH. Relative to the BEAS-2B cells, a marked decrease was observed in the BEAS-2B _NNK cells. Similarly, a loss in cytoplasmic granularity of the autofluorescence was observed. The middle images show emission consistent with endogenous flavins. A similar autofluorescence decrease was observed. The bottom figures show corresponding DIC images of each field.

4.

Discussion

To our knowledge, this study was the first investigation of the autofluorescence characteristics of immortalized BEAS-2B and the carcinogen-transformed BEAS-2B _NNK human bronchial epithelial cells. These cells serve as models for studies seeking to distinguish between immortalized and transformed (malignant) human bronchial epithelial cells. ^{72 73 74} The utility of this model system has been noted previously in the literature. For example, studies have used these cell lines to monitor responses to chemopreventive agents such as retinoids. ^{72 73 74} Since BEAS-2B _NNK cells were derived from BEAS-2B cells after carcinogenic transformation, it is possible to make direct comparisons between these two cell lines. This has led to the identification of a chemopreventive pathway that affects cell cycle progression at G1. ^{73 74} Based on these findings, it was hypothesized that aberrant expression of these cyclins would occur in bronchial pre-neoplasia. This hypothesis has been recently confirmed. ⁷⁵ Thus, these cellular models may reflect closely changes that occur during lung carcinogenesis in vivo.

The spectrofluorimetric results suggested that the predominant emission from both the BEAS-2B and BEAS-2B _NNK cells arose from tryptophan residues. Figures 2 and 3 showed that pure tryptophan emission was at approximately 350 nm, while that of the cells was shifted by 20 nm to shorter wavelengths (higher energies). Similar shifts have been noted in proteins ³⁸ and human laryngeal epithelial cells. ⁵⁶ In both of these studies, a shift of approximately 10 nm to shorter wavelengths, relative to free tryptophan, was observed. Our observations agreed with these previous studies, and found that the shifts occurred in both the immortalized and the carcinogen-transformed human bronchial epithelial cells. This well-known shift is the result of changes in the local chemical environment and conformational conditions of the protein-bound tryptophan, relative to the free amino acid. ^{50 77} Shifts to higher energy result from orientation constraints of the amino acid, since limited molecular rearrangement subsequently raises dipole interaction energies. ⁵⁰

In general, the use of endogenous cellular fluorescence to discriminate between normal and malignant cells relies on monitoring changes in fluorescence characteristics. Two studies that were specific to human oral epithelial cells found little or no change in the emission attributed to tryptophan. ^{55 78} In contrast, a study on human laryngeal cells reported a decrease in the longer-wavelength edge of the tryptophan emission of malignant cells relative to nontumorigenic cells. ⁵⁶ For the human bronchial epithelial cells studied here, we did not observe any change in tryptophan excitation or emission fluorescence between immortalized and carcinogen-transformed cells (Figures 2 and 3).

We note that the much weaker emission band centered at 655 nm was highly reproducible in the two human bronchial epithelial cell lines studied. The origins of this band are unknown, as are emission peaks observed in other cell types. ⁵⁸ Further investigations concerning this emission band will be the subject of future work.

Confocal fluorescence imaging was used to measure weak autofluorescence contributors undetected by spectrofluorimetry and to study the spatial localization of these fluorophores in individual cells. Here, we report an appreciable decrease in the NADH autofluorescence signal of the tobacco-carcinogen-transformed BEAS-2B _NNK cells relative to BEAS-2B cells (Figure 5). Although detailed findings have varied depending on cell type and fluorophore monitored, previous studies have reported generally less tryptophan, NADH, and/or flavin autofluorescence emission from malignant cells than from normal cells. ^{45 56 79 80} For example, a decrease in NADH autofluorescence was observed in malignant human laryngeal epithelial cells relative to nontumorigenic human laryngeal epithelial cells. ⁵⁶ It has been suggested that the reduction in NADH autofluorescence is most likely due to changes in the redox state of the coenzyme. ^{1 81} A loss in autofluorescence may suggest a concentration shift in the carcinogen-transformed cell’s reduced NADH to the oxidized (and low quantum yield) NAD+. We note that in the studies reported here, data were collected on various proliferative states of both BEAS-2B and BEAS-2B _NNK cells, with decreases similar to that shown in Figure 5 recorded for each state.

In contrast to the observations reported above, a recent report compared steady-state and time-resolved autofluorescence in normal human bronchial epithelial (HBE) cells, nonmetastatic, non-small cell lung cancer (NSCLC) cells, and metastatic small cell lung cancer (SCLC) cells. ⁷⁰ The study found that NSCLC cells emitted more NADH and tryptophan fluorescence than normal HBE cells, while SCLC emitted more NADH fluorescence and similar levels of tryptophan fluorescence, when compared to normal HBE cells. It is important to point out that these measurements were reported on one sample only and were made at high cell counts using a spectrofluorimeter with front-face collection geometry. In our confocal fluorescence microscopy studies, we observed a consistent decrease in the carcinogen-transformed cell fluorescence versus immortalized cell fluorescence over many measurements using independent cellular cultures.

Finally, as seen in Figures 4 and 5(a), the BEAS-2B cells exhibit cytoplasmic, granular localization of the NADH fluorescence. This has been shown previously in a number of cell lines and was reported to be partially, but not totally, localized in the mitochondria. ^{42 44 82} It is interesting to note that granular cytoplasmic autofluorescence appeared less intense and less localized in the BEAS-2B _NNK cells, relative to the BEAS-2B cells. A clear example is seen at the top of Figure 5(b). This loss of granular structure may be the result of changes in the mitochondrial morphology. A recent study noted that normal Galliera rat fibroblasts contain many filamentous mitochondria, while rat sarcoma lines contain a random cytoplasmic distribution of short spherical mitochondria. ⁴⁴

An appreciable reduction of the total flavin autofluorescence of the carcinogen-transformed human bronchial epithelial cells was also evident in Figure 5. A similar decrease in cellular flavin emission, from tumor human urothelial cells relative to normal human urothelial cells, was noted previously. ⁴⁵ It was suggested, based on previous work, ⁷⁹ that flavin fluorescence may be reduced in tumor cells, because deficient aerobic oxidation systems like tumors show reduced riboflavin concentrations. A second possible contributing factor may be redox changes similar to those hypothesized with NADH. Thus, biochemical pathways in the carcinogen-transformed cells may cause changes in the fluorescence by reducing FAD to the nonfluorescent FADH ₂ . ^{45 62} Our data are consistent with the observation of reduced total flavin emission in malignant versus normal cells.

These changes in cellular autofluorescence between immortalized and carcinogen-transformed human bronchial epithelial cells may indicate a potential method for noninvasively detecting pre-neoplasia in lung tissue using fluorescence spectroscopy and imaging. Much of the work on human bronchial tissue fluorescence ^{13 14 15 16 62 69} has revealed a consistent decrease in the autofluorescence of cancerous lesions versus normal bronchial tissue. ⁶² One cannot attribute all of the differences observed in tissues only to cellular fluorophores. Also contributing to measured tissue autofluorescence are fluorophores in the extracellular matrix, as well as morphological changes associated with pre-neoplastic tissue (including mucosal thickening and increased hemoglobin content) that are likely to influence light absorption and scattering. Nonetheless, the studies described here represent the first investigation of the autofluorescence properties of human bronchial epithelial cell lines that serve as model systems for neoplastic changes. By monitoring the changes in fluorescence emission from distinct fluorophores, such as NADH and flavins, or combinations of fluorophores, a sensitive and specific method for pre-neoplasia detection in the lung may be developed.

5.

Conclusion

Using both spectrofluorimetry and confocal fluorescence microscopy, we characterized, for the first time, the spectral cellular autofluorescence properties of immortalized and distinct tobacco-carcinogen-transformed human bronchial epithelial cells. Differences were observed between the two cell lines, as detected in confocal autofluorescence images that were specific to NADH and flavin emission. The observed loss in autofluorescence for carcinogen-transformed cells relative to immortalized bronchial epithelial cells may be the result of changes in fluorophore oxidation state. Future studies with improved sensitivity will seek to quantify both steady-state and time-resolved autofluorescence characteristics of these and other relevant cell lines, including chemoprevented cells and primary cultures of neoplastic cells.