2.1.

Preparation of Specimens

Clinically healthy adult (about 3 kg body weight) male gray chinchilla rabbits ($n=13$) were kept under standard housing conditions. Arthritis was induced in the right knee of the animals ($n=9$) by applying an antigen following the procedures described in Ref. 23. Based on the clinical signs of the joint swelling, treatment with methyl 5-aminolevulinate hydrochloride (ALA-Me) (Fluka Chemie GmbH, Germany) was performed not earlier than 3 days after the injection of ovalbumin. ALA-Me was dissolved in a sterile saline ($c=16\mathrm{mM}$, $2.92\mathrm{mg}/\mathrm{ml}$, i.e., $0.97\mathrm{mg}/\mathrm{kg}$) and 1 ml of the solution was injected intra-articularly into the inflamed knee. After in vivo spectroscopic measurements,²⁴ one rabbit treated with ALA-Me and one control rabbit from the group of healthy untreated rabbits ($n=4$) were selected for the visualization study of sensitized and control tissues. The rabbits were euthanized with the intravenously administered thiopental ($50\mathrm{mg}/\mathrm{kg}$) (Biochemie GmbH, Vienna, Austria). Ten specimens were prepared from both the synovium and cartilage tissues of each rabbit. The study was approved by the Lithuanian Laboratory Animal Use Ethical Committee under the State Food and Veterinary Service (No. 0218).

The microscopy and spectroscopy of endogenously produced porphyrins were performed ex vivo by measuring the fluorescence spectra of the synovium and cartilage specimens taken from healthy and inflamed rabbit knee joints. Specimens of the inflamed joint were taken 3 h after intra-articular injection of ALA-Me. Then control and sensitized specimens were immediately frozen with a spray, Solidofix®-Cryo (Carl Roth GmbH + Co. KG). Frozen tissues were microsectioned with a cryotome Leica CM1100 (Leica Biosystems, Germany). Frozen sections of $15\text{-}\mu \mathrm{m}$ thick were collected on glass-slides Histobond®+ (Carl Roth GmbH + Co. KG) and air-dried at room temperature. The same $15\text{-}\mu \mathrm{m}$ thick cryostat sections were studied by means of fluorescence intensity and lifetime microscopy as well as spectroscopy techniques. Later, the morphological structures of those samples were distinguished by express staining with hematoxylin 2 and eosin (Fisher Scientific, United Kingdom).

2.2.

Fluorescence Detection

Tissue specimens were imaged by means of a confocal fluorescence microscope “Eclipse TE2000” (Nikon, Japan), with a confocal scanning C1si system performing sequential scanning with a beam of a CW diode laser (405 nm) (56RCS/S2780, Melles Griot), using a $40\times$ objective Plan Apo, NA 0.95 (Fig. 1). The three-channel RGB detector (bandpass filters: 433 to 467 nm, 500 to 590 nm, and 621 to 755 nm for blue (B), green (G), and red (R) channels, respectively) was used for fluorescence detection. Image processing was done using the software “Nikon EZ-C1 Bronze v.3.80” and “ImageJ 1.43.”

A pulsed diode laser emitting at 405 nm (LDH-P-C-405B, average power 1.6 mW, controlled by a driver PDL 800-B, PicoQuant GmbH, Germany) was coupled to the microscope to measure the lifetimes of tissue autofluorescence and sensitized tissue fluorescence. The pulse repetition rate was set to 20 MHz (pulse energy 80 pJ), which allowed using a measurement time interval of 50 ns. FLIM imaging was performed using a time-correlated single photon counting (TCSPC) module PicoHarp 300, and 3200 channels of a detector were used to register a fluorescence decay curve. A single channel single photon avalanche diode (SPAD) detection unit was used for photon counting at $650/160$- or $637/13\text{-}\mathrm{nm}$ spectral ranges (FWHM) by changing filters. Each fast FLIM image representing the average fluorescence lifetimes was obtained by collecting 1000 counts at the peak value, and the image resolution was set at $512\times 512\text{pixels}$ (the typical acquisition time for healthy tissues $\sim 20\min$, for sensitized tissues $\sim 2\min$). The different colors of each pixel in the fast FLIM images denote the different average lifetimes and the mean arrival time of fluorescence photons after the excitation pulse. The intensity of each pixel is proportional to the overall fluorescence intensity. The fluorescence decay curve, representing data from all image pixels, was fitted with a three-exponential decay function $\{I(t)=A_0+A_1*\exp [-(t-t_0)/{\tau }_1]+A_2*\exp [-(t-t_0)/{\tau }_2]+A_3*\exp [-(t-t_0)/{\tau }_3]\}$ to obtain characteristic fluorescence lifetime values for the image. A three-exponential decay function provided a superior fit with smaller weighted residuals and a ${\chi }^2$ close to unity in comparison with a biexponential decay function. A three-exponential reconvolution fit was performed on each pixel using the obtained fluorescence lifetimes ${\tau }_i$ to calculate the FLIM images by means of the software SymPhoTime v.5.2 (PicoQuant GmbH, Berlin, Germany). FLIM images are displayed in three colors representing three calculated fluorescence lifetimes. The brightness of each color corresponds to the amplitude of the calculated lifetime. The instrument response function (IRF) was measured using a dye, erythrozine B.²⁷

A spectrometer QE65000 (Ocean Optics Inc.) was connected to the confocal scanning C1si system of the microscope to measure the integral fluorescence spectra using certain integration times during the sequential scanning of samples with a 405 nm beam of the CW diode laser. The light emitted from the tissue was collected through a 500 nm long-pass filter (“ЖC 18,” LOMO, Russia) and directed to the tip of a fiber connected directly to the spectrometer.

3.1.

Healthy Cartilage

Fluorescence intensity microscopy and lifetime microscopy images together with a corresponding histological image of the cartilage specimen from a healthy rabbit are presented in Fig. 2. The highest registered autofluorescence intensity of healthy cartilage was detected in the green/orange spectral region (520 to 590 nm) which falls into a green channel of the microscope, and no fluorescence was registered in the red spectral region (621 to 755 nm) [Fig. 2(a)]. Based on the histological image [Fig. 2(b)], the tissue was defined as a healthy transitional layer of articular cartilage. Dark spots seen throughout the histological image are cells’ (namely chondrocytes) nuclei, and the light pink area is the extracellular matrix. The brighter green spots seen in the fluorescence image clearly correspond to the dark spots seen throughout the histological image [Fig. 2(b)]. Fast FLIM imaging [Fig. 2(c)] revealed that the average lifetimes of intrinsic fluorophores of the healthy cartilage were limited between 5 and 10 ns [Fig. 2(d), curve I]. The cells of a healthy cartilage had shorter average lifetimes than extracellular matrix. Since a fast FLIM technique allows one to gate average lifetimes, a combined fast FLIM image has been split into several pictures depicting average fluorescence lifetime values in shorter intervals. As is seen in the gated images [Figs. 2(e) and 2(f)], the specific structures distinguishable in tissue specimens at average lifetimes between 5 and 6.5 ns [Fig. 2(e)] correspond to cells, while the structures seen at lifetimes between 6.5 and 10 ns [Fig. 2(f)] can be attributed to extracellular matrix when their locations are compared with the histological image.

Fig. 1

Measurement system [adapted from www.picoquant.com].

Fig. 2

Images of the specimens of healthy rabbit cartilage: (a) a fluorescence image taken combining G and R channels; (b) a H&E stained histological image; (c) a fast fluorescence lifetime imaging microscopy (FLIM) image in $650/160\text{-}\mathrm{nm}$ spectral range; (d) normalized distribution of average fluorescence lifetimes of healthy cartilage in $650/160\text{-}\mathrm{nm}$ spectral range (curve I) as well as sensitized cartilage in $650/160\text{-}\mathrm{nm}$ (curve II) and $637/13\text{-}\mathrm{nm}$ (curve III) spectral ranges, time gates: yellow—3 to 5 ns, blue—5 to 6.5 ns, green—6.5 to 10 ns, and red—10 to 15 ns; (e), (f) time-gated fluorescence lifetime images (gates are shown on pictures). Total magnification $40\times$.

3.2.

Sensitized Cartilage

Typical fluorescence spectra registered ex vivo on the healthy and sensitized inflamed cartilage specimens taken postmortem are presented in Fig. 3(a). Contrary to the spectra of the healthy rabbit specimens, the highest fluorescence intensity of sensitized cartilage was detected in the red spectral region, which falls into the range of the red channel of the microscope (620 to 700 nm). Two distinct bands in the fluorescence spectra clearly indicated the presence of endogenous porphyrins. It is well known that the typical fluorescence spectrum of PpIX in tissue possesses a dominant fluorescence peak at 635 nm with a second peak around 700 nm.^28,29 The microscopic fluorescence imaging of the sensitized cartilage [Fig. 3(b)] revealed, however, only weak fluorescence in the red channel. This fluorescence pattern corresponds very well with the intratissue location of the cells identified from the histological image [Fig. 3(c)]. Besides the dark spots seen throughout the histological image that are cell nuclei, and the light pink area resembling the extracellular matrix as in the control cartilage, a tangential zone of the cartilage with flattened chondrocytes is seen in the top of the image. Below it lies a transitional cartilage zone, while a calcified cartilage and a subchondral bone (dark area) are present in the bottom of the sample. No damage of cartilage is seen in the histological image.

Fig. 3

Images of the sensitized cartilage specimen: (a) normalized fluorescence spectra of healthy and sensitized cartilage, color areas represent the spectral ranges of fluorescence microscope “green” (G) and “red” (R) channels; (b) a fluorescence image taken combining G and R channels; (c) H&E stained histological image; (d) a fast FLIM image in $650/160\text{-}\mathrm{nm}$ spectral range; (e)–(h) time-gated fluorescence lifetime images (lifetimes gates shown on pictures). Total magnification $40\times$.

Figure 3(d) represents a fast FLIM image of sensitized inflamed cartilage, which is completely different from a healthy cartilage image with regard to emerged shorter and longer average lifetimes. The distribution of average lifetimes of fluorophores in the sensitized cartilage in a 650/160-nm spectral range was limited between 3 and 13 ns [Fig. 2(d), curve II]. The gating of fast FLIM images [Figs. 3(e)–(h)] revealed that the shortest lifetimes in the range of 3 to 5 ns belong to the fluorescing subchondral bone structure [Fig. 3(e)]. The pattern of the extracellular matrix [Fig. 3(g)] can be distinguished at the same average lifetimes as in the healthy cartilage—between 6.5 and 10 ns [Fig. 2(f)]. However, contrary to the gated image of healthy cartilage, lifetimes between 5 and 6.5 ns cannot be attributed to the pattern of cells, but to a dim pattern of calcified cartilage. Instead, the cells of the sensitized cartilage can be distinguished in the image gated at the longest fluorescence lifetimes—10 to 15 ns [Fig. 3(h)], and this pattern corresponds to the location of cells in a histological image [Fig. 3(c)] as well as with a red fluorescence pattern in a fluorescence image [Fig. 3(b)]. The fluorescence spectra of sensitized cartilage [Fig. 3(a)] imply that the longest fluorescence lifetimes in the $650/160\text{-}\mathrm{nm}$ spectral range are determined by endogenously synthesized PpIX, which accumulates in cells.

To assure that the longest average fluorescence lifetimes registered in specimens of the sensitized cartilage do belong to PpIX, the fast FLIM imaging of the same sample has been repeated in a narrow spectral range—$637/13\mathrm{nm}$ [Fig. 4(b)]. At this spectral range, the highest fluorescence intensity originates from PpIX, while the autofluorescence of cartilage is weak [Fig. 4(a)]. The distribution of measured average fluorescence lifetimes in the sensitized cartilage was limited between 10 and 16 ns [Fig. 2(d), curve III], including even longer lifetime values than those observed at the broader spectral range. Gating of fast FLIM images at the longest fluorescence lifetimes (10 to 15 ns), revealed the same pattern of intracellular fluorescence in both broad and narrow spectral ranges [Figs. 3(h) and 4(c)].

Fig. 4

(a) A fluorescence spectrum of sensitized cartilage, colored areas represent the spectral ranges of FLIM: green—$650/160\mathrm{nm}$ and red—$637/13\mathrm{nm}$; (b) a fast FLIM image in $637/13\text{-}\mathrm{nm}$ spectral range; (c) a gated fluorescence lifetime image of sensitized cartilage in $637/13\text{-}\mathrm{nm}$ spectral range; (d) an FLIM image: green color—${\tau }_1=0.8\mathrm{ns}$ (relative intensity $A_1$—3%), blue color—${\tau }_2=3\mathrm{ns}$ ($A_2$—13%), and red color—${\tau }_3=15.8\mathrm{ns}$ ($A_3$—84%). Total magnification $40\times$.

The fitting of the measured average fluorescence decay curves with a three-exponential decay function yielded the characteristic temporal values ${\tau }_i$ for the fluorescence decay in healthy and sensitized tissues. Those values calculated for all studied tissues are presented in Table 1. The amplitudes of characteristic lifetimes are not indicated due to the big variation throughout each specimen.

Table 1

Fluorescence lifetimes of fluorophores in samples of synovium (n=10) and cartilage (n=10) tissues, calculated from FLIM measurements in 650/160- and 637/13-nm spectral ranges. Data expressed as mean±SD.

The FLIM image of sensitized cartilage recalculated after the detection in a 637/13-nm spectral range [Fig. 4(d)], resembled the image recalculated when the imaging has been performed in a $650/160\text{-}\mathrm{nm}$ spectral range [Fig. 5(e)]; however, the calculated ${\tau }_i$ and the relative contributions of their amplitudes were different [Figs. 4(d) and 5(h)]. The value of the middle lifetime ${\tau }_2$ (3 ns) became shorter in comparison with the ${\tau }_2$ value obtained for a broader spectral range (Table 1), while the values of the longest lifetime ${\tau }_3$ were essentially the same.

Fig. 5

(a and b) Fluorescence lifetime images of the sensitized cartilage, the marked area represents region of interest (ROI) before (a) and after (b) exposure; (c) normalized distribution of average fluorescence lifetimes of sensitized cartilage in ROI before (I) and after (II) photobleaching in $650/160\text{-}\mathrm{nm}$ spectral range, time gates: yellow—3 to 5 ns, blue—5 to 6.5 ns, green—6.5 to 10 ns, and red—10 to 15 ns; (d) fast FLIM image in $650/160\text{-}\mathrm{nm}$ spectral range; FLIM images before (e) and after (f) exposure: green color—${\tau }_1=1.2\mathrm{ns}$, blue color—${\tau }_2=4.2\mathrm{ns}$, and red color—${\tau }_3=15.87\mathrm{ns}$. Total magnification $40\times$; (g) the changing fluorescence spectra registered during exposure of sensitized cartilage to 405 nm light; (h) fluorescence lifetimes and their relative amplitudes calculated before (A1) and after (A2) exposure.

To elucidate the effect of illumination on the fluorescence lifetimes during the microscopic observation of sensitized specimens, they were repeatedly scanned for a period of 10 min with a 405 nm beam of the CW laser connected to the scanning fluorescence microscope, the same system that had been used for fluorescence imaging.

Figures 5(a) and 5(b) show the fast FLIM images of the sensitized cartilage with a marked region of interest (ROI) before and after its extended exposure to the laser radiation. The initial distribution of average fluorescence lifetimes in ROI of the sensitized cartilage was limited between a range of 6.5 to 13 ns [Fig. 5(c), curve I], which is typical for other specimens of sensitized cartilage [see Fig. 2(d)] in regard to the longest fluorescence lifetimes. However, due to the absence of calcified cartilage and subchondral bone, there were no fluorescence lifetimes shorter than 5 ns. The distribution of average fluorescence lifetimes in ROI after the exposure shifted to shorter lifetimes (4 to 10 ns) [Fig. 5(c), curve II] and became like that in healthy cartilage [Fig. 2(d), curve I].

Comparison of the fast FLIM image of the sensitized cartilage specimen with the FLIM image indicated the resemblance between the distribution patterns of the fluorophores with the longest average lifetimes and those of the fluorophores with the calculated longest lifetime ${\tau }_3$ [Figs. 5(d) and 5(e)]. The highest amplitude of the fluorescence signal corresponding to the longest lifetime ${\tau }_3$ was detected in the cells [Fig. 5(e)]. The decreased PpIX fluorescence bands and the increased fluorescence bands of a photoproduct were seen in the fluorescence spectra [Fig. 5(g)] registered after the exposure of specimens to 405-nm radiation. The FLIM image calculated after the exposure [Fig. 5(f)] strongly differed from that calculated before the exposure [Fig. 5(e)]. This difference was reflected to the different inputs of the amplitudes of three calculated decay times. The highest amplitude before exposure was of ${\tau }_3$, while after exposure that of ${\tau }_2$ took the lead [Fig. 5(h)].

3.3.

Healthy Synovium

The fluorescence image as well as fast FLIM and histological images of the synovium specimen from a healthy rabbit are shown in Fig. 6. The green tissue autofluorescence was dominant while the red fluorescence was not distinguished in the combined fluorescence image [Fig. 6(a)]. Dark spots seen throughout the histological image mark discrete synoviocytes, and the light pink area represents fibrous elements of the synovial tissue [Fig. 6(b)]. The specimen was described morphologically as healthy synovium tissue. The characteristic average lifetimes of fluorophores in the fast FLIM image [Fig. 6(c)] of the healthy synovium were distributed from 4 to 8 ns [Fig. 6(d)]. The gating procedure of the fast FLIM image did not allow distinguishing between cells and fibrous elements of healthy synovium.

Fig. 6

Images of the healthy synovium specimen: (a) a fluorescence image taken combining G and R channels; (b) a H&E stained histological image; (c) a fast FLIM image in a $650/160\text{-}\mathrm{nm}$ spectral range, total magnification $40\times$; (d) normalized distribution of average fluorescence lifetimes of healthy (I) and sensitized (II) synovium in $650/160\text{-}\mathrm{nm}$ spectral range, time gates: green—4 to 8 ns and red—8 to 10 ns.

3.4.

Sensitized Synovium

The characteristic greenish autofluorescence was also dominant over dim red fluorescence in a sensitized synovium specimen [Fig. 7(a)]. The signs of tissue inflammation in the histological image showed itself as the activation of synovial fibroblasts and the accumulation of inflammatory cells, which were distinguished by the dark circular areas of nuclei [Fig. 7(b)]. However, the fast FLIM image of sensitized inflamed synovium [Fig. 7(c)] differed from that of healthy synovium by the presence of a fluorescence pattern with longer lifetimes. Average fluorescence lifetimes of fluorophores in the sensitized synovium were prolonged [Fig. 6(d)] in comparison with lifetimes of healthy synovium and were found to be limited between 4 and 10 ns. The gating of fluorescence lifetime images [Figs. 7(d) and 7(e)] revealed that the pattern of fibrous elements [Fig. 7(d)] can be distinguished at the same average lifetimes as in the healthy synovium—between 4 and 8 ns. The longest average fluorescence lifetimes—8 to 10 ns [Fig. 7(e)] were measured in the cells of the sensitized synovium, and this gated pattern corresponded well with the pattern of cells [Fig. 7(f)], which was retrieved from the histological image.

Fig. 7

Images of the sensitized synovium specimen: (a) a fluorescence image taken combining G and R channels; (b) a H&E stained histological image; (c) a fast FLIM image in a $650/160\text{-}\mathrm{nm}$ spectral range; (d and e) time-gated fluorescence lifetime images (lifetimes gates shown on pictures); (f) a time-gated fluorescence lifetime image (e) interlaced with a pattern of cells from a histological image; (g) an FLIM image: green color—${\tau }_1=1.3\mathrm{ns}$ (14%), blue color—${\tau }_2=4.1\mathrm{ns}$ (44%), and red color—${\tau }_3=15.4\mathrm{ns}$ (42%); (h) an FLIM image (g) interlaced with a pattern of cells from a histological image. Total magnification $40\times$.

As is seen in the calculated FLIM image of the sensitized synovium, the fluorophores with different fluorescence lifetimes were localized in different sites of the specimen [Fig. 7(g)]. In this specimen, the inputs of amplitudes of ${\tau }_2$ (4.1 ns) and ${\tau }_3$ (15.4 ns) in a $650/160\text{-}\mathrm{nm}$ spectral range were almost the same. The distribution pattern of the longest fluorescence lifetime ${\tau }_3$ in the calculated FLIM image showed a close resemblance to the pattern of cells in a histological image of synovium tissue [Fig. 7(h)].