1.

Introduction

Noninvasive in vivo evaluation of tissue iron concentration in animal models is important for the study of iron-related disorders and their treatments.¹ Presently, magnetic resonance imaging (MRI) is the gold standard technique for tissue iron quantification in vivo. It measures iron indirectly by its effect on the $T1$, $T2$, and $T2*$ relaxation times of water.² Since most of the nonheme iron deposited in the tissues is stored inside the ferritin, and ferritin can be detected by photoacoustic devices, it was recently proposed by the use of photoacoustic molecular imaging (PMI) techniques to detect endogenous ferritin in vivo.³ Multispectral optoacoustic tomography (MSOT) is emerging as a powerful noninvasive PMI technique for resolving tissue chromophores at the high spatial resolution provided by ultrasound detection,^4,5 thus becoming an alternative technique for in vivo imaging in small animals.⁴ The method is based on the measure of the acoustic waves generated by the thermoelastic expansion of the environment surrounding molecules that absorb an electromagnetic radiation. The amplitude of the generated acoustic waves depends on the optical absorption profile of the tissue.⁶ When the emitted acoustic waves reach an ultrasound detector, they can be combined for the reconstruction of high-resolution images deep in tissue, since ultrasonic scattering is about two to three orders of magnitude weaker than optical scattering.^7,8 In vivo, the system needs probes that absorb in the near-infrared (NIR) region, as these wavelengths are able to penetrate through the tissues up to 1 cm deep. For example, oxy- and deoxy-hemoglobin are important endogenous chromophores with absorption bands in the NIR region that have been used to monitor blood flow and angiogenesis using photoacoustic approaches.⁹ Other endogenous photoacoustic probes present in vivo include melanin, bilirubin, and lipids.¹⁰ Melanin is located in the skin of nonalbino mice, bilirubin can be detected in the lumen of the intestine, and lipids show a photoacoustic signal at wavelengths above 900 nm. Recently, it was shown that ferritin overexpressed in cultured cells can be monitored by PMI due to its capacity to form an iron core that absorbs in the infrared region.³ This observation prompted us to verify whether ferritin can also be used as an in vivo endogenous probe, based on the fact that ferritin evaluation gives a reliable index of iron loading because its expression is strongly stimulated by iron, particularly in the liver.¹¹ In the present work, we used MSOT to detect and quantify liver ferritin/iron accumulation in vivo in mice before and after experimental iron loading.

2.

Materials and Methods

3.

Results and Discussion

Ferritin photoacoustic properties were recently analyzed in vitro in cells³ in the absence of hemoglobin (Hb) and oxygenated hemoglobin (${\mathrm{HbO}}_2$), the two main endogenous iron-containing in vivo photoacoustic absorbers.¹⁶ Thus, we started comparing the photoacoustic properties of iron-loaded ferritin with those of the Hbs. The absorption spectra of ferritin and Hb in the 680 to 900 nm range are similar with the exception of an Hb peak at 760 nm, while the ${\mathrm{HbO}}_2$ spectrum shows a wide absorption peak at higher wavelengths (Fig. 1).

Fig. 1

NIR absorption spectra of ferritin, Hb, and ${\mathrm{HbO}}_2$. The absorption and the photoacoustic spectra of ferritin and Hb in the 680 to 900 nm range are similar with the exception of an Hb peak at 760 nm, while the ${\mathrm{HbO}}_2$ spectrum shows a wide absorption peak at higher wavelengths.

For signal deconvolution, the ViewMSOT™ 3.6 software provided by the inVision-128 small animal scanner uses a linear algorithm. Unmixing the ferritin signal from that of ${\mathrm{HbO}}_2$ does not pose problems, while the unmixing from that of Hb may suffer of some limitations, since it is based only on the difference at 760 nm between their absorption spectra. Then, to verify whether ferritin/iron can be monitored by MSOT using our equipment (inVision-128 small animal scanner), twofold serial dilutions of ferritin and iron dextran were loaded in a scattering and nonabsorbing tissue-mimicking phantom device using water as the control reference. In the concentration range compatible with that expected in some tissues of living mice (95 to $380\mu \mathrm{g}/\mathrm{mL}$),¹⁷ the MSOT signal obtained was positively related to the loaded concentration of ferritin (Fig. 2) and iron (data not shown). The relationship between the photoacoustic signal obtained with the MSOT and the ferritin/iron concentration supported the possibility to use ferritin as an endogenous photoacoustic contrast agent in vivo in mice.

Fig. 2

Photoacoustic properties of ferritin/iron. (a) Photoacoustic tomographically reconstructed images. Ferritin dilutions were inserted in a scattering and nonabsorbing tissue-mimicking phantom device using water as control. (b) Photoacoustic signal quantification of ferritin. The ferritin/iron photoacoustic signal was obtained from the ROI area drawn in correspondence of the samples using the transverse photoacoustic images provided by ViewMSOT™ 3.6 software. The graph shows the relationship between the MSOT signal and the ferritin concentration (95 to $380\mu \mathrm{g}/\mathrm{mL}$).

Next, we analyzed six untreated mice of the CD1 albino background using the settings for ferritin/iron detection. The anatomical localization of the signal was obtained comparing the photoacoustic background image at 850 nm, where no ferritin/iron photoacoustic signal is present, with the corresponding transverse image of the mouse atlas provided by the ViewMSOT™ 3.6 software. This allowed precise drawing of ROI areas over the liver, spleen, and other organs, for the quantitation of the photoacoustic signal. An evident ferritin/iron photoacoustic signal was found in correspondence of the liver and spleen (Fig. 3), which are the major iron storage organs, but was not found in the kidneys and the heart (data not shown).

Fig. 3

In vivo photoacoustic tomographically reconstructed image of a transverse slice of an untreated mouse. (a) The grayscale image represents a single wavelength (850 nm) and was used as background for the anatomical localization of the signal. The colored scale shows the ferritin/iron photoacoustic signal. All the mice show a stronger ferritin signal (1) in the liver and (2) in the spleen. On the contrary, no ferritin absorbance was detected (3) in the kidneys. (b) Corresponding transverse image of the mouse atlas provided by ViewMSOT™ 3.6 software.

This result suggested that ferritin/iron could be quantified in the liver of living mice by photoacoustic analysis. To verify this, three CD1 mice were treated intraperitoneally (IP) with iron-dextran solution ($5\mathrm{mg}/\mathrm{IP}$, twice a week for three consecutive weeks) to induce iron overload, while the other three control mice were treated in the same way with saline solution. The skin color of the shaved iron-loaded mice at the end of the treatment was darker than that of the saline-treated ones (Ctrl) [Fig. 4(a)], and after the sacrifice the liver of iron-treated mice showed a color remarkably stronger than that of the matched controls [Fig. 4(b)].

Fig. 4

Analysis of the iron-treated and untreated mice. (a) The skin color of the shaved mice after the 3-week iron treatment was darker than that of the controls. (b) After the sacrifice, tissue samples were collected for biochemical analysis. A darker color in the liver of iron-treated mice is evident compared to the controls, while the color of the kidney is similar, indicating the correct perfusion of the animals. (c) Photometric iron evaluation shows a 15-fold increase of the liver nonheme iron concentration in the treated mice. (d) Immunoblotting on liver protein extracts, performed using the specific primary antibody, shows a 40-fold ferritin increase in iron-loaded mice.

In addition, postmortem analyses showed that nonheme iron content in liver was 15-fold higher in iron-treated mice compared to the controls [Fig. 4(c)], and the ferritin level measured by western blotting was about 40-fold above the basal level [Fig. 4(d)]. Altogether, the biochemical analyses confirmed that the iron treatment was effective and induced a massive iron loading in the liver. MSOT acquisition was performed for all the mice before, during, and at the end of the treatment, before the sacrifice. Figure 5 shows the MSOT cross sections of the mice after the 3-week iron treatment.

Fig. 5

In vivo and ex vivo photoacoustic analysis of the liver of iron-loaded and control mice. Photoacoustic tomographically reconstructed images of (d, e, f) iron-loaded mice show a stronger ferritin/iron signal both in the liver (right ROI) and in the spleen (left ROI) compared to (a, b, c) saline-treated mice. The gray-scale image taken at 850 nm was used for the anatomical localization of the signal. The colored scale shows the ferritin/iron photoacoustic signal. (g) Quantitation of the ferritin/iron photoacoustic signal was obtained from the ROI areas drawn over the liver using the transverse photoacoustic images provided by ViewMSOT™ 3.6 software. The mean photoacoustic signal of the liver of iron-treated animals was 1.6-fold higher than that of the control ones. A similar signal dispersion is observed in the iron-loaded mice before and after the treatment, while no differences are observed in the saline-treated ones. The iron-loaded mice showed an increase of the ferritin/iron photoacoustic signal also in peripheral areas in proximity of the liver and the spleen (yellow arrows). (h) Photoacoustic tomographically reconstructed images. Liver homogenates of iron-loaded and control mice were inserted in a scattering and nonabsorbing tissue-mimicking phantom. (i) Photoacoustic signal quantification of ferritin. The ferritin/iron photoacoustic signal was obtained from the ROI area drawn in correspondence of the samples using the transverse photoacoustic images provided by ViewMSOT™ 3.6 software. The graph shows a signal 1.5-fold higher in the iron-loaded mice compared to the controls, a difference not statistically significant.

A photoacoustic signal is evident in the ROI areas drawn over the liver and spleen region confirming that the ferritin/iron deposits are actually detected by the system in physiological conditions [Figs. 5(a), 5(b), and 5(c)]. A careful quantification of the pixel intensity of the similar ROI areas showed that the signal of the three control mice did not change significantly during the period of the treatment, while that of the iron-treated animals increased. In two mice it increased about 20%, and in one mouse it increased about 90%, with a mean signal 1.6-fold higher in the iron-loaded mice compared to the controls, a statistically insignificant difference [Fig. 5(g)]. A similar increase in the photoacoustic signal was also found analyzing the liver homogenates inserted in an agar phantom [Fig. 5(h)]. The iron-loaded mice showed an increase of the ferritin/iron photoacoustic signal also in peripheral areas in proximity of the liver and the spleen [Figs. 5(d), 5(e), and 5(f) yellow arrows], which was attributed to the skin and probably not directly linked to the iron overload. We are not aware that iron overload may cause an accumulation of iron or NIR probes in the skin; thus, further experiments are needed to verify the reasons leading to an increase of the photoacoustic signal in the skin after iron treatment.

In conclusion, we found that although MSOT can detect in vivo ferritin/iron overload in mice liver and spleen, its sensitivity is far too low for many applications, and is not comparable to that of postmortem analyses. Therefore, it seems to us unlikely that photoacoustic tomography could replace MRI for in vivo analysis of iron status in mice using endogenous ferritin as a photoacoustic reporter. The low sensitivity of MSOT in this study may be due to the low ferritin/iron absorbance in the NIR region compared to the high background signal given by Hb, which interferes with that of the ferritin.