Reporter Expression and Tissue Depth Quantification Using Bright and Broad-Range Spectrum Bioluminescence Probes

Bioluminescence imaging is widely used to study biological processes in small animal models. Such technique suffers from absorptive and scattering properties of tissues preventing to assess depth and expression of a given bioluminescent reporter. We developed a broad-band highly bioluminescent probe and studied its selective absorption according to tissue depth using tissue-mimicking phantom. Our probe, named YFP-Nluc consisting of the Yellow Fluorescent Protein (YFP) fused with the Nluc luciferase (Nluc), emits two distinctive peaks at 444 nm (Nluc) and 530 nm (YFP) upon bioluminescence resonance energy transfer (BRET). Upon expression in HEK cells, we show that the YFP-Nluc emission spectra is selectively and gradually modified upon increase of the hemoglobin concentration simulating multiple tissue depths. Furthermore, we found a linear correlation between the hemoglobin concentration and the bioluminescence ratio of YFP-Nluc corresponding to the spectra band-pass of 530-550 and 570-590 nm. Such probe opens possibility to determine, in the context of dual bioluminescence small animal imaging, the depth and signal strength without requiring the use of complex and time-consuming tomography based methods.


Introduction
In vivo optical imaging enables quantitative or semi-quantitative analysis of bioluminescent reporter probes to assess pathologies [1,2]. Bioluminescence imaging (BLI) approaches suffer from substantial limitations due to light scattering [3] leading to low spatial resolution and tissue absorption [4] affecting the signal intensity in relation to the source depths [5,6]. As a result, various technical methods have been developed to take into account those issues. First, to compensate for the loss of photons absorbed in tissues [7], in vivo imaging system are built with cooled CCD [8] camera enabling high sensitivity for deep image source or weak bioluminescence signal. Second, although the majority of in vivo bioluminescence studies do not take into consideration the depth of probe location while quantifying bioluminescence signal, few examples described in the literature are proposing alternative solutions using tomography.
Such method named Bioluminescence Tomography (BLT) [9] requires multiple acquisitions of bioluminescence performed with different angles. Next, the in vivo light path must be reconstructed using complex algorithms and mathematical formulas [10][11][12] leading upon small disturbances during the recording to significant errors [13]. As a result, lengthy time acquisition and complex kinetics of in vivo bioluminescence substrate, reported to lead to time-dependent variations of bioluminescence intensity [14] interfere with the 3D reconstruction.
To address this issue, costly multi-camera approach can be developed to collect simultaneously the bioluminescent signal from different angles and hence reduce the imaging time [15]. Alternative approaches using hyperspectral and multispectral bioluminescence optical tomography take advantage of the wavelength dependence of imaging depth [16,17]. Such method based on light collection through several narrow band-pass filters is not wildly adopted, although it was shown to reduce localization errors upon software reconstruction [18]. Interestingly as today, BLT 3D method applied in laboratories, mostly rely on dual-modality involving BLI together with MRI [19], SPECT [20] or micro-CT scan [19] in order to gather structural and anatomical information [21]. A solution that remains time-consuming and extremely expensive.
For this project, we decided to follow the multispectral bioluminescence strategy and develop a probe with sensitivity build in tissue depth signature. To do so, we engineered a bioluminescent fusion protein named YFP-Nluc composed of a bioluminescent donor Nanoluciferase (Nluc) and the Yellow Fluorescent Protein (YFP). Our model takes advantage of the bioluminescent and fluorescent reporters.
The Nluc extreme brightness known to be 150-fold superior to Renilla and firefly luciferases [22] provide sufficient bioluminescent signal to avoid the need for high cost cooled camera [23] and to suffer from long acquisition time due to narrow spectral detection. In addition, the YFP-Nluc bioluminescent reporter emits two distinctive peaks at 444 nm and 530 nm corresponding to the Nluc and YFP respectively. The resulting broad-band emission of YFP-Nluc improve the probe depth analysis upon emission spectrum as demonstrated in the following study with phantoms composed by gelatin, lipids and different concentrations of hemoglobin to simulate probe depth in tissues.
Phantom construction: Tissue mimicking phantoms were built using TBS Tri-buffered saline (50 mM Tris, pH 7.4 and 150 mM sodium chloride) as a buffer mixed with 10% gelatin and 0.1% sodium azide [24]. The solution was warmed up to 50°C while stirring until the complete dissolution of gelatin in the buffer. The solution was cooled down to 4°C while stirring and 1% intralipid was added as well as the desired amount of hemoglobin. The solution was poured into a 384 well plate (75 μL per well) and stored overnight at 4°C to allow the solidification of the phantom.
Hemoglobin absorbance spectrum: Several concentrations of hemoglobin diluted into TBS (from 0 to 34 g/L, dilution 1/10 compared to the phantom concentration) were poured in a black transparent 384 well plate (75 μL per well). Absorbance spectra of hemoglobin solutions were performed using SpectraMax M5 microplate reader (Molecular Devices, CA) from 350 to 650 nm with a 2-nm step increment.

Suspension and count of the transfected cells
The spent media was removed, HEK cells were washed using 10 μL PBS. 10 μL Trypsin-EDTA (Gibco, ThermoFisher Scientific, Waltham. MA) was added in each well. After rotating the 384 well plate to cover all the cell layer with trypsin, the plate was incubated at 37°C in a humidified atmosphere of 5% CO2 for 3 min. Transfected HEK cells were then collected in an Eppendorf tube with cell media. After 3 min centrifugation at 900 rpm, HEK cells were suspended in 100 μL of fresh cell media and counted using a hematocytometer.

Hemoglobin absorption spectrum
Hemoglobin, the main absorber of visible light in vivo, is wellknown to limit BLI sensitivity. Using the microplate reader SpectraMax M5, we measured the hemoglobin absorptive properties at different concentrations (0 to 8.5 g/L) in TBS (Tri-buffered saline), upon spectral analysis from 350 to 650 nm wavelength. As shown in the Figure 1A, the absorption spectra of hemoglobin at 2.1 g/L concentration and bellow were found to be within the non-saturating detection range of the SpectraMax M5.
Regardless of its concentration, the hemoglobin absorption showed a major pic at 410 nm and minor pics at 500 nm and above. As compared to the absorption spectra of hemoglobin, the intralipid (0.1%, red line) and the gelatin (10%, blue line) absorption spectra were less heterogeneous and showed a smooth drop of their absorption properties toward higher wavelengths.

Characterization of Nluc and YFP-Nluc Protein Emission Spectra Through Tissue-Mimicking Phantom
We next assessed the bioluminescence probes Nluc and YFP-Nluc emission spectra after absorption through tissue-mimicking phantom. In our experimental setup, live HEK cells transiently expressing our probes were maintained in cell media solution in presence of Furimazine substrate. The emitted light from the cells located on top of the solid phantom was characterized using the experimental setup as described in Figure 1B. The visible emission spectra region (350-650 nm) of our probes were first measured in presence of gelatin (10%) and intralipid scattering agent (1%), but in absence of hemoglobin ( Figures  1C and 1D  with two distinctive peaks: the first one at 444 nm and the second one at 530 nm. The first peak matched with the emission peak of Nluc at 444 nm and the second peak with the emission peak of YFP at 530 nm.
The presence of two peaks on the emission spectrum of the cells transfected with YFP-Nluc, highlighted BRET from Nluc to YFP. This energy transfer was enabled by the overlap between the emission spectrum of Nluc as a donor and the excitation spectrum of YFP as an acceptor [23].
We next aimed to characterize the spectrum of bioluminescence probes emitted through the solid phantom in presence of fix composition of gelatin (10%), and intralipid (1%), but variable concentrations of hemoglobin (main absorber in vivo) to simulate the optical properties of tissues. The complex absorption properties of hemoglobin ( Figure 1A) are expected to drastically influence the light propagation and absorption of our bioluminescence probes through the gelatin phantom according to the wavelength. As shown in ( Figures  1C and 1D), we normalized each emission spectrum (color lines) with their maximum emission at 444 nm (emission peak of Nluc) measured through the phantom in absence of hemoglobin (black bold curve).
The comparison between emission spectra of Nluc and YFP-Nluc after their absorption through the tissue-mimicking phantom, showed as expected, a major absorption of the emitted light with an increase of concentration of hemoglobin ( Figure 2). While focusing on the global shape of emission spectra of both Nluc and YFP-Nluc, we could notice the emission peak at 444 nm became more splayed with the increase of hemoglobin concentration ( Figure 1C). Furthermore, the relative intensity of the peaks at 444 nm and 530 nm seemed to be reversed ( Figure 1D). Indeed, the peak at 530 nm became predominant over the peak at 444 nm for higher concentration of hemoglobin.

Bioluminescence Wavelength Dependence of YFP-Nluc Hemoglobin Concentration
Each bioluminescence spectrum emitted from Nluc or YFP-Nluc through the phantom were next normalized using the maximum emission peak of the Nluc (Figures 3A and 3B). Upon observation of Nluc spectra the maximum intensity peak shift toward higher wavelength upon increase concentration of hemoglobin ( Figures 3A,  arrow-1). No significant correlation was found between the maximum emission wavelength of Nluc and the hemoglobin concentration (R2=0.4153).
Similarly, the first maximum intensity peak at 440 nm of the YFP-Nluc was shifted in presence of hemoglobin (332.8 g/L) by 16 nm ( Figure 3B, arrow-1). In Figure 1A, there is an emission wavelength shift of Nluc that is observed in the presence of hemoglobin. This can be explained by the steep drop of the hemoglobin absorption wavelengths above 450 nm which leads to a strong absorption of the bioluminescent signal of the Nluc bellow 450 nm. As a result, the 440nm maximum intensity peak of the Nluc is strongly reduced, and emission wavelengths above 450 nm are increasingly dominant in presence of hemoglobin.  3B). Next in order to quantify the spectral changes according to the hemoglobin concentration but independently from the bioluminescence signal, ratios of bioluminescence signal obtained with the narrow band-pass were calculated. As shown in Figures 3C and 3D, we found the best correlation between hemoglobin concentration and the ratio (570-590)/(530-550) and (570-590)/(450-470) for the YFP-Nluc probe with the R² value at 0.935 and 0.878 respectively. The (530-550)/(450-470) ratio from the Nluc or YFP-Nluc however showed poor correlation with the hemoglobin concentration. Our results showing that at given concentration of hemoglobin, YFP-Nluc BP3/BP-2 bioluminescence ratio strongly correlated to tissue-mimicking phantom hemoglobin concentration, suggest possibility to determine depth localization in living animal tissues. Furthermore, it should be noticed that the hemoglobin concentrations used in this experiment from 0 to 332 g/L covered the human blood which is approximately 160 g/L. However, in order to further validate YFP-Nluc bioluminescence probe additional experiments needed to be performed using (1) either solid optical phantoms as the ones used to calibrate optical imaging instruments or with (2) small animal models. Indeed, we should in future studies take into consideration tissue heterogeneity known to dramatically redirect the light path due to various tissue composition and density.

Conclusion
By comparing the proportion of light collected by the camera at different wavelength the YFP-Nluc probe such as already reported dual-wavelength fluorescent [25] or bioluminescent [26][27][28][29] probes enable a straightforward method to estimate the depth of the signal. Altogether the extreme brightness of the Nluc together with the broadspectra bioluminescent probe is an alternative to current in vivo optical tomography approaches to assess the depth of the reporter together with expression quantification without the need for expensive camera and tomography setup.