Label-free whole-kidney metabolic optical imaging using in vivo insulated cryofixation and cryo-micro-optical sectioning tomography

Abstract

Cryo-imaging has the potential to obtain and visualize the metabolic state of the whole kidney without labeling. However, uneven fixation of metabolic information and incomplete organ morphology in three dimensions limit cryo-imaging application. Here, a pipeline of in vivo insulated cryofixation combined with cryo-micro optical sectioning tomography (cryo-MOST) was established to achieve uniform and complete cryofixation and three-dimensional visualization of renal metabolic mapping at a micron-scale resolution. By this pipeline, we discovered an increased renal redox ratio of db/db mice with type 2 diabetes, indicating the presence of metabolic disorders. The results demonstrate that our convenient optical imaging tool provides a micro-resolution, quantitative assessment of the metabolic state of the whole kidney and potentially extends to other organs.

1. Introduction

The kidney has a vigorous metabolism, second only to the heart regarding mitochondria abundance and oxygen consumption [1]. Disorders of energy metabolism usually accompany nephropathy. Thus, observing energy metabolic profiles is conducive to the pathological investigations of kidney diseases. Reduced nicotinamide adenine dinucleotide (NADH) and oxidized flavin adenine dinucleotide (FAD) are located at both ends of the mitochondrial respiratory chain, and function as crucial sources of free energy and electrons for its reactions [2], which are important indicators to characterize the metabolic state of the kidney.

Many acute kidney injuries (AKI) and chronic kidney diseases (CKD) have been reported to be related to mitochondrial damage and dysfunction [3]. In AKI, the significant decrease in the levels of NAD⁺ impedes energy generation and ultimately impairs the core function of selective transport in the kidney. Singh et al. reported significant reductions in ATP, NAD⁺, and the NAD⁺/NADH ratio in the whole-kidney of heme protein-induced AKI (HP-AKI) models [4]. Similarly, dysregulation of NAD⁺ metabolism in CKD may contribute to obesity, metabolic syndrome, diabetes, and diabetic end-organ damage [5]. Zhu et al. demonstrated that the Wld^S gene can improve metabolic disorder and morphology changes in kidney injury of type 1 diabetes mice by slowing the decrease of the NAD⁺/NADH ratio [6]. However, these reports were only able to obtain the results of NAD⁺/NADH concentration changes in the whole kidney by using NAD⁺/NADH Assay Kits, and their three-dimensional (3D) metabolic distribution information across different renal subregions was missing. Each renal subregion has unique biochemical compositions and metabolic rates, with distinct capacities to handle oxidative stress so that different renal subregions may respond differently to the diseases [7–9] . Given these differences, there is a compelling need for advanced imaging techniques to provide comprehensive 3D metabolic information to analyze the metabolic differences within these renal subregions.

Chance et al. found that the FAD/NADH ratio in isolated mitochondria and liver tissue significantly correlates with the NAD⁺/NADH ratio [10]. NADH and FAD exhibit endogenous fluorescence characteristics, and their fluorescence intensities correlate positively with their concentrations. Therefore, they developed fluorometry-based methods and proposed the redox ratio (RR) as an index to reflect the mitochondrial metabolic state [11]. It should be noted that the NADH and FAD signals of tissue undergo rapid changes within seconds of disrupting blood oxygen transport [12]. Therefore, the traditional optical imaging technology of isolated tissue cannot accurately obtain the in vivo metabolic information of the kidney. In addition, NADH and FAD exhibit weak fluorescence intensities at 20 °C, posing challenges for detection. Two-photon or multi-photon fluorescence microscopy enables imaging of NADH and FAD signals in vivo at 20 °C using a high-intensity pulsed laser [13]. However, imaging speed and depth limitations hinder the acquisition of metabolic information for the whole organ.

Cryo-optical microscopic imaging technology provides a novel approach to 3D metabolic imaging of whole organs. At low temperatures, biochemical reactions, including enzymatic activities, are significantly attenuated, so the sample's structural and metabolic information can be fixed by snap-freezing. Thus, to accurately reflect the frozen sample's in vivo metabolic state, minimizing the interval between regular cycle interruption and metabolic process cessation is crucial while maintaining the sample at low temperatures during imaging [10]. Different research groups have tailored cryo-imaging techniques to specific applications. Quistorff et al. built a redox ratio-scanning instrument based on liquid nitrogen (LN₂). They realized 3D metabolic imaging of 3 × 3 × 2 mm rat organs through sample milling and point-by-point scanning with movable microlight guides [10]. However, this point scanning is relatively time-consuming and not suitable for 3D metabolic imaging of large-scale samples. Xu et al. developed a charged coupled device cryogenic redox imager and demonstrated its application in cancer research [14]. They used LN₂ to freeze the whole anesthetized mice quickly, and then the tumors were surgically excised by a saw and performed cryo-optical microscopic imaging. They found that the RR of the aggressive tumors was much higher than that of the indolent tumors. Ranji et al. developed a 3D cryo-imager at −80 °C to obtain 3D RR images of type 1 diabetic mice kidneys and found that oxidative stress (OS) in the kidneys increases as diabetes progresses [15]. The team used isopentane pre-cooled by LN₂ to freeze the ex vivo samples [16]. These works extend the application of cryo-imaging to whole organs. However, some problems remain, such as uneven freezing, sample fracture, and the challenge of virtually maintaining an in vivo metabolic state in ex vivo samples, which reduces the integrity and reliability of cryo-imaging results. Therefore, there remains an urgent need to achieve both rapid and uniform cryofixation and 3D mesoscopic visualization of the kidney-wide structure and metabolic state, to further enhance our understanding of renal organ function and disease progression.

In this study, we developed a label-free whole-kidney metabolic optical imaging pipeline using in vivo insulated cryofixation and cryo-MOST. The in vivo insulated cryofixation method was employed to enhance both the freezing speed and uniformity of the kidney compared to traditional cryofixation methods, and cryo-MOST enables 3D metabolic and structural imaging of the frozen samples with micron resolution. Utilizing this pipeline, we compared the metabolic states in specific renal subregions between type 2 diabetic db/db mice kidneys and their control counterparts, db/m mice kidneys. The results highlight that the combination of in vivo insulated cryofixation and cryo-MOST offers a straightforward yet robust means to quantitatively evaluate the metabolic state and structure of the whole kidney. It potentially facilitates research in pathology and therapeutic interventions.

2. Materials and methods

2.1. Animals

The 8-week-old male C57BL/6J mice were purchased from Vital River Laboratories (Beijing, China). All male db/db mice (B6.BKS(D)-Lepr $<\mathrm{d}\mathrm{b}>$ /J) and age-matched heterozygote non-diabetic mice (db/m) were purchased from Changzhou Cavens laboratory animal co. LTD (Changzhou, China). All experimental procedures were performed according to animal experiment guidelines of the Experimental Animal Management Ordinance of Hubei Province, P. R. China, and have been approved by the Institutional Animal Ethics Committee of Huazhong University of Science and Technology.

2.2. Cryofixation method of mice kidneys

Here, we have developed in vivo insulated cryofixation methods to enhance the efficiency and uniformity of frozen kidneys. This method insulates the kidney from the surrounding tissues by positioning a specialized insulated component beneath it, as illustrated in Fig. 1(a). The upper part of this component is an open-box-type structure with a base plate and three side plates made of pure copper. The kidneys are placed in the open-box-type structure. The three side panels are designed to minimize cryogenic fluid's outflow, thereby enhancing cryogen usage efficiency. Furthermore, the heat at the renal dorsal can be quickly conducted out through the base plate, increasing the freezing rate at the renal dorsal. The lower part of the component is composed of two corner columns and a sloping top plate made of high-density polyethylene. The two corner columns can insulate the kidney from the tissue below the kidney. Air is an excellent insulating medium that can effectively block the heat transfer from the tissue to the kidney, conducive to freezing the bottom of the kidney. In addition, the space between the two corner columns also facilitates the harvest of frozen kidneys. The sloping top plate is designed to minimize the impact of the component on the position of the kidneys.

Fig. 1.

Label-free whole-kidney metabolic optical imaging pipeline. (a) Scheme of in vivo insulated cryofixation. (b) Scheme of cryo-MOST system.

To verify the effect of in vivo insulated cryofixation methods on mouse kidney freezing, we compared it with other traditional freezing methods. The process is as follows. The mice were anesthetized with 2% chloral hydrate and 10% ethyl urethane (8 mL/kg body weight, i.p.) and immobilized in a supine position. The abdominal cavity was opened to expose the kidney, and then the kidney was harvested according to different cryofixation methods as follows:

• (1)Ex vivo-Aluminum foil (EV-AF) cryofixation: Fold the aluminum foil into a small box shape and let it float on the surface of LN₂. The kidneys were harvested from the anesthetized mouse and placed in the aluminum foil box.
• (2)Ex vivo-LN₂ (EV-LN₂) cryofixation: The kidneys were harvested from the anesthetized mouse and then dropped directly into LN₂.
• (3)Ex vivo-Isopentane (EV-Iso) cryofixation: Isopentane was poured into an aluminum foil box floating on LN₂ and pre-cooled to −160 °C. The kidneys were harvested from the anesthetized mouse and dropped directly into pre-cooled isopentane.
• (4)In vivo-LN₂ (IV-LN₂) cryofixation: LN₂ was poured onto the kidneys while the anesthetized mouse remained alive, then the frozen kidneys were harvested.
• (5)In vivo-Isopentane (IV-Iso) cryofixation: Pre-cooled isopentane was poured onto the kidneys while the anesthetized mouse remained alive, then the frozen kidneys were harvested.
• (6)In vivo insulated-LN₂ (IVI-LN₂) cryofixation: Two insulated components were positioned beneath each kidney of the anesthetized mouse to keep the kidneys horizontal. LN₂ was poured onto the kidneys while the mouse remained alive, and then the snap-freezing kidneys were harvested.
• (7)In vivo insulated-Isopentane (IVI-Iso) cryofixation: The insulated components were placed as described above. Pre-cooled isopentane was poured onto the kidneys while the anesthetized mouse remained alive, and then the snap-freezing kidneys were harvested.

After being harvested, the kidneys were photographed on a grid paper (2 mm spacing) to assess their morphological integrity.

2.3. Cryo metabolic optical imaging

Figure 1(b) illustrates the schematic of the cryo-MOST system. Details of this system have been previously reported [8]. In short, the system comprises optical imaging, mechanical cutting, and other system components. We use halogen Lamp (X-Cite exacte, Olympus, Japan) as a light source. Multichannel imaging is achieved by rotating the filter wheel of a stereomicroscope (MVX10, Olympus, Japan). The NADH (excitation filter: FF01-360/12, Semrock, dichroic mirror: FF390-Di01, Semrock, emission filter: FF01-470/28, Semrock) and FAD (excitation filter: FF01-434/17, dichroic mirror: FF510-Di02, Semrock, emission filter: FF01-536/40, Semrock) autofluorescence channels are used for renal metabolic imaging. The excited fluorescence signal is collected by a sCMOS camera (ORCA-Flash 4.0, Hamamatsu, Japan). The ZOOM module of the stereomicroscope allows the lateral resolution to adjust between 0.52 - 5.16 µm/pixel. A 3D transmission stage drives the sample (X-axis: M531, Y-axis: M521, Z-axis: HPS-170, Physik Instrumente, Germany) to move between the milling and imaging modules. In this study, the lateral pixel resolution, axial step, and exposure time in cryo-MOST imaging are 4 × 4 µm, 200 µm, and 200 ms, respectively. Imaging takes approximately 3 minutes per layer, the overall imaging time is about 5 hours, and the duration may vary slightly depending on the size of the kidney.

2.4. Freezing rate measurement

To compare the freezing rate of different cryofixation methods, we used three T-type thermocouples (temperature probe diameter: 0.13 mm, operating temperature: -200 °C – −150 °C) (Jingdu Electric, China) and multichannel temperature acquisition module DAM-4501 (Xunyan Electronics, China) for temperature monitoring. We performed temperature monitoring at the ventral, center, and dorsal regions within the kidney. Given the kidney's thickness of approximately 4 mm in C57 mice (8 weeks old), the temperature monitoring points were strategically positioned at 0.5 mm, 2 mm, and 3.5 mm from the upper surface. Considering that biochemical reactions persist during the freezing process, we estimated the changes in kidney metabolites during freezing according to Arrhenius's empirical formula: [17]

$$k=A{e}^{\frac{-E_a}{R(T+273.15)}}$$ (1)

where k is the rate constant, A is the pre-exponential factor, R is the universal gas constant, about 8.314 J/(mol·K), T is the temperature expressed in centigrade (°C), and E_a is the activation energy of the reaction. Here, we take the catalase reaction as an example to calculate the changes in kidney metabolites. It has the minimum activation energy (29.274 kJ/mol) in known biological tissues, and its reaction rate is the least sensitive to temperature changes. We observed that the initial temperature of kidneys in mice was approximately 20 °C immediately after laparotomy, thus setting 20 °C as the initial temperature. Previous studies have shown that the fluorescence intensities of NADH and FAD in tissues below −100 °C change slowly, which is a suitable temperature window for metabolic imaging [18]. In this study, the cross-sectional temperature of the kidney was approximately −100 °C during the image acquisition process. Therefore, we mainly focus on comparing the metabolic changes of different cryofixation methods in the 20 °C – −100 °C temperature range. We define the state change, denoted as Δ_A, to represent the cumulative effect of temperature-dependent biochemical reactions over time:

$${\mathrm{\Delta }}_A=\sum _{T=-100}^{20}{k}_T{t}_T$$ (2)

where k_T is the biochemical reaction rate at temperature T; t_T is the duration at temperature T, expressed in seconds.

2.5. Cryo metabolic optical imaging

The 16-bit grayscale images of NADH and FAD channels for the kidney were analyzed using MATLAB (2017a, Mathworks). The generation of RR images includes the following four steps:

• Step1. Flat-field correction. Fluorescence images are corrected by the distribution map of illumination intensities obtained by imaging a standard fluorescent plate to eliminate the difference in fluorescence brightness due to uneven excitation illumination.

• Step2. Image segmentation. The cross-section image of the kidney was extracted, and the background pixels were set to 0 by using the Graph Cut semiautomatic segmentation tool in MATLAB.

• Step3. Image registration. Based on rigid registration, the FAD-channel image is translated and scaled to register with the NADH channel.

• Step4. Calculation of RR. Based on the dual-channel fluorescence image, the RR images can be calculated by the following formula [19]:

  $$R{R}_{F_N}=\frac{I_N}{I_N+I_F}$$ (3)

  where I_N and I_F are the intensities of NADH and FAD channels at the corresponding voxels in the image.

To analyze the uniformity of RR in the whole kidney, we randomly selected the image stacks of 400 × 400 × 600 µm at the cephalic, ventral, lateral, dorsal, and caudal sites from each renal cortex and calculated their mean of RR for each site, the statistical indicators were calculated as follows:

• (1)
  Mean:

  $$Mea{n}_{RR}=\frac{{\sum }_{i=1}^n{x}_i}{n}$$ (4)
• (2)
  Standard deviation:

  $$SD=\sqrt{\frac{{\sum }_{i=1}^n{(x_i-\stackrel{-}{x})}^2}{n-1}}$$ (5)

  where x_i is the i-th element and n is the sample size.
• (3)
  Relative standard deviation:

  $$RSD=\left(\frac{SD}{M\text{ea}{\text{n}}_{RR}}\right)\times 100\mathrm{\%}$$ (6)
• (4)
  Range:

  $$R=x_{max}-x_{min}$$ (7)

  where x_max is the maximum value and x_min is the minimum value.

We performed the statistical analysis and graph construction of the data using GraphPad Prism software (v 7.00, GraphPad). All data are presented as mean ± SEM. Two-tailed unpaired t-tests were used to compare differences between different groups. In this study, p < 0.05 was considered significant (* P < 0.05, ** P < 0.01, and *** P < 0.001).

3. Results

3.1. Comparison of freezing rates

To evaluate the improvement of seven insulated cryofixation methods on the freezing rate and freezing uniformity of the kidney, we monitored the temperature change during the process of kidney freezing, as shown in Fig. 2(a). In the experiment of ex vivo freezing, we found that the average time from kidney extraction to freezing was 5 seconds. Hence, these 5 seconds were included in the total time of ex vivo cryofixation methods. Compared with other freezing methods, the EV-AF cryofixation method (Fig. 2(b)) has the slowest freezing rate, taking up to tens of seconds from 20 °C to −100 °C, highlighting its slowest freezing efficiency. EV-LN₂ (Fig. 2(c)) and EV-Iso (Fig. 2(d)) cryofixation methods require about 15 s, while IV-LN₂ (Fig. 2(e)) and IV-Iso (Fig. 2(f)) cryofixation methods require 15 – 35 s from 20 °C to −100 °C. However, it takes less than 15 seconds for the IVI-LN₂ (Fig. 2(g)) and IVI-Iso (Fig. 2(h)) cryofixation methods, indicating that the in vivo insulated cryofixation methods accelerate the overall freezing rate of the kidney. We also identified notable discrepancies between the temperature curves at the ventral, center, and dorsal regions in both the IV-LN₂ and IV-Iso cryofixation methods (Figs. 2(e) and 2(f)). These disparities suggest non-uniform freezing within the kidney tissue. In contrast, the temperature curves for the IVI-LN₂ and IVI-Iso cryofixation methods are much more closely aligned (Figs. 2(g) and 2(h)), especially in the temperature range from 20 °C to 0°C, which has a greater impact on changes in metabolism. It indicates a more uniform cryofixation of the kidney than other methods.

Fig. 2.

Measurement of the temperature of the kidneys during freezing using different cryofixation methods. (a) Three temperature monitoring sites in the kidney. (b) – (h) Average measured temperature curves of EV-AF (b), EV-LN₂ (c), EV-Iso (d), IV- LN₂ (e), IV-Iso (f), IVI-LN₂ (g), and IVI-Iso (h) cryofixation methods. (N = 9 for EV-AF, N = 8 for EV-LN₂, N = 7 for IV-LN₂ and EV-Iso, N = 6 for IV-Iso, IVI- LN₂ and IVI-Iso).

We further analyzed the average time of the kidney from 20 °C to −100 °C for different cryofixation methods. We calculated its normalized average state change, as shown in Fig. 3. The time required to reach −100 °C from 20 °C is shorter for ex vivo cryofixation methods (EV-LN₂ and EV-Iso) compared to in vivo methods (IV-LN₂ and IV-Iso), as shown in Fig. 3(a). However, the normalized reaction amount of ex vivo cryofixation methods was much higher than in vivo cryofixation methods (Fig. 3(b)). It leads to the 5 s required for kidney extraction, contributing significantly to the normalized reaction amount. The freezing time and the normalized reaction amount of the EV-AF cryofixation methods were significantly higher than that of other methods.

Fig. 3.

Statistics of freezing efficiencies from 20 °C to −100 °C using different cryofixation methods. (a) Cooling time. (b) Normalized reaction amount. (* P < 0.05, ** P < 0.01, *** P < 0.001, and N = 9 for EV-AF, N = 8 for EV-LN₂, N = 7 for IV-LN₂ and EV-Iso, N = 6 for IV-Iso, IVI- LN₂ and IVI-Iso).

Regardless of whether LN₂ or isopentane is used as the cryogen, the freezing rates of in vivo insulated cryofixation methods (IVI-LN₂ and IVI-Iso) are markedly higher than that of in vivo cryofixation methods (IV-LN₂ and IV-Iso). Compared with in IV-LN₂ group and IV-Iso group, the time from 20 °C to −100 °C at the renal dorsal was reduced by 27.4% (IV-LN₂: 18.4 ± 2.3 s, IVI-LN₂: 13.3 ± 1.3 s) and 69.3% (IV-Iso: 32.0 ± 2.5 s, IVI-Iso: 9.8 ± 1.5 s) in the IVI-LN₂ group and IVI-Iso group, respectively, and the corresponding normalized reaction amounts were decreased by 52.5% (IV-LN₂: 0.389 ± 0.042, IVI-LN₂: 0.185 ± 0.029) and 52.9% (IV-Iso: 0.253 ± 0.043, IVI-Iso: 0.119 ± 0.021) in IVI-LN₂ group and IVI-Iso group, respectively. These results confirmed that in vivo insulated cryofixation methods could effectively improve the freezing speed at the renal dorsal. Interestingly, the freezing time from 20 °C to −100 °C at the renal center was also significantly reduced by 27.5% (IV-LN₂: 16.8 ± 1.3 s, IVI-LN₂: 12.2 ± 0.8 s) and 63.5% (IV-Iso: 24.7 ± 2.9 s, IVI-Iso: 9.00 ± 1.0 s) in IVI-LN₂ and IVI-Iso groups, respectively, and the normalized reaction amounts were reduced by 31.4% (IV-LN₂: 0.217 ± 0.026, IVI-LN₂: 0.149 ± 0.013) and 10.2% (IV-Iso: 0.141 ± 0.022, IVI-Iso: 0.127 ± 0.012), respectively, compared with in IV-LN₂ group and IV-Iso group. In IVI-LN₂ group and IVI-Iso group, the time of the renal ventral from 20 °C to −100 °C was decreased by 34.7% (IV-LN₂: 14.9 ± 1.10 s, IVI-LN₂: 9.8 ± 0.7 s) and 54.2% (IV-Iso: 16.9 ± 3.4 s, IVI-Iso: 7.8 ± 0.7 s), respectively, and the normalized reaction amount was reduced by 32.3% (IV-LN₂: 0.102 ± 0.019, IVI-LN₂: 0.069 ± 0.009) and 1.6% (IV-Iso: 0.058 ± 0.010, IVI-Iso: 0.057 ± 0.006), respectively, compared with in IV-LN₂ group and IV-Iso group. The accelerated freezing rate of the renal center and ventral may be because the insulated component reduces the outflow of cryogen and improves the use efficiency of cryogen. Meanwhile, the continuous pouring of cryogen will ensure that the heated cryogen in the insulated component is constantly replaced. These data indicate that the insulated cryofixation methods can achieve the design purpose of improving the overall freezing rate of mouse kidneys.

It also can be observed that the freezing rate of the renal dorsal is much lower than that of the renal center and ventral in both the IV-LN₂ group and IV-Iso group. For example, the normalized reaction amounts at the renal dorsal are 79.1% (dorsal: 0.389 ± 0.042, center: 0.217 ± 0.026) and 79.6% (dorsal: 0.253 ± 0.043, center: 0.141 ± 0.022), respectively, higher than those at the center in the IV-LN₂ group and IV-Iso group. In contrast, the normalized reaction amount at the renal dorsal is 24.0% (dorsal: 0.185 ± 0.029, center: 0.149 ± 0.013) higher and 5.8% (dorsal: 0.119 ± 0.021, center: 0.127 ± 0.012) lower than those at the center in the IVI-LN₂ group and IVI-Iso group. These data indicate that the in vivo insulated cryofixation methods can improve the uniformity of freezing the whole kidney.

3.2. Evaluation of renal morphological integrity

To demonstrate the ability of in vivo insulated cryofixation methods to maintain the morphological integrity of kidneys, we photographed the frozen kidneys acquired by different cryofixation methods, as shown in Fig. 4(a). We can observe that most of the kidneys showed cracking in the EV-LN₂ and EV-Iso groups, and a few kidneys showed local bulges. It likely occurs because the kidneys’ outer layers freeze more rapidly into a solid state while the central part freezes more slowly. The slower the freezing rate, the more ice crystals are formed and the greater the resulting volumetric expansion. Consequently, the increased internal expansion pressure ultimately leads to kidney cracking or bulging. In the IV-LN₂ group, a few kidneys were missing a little tissue since these kidneys were frozen together with the underlying tissue and could not be easily harvested. A few kidneys also appeared to have similar cracking in the IV-Iso group. Conversely, the kidneys in the IVI-LN₂ and IVI-Iso groups exhibit no bulging or cracking. As shown in Fig. 4(b), we proposed the structural integrity rate to quantify the effect of different cryofixation methods. It is calculated as the ratio of structurally intact kidneys to the total number in the group. The sample without rupture or significant structural deformation in macroscopic and cryo-MOST images, indicated by blue arrows in Fig. 4(a), was considered structurally intact. The structural integrity rates were higher in the IVI-LN₂ and IVI-Iso groups than in the other groups, as shown in Fig. 4(b). Thus, in vivo insulated cryofixation methods effectively preserve the whole kidney's morphology integrity during freezing.

Fig. 4.

Evaluation of renal morphological integrity by using different cryofixation methods. (a) The appearance of the kidney. The blue arrows indicate the cracked or defective in the frozen kidneys. (b) Renal intact ratio (n = 10).

3.3. Comparison of renal metabolic state

To evaluate the ability of different cryofixation methods to maintain the metabolic states of the kidneys, we performed autofluorescence cryo-MOST metabolic imaging. Figure 5(a) shows NADH, FAD, and RR images of typical cross-sections of kidneys using seven cryofixation methods. The NADH and RR values of ex vivo cryofixation methods were significantly higher than those of in vivo cryofixation methods. In addition, we also observed obvious damages using ex vivo cryofixation methods indicated by red arrows in Fig. 5(a), consistent with the results in Fig. 4. The RR of IV-LN₂ and IV-Iso in the dorsal of the kidney was significantly higher than that of other sites. At the same time, the RR of IVI-LN₂ and IVI-Iso was uniform in the whole kidney. It is also consistent with the results previously found that the dorsal kidney freezing rate of IV-LN₂ and IV-Iso is slower than that of the ventral, while the whole kidney freezing rate of IVI-LN₂ and IVI-Iso is less different (Fig. 2).

Fig. 5.

Comparison of renal metabolic states by using different cryofixation methods. (a) NADH, FAD, and RR images of typical kidney cross-sections by different cryofixation methods. The red arrows indicate kidney cracking, (b) RR statistical analysis. (MeanRR: mean value of RR; R: range; SD: standard deviation; RSD: relative standard deviation. * P < 0.05, ** P < 0.01, *** P < 0.001, and N = 5).

To quantitatively study the uniformity of renal metabolic states obtained by different freezing fixation methods, we selected 400 × 400 × 600 µm image blocks at five positions of the kidney, cephalic, ventral, lateral, dorsal, and caudal, for statistical analysis. As shown in Fig. 5(b), the mean_RR values of ex vivo cryofixation methods are significantly higher than the other groups. At the same time, there is no significant difference in RR values between in vivo cryofixation methods and in vivo insulated cryofixation methods. These results are consistent with Fig. 3(b), indicating that this is likely due to the metabolic process of kidneys continuing under the premise of interrupted oxygen supply. The lack of oxygen during this period is enough to shift the energy metabolism of the kidney [2,20], manifesting as a high NADH signal in the group. The more changes of RR observed in the cortex compared to the inner medulla may be explained by the cortex being more energy-demanding and particularly sensitive to hypoxia. Hence, the ex vivo cryofixation methods are unsuitable for acquiring precise metabolic information about the tissue. Then, we used range (R), standard deviation (SD), and relative standard deviation (RSD) for uniformity analysis. As for the IVI-LN₂ and IVI-Iso, the R decreased by 71.6% (IVI-LN₂: (2.34 ± 0.28) × 10⁻², IV-LN₂: (8.23 ± 1.73) × 10⁻²) and 79.4% (IVI-Iso: (2.05 ± 0.35) × 10⁻², IV-Iso: (9.96 ± 1.39) × 10⁻²), respectively, the SD decreased by 70.8% (IVI-LN₂: (0.88 ± 0.13) × 10⁻², IV-LN₂: (3.01 ± 0.61) × 10⁻²) and 78.0% (IVI-Iso: (0.81 ± 0.13) × 10⁻², IV-Iso: (3.69 ± 0.57) × 10⁻²), respectively, and the RSD decreased by 69.8% (IVI-LN₂: (2.03 ± 0.32) × 10⁻², IV-LN₂: (6.72 ± 1.29) × 10⁻²) and 77.4% (IVI-Iso: (1.93 ± 0.31) × 10⁻², IV-Iso: (8.54 ± 1.18) × 10⁻²), respectively, compared with the IV-LN₂ and IV- Iso. It is likely due to the larger time interval between the freezing of the renal ventral and dorsal areas in the IV-LN₂ and IV-Iso, resulting in more pronounced differences in the metabolic state of the renal cortex than in the IVI-LN₂ and IVI-Iso. In addition, the R and SD of IVI-LN₂ and IVI-Iso are also lower than those of EV-LN₂ and EV-Iso. Therefore, the metabolic homogeneity of frozen kidneys obtained by in vivo insulated cryofixation methods is optimal, which verifies the results of freezing rate measurements shown in Fig. 3(b). These results indicate that the in vivo insulated cryofixation methods substantially enhance the freezing uniformity of the kidneys for cryo metabolic imaging, and the result can better reflect the state of the mouse before death. There was little statistical difference between the IVI-LN₂ and IVI-Iso groups, given the complexities associated with using isopentane, such as the requirement for pre-cooling, post-use recovery, and potential safety concerns. Therefore, LN₂ was used as a cryogen in subsequent experiments due to its ease of use.

3.4. Analysis of metabolic states of whole diabetic mice kidneys

The db/db mouse, characterized by a leptin receptor mutation, serves as a well-established model for type 2 diabetes research [21]. Some biochemical analyses of the whole kidney show that the balance of renal redox metabolism is disrupted in db/db mice, leading to kidney injury [22]. However, the energy consumption and damage of different subregions of kidneys are not the same [7,9]. So, it is necessary to analyze the 3D metabolic information of each renal subregion. We assessed the renal metabolic states in 15-week-old db/db mice and their db/m control counterparts. Typical renal metabolic image sequences of two group mice kidneys at 1.2 mm-interval are shown in Fig. 6(a). It is observable that the whole kidneys exhibit no significant cracking or deformation. There was no significant difference in RR values between the ventral and dorsal sides of the kidney, and the metabolic images were uniform overall. The results indicated that the IVI-LN₂ cryofixation method combined with cryo-MOST imaging could achieve complete and uniform 3D whole-organ metabolic information acquisition. The highlighted areas in the pelvis of FAD and NADH fluorescence images indicated hydronephrosis in db/db mice. Our previous results have shown that this hydronephrosis has no significant effect on the autofluorescence intensity of renal parenchymal [8]. Enlarged views of the typical sections indicated by red rectangles in Fig. 6(a) show inconsistent metabolic states across various renal subregions, as shown in Fig. 6(b). The outer stripe of the outer medulla (OSOM) region in both db/db and db/m mice kidneys exhibits lower RR values and higher NADH and FAD intensities compared to other subregions, indicating a more vigorous energy metabolism. We selected 400 × 400 × 600 µm image blocks on different subregions of the kidney in two groups of mice to perform further statistical analysis, as depicted in Fig. 6(c). The RR values in the renal cortex, OSOM, and the inner stripe of the outer medulla (ISOM) regions of the db/db mice kidneys are increased by 10.7% (db/db: 0.470 ± 0.005, db/m: 0.424 ± 0.003), 11.4% (db/db: 0.417 ± 0.004, db/m: 0.375 ± 0.004), and 6.68% (db/db: 0.467 ± 0.006, db/m: 0.437 ± 0.004), respectively, than those in the db/m mice, with no significant difference observed in the inner medulla (IM) region between the two groups. It suggests that metabolic disorders may have occurred in the cortex, ISOM, and OSOM of the db/db mice while the IM was normal.

Fig. 6.

Comparison of renal metabolic states between 15-week db/m mice and db/db mice. (a) Typical renal metabolic image sequence for db/db and db/m mice. The spacing of adjacent images in the image sequence is 1.2 mm. (b) Enlarged views of the red box in (a) and different subregions are marked by black lines. (a: IM, b: ISOM, c: OSOM, d: Cortex). (c) Statistical analysis of RR. (** P < 0.01, *** P < 0.001, and N = 5).

4. Discussion and conclusions

In this study, we proposed combining the in vivo insulated cryofixation method and cryo-MOST to obtain 3D structural and metabolic information on whole mice kidneys without labeling. Compared with other cryofixation methods, the in vivo insulated cryofixation method can improve the freezing rate and uniformity while maintaining the integrity of the kidney. Cryo-MOST imaging technology can use autofluorescence to obtain and display 3D metabolic information of the kidney with micron resolution. Using this pipeline, we found that the renal cortex, OSOM, and ISOM of 15-week-old db/db mice showed abnormal metabolic states, while the IM region was typical. These results demonstrate the ability of this pipeline to obtain and visualize 3D metabolic maps of complete organs, which provides potential help for studying the spatiotemporal progression of metabolic diseases.

The kidney's structure and function are complex, and its metabolic information is equally intricate [23–25]. So, the study of renal 3D metabolic information is essential. However, conventional metabolic assays, which rely on kit-based methods, often sacrifice this 3D context [10]. Cryo-optical microscopic imaging technology promises to capture intact organs’ complete 3D metabolic state. Snap cryofixation of the in vivo metabolic state is a prerequisite for realizing this potential. For the frozen sample's metabolic state to reflect the in vivo state, the freezing process must be controlled, i.e., the time interval between interruption of blood circulation and metabolic cessation should be as short as possible. Previous in vivo monitoring studies have demonstrated that detectable NADH changes were observed in mice upon induction of circulatory hypoxia for 6 s or hypoxic hypoxia for 3.9 s [12]. In addition, to maintain the reliability of statistical results, the samples should also be kept intact and uniformly frozen as much as possible [10]. However, the traditional snap-freezing methods make it challenging to achieve a uniform and complete cryofixation [26,27]. Here, we proposed in vivo insulated cryofixation methods to improve the quality of kidney cryofixation. In the IVI-LN₂ group and IVI-LN₂ group, Our temperature monitoring indicates that the time of the kidneys frozen from 20°C to 0°C (a temperature range significantly impacting metabolic changes) is less than 3s. It suggests that in vivo insulated cryofixation methods can better maintain the metabolic states of kidneys. Our previous studies also showed that the overall shape of the kidneys in ev vivo cryofixation groups was changed, and its volume was reduced due to blood outflow compared with the in vivo shape obtained by micro-CT, while in vivo cryofixation methods could better maintain its in vivo morphology [8]. In this paper, The cryo-MOST results indicate that the in vivo insulated cryofixation methods substantially enhance the freezing uniformity of the kidneys and effectively preserve the morphology integrity of the whole kidney during the freezing process.

All methods are simple in terms of technical complexity, and the entire surgical process can be completed within 5 minutes. Compared with other methods, ex vivo cryofixation methods are the simplest but have the worst effect on preserving structural and metabolic information. There is only one more step to place the insulated components in the body compared with the in vivo cryofixation methods, but it will be convenient for kidney extraction. In vivo insulated cryofixation methods combined with Cryo-MOST imaging exceed limitations on optical imaging depth and can obtain label-free 3D metabolic and structural information on kidneys. This is conducive to a comprehensive assessment of physiological, structural changes, and metabolic abnormalities caused by kidney disease.

Different subregions of kidneys have different structures and functions, so the metabolic states of different subregions may be different. However, previous studies on kidney metabolism were primarily based on the whole kidney [4,6,15,23]. Cryo-MOST can perform micron-resolved 3D metabolic imaging of kidneys and identify the boundaries of the renal cortex, OSOM, ISOM, and IM subregions. Therefore, we analyzed the metabolic states of different renal subregions of db/db mice, respectively. We found that the change of RR in different renal subregions of db/db mice was indeed different. Compared with db/m mice, RR of the renal cortex, ISOM, and OSOM were increased in db/db mice, while IM was unchanged. This result indicates that it is necessary to analyze the different subregions of the kidney in 3D.

We propose two potential reasons for the increased renal RR in specific db/db mice subregions. In our previous study, structural analysis of db/db mouse kidneys revealed significant volume expansion in the cortex and ISOM, which may be closely associated with proximal convoluted and straight tubule dilation. In contrast, no significant volume changes were observed in the IM [8]. The dilation of renal tubules may compress capillaries, impairing local blood circulation. Furthermore, we identified morphological deformations in particular interlobar veins of db/db mice due to compression, which exacerbates disruptions in normal hemodynamics. Notably, the cortex, OSOM, and ISOM are metabolically active regions with high mitochondrial content and oxygen consumption, rendering them particularly sensitive to ischemic hypoxia [28]. The resultant hypoxia from circulatory disturbances likely contributes directly to the increased RR in these regions. In contrast, the IM exhibits lower metabolic demand and reduced responsiveness to hypoxia, which may explain the absence of significant RR changes in this area [9]. Secondly, some studies have demonstrated that hyperglycemia induces excessive NADH production, disrupting the NADH/NAD⁺ redox balance and causing pseudohypoxia, thereby elevating RR [29,30]. Pseudohypoxia may trigger reductive and subsequent oxidative stress, ultimately promoting cell death and tissue dysfunction [31]. The elevated RR in db/db mouse kidneys observed in this study indicates a profound energy metabolic imbalance. This imbalance could impair ATP synthesis, compromise active transport functions in renal tubules, and exacerbate tubular injury, accelerating the progression of diabetic kidney disease [32]. Therefore, analyzing db/db mouse subregions can further deepen the understanding of disease changes in different renal subregions.

NADH concentrations are significantly higher than NADPH, and collagen, elastin, and lipofuscin are not abundant in renal tissue [33–37]. So, in this study, we assume that the interference from NADPH and other endogenous fluorophores in the tissue is minimal in detecting the RR. In addition, we have only preliminarily explored the subregional structure and metabolic changes of the kidneys in db/db mice. We can also combine metabolism state and vascular morphological changes to further study the etiology of diabetic nephropathy and verify the improvement of different drugs on its metabolism, which provides a new way for efficacy evaluation. In addition, due to the widefield imaging's lack of background inhibition, some adjacent tissues with strong fluorescence (such as fat, etc.) may interfere with the acquisition of metabolism images of the kidney. Next, we will use optical sectioning methods like line scanning to solve the problem.

In summary, in vivo insulated cryofixation methods combined with cryo-MOST imaging can provide label-free 3D delicate structure information and metabolic information in situ of the kidney, which provides convenience for the research on the pathogenesis and therapeutic effect evaluation of kidney diseases. This method is also widely applicable to other abdominal organs, providing a simple and effective tool for studying metabolic diseases in whole organs.

Funding

National Natural Science Foundation of China10.13039/501100001809 (62325502, 81827901).

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available but may be obtained from the authors upon reasonable request.