Double resin casting micro-computed tomography, or DUCT, enables visualization, digitalization, and segmentation of two tubular systems simultaneously to facilitate 3D analysis of organ architecture. DUCT combines ex vivo injection of two radiopaque resins followed by micro-computed tomography scanning and segmentation of the tomographic data.
The liver is the biggest internal organ in humans and mice, and high auto-fluorescence presents a significant challenge for assessing the three-dimensional (3D) architecture of the organ at the whole-organ level. Liver architecture is characterized by multiple branching lumenized structures, which can be filled with resin, including vascular and biliary trees, establishing a highly stereotyped pattern in the otherwise hepatocyte-rich parenchyma. This protocol describes the pipeline for performing double resin casting micro-computed tomography, or "DUCT". DUCT entails injecting the portal vein and common bile duct with two different radiopaque synthetic resins, followed by tissue fixation. Quality control by clearing one lobe, or the entire liver, with an optical clearing agent, allows for pre-screening of suitably injected samples. In the second part of the DUCT pipeline, a lobe or the whole liver can be used for micro-computed tomography (microCT) scanning, (semi-)automated segmentation, and 3D rendering of the portal venous and biliary networks. MicroCT results in 3D coordinate data for the two resins allowing for qualitative as well as quantitative analysis of the two systems and their spatial relationship. DUCT can be applied to postnatal and adult mouse liver and can be further extended to other tubular networks, for example, vascular networks and airways in the lungs.
Organ resin casting is a technique that dates back to the 17th century1. One of the first examples of modern resin casting was performed on the human liver from an autopsy. Intrahepatic bile ducts were filled with a contrast agent mixed with gelatin, followed by imaging with an x-ray CT scan2. The aim of the DUCT technique is to visualize, digitalize and analyze two tubular resin-casted networks, in tandem, in 3D.
DUCT is based on the extensive existing knowledge of single-system liver resin casting3,4,5,6,7,8 and extends to simultaneous 3D visualization and analysis of two systems9. DUCT advanced single resin casting to double resin casting by mixing two radiopaque resins of different contrast and injecting these resins into two different networks, specifically the common bile duct and portal vein. DUCT can be applied to young postnatal mice with reproducible results as early as postnatal day 15 (P15). Compared to microscopy-based imaging techniques, the main advantage is that DUCT is faster, antibody-free, and liver tissue autofluorescence does not interfere with imaging. Further, DUCT provides quantitative data describing lumenization status, internal diameter, network connectivity, and perfusion. Differentiating between the presence of lumen-forming cells and their de facto morphogenesis into tubes is essential for analyzing organs in which ductular cells are present but do not form tubes, as can be the case in Alagille syndrome10. The main disadvantage of DUCT is the limited penetration of the resin, which is viscous and does not enter tubes with a small caliber (<5 µm). DUCT can be applied for any tubular structure after determining the injection entry point, such as the arterial and venous circulatory systems, airways, the extrahepatic bile duct, or lymphatic vessels. It could thus facilitate whole organ architecture analysis of other tissues such as lungs and pancreas.
MicroCT segmented images can be processed using commercially available imaging software, such as ImageJ, or custom-written pipelines (e.g., MATLAB). The resin-injected liver can be analyzed qualitatively for network expansion and connectivity or quantitatively for volume, length, branching, tortuosity of a single system, and the interaction between two systems such as the distance between two systems, or branchpoint dependence (does system 1 branch in proximity to system 2 branching?). The DUCT pipeline encompassing resin injection, microCT scanning, and CT data segmentation, combined with detailed quantitative analysis of architectural mechanisms of two tubular systems, could provide a standard for whole liver analysis in animal models.
The protocol described in this study was approved and follows the animal welfare rules and regulations of the Stockholms Norra Djurförsöksetiska nämnd (Stockholm animal research ethics board). The animals used in this study were wild type or homozygous Jag1H268Q mutant mice on a mixed C3H/N and C57bl6J background. Both males and females were included in the study. The animals were used at postnatal day 15 or as adults between 3 – 8 months.
1. Double resin injections
2. Micro-computed tomography
3. Analysis and data segmentation
NOTE: The following steps are image-processing software-dependent; specific settings and actions may differ depending on the software used.
What to do
Successful double resin injection is achieved when both the intrahepatic bile ducts and portal vein vasculature are well filled. As a quality control step, clearing one lobe (for example, the left lateral lobe) allows for verification of a successful injection, followed by imaging of lobes of interest. The optically cleared lobe can be scanned later using microCT; hence it is possible to optically clear the whole liver. In well-injected mouse liver, the portal vein vasculature should be filled with resin until the liver periphery and resin should be visible in side-branches (Figure 4), and this architecture is faithfully recapitulated in microCT scanned and segmented data. Further, well-injected intrahepatic bile ducts should be visible next to the main portal vein branches extending almost to the periphery, and resin should be visible in the major side branches. If the control lobe passes the quality control step, the lobes of interest (included the optically cleared one) can be scanned with microCT. The result of the segmented data from a well-injected liver is shown for a P15 mouse (Figure 4A,B) and an adult mouse (Figure 4C,D).
What not to do
Intact liver tissue is a prerequisite for successful injection. Take extra care when cutting the abdominal cavity and diaphragm not to accidentally nick the liver tissue. If there is physical damage to the liver during this procedure, the resin is very likely to leak out during portal vein injection (Figure 5A). It is not possible to achieve a good injection of the vascular system if the liver is physically damaged.
One of the common mistakes is underfilling the liver with resin that can lead to challenges for visualization or analysis. One of the causes for system underfilling is resin hardening prematurely in the needle or the tip of the tubing before the injection is completed (Figure 5B, blue arrowheads, brackets depict large bubbles). A good practice is to use one injection set per animal and work fast after the curing agent is added to the resin. If the resin hardens during the injection (which can be observed by a half-filled system, here exemplified with a half-filled portal vein vasculature) remove the tubing, cut the tip of the tubing (always diagonally to create a beveled tip), and push the plunger. If resin begins to drip again, carefully re-insert the tubing and secure it with the suture. If the resin has hardened in the needle, replace the tubing completely, fill it with resin (avoiding bubbles), carefully re-insert the tubing, and secure it with the suture. It can be challenging to replace the tubing, especially in young postnatal mice <P30, as the tissue is more fragile. Another cause of poor resin filling of portal vein vasculature can be insufficient transcardial perfusion (Figure 5C, blue arrowheads denote blood visible in terminal branches). This can be observed when the tips of the vessels are filled with blood instead of resin. To avoid this, ensure that the portal vein (outside the liver) does not contain any blood before the injection. The third cause of underfilled liver is when the tubing is inserted too deep into the liver and enters a branch towards one of the lobes. To prevent this, insert the tubing at a minimum of 0.5 cm from the entry to the liver.
Conversely, one, or both, of the systems become overfilled with resin (Figure 5D). It is necessary to visually monitor the liver throughout the injection. Biliary system casting with resin is more challenging than portal vein resin casting since resin-filled ducts are only faintly visible on the liver surface, and it is difficult to assess when the system is nearly full and when to stop. When small yellow resin dots appear on the liver surface (Figure 2Ci, blue arrowhead), this is a sign that the biliary system is completely filled, and the resin is starting to leak out of the ducts. Minor resin leakage can be manually corrected during microCT data segmentation (Figure 5D, right panels).
If the injection pressure is too high, this can cause vessels or ducts to rupture (Figure 5E), irreversibly damaging vessel or duct architecture. The liver will not be suitable for microCT scanning or analysis. To avoid resin overfilling, optimize the right volume and pressure used for injection in each mouse model. When working with mice that have been challenged with a toxic diet, genetic modification, or liver injury that affects the biliary or venous systems, or liver stiffness, the injection pressure and volume may need to be adjusted as volume and pressure tolerated may be different from the wild type mice. This protocol describes the manual injection of the two systems, but it is possible to connect the syringe to a pump to standardize the injection pressure. Bubbles are another very common injection artifact that leads to sparse filling of the tubular networks (Figure 5F–H, blue arrowheads). To avoid bubble formation, make sure that the syringe and tubing do not contain any bubbles, are completely filled with resin, and the resin is dripping from the tip of the tubing before injection. Small bubbles that appear as negative areas on the microCT data can be manually corrected during post-processing steps, although this is laborious.
Fresh is the best
Using fresh yellow resin is a crucial factor, significantly affecting the contrast of the two resins and the microCT data segmentation. When the freshly opened resin is used (Figure 6A) there is a clear difference in contrast between the yellow resin-injected bile ducts (bright white) and the green resin-injected portal veins (bright grey). Liver that is injected with fresh resin is easily processed using automated global thresholding. With prolonged storage, the resin precipitates, and the contrast diminishes. After 3 months of storage, the contrast can still be sufficient to distinguish the portal vein from the bile duct (Figure 6B), but precipitation affects the mixing of the two resins, which is visible as a heterogeneous opacity in the filled portal vein (Fig 6B, blue arrowheads). Heterogeneous contrast negatively affects the automated thresholding and necessitates manual corrections, which increases the processing time. If the resin is older than six months, the contrast has degraded to a point at which it is not possible to distinguish the yellow-injected bile duct from the green-injected portal vein based on their contrast alone (Figure 6C). In this case, the bile duct and portal vein must be segmented manually based on their diameter and position in the hilar region and followed manually throughout the entire microCT data. This procedure is extremely time-consuming and best avoided.
Figure 1: Injection set for resin casting. (A) Injection set #1 comprises a 30 G needle and PE10 tubing that is ~30 cm long. Injection set #2 is composed of a 23 G needle and PE50 tubing ~30 cm long. (B) The tip of the tubing is stretched and cut at an angle to create a beveled tip. The ruler in A and B is a centimeter ruler, with major increments of 1 cm, intermediate increments of 5 mm, and minor increments of 1 mm. Please click here to view a larger version of this figure.
Figure 2: Double resin casting flow chart. (A) Schematic showing the murine venous circulatory system and the heart with highlighted right atrium, which should be cut away before perfusion, and the left ventricle into which the needle should be inserted for perfusion to wash away the blood from the circulatory system. The inferior vena cava should be severed under the kidneys to relieve vascular pressure. (B) Bile duct resin injection flow chart. (i) Zoom image of (A) depicting where to sever the IVC. (ii) Image depicting the common bile duct (yellow dotted line) from the liver hilar region to the sphincter of Oddi (black arrowheads), with suture thread under the cleared common bile duct. (iii) Suitable position for loose overhand knot around common bile duct. (iv) The yellow dotted line and black arrowheads label the sphincter of Oddi, demonstrating the oblique angle for incision and how the opening should appear after the oblique angle incision. (v) Schematic demonstrating the orientation of the PE10 tubing bevel opening (upwards) upon insertion. (vi) Appearance of yellow resin being injected; resin should easily pass the loosely tied knot. IVC, inferior vena cava; CBD, common bile duct. (C) Portal vein resin injection flow chart. (i) The green dotted line marks the portal vein from the hilar region. The blue arrowheads label the overfilled biliary system. (ii) Suitable location for loose overhand knot around portal vein. (iii) Schematic demonstrating the bevel opening (upward) upon insertion. (iv) Appearance of liver upon injection of green resin and yellow resin; note the resin-filled blood vessel in the liver periphery. PV, portal vein. Figure 2A was created with Biorender.com. Please click here to view a larger version of this figure.
Figure 3: Micro CT data processing in volume graphics software. (A) Surface determination, (i) overestimated isovalue (current preview of selection is shown in yellow color), (ii) underestimated isovalue, (iii) optimal selection of isovalue for proper surface determination of portal vein and bile ducts. (B) Splitting region of interest (ROI) created by surface determination, (i) set the value in dialog window high enough that only one segment (the largest one) will remain, (ii) in yellow frames the smaller (excluded) particles are shown. (C) Surface smoothing of the data, (i) smoothing function is on the left panel, (ii) set the smoothing strength to 1 (max. 2) and create new smoothed ROI, (iii) smoothed data. (D) Separation of individual tubular systems, (i) in surface determination function set the isovalue so that only the biliary system is included in selection (current preview of selection is shown in yellow color), (ii) mark the ROI of both systems and ROI of only the biliary system and subtract biliary system ROI from ROI of both systems, (iii) Portal vein shown in grey, the biliary system shown in green. Please click here to view a larger version of this figure.
Figure 4: Well-injected bile duct (BD) and portal vein (PV) systems. (A) Optically cleared right medial lobe (RML) of postnatal day 15 (P15) liver injected with two resins into the two systems. Scale bar 1 mm. (B) 3D rendering of P15 RML shown in (A) depicting portal vein vasculature in white and biliary system in green. Scale bar 1 mm. (C) Optically cleared RML of adult liver injected with two resins into the two systems. Scale bar 1 mm. (D) 3D rendering of adult RML shown in (C) depicting portal vein vasculature in white and biliary system in green. H = hilar, P = peripheral. Scale bar 1 mm. Panels A, B, D are adapted with permission from Hankeova et al.9. Please click here to view a larger version of this figure.
Figure 5: Common challenges of double resin liver injections. (A) The image depicts a liver that was accidentally nicked during the initial opening of the abdominal cavity, and the resin is leaking through the cut (blue arrowhead). (B) Poorly injected portal vein system due to resin hardening. Blue arrowheads label empty terminal branches, and red brackets label large bubbles. Scale bar 1 mm. (C) Poorly injected portal vein system due to poor transcardial perfusion. Blue arrowheads label blood visible in the terminal branches. Scale bar 1 mm. (D) Overfilled biliary system manifested by isolated balls of resin. The left panels show the optically cleared liver, and the right panels show the 3D microCT rendered image. The blue dotted outlines depict zoom-in regions. The black arrowheads label a part of the liver that was damaged during the optical clearing after microCT scanning. Scale bar 1 mm. (E) High pressure during resin injection can cause rupture of the portal vein (the animal in this panel carries a Jag1H268Q mutation), marked by blue arrowheads. Scale bar 1 mm. (F) Bubbles in the resin during portal vein injection (blue arrowheads) and (G) biliary system injection (blue arrowhead), scale bar 1 mm. (H) MicroCT scan of bubbles (blue arrowhead), poorly filled terminal branches (red brackets) and resin leakage (yellow arrowhead), scale bar 1 mm. Please click here to view a larger version of this figure.
Figure 6: Differential resin contrast. (A) Freshly opened yellow resin generates sufficient contrast to distinguish resin injected portal vein (grey) and bile ducts (white). (B) Three months storage of yellow resin lead to precipitation of the resin resulting in heterogeneous opacity (grey-white portal vein, blue arrowhead). (C) Prolonged storage (>6 months) of yellow resin diminishes the contrast between the portal vein (grey) and bile ducts (grey). Scale bar 100 µm. Please click here to view a larger version of this figure.
Several critical steps determine DUCT success, from sample preparation to the parameters of the CT device. To achieve the best results, well contrasted, well-injected, and bubble-free resin should be used to allow straightforward digital processing with automated thresholding to obtain 3D data, images, and movies. With training and following this protocol, 90% of injections are successful and result in reproducible data. It is important to use fresh yellow resin to achieve the best contrast between the two injected systems. The yellow resin has a very strong radiopacity, while the blue resin has undetectable radiopacity. Top results are achieved within the first three months after opening a new yellow resin bottle. With time, resin precipitates, and after longer storage (>6 months), the yellow and the green resins will no longer be distinguishable in CT scans. Images with poor contrast necessitate extensive and time-consuming manual tracing and segmentation of the two systems. Next, well-stretched tubing is indispensable to fit into the common bile duct of adult mice and common bile duct and portal vein of postnatal mice. The entry point for the injection must be created with care. If the common bile duct is cut open transversally, it is likely to detach from the surrounding tissue, preventing successful entry of the tubing. This step is especially delicate for postnatal mice in which the common bile duct retracts and "curls up" if it has detached from its surrounding tissue, making insertion of the tubing extremely challenging. The common bile duct entry and injection may require some practice. While preparing the tubing with resin and throughout injection, avoid bubble formation as bubbles will create negative space in the CT images and require time-consuming manual correction. It is important to gently massage the liver by rolling over its surface with a wetted cotton swab during and after the injection procedure as this facilitates even resin spreading. After the completion of injection and removal of the tubing, the silk suture knot must be tightened quickly and carefully, so the resin does not flow out of the liver before it polymerizes completely. For successful microCT imaging, the sample must be properly fixed in place with agarose and thermally adapted to eliminate movement artifacts in the CT data. The acquisition settings are also of key importance, which should be optimized to reach an adequate spatial resolution to resolve fine structures.
Technical modifications to the injection procedure can be made to achieve injection in younger mice. Currently, resin casting of younger mouse livers is limited by the availability of sufficiently thin tubing, with PE10 being the smallest commercially available tubing. Tanimizu et al. successfully injected carbon ink into embryonic day 17 (E17) common bile duct using glass capillaries11. Future testing of whether resin can be delivered via glass capillary would therefore be of interest. DUCT was further adapted to inject other tubular systems such as the airways and pulmonary artery vasculature of the lungs9. The double resin injection could also be modified to be used with other commercially available resins, or this protocol could be used for injections with carbon ink.
One of the main limiting factors of the DUCT pipeline is the resin viscosity. DUCT can only be used for resin casting of tubular structures above a diameter of 5 µm. In this data set, the resin could penetrate tubes with the smallest diameter of 5 µm9. This size limitation precludes the analysis of fine ductules and small capillaries. To further advance the DUCT pipeline to smaller caliber vessels, other commercially available resins should be tested, or the development of new low-viscosity radiopaque agents may improve the lumen penetration.
In Hankeova et al.9, DUCT was compared to two other commonly used techniques, double carbon ink injections followed by tissue clearing and standard photography, and iDISCO+ with staining of the blood vessels with alpha-smooth muscle cell actin and bile ducts with cytokeratin 7, followed by 3D imaging9. DUCT outperformed the other two methods in terms of dual analysis (which was challenging for iDISCO+ due to high liver autofluorescence), 3D imaging, and quantification (not possible with carbon ink injection), and lumenization (DUCT provides data for the internal lumen architecture and system perfusion). As mentioned above, the main limitation of DUCT is the minimum lumen size that can be injected and analyzed (5 µm limit), a parameter in which both carbon ink injection and iDISCO+ performed better. DUCT is superior to single system resin casting3,5,6 because it allows analysis of each injected system separately and also facilitates dual 3D investigation to study the architectural relationship between the two systems.
DUCT can be applied to study any two tubular networks in 3D. As proof of principle, DUCT was used to visualize the liver biliary and portal vein systems and the pulmonary artery vasculature and airways in lung9. The intrahepatic bile ducts develop adjacent to the portal vein, and the portal vein provides a structural template and signaling center that regulates the growth and differentiation of the biliary tree12. In Hankeova et al.9, DUCT explored biliary regeneration in a mouse model for the human pediatric disease Alagille syndrome. DUCT revealed previously unreported architectural mechanisms that the biliary system used to achieve a wild-type-like volume9. The Alagille syndrome mice utilized two different strategies: (1) in the hilar and central regions of the liver, the biliary system increased its branching, and (2) in the liver periphery, the de novo-generated bile ducts were highly tortuous. These two factors combined to yield a near-normal biliary system volume, despite the abnormal architecture. Furthermore, DUCT detected abnormal bile duct branching that occurred independent of portal vein branching and bile ducts forming connecting bridges between two portal veins9. These phenotypes would be impossible to detect in single resin casting and could be misinterpreted in 2D histological sections as bile duct proliferation. DUCT thus provides data describing the 3D architecture of two tubular networks at the whole organ or lobe level with the possibility of qualitative and in-depth quantitative analysis. DUCT could be a new standard for postnatal liver development and liver regeneration analyses in different animal models.
The authors have nothing to disclose.
We thank Kari Huppert and Stacey Huppert for their expertise and help regarding bile duct cannulation and their laboratory hospitality. We also thank Nadja Schultz and Charlotte L. Mattsson for their help with common bile duct cannulation.
We thank the following Granting Agencies for their support:
For work in ERA Lab: Karolinska Institutet (2-560/2015-280), Stockholms Läns Landsting (CIMED (2-538/2014-29)), Ragnar Söderbergs stiftelse (Swedish Foundations' Starting Grant), European Association for the Study of the Liver (Daniel Alagille Award), Swedish Heart-Lung Foundation (20170723), and Vetenskapsrådet (2019-01350).
For work in JK Lab: We acknowledge CzechNanoLab Research Infrastructure supported by MEYS CR (LM2018110). J.K. thanks to the support of grant FSI-S-20-6353.
1.5 mL SafeSeal micro tubes | Sarstedt | 72.706 | |
23 G butterfly needle with tubing | BD bioscience | 367283 | |
25 G needle | BD bioscience | 305122 | |
30 G needle | BD bioscience | 305106 | |
Agarose | Top-Bio | P045 | |
Benzyl alcohol | Sigma Aldrich | 108006 | |
Benzyl benzoate | Sigma Aldrich | B6630 | |
Corning 50 mL tubes | Sigma Aldrich | CLS430829-500EA | polypropylene |
Cotton swabs | Medicarier | 60406 | |
Dissection Microscope | Leica Camera AG | Leica M60 | |
Dulbecco's phosphate-buffered saline | ThermoFisher Scientific | 14190144 | |
Ethanol 70% | VWR | 83801.41 | |
Falcon tube 15 mL | Verkon | 331.850.084.006 | |
Forceps curved | Fine Science Tools | 11051-10 | Fine Graefe 10 cm curved |
Forceps straight | Fine Science Tools | 11050-10 | Fine Graefe 10 cm straight |
Formaldehyde solution | Sigma Aldrich | F8775 | |
GE Phoenix v|tome|x L 240 | Waygate Technologoies | micro computed tomography scanner | |
Hanks' Balanced Salt Solution | ThermoFisher Scientific | 14025092 | |
Heparin | Leo Pharma | B01AB01 | 5000 IE/mL |
Isolfurane | Baxter | FDG9623 | |
Methanol | ThermoFisher Scientific | 11413413 | |
MICROFIL | Flowtech | MV-122 | synthetic resin yellow |
MICROFIL | Flowtech | MV-120 | synthetic resin blue |
MICROFIL | Flowtech | MV-diluent | clear resin diluent |
Pasteur pipette | Verkon | 130.690.424.503 | |
Peristaltic pump | AgnThos | 010.6131.M20 | |
phoenix datos|x 2.0 software | Baker Hughes | CT data reconstruction software | |
Rocker | VWR | 444-0142 | |
Silk suture | AgnThos | 14757 | Black silk, 4-0, sterile, 100 m |
Skin scissor | Fine Science Tools | 14058-09 | Iris straight tip 9 cm |
Spring scissor | Fine Science Tools | 15000-03 | Vannas micro, straight tip 2 mm |
Syringe 1 mL Luer | BD bioscience | 303172 | |
Tubing PE10 | BD bioscience | 427401 | |
Tubing PE50 | BD bioscience | 427411 | |
VG Studio MAX 3.3 software | Volume Graphics GmbH | CT data processing and analysis software |