|Home | About | Journals | Submit | Contact Us | Français|
Optical projection tomography (OPT) is a technology ideally suited for imaging embryonic organs. We emphasize here recent successes in translating this potential into the field of live imaging. Live OPT (also known as 4D OPT, or time-lapse OPT) is already in position to accumulate good quantitative data on the developmental dynamics of organogenesis, a prerequisite for building realistic computer models and tackling new biological problems. Yet, live OPT is being further developed by merging state-of-the-art mouse embryo culture with the OPT system. We discuss the technological challenges that this entails and the prospects for expansion of this molecular imaging technique into a wider range of applications.
Optical projection tomography (OPT) is a young 3D imaging technology.1 OPT was created to fill a gap (1–10 mm across) in the range of specimen sizes encompassed by existing 3D imaging techniques.2,3 An overview of the standard OPT methodology, the consequences of using light for computed tomography (rather than, for example, X-rays or electron beams), and the basics of the reconstruction and visualization approaches can be found in ref. 4.
It will take more time to explore all potential biomedical applications of OPT, yet OPT has already had a significant impact in the field of organogenesis. Three main reasons can be identified: (1) OPT computes tomographic reconstructions and provides 3D global visualization in both transmission and fluorescent modes, a clear advantage for elucidating the anatomical and molecular complexity of organogenesis;5 (2) the size and optical properties of many organs or embryonic organ rudiments are suitable for OPT (especially for the mouse, the main biomedical animal model); and (3) OPT can play multiple imaging “roles” that can all benefit organogenesis research: virtual histology, 3D morphology, gene expression 3D pattern, pinpointing labeled cells, atlases of development, genetic screens and phenotyping mutants, and analysis of disease models.6
After summarizing the current applications of OPT 3D imaging (especially on fixed and cleared mouse embryonic specimens), the bulk of the review will be devoted to the examination of the most recent successes in extending OPT technology into the field of live imaging of organogenesis.7 Live OPT, in which the OPT set up and ex vivo organ culture were made compatible, has exciting potential as a molecular imaging approach to track tissue movements and dynamic gene expression in 3D over time. We will end discussing challenges ahead, in particular for approaching le milieu intérieur8 inside the tomograph. The goal is to describe organogenesis quantitatively and faithfully in order to generate an accurate computational modeling of organ development and to tackle new problems in organogenesis.
Since its implementation,1 OPT has mostly been used on fixed specimens after optical clearing in order to reduce photon scattering. The range of species successfully imaged now includes human,9 mouse,10,11 chick,12 reptile species,13 zebrafish,14 Drosophila15 and Arabidopsis.16 Mouse embryos are routinely imaged at E9.5 up to E12.5,17 and, for older specimens, organs in isolation are scanned, for example, the developing limb18 or brain19 at E14.5. The Edinburgh Mouse Atlas Project (EMAP) has also performed purely anatomical scanning (of autofluorescence) for a few whole embryos up to an age of ~E15.17 Thus, all the organogenesis stages of mouse development have been imaged with OPT. Younger and smaller embryos can also be imaged by OPT, but because previous techniques (such as confocal microscopy) were already able to capture these smaller specimens, they have not been considered an important application for OPT itself. Successes with bigger specimens such as whole organs taken from the adult mouse have also now been reported, for brain,20,21 pancreas,20 kidney5 and lungs.22 This opens up exciting new applications such as preclinical disease research, and accordingly, the whole pancreas imaging was performed quantitatively to compare the mass of β-cell tissue in normal versus diabetic specimens, using the NOD mouse model.20 A related type of soft tissue, but with a very different application, is human biopsy material. Although still at the exploratory phase, early tests look encouraging.6
The incentive is always strong for adapting a technological advance made in “static” imaging—such as OPT—to its “live” counterpart. Capturing the dynamics of a process in a single living individual over time often yields important information not obtainable from a time series of fixed samples. In the domain of optical tomography, recent years have seen the development of approaches—such as diffuse optical tomography (DOT) or fluorescence mediated tomography (FMT)—for producing 3D images of entire living adult mice.23 In these cases, photons must pass through 1 to 2 cm of living heterogeneous tissue, so the signals emerging are unavoidably weak and highly scattered and can only be reconstructed into fairly low-resolution images. Nevertheless, these data revealed biologically useful information, highlighting that a useful compromise always exists between our desire for high spatial resolution and the desire for longitudinal studies that capture the dynamics of a single individual over time.23 The number of scattering events a photon will experience depends on both the opacity of the tissue and the thickness of tissue to penetrate. If useful information can be obtained using FMT from scattered photons traveling through centimeters of living tissue, it therefore follows that imaging through millimeters of living tissue should provide higher resolution and could be useful for certain experiments. This is despite the fact that this resolution will nevertheless be significantly lower than when the sample is fixed and cleared (by which scattering is reduced to almost zero). For these reasons, it has been worth extending OPT technology into the field of time-lapse imaging—it again fills an optical imaging gap here for living specimens between optical sectioning techniques on the one hand (such as confocal microscopy and SPIM24) and diffuse optical tomography approaches on the other (such as DOT and FMT). The goal therefore is to image small living specimens 1 to 2 mm across.
Although a range of animal model systems exists whose embryos or organ explants grow happily on a microscope stage at room temperature (e.g., the anamniotes zebrafish and Xenopus), rodents are the primary research models for biomedical research. Unfortunately, with placental mammals, time-lapse microscopy with adequate resolution is not easily compatible with embryos developing in normal conditions, especially at post-implantation stages of organogenesis as it involves explantation and drastic environmental changes. This is even more delicate with 3D imaging (compared to 2D time-lapse) as it often requires longer and potentially phototoxic scans. Thus a particularly challenging line of research has been pursued to combine OPT technology with ex vivo organ culture techniques.7 The organ chosen was the developing limb bud of the mouse embryo since it is visually accessible (unlike the heart for example, which is concealed by other tissues) and is about 1 mm in size, therefore allowing reasonable transmission of photons.
A life-support chamber was designed to be compatible with the rotary stage required for OPT and with the growth of the limb bud (Fig. 3A). This chamber required reliable temperature control, the ability to avoid evaporation from the growth medium, and a method of supplying gas (in particular oxygen). A particularly important but difficult issue to solve was the need for controlling the angle of the organ explant with respect to the axis of rotation while the specimen was already inside the tomograph, without the user opening the chamber—as this would impair the careful temperature control. A micromanipulator was designed with six mechanical degrees-of-freedom (including three translations, one simple rotation and a 2D tilting surface) all controllable from outside the life-support chamber (Fig. 3B). Owing to these new hardware and the development of control software, a series of experiments for quantifying aspects of limb development were launched.
The goal of the first experiment was to build up a dynamic picture of limb growth by tracking the tissue-level movements of surface ectoderm. A method for creating fluorescent landmarks and tracking their 3D movements over time was developed. The specimen was imaged from 200 angles (every 1.8°) at 15-minute intervals in both fluorescent and transmission modes—the former to track the artificial landmarks and the latter to measure the overall shape of the organ. The datasets that were produced on tissue movements were both global—i.e., capturing the whole developing organ, rather than subregions—and dynamical. These dynamical measurements provided both rates of tissue displacement over time (Fig. 1A–C) and rates of surface expansion (Fig. 1D–F). Quantitative maps on tissue movements were extracted: a tissue velocity map (Fig. 1C) showing for the first time that a twisting motion is involved in normal limb development, and a map of regional expansion rates across the limb bud (Fig. 1E and F) showing a surprising degree of spatial variation.
A second major achievement of live OPT was made in the realm of molecular imaging. Live OPT was able to monitor a dynamically changing 3D gene expression pattern throughout the volume of a living tissue.7 The same model system as above was used, the mouse limb bud developing in culture, but this time the signal tracked over time was the signal emitted by the green fluorescent protein (GFP) reporting for transcriptional activities of the gene Scleraxis deep inside the mesenchyme of the organ rudiment.25 Live OPT imaging of the GFP fluorescence revealed how the expression domain of Scleraxis changes its shape and size in 3D over time (reviewed in ref. 7). It was the first direct 4D observation of dynamic spatial patterning of the mesenchymal tissue in a mammalian organ. Therefore a useful compromise between imaging depth and imaging resolution can also be found with the molecular modality of live OPT to extract new and significant 4D biological information.
A third, proof-of-concept experiment was also performed on a different sample: the head of a 9-day-old mouse embryo from a transgenic line expressing GFP driven by the control elements of the Pax6 gene.26 Figure 2 shows that good tomographic reconstructions through the head were achieved highlighting Pax6-GFP expression in the developing eye and brain. During the 6-hour time-lapse, virtual sections through the eye were able to record the early stages of lens induction and the resulting morphogenic shape changes. Thus, live OPT is not restricted to the study of limb development, but could become a general tool for a number of different model systems.
As mentioned above, useful biological information as gained with live OPT using organ culture could not have been obtained in vivo stricto sensu. In vivo, before reaching the mouse conceptus, photons would have to penetrate millimeters of adult opaque tissues and only low resolution images can be reconstructed. Therefore, imaging improvements beyond the 6 hours of normal organogenesis achieved in the limb bud explants will most likely come from a protocol which is less disruptive to the embryo, but which still minimizes the amount of tissue in the photon path. Such protocol is worth finding because: (1) 6 hours of normal growth is not quite enough for embracing many developmental processes, (2) preserving the integrity of the embryo could lead to imaging of developmental events for which there are no existing organotypic culture method available, (3) it is likely that most existing organotypic culture models cause subtle divergence from natural development due the absence of systemic influences such as the cardiovascular system, and that preserving blood circulation27 may be critical for faithful quantitative recording of organogenesis, (4) the main strength of the mouse model is in its molecular genetics and so, to take full advantage of investment into subtle germ line modifications (such as cell- or tissue-specific fluorescent protein labeling of deep tissues), it is a sensible endeavor to develop less invasive and durable live imaging technology.
Reducing the thickness of tissue to penetrate in order to improve resolution while keeping in vivo-like conditions has been so far attempted in two different manners: “in utero” after laparotomy and exposure of embryo's head at E16, or using whole embryo culture. The former requires the mother to be immobilized on the microscope stage for imaging28 which is difficult to combine with an OPT approach which requires the specimen to rotate. Whole embryo culture offers little help if any at organogenesis stages over ex vivo explant culture. In fact, it mostly leads to new problems as it requires rolling the embryo in a bottle which in principle precludes time-lapse imaging. Attempts have been made for static imaging of cultured embryos but the results concern at best the first day of organogenesis and are so far in 2D.29 Moreover, development of highly-vascularized structures, such as the limb buds, is deficient in this arrangement. A third option, imaging organogenesis while culturing the complete mouse conceptus (embryo + extraembryonic membranes) has to our knowledge never been explored. Our previous submerged, “semi-static” arrangement of culture inside the OPT chamber as designed for limb bud explant live imaging could be advantageous for culturing concepti in many ways. The minimal-support rotary arm using tungsten needles would insure that the conceptus suspended in the medium above the perfluorodecalin layer (Fig. 3C) is free of mechanical pressure, is rotated and properly oxygenated and already fully integrated into a 4D imaging system. Moreover, keeping the yolk sac and placenta intact was likely to provide some growth advantage for the embryo.30 However, it was not known whether the conceptus would survive the many steps of the surgical procedure, and whether any imaging gain could be obtained. Preliminary exploration of this new approach illustrates a potential route for the future. Mouse concepti of 10 days of age were cultured inside the live OPT scanner. Extra-embryonic membranes of the conceptus were windowed to facilitate tomography of the limb bud and improve transfer of gases and nutrients. This procedure fulfilled its first goal as it enabled healthy development of the limb bud to go beyond the previous 6-hour limit, allowing for long-term tomographic recording in the transmission mode of limb bud growth (Fig. 3). Moreover, the conceptus could be kept alive for up to 36 h—as judged by heart beating, and blood flowing in the embryo proper (especially in the marginal vein of the limb bud) and in the two extraembryonic circuits of the placenta and the yolk sac (Fig. 3). The latter implies that many events of organogenesis are now potentially accessible to 4D imaging during the 10th and 11th day of mouse embryo development. Because this method had been specifically developed for culturing and imaging the conceptus inside the live OPT scanner, it was given an acronym: LOOCC (Live OPT Opened Conceptus Culture) that underscores the mutual dependency of these two technologies.
The cardiovascular system is the first organ system at work in vertebrate embryos. In mouse embryo, circulation develops and sustains growth through the coordinated emergence of cardiac function, a vascular network and maturing red cells between the 8th and 10th day of gestation,31 i.e., at the onset of organogenesis. The heart starts beating during the 8th day of gestation and by the 10th day reaches a rate of 200 beats per minute.31 The cardiovascular system is therefore a source of embryonic motion during all stages of organogenesis, even before the appearance of first embryonic muscular contractions during the 12th day.32
Thus, efforts to preserve an intact circulatory system for ex utero imaging as in LOOCC come with a reward: records recapitulate more faithfully natural organogenesis. Indeed, organ rudiments such as the limb bud are early and highly vascularized.33,34 The blood flow is clearly visible in the marginal vein of limb buds of 10-day embryo developing in LOOCC (Fig. 3E). However, there is also a fundamental drawback: heartbeat and blood flow induce motions that greatly complicates 4D image processing especially with the fairly long exposure time required for fluorescence signal acquisition during each tomographic scan. There is the risk, while gaining in healthier and longer development, to lose spatial resolution attained with organ culture approaches. It is crucial though to accept and cope with this negative impact of heartbeat-induced motions to overcome limitations on longitudinal recording of the dynamics of normal organogenesis. Thus, the next challenge for live OPT will be to work out the conditions for 4D fluorescence imaging that cope with motion artifact caused by heartbeats. The ultimate goal is to track fluorescent signals in transgenic concepti using live OPT, as was done previously in organ explants. Technical improvements to the imaging and computational processing have to be explored to correct for these movements.35,36
Besides computational improvements, progress may also come from better cameras, better labels such as in transgenic mice carrying new-generation fluorescent proteins37 or from other fields such as similar live imaging research on different mouse stages38 or on organogenesis in different species.39–41 Also, extending LOOCC beyond the 11th day of embryonic development to access further organogenesis stages should benefit from research on blood circulation in the conceptus,30,42 culture media used for whole embryo culture and refinements in the hardware of the live tomograph itself. In parallel, it is likely that progress will be made in developing organotypic culture models43 allowing for live OPT imaging of many developmental events. It is through the combination of these explorations that the range of applications of live OPT will expand and that a more complete representation of organogenesis will be obtained.
We wish to show appreciation for Dr. Marit Boot's pioneering contributions to the development of the live OPT approach. Work is supported by CRG, ICREA, the Spanish Ministry of Science (MEC) and European Community.
Previously published online: www.landesbioscience.com/journals/organogenesis/article/10426