Number and percentage of children reporting cognitive features of robot (N = 184).
\r\n\t
",isbn:"978-1-83969-150-8",printIsbn:"978-1-83969-149-2",pdfIsbn:"978-1-83969-151-5",doi:null,price:0,priceEur:0,priceUsd:0,slug:null,numberOfPages:0,isOpenForSubmission:!1,hash:"7409b2acd5150a93004300800918b736",bookSignature:"Prof. Karmen Pažek",publishedDate:null,coverURL:"https://cdn.intechopen.com/books/images_new/10548.jpg",keywords:"Lean Manufacturing, Agriculture, Production and Process, Costs Reduction, Lean Principles, Industry, Tools, Implementation, Sustainability, Modeling, Environment, Planning",numberOfDownloads:7,numberOfWosCitations:0,numberOfCrossrefCitations:0,numberOfDimensionsCitations:0,numberOfTotalCitations:0,isAvailableForWebshopOrdering:!0,dateEndFirstStepPublish:"October 20th 2020",dateEndSecondStepPublish:"November 17th 2020",dateEndThirdStepPublish:"January 16th 2021",dateEndFourthStepPublish:"April 6th 2021",dateEndFifthStepPublish:"June 5th 2021",remainingDaysToSecondStep:"2 months",secondStepPassed:!0,currentStepOfPublishingProcess:4,editedByType:null,kuFlag:!1,biosketch:"Dr. Pažek is Head of the undergraduate study program Agricultural economics and rural development and Vice-dean for education. She is the author or co-author of 61 scientific papers, 6 scientific books, and 24 book chapters.",coeditorOneBiosketch:null,coeditorTwoBiosketch:null,coeditorThreeBiosketch:null,coeditorFourBiosketch:null,coeditorFiveBiosketch:null,editors:[{id:"179642",title:"Prof.",name:"Karmen",middleName:null,surname:"Pažek",slug:"karmen-pazek",fullName:"Karmen Pažek",profilePictureURL:"https://mts.intechopen.com/storage/users/179642/images/system/179642.jpg",biography:"Karmen Pažek achieved her Ph.D. at University of Maribor, Faculty of Agriculture in 2006. She is active as Full Professor for Farm management in the Department for Agriculture Economics and Rural Development on Faculty of Agriculture and Life Sciences, University of Maribor. Her research includes development of decision support tools and systems for farm management (simulation modeling, multi-criteria decision analysis, option models, investment analysis) and economics of agricultural production. She is involved in teaching activities as thesis supervisor at postgraduate study programs and involved in national and international research projects. She is author or coauthor of 61 scientific papers (including 34 papers in journals with impact factor), 6 scientific books and 24 book chapters. Currently she is Head of the undergraduate study program Agricultural economics and rural development and Vice dean for education. \r\n\r\nAcademic activities\r\nResearch:\r\n-\tFarm management\r\n-\tDecision support, simulation, forecasting, multi criteria decision making in the area of agriculture with emphasis on field crops, farm tourism and fruit producon\r\n\r\nCurrent Research work:\r\n- Financial parameters assessment based on perfect and in-perfect information in agrifood \r\n systems \r\n- Option modeling of agrifood projects\r\n-\tEfficiency assessment in farm tourism \r\n-\tEfficiency of sugar beet production systems \r\n\r\nTeaching:\r\nUndergraduate Programmes and Courses\r\n-\tFarm management I and II\r\n-\tIntroduction to decision theory\r\n-\tOrganic fam management\r\n-\tManagement od supplementary activities\r\n-\tEconomics and management of rural tourism\r\n-\tSelected issues in agricultural entrepreneurship\r\n\r\nMaster Programmes and Courses\r\n\r\n-\tResearch methods in farm management\r\n-\tDecision theory\r\n-\tProject planning and quality management\r\n-\tOrganic fam management\r\n\r\n \r\nPhD Programme and Course\r\n\r\n-\tProject management (transferable skills)\r\n-\tSelected issues in farm management",institutionString:"University of Maribor",position:null,outsideEditionCount:0,totalCites:0,totalAuthoredChapters:"3",totalChapterViews:"0",totalEditedBooks:"0",institution:{name:"University of Maribor",institutionURL:null,country:{name:"Slovenia"}}}],coeditorOne:null,coeditorTwo:null,coeditorThree:null,coeditorFour:null,coeditorFive:null,topics:[{id:"11",title:"Engineering",slug:"engineering"}],chapters:[{id:"74769",title:"Development of Integrated Lean Six Sigma-Baldrige Framework for Manufacturing Waste Minimization: A Case of NAS Foods Plc",slug:"development-of-integrated-lean-six-sigma-baldrige-framework-for-manufacturing-waste-minimization-a-c",totalDownloads:7,totalCrossrefCites:0,authors:[null]}],productType:{id:"1",title:"Edited Volume",chapterContentType:"chapter",authoredCaption:"Edited by"},personalPublishingAssistant:{id:"247865",firstName:"Jasna",lastName:"Bozic",middleName:null,title:"Ms.",imageUrl:"https://mts.intechopen.com/storage/users/247865/images/7225_n.jpg",email:"jasna.b@intechopen.com",biography:"As an Author Service Manager, my responsibilities include monitoring and facilitating all publishing activities for authors and editors. From chapter submission and review to approval and revision, copyediting and design, until final publication, I work closely with authors and editors to ensure a simple and easy publishing process. I maintain constant and effective communication with authors, editors and reviewers, which allows for a level of personal support that enables contributors to fully commit and concentrate on the chapters they are writing, editing, or reviewing. I assist authors in the preparation of their full chapter submissions and track important deadlines and ensure they are met. I help to coordinate internal processes such as linguistic review, and monitor the technical aspects of the process. As an ASM I am also involved in the acquisition of editors. Whether that be identifying an exceptional author and proposing an editorship collaboration, or contacting researchers who would like the opportunity to work with IntechOpen, I establish and help manage author and editor acquisition and contact."}},relatedBooks:[{type:"book",id:"1591",title:"Infrared Spectroscopy",subtitle:"Materials Science, Engineering and Technology",isOpenForSubmission:!1,hash:"99b4b7b71a8caeb693ed762b40b017f4",slug:"infrared-spectroscopy-materials-science-engineering-and-technology",bookSignature:"Theophile Theophanides",coverURL:"https://cdn.intechopen.com/books/images_new/1591.jpg",editedByType:"Edited by",editors:[{id:"37194",title:"Dr.",name:"Theophanides",surname:"Theophile",slug:"theophanides-theophile",fullName:"Theophanides Theophile"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"3092",title:"Anopheles mosquitoes",subtitle:"New insights into malaria vectors",isOpenForSubmission:!1,hash:"c9e622485316d5e296288bf24d2b0d64",slug:"anopheles-mosquitoes-new-insights-into-malaria-vectors",bookSignature:"Sylvie Manguin",coverURL:"https://cdn.intechopen.com/books/images_new/3092.jpg",editedByType:"Edited by",editors:[{id:"50017",title:"Prof.",name:"Sylvie",surname:"Manguin",slug:"sylvie-manguin",fullName:"Sylvie Manguin"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"3161",title:"Frontiers in Guided Wave Optics and Optoelectronics",subtitle:null,isOpenForSubmission:!1,hash:"deb44e9c99f82bbce1083abea743146c",slug:"frontiers-in-guided-wave-optics-and-optoelectronics",bookSignature:"Bishnu Pal",coverURL:"https://cdn.intechopen.com/books/images_new/3161.jpg",editedByType:"Edited by",editors:[{id:"4782",title:"Prof.",name:"Bishnu",surname:"Pal",slug:"bishnu-pal",fullName:"Bishnu Pal"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"72",title:"Ionic Liquids",subtitle:"Theory, Properties, New Approaches",isOpenForSubmission:!1,hash:"d94ffa3cfa10505e3b1d676d46fcd3f5",slug:"ionic-liquids-theory-properties-new-approaches",bookSignature:"Alexander Kokorin",coverURL:"https://cdn.intechopen.com/books/images_new/72.jpg",editedByType:"Edited by",editors:[{id:"19816",title:"Prof.",name:"Alexander",surname:"Kokorin",slug:"alexander-kokorin",fullName:"Alexander Kokorin"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"1373",title:"Ionic Liquids",subtitle:"Applications and Perspectives",isOpenForSubmission:!1,hash:"5e9ae5ae9167cde4b344e499a792c41c",slug:"ionic-liquids-applications-and-perspectives",bookSignature:"Alexander Kokorin",coverURL:"https://cdn.intechopen.com/books/images_new/1373.jpg",editedByType:"Edited by",editors:[{id:"19816",title:"Prof.",name:"Alexander",surname:"Kokorin",slug:"alexander-kokorin",fullName:"Alexander Kokorin"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"57",title:"Physics and Applications of Graphene",subtitle:"Experiments",isOpenForSubmission:!1,hash:"0e6622a71cf4f02f45bfdd5691e1189a",slug:"physics-and-applications-of-graphene-experiments",bookSignature:"Sergey Mikhailov",coverURL:"https://cdn.intechopen.com/books/images_new/57.jpg",editedByType:"Edited by",editors:[{id:"16042",title:"Dr.",name:"Sergey",surname:"Mikhailov",slug:"sergey-mikhailov",fullName:"Sergey Mikhailov"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"371",title:"Abiotic Stress in Plants",subtitle:"Mechanisms and Adaptations",isOpenForSubmission:!1,hash:"588466f487e307619849d72389178a74",slug:"abiotic-stress-in-plants-mechanisms-and-adaptations",bookSignature:"Arun Shanker and B. Venkateswarlu",coverURL:"https://cdn.intechopen.com/books/images_new/371.jpg",editedByType:"Edited by",editors:[{id:"58592",title:"Dr.",name:"Arun",surname:"Shanker",slug:"arun-shanker",fullName:"Arun Shanker"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"878",title:"Phytochemicals",subtitle:"A Global Perspective of Their Role in Nutrition and Health",isOpenForSubmission:!1,hash:"ec77671f63975ef2d16192897deb6835",slug:"phytochemicals-a-global-perspective-of-their-role-in-nutrition-and-health",bookSignature:"Venketeshwer Rao",coverURL:"https://cdn.intechopen.com/books/images_new/878.jpg",editedByType:"Edited by",editors:[{id:"82663",title:"Dr.",name:"Venketeshwer",surname:"Rao",slug:"venketeshwer-rao",fullName:"Venketeshwer Rao"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"4816",title:"Face Recognition",subtitle:null,isOpenForSubmission:!1,hash:"146063b5359146b7718ea86bad47c8eb",slug:"face_recognition",bookSignature:"Kresimir Delac and Mislav Grgic",coverURL:"https://cdn.intechopen.com/books/images_new/4816.jpg",editedByType:"Edited by",editors:[{id:"528",title:"Dr.",name:"Kresimir",surname:"Delac",slug:"kresimir-delac",fullName:"Kresimir Delac"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"3621",title:"Silver Nanoparticles",subtitle:null,isOpenForSubmission:!1,hash:null,slug:"silver-nanoparticles",bookSignature:"David Pozo Perez",coverURL:"https://cdn.intechopen.com/books/images_new/3621.jpg",editedByType:"Edited by",editors:[{id:"6667",title:"Dr.",name:"David",surname:"Pozo",slug:"david-pozo",fullName:"David Pozo"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}}]},chapter:{item:{type:"chapter",id:"15865",title:"Robot Arm-Child Interactions: A Novel Application Using Bio-Inspired Motion Control",doi:"10.5772/16704",slug:"robot-arm-child-interactions-a-novel-application-using-bio-inspired-motion-control",body:'Robot arms were originally designed in the 1960s for intended use in a wide variety of industrial and automation tasks such as fastening (e.g., welding and riveting), painting, grinding, assembly, palleting and object manipulation). In these tasks humans were not required to directly interact or cooperate with robot arms in any way. Robots, thus, did not require sophisticated means to perceive their environment as they interacted within it. As a result, machine type motions (e.g., fast, abrupt, rigid) were suitable with little consideration made of how these motions affect the environment or the users. The application fields of robot arms are now extended well beyond their traditional industrial use. These fields include physical interactions with humans (e.g., robot toys) and even emotional support (e.g., medical and elderly services).
In this chapter we begin by presenting a novel motion control approach to robotic design that was inspired by studies from the animal world. This approach combines the robot’s manipulability aspects with its motion (e.g., in case of mobile robots such as humanoids or traditional mobile manipulators) to enable robots to physically interact with their users while adapting to changing conditions triggered by the user or the environment. These theoretical developments are then tested in robot-child interaction activities, which is the main focus of this chapter. Specifically, the children’s relationships (e.g., friendship) with a robotic arm are studied. The chapter concludes with speculation about future use and application of robot arms while examining the needs for improved human-robot interactions in a social setting including physical and emotional interaction caused by human and robot motions.
There are many different fields of human-robot interaction that have been developed within the last decade. The intelligent fusion scheme for human operator command and autonomous planner in a telerobotic system is based on the event based planning introduced in Chuanfan, 1995. This scheme integrates a human operator control command with an action planning and control for autonomous operation. Basically, a human operator passes his/her commands via the telerobotic system to the robot, which, in turn, executes the desired tasks. In many cases both an extender and material handling system are required during the implementation of tasks. To achieve proper control, force sensors have
been used to measure the forces and moments provided by the human operator [e.g., Kim, 1998]. The sensed forces are then interpreted as the desired motion (translational and rotational) while the original compliant motion for the robot remains effective. To improve previous works, video and voice message has been employed, [e.g., Wikita, 1998], for information sharing during the human-robot cooperation. The projection function of the video projector is to project the images of the messages from the robot into an appropriate place. The voice message has the function to share the event information from the robot to the human. Fukuda et al. proposed a human-assisting manipulator teleoperated by electromyography [Fukuda, 2003]. The works described above simplify the many different applications in the field of human-robot interaction. The control mechanism presented herein allows robots to cooperate with humans where humans practically employ no effort during the cooperation task (i.e., minimal effort during command actions). Moreover, in contrast to previous work, where the human-robot cooperation takes place in a well structured engineered environment, the proposed mechanism allows cooperation in outdoor complex/rough terrains.
Human-robot arm manipulator coordination for load sharing
Several researchers have studied the load sharing problem in the dual manipulator coordination paradigm [e.g., Kim, 1991]. Unfortunately, these results cannot be applied in the scope of the human-arm-manipulator coordination. The reason is that in the dual manipulator coordination, the motions of the manipulators are assumed to be known. However, in the human-arm-manipulator coordination, the motion of the object may be unknown to the manipulator. A number of researchers have explored the coordination problem between a human arm and a robot manipulator using compliant motion, predictive control and reflexive motion control [Al-Jarrah, 1997; Al-Jarrah and Zheng, 1997; Iqbal, 1999]. In such scenarios the human-arm, by virtue of its intelligence, is assumed to lead the task while the manipulator is required to comply with the motion of the arm and support the object load. The intelligence of the arm helps perform complex functions such as task planning and obstacle avoidance, while the manipulator only performs the load sharing function. By coordinating the motions of the robotic arm with the user’s arm, the uncertainty due to the environment can be reduced while load sharing can help reduce the physical strain in the human.
Complaint control\n\t\t\t\t
The basic ability for a robot to cooperate with a human is to respond to the human’s intentions. Complaint motion control has been used to achieve both load sharing and trajectory tracking where the robot’s motion along a specific direction is called complaint motion. This simple but effective technique can be used to guide the robot as it attempts to eliminate the forces sensed (i.e., precise human-robot interaction). However, diverse problems might occur that require different control approaches.
Predictive control\n\t\t\t\t
The problem in the framework of model-based predictive control for human-robot interaction has been addressed in numerous papers [e.g., Iqbal, 1999]. First, the transfer function from the manipulator position command to the wrist’s sensor force output is defined. Then, the desired set point for the manipulator force is set to equal the gravitational force. Numerous results reported in the literature indicate that predictive control allows the manipulator to effectively take over the object load, and the human’s forces (effort) stays close to zero. Moreover, manipulators have been shown to be highly responsive to the human’s movement, and relatively small arm force can effectively initiate the manipulation task. However, difficulties still remain when sudden large forces are exerted to the robot to change the motion of the shared object (load) as the robot arm acts as another automated load to the human.
Reflexive motion control
Al-Jarrah [1997] proposed reflexive motion control for solving the loading problem, and an extended reflexive control was shown to improve the speed of the manipulator in response to the motion of the human. The results show that the controller anticipated the movements of the human and applied the required corrections in advance. Reflexive control, thus, has been shown to assist the robot in comprehending the intentions of the human while they shared a common load. Reflexive motion is an inspiration from biological systems; however, in reflexive motion control it is assumed that the human and the manipulator are both always in contact with an object. That is, there is an object which represents the only communication channel between the robot and the human. This is not always possible. Thus, mechanisms that allow human-robot cooperation without direct contact are needed.
In an attempt to enhance pure human-robot arm cooperation, human-mobile manipulator cooperation applications have been proposed [e.g., Jae, 2002; Yamanaka, 2002; Hirata, 2005; Hirata, 2007]. Here the workspace of the cooperation is increased at the expense of the added complexity introduced by the navigation aspects that need to be considered. Accordingly, humans cooperate with autonomous mobile manipulators through intention recognition [e.g., Fernandez, 2001]. Herein mobile-manipulators refer to ground vehicles with robot arms (Fig. 1a), humanoid robots, and aerial vehicles having grasping devices (Fig. 1b). In contrast to human-robot arm cooperation, here the cooperation problem increases as the mobile manipulator is not only required to comply with the human’s intentions but simultaneously perceives the environment, avoids obstacles, coordinates the motion between the vehicle and the manipulator, and copes with terrain/environment irregularities/uncertainties, all of this while making cooperation decisions, not only between human and robot but also between the mobile-base and robot arm in real-time. This approach has been designated as active cooperation where diverse institutions are running research studies. Some work extends the traditional basic kinematic control schemes to master-slave mechanisms where the master role of the task is assigned to the actor (i.e., human) having better perception capabilities. In this way, the mobile manipulator not only is required to comply with the force exerted by the human while driving the task, but also contributes with its own motion and effort. The robot must respond to the master’s intention to cooperate actively in the task execution. The contribution of this approach is that the recognition process is applied on the frequency spectrum of the force-torque signal measured at the robot’s gripper. Previous works on intention recognition are mostly based on monitoring the human’s motion [Yamada, 1999] and have neglected the selection of the optimal robot motion that would create a true human-robot interaction, reducing robot slavery and promoting human-robot friendship. Thus, robots will be required not only to help and collaborate, but to do so in a friendly and caring way. Accordingly, the following section presents a simple yet effective robot control approach to facilitate human-robot interaction.
Schematic diagrams of: a) Mobile manipulator, and b) Aerial robot with robotic arm.
The objective of this section is to briefly present, without a detailed mathematical analysis, a simple yet effective human-robot cooperation control mechanism capable of achieving the following two objectives: i) Cooperation between a human and a robot arm in 3D dimensions, and ii) Cooperation between a human and a mobile-manipulator moving on rough terrain. Here the focus is placed on the former aspect as it is directly related to the experiments discussed in Section 3.
Many solutions have been developed for human-robot interaction; however, current techniques work primarily when cooperation occurs on simple engineered environments, which prevents robots from working in cooperation with humans in real human settings (e.g., playgrounds). Despite the fact that the control methodology presented in this section can be used in a number of mobile manipulators (e.g., ground and aerial) cooperating with humans, herein we focus on the cooperation between a human and a robot arm in 3D dimensions. This application requires a fuzzy logic force velocity feedback control to deal with unknown nonlinear terms that need to be resolved during the cooperation. The fuzzy force logic control and the robot’s manipulability are used and applied to the control algorithm. The goal of using these combined techniques is to ensure that the design of the control system is stable, reliable, and applicable in a wide range of human cooperation areas. Herein, we specially consider those areas and settings where the associated complexities that humans and their environments impose on the system (robot arm) have a significant impact. When interaction occurs, the dynamic coupling between the end-effector (i.e., robot arm) and the environment becomes important. In a motion and force control scenario, interaction affects the controlled variables, introducing error upon which the controller must act. Even though it is usually possible to obtain a reasonably accurate dynamic model of the manipulator, the main difficulties occur from the dynamic coupling with the environment and similarly with the human. The latter is, in general, impossible to model due to time variation. Under such conditions a stable manipulator system could usually be destabilized by the environment/human coupling. Although a number of control approaches of robot interaction have been developed in the last three decades the compliant motion control can be categorized as the one performing well within the above described problems. This is due to the fact that compliant motion control uses indirect and direct force control. The main difference between these two approaches is that the former achieves force control via motion control without an explicit force feedback loop, while the latter can regulate the contact (cooperation) force to a desired value due to the explicit force feedback control loop. The indirect force control includes compliance (or stiffness) and impedance control with the regulation of the relation between position and force (related to the notion of impedance or admittance). The manipulator under impedance control is described by an equivalent mass-spring-damper system with the contact force as input. With the availability of a force sensor, the force signal can be used in the control law to achieve linear and decoupled impedance. Impedance control aims at the realization of a suitable relation between the forces and motion at the point of interaction between the robot and the environment. This relation describes the robot’s velocity as a result of the imposed force(s). The actual motion and force is then a result of the imposed impedance, reference signals, and the environment admittance.
It has been found by a number of researchers that impedance control is superior over explicit force control methods (including hybrid control). However, impedance control pays the price of accurate force tracking, which is better achieved by explicit force control. It has also been shown that some particular formulations of hybrid control appear as special cases of impedance control and, hence, impedance control is perceived as the appropriate method for further investigation related to human-robot arm cooperation. Hybrid motion/force control is suitable if a detailed model of the environment (e.g., geometry) is available. As a result, the hybrid motion/force control has been a widely adopted strategy, which is aimed at explicit position control in the unconstrained task direction and force control in the constrained task direction. However, a number of problems still remain to be resolved due to the explicit force control in relation to the geometry.
Control architecture of human robot arm cooperation
To address the problems found in current human-robot cooperation mechanisms, a new control approach is described herein. The approach uses common known techniques and combines them to maximize their advantages while reducing their deficiencies. Figure 2 shows the proposed human-mobile robot cooperation architecture that is used in its simplified version in human-robot arm cooperation described in Section 3.
In this architecture the human interacting with the robot arm provides the external forces and moments to which the robot must follow. For this, the human and the robot arm are considered as a coupled system carrying a physical or virtual object in cooperation. When a virtual object is considered, virtual forces are used to represent the desired trajectory and velocities that guide the robot in its motion. In this control method the human (or virtual force) is considered as the master while the robot takes the role of the slave. To achieve cooperation, the changes in the force values, which can be measured via a force/torque (F/T) sensor, must be initialized before starting the cooperation. Subsequently, when the cooperation task starts, the measured forces will, in general, be different than the initialized values. As a result, the robot will attempt to reduce such differences to zero. According to the force changes, the robot determines its motion (trajectory and velocity) to compensate the changing in F/T values. Thus, the objective of the control approach is to eliminate (minimize) the human effort in the accomplishment of the task. When virtual forces are used instead of direct human contact with the robot the need to re-compute the virtual forces is eliminated.
Flow chart of the human-mobile robot cooperation.
Motion decomposition of the end-effector
The manipulability (w) of the robot arm captures the relation between the singular point and the gripper’s end point. Here, the manipulability function of the robot arm (Fig. 2) is used to decompose the end-effector’s desired motion based on the value of w. First the maximum w value of the arm has to be known before it can be used. If the manipulability is small, the end point of the robot’s gripper is close to the singular point of the manipulator. That is, the capability of the robot arm to effectively react to the task while cooperating is reduced. On the other hand, if the value of w (manipulability) is large, the end point of the robot is far from the its singular point and the manipulator will find it easier to perform cooperating actions.
Thus the goal is to maintain the manipulability of the arm (and the mobility of the vehicle if working with a mobile manipulator) as large as possible, thus allowing the arm (and the vehicle when used) to effectively react to the unknown conditions of the environment and the cooperation tasks simultaneously. The fuzzy logic controller in Figure 2 is important in this case as the fuzzy rules can easily be tuned and used to distribute the robot arm’s motion based on the manipulability value and the geometry of the environment (e.g., as the robot arm overcomes obstacles).
Control architecture of human-mobile manipulator cooperation
To finalize this section the cooperation between a human and a mobile manipulator is described for completeness. The motion of a mobile base is subject to holonomic or nonholonomic kinematics constraints, which renders the control of mobile manipulators very challenging, especially when robots work in non-engineered environments. To achieve the cooperation between the human and a mobile manipulator, a set of equations to represent the changes in forces and torques on the robot’s arm caused by the interaction of the mobile manipulator on rough terrains is required. These equations can take different forms depending on the type or robot systems used (e.g., sensors). However, all forces and torques should be a function of the roll, pitch, and yaw angles of the vehicle as it moves. These formulations will indicate what portion of the actual sensed force must be considered for effective cooperation (i.e., human intention) and which portion is to be neglected (i.e., reaction forces due to the terrain or the disturbances encountered by the robot).
The control system of the manipulator for human-robot cooperation/interaction was designed considering the operational force by the human (operator) and the contact force between the manipulator and the mobile robot. The interacting force can be measured by a F/T sensor which can be located between the final link of the manipulator and the end-effector (i.e., the wrist of the manipulator). The human force and the operational force applied by the human operator denote the desired force for the end-effector to move while compensating the changing in the forces. The final motion of the manipulator is determined by the desired motion by the human force controller. To allow the arm to be more reactive to unknown changes (due to the human and the environment) the manipulability of the arm must be continuously computed. As the arm approaches the limits of its working environment the motion of the mobile manipulator relies more on the mobile base rather than the arm. In this way, the arm is able to reposition itself in a state where it is able to move reactively. In the experiments used in the next section the mobile base was removed. This facilitated the tests while simultaneously enhancing the cooperation.
The above control mechanism (Fig. 2) not only enhances human-robot cooperation but also enhances their interaction. This is due to the fact that the robot reacts not only to the human but also to the environmental conditions. This control mechanism was implemented in the studies presented in the following section.
We designed a series of experiments to explore children’s cognitive, affective, and behavioral responses towards a robot arm under a controlled task. The robot is controlled using a virtual force representing a hypothetical human-robot interaction set a priori. The goal of using such control architecture was to enable the robot to appear dexterous, flexible while operating with smooth, yet firm biological type motions. The objective was to enhance and facilitate the human-robot cooperation/interaction with children.
Experimental setup
A robot arm was presented as an exhibit in a large city Science Centre. This exhibit was used in all the experimental studies. The exhibit was enclosed with a curtain within a 20 by 7 foot space (including the computer area). A robot arm was situated on a platform with a chair placed.56 meters from its 3D workspace to ensure safety. Behind a short wall of the robot arm was one laptop used to run the commands to the robotic arm and a second laptop connected to a camera positioned towards the child to conduct observations of children’s helping and general behaviors.
All three studies employed a common method. A researcher randomly selected visitors to invite them to an exhibit. The study was explained, and consent was obtained. Each child was accompanied behind a curtain where the robot arm was set up, with parents waiting nearby. Upon entering an enclosed space, the child was seated in front of a robot arm. Once the researcher left, the child then observed the robot arm conduct a block stacking task (using the bio-inspired motion control mechanisms described in Section 2). After stacking five blocks, it dropped the last block, as programmed.
Design and characteristics of the employed robot arm
The robot arm used in these experiments was a small industrial electric robot arm having 5 degrees of freedom where pre-programmed bio-inspired control mechanisms were implemented. To aesthetically enhance the bio-inspired motions of the robot the arm was “dressed” in wood, corrugated cardboard, craft foam, and metal to hide its wires and metal casing. It was given silver buttons for eyes, wooden cut-outs for ears, and the gripper served as the mouth. The face was mounted at the end of the arm, creating an appearance of the arm as the neck. Gender neutral colors (yellow, black, and white) were given to a non-specific gender. Overall, it was decorated to appear pleasant, without creating a likeness of an animal, person, or any familiar character yet having smooth natural type motions.
In addition to these physical characteristics, its behaviour was friendly and familiar to children. That is, it was programmed to pick up and stack small wooden blocks. Most children own and have played with blocks, and have created towers just as the robot arm did. This familiarity may have made the robot arm appear endearing and friendly to the children.
The third aspect of the scenario that was appealing to the children was that it was programmed to exhibit several social behaviours. Its face was in line with the child’s face to give the appearance that it was looking at the child. Also, as it picked up each block with its grip (decorated as the mouth), it raised its head to appear to be looking at the child before it positioned the block in the stack. Such movement was executed by the robot by following a virtual pulling force simulating how a human would guide another person when collaborating in moving objects. Then, as it lifted the third block, the mouth opened slightly to drop the block and then opened wider as if to express surprise at dropping it. It then looked at the child, and then turned towards the platform. In a sweeping motion it looked back and forth across the surface to find the block. After several seconds it then looked up at the child again, as if to ask for help and express the inability to find the block.
Five degree of freedom robot arm on platform with blocks.
Measures
The child’s reactions to the robot arm were observed and recorded. Then the researcher returned to the child to conduct a semi-structured interview regarding perceptions of the robot arm. In total, 60 to 184 boys and girls between the ages of 5 to 16 years (M = 8.18 years) participated in each study. We administered 15 open-ended questions. Three questions asked for general feedback about the arm’s appearance, six questions referred to the robot’s animistic characteristics, and six questions asked about friendship. These data formed the basis of three separate areas of study. First, we explored whether children would offer assistance to a robot arm in a block stacking task. Second, we examined children’s perceptions of whether the arm was capable of various thoughts, feelings, and behaviours. Finally, the children’s impressions about friendship with the robot arm were investigated.
Only a generation ago, children spent much of their leisure time playing outdoors. These days, one of the favourite leisure activities for children is using some form of advanced technological device (York, Vandercook, & Stave, 1990). Indeed, children spend 2-4 hours each day engaged in these forms of play (Media Awareness Network, 2005). Robotics is a rapidly advancing field of technology that will likely result in mass production of robots to become as popular as the devices children today enjoy. With robotic toys such as Sony’s AIBO on the market, and robots being developed with more advanced and sensitive responding capabilities, it is crucial to ask how children regard these devices. Would children act towards robots in a similar way as with humans? Would children prefer to play with a robot than with another child? Would they develop a bond with a robot? Would they think it was alive? Given that humans are likely to become more reliant upon robots in many aspects of daily life such as manufacturing, health care, and leisure, we must explore their psycho-social impact. The remainder of this chapter takes a glimpse on this potential impact on children by determining their reactions to a robot arm. Specifically, this section will explain whether children would offer assistance to a robot, perceive a robot as having humanistic qualities, and would consider having a robot as a friend.
Study 1: Assistance to a Robot Arm
Helping, or prosocial behaviours are actions intended to help or benefit another individual or group of individuals (Eisenberg & Mussen, 1989; Penner, Dovidio, Pilavin, & Schroeder, 2005). With no previous research to guide us, we tested several conditions in which we believed children would offer assistance (see Beran et al. 2011). The one reported here elicited the most helping behaviors.
Upon sitting in front of the robot arm the researcher stated the following:
Are you enjoying the science centre? What’s your favorite part?
This is my robot (researcher touches platform near robot arm). What do you think?
My robot stacks blocks (researcher runs fingers along blocks).
I’ll be right back.
The researcher then exited and observed the child’s behaviors on the laptop. A similar number of children, who did not hear this introduction, formed the comparison group. As soon as children in each group were alone with the robot arm, it began stacking blocks. A significantly larger number of children in the introduction group (n = 17, 53.1%), than in the comparison group (n = 9, 28.1%), helped the robot stack the blocks, X2(1) = 4.15, p = 0.04. Thus, children are more likely to offer assistance for a robot when they hear a friendly introduction than when they receive no introduction. We interpret these results to suggest that the adult’s positive statements about the robot modeled to the child positive rapport regarding the robot arm, which may have created an expectation for the child to have a positive exchange with it. Having access to no other information about the robot, children may have relied on this cue to gauge how to act and feel in this novel experience. Interestingly, at the end of the experiment, the researcher noted anecdotally that many children were excited to share their experience with their parents, asked the parents to visit the robot, and explained that they felt proud to have helped the robot stack blocks. Other children told their parents that they did not help the robot because they believed that it was capable of finding the block itself. Overall, we speculate that the adult’s display of positive regard towards the robot impacted children’s offers of assistance towards it.
Study 2: Animistic impressions of a Robot Arm
Animism as a typical developmental stage in children has been studied for over 50 years, pioneered by Piaget (1930; 1951). It refers to the belief that inanimate objects are living. This belief, according to Piaget, occurs in children up to about 12 years of age. The disappearance of this belief system by this age has been supported by some studies (Bullock, 1985; Inagaki and Sugiyama, 1988) but not others (Golinkoff et al., 1984; Gelman and Gottfried, 1983). Nevertheless, the study of animism is relevant in exploring how children perceive an autonomous robot arm.
Animism can be divided and studied within several domains. These may include cognitive (thoughts), affective (feelings), and behavioural (actions) beliefs, known as schemata. In other words, people possess schemata, or awareness, that human beings have abilities for thinking, feeling, and acting. More specifically, thinking abilities may include memory and knowledge; feeling abilities include pleasant and unpleasant emotions; and behaviour abilities can refer to physical abilities and actions. Melson et al. (2009) provide some initial insights into several of these types of beliefs children hold towards a robotic pet (Sony’s AIBO). Also, Melson et al. (2005) found that many children believed that such a robot was capable of the feelings of embarrassment and happiness, as well as recognition. Additional evidence of animism towards a robot was obtained by Bumby and Dautenhahn (1999) who reported that children may include human characteristics about robots in stories they create.
The most recent study on animism presents surprising insights about animism. A team of researchers from the University of Washington\'s Institute for Learning and Brain Sciences [I-LABS, 2010] found that, “babies can be tricked into believing robots are sentient”. The researchers used a remote-controlled robot in a skit to act in a friendly manner towards its human (i.e., adult) counterpart. When the baby was left alone with the robot, in 13 out of 16 cases the baby followed the robot\'s gaze, leaving researchers to conclude that the baby believed it was sentient. We extend these insightful findings of animism to children’s cognitive, affective, and behavioural beliefs about a robot arm in the present study.
Responses to questions about the arm’s appearance and animistic qualities were coded for this study. Two raters were used to determine the reliability of the coding, with Cohen’s Kappa values ranging from 0.87 to 0.98, with a mean of 0.96 indicating very good inter-rater agreement. The majority of children identified the robot as male, and less than a quarter of the children identified it as female. One child stated the robot was neither, and about 10% did not know. The child’s sex was not related to their response. About a third of the children assigned human names to the robot such as ‘Charlie’. About a third gave names that refer to machines, such as ‘The Block Stacker’. A pet name was rarely assigned, such as ‘Spud’, or a combined human-machine type name ‘Mr. Robot’. When asked about their general impressions of the robot, a large majority gave a positive description, such as cool/awesome, good/neat, nice, likeable, interesting, smart, realistic, super, fascinating, and funny. Two children reported that the robot had a frightening appearance, and three children thought it looked like a dog. Another 17 did not provide a valid response.
Regarding its cognitive characteristics, more than half of the children stated the robot had recognition memory due to the ability to see their face, hair, and clothes; and that the robot was smart and had a brain (see Table 1). Other children provided a mechanical reason by stating it had a memory chip, camera, or sensors, or may have been programmed. Over a third of the children stated the robot could not remember them, for various reasons shown in the table. Children’s perceptions about the robot’s cognitive abilities in regards to knowledge are also shown in Table 1. About half of the children thought the robot did not have this capability, due to reasons such as not having a brain or interactions with them. Almost a third indicated that they believed the robot does know their feelings for various reasons such as from seeing the child and being programmed with this ability.
Regarding affective characteristics, the majority of children thought that the robot liked them, as shown in Table 2. A few children believed that the robot did not like them. Similarly, the majority of children reported that they thought the robot would feel left out if they played with a friend. Over a quarter of the children stated the robot would not feel left out, but provided explanations that would seemingly protect the robot from harm.
Robot can remember you | Robot knows your feelings | ||
Yes | 97 (52.7%) | Yes | 54 (29.3%) |
Can see me | 37 | Can see me | 18 |
Has memory chip, sensors | 15 | Has memory chip, sensors | 5 |
Smart, has brain | 3 | Smart, has brain | 3 |
If has a brain | 6 | Do not know why | 17 |
If short duration | 5 | Not coded | 11 |
If programmed | 1 | ||
Do not know why | 24 | ||
Not coded | 6 | ||
No | 68 (37.0%) | No | 103 (56.0%) |
No brain, eyes, or memory | 30 | No brain, eyes, or memory | 37 |
Too many people to remember | 14 | No interaction with me | 19 |
Robot does not like me | 3 | If not programmed | 8 |
If no brain | 3 | Do not know why | 31 |
If long duration | 2 | Not coded | 8 |
If not programmed | 2 | ||
Do not know why | 11 | ||
Not coded | 3 | ||
Do not know | 19 (10.3%) | Do not know | 27 (14.7%) |
Number and percentage of children reporting cognitive features of robot (N = 184).
Robot likes you | Robot feels left out | ||
Yes | 118 (64.0%) | Yes | 127 (69.0%) |
Looks/smiles at me, friendly | 38 | No one to play with | 62 |
I was nice/did something nice | 20 | Hurt feelings | 36 |
Did not hurt me | 13 | I would include robot | 9 |
It had positive intentions | 9 | Not fair | 2 |
Do not know why | 33 | Do not know why | 11 |
Not coded | 5 | Not coded | 7 |
No | 16 (8.7%) | No | 53 (28.8%) |
Ignored me/didn’t let me help | 10 | No thoughts/feelings | 29 |
No thoughts/feelings | 4 | Would include robot | 16 |
Do not know why | 2 | Does not understand | 3 |
Not coded | 0 | Do not know why | 5 |
Not coded | 0 | ||
Do not know | 50 (27.3%) | Do not know | 4 (2.2%) |
Number and percentage of children reporting affective features of robot (N = 184).
In regards to its behavioral characteristics (Table 3), more than a third of the children stated the robot was able to see the blocks, with just over half of the children indicating that the robot could not see the blocks. A higher endorsement of the robot’s ability to show action is evident in the table. That is, a large majority stated the robot could play with them, and even provided a variety of ideas for play. Examples include block building, and Lego®, catch with a ball, running games, and puzzles.
Robot sees blocks | Robot plays with you* | ||
Yes | 77 (41.8%) | Yes | 154 (83.7%) |
Has eyes | 32 | Construction | 103 |
Stacking | 20 | Ball game | 26 |
Sensors, camera | 13 | Running game | 12 |
Trained | 5 | Board game | 12 |
Other | 0 | Other | 17 |
Do not know why | 7 | Do not know why | 5 |
Not coded | 0 | Not coded | 5 |
No | 94 (51.1%) | No | 25 (13.6%) |
Eyes not real | 49 | Physical limitation | 11 |
Sensors, camera | 19 | Other | 4 |
Missed a block | 19 | Do not know why | 6 |
Guessed | 1 | Not coded | 4 |
Do not know why | 5 | ||
Not coded | 1 | ||
Do not know | 13 (7.1%) | Do not know | 5 (2.7%) |
Note* Many children provided more than one response.Number and percentage of children reporting behavioral features of robot (N = 184).
To further determine whether children considered the robot to be animate or inanimate, we analyzed the pronouns children used when talking about the robot arm. Almost a quarter of the children used the pronoun “it” in reference to the robot, another quarter stated “he”, and half used both.
In summary, children seemed to adopt many animistic beliefs about the robot. Half thought that it would remember them, and almost a third thought it knew how they were feeling. Affective characteristics were highly endorsed. More than half thought that the robot liked them and that it would feel rejected if not played with. In their behavioral descriptions, more than a third thought it could see the blocks, and more than half thought the robot could play with them. It is evident that children assigned many animistic abilities to the robot, but were more likely to ascribe affective than cognitive or behavioral ones. There was additional evidence of human qualities according to the names children gave it, their descriptions of it, and the pronouns they used to reference it in their responses. These animistic responses, moreover, were more apparent in younger than older children.
Although some responses suggest that children believed the robot held human characteristics because of programming and machine design, the majority of statements referred to human anatomy (e.g., eyes, facial features, and brain), emotions, and intentions. We explain these findings in two ways. First, the robot arm presented many social cues. That is, the eyes were at the same eye level as the children’s, giving the impression of ‘looking’ at child, and it returned to this position many times while scanning for the block. Children may have interpreted this movement as expression of interest and closeness, which is one of the reactions to frequent eye contact among people (Kleinke, 1986; Marsh, 1988). Second, children may have projected their own feelings, thoughts, and experiences onto the robot arm, which Turkle (1995) has reported may occur with robots. This was particularly evident in the surprising finding that so many children believed that the robot would feel rejected and lonely if not included in play, as well as that the arm could engage in forms of play that clearly it could not (e.g., running). Third, children may have lacked knowledge of terms and principles to explain the robot’s actions, thereby relying on terms that express human qualities such as ‘remembering’, ‘knowing’, and ‘liking’. Fourth, because the arm moved autonomously, children may have developed the impression that it has intentions and goals, as is a typical reaction to any independently moving object (Gelman, 1990; Gelman and Gottfried, 1996; Poulin-Dubois and Shultz, 1990).
Study 3: Children’s impressions of friendship towards a Robot Arm
Friendships are undoubtedly important for childhood development, and, as such, set the stage for the development of communication skills, emotional regulation, and emotional understanding (Salkind, 2008). In this study, and given the animistic responses obtained in the previous study, we set out to determine the extent to which children would hold a sense of positive affiliation, social support, shared activities, and communication towards a robot; all of which exemplify friendship. In addition, we questioned whether children would share a secret
Robot can cheer you up | Robot can be your friend | ||
Yes | 145 (78.8%) | Yes | 158 (85.9%) |
Perform action for me | 61 | Conditional | 31 |
Perform action with me | 12 | Being or doing things together | 30 |
Cheerful appearance | 20 | Helpful | 17 |
Connects with me | 20 | Knows me | 12 |
Help me | 7 | Kind | 11 |
Do not know why | 17 | Friendly | 6 |
Not coded | 8 | Likeable | 7 |
No | 27 (14.7%) | Friend to robot | 4 |
Limited abilities | 16 | Do not know why | 28 |
Does not like me | 1 | Not coded | 12 |
No | 19 (10.3%) | ||
Limited mobility | 3 | ||
Limited communication | 2 | ||
No familiarity | 3 | ||
No brain, feelings | 4 | ||
Do not know why | 8 | Do not know why | 4 |
Not coded | 2 | Not coded | 3 |
Do not know | 12 (6.5%) | Do not know | 7 (3.8%) |
Number and percentage of children reporting positive affiliation (N = 184).
with a robot, as this behavior may also signify friendship (Finkenauer, Engels & Meeus, 2002).
As shown in Table 4, more than three quarters of the children stated that the robot could improve their mood, with reasons varying from its actions to its appearance. Moreover, more than three quarters stated the robot could be their friend. Many reasons were given for this possibility. They included enjoying activities together, helping each other, kindness, likeability, and shared understanding.
According to Table 5, the majority of children stated they would talk to the robot and share secrets with it. Most children had difficulty explaining their reasons for their answers. Rather, they provided answers that described what they would talk about, such as what to play together. Interestingly, many children stated that they liked the robot and wanted to spend time becoming acquainted. This desire for a greater connection to the robot is also exemplified in their responses to sharing secrets. More than a third of the children stated they would tell the robot a secret. Some children (n = 24) stated that they thought it was wrong to tell secrets, suggesting that of those children who would generally tell secrets (n = 160), half of them (n = 84, 52.50%) would tell a robot. The most frequent reason given was because they believed that the robot would not share it – seemingly because the robot arm could not speak. Many of them also stated, however, that they considered the robot arm to be friendly.
Talk to robot | Tell robot secrets | ||
Yes | 124 (67.4%) | Yes | 84 (45.7%) |
I like the robot | 16 | Robot will keep secret | 30 |
To get to know each other | 6 | Friendship with robot | 13 |
Robot has mouth | 6 | Positive response to secret | 7 |
If robot could talk | 22 | Other | 4 |
Gave examples | 30 | ||
Do not know why | 37 | Do not know why | 22 |
Not coded | 7 | Not coded | 8 |
No* | 53 (28.8%) | No | 92 (50.0%) |
Robot cannot talk | 20 | Secrets are wrong | 24 |
Robot cannot hear | 6 | Robot has limitations | 18 |
Not human | 5 | Robot not trustworthy | 24 |
Looks unfriendly | 9 | Robot is not alive | 9 |
Do not know why | 11 | Do not know why | 12 |
Not coded | 4 | Not coded | 5 |
Do not know | 7 (3.8%) | Do not know | 8 (4.3%) |
Some children provided more than one reasonNumber and percentage of children reporting communication (N = 184).
The majority of children responded affirmatively to questions about affiliation, receiving support, communicating, and sharing secrets, which typically characterize friendship. Regarding affiliation, almost two thirds of the children thought the robot liked them, and many explained that it was because the robot appeared friendly. Children also attributed positive intentions to the robot, likely because it was moving independently and engaging in a child friendly task. More than three quarters of the children did believe that the robot could offer them support. The action of stacking blocks was often explained as a means of providing this support, perhaps to distract and entertain the child. A large majority of children stated that they would play with the robot in a variety of games. It is not surprising that many of them suggested building with blocks, considering that they had just observed this activity. Finally, about two thirds of the children stated they would talk to the robot and more than a third stated that they would share secrets. Again, these results suggest that children are willing to develop a bond with the robot.
Many children in our study stated that they would not engage in these friendship-behaviors with a robot and explained that the robot did not have the capabilities to do so. Reasons for these different perceptions of the robot have not been explored in the research but may plausibly include variation in children’s knowledge of the mechanics of robots. In addition, a considerable proportion of children did not or could not provide an answer to the questions about friendship. It is possible that children were unable to differentiate human from robot characteristics, lacked sufficient understanding about the mechanics of robots, or were generally confused about the robot’s abilities. Our use of terms in the interview, such as whether the robot would ‘feel left out’, describe human characteristics and may have mislead children into positively responding. Clearly, the results raise many questions for research, not the least of which is whether children actually do develop a friendship with a robot. Over time, and as a result of interactions with robots, children may develop a new system or schema of understanding, and subsequent vocabulary to articulate their sense of friendship with a robot, that is likely distinct from their friendships with children.
The fact that so many children ascribed life characteristics to the robot suggests that they have high expectations of them and are willing to invite them into their world. This presents a challenge to robot designers to match these expectations, if the purpose of the robot is to garner and maintain interest from children. Children may be primed for these interactions. In fact, children may become frustrated when a robot does not respond to their initiations and may actually persevere at eliciting a response (Weiss et al., 2009). Therefore, the robot may not need to be programmed to respond in an identical fashion to a specific initiation, as humans certainly do not, which may actually increase the child’s engagement with a robot. This principle is well known as variable ratio reinforcement according to behaviourism learning theory (Skinner, 1969). Of course, children may become discouraged if the robot’s response is erratic. Instead, we propose that a high, but not perfectly predictable response to the child’s behaviours will lead to the longest and most interesting interactions.
In addition, our studies suggest that children can develop a collaborative relationship with a robot when playing a game together. This gives some suggestion of the nature of the relationship children may enjoy with a robot: one that allows give and take (Xin & Sharlin, 2007). This may enhance a child’s sense of altruism and, hence, increase engagement with it. It is, thus, recommended that developers of such robots consider designing them to not only offer help, but be able to receive it.
The studies and tests reported in this chapter have certain limitations that one must consider when interpreting the results. The tests and associated observations made during the study can be reproduced by using a variety of robot arms and even include mobile manipulators from where more detailed children-robot interaction studies can be made. Although the bio-inspired control mechanism used in this study worked well, tests using such control approaches should be performed on other robot types including humanoids and mobile robots. Such control architecture should also be tested in physical children-robot interaction to determine its suitability towards enabling seamless active engagement between children (humans) and robots.
Robot arms have indeed changed from their original industrial and automotive applications in the 1960s. Our studies show that children are ready to accept them as social objects for sharing personal information, offering mutual support and assistance, and regarding them as human in various ways. In the near future, we expect that humans will not only frequently and directly interact with and rely on robot arms and robots of diverse types for daily activities, but perhaps treat them and regard them as possibly human. Our studies cannot begin to address the numerous complex questions about the nature of the interactions people will have with robots. We offer a glimpse, however, of children’s willingness to do so. Overall, the results are rather surprising given that the robot arm did not speak, performed only one task, and did not initiate physical interaction with the child. Are children merely responding to the robot arm as if it is a fancy puppet, and they are presenting their imagination in their responses? Perhaps, but regardless of the explanation, children in these studies demonstrated overwhelmingly their predisposition towards active engagement for bio-inspired motion control.
We give special thanks to the TELUS World of Science - Calgary for collaborating with us. This research would not have been possible without their support.
The domestic pig has been widely used as a large animal model in biomedical research, as it is similar to humans with respect to the size of body and internal organs, longevity, anatomy, physiology, and metabolic profile [1]. Modification of the porcine genome is also important for studying the mechanisms underlying genetic disorders, developing therapeutic drugs, and improving pig meat production yields [2, 3]. Over the past three decades, attempts have been made to modify the porcine genome using genetic engineering technology, starting after Gordon et al. [4] first reported DNA microinjection (MI)-based production of transgenic (Tg) mice. Hammer et al. [5] first reported the successful production of Tg piglets using the technique reported by Gordon et al. [4], but attaining this result was more difficult than for rodents, where pronuclei are clearly visible using an optical microscope. In the case of porcine zygotes, pronuclei are difficult to see due to the presence of high lipid content in the cytoplasm. Researchers must briefly centrifuge zygotes to visualize the pronuclei prior to MI [5], which is labor-intensive and requires skill. Moreover, MI-mediated transgene integration into host chromosomes occurs randomly, which often causes gene silencing [6]. However, for precise and efficient genetic modification in the porcine genome, homologous recombination (HR)-based gene targeting technology may be recommended, which was first developed by Smithies’ group in mice [7]. In this case, the use of germline-competent embryonic stem (ES) cells is a prerequisite. These ES cells are first transfected with a targeting vector and then recombinant ES clones showing successful targeting are obtained. This vector usually contains a gene of interest (GOI) to be integrated into the target locus, together with a selection marker gene such a neomycin resistance gene (neo), and DNA sequences of appropriate length, termed homology arms (HA), that correspond to the endogenous target gene, are placed at both ends of the DNA containing the GOI and a marker gene. Chimeric mice can be obtained through blastocyst injection with the targeted ES clones, and the resulting chimeric mice would contribute to produce heterozygous mice carrying mutated traits (GOI/selection marker gene) in the target locus [8]. Unfortunately, there are no germline-competent porcine ES cells, despite extensive efforts [9, 10, 11, 12, 13]. Thus, to date, production of gene-targeted pigs derived from recombinant porcine ES cells has not yet been successful.
In 1996, scientists at the Roslin Institute (Wilmut and colleagues) first succeeded in producing cloned sheep using somatic cell nuclear transfer (SCNT) technology. They used fetal fibroblasts [14] or adult mammary gland-derived fibroblasts [15] as SCNT donors. Notably, prior to these reports, an attempt to produce cloned embryos by the similar technique shown by the Roslin’s group has been made by Prather et al. [16], who performed transfer of nuclei from two, four and eight-cell embryos to the enucleated oocytes and successfully produced one piglets that had been derived from the four-cell embryo’s nucleus. This approach is called “blastomere transplantation,” which is basically different from the somatic cell-based SCNT. Generally, fibroblasts used as SCNT donors can proliferate actively in vitro, and therefore are considered to be ideal cells for transgenesis or gene targeting. If these engineered fibroblasts are used for SCNT, the resulting cloned embryos and piglets should have the engineered traits in their genome. This possibility was first proven by the scientists at the Roslin Institute [17, 18] who successfully produced genetically engineered (GE) cloned sheep through SCNT using GE fetal fibroblasts. Since then, SCNT using GE cells as SCNT donors has been a common approach for production of knockout (KO) or Tg piglets [19, 20]. However, as mentioned below, the efficiency of producing cloned GE piglets is extremely low and the preparation of GE donor cells is laborious and time-consuming [21]. During the past two decades, production of only a few GE (KO) piglets has been reported by traditional approaches [22, 23, 24, 25, 26, 27]. Moreover, almost all resulting KO piglets were heterozygous with respect to the KO allele, thus requiring additional tasks such as breeding (likely one or two generations), sequence targeting, or in vitro cell cloning to obtain homozygous KO animals [28], which is also laborious, expensive, and time-consuming.
However, this situation drastically changed when new gene-targeting technologies emerged for precisely manipulating mammalian genomes, called “second-generation genome editing.” These technologies require the design of site-specific engineered nucleases which can be zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs) or clustered regularly interspaced short palindromic repeat-associated protein 9 (CRISPR/Cas9) nucleases, all of which induce a double-stranded break (DSB) at a specific site in the genome. This DSB facilitates genetic modification such as nonhomologous end-joining (NHEJ) and homology directed repair (HDR) [29], as described below. Using these genome-editing systems, many GE piglets have been produced using SCNT of genome-edited cells, or direct microinjection of genome-editing components (including engineered endonucleases) into the cytoplasm of zygotes, as described below in more detail.
As mentioned above, site-specific engineered nucleases are used in these genome-editing techniques. ZFNs, TALENs, and CRISPR/Cas9 can all bind to DNA and induce DSB, which triggers endogenous DNA repair. If the template DNA is absent, the DSB is repaired via the NHEJ pathway where insertion or deletion of nucleotides (hereinafter called “indels”) can happen in the cleaved area. These indels often cause frameshift of the amino acid sequence, leading to the generation of abnormal proteins or formation of a premature stop codon leading to cessation of protein synthesis. If template DNA homologous to the target site is present, it is inserted into the cleaved area via a site-specific HR event which is called HDR. Generally, NHEJ occurs in cells independent of its cell cycle, but HDR occurs primarily in dividing cells [30].
The ZFN technique uses the ZF protein (which binds to the target DNA) and the endonuclease Fok I (which cleaves DNA) [31]. ZF protein has several protein motifs capable of recognizing specific sequences of three nucleotides and binding to them. Notably, Urnov et al. [32] first demonstrated that ZFN is effective to induce DNA editing at the endogenous target gene in mammalian cells. Its targeting efficiency was over 18% in the absence of drug selection, which is ~1000-fold higher than that achieved by traditional gene targeting.
The TALEN technique uses proteins, termed transcription activator-like effectors (TALEs), which contain 33–35 amino acid repeats that flank a central DNA binding region (amino acids 12 and 13), and Fok I nuclease, as in ZFN, thus the term TALE nucleases (TALENs) [33, 34, 35]. Notably, the design and engineering of TALENs is simpler than that of ZFN, and thus can be done faster [35, 36].
CRISPR/Cas9 employs a short (20 bp) RNA sequence called single-guide RNA (sgRNA) which can bind to the specific chromosomal DNA site together with the Cas9 endonuclease [37, 38, 39, 40]. Once bound, two independent nuclease domains in Cas9 each cleave one of the DNA strand’s three bases upstream of the protospacer adjacent motif (PAM), introducing DSB at the target site of the host chromosome, which is then repaired by NHEJ. This system is different from the other genome editing tools such as ZFNs and TALENs, and thus synthesis of sgRNA is a prerequisite for this system. This development dramatically reduced both the complexity and time required for the design and implementation of gene editing.
Table 1 lists instances of production of GE piglets with genome editing technology from 2011 to 2018. This section provides a brief explanation on the background of GE pig production.
Method | Genome editing tool (mode for gene modification) | Method for gene modification | Outcome | Target gene | References |
---|---|---|---|---|---|
SCNT | ZFN (indels) | Using adult porcine ear fibroblasts hemizygous for the eGFP transgene | Seven of nine embryos (Day 12) exhibited loss of fluorescence | eGFP transgene | [41] |
SCNT | ZFN (indels) | Using porcine fibroblasts transfected with ZFN plasmid | Of 10 live piglets delivered, two carried the predicted ZFN-induced mutation; lower expression of both PPAR-γ1 and PPAR-γ2 was observed in those clones | PPARγ | [42] |
SCNT | ZFN (indels) | Using porcine fetal fibroblasts transfected with ZFN plasmid | Of six fetuses, all completely lacked α-Gal epitopes | GGTA1 | [43] |
MI | TALEN (indels) | Cytoplasmic MI of TALEN mRNA toward IVF-derived zygotes | CI of TALEN mRNAs inducing gene KO in up to 75% of embryos; Of the 18 live-born clones, eight contained monoallelic mutations and 10 contained biallelic modifications of the LDLR gene | LDLR | [44] |
MI | ZFN, TALEN (indels) | Cytoplasmic MI of ZFN or TALEN mRNA toward in vivo fertilized zygotes | Of 39 piglets produced, eight carried TALEN-derived editing events (21%); of nine piglets produced, one carried an editing event at the ZFN target site (11%) | NF-kappaB subunit | [45] |
SCNT | TALENs (indels) | Using porcine fetal fibroblasts transfected with TALEN plasmid | Three piglets with biallelic mutations of the GGTA1 gene exhibited loss of α-Gal epitopes on the surface of cells | GGTA1 | [46] |
SCNT | TALENs (indels/KI) | Using porcine fibroblasts transfected with TALEN mRNA + ssODN | Of eight piglets born from DAZL-modified cells, three are still born; of the six piglets from APC-modified cells, only one alive | DAZL, APC | [47] |
SCNT | ZFNs (indels) | Using porcine fetal fibroblasts transfected with ZFN mRNA | The resulting IL2RG KO pigs completely lacked a thymus and were deficient in T and NK cells, similar to human X-linked SCID patients | IL2RG | [48] |
SCNT | ZFN (indels) | Using porcine adult liver-derived cells transfected with ZFN plasmid through the two-steps | Four viable and healthy cloned pigs obtained exhibited disruption of the GGTA1 and the CMAH loci | GGTA1, CMAH | [49] |
SCNT | ZFN (KI) | Using porcine fibroblast cells transfected with ZFN plasmid and donor DNA | Successfully produced healthy monoallelic/biallelic CMAH KO pigs | CMAH | [50] |
SCNT/MI | CRISPR/Cas9 (KI/indels) | Using porcine fetal fibroblasts transfected with sgRNA-Cas9 plasmids + donor DNA/cytoplasmic MI of Cas9 mRNA/sgRNA toward IVF-derived zygotes | Of the CD163 recipients, five delivered healthy piglets by cesarean section; 12 of the 13 piglets contained either a biallelic or homozygous deletion of CD1D | eGFP, CD163, CD1D | [51] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA/sgRNA toward in vivo fertilized zygotes | Ten of 16 resulting piglets had indels with an efficiency of 63% and were comprised by cells with monoallelic mutant; they can be a model for von Willebrand disease | vWF | [52] |
SCNT | CRISPR/Cas9 (indels) | Using porcine fetal fibroblasts transfected with Cas9/sgRNA expression plasmid | A total of three piglets were obtained; fibroblasts from all three animals were negative for class I SLA cell surface expression | class I MHC | [53] |
SCNT | TALENs (indels) | Using pig fetal fibroblasts transfected with TALEN plasmid | Of 27 live cloned piglets obtained, nine were targeted with biallelic mutations in RAG1, three were targeted with biallelic mutations in RAG2, and 10 were targeted with a monoallelic mutation in RAG2 | RAG1, RAG2 | [54] |
SCNT | ZFN (indels) | Using pig fetal fibroblasts transfected with ZFN plasmid | Three GGTA1 null piglets showing loss of α-Gal epitope expression were born | GGTA1 | [55] |
SCNT | CRISPR/Cas9 (indels) | Using pig liver-derived cells transfected with two or three plasmids expressing Cas9 and sgRNA targeting to GGTA1, CMAH, or putative iGb3S genes | Of 10 fetuses obtained, five had mutations in both the GGTA1 and CMAH genes | GGTA1, CMAH, putative iGb3S | [56] |
SCNT | ZFNs (indels) | Using pig fetal fibroblasts transfected with ZFN plasmid | The MSTN-mutant pigs grew normally, had increased muscle mass with decreased fat accumulation | MSTN | [57] |
SCNT | TALENs (indels) | Using pig liver-derived cells transfected with TALEN plasmid | Livers from ASGR1−/− pigs exhibit decreased human platelet uptake | ASGR1 | [58] |
SCNT | ZFN/TALEN (indels) | Using pig fetal fibroblasts transfected with TALEN or ZFN mRNA | One of the cloned pigs generated GalT/CMAH-double homozygous KO pigs | CMAH | [59] |
SCNT | CRISPR/Cas9 (KI) | Using pig fetal fibroblasts transfected with targeting donor vector and two expression vectors for sgRNA and Cas9 | Highly efficient KI (up to 54%) was achieved after drug selection; one cloned piglet obtained showed correct targeting | H11 | [60] |
SCNT | CRISPR/Cas9 | Using pig fetal fibroblasts transfected with sgRNA, Cas9 expression plasmids | Four cloned double KO piglets showing loss of expression for both PARK2 and PINK1 were produced | TYR, PARK2, PINK1 | [61] |
MI | CRISPR/Cas9 (KI) | Cytoplasmic MI of Cas9 mRNA + sgRNA + ssODN toward in vivo fertilized zygotes | Two live-born piglets obtained showed the white coat-color phenotype over its entire body | MITF | [62] |
MI | CRISPR/Cas9 (KI) | Cytoplasmic MI of Cas9 mRNA + sgRNA + circular vector toward in vivo fertilized zygotes | All 16 piglets born were healthy and carried the expected KI allele; the KI allele was successfully transmitted through germline | Alb | [63] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA and sgRNA toward in vivo fertilized zygotes | Bi-allelic modifications of pig Npc1l1 were achieved at the efficiency as high as 100% | Npc1l1 | [64] |
SCNT | CRISPR/Cas9 (indels) | Using pig fetal fibroblasts transfected with sgRNA-Cas9 encoding vector | Of eight marker-gene-free cloned pigs with biallelic mutations obtained, some showed phenotypes similar to DM | MSTN | [65] |
SCNT | CRISPR/Cas9 (indels) | Using liver-derived cells transfected with sgRNA-Cas9 encoding vectors | One triple knockout pig was obtained; Cells from this cloned pig exhibited reduced human IgM and IgG binding | GGTA1, CMAH, β4GalNT2 | [66] |
SCNT | CRISPR/Cas9 (indels) | Using pig fetal fibroblasts transfected with sgRNA-Cas9 encoding vector | Three cloned piglets with biallelic mutation produced showed no antibody-producing B cells | IgM JH | [67] |
EP | CRISPR/Cas9 (indels) | Using Cas9 protein and sgRNA (RNP) toward IVF-derived zygotes | The use of gene editing by electroporation of Cas9 protein (GEEP) resulted in highly efficient targeted gene disruption and efficient production of Myostatin mutant pigs | MSTN | [68] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA and sgRNA toward in vivo fertilized zygotes | Of two piglets obtained, one piglets exhibited DMD phenotype, as exemplified by degenerative and disordered skeletal and cardiac muscle | DMD | [69] |
SCNT | ZFN (indels) | Using pig fetal fibroblasts transfected with ZFN-encoding mRNA | The heterozygous FBN1 mutant pigs obtained exhibited abnormal phenotype, which resembles MFS found in humans | FBN1 | [70] |
SCNT | CRISPR/Cas9 (KI) | Using pig fetal fibroblasts transfected with Cas9-sgRNA expression vector + donor DNA containing Cre/loxP system | Two male live piglets with mono-allelic MSTN KO obtained exhibited enhanced myofiber quantity, but the myofiber size remained unaltered | MSTN | [71] |
SCNT | TALENs (indels) | Using pig fetal fibroblasts transfected with TALEN plasmids | In total, 12 live and two stillborn piglets were collected; all fetuses and piglets exhibited homozygous GGTA1-null mutation | GGTA1 | [72] |
SCNT | TALENs (indels) | Using porcine fetal fibroblasts co-transfected with TALEN and hDAF expression plasmids | Six live-born piglets and three stillborn piglets were obtained; the piglets showed eight base mono-allelic mutations of GGTA1 and hDAF expression | GGTA1 | [73] |
SCNT | CRISPR/Cas9 (indels) | Using fetal fibroblast cells transfected with sgRNA-Cas9 encoding vector | Four live RUNX3 KO piglets with monoallelic mutation showed the lack of RUNX3 protein in their internal organ system | RUNX3 | [74] |
SCNT | TALENs (indels) | Using dermal fibroblasts transfected with TALEN plasmids | MSTN KO piglets exhibited a double-muscled phenotype, possessing a higher body weight and longissimus muscle mass measuring 170% that of wild-type piglets, with double the number of muscle fibers | MSTN | [75] |
SCNT | TALENs/Cas9 (KI) | Using fetal fibroblast cells transfected with Cas9/sgRNA or TALEN vector + ssODN | Of seven cloned piglets, some expressed human insulin | INS | [76] |
SCNT | CRISPR/Cas9 (KI) | Using fetal fibroblast cells transfected with sgRNA-Cas9 encoding vector + ssODN | One cloned stillborn piglet harbored the orthologous p.C313Y mutation at the MSTN locus | APP, LRRK2, MSTN | [77] |
SCNT | TALENs (indels) | Using ear fibroblasts transfected with TALEN vectors | Thirty GGTA1 biallelic KO piglets were successfully delivered and grew normally. | GGTA1 | [78] |
MI | CRISPR/Cas9 (KI) | Cytoplasmic MI of Cas9 mRNA + sgRNA + ssODN toward in vivo fertilized zygotes | Of five piglets delivered alive, three exhibited pigmentary disorders with light-colored iris in eye, which was observed in patients harboring Sox10 mutations | Sox10 | [79] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of sgRNA-Cas9 encoding vector toward in vivo fertilized zygotes | Of six healthy fetuses recovered, four exhibited loss of α-Gal epitope expression, indicating a biallelic KO of GGTA1 | GGTA1 | [80] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA + three types of sgRNAs toward in vivo fertilized zygotes | Of two live-born piglets delivered, one piglet showed biallelic modification of all three genes, and another showed biallelic modification of the DJ-1 and PINK1 genes and monoallelic mutations of parkin gene | Parkin, DJ-1, PINK1 | [81] |
SCNT-MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of RNP toward SCNT embryos | Six fetuses recovered revealed that all fetuses carried biallelic edits for the GRB10 gene (6/6, 100%) | GRB10 | [82] |
SCNT | CRISPR/Cas9 (indels) | Using kidney fibroblasts transfected with ZFN vectors | Two healthy normal females with GGTA1, CMAH double KO phenotypes are currently being raised for mating | GGTA1, CMAH | [83] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA + sgRNA toward IVF-derived zygotes | Seventeen live piglets and two stillborn were produced; all had mutations in both genes (no pigs with wild-type sequence) | RAG2, IL2RG | [84] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA + sgRNA toward in vivo fertilized zygotes | Eighteen piglets recovered showed either mono- or bi-allelic modifications and no wild-type animals; NANOS2 KO pigs phenocopied KO mice with male specific germline ablation | NANOS2 | [85] |
SCNT | CRISPR/Cas9 (indels) | Using fetal fibroblasts transfected with sgRNA and Cas9 expression vectors | Six biallelic KO pigs with mutations in ApoE and LDLR genes were obtained successfully in a single step. | ApoE, LDLR | [86] |
SCNT | TALENs | Using fetal fibroblasts transfected with TALEN plasmid | All six live piglets obtained carried biallelic mutations in the P53 locus | P53 | [87] |
SCNT | TALEN (indels) | Using fetal fibroblasts transfected with TALEN plasmid | A total of 18 live piglets were obtained; they showed hypermuscular characteristics | MSTN | [88] |
SCNT | CRISPR/Cas9 (indels) | Using fetal fibroblasts transfected with sgRNA-Cas9 encoding vectors | A total of 37 PERV-inactivated piglets were generated; 15 piglets remain alive | PERV | [89] |
SCNT | CRISPR/Cas9 (indels) | Using fetal fibroblasts transfected with sgRNA-Cas9 encoding vectors | Of 26 female piglets delivered, 23 piglets carried mutations in the MSTN locus; the bi-allelic KO pigs were viable and exhibited partial double-muscled phenotype | MSTN | [90] |
SCNT | CRISPR/Cas9 (KI) | Using fetal fibroblasts transfected with Cas9-gRNA plasmid and targeting vector | Twelve male piglets were born and expressed UCP1 in a tissue-specific manner | UCP1 | [91] |
MI | CRISPR/Cas9 (KI) | Cytoplasmic MI of RNP toward in vivo fertilized zygotes | A total of 18 fetuses/born piglets were obtained; successful insertion of pseudo attP sites within the COL1A locus was observed | COL1A | [92] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA + sgRNAs toward in vivo-derived zygotes | Indels in 92–100% of the embryos analyzed; all resulting 12 piglets had biallelic edits of TMRPSS2 | TMPRSS2 | [93] |
MI | CRISPR/Cas9 (indels) | Direct pronuclear microinjection of Cas9-gRNA plasmid | Of seven born piglets, one exhibited biallelic KO phenotype and one did monoallelic KO one | GGTA1 | [94] |
MI | CRISPR/Cas9 (indels) | Cytoplasmic MI of Cas9 mRNA and dual sgRNAs toward in vivo fertilized zygotes | Of nine fetuses examined, three exhibited bi-allelic mutations at the PDX1 locus; in those fetuses pancreatic primordium was highly disorganized | PDX1 | [95] |
SCNT (Handmade cloning) | CRISPR/Cas9 (indels) | Using fetal fibroblasts transfected with sgRNA and Cas9 expression vectors | Eleven live bi-allelic GGTA1/CMAH double KO piglets were obtained with the identical phenotype | GGTA1, CMAH | [96] |
EP | CRISPR/Cas9 | Using Cas9 protein and sgRNA (RNP) toward IVF-derived zygotes | Of 11 piglets born, nine survived; six of nine carried mutations in TP53; three of the genome-edited pigs (50%) exhibited various tumor phenotypes | TP53 | [97] |
SCNT | CRISPR/Cas9 (KI) | Using fetal fibroblasts transfected with sgRNA- Cas9 plasmid + donor DNA | Three piglets born grew and developed normally; all these piglets had fat-1 KI at the Rosa26 locus | Rosa26 | [98] |
SCNT | CRISPR/Cas9 (KI) | Using fetal fibroblasts transfected with sgRNA- Cas9 plasmid + donor DNA | Of seven naturally delivered piglets, six showed successful KI; the KI allele was successfully transmitted through germline | HTT | [99] |
SCNT | CRISPR/Cas9 (indels) | Using fetal fibroblasts transfected with sgRNA- Cas9 plasmid | Of a total of 17 piglets obtained, 12 appeared healthy; all had mutations at the target locus | FBXO40 | [100] |
Summary of production of genome-edited pigs.
Abbreviations: APC, adenomatous polyposis coli; Alb, albumin; GGTA1, α-1,3-galactosyltransferase gene; APP, amyloid precursor protein; ApoE, apolipoprotein E; ASGR1, asialoglycoprotein receptor; β4GalNT2, β1,4-N-acetylgalactosaminyl transferase; class I MHC, class I major histocompatibility complex; CRISPR/Cas9, clustered regulated interspaced short palindromic repeat and CRISPR-associated Cas; COL1A, collagen type I alpha 1 chain; CMAH, cytidine monophosphate-N-acetylneuraminic acid hydroxylase; DAZL, deleted in azoospermia-like; DM, double muscling; DMD, Duchenne muscular dystrophy; PARKIN, E3 ubiquitin ligase PARK2; EP, electroporation; eGFP, enhanced green fluorescent protein; FBXO40, F-box protein 40; FBN1, fibrillin-1; GM, gene-modified; GRB10, growth-hormone receptor binding protein-10; H11, Hipp11; hDAF, human decay-accelerating factor; HTT, huntingtin; indels, insertion or deletion of nucleotides; INS, insulin; IL2RG, interleukin-2 receptor gamma; IVF, in vitro fertilized; iGb3S, isogloboside 3 synthase; IgM JH, JH region of the pig IgM heavy chain; KI, knock-in; KO, knockout; LRRK2, leucine-rich repeat kinase 2; LDLR, low density lipoprotein receptor; MFS, Marfan syndrome; MI, microinjection; SCNT-MI, microinjection following somatic cell nuclear transfer; MITF, Microphthalmia-associated transcription factor; MSTN, myostatin; NANOS2, nanos C2HC-type zinc finger 2; fat-1, n-3 fatty acid desaturase; Npc1l1, Niemann-Pick C1-Like 1; PDX-1, pancreas duodenum homeobox 1; DJ-1, PARK7; PARK2, parkin; PPARγ, peroxisome proliferator-activated receptor-gamma; PERV, porcine endogenous retrovirus; PINK1, PTEN-induced putative kinase 1; RNP, ribonucleoprotein; RUNX3, Runt-related transcription factor 3; sgRNA, single guide RNA; ssODN, single-stranded DNA oligonucleotides; SCNT, somatic cell nuclear transfer; Sox10, SRY (sex determining region Y)-box 10; TALENs, transcription activator-like effector nucleases; TMPRSS2, transmembrane protease, serine S1, member 2; TYR, tyrosinase; UCP1, uncoupling protein 1; vWF, von Willebrand factor; ZFNs, zinc finger nucleases.
In 2011, three types of GE piglets were produced using ZFNs from different laboratories. All of these piglets were produced by SCNT using GE cells as a SCNT donor. The first report showing successful production of GE piglets involved the disruption of enhanced green fluorescent protein (EGFP) gene in a hemizygous manner. Whyte et al. [41] demonstrated that a ZFN pair efficiently inactivated the expression of EGFP that was integrated into the chromosomes of porcine fibroblasts via the NHEJ pathway with an efficiency of ~5%. From this experiment, it was found that the endogenous NHEJ pathway is effective for inducing mutation in a porcine target gene. Furthermore, SCNT using GE cells with the mutated allele as a SCNT donor demonstrated that seven of the nine resulting cloned fetuses (at Day 12) stopped expressing EGFP. Yang et al. [42] co-transfected porcine fibroblasts with ZFN and pcDNA3.1 plasmids (providing neo) by electroporation (EP), performed SCNT using these GE cells, and finally obtained peroxisome proliferator-activated receptor γ (PPARγ) mono-allelic KO pigs, which are expected to generate a porcine cardiomyopathy model. Hauschild et al. [43] attempted to destroy GGTA1, an endogenous gene encoding an enzyme required for production of a xenogeneic antigen called α-Gal epitope, using ZFNs in porcine fetal cells. They found that α-Gal epitope-negative cells with the bi-allelic KO phenotype can be efficiently isolated by FACS, and the resultant GE cells have the potential to make cloned piglets. Importantly, this experiment suggests that it is possible to produce individuals with a bi-allelic KO phenotype with this technology. In other words, bi-allelic KO piglets can be directly created without breeding or subcloning, which contrasts with past instances where only heterozygous KO piglets have been produced through traditional gene targeting. Hauschild et al. [43] also showed that neither off-target cleavage nor integration of the ZFN-coding plasmid occurred.
The successful production of genome-edited piglets with bi-allelic KO genotype obtained after cytoplasmic MI of in vivo-derived porcine zygotes using either ZFN or TALEN mRNA was first reported by Lillico et al. [45]. This MI-based production of GE animals was also shown to be successful in mice [101, 102, 103, 104].
Hai et al. [52] first demonstrated that GE pigs can be produced using the CRISPR/Cas9 system. They performed cytoplasmic MI with Cas9 mRNA and sgRNA targeted to von Willebrand factor gene (vWF) to produce a pig model for type 1 von Willebrand disease. In this study, 10 of 16 resulting piglets had indels with an efficiency of 63%, and most pigs contained more than two different alleles, suggesting mono-allelic mutants.
Successful knock-in (KI) of a GOI into the target locus was first reported in pigs by Ruan et al. [60] and Peng et al. [63]. Zhou et al. [61] demonstrated the production of SCNT-treated piglets with mutations in multiple genes after a single transfection.
Fischer et al. [83] first succeeded in producing GE pigs by cytoplasmic MI of a Cas9 protein/gRNA complex called a ribonucleoprotein (RNP). Furthermore, GE pigs could be efficiently produced by in vitro EP in the presence of RNP [68]. Sheets et al. [82] produced GE fetuses by cytoplasmic MI of RNP into oocytes reconstituted with intact cells using the SCNT technology. By this treatment, they reported highly efficient (100%) generation of bi-allelic modification in the resultant cloned fetuses. The significance of this approach is be that researchers can obtain GE pigs with a defined genetic background, even though the starting oocytes are derived from ovaries obtained from a slaughterhouse.
For the production of GE pigs, the choice of delivery method for genome editing components in porcine zygotes is important. As shown in Table 1, the methods for the production of GE pigs achieved by delivering genome editing reagents at earlier stages of development can be largely divided into four groups: the first is MI of genome editing reagents (in a form of DNA, mRNA or protein) into zygotes (Figure 1A); the second is SCNT using GE cells as the SCNT donor (Figure 1B); the third is in vitro EP of zygotes in the presence of genome editing reagents (Figure 1C); the fourth is MI of genome editing reagents into SCNT-treated embryos reconstituted using a normal cell (Figure 1D). Furthermore, we will provide a new approach based on in vitro EP of the SCNT-treated embryos reconstituted using a normal cell as the fifth group of methods for possible production of GE pigs (Figure 2A). In the following sections, each of these methods are described.
Several methods to create genome-edited pigs. (A) Microinjection (MI)-based method using zygotes. (B) Somatic cell nuclear transfer (SCNT)-based method using gene-engineered (GE) cells as a SCNT donor. (C) Electroporation (EP)-based method using zygotes. (D) MI-based method using the SCNT-treated embryos.
A new method for production of genome-edited pigs. (A) EP-based method using the SCNT-treated embryos, which is termed “GENTEP.” (B) Experimental outline for checking the validity of GENTEP.
MI is an important tool in the creation of GE piglets. To date, about 30% (17/60) of studies (Table 1) have employed this approach. For example, in the case of MI with CRISPR/Cas9-related mRNA, a single cytoplasmic MI of 2–10 pL containing 125 ng/μL Cas9 mRNA and 12.5 ng/μL sgRNA was adopted [62]. Yu et al. [69] employed Cas9 mRNA (20 ng/μL) and sgRNA (10 ng/μL) mixtures for cytoplasmic MI.
Is MI of these components deleterious to the development of porcine zygotes? According to Hai et al. [52], the in vitro developmental efficiencies of embryos injected with Cas9 mRNA/sgRNA (~79%) and embryos injected with water (~77%) were both very high and comparable with each other, suggesting that the MI and the Cas9 mRNA/sgRNA had little effect on early embryonic development. On the contrary, Whitworth et al. [51] reported that a higher concentration of sgRNA induces toxicity in porcine embryos. According to Whitworth et al. [51], 10 ng/μL of sgRNA and Cas9 mRNA are recommended.
Selecting appropriate zygotes also appears to be an important factor in the production of GE pigs. For acquisition of viable zygotes, there are at least two methods. One is isolation of zygotes from oviducts of a female that has been inseminated, hereinafter called “in vivo-derived zygotes,” and the other is acquisition of zygotes by in vitro fertilization (IVF) between in vitro matured oocytes (derived from the ovaries obtained from the slaughterhouse) and sperm. Generally, it is believed that the in vivo-derived zygotes exhibit superior development performance comparing to the IVF-derived zygotes [51, 62]. Indeed, the number of laboratories using IVF-derived zygotes for production of GE pigs is low (~30% (5/17); see Table 1). However, acquisition of viable in vivo-derived zygotes is laborious and often associated with the sacrifice of pregnant females, which appears to be one of the major goals for improvement in genetic engineering technology using pigs.
The frequent generation of individuals with mosaic genotypes is also a serious problem associated with MI-based GE pig production. Sato et al. [105] demonstrated that cytoplasmic MI of parthenogenetically activated porcine embryos (hereinafter called “parthenotes”) with Cas9 mRNA + sgRNA caused frequent mosaicism in the offspring (blastocysts) with cells with mixed genotype, so-called normal wild-type cells and mutated cells, when they were subjected to cytoplasmic MI immediately after oocyte activation. Notably, Carlson et al. [44] suggested that 100% of bovine embryos exhibited fluorescence expression after cytoplasmic MI of EGFP mRNA, but only ~40% of porcine embryos did. It is probable that an endogenous system for translation to protein from mRNA may not be sufficiently established in those porcine embryos, especially at the stage immediately after fertilization or zygotic activation. Indeed, Sato et al. [106] demonstrated that this mosaicism can be partially improved when cytoplasmic MI is performed with oocytes 12 h after activation. In contrast, other researchers reported that only 10–20% of MI-treated embryos exhibited mosaicism [51, 84]. Notably, Whitworth et al. [51] performed cytoplasmic MI with Cas9 mRNA + sgRNA toward fertilized oocytes at 14 h post-fertilization. In this context, the use of sgRNA and an RNP instead of Cas9 mRNA may be the key to solving this issue of mosaicism, as the Cas9 protein is more rapidly translated, folded, and complexed with sgRNAs prior to editing, unlike the Cas9 mRNA [107, 108, 109]. For example, in mice, delivering RNPs into zygotes causes rapid genome editing in the target locus, which also maximizes efficiency while minimizing mosaicism [110, 111, 112]. Indeed, Sheets et al. [82] demonstrated that after MI with RNP, 100% of piglets produced had the bi-allelic KO genotype.
Interestingly, Petersen et al. [80] demonstrated that cytoplasmic MI of DNA vectors coding for CRISPR/Cas9 targeting the porcine GGTA1 gene enabled biallelic knockout of GGTA1 in 7/12 fetuses and piglets (58.3%). As mentioned previously, it is difficult to visualize porcine pronuclei at zygote stage under normal conditions due to high lipid content in the cytoplasm. Researchers therefore must centrifuge them briefly prior to MI. The fact that cytoplasmic MI of DNA vectors can induce genome editing at a target locus may be beneficial for researchers, because preparation of plasmid DNA is easier than that of mRNA, and it is generally more resistant against degradation than mRNA. According to Petersen et al. [80], it currently remains unknown how the circular DNA plasmid translocates from the cytoplasm to the nucleus. They speculate that the SV40 nuclear translocation signal of the CRISPR/Cas9 plasmid could play an important role by facilitating nuclear translocation via association with ubiquitous transcription factors.
SCNT using GE cells as an SCNT donor is another way to produce GE pigs. The merit of this approach is the use of in vitro cultivated cells such as fetal fibroblasts to which various genetic engineering techniques (i.e., introduction of multiple KO, KI, and transgenes) can be applied easily. After gene transfer, these cells are subjected to cell selection through drug selection or fluorescence activated cell sorting (FACS) to enrich GE cells as a pure population. Thus, it is highly probable that the resulting SCNT-derived GE founder pigs have a predictable genotype and low rates of mosaicism. Unfortunately, as mentioned previously, the efficiency of SCNT to produce cloned piglets is still very low. Much effort has been focused on improving the low efficiency associated with the SCNT, which includes improvement of the oocyte/zygote culture system and application of chemical reagents to alter the epigenetic status of transferred nuclei. For improving the culture method, researchers have used vitamin C [113], α-tocopherol [114], melatonin [115] or alanyl-glutamine dipeptide (instead of glutamine) [116]. For altering the epigenetic status, researchers have used histone deacetylase inhibitors (HDACi) such as trichostatin A (TSA) [117, 118], valproic acid (VPA) [119, 120, 121], scriptaid [122, 123, 124], LBH589 (panobinostat) [125], oxamflatin [126], PXD101 (belinostat) [127], quisinostat [128], MGCD0103 [129], or histone methyltransferase inhibitors such as MM-102 [130]. Lin et al. [131] employed tauroursodeoxycholic acid (TUDCA), an inhibitor of endoplasmic reticulum (ER) stress, and demonstrated that TUDCA can enhance the developmental potential of porcine SCNT embryos by attenuating ER stress and reducing apoptosis. Wang et al. [132] demonstrated that administration of siRNA or microRNA-148a, both of which can suppress the function of DNA methyltransferase 1 (DNMT1) at a transcriptional level, is effective for enhancing the developmental potential of SCNT embryos. Furthermore, Matoba et al. [133] succeeded in drastically increasing SCNT efficiency by cytoplasmic MI of mRNA coding for histone demethylase (Kdm4d) in mice.
EP is known to be a useful and powerful gene delivery tool enabling transfer of exogenous substances (i.e., DNA) into a cell and was first applied to rat zygotes for genome editing by Kaneko et al. [134]. Since then, many researchers have successfully induced gene edits by using this technology in mice [135, 136], bovines [137] and pigs [68]. The merit of this technology is that it is simple, rapid and convenient for genome editing in zygotes, compared to the previous MI-based technique. Notably, about 30–50 zygotes can be edited with one pulse of EP. Furthermore, EP only requires a square pulse generator called an electroporator, and not a more expensive micromanipulator system.
As mentioned previously, Tanihara et al. [68] first applied EP to porcine IVF-derived zygotes and produced genome-edited pigs. They used CRISPR/Cas9-based RNP for knock-in of a target gene, and achieved reduced mosaicism and higher efficiency of genome-edited pig production with EP (30 V, square pulse 1.0 ms in duration repeated five times) using an electrode (#LF501PT1-20; BEX Co. Ltd., Tokyo, Japan) connected to a CUY21EDIT II electroporator (BEX Co. Ltd.). Notably, they reported no appreciable reduction in the developmental ability of the EP-treated embryos.
Although direct modification of zygotic genomes provides some advantages, SCNT also provides a significant advantage by permitting the isolation of cells containing precise modifications before the expense of animal production is incurred. As mentioned previously, Sheets et al. [82] successfully produced genome-edited cloned pigs by combining SCNT with CRISPR/Cas9 MI, which is beneficial for researchers as they do not need to manage a founder herd, and can eliminate the need for laborious in vitro culture and screening. In this study, all (6/6) of the resultant clone fetuses exhibited 100% bi-allelic modification. Unfortunately, they failed to describe successful production of live birth piglets, but it seems that this approach is a powerful tool for GE pig production.
Similar to the approach shown by Sheets et al. [82], we tried to obtain cloned GE piglets through in vitro EP in the SCNT-treated embryos, which is called Genome Editing via Nuclear Transfer and subsequent Electroporation or GENTEP (Figure 2A). Some results obtained from GENTEP-related experiments are presented below.
SCNT-derived embryos were obtained by inserting fetal fibroblasts derived from microminiature pigs (MMP) [138] into the perivitelline space between enucleated porcine oocytes (derived from ovaries obtained from a slaughterhouse) and zona pellucida, according to the method described by Miyoshi et al. [119] (Figure 2B). The resulting SCNT-derived embryos were then subjected to electric activation following electric fusion between an egg and a cell (Figure 2B). Six or 12 h after activation, the SCNT-treated embryos were subjected to in vitro EP in the presence of RNP targeted to the pig low density lipoprotein receptor (LDLR) gene (Figure 2B). Parthenotes (~6 h after electric activation) were also used for in vitro EP using tetramethylrhodamine-labeled dextran 3 kDa (used as an indicator for successful gene delivery) (as shown in Figure 3A) or as controls for immunocytochemistry using anti-LDLR antibody (as shown in Figure 3B).
Validity check of GENTEP. (A) EP of parthenotes (6 h after activation) in the presence of tetramethylrhodamine-labelled dextran 3 kDa. Porcine parthenotes were subjected to in vitroEP, and then cultured for 7 days up to blastocysts. Note that almost all of the EP-treated blastocysts are fluorescent (arrows in a and b), while intact parthenotes do not fluoresce (c and d). Bar = 100 μm. (B) Staining with anti-LDLR antibody. The intact parthenote (hatched blastocysts) exhibits the reactivity to the antibody (arrowheads in a–c), but not to the second antibody alone (arrowheads in d–f). No reactivity to the antibody was also seen in the GENTEP-derived blastocyst (arrowheads in g–i). Nuclear staining with DAPI was performed after staining with the second antibody. Note that porcine zona pellucida was slightly stained with the second antibody, since it was found to be reactive with the second antibody alone (see d–f). Bar = 100 μm.
First, we examined whether the in vitro EP we used here is effective for successful gene delivery to porcine embryos and does not cause any deleterious effects on their embryonic development, using porcine parthenotes (6 h after activation). The EP procedure was based on the method described by Hashimoto and Takemoto [135]. An electroporation chamber (#LF610P4-4_470; BEX Co. Ltd.), in which two platinum block electrodes were situated with a 1-mm gap between them (Figure 1C), was placed under a stereoscopic microscope and connected to an electric pulse generator (CUY21EDITII Genome Editor™, BEX Co. Ltd.). About 20 parthenotes were placed into a 5-μL drop containing 2 μg/μL tetramethylrhodamine-dextran 3 kDa (#D3307; Thermo Fisher Scientific Inc., Waltham, MA, USA) in Opti-MEM (Invitrogen, Carlsbad, CA, USA) between the electrodes (Figure 1C). EP was performed under these conditions: 30 V, square pulses, 1.0 ms in duration at 99 ms intervals, repeated seven times. The EP-treated parthenotes were then cultured for 7 days up to blastocysts to evaluate the in vitro developmental rate and uptake of fluorescent dye into the embryos. Approximately 40% of the EP-treated parthenotes developed to the blastocyst stage and ~80% of them exhibited bright tetramethylrhodamine-derived fluorescence [arrows in Figure 3A(a,b)]. This result suggests that our EP condition is useful for effective delivery of a foreign substance into porcine embryos and not harmful for their development.
Second, we performed CRISPR/Cas9-based genome editing (targeted to the endogenous LDLR gene) with porcine SCNT-treated embryos. We designed sgRNA capable of recognizing a 20 bp sequence spanning the translation initiation codon (ATG) upstream of the protospacer adjacent motif (PAM) sequence (CGG) on the first exon of porcine LDLR (left panel of Figure 4A). The sgRNA was synthesized by Integrated DNA Technologies, Inc. (IDT; Coralville, Iowa, USA) as Alt-R™ CRISPR crRNA product. The crRNA and tracrRNA (purchased from IDT) were combined for annealing and then mixed with recombinant Cas9 protein (TaKaRa Shuzo Co. Ltd., Shiga, Japan) to form RNP, according to the method of Ohtsuka et al. [139]. The final concentrations of the components in RNP were 30 μM/mL (for crRNA/tracrRNA) and 1 mg/mL (for Cas9 protein). The SCNT-treated embryos 6 or 12 h after activation were transferred to a 5-μL drop (containing RNP in Opti-MEM) and immediately subjected to in vitro EP under these conditions: 30 V, square pulses, 1.0 ms in duration at 99 ms intervals (or 0.5 ms in duration at 99.5 ms intervals), both repeated seven times. After EP, the embryos were promptly cultivated in normal medium for 7 days up to blastocysts and then subjected to analysis of molecular biology (possible mutations in the first exon of LDLR) and immunocytochemistry (possible loss of LDLR protein synthesis) parameters, as described in Figure 2B. In each group, 8–17% of the EP-treated embryos developed to blastocysts (Table 2). These rates appear to be comparable to the yield (24.2%) in experiments performed using intact MMP fetal fibroblasts as SCNT donors [119]. All of the blastocysts obtained were then fixed with 4% paraformaldehyde, and a section of these embryos was subjected to immunocytochemical staining using anti-LDLR antibody (Figure 2B). The EP-treated cloned blastocyst (termed GENTEP-1) was unreactive to anti-LDLR (arrows in Figure 3B(g–i)). In contrast, a parthenote (blastocyst) exhibited positive reactivity to anti-LDLR (arrow in Figure 3B(a–c)). Staining with the second antibody alone failed to react with the antibody (arrow in Figure 3B(d–f)). Furthermore, each of these fixed blastocysts was subjected to genomic DNA isolation to examine possible mutations at the individual embryo level (Figure 2B). Next, GenomiPhi-based whole genome amplification (WGA) was performed using the isolated genomic DNA as a template, as described previously [140]. PCR was then performed using the WGA products as a PCR template. The primer sets used are LDLR-S (5′-AAACCTCACATTGAAATGCTG-3′)/LDLR-RV (5′-CCTAAACTCTCGCGCCCCCCT-3′) for the first round of PCR and LDLR-2S (5′-CTGCAAATGACTGGGGCCCCG-3′)/LDLR-2RV (5′-CTCCAACCACGTAAGAATGAC-3′) for nested PCR (left panel of Figure 4A). Nested PCR using the LDLR-2S/LDLR-2RV primer set yields 355-bp products (left panel of Figure 4A). The typical example when the nested PCR products (lanes 1–9) are loaded onto a 2% agarose gel is shown in the right panel of Figure 4A. Almost all of the samples tested exhibited 355-bp products, except for lane 3 showing bands of reduced size, suggesting occurrence of a large deletion (probably over 100 bp) around the LDLR sequence recognized by sgRNA. In Figure 4B, an example of the results obtained from direct sequencing of nested PCR products using LDLR-2S primer is shown. The sample GENTEP-1, which has been shown to exhibit loss of the reactivity to anti-LDLR (see Figure 3B-g-i), had a large deletion including a sequence spanning ATG and PAM. Notably, there was no appreciable overlapping in ideograms of the sample, suggesting a homozygous bi-allelic KO phenotype (Figure 4C). Subcloning of the PCR products derived from the sample GENTEP-1 into TA cloning vector and subsequent sequencing demonstrated that all six clones obtained exhibited the same sequence as the parental product (data not shown). When the remaining PCR products were sequenced it was found that almost all (82%, 9/11) of the samples exhibited the homozygous bi-allelic KO genotype (Figure 4C).
Molecular biological analysis of the GENTEP-treated embryos (termed GENTEP-1 to -11) at a single embryo level. (A) Structure of porcine LDLR gene and a target sequence recognized by sgRNA (left panel), and the results of nested PCR (right panel). The target sequence (shown in blue) spanning ATG (shown in red) is located on the first exon of LDLR. PAM, protospacer adjacent motif. Primers used for first PCR and nested PCR are shown above the LDRL. In the right panel, a part of the nested PCR products (lanes 1–9) loaded onto 2% agarose gel is shown. Arrow indicates the PCR products of 355 bp in size. M, 100-bp ladder markers. (B) Ideogram pattern in the GENTEP-1 and -2 samples obtained after direct sequencing of the nested PCR products using LDLR-2S primer. (C) Various indels found in each GENTEP-treated embryo. ATG is shown in red. Sequence recognized by sgRNA is shown in blue.
Stage at EP after activation of SCNT-treated embryos | EP condition2 | Total number of SCNT-treated embryos examined | No. of embryos cleaved to the two-cell stage (%) | No. of embryos developed to blastocysts (%) |
---|---|---|---|---|
6 h | 0.5 | 12 | 8 (66.7) | 1 (8.3) |
1.0 | 12 | 11 (91.7) | 2 (16.7) | |
12 h | 0.5 | 33 | 21 (63.6) | 5 (15.2) |
1.0 | 35 | 23 (65.7) | 3 (8.6) |
Summary of the properties of blastocysts derived from EP1 toward the SCNT-treated porcine embryos.
EP in the presence of RNP [10 μM of crRNA/tracrRNA mixture (targeted to LDLR gene) + 0.3 μg/μL of Cas9 protein] is performed on SCNT-treated embryos 6 or 12 h after activation. The EP-treated embryos were then cultured for 7 days to the blastocyst stage for the presence of mutations in the target gene at molecular biological and immunocytochemical levels.
EP was performed under the electric condition of 30 V in voltage, 0.5 ms in length of square pulse with 99.5-ms intervals (0.5) or 1.0 ms in length of square pulse with 99-ms intervals (1.0), and seven times of pulse stimulation using an electroporation chamber (#LF610P4-4_470; BEX Co. Ltd.) connected to an electric pulse generator (CUY21EDITII. Genome Editor™, BEX Co. Ltd.).
As shown above, genome editing tools such as ZFNs, TALENs and CRISPR/Cas9 are considered useful in enabling site-specific gene modification in livestock such as pigs. However, there are still several techniques and factors that influence performance which must be addressed. These include the single embryo assay, off-target cutting, multiplexed genome engineering, KI, and Cas9 pigs. In this section, these techniques or factors are described in greater detail.
To increasing the efficiency of genome editing systems, it is important to select suitable sets of ZFNs (or TALENs) or sgRNA (in the case of CRISPR/Cas9). Researchers therefore must check the efficiency of these reagents by introducing them into cultured cells, but at this point it remains unknown whether they will function in vivo. Unlike small animals such as mice and rats, large animals have long gestation periods and it is costly to prepare large animal recipients. Therefore, this testing in vivo appears to be difficult in larger animals such as pigs. To overcome this issue, a single embryo (blastocyst) assay to evaluate the operability of the genome editing reagents prepared was provided by Wang et al. [62] who later re-validated those sets using porcine parthenotes. To our knowledge, this assay was first developed using mice by Sakurai et al. [141] who reported that it is useful for confirming the fidelity of sgRNAs used.
It may be required to confirm at a molecular level whether the genome-edited embryos have mutations. In this case, WGA has often been employed for amplifying the whole genome of an embryo (blastocyst) using genomic DNA isolated from a single embryo as the DNA template [140, 141], since the blastocyst DNA is often too small to generate a sufficient amount of PCR product. The effectiveness of WGA-based amplification of blastocyst DNA has already been confirmed by ours [142] and others [44]. The resulting products obtained after PCR using WGA-derived DNA as the template are then subjected to direct sequencing for identification of possible mutations in the target gene, as shown in Figure 4B.
Since sgRNA used in the CRISPR/Cas9 system can recognize only a short sequence (20 bp) at the target gene where Cas9 cleaves, other genes with a similar sequence to the sgRNA may be susceptible to Cas9-mediated DNA cleavage, which leads to the occasional generation of off-target cutting [29, 143]. This unintended cutting is considered a serious problem to be resolved.
Several strategies to minimize off-target cutting have been employed including the use of the double nickase mutant form of Cas9, which induces a single-strand break instead of DSB [144]; the use of RNP, whose half-life is shorter than the duration of transcription of plasmid or viral nucleic acids [110, 145]; or the fusion of catalytically inactive Cas9 with Fok I nuclease domain (fCas9) to improve the DNA cleavage specificity [146]. Recently, it was reported that Cpf1, a putative Class 2 CRISPR effector, mediates target DNA editing differently from Cas9 [147]. It generates a 5-nucleotide staggered cut with a 5′ overhang, which is particularly advantageous in facilitating an NHEJ-based KI into a genome. Several unique enzymes that can decrease the probability of off-target cleavage have also been produced. For example, two engineered enzymes produced from SpCas9 from Streptococcus pyogenes with the goal of enhancing specificity, called eSpCas9 [148] and SpCas9-HF [149], are reported to reduce the probability of mismatched DNA binding. A hybrid enzyme combining the Cas9-nickase and PmCDA1, an activation-induced cytidine deaminase (AID) ortholog, could perform targeted nucleotide substitution [150]. Furthermore, a CRISPR system using a new Cas-related enzyme called Cas13a that targets RNA has also been recently developed [151].
Notably, in the case of GE pigs and embryos, there have been no reports of off-target mutagenesis as shown by the following papers: [43, 50, 61, 62, 63, 64, 69, 74, 77, 78, 80, 86, 100, 105]. This suggests a very low probability of off target-cleavage in GE pigs.
The CRISPR/Cas9 system can confer multigene KO in one shot of gene delivery [152, 153]. This property is especially beneficial for the purpose of creating disease model animals, as certain types of diseases are known to be caused by multigene defects. Interestingly, Sakurai et al. [154] demonstrated that at least nine endogenous genes can be knocked out simultaneously through a single shot of cytoplasmic MI of 12 sgRNAs together with Cas9 mRNA into murine zygotes. In pigs, Zhou et al. [61] demonstrated successful generation of PARK2 (parkin) and PTEN-induced putative kinase 1 (PINK1) double-KO pigs through SCNT with GE fetal fibroblasts after co-transfection of Cas9, PARK2-sgRNA, and PINK1-sgRNA-expressing vectors by electroporation. The percentage of PARK2−/−/PINK1−/− double-KO cells was up to 38.1%. SCNT using these double-KO cells resulted in the birth of 20 cloned piglets. Of these, four piglets developed normally, and both parkin and PINK1 in those individuals were depleted at the protein level. Estrada et al. [66] also succeeded in obtaining one triple-KO cloned piglet with mutations in GGTA1 (coding for α-1,3-galactosyltransferase), CMAH (coding for cytidine monophosphate-N-acetylneuraminic acid hydroxylase) and β4GalNT2 (coding for β1,4-N-acetylgalactosaminyl transferase) after SCNT. Wang et al. [81] generated PARK7 (DJ-1)/parkin/PINK1 triple-gene modified pigs using the CRISPR/Cas9 system in one step through direct zygote injection of Cas9 mRNA and three types of sgRNAs. According to Wang et al. [81], of two live-born piglets delivered, one piglet showed biallelic modification of all three genes, and another showed biallelic modification of the DJ-1 and PINK1 genes and monoallelic mutation of the parkin gene.
As shown in Table 1, in 2015 successful KI in pigs was reported by several groups. For example, Wang et al. [62] performed MI with in vivo fertilized zygotes (derived from colored pigs) using Cas9 mRNA + sgRNA + single-stranded DNA oligonucleotides (ssODN), targeting microphthalmia-associated transcription factor (MITF), a master regulator gene of melanocyte development, and obtained two live-born piglets showing the white coat color phenotype over its entire body. Peng et al. [63] tried to create KI piglets with a MI approach using a circular vector as donor DNA. They designed an sgRNA targeting the starting codon region (including the adjacent 5′ and ATG) and generated a targeted fragment (donor for HR) with the insert flanked by 1-kb HA on both sides. They performed cytoplasmic MI of Cas9 mRNA + in vitro synthesized sgRNA + circular vector containing the targeting fragment, and finally obtained 16 live piglets, all of which were found to carry the expected KI allele. Notably, they confirmed expression of human albumin (Alb) protein generated from the KI allele in the plasma of these cloned pigs. This means that expression of a transgene (human Alb as GOI) is possible under the control of an endogenous promoter system (in this case, Alb promoter).
Ruan et al. [60] demonstrated production of GE pigs with successful KI of GOI into the target Hipp11 (H11) locus, which is considered as “safe harbor” genomic locus that allows gene expression without disrupting internal gene function, like the Rosa26locus. They utilized a positive and negative selection method to insert GFP into the pH 11 locus in pig fetal fibroblast cells by electroporation. The targeting donor vector (4.2 kb in size) contains a reporter cassette with neo and GFP genes which are flanked by a 0.8-kb HA to the H11 locus on each side with the diphtheria toxin A (DTA) gene at the 3′ end. Cells were transfected with the linearized donor vector and two expression vectors for sgRNA (targeted to the H11 locus) and Cas9. After drug selection, they obtained GE cells with successful KI at the H11 locus with efficiencies up to 54%. Next, they performed SCNT using these correctly targeted clones, and obtained one cloned piglet which was later confirmed to show correct targeting.
Generally, it is believed that HDR-mediated KI is more difficult than NHEJ-based indels. For example, in proliferating human cells, NHEJ has been reported to repair 75% of DSBs, while HDR repaired the remaining 25% [155]. To enhance the HDR efficiency, several approaches are now being attempted. For examples, co-injection of murine zygotes with a mixture containing Cas9 mRNA, sgRNA, template ssODNs and Scr7 (an inhibitor for DNA ligase IV) significantly improved the efficiency of HDR-mediated insertional mutagenesis [156]. Chu et al. [157] also demonstrated usefulness of Scr7 for abolishing NHEJ activity and increasing HDR in both human and mouse cell lines. However, the function of Scr7 in promoting HDR remains controversial. Some researchers demonstrated that Scr7 failed to increase HDR rates in rabbit embryos [158] and porcine fetal fibroblasts [159]. On the contrary, Li et al. [160] demonstrated that Scr7 promoted HDR efficiency in porcine fetal fibroblasts. The same group also showed that other reagents L755507 (β-3 adrenergic receptor agonist) and resveratrol (small-molecule compound found in grapes) also showed similar effects (promotion of HDR efficiency) in porcine cells.
As mentioned previously, the current generation of gene-edited pigs has mostly been produced through either MI or SCNT approaches, which are both expensive and time-consuming. In mice, several Tg lines carrying a Cas9-expressing cassette have been created [154, 161, 162]. These Tg mice are thought to be useful animals for direct in vivo genome editing experiments, because successful delivery of the expression vectors of sgRNAs alone or RNA itself into selected tissues caused generation of genome-edited tissues. For example, Platt et al. [161] demonstrated that in vivo viral administration of Kirsten rat sarcoma viral oncogene homolog (Kras), transformation related protein 53 (Trp53), and serine/threonine-protein kinase 11 (Stk11)-gRNAs to the Cas9-expressing line caused lung carcinomas within a short period. This suggests that if a Cas9-expressing pig is produced, it will provide an easy and efficient way to produce genetic modifications, which should substantially facilitate studying gene functions, modeling human diseases, and promoting agricultural productivity. Based on this concept, Wang et al. [163] first produced Cre-dependent Cas9-expressing pigs to enable efficient in vivo genome editing. They first transfected the linear-targeting donor containing Cre-dependent Cas9-expression cassette and TALEN plasmids directed to Rosa26 locus into porcine fetal fibroblasts and finally selected clones carrying KI cassette. These clones were then used for SCNT to produce cloned GE piglets. They showed that cells isolated from several organs of GE pigs exhibited Cre-induced activation of Cas9 expression. This Cas9 pig line will be used for various studies as indicated above.
Because pigs are similar to humans in physiological, anatomical, and genetic aspects, they are now seen as a leading animal model for biomedical research. Recent advances in genome editing technology have led to accelerated production of GE pigs within a relatively short time period, which is beneficial due to cost savings in propagation of GE animals and maintaining animals for breeding. Production of GE pigs can be largely categorized into two approaches, so-called MI/EP-mediated production of GE zygotes and SCNT using GE cells as the SCNT donor. There are advantages and drawbacks for both these approaches. For example, the former is simpler, more convenient, and cost-effective than the latter. However, the available genetic background is limited. In this context, the latter is beneficial for the flexibility of choosing any type of genetic background, because the genetic background of SCNT-derived cloned pigs is determined by that of donor cells used for SCNT. Unfortunately, the efficiency of SCNT is extremely low at present. MI/EP with SCNT-treated embryos may compensate for these disadvantages associated with MI/EP or SCNT-mediated production of GE piglets, if the efficiency of SCNT is greatly improved in future.
We thank Shogo Matsunaga for their support in the GENTEP-related experiment, shown in Figures 3 and 4. This study was partly supported by a grant (no. 19K06372 for Masahiro Sato; nos. 25450475 and 16K08085 for Kazuchika Miyoshi; no. 18K09839 for Emi Inada; no. 17H04412 for Issei Saitoh; no. 16H05176 for Akihide Tanimoto) from the Ministry of Education, Science, Sports, and Culture, Japan.
The founding sponsors had no role in the design of the study, collection, analyses, or interpretation of data, writing of the manuscript, and decision to publish the results.
Masahiro Sato designed the study and drafted the manuscript; Kazuchika Miyoshi and Hiroaki Kawaguchi involved in the GENTEP-related experiment; Emi Inada and Issei Saitoh critically revised the manuscript; Akihide Tanimoto supervised the manuscript.
If your research is financed through any of the below-mentioned funders, please consult their Open Access policies or grant ‘terms and conditions’ to explore ways to cover your publication costs (also accessible by clicking on the link in their title).
\n\nIMPORTANT: You must be a member or grantee of the listed funders in order to apply for their Open Access publication funds. Do not attempt to contact the funders if this is not the case.
",metaTitle:"List of Funders by Country",metaDescription:"If your research is financed through any of the below-mentioned funders, please consult their Open Access policies or grant ‘terms and conditions’ to explore ways to cover your publication costs (also accessible by clicking on the link in their title).",metaKeywords:null,canonicalURL:"/page/open-access-funding-funders-list",contentRaw:'[{"type":"htmlEditorComponent","content":"Book Chapters and Monographs
\\n\\nMonographs Only
\\n\\nBook Chapters and Monographs
\\n\\n\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\nMonographs Only
\\n\\nLITHUANIA
\\n\\nBook Chapters and Monographs
\\n\\n\\n\\nBook Chapters and Monographs
\\n\\n\\n\\nBook Chapters and Monographs
\\n\\n\\n\\nSWITZERLAND
\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\\n\\n\\n\\nBook Chapters and Monographs
\\n\\nBook Chapters and Monographs
\n\nMonographs Only
\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\nBook Chapters and Monographs
\n\nBook Chapters and Monographs
\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\nBook Chapters and Monographs
\n\n\n\nMonographs Only
\n\n\n\nLITHUANIA
\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\n\n\nSWITZERLAND
\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\n\n\nBook Chapters and Monographs
\n\n