IROS2019国际学术会议论文集 0604

Abstract Introducing alterations in the mtDNA sequence is challenging but needed for potential therapies and basic studies Direct microinjection of mitochondria into small cells has been considered inefficient and impractical To address this issue we present a highly efficient and precise robotic approach for automatically transferring mitochondria from one single cell to another A microfluidic cell positioning device is used to pattern two different types of cells in one dimensional array and an image processing algorithm is applied to identify the location of the mitochondria and cell A visual feedback control mechanism is developed to enhance the mitochondrial extraction efficiency A robust adaptive sliding control algorithm is developed to precisely control an X Y stage to accomplish the extraction of mitochondria from A type cell followed by injection of the mitochondria into B type cell automatically The system can transfer mitochondria from one cell to another with an average duration of 15 s mitochondria Experiments of mitochondrial transfer from THP1 and NB4 cells to THP1 cells and fibroblasts are conducted to show the effectiveness of the developed approach Index Terms Mitochondrial transfer robotic surgery mitochondria single cell injection I INTRODUCTION Mitochondria are important organelles of biological cells They are generally known as the powerhouse of cells and play a key role in essential biological functions Mitochondria regulate apoptosis intracellular signalling and Ca2 homeostasis and generate reactive oxygen species during cell respiration 1 2 Mitochondrial dysfunction often plays an essential role in tumour progression and many genetic diseases 3 For example mitochondrial dysfunction has been associated with ageing and neurodegenerative disorders 4 5 as well as epilepsy 6 and type 2 diabetes 7 8 Currently mitochondrial diseases are treated by This work was supported by a grant from the Research Grants Council of Hong Kong Special Administrative Region China Reference No 11209917 and T42 409 18 R Adnan Shakoor Fei Pan and Jiayu Sun are with the Department of Biomedical Engineering City University of Hong Kong Hong Kong email ashakoor2 c my cityu edu hk fei pan my cityu edu hk jiayusun4 c my cityu edu hk Mingyang Xie is with the College of Automation Engineering Nanjing University of Aeronautics Fax 852 3443 0172 Email medsun cityu edu hk medical approaches that are only palliative Therefore the development of approaches could replace or transfer healthy mitochondria can be potential solution for those who are suffering from mitochondrial disease triggered by mutations in their DNA The expression and level of the diseases occur by Mutations in the mtDNA during life depends on the amount of mitochondria that are spoiled in contrast with functional copies in the cell If the quantity of damaged mitochondria crosses a certain threshold the disease will manifest 9 10 Here in these cases the transfer of healthy mitochondria to damaged tissues or cells can help for the treatment of these diseases Few mitochondria transfer approaches have been developed to repair cells with dysfunction or diseased mitochondria by transfer of functional and healthy mitochondria 11 Mitochondrial transfer between cells and the transfer of isolated mitochondria in co culture or in vivo have been reported in 12 15 However it s still needed to be investigated that uptake of isolated mitochondria during co culture with cells or the mitochondrial transfer by tunnelling nanotube are cell type specific or general mechanisms Yang developed a method of mitochondrial transfer into cells by firstly injecting the mitochondria into rodent oocyte then withdrawing the part of the oocyte containing the injected mitochondria and fusing it with mammalian cells 16 However in this technique other rodent egg impurities were transferred along with the mitochondria Microinjection enables the mechanical delivery of foreign material into cells Several robotic cell injection and micromanipulation methods 17 22 have been developed in literature Endogenous mitochondria were replaced by direct microinjection of the isolated mitochondria 23 24 Recently the metabolite profile in mammalian cells was restored with mitochondrial transfer by photothermal nanoblade 25 However the efficiency of direct mitochondrial injection and photothermal nanoblade was only 0 3 and 2 respectively Manual control and frequent clogging of backfilled isolated mitochondria into the small micropipette tips are core issues that contribute to the low efficiencies of mitochondrial injection in mammalian cells Unlike cell microinjection mitochondrial injection into mammalian cells is more challenging mainly because of the size of the micropipette required to be used for mammalian cells Micropipette tips normally used for microinjection are less than 1 m which are not appropriate for the injection of mitochondria with sizes of 1 3 m Using a micropipette larger than 2 m can significantly damage mammalian cells Thus existing direct mitochondrial transfer systems are inefficient manually A Robotic Surgery Approach to Mitochondrial Transfer Amongst Single Cells Adnan Shakoor Mingyang Xie Fei Pan Wendi Gao Jiayu Sun and Dong Sun 2019 IEEE RSJ International Conference on Intelligent Robots and Systems IROS Macau China November 4 8 2019 978 1 7281 4003 2 19 31 00 2019 IEEE659 controlled and incapable of distinguishing between healthy and non healthy mitochondria prior to injection Furthermore mitochondrial transfer methods use isolated mitochondria for transfer and obtaining 100 pure functional isolated mitochondria is difficult Isolated mitochondria are possibly a combination of functional mitochondria dead mitochondria and cell debris These methods disregard the real time functionality status of the mitochondria during mitochondrial transfer Therefore dead mitochondria and cell debris which are potentially harmful to the transferred cells can be transferred during the transfer process Moreover isolated mitochondria are obtained from thousands of cells simultaneously hence the above methods ignore the heterogeneity of mitochondria within cells which is another level of mitochondrial complexity 26 Approaches that involve whole tissues can only offer the average results of numerous actions that happens in various cells Studies of single cell can reveal that genetic variations triggered by tumorigenesis related signalling reveal clues to the variation of a normal cell to a tumour cell 27 Hence biological analysis at the single cell and organelle level provides basic information and has recently gained worldwide attraction We recently developed a robotic organelle biopsy system to extract mitochondria and nuclei from small single cells 28 29 The robotic system was able to extract functional mitochondria from single cells In the present study we further enhanced our robotic system to transfer the healthy and functional mitochondria from A type cells to B type cells automatically Microinjection and micropipette parameters were optimised to inject the mitochondria into mammalian cells with high efficiency The core contributions of this study are as follows Firstly this system can perform complete automated single cell mitochondrial transfer tasks including mitochondrial extraction from one cell followed by transfer to another cell with high efficiency Bypassing the mitochondria isolation method for the microinjection of mitochondria our system extracts fresh and healthy mitochondria from a single cell followed by injection into small single cells without compromising the mitochondria and cell viability Secondly the efficiency of mitochondrial extraction presented in 29 is further improved by enabling the system to detect and extract mitochondria on the basis of their size and location and by integrating the visual feedback control scheme for mitochondrial extraction Image processing algorithms are developed to detect mitochondria and injection cells Unlike previous limitations of extracting one or two mitochondria the current system can extract single and multiple mitochondria as per mitochondrial transfer requirement Moreover a robust adaptive sliding controller is developed that deals with unmolded uncertainties and external disturbances during the procedure of stage movement Thirdly automatic transfer of mitochondria can applied to a wide range of cells Experiments are conducted to transfer mitochondria from NB4 and THP1 cells to NB4 cells and fibroblasts HDF Mitochondrial injection parameters and the impact of micropipette tip size on injection cells are further analysed The rest of this paper is organised as follows Section II describes the mitochondrial transfer system setup Section III introduces the detail of the mitochondrial transfer process and motion control Section IV provides the details of the image processing Section V presents the experimental results and effectiveness of the robotic mitochondrial transfer system Finally the conclusion of this study is given in Section VI II SYSTEM DESIGN A System setup The robotic mitochondrial transfer system comprises of modules namely sensing module executive module and control module The sensing module contains a microscope Nikon TE2000 a CCD camera FOculus FO124SC and a 60 objective The executive module comprises a 3 DOF micromanipulator uMp Sensapex and an X Y Z stage Prior Scientific ProScan A micropipette B 100 78 10 Sutter Instrument is connected to a programmable microinjector PM6000 MicroData Instrument A microfluidic cell holding device is used for patterning cells The control module comprises of two motion controllers one for controlling the X Y Z positioning stage and the other for 3 DOF manipulator Fig 1 A robotic surgery system for the mitochondria transfer B Microfluidic chip design A microfluidic chip was designed to pattern two different types of cells in one 1D array in an alternative manner The detail of the design and fabrication process can be found in 29 The store presented in a previous design is replaced with microfluidic channels to pattern extraction and injection cells in an alternative manner C Glass micropipette Glass capillaries O D 1 mm and I D 0 78 mm were first rinsed with a siliconising reagent Sigmacote Sigma Then sigmacote was blown out of the capillaries by applying air pressure from one end of the capillaries The capillaries were subsequently dried in an oven overnight at 80 C To obtain a gradual taper with the final tip size of approximately 0 7 m the capillaries were pulled with a micropipette puller P 2000 Sutter Instrument Then the micropipette tip was 660 bevelled at 15 with a microelectrode beveller BV 10 Sutter Instrument with the final opening of 1 2 m D Mitochondria transfer process Fig 2 illustrates the schematics of automatic mitochondrial transfer from A type cell to B type cell After patterning both types of cells in a microfluidic chip see Fig 2 a the micropipette and cell are brought at same position Then the operator defines the number of mitochondria to be extracted The remaining mitochondrial transfer process is conducted automatically in the following manner Firstly mitochondria are detected by image processing then the stage drives along the y axis to bring into line the pipette with one of the selected mitochondria see Fig 2 b After the alignment the stage travels along the x axis to bring the mitochondria and pipette tip at the same position see Fig 2 c The injector then applies negative pressure to aspirate the mitochondria inside the pipette see Fig 2 d followed by the positioning of the stage to align the micropipette with the next mitochondria see Fig 2 e This process repeats until the operator defined number of mitochondria is aspirated see Fig 2 f h After extracting mitochondria the stage moves back to extract the micropipette see Fig 2 i and then the system spots the centre of the B type cell The stage align the pipette with the centre of the cell by moving along the y axis see Fig 2 j and then moves towards the pipette until the pipette reaches inside the cell at a distance equal to 1 3 of the cell radius see Fig 2 k The injector applies positive pressure to inject the mitochondria into the cell see Fig 2 l The stage subsequently moves back towards the x axis to move the micropipette out of the cell see Fig 2 m and bring the next cell within the field of view see Fig 2 n The system then detects the mitochondria of this cell and the processes continue until all cells present in the microfluidic chip are processed On average the system takes 15s per mitochondrial transfer from one cell to another Fig 2 Schematics of the automatic mitochondrial transfer from an A type cell to a B type cell a i Mitochondrial extraction from an A type cell j n Mitochondrial injection process into a B type cell E Motion control Lagrange s motion equation can be used to express the dynamic equation of a 3 DOF motorised stage as follows MqBqG q 1 Where 3 1 G is the vector of the gravity force 3 3 M represents the inertial matrix 3 3 B represents the damping friction coefficient matrix 3 1 denotes the input torque and 3 1 q is the position coordinate of motorised stage Both B and M are positive definite and diagonal The aforementioned dynamic equation can be linearly parameterised as MqBqG qY q q q 2 where 3 p Y q q q represents the regressor matrix and 1p denotes the vector of the model parameters Considering the unmodelling uncertainty and external disturbances e g mechanical vibration the dynamic model 1 can be generalised as follows d MqBqG 3 where d denotes the external disturbance and unmodeled uncertainty A robust control strategy is proposed for the trajectory tracking of the motorised stage in the presence of non parameter uncertainties and external disturbances Firstly a sliding variable is defined as follows rdeee Sqqqqqqq 4 where 3 3 is the positive definite diagonal matrix and ed qqq is the position tracking error and The dynamic model of the motorised stage in Eq 3 can be expressed in terms of the sliding variable S as drrrr MSBSMqBqG qY q q q q 5 Then to achieve trajectory tracking a robust sliding controller is designed as follows eqs 6 where eq Y 7 s KS s 8 T Y S 9 In the above formulas K and are both positive definite matrices and 0 The first term in 6 deals with system parameter uncertainties whereas the second term in 6 compensates for the unmodelled uncertainties and external disturbances Theorem 1 The proposed robust controller 6 together with the updated law 9 converges the motorised stage tracking error e q to zero under the following two assumptions Assumption 1 0d kS where 0 0k Assumption 2 min0 Kk where min denotes the minimum eigenvalue of matrix K Substituting 6 together with 7 9 into 3 produces the closed loop dynamics as follows KS MSBSY S 10 661 where is the estimation error of the system parameter Analysis of the stability of closed loop dynamics 10 can be performed with a Lyapunov function candidate defined as follows 11 22 TT VS MS 11 The developed robust adaptive sliding controller deals with unmodeled uncertainties and external disturbances during the procedure of stage movement III IMAGE PROCESSING A Mitochondria detection Fig 3 Mitochondrial detection depending on their sizes and locations a d Mitochondrial thresholding and contour detection g h Mitochondrial detection depending on their sizes and locations In our previous work 29 the mitochondrial region was detected on the basis of the HSV values of the mitochondria under fluoresce microscopy After detecting the region a random region was chosen as the final position of the mitochondria In this study we further enhanced our mitochondrial detection method by enabling the system to detect mitochondria depending on their relative location and sizes relative to the micropipette tip Extraction of mitochondria that are near the cell membrane causes less damage to the cell by preventing deep penetration of the micropipette Selecting mitochondria depending on their size relative to the dimensions of the micropipette tip can greatly enhance the mitochondrial extraction rate For example mitochondria of 3 m may not be extracted into a micropipette with a tip of 1 m size Therefore choosing mitochondria depending on their sizes and locations is an important aspect of mitochondrial extraction and transfer For detection the functional isolated mitochondria were stained with membrane potential dye Jc 1 Fig 3 a The threshold image was created from the original image containing the mitochondria and the location of each mitochondrion was detected by edge detection The algorithm can also detect the mitochondria depending on their sizes For example in Fig 3 g and Fig 3 h mitochondria larger than 2 m and 0 5 1 m were detected respectively B Injected cell detection For detection under fluorescence microscopy the image of the cell trapped in a channel was captured and converted into a greyscale image Gaussian filter was used to smooth the image or reduce any noise present in the image Canny edge detection was applied to detect the boundaries exist in the image To remove small circles and enhance the performance of Hough circle transform image dilation was applied as shown in Fig 4 d Finally the Hough circle transform was applied to detect circular objects present in the image as shown in Fig 4 e By applying the above methodology the Hough transform can more accurately detect cells compared with directly applying the Hough circle transform on the original image Fig 4 Cell detection for mitochondrial transfer a Cell under a bright field b Cell under fluorescence microscopy c Threshold image d Contour detection e and f Final position selection C Visual feedback control for mitochondria extraction Fig 5 Visual feedback control mechanism for mitochondrial extraction a Cell stained with jc 1 mitochondria membrane potential before mitochondrial extraction b Threshold image whi