Methods and compositions for microspore imaging, analysis, culturing, and / or sorting

Abstract

Provided herein is a method of imaging and analyzing plant microspores. The method comprises obtaining one or more two-dimensional micrographs of a microspore population, wherein the microspore population has not been labeled; sending the one or more two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented; and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry. Also provided herein is a method of imaging and analyzing plant microspores comprising determining microspore viability.

Description

METHODS AND COMPOSITIONS FOR MICROSPORE IMAGING, ANALYSIS, CULTURING, AND / OR SORTINGBACKGROUND

[0001] Microspores can be used in plant variety breeding programs. For example, microspores can be used to generate doubled haploids. Microspores are also relatively abundant and easily accessible, compared to their female counterparts. However, there remains a need to increase the ability of researchers and breeders to image, analyze, categorize, and / or sort microspores.SUMMARY

[0002] The present disclosure provides methods for rapidly imaging and analyzing microspore populations in order to determine, for example, viability, nucleus morphology, and / or cargo loading status. The methods disclosed herein do not require any type of labeling (e.g., no tag, dye, or pigment is used) and can be conducted using relatively simple white light microscopy. These methods allow micropore populations to be rapidly analyzed and then used for downstream work, without the addition of any molecular label.

[0003] Provided herein is a method of imaging and analyzing plant microspores. The method comprises obtaining one or more two-dimensional micrographs of a microspore population; sending the two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; obtaining one or more z- stacks (e.g., a sequence of 2D images taken at progressive optical focal planes to render a volumetric image) of the population of segmented microspores or the population of microspores to be segmented; and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and geometry. The microspore population is not labeled.BRIEF DESCRIPTION OF DRAWINGS

[0004] The present disclosure will be more fully appreciated by reference to the following drawings.

[0005] FIG. 1 shows a brightfield micrograph image of a dense population of microspores and the result of automatic segmentation of the maize microspores, which can be the first step in both viability analysis and nuclear morphology analysis. Each cell is shown in a unique hue(here, using a psuedo-RGB value). Alternatively, each cell can be identified by a unique value (e g the unique pseudo-(RGB) value or other unique value) for processing purposes.

[0006] FIG. 2 shows the results of automatic labeling and classification of segmented maize microspores as viable or non-viable based on area and pixel intensity, as described herein.

[0007] FIG. 3 shows the results of nuclei edge detection for maize microspores, using multiple images from a z-stack. The dotted pixels in the interior of the microspore (or which alternatively may be colored pixels) represent projected nuclei edges detected in any one or more images in the z-stack, and so provide a composite image of the entire z-stack illustrating where any nuclei edge was detected. Illuminated (or optionally, colored) pixels may be one pixel per point of nuclei edge detection. Alternatively, a larger area such as a 2x2 pixel grid per point of nuclei edge detection may be used for increased visual representation. The microspore shown marked with an asterisk is an example of a microspore with no illuminated pixels, because it was identified as non-viable based on area and pixel intensity, and therefore was excluded from nuclei detection.

[0008] FIG. 4A (unlabeled) and Fig. 4B (labeled) show the results of nuclei edge filling of the nuclei detected in FIG. 3, performed on segmented and viable cells. Fig. 4A and 4B are inverted views of the same image. Solid-filled clusters in Fig. 4B show filled nuclei edges projected to the microspore equatorial plane. The asterisk graphic marks a single microspore as an example, which indicates the same microspore shown in both FIG. 4 and FIG. 5. The asterisk marked microspore has three distinct illuminated edge filled segmented regions indicating candidate nuclei. The arrow indicates a smaller segmented region of interest that was ultimately not identified as a nucleus (a false positive) that was removed after intensity line signal processing. As a result, that smaller segmented region is no longer shown as part of the asterisk marked microspore in FIG. 5. The other two larger regions in the asterisk marked microspore were correctly identified as nuclei in FIG. 4 and so remain shown as true nuclei in FIG. 5.

[0009] FIG. 5 shows the microspore marked with an asterisk in Fig. 4 with the false nuclei signal removed by automated grouping and thresholding of local maxima, peak width, and peak prominence across all image planes to detect and verify true nuclei signals (as shown in the top panels of Fig. 7). The true nuclei pixels confirmed by this signal processing can be compared with the filled nuclei edges identified in FIG. 4 in order to allow for the removal of filled edges that are not identified as true nuclei after signal processing. A filled edge is considered as a segmented true nucleus if it contains at least one pixel identified as a nucleus from the signal processing step (as shown in the top panels of Fig. 7). The filled edge removed from theasterisk marked microspore in Fig. 4 did not pass the signal processing step and therefore was removed from the image.

[0010] FIG. 6 shows how virtual lines are drawn on each z-stack image in order to identify and distinguish local pixel maxima that can correspond to any of nuclei edges, cell edges, or germination pore edges. The filled circles of increasing and decreasing sizes represent the cell as a whole viewed in each focal plane of the z-stack. The dotted lines are the virtual lines.

[0011] FIG. 7A, 7B, and 7C show data from line scans across segmented microspores. The values are either subjected to a Gaussian fit (smoother curves) or not (rougher curves). FIG. 7A shows an example across a nucleus. FIG 7B and 7C show 4 examples of line scans that are not across a nucleus. Each of FIG. 7A, 7B, and 7C show examples with and without a Gaussian fit applied.DETAILED DESCRIPTION

[0012] Definitions

[0013] The disclosure is not limited to particular examples, which can, of course, vary. The terminology and exemplary examples used herein are for the purpose of describing aspects of the disclosure only and are not intended to be limiting. As used herein, terms in the singular and the singular forms “a”, “an” and “the”, for example, include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to “plant”, “the plant” or “a plant” also includes a plurality of plants; also, depending on the context, use of the term “plant” can also include genetically similar or identical progeny of that plant.

[0014] Unless defined otherwise, numeric ranges recited within the specification are inclusive of the numbers defining the range and include each integer within the defined range. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure pertains.

[0015] As used herein, “about” or “approximately” means plus or minus 10% of the recited value. For example, “approximately 10 meters” or “about 10 meters” means any value from 9 meters to 11 meters, inclusive.

[0016] “Nonviable” and “unviable” are used herein interchangeably to refer to microspores which are not expected to be appropriate for further use (e.g., haploid induction, haploid doubling, and / or incapable of producing a regenerated plant). In contrast, “viable” microspores are those that are appropriate for further use.

[0017] As used herein “labeling” of microspores, “labeled” microspores, or similar phrases refers to any biotechnology or tissue culture process which adds a non-natural substance or genetic or polypeptide sequence to the micropores, or a portion thereof, in order to produce oradd a signal detectable via microscopy. Examples of labeling include, but are not limited to, fluorescent tagging of a gene product or small molecule (e.g., GFP tagging of a protein, fluorescence in situ hybridization (FISH), or antibody-based labeling), dyes (e.g., trypan blue or propidium iodide), and pigments / stains such as crystal violet. As used herein, “non-labeled”, “unlabeled” and similar phrases made with respect to microspores refers to microspores without a non-natural substance added for detection. An advantage of the present invention is that it may be practiced on non-labeled microspores, because labeling of microspores can be challenging due to dye or other label blockage by the rigid cell wall, significant viability reduction as a result of the label, and wide ranging autofluorescence when a fluorescent label is used.

[0018] Methods of imaging and analyzing plant microspores comprising determining if the microspores are uninucleate or binucleate

[0019] Provided herein is a method of imaging and analyzing plant microspores. The method comprises obtaining one or more two-dimensional micrographs of a microspore population. The microspore population is not labeled.

[0020] In some examples, the microspore population is obtained from a dicot plant species. In some examples, the microspore population is obtained from a tobacco plant, an Arabidopsis plant, a soybean plant, a canola plant, a cotton plant, or an alfalfa plant. In some examples, the microspore population is obtained from a monocot plant species. In some examples, the microspore population is obtained from a Poaceae plant. In some examples, the microspore population is obtained from a corn, wheat, sorghum plant, or rice. In some examples, the microspore population is obtained from a Panicoideae plant. In some examples, the microspore population is obtained from a Pooideae plant.

[0021] Any microscope suitable to capture individual two-dimensional micrographs and z-stacks of individual micrographs (optionally, both using white light and / or brightfield) can be used in the method. In some examples, the microscope comprises an objective lens capable of at least 2X magnification. In some examples, it is not necessary for the microspore to comprise an objective lens capable of more than 40X magnification. Additional elements of the microscopy procedure and microscope hardware can be accounted for as needed in specific scenarios. For example, a motorized stage capable of moving in the x and y directions and / or automated micrograph capture can be employed to increase throughput.

[0022] The method comprises sending the one or more two-dimensional micrographs to an image processing system. The image processing system can comprise a general purpose computer capable of receiving the micrographs (e.g., receiving them electronically via either awired or wireless connection). The general purpose computer can comprise both a CPU and, optionally, a GPU. Furthermore, the image processing system should be capable of running software program(s) to detect, identify, and discriminate between subcellular structures such as nuclei, germination pores, and cell walls of the subject microspores.

[0023] The method comprises segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a population of segmented microspores. In some examples, the segmenting comprises binary segmentation, filling of enclosed boundaries, and removal of outlier binary components. In such examples, a series of image transformation functions can also be utilized to successfully segment contiguous regions (e.g., microspores that are contacting each other) into distinct components (e.g., distinct cells). In some examples, remaining contiguous objects that have been erroneously identified as single cells can be corrected by quantification and thresholding of the area and circularity of segmented cells. In some examples, the image transformation functions can comprise computing the Euclidean distance transform of the binary image, a watershed transform, an extended minima transform, and / or imposing regional minima. In some examples, an image transformation function can be performed multiple times and / or in combination with different transformation functions. Image transformation function can be followed by the removal of the final watershed lines or ridges from the filled binary components. An example of the results of the auto segmentation step is shown in FIG. 1 , which illustrates that this method of auto segmentation is effective in distinguishing individual microspores even in a crowded field of view. This auto segmentation may be utilized as an initial step in determining microspore viability, determining whether the microspore is uninucleate or binucleate, and / or assessing microspore morphology.

[0024] The method comprises obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented. The method of imaging and analyzing plant microspores disclosed herein makes use both of individual, two-dimensional micrographs (e.g., during the image segmentation) and of z-stacks of several two-dimensional micrographs taken as a series with progressively changing focal length (e.g., to find nucleus edges). The individual micrographs used during image segmentation can comprise a member of the z-stack (e.g., a mid-plane image), or the individual micrograph(s) can be captured separately from the z-stack(s). If the two-dimensional micrograph(s) are captured separately from the z-stack(s), both sets of images can be captured in either order. Thus, the z-stacks can be obtained before or after segmentation. In some examples, it can be beneficial to capture allimages of a population such that movement of microspores relative to each other and over time is reduced or eliminated.

[0025] The method comprises (a) using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and (b) determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry. Upon investigation, it was observed that nucleus maximum pixel intensity is a local maximum in 3 dimensions (see Fig. 7), an observations that is used herein to distinguish the nucleus local maximum from other local maxima such as those associated with other cytoplasmic components, the microspore germination pore, and / or the microspore border / cell wall. In some examples and in order to measure local maxima, locations of local maxima, prominence of local maxima, and width of local maxima (in individual focal planes and / or in z-projected images of segmented viable cells) signal processing can be used to analyze these parameters in 3 dimensions. These measurements and qualities are used to distinguish nuclei signals from other sources of local maxima.

[0026] In some examples, the two-dimensional micrographs are captured using white light and no optical filter. In some examples, the z-stacks are captured using white light and no optical filter. Advantageously, because no labeling is required for the methods disclosed herein, it is not necessary to use light of a particular wavelength. Nonetheless, filtered light or light of a particular wavelength may be used in some examples of the present method.

[0027] In some examples, the two-dimensional micrographs (and / or z-stacks thereof) are obtained via an objective lens providing from 2X to 40X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 10X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 20X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 30X magnification.

[0028] In some examples, the method comprises determining microspore viability. Microspores can be labeled as “viable” or “nonviable”. In some examples, only viable cells are determined to be uninucleate or binucleate. The determination of viability can be made via the image processing system.

[0029] These automated determinations of viability, nucleation, morphology and / or developmental stage can be utilized as part of a fluidic or microfluidic system, wherein a portion of such fluidic or microfluidic system comprises in imaging step in which microspores are imaged for sorting, re-sorting and purification, for ultimate use in transformation, cargo delivery, tissue culture and / or plant regeneration. In one instance, the imaging and automateddeterminations of viability, nucleation, morphology and / or developmental stage may occur in populations of microspores prior to utilizing such populations in a fluidic or microfluidic system, as an initial quality check on the characteristics of the microspore. In other instances, when used as part of the system, the imaging step may be used in tandem with a sorting step, to assess the output of the sorting step and / or potential need to re-sort the population of microspores, with the ultimate object of obtaining a purified population of viable microspores, and sometimes even more specifically, a purified population of viable uninucleate microspores and / or a purified population of viable binucleate microspores. Microspores may also be selected based on morphology, such as size, and / or developmental stage. Sorting may be accomplished in one or more ways, including by using filters or density gradients. One advantageous method of sorting that may be used in combination with the imaging and automated determinations of viability and / or nucleation is sorting by electrophoresis, which has been shown to be capable of sorting viable and non-viable microspores. For example, see W02021212102, a tunable microfluidic dielectrophoresis sorter, which is incorporated herein by reference.

[0030] In cell types such as microspores, organelle disruptions due to viability loss can create high deviations in cytoplasmic pixel intensities (e.g., brightness). This discovery regarding viability and pixel intensity was surprising since mammalian cells (as distinguished from plant microspores) were previously seen to change shape, rather than change pixel intensity, when viability was lost. In some examples, the viability determination therefore comprises statistical analysis of the pixel intensity matrix of the segmented cell interior. In some examples, the viability determination is at least partly based on the ratio of plant microspore average pixel intensity to maximum pixel intensity.

[0031] In some examples, the viability determination involves measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse. In some examples, these properties can be measured to provide an area measurement for the microspore. FIG. 2 shows the determination of viable and nonviable microspores based upon analysis of both area and pixel intensity.

[0032] In some examples, the nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack and (ii) pixel intensity signal processing along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, and bounded within individual microspores to identify the geometry of refined edge-filled images to segmentnuclei and determine nucleus geometrical properties. The nucleus geometrical properties to be measured can comprise equivalent diameter, area, perimeter, perimeter-to-square root of area, aspect ratio, and circularity. A schematic drawing of the analysis of pixel intensity profiles along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, bounded within individual microspores, is shown in FIG. 6. In some examples, the pixel intensity signal processing along parallel lines in the x, y, or x and y direction comprises curve fitting and peak analysis.

[0033] Example results of (i) edge detection at multiple sensitivity thresholds are shown in FIG. 3. Filling of those edges is shown in FIG. 4. The results shown in FIG. 4 can be modified or adjusted via (ii) pixel intensity signal processing along parallel lines for each component image of the z-stack to identify the geometry of refined edge-filled images to segment nuclei and determine nucleus geometrical properties and to identify true nuclei signals (FIG. 5 and top of Fig. 7) thereby allowing for the identification of geometry of corrected edge-filled images to segment nuclei and determination of nucleus geometrical properties. “Refined” in this sense means that the filled edges of FIG. 4 are adjusted based on the analysis of pixel intensity profiles performed in (ii) to produce the true nuclear signals of FIG. 5. The analysis of pixel intensities can use a set of 3 parameters. Local maxima (or value), thickness (or width), and prominence of fitted curves to intensity line profiles crossing the detected edges can be used to detect the true nucleus signal (Fig. 7). This can be used as a feedback for the edge filled data (e.g., FIG. 4) to eliminate the edges that do not belong to the nucleus, thereby refining the previous edge filling data.

[0034] In some examples, uninucleate and binucleate microspores can be further classified as early stage uninucleate, late stage uninucleate, early stage binucleate, or late stage binucleate. In some examples, K-means clustering is used to distinguish between at least some uninucleate and binucleate cells. In some examples, K-means clustering is used to distinguish between late uninucleate cells and early binucleate cells.

[0035] In some examples, pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit. In some examples, the pixel intensity information comprises grouping of peaks of the Gaussian curve, location of peaks of the Gaussian curve, thickness / width of peaks of the Gaussian curve, and / or peak prominence of peaks of the Gaussian curve. The arrows of FIG. 7 show examples of peaks, peak prominence, and width. In some examples, the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores. Surprisingly, it was found that a Gaussian model performed relativelywell in this respect, while other fitting models (e.g., spline fitting, polynomial fitting, and exponential fitting) were found to perform undesirably.

[0036] FIG. 7 (upper two panels) shows a line scan across a nucleus with (right) and without (left) Gaussian fit. The lower panels show four examples of other scans that do not cross nuclei, with and without Gaussian fit.

[0037] In some examples, the method comprises determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing. Determining this status can provide information about whether or not cargo (e.g., a polypeptide, polynucleotide, a drug, or any other molecule of interest) has been loaded into the cell via, for example, a microinjection.

[0038] In some examples, the method can be performed at least partly automatically, according to software code, with inputs provided by a computer user, for example, via a graphical user interface (GUI). For example, upon appropriate inputs, software can perform microspore segmentation, determine nuclear geometry (e.g., uninucleate or binucleate microspores), and / or determine microspore viability.

[0039] Methods of imaging and analyzing plant microspores comprising determining microspore viability

[0040] Provided herein is a method of imaging and analyzing plant microspores. The method comprises obtaining one or more two-dimensional micrographs of a microspore population. The microspore population is not labeled.

[0041] In some examples, the microspore population is obtained from a dicot plant species. In some examples, the microspore population is obtained from a tobacco plant, an Arabidopsis plant, a soybean plant, a canola plant, a cotton plant, or an alfalfa plant. In some examples, the microspore population is obtained from a monocot plant species. In some examples, the microspore population is obtained from a Poaceae plant. In some examples, the microspore population is obtained from a corn, wheat, sorghum plant, or rice. In some examples, the microspore population is obtained from a Panicoideae plant. In some examples, the microspore population is obtained from a Pooideae plant.

[0042] Any microscope or microscopy system suitable to capture individual two-dimensional micrographs and z-stacks of individual micrographs (optionally, both using white light and / or a brightfield) can be used in the method. In some examples, the microscope comprises an objective lens capable of at least 2X magnification. In some examples, it is not necessary for the microspore to comprise an objective lens capable of more than 40X magnification. Additionalelements of the microscopy procedure and microscope hardware can be accounted for as needed in specific scenarios. For example, a motorized stage capable of moving in the x and y directions and / or automated micrograph capture can be employed to increase throughput. Such motorized stage may be implemented in an overall fluidic or microfluidic system for imaging, including as part of a pre or post sorting stage or a final quality control check.

[0043] The method comprises sending the one or more two-dimensional micrographs to an image processing system. The image processing system can comprise a general purpose computer capable of receiving the micrographs (e.g., receiving them electronically via either a wired or wireless connection). The general purpose computer can comprise both a CPU and, optionally, a GPU. Furthermore, the image processing system should be capable of running software program(s) to detect, identify, and discriminate between subcellular structures such as nuclei, germination pores, and cell walls of the subject microspores.

[0044] The method comprises segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a population of segmented microspores. In some examples, the segmenting comprises binary segmentation, filling of enclosed boundaries, and removal of outlier binary components. In such examples, a series of image transformation functions can also be utilized to successfully segment contiguous regions (e.g., microspores that are contacting each other) into distinct components (e.g., distinct cells). In some examples, remaining contiguous objects that have been erroneously identified as single cells can be corrected by quantification and thresholding of the area and circularity of segmented cells. In some examples, the image transformation functions can comprise computing the Euclidean distance transform of the binary image, a watershed transform, an extended minima transform, and / or imposing regional minima. In some examples, an image transformation function can be performed multiple times and / or in combination with different transformation functions. Image transformation function can be followed by the removal of the final watershed lines or ridges from the filled binary components. An example of the results of the segmentation is shown in FIG. 1.

[0045] In some embodiments the method comprises determining microspore viability. Microspores can be labeled as “viable” or “nonviable”. In some examples, a determination is made individually for each member of the microspore population. In some examples, each determination for an individual member of the population can be used to generate a value for the proportion of microspores which are viable. In some examples, the population of microspores in the one or more two-dimensional micrographs can be used to estimate theviability of a larger population. The determination of viability can be made via the image processing system.

[0046] In cell types such as microspores, organelle disruptions due to viability loss can create high deviations in cytoplasmic pixel intensities (e.g., brightness). This discovery regarding viability and pixel intensity was surprising since mammalian cells (vs plant microspores) were previously seen to change shape, rather than change pixel intensity, when viability was lost. Therefore, the approach described herein specific to plant microspores was developed. In some examples, the viability determination therefore comprises statistical analysis of the pixel intensity matrix of the segmented cell interior. In some examples, the viability determination is at least partly based on the ratio of plant microspore average pixel intensity to maximum pixel intensity.

[0047] In some examples, the viability determination involves measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse. One of ordinary skill in the art could utilize some or all of these morphological measurements in a machine learning model to develop criteria indicative of viability. In some examples, these properties can be measured to provide an area measurement for the microspore, where a threshold area may be used as an indication of viability, as described below in Example 2. The threshold area may be adjusted as appropriate for the plant species and characteristics of the microspore population. For example, for certain maize microspore populations, a threshold area above 2500 pm2, 2600 pm2, 2700 pm2, 2800 pm2, 2900 pm2, or 3000 pm2may be used as an indication of viability. FIG. 2 shows the determination of viable and nonviable microspores based upon analysis of both area and pixel intensity. In some embodiments, area can be based on a threshold area as described above. In some embodiments, pixel intensity may be based on the average bright field pixel intensity of an individual microspore divided by the maximum bright field pixel intensity of the same microspore. Microspore viability determination parameters may be established and / or confirmed by comparison to fluorescein diacetate (FDA) viability staining.

[0048] In some examples, the method further comprises obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented. The method of imaging and analyzing plant microspores disclosed herein makes use both of individual, two-dimensional micrographs (e.g., during the image segmentation) and of z-stacks of several two-dimensional micrographs taken as a series with progressively changing focal length (e.g., to find nucleus edges). The individual micrographs used during imagesegmentation can comprise a member of the z-stack (e.g., a mid-plane image), or the individual micrograph(s) can be captured separately from the z-stack(s). If the two-dimensional micrograph(s) are captured separately from the z-stack(s), both sets of images can be captured in either order. Thus, the z-stacks can be obtained before or after segmentation. In some examples, it can be beneficial to capture all images of a population such that movement of microspores relative to each other and over time is reduced or eliminated.

[0049] In some examples, the method comprises (a) using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and (b) determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry. Upon investigation, it was found that nucleus maximum pixel intensity is a local maximum in 3 dimensions. Other local maxima include those associated with other cytoplasmic components, the microspore germination pore, and / or the microspore border / cell wall, and in embodiments of this invention, those other local maxima are differentiated from the nucleus local maxima by measurements of the prominence of the local maxima and width of the local maxima (in individual focal planes and / or in z-projected images of segmented viable cells) with signal processing to analyze these parameters in three dimensions (Fig. 7). These three-dimensional measurements and qualities are used to distinguish nuclei signals from other sources of local maxima.

[0050] In some examples, only viable cells are determined to be uninucleate or binucleate.

[0051] In some examples, the two-dimensional micrographs are captured using white light and / or brightfield and no optical filter. In some examples, the z-stacks are captured using white light and / or brightfield and no optical filter. Advantageously, because no labeling is required for the methods disclosed herein, it is not necessary to use light of a particular wavelength. Nonetheless, filtered light or light of a particular wavelength may be used in some examples of the present method.

[0052] In some examples, the two-dimensional micrographs (and / or z-stacks thereof) are obtained via an objective lens providing from 2X to 40X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 10X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 20X magnification. In some examples, the two-dimensional micrographs are obtained via an objective lens providing from 2X to 30X magnification.

[0053] In some examples, the nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack and (ii) pixel intensity signal processing along parallellines in the x, y, or x and y direction, for each component image of the z-stack, and bounded within individual microspores to identify the geometry of refined edge-filled images to segment nuclei and determine nucleus geometrical properties. The nucleus geometrical properties to be measured can comprise equivalent diameter, area, perimeter, perimeter-to-square root of area, aspect ratio, and circularity. A schematic drawing of the analysis of pixel intensity profiles along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, bounded within individual microspores, is shown in FIG. 6. In some examples, the pixel intensity signal processing along parallel lines in the x, y, or x and y direction comprises curve fitting and peak analysis.

[0054] Example results of (i) edge detection at multiple sensitivity thresholds are shown in FIG.3. Filling of those edges is shown in FIG. 4. The results shown in FIG. 4 can be modified or adjusted via (ii) (pixel intensity signal processing along parallel lines for each component image of the z-stack) in order to identify true nuclei signals (FIG. 5) thereby allowing for the identification of geometry of corrected edge-filled images to segment nuclei and determination of nucleus geometrical properties. “Refined” in this sense means that the filled edges of FIG. 4 are adjusted based on the analysis of pixel intensity profiles performed in (ii) to produce the true nuclear signals of FIG. 5. The analysis of pixel intensities can use a set of three parameters. Local maxima, thickness, and prominence of fitted curves to intensity line profiles crossing the detected edges can be used to detect the true nucleus signal. This can be used as a feedback for the edge filled data (e.g., FIG. 4) to eliminate the edges that do not belong to the nucleus, thereby refining the previous edge filling data.

[0055] In some examples, uninucleate and binucleate microspores can be further classified as early stage uninucleate, late stage uninucleate, early stage binucleate, or late stage binucleate. In some examples, K-means clustering is used to distinguish between at least some uninucleate and binucleate cells. In some examples, K-means clustering is used to distinguish between late uninucleate cells and early binucleate cells.

[0056] In some examples, pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit. In some examples, the pixel intensity information comprises grouping of peaks of the Gaussian curve, location of peaks of the Gaussian curve, thickness / width of peaks of the Gaussian curve, and / or peak prominence of peaks of the Gaussian curve. The arrows of FIG. 7 show examples of peaks, peak prominence, and width. In some examples, the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores. Surprisingly, it was found that a Gaussian model performed relativelywell in this respect, while other fitting models (e.g., spline fitting, polynomial fitting, and exponential fitting) were found to perform undesirably.

[0057] FIG. 7 (upper two panels) shows a line scan across a nucleus with (right) and without (left) Gaussian fit. The lower panels show four examples of other scans that do not cross nuclei, with and without Gaussian fit.

[0058] In some examples, the method comprises determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing. Determining this status can provide information about whether or not cargo (e.g., a polypeptide, a polynucleotide, a drug, or any other molecule of interest) has been loaded into the cell via, for example, a microinjection. Such analysis may be performed as part of a fluidic or microfluidic system.

[0059] In some examples, the method can be performed at least partly automatically, according to software code, with inputs provided by a computer user, for example, via a graphical user interface (GUI). For example, upon appropriate inputs, software can perform microspore segmentation, determine nuclear geometry (e.g., uninucleate or binucleate microspores), and / or determine microspore viability.

[0060] The present disclosure will be more fully understood by reference to the following clauses. a. A method of imaging and analyzing plant microspores, the method comprising: obtaining one or more two-dimensional micrographs of a microspore population, wherein the microspore population has not been labeled; sending the one or more two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented; and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry. b. The method of clause a, wherein the two-dimensional micrographs are captured using white light and no optical filter. c. The method of clause a or b, further comprising determining microspore viability.d. The method of clause c, wherein only viable cells are determined to be uninucleate or binucleate. e. The method of clause c or d, wherein the viability determination involves measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse. f. The method of clause any one of clauses a-e, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 40X magnification. g. The method of clause f, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 10X magnification. h. The method of any one of clauses a-f, wherein nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack and (ii) pixel intensity signal processing along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, and bounded within individual microspores to identify the geometry of refined edge- filled images to segment nuclei and determine nucleus geometrical properties. i. The method of clause h, wherein pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit. j. The method of clause i, wherein the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores. k. The method of any one of clauses a-j, further comprising determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing. l. The method of any one of clauses a-k, wherein K-means clustering is used to distinguish between at least some uninucleate and binucleate cells. m. A method of imaging and analyzing plant microspores, the method comprising: obtaining one or more two-dimensional micrographs of a microspore population, wherein the microspore population has not been labeled; sending the one or more two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; and determining microspore viability.n. The method of clause m, further comprising obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry. o. The method of clause m or n, wherein the two-dimensional micrographs are captured using white light and no optical filter. p. The method of clause n, wherein only viable cells are determined to be uninucleate or binucleate. q. The method of any one of clauses m-p, wherein the viability determination comprises measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse. r. The method of any one of clauses m-q, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 40X magnification. s. The method of clause r, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 10X magnification. t. The method of clause n or p, wherein nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack, (ii) analysis of pixel intensity profiles along parallel lines in the x, y, or x and y direction, for each component image of the z- stack, and bounded within individual microspores, to identify the geometry of refined edge-filled images to segment nuclei and determine nucleus geometrical properties. u. The method of clause t, wherein pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit. v. The method of clause u, wherein the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores. w. The method of any one of clauses m-v, further comprising determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing. x. The method of clause n, p, or t wherein K-means clustering is used to distinguish between at least some uninucleate and binucleate cells.y. The method of any one of clauses m-x, wherein the viability determination comprises statistical analysis of the pixel intensity matrix of the segmented cell interior.

[0061] Examples - The present disclosure will be more fully appreciated with reference to the following non-limiting examples.

[0062] Example 1 - Collection, handling, and culturing of maize microspores

[0063] Maize microspores of Sx19 (a hybrid resulting from a cross between B73 and Mo17) and a proprietary inbred variety were collected via standard techniques. Briefly, tassels of plants from 42-55 days after planting were collected from maize plants. The collection time after planting varied with the growing season and weather. Microspores were released from anthers by mechanical disruption and purified via centrifugation.

[0064] Once collected, microspores were maintained in a basal culturing medium comprising sucrose at pH 5.8. Microspores were transferred to storage medium comprising 0.6 M mannitol and proline for microscopy.

[0065] Example 2 - Microscopy, microspore segmentation, image analysis, and viability determination

[0066] Once microspore cells were prepared for microscopy, the cells were transferred to an inverted light microscope for diascopic imaging. Z-stack image sequences were captured using a 10x objective lens and z-interval of 2 pm resulting in 50 images per region covering the bottom to top planes of microspores.

[0067] After z-stacks were captured, the file directory containing the z-stacks, the objective magnification, z-interval, and intensity / morphology thresholds were input into the image analysis program. The following intensity / morphology thresholds were used:First and last image numbers within z-stack sequence to be processed. Default values are 1 and the last image number in folder.The ratio of average to maximum microspore pixel intensity was measured for viability determination. The default value was set to 0.7 for our light microscope equipped with a 10x objective lens.Microspore area for viability determination. The default value was set to 2700 pm2, below which microspores are considered unviable.Normalized intensity threshold for elimination of high intensity pixels from corresponding binary images in z-stack. These signals are normally located on cell pore and wall and interfere with nuclei detection. Default value is set to 0.7.

[0068] Regarding microspore area, this value can be altered as needed for use of this method with microspores of different plant varieties or species to account for differences in the morphology of the microspores.

[0069] After user inputs were complete, the image analysis program was run on a computer. The code performed the following initial operations to complete segmentation of microspores. Z- stack images were assembled, inverted, and stored in 3D matrices and / or MATLAB cells. High- intensity pixels from raw microspore images were eliminated from the data.

[0070] Once initial processing was complete, microspores were segmented. The mid-plane image of the z-stack was binarized and inverted. Matrix operations including transformations were then performed to segment contiguous microspores. The step-by-step operations are listed below:1- Image inversion2- Hole filling of image regions3- Small object removal from binary image4- Watershed transform.5- Extended minima transform.6- Impose minima operation to modify image using morphological reconstruction.7- Watershed transform

[0071] After matrix transformations, binary components from the final transformant were connected and the geometrical properties and pixel intensities of connected component regions were measured and labeled. Morphological thresholding using the measurements was employed to eliminate unwanted connected components from the labeled image. These operations produced the final segmented image / matrix. An example of segmented microspores is shown in FIG. 1. The same microspores are labeled as viable or unviable in FIG. 2. Viability was determined based on (1) the average bright field pixel intensity of an individual microspore divided by the maximum bright field pixel intensity of the same microspore and (2) microspore area. Maize microspore viability determinations were previously confirmed by comparison to fluorescein diacetate (FDA) viability staining.

[0072] Example 3 - Nuclei edge detection image analysis

[0073] After segmentation was performed, a first binary detection of the edges of the microspore walls, the germination pores, and the nuclei was performed for all image planes. A second binary detection was also performed for the edges of the microspore walls and the germination pores in all image planes, without detection of the nuclei. Logical subtraction of dilated edges from the first and second binary detections was performed to provide nuclei edges in all image planes.

[0074] In order to provide binary nuclei edges in all image panes, secondary dilations of edges in the second binary detection were calculated using morphological structuring elements with specific radii. These secondary dilations were also subtracted from the first binary detection in order to provide binary nuclei edges in all image planes.

[0075] Binary nuclei edges were dilated and filled resulting in filled binary nuclei edges in all image planes. The filled edges were then multiplied by the corresponding bright field image at each focal plane to get intensity matrix of the detected nuclei. The output stack was then concatenated to get 3D intensity matrix of nuclei edges. Maximum intensity projection as well as dilation was performed on the concatenated matrix for displaying the output or performing k- means clustering for nuclei classification. The results of the initial nuclei edge detection in z- stacks are shown in FIG. 3. A projected image of the filled edges is shown in FIG. 4.

[0076] Example 4 - Line scanning and intensity signal processing

[0077] Pixel values (intensity), rows (y coordinate) and columns (x coordinate) of segmented microspores for all images within the z-stack were vectorized and tabulated.

[0078] The pixel columns (vertical pixel lines) which cross filled binary nuclei edges were identified.

[0079] Pixel values of each column were normalized to the focal plane’s maximum microspore intensity. The normalized pixel values and associated row numbers (y coordinate) were stored as vector pairs.

[0080] Gaussian curve fitting was performed on vector pairs at all focal planes of the segmented microspores.

[0081] Peak value, peak thickness, and peak prominence for fitted curves were identified and stored as 3-component groups.

[0082] Certain peak groups with specific ranges for group components belong to a nucleus signal. If all 3 components of a group fall within these ranges and their x and y coordinates belong to a nucleus signal detected from the edge filling operation, the group along with itscoordinates are stored as a true nucleus signal. Examples of such signal processing are shown with respect to an exemplary microspore in Fig. 5.

[0083] Image transformations (e.g., watershed transform), connected component analysis, and K-means clustering of the segmented nuclei (Fig. 4) as well as image signal processing (Fig. 5 and top of Fig. 7) will be performed to connect separate segments of a single nucleus to identify the nucleus as one object or to identify a single segment as two nuclei when a nucleus has just divided into two nuclei (e.g., early bi-nucleate microspore). Thus, the microspores can be identified as uninucleate or binucleate.

Claims

Claims1 . A method of imaging and analyzing plant microspores, the method comprising: obtaining one or more two-dimensional micrographs of a microspore population, wherein the microspore population has not been labeled; sending the one or more two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; and determining microspore viability.

2. The method of claim 1 , further comprising obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry.

3. The method of claim 1 , wherein the two-dimensional micrographs are captured using white light and no optical filter.

4. The method of claim 2, wherein only viable cells are determined to be uninucleate or binucleate.

5. The method of claim 1 , wherein the viability determination comprises measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse.

6. The method of claim 1 , wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 40X magnification.

7. The method of claim 2, wherein nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack, (ii) analysis of pixel intensity profiles along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, and bounded within individual microspores, to identify the geometry of refined edge-filled images to segment nuclei and determine nucleus geometrical properties.

8. The method of claim 7, wherein pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit.

9. The method of claim 8, wherein the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores.

10. The method of claim 1 , further comprising determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing.

11. The method of claim 2, wherein K-means clustering is used to distinguish between at least some uninucleate and binucleate cells.

12. The method of claim 1 , wherein the viability determination comprises analysis of pixel intensity.

13. The method of claim 1 , wherein the viability determination comprises statistical analysis of the pixel intensity matrix of the segmented cell interior.

14. A method of imaging and analyzing plant microspores, the method comprising: obtaining one or more two-dimensional micrographs of a microspore population, wherein the microspore population has not been labeled; sending the one or more two-dimensional micrographs to an image processing system; segmenting, via the image processing system, the one or more two-dimensional micrographs to distinguish individual microspores, thereby creating a segmented population of microspores; obtaining one or more z-stacks of the population of segmented microspores or the population of microspores to be segmented; and using one or more of pixel intensity information and intensity signal processing within the one or more z-stacks of segmented cells to find nuclear edges and nuclear geometry and determining whether cells are uninucleate or binucleate based on the nuclear edges and nuclear geometry.

15. The method of claim 14, wherein the two-dimensional micrographs are captured using white light and no optical filter.

16. The method of claim 14, further comprising determining microspore viability.

17. The method of claim 16, wherein only viable cells are determined to be uninucleate or binucleate.

18. The method of claim 16, wherein the viability determination involves measurement and analysis of one or more of the following properties for each microspore candidate: equivalent diameter, area, perimeter, circularity, semi-major axis, semi-minor axis, aspect ratio, eccentricity, and score of morphological fit into an ellipse.

19. The method of claim 14, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 40X magnification.

20. The method of claim 19, wherein the two-dimensional micrographs are obtained via an objective lens providing from 2X to 10X magnification.

21. The method of claim 14, wherein nuclear edges and nuclear geometry are found using (i) edge detection at multiple sensitivity thresholds to find and fill the edges of microspores, germination pores, and nuclei through the z-stack and (ii) pixel intensity signal processing along parallel lines in the x, y, or x and y direction, for each component image of the z-stack, and bounded within individual microspores to identify the geometry of refined edge-filled images to segment nuclei and determine nucleus geometrical properties.

22. The method of claim 21 , wherein pixel intensity information along the parallel lines is modeled as a multi-term Gaussian fit.

23. The method of claim 22, wherein the image processing system uses the modeled pixel intensity information to distinguish between nuclei and germination pores.

24. The method of claim 14, further comprising determining whether or not the microspore has been pierced based on the pixel intensity information and / or intensity signal processing.

25. The method of claim 14, wherein K-means clustering is used to distinguish between at least some uninucleate and binucleate cells.

Images