Reconstruction of bovine spermatozoa substances distribution and morphological differences between Holstein and Korean native cattle using three-dimensional refractive index tomography

Measurements of the three-dimensional (3D) structure of spermatozoon are crucial for the study of developmental biology and for the evaluation of in vitro fertilization. Here, we present 3D label-free imaging of individual spermatozoon and perform quantitative analysis of bovine, porcine, and mouse spermatozoa morphologies using refractive index tomography. Various morphological and biophysical properties were determined, including the internal structure, volume, surface area, concentration, and dry matter mass of individual spermatozoon. Furthermore, Holstein cows and Korean native cattle spermatozoa were systematically analyzed and revealed significant differences in spermatozoa head length, head width, midpiece length, and tail length between the two breeds. This label-free imaging approach provides a new technique for understanding the physiology of spermatozoa.


Results
General 2D and 3D imaging of bovine, porcine, and mouse spermatozoa. To non-invasively investigate the 3D morphology of spermatozoa cells, the RI tomograms of individual spermatozoa cells from bovine, porcine, and mice were measured using a commercial HT instrument, which could continuously monitor the spermatozoa (Video 1-6). The representative cell RI tomograms and 2D optical phase images are shown in Fig. 1. The 2D optical phase images and 3D RI tomograms of the spermatozoa cells clearly visualized their distinct anatomical structures. The 2D optical phase images showed the characteristic morphology of the mouse spermatozoa, which was different from that of the bovine and porcine spermatozoa. However, the morphologies of the bovine and porcine spermatozoa were not easily distinguishable using the 2D optical phase images. In contrast, the 3D RI tomograms of the spermatozoa showed a significant difference in the morphology of individual cells, distinguishing all three groups. The sperm head of the bovine (Fig. 1A), porcine (Fig. 1B), and mouse (Fig. 1C) showed noticeably different sphericity. The head of the bovine spermatozoa had the highest sphericity, followed by the porcine spermatozoa, and finally the mouse spermatozoa. The 3D tomograms of spermatozoa from the three groups presented bilateral symmetry when seen from the outside. The midpiece portion of the porcine and mouse spermatozoa were more pronounced than those of the bovine spermatozoa. The head shape of the bovine and mouse spermatozoa had distinct back, abdominal, left, and right sides, and the front part of the mouse spermatozoan head was eagle beak-shaped.

Morphological analysis of different bovine spermatozoa.
To further investigate the morphology of individual spermatozoon, we assessed the spermatozoa of HO cows and KN cattle and conducted an in-depth morphological analysis (Fig. 1D,E). The 2D optical phase images enabled the length and width of the head, and length of the midpiece and tail of the HO cow and KN cattle spermatozoa to be analyzed. The 2D morphological analysis results showed that the head lengths of the HO cow and KN cattle spermatozoa were 9.053 ± 0.310 µm and 9.229 ± 0.297 µm, respectively (P < 0.05); the head widths were 4.954 ± 0.268 µm and 5.241 ± 0.297 µm, respectively (P < 0.01); the midpiece lengths were 13.926 ± 0.547 µm and 13.245 ± 0.299 µm, respectively (P < 0.01); and the tail lengths were 46.415 ± 1.386 µm and 47.750 ± 1.373 µm (P < 0.01), respectively.
To perform volumetric analysis of the HO cow and KN cattle spermatozoa, we first obtained ten independent measurements from one bovine spermatozoon and visualized the reconstructed RI tomograms of the entire spermatozoa using the detectable RI range, which showed high specificity and accuracy of substance distribution between 1.345 and 1.380, and a gradient of 0.005 (Fig. 2 and Supplementary Table S1). Then, by optimizing the parameter settings based on the entire RI range (1.345 to 1.380), we found that three RI ranges, namely, RI-I, RI-II, and RI-III, effectively represented the overall cell morphology and internal structures, including the approximate location of the mitochondria. We used these parameter settings for subsequent analyses.

Biophysical analysis of the bovine spermatozoa.
To perform biophysical analysis of the individual cattle spermatozoon, we retrieved the biophysical parameters from the measured RI tomograms, including volume, surface area, concentration, and dry mass. Because the RI value of the cell cytoplasm is linearly proportional to its concentration 29 , the concentration of spermatozoa can be obtained from the measured RI tomogram. In addition, the spermatozoa dry mass can be calculated from the retrieved concentration and the measured cell volume.
To further analyze the components of the spermatozoa, the individual spermatozoon were divided into three components: head, midpiece, and tail, and quantitative imaging analysis was performed on each component (Table 1 and Supplementary Fig. S2). No significant differences were found among the dry mass of the sperm head, midpiece, and tail. For the substances of sperm head, HO cattle had a much higher concentration and sphericity but a smaller volume and surface area than that of KN cattle. For the substances of sperm midpiece, HO cattle had a larger volume and higher concentration than that of KN cattle, whereas no significant difference was found in surface area. For the substances of sperm tail, no significant differences were observed in volume, surface area, and concentration between the two breeds.
3D tomography analysis of HO cow and KN cattle spermatozoa inner structure. After we confirmed that we could accurately and specifically distinguish substances with different RI parameters, we utilized www.nature.com/scientificreports www.nature.com/scientificreports/ the same analytical parameters (i.e., RI-I, RI-II, and RI-III) to show that the head ( Fig. 3 and Video 7), midpiece ( Fig. 4 and Video 8), and tail ( Fig. 5 and Video 9) of the HO and KN spermatozoa exhibited significant differences. As shown in Table 2, RI-I range substances were significantly different in volume, surface area, and dry mass of the sperm head between HO cows and KN cattle. RI-II range substances were significantly different in surface area, concentration, and dry mass of the midpiece between HO cows and KN cattle, as well as the surface area and dry mass of the tail. RI-III range substances were significantly different in volume, surface area, concentration, and dry mass of the sperm head between HO cows and KN cattle, as well as the volume and concentration of the midpiece.
In general, the substances with different RIs of the entire spermatozoa exhibited a bilateral symmetry distribution, not a radial symmetry distribution (Figs 2, 6-8). The composition of the entire spermatozoa substance was generally the same as that of the tail, except for the midpiece that contained regions with high RI. The midpiece and the portion linked to the head contained a substance within the RI-II range.
In the sperm head ( Fig. 6 and Video 10), substances in the RI-I range were mainly distributed on the surface of the sperm, and substances in the RI-II range constituted the "skeletal system" of the sperm head (Fig. 3, Supplementary Fig. S3A and Supplementary Fig. S3B). Most of the substances in the RI-III range were only located at the bottom of the sperm head (near the midpiece). However, the substances in the RI-III range were also detected on both sides of the head, and in the acrosome position in a few spermatozoa (Supplementary  www.nature.com/scientificreports www.nature.com/scientificreports/  Supplementary Fig. S3D). Specifically, the RI-II substances started from the bottom of the head and gradually decreased toward the acrosome until they were completely undetectable ( Fig. 6A-G). Throughout this disappearing process, RI-II substances first decreased on the back and abdomen of the sperm head (Fig. 6D,E), and then disappeared from both sides (Fig. 6F,G).
RI-II range substances were distributed continuously; however, they did not exhibit an absolute ring-shaped structure. When their position reached approximately 1/5 of the length from the bottom of the sperm head, a cavity-like region with a relatively low RI value was observed, which extended toward the top of the head ( Fig. 6H-J). This cavity reached approximately 2/5 of the length from the top of the sperm head. The spermatozoon acrosome was mainly composed of substances in the RI-II range (Fig. 6K,L). The size of the cavity varied among different sperm cells (Video 10, Supplementary Fig. S3E-G).
The substance in the RI-III range was widely distributed in the midpiece. Although some of the substances were distributed in the head, no substance in the RI-III range was detected at the linkage position of the midpiece and the head (Fig. 7A). The substance in the RI-II range was separated from the substance in the RI-III range (Fig. 7C,E and G) and exhibited a discontinuous distribution (Fig. 7B-H). The substance in the RI-III range disappeared at the end of the midpiece (Fig. 7I and Video 11).
From the end of the spermatozoan midpiece to the tail, there was no positional distribution of RI-III ( Fig. 8B-F), and the components that constitute the sperm tail belonged mainly to RI-I and RI-II. We also found a case where the RI-II range substance was intermittently distributed (Fig. 8B-D) and could no longer be detected at the end of the tail (Fig. 8E,F; Video 12).

Discussion
The morphology and internal structure of spermatozoa have a great influence on their motility, affecting slither swimming mode, oocyte fertilization rate, and non-return rates, and are affected by environmental factors, genetics, and the physiological state of individuals 30 . The evaluation of spermatozoa quality has been highlighted as a predictor of fertilization success 31 . Although they can provide valuable information regarding the overall cell shape, 2D label-free imaging methods can only detect abnormal spermatozoa with significant morphological changes and incomplete structures; other methods are mostly based on the chemical labeling of specific substances.
Recently, quantitative phase and label-free imaging techniques have been successfully applied to spermatozoan imaging and digital structure construction. Similar to other studies, the present approach provided a label-free 3D imaging method, which could noninvasively measure the biochemistry and morphology of spermatozoa cells. The physiological state of the spermatozoa was not altered, yet both the external and internal structures could be comprehensively analyzed. QPI techniques are suitable for quick detection and imaging while www.nature.com/scientificreports www.nature.com/scientificreports/ correlative spectroscopy imaging (like Raman spectroscopy) is good at molecular characterization 32 . For example, Raman spectroscopy can reliably and independently detect the activation state of cells, and provides information at the molecular level, allowing for the investigation of functional groups, bonding types, and molecular conformations, and correlating them to the cell structural properties 33,34 . Utilizing QPI techniques, the present study not only acquired the full 3D structure of the studied bovine spermatozoa, but also revealed their 3D translational head motion and the angular velocity of their head spin as well as the 3D flagellar motion and the preference of bovine spermatozoa for helix-shaped 3D swimming trajectories 35,36 . Moreover, the substance distribution in spermatozoa was distinguished accurately and specifically with a 0.005 gradient range. Therefore, we were able to use specific RIs to identify the distribution of related substances and to determine substances in an abnormal state.
Studies have shown that the length of the sperm midpiece is significantly different among different breeds of cattle with different feeding purposes such as beef cattle or dairy cows, and there are large differences among individuals within each breed 37 . Previous studies have shown that a sperm midpiece possesses a plasma membrane, mitochondrion, dense fibers, doublet, central sheath, tubules, and other important organelles from the outside to the inside 38 . In the present study, we found that the RI-III range substance was widely and almost uniquely distributed in the midpiece. The RI-II range substance was separated by the RI-III range substance and exhibited discontinuous distribution; therefore, we hypothesized that the RI-III range substance was a mitochondria-related substance, which was mainly distributed in the midpiece with tyrosine phosphorylated proteins, two of which are phospholipid hydroperoxide glutamate peroxidize and ATP synthase beta subunit, and play a primary role in www.nature.com/scientificreports www.nature.com/scientificreports/ spermatozoan motility 39 . Undoubtedly, structural abnormality of the sperm midpiece has a negative effect on the spermatozoan quality. In the present study, a combination of 2D and QPI imaging was used to analyze the spermatozoa midpiece. We first identified the different midpiece lengths of the HO cow and KN cattle spermatozoa. Then, the different volumes and distributions of the spermatozoa in the two breeds were measured with different RIs. The results showed that the corresponding mitochondrial distribution range (or volume) might also be different. Therefore, there are potential differences between HO cow and KN cattle spermatozoa motility, patterns, and adaptation to the environment. Our analysis provides the basis for vibrant sperm cell selection in artificial insemination and in vitro fertilization. However, whether and how the length and inner structure of the midpiece is meaningful for animal breeding remains unclear.
An abnormal spermatozoon head is often caused by genetic defects 40 , food intake 41 , chemical substances, or radio-frequency radiation 42 , which results in a loss of fertilization ability. Simultaneously, conventional staining and labeling-based imaging techniques may change the morphometric dimensions of a spermatozoon head, with the plasma membrane, acrosome, and nucleus being the most important components 43 . Therefore, a QPI technique that can shorten the analysis time and maintain the morphology and structural integrity of a spermatozoon head would be highly beneficial for subsequent artificial insemination and assisted reproductive technology. The present study found that the RI distribution inside the spermatozoon head is complicated, and there were significant differences in the spermatozoa between the two bovine breeds, especially in the nucleus and acrosome. First, the substance corresponding to the same RI exhibited a discontinuous distribution along the x-y, y-z, and x-z planes inside the head. These findings are similar to previous studies that showed the existence of spermatozoa chromatin positioning factors specific to individual chromosomes. The nucleus can be divided into several asymmetric areas. These non-random preferential lateral and longitudinal intranuclear positionings have potential effects on the fertilization process and chromatin remodeling events during the early stages of embryogenesis 44 . Second, there are differences in the nuclear space of different bovine breeds and the same RI distribution inside them. Whether this affects the ability of spermatozoa to achieve insemination between breeds is unclear. However, this might be a new direction and new ideas for screening the morphology, function, and quality of the spermatozoon head.
The present method can potentially be used in various applications, including sperm quality monitoring and rapid selection. It can also provide information regarding unexplained failures of multiple assisted reproductive techniques and recurrent abortions. For example, the general morphology and structure of spermatozoa are known to be related to sperm motility and competition ability 45,46 . However, one limitation of the present method is a lack of molecular specificity. Unlike fluorescent labeling techniques, RI distribution cannot provide specific molecular information because the RI increments that are necessary to convert RI values to the local concentration of intracellular components are too similar 28 . Although the molecular specificity of the present RI measurement technique may not be as high as labeling methods, it could be improved by measuring spectroscopic RI distributions, which can separate molecules spectrally owing to their different optical dispersion properties 47 . We also performed measurements of the dry mass at the cellular and subcellular level. This approach can be utilized for animal reproduction and human assisted reproduction, especially spermatozoa selection during preparation for artificial insemination 8 .    www.nature.com/scientificreports www.nature.com/scientificreports/ In summary, we presented the measurements of 3D RI tomograms of individual spermatozoon and performed a systematic analysis of the morphology and internal structures of spermatozoa from HO cows and KN cattle. The results showed that HO cow and KN cattle spermatozoa have significant differences in morphological parameters such as length, width, and volume of the head, midpiece, and tail. There were also significant differences in the distribution of intracellular components containing the same RI inside the head, midpiece, and tail. These results provide new techniques for further in-depth structural analysis, quality monitoring, and spermatozoa selection in a rapid and label-free manner. Whether the internal RI distribution of spermatozoa can be used as a diagnostic parameter to evaluate the state of a spermatozoa and fertilization ability remains an open question; however, it is now accessible to direct experimental study. Principles of optical diffraction tomography. To measure 3D RI tomograms of individual spermatozoon, we used a commercial ODT system (HT-1H, Tomocube Inc., Republic of Korea) 48 . CT reconstructs X-ray absorptivity in 3D, whereas ODT reconstructs RI distributions using information regarding light absorptivity and RI. ODT reconstructs the 3D RI tomogram of a sample from the measurements of multiple 2D holograms with various illumination angles 28 . The experimental setup is based on an off-axis Mach-Zehnder interferometer equipped with a digital micromirror device (DMD) 49,50 (Supplementary Fig. S4). sample collection and preparation. Mouse spermatozoa were isolated from two-month-old ICR mice.

Materials and
Longitudinal cuts were made in the cauda epididymis and the tissue was incubated in phosphate-buffered saline (PBS) solution at 37 °C to enable motile, mature spermatozoa to swim out. The Large White pig semen samples were obtained from boars in the local farm (Darby genetics, Anseong, Gyeonggi-do, Korea) using the gloved-hand method and delivered to the laboratory immediately under 37 °C. Semen samples were washed three times with 0.85% NaCl (w:v) containing 100 mg/L bovine serum albumin. Frozen semen from five Holstein (HO, a well-known dairy cattle) bulls and five Korean native (KN, a beef cattle with good meat quality) bulls were obtained from local farms (Nonghyup Economic Landholding Hanwoo Improvement offices, Chungnam, Korea). Their spermatozoa were separated using the Percoll method 51 , gently washed with PBS, and the number of sperm was adjusted to a density suitable for observation. One drop (approximately 25 µL) of PBS containing spermatozoa from the different breeds was placed on a 0.13-0.17 mm NEO micro glass slide (Matsunami, Tokyo, Japan) and measured using the HT-1H system. Six spermatozoa from each semen sample were randomly selected for observation.

Image acquisition and 3D RI model reconstruction.
For tomographic reconstruction, a total of 30 holograms of the single spermatozoon sample at various illumination angles were measured. Using a phase retrieval algorithm 52,53 , optical field images containing both the amplitude and phase maps of the sample were retrieved from the measured holograms. Based on the Fourier diffraction theorem with Rytov approximation 54 , the 3D RI tomogram of the sample was reconstructed from the multiple 2D optical field images. The lateral and axial optical resolutions of the ODT system were 110 nm and 360 nm, respectively, according to the Lauer criterion 55,56 . www.nature.com/scientificreports www.nature.com/scientificreports/ Due to the limited numerical apertures of the condenser and objective lenses, side scattering information was not retrieved, resulting in inaccuracy in tomographic reconstruction, the so-called "missing cone" problem. To remedy this, an iterative regularization algorithm based on non-negativity was applied to the reconstruction process 57 .
Quantitative image analysis. The structures of the head, midpiece, and tail of HO and KN spermatozoon cells were systematically investigated by retrieving morphological properties from the measured 3D RI maps. In each 3D RI map, the cell and subcellular compartments were determined from the RI values. The RI range (RI start-end ) in spermatozoa cells was divided into three groups (RI-I: 1.3451-1.3520; RI-II: 1.3521-1.3640; and RI-III: 1.3641−1.382). For visualization, substances belonging to RI-I, RI-II, or RI-III were digitally color-coded with purple, yellow, or green, respectively, unless stated otherwise. The volumes of the whole spermatozoa cells and their compartments (head, midpiece, and tail) were retrieved by multiplying the unit volume of a voxel by the total number of voxels in the occupied region. The surface area was calculated as the area of the outermost surface of the occupied region. The sphericity index of each spermatozoon head was calculated from the volume and surface area. A higher sphericity index indicated a stronger resemblance to a sphere.
The concentration of dry matter mass in a spermatozoon sample can be calculated from its RI value because the cytoplasm protein and lipid concentrations are linearly proportional to their RI value 29,58 by a constant called the refractive index increment. Most non-aqueous small molecules in the spermatozoa, including proteins, DNA, and lipids, have similar RII values 59 . The dry mass of the occupied region in sperm cells was then calculated by multiplying the dry mass concentration by the occupied volume. statistical analysis. Results are presented as the mean ± standard deviation (SD). Data obtained from two groups were compared using Student's t-test. All statistical analyses were performed using SPSS version 22.0 (IBM, IL, USA) software. Solid dots and open circles represent HO and KN measurements, respectively. Significant differences are represented with *(P < 0.05) and **(P < 0.01).