Skip to main content
NCBI home page
Royal Society of Chemistry logoLink to Royal Society of Chemistry
. 2026 Jul 2;26(16):4533–4543. doi: 10.1039/d6lc00399k

A sperm racetrack to separate sperm by swim speed

James Baldwin-Brown a,✉, Lauren Hulse a, Sabin Nepal c, Ata Ullah c, Utpal Saha c, Kenneth Ivan Aston b, Bruce Gale c, Nitin Phadnis a,✉
PMCID: PMC13324299  PMID: 42387989

Abstract

The genetic basis for sperm swimming ability in humans is underinvestigated. In particular, selfish genes known as segregation distorters dramatically influence sperm swimming in organisms like mice. Segregation distorters are difficult to screen for in humans because no tool exists to compete sperm by swimming ability and collect the separated fast and slow sperm for further study. We designed, built, and tested a “sperm racetrack”, a microfluidic device that races sperm against each other and allows collection of sperm based on the distance travelled. This biologically-inspired PDMS device contains a long, straight channel with counterflow against which sperm swim naturally, and contains regularly-spaced ports along the channel for collection of sperm after the race. We performed a series of test races and found that sperm in the fourth channel segment are significantly faster swimming according to video analysis than those in the second channel segment. Sperm from these same segments are significantly different in speed after extraction from the channel as well. We characterized several swimming behaviors for these faster and slower sperm, including linearity (ability to swim in a straight line) and wobble (amount of side-to-side head movement per beat), and found a significant association between higher sperm wobble and distance travelled along the channel in extracted sperm. Together, we show that the sperm racetrack separates sperm according to swimming speed, and this sorting of sperm is faithfully maintained after extraction from the device as fast- and slow-swimming groups. The sperm racetrack may prove useful in understanding the genetic and physiological causes of sperm immotility, and in identifying swimming-based segregation distorters in humans. The sperm racetrack may also serve as a tool for enriching fast-swimming sperm for selection in assisted reproductive technologies such as in vitro fertilization.

A sperm racetrack device that competes sperm against each other by swimming speed. We show that the device successfully separates sperm into faster and slower aliquots using computer-aided sperm analysis.graphic file with name d6lc00399k-ga.jpg

Introduction

Segregation distorters are selfish genes that manipulate spermatogenesis by sabotaging the development or swimming ability of sperm carrying their competing homologous chromosomes.1 Because distorters operate by either killing or inactivating sperm, they are a potential cause of infertility. Consistent with this idea, segregation distorters have been identified across a wide range of species and are often associated with reduced male fertility.2,3 The role of segregation distortion in causing male infertility in humans, however, remains under-investigated.

Some distorters inactivate rival sperm by reducing swimming ability. Among distorters that weaken sperm swimming, the best characterized system is the mouse t-locus. In males heterozygous for the t-haplotype on chromosome 17, nearly all offspring inherit the t-haplotype. This is because sperm bearing the t-haplotype reduce the progressive motility of sperm bearing the wild-type haplotype3 (Fig. 1). The mouse t-haplotype drive phenotype was discovered serendipitously because the t-haplotype drive loci are linked to a marker that causes taillessness.4–8 Heterozygous t-locus mice produced mostly t-locus offspring; later research uncovered the inactivation of wild type sperm by t-locus sperm as the cause of this non-Mendelian segregation.

Fig. 1. A. The mouse t-locus is a distorter that reduces the swimming ability of rival sperm. Sperm swimming through the mouse oviduct to fertilize eggs. T-haplotype sperm (red) impart a mobility cost on wild type sperm (blue). Fertilized offspring predominantly carry the t-haplotype. This figure shows a crucial period when sperm travel from the uterus to the fallopian tube. B. Like the fallopian tube, the sperm racetrack separates sperm by swimming ability. In this simplified diagram, sperm are injected into the sperm input port, “race” against counterflow, and separate into better and worse swimmers. We then extract aliquots of faster and slower sperm for comparison at evenly spaced extraction ports (open circles).

Fig. 1

Open in a new tab

Such sperm swimming-related distorters may exist in humans, but they have not been identified because the approach used to discover the t-locus in mouse is not applicable in humans. In addition, few tools can agnostically detect the presence of such distorters. In a race between sperm from an individual, sperm that bear a distorting allele are expected to swim farther and outcompete those carrying wild type alleles. This provides a way to detect swimming-based distorters: they manifest as allelic biases between the best and worst swimming sperm. Such a test would require the ability to “race” sperm against each other and extract the best swimmers for detailed study. While we primarily discuss distortion here, such a device could actually separate sperm that differ in swimming ability for any reason. Past studies have shown that both environment (especially toxins) and genetics play a role in sperm swimming.9–11 A device for racing sperm against each other could be used to identify candidate sperm that differ for any of these reasons, and if the sperm can be extracted from the device, follow-up studies may distinguish between these potential effectors of swim speed.

Many existing devices sort sperm one way or another.12–26 The mechanisms used to induce and measure sperm movement are highly varied, including acoustofluidics,19 passive pumping of immotile sperm with26 or without a specialized syringe,13,15 simple rheotaxis,14,22,24 rheotaxis with obstacles,21,25,27 obstacles without rheotaxis,12 or Dean flow.16 These various mechanisms show the diversity of thought in solving the problem of sperm sorting. Here, we distinguish between motility and swimming ability. When we discuss motility we are referring to the ability of sperm to swim at all, and when we discuss swimming ability, we are referring to quantitative differences in the ability of sperm to move quickly in a desired direction. Existing devices can be roughly split into those that separate sperm into high-motility and low-motility aliquots for extraction and further use,12,13,15,16,19,21–26 and those that quantitatively measure sperm speed within a device.12,14

By necessity, these devices have at least some biomimetic features because they are attempting to measure or separate sperm based on their motility and/or swim performance in a manner that resembles these same metrics in their natural environment. Sperm are usually measured while swimming in a biomimetic solution such as human tubal fluid, and devices that use rheotaxis mimic the natural flow of fluid in the female reproductive system.28 Some devices take this a step further and attempt to mimic a particular portion of the female reproductive system, such as devices that resemble the cervical opening or the oviduct.23

Many devices that separate sperm by motility distinguish between sperm that are motile and non-motile, but not degrees of swimming ability. A notable exception to this is a recent device that uses acoustofluidics to quantitatively separate faster and slower sperm and collect them into aliquots.19 Here, we wanted to design a simpler and cheaper approach to quantitative sperm separation by swimming ability, and developed a new device, the “sperm racetrack”. The sperm racetrack separates sperm quantitatively by swimming ability and allows us to extract them for further study. The ability to quantitatively separate the best swimming sperm from those that are outcompeted has useful applications in research and medicine.

Materials and methods

Design

We designed a ‘sperm racetrack’ device that uses a biology-inspired design to harness sperm's natural tendency to swim against counterflow (rheotaxis)29,30 and sort sperm by swimming speed. It contains a straight channel with evenly spaced extraction ports and applies counterflow from one end of the channel to induce rheotaxis in sperm. The device also includes a vacuum layer, positioned beneath the flow layer and separated by a silicone membrane, to remove bubbles that obstruct sperm movement and disrupt counterflow. All components are fabricated using polydimethylsiloxane (PDMS), a transparent and gas permeable silicone polymer commonly used in microfluidics.

The microfluidic device was designed in AutoCAD (Product Version V.58.M.214, AutoCAD 2025) (Fig. 2). The vacuum layer and flow layer are separated by a 100 μm silicone membrane. The flow layer holds the counterflow media and sperm sample and features a 6.5 cm long, 2 mm wide, straight channel. The channel height is approximately 0.076 mm. The channel height was determined by the thickness of tape used in fabrication. The channel width was chosen as a compromise between (1) a more even velocity profile and larger total sperm capacity (a larger channel) and (2) a more biomimetic width and lower chance of issues with bubbles (a smaller channel). Preliminary tests showed that a 5 mm wide channel could not reliably clear bubbles, so a 2 mm width was chosen. The thickness of the membrane was chosen to match existing devices that remove bubbles from a flow channel.31 The spacing of ports was a compromise between more fine-grained separation of swimmers and practical ease of use. We settled on seven extraction ports because each manual port extraction takes about one minute and we wanted to minimize handling time. We chose the spacing of the ports based on preliminary tests that showed sperm rarely swim more than 5 cm in 30 minutes.

Fig. 2. The two-layer design of the sperm racetrack. The flow layer and vacuum layer, shown here, are each molded out of PDMS and then are bonded together into a complete device with a silicone membrane between the layers. Small circles represent injection/extraction ports cut through the device. Left: the overhead view of the flow layer alone, with the counterflow port at the top of the diagram and the sperm injection port (E1) and waste port (W) at the bottom. Extraction ports are labeled E1 to E7. Middle: schematic of the vacuum layer. Cross shapes here represent solid PDMS columns that support the sandwiched membrane when vacuum is applied. The port to the left of the channel is the vacuum input port. Right: an overlay of the flow layer and vacuum layer.

Fig. 2

Open in a new tab

Seven extraction ports, labeled E1 through E7 in Fig. 2, are evenly spaced along the channel at 5 mm intervals. These ports separate sperm into six groups based on distance travelled. Solid 1.0 mm-diameter optical fiber plugs block these ports during operation but allow sample extraction afterward. The ports serve distinct functions (Fig. 2). At the top, the counterflow port (C) is attached to a pump that pushes media to the device to generate counterflow. The next six ports are extraction ports for sperm retrieval. The injection port (E1) is used to introduce a semen sample, and the waste port (W) removes excess media and immotile sperm. A 30 cc syringe attaches to the vacuum port to create the vacuum effect via a syringe lock. We were concerned that the membrane separating the vacuum and flow layers could become curved downward toward the vacuum when negative pressure was applied, which would change the cross-sectional profile of the flow channel and affect the flow in the device. To prevent this, we added cross-patterned ‘columns’ within the vacuum layer that prevent membrane deflection, ensuring consistent counterflow rates. The cross shape was chosen because it provides support across a large portion of the membrane while still allowing diffusion into the vacuum layer.

One complication in any pipe-like device is the velocity profile of fluid flow. The rectangular cross-section of the main flow channel in this device should produce a laminar flow with a parabolic velocity profile.32 Compared to a flat velocity profile, this design will have reduced velocity around the edges of the pipe. Sperm in these low-velocity regions may either fail to orient by rheotaxis, leading to a disadvantage in the “race”, or may orient correctly and have an advantage in that they are swimming against a slower countercurrent. It is, thus, difficult to predict the effects of the parabolic velocity profile, and we acknowledge it as a caveat that probably reduces the efficiency of sorting fast vs. slow sperm. The much higher width than height of the channel helps to alleviate some of these concerns by increasing the overall uniformity of the flow.

Fabrication

The device is fabricated using rapid prototyping and soft lithography. Broadly, molds for all layers are laser-cut and cast in polydimethylsiloxane (PDMS), then bonded through plasma etching, following standard soft lithography procedures. AutoCAD channel designs are laser-cut from 3M Scotchcal Marking Film (3M, product ID: SC 50-12 white; thickness 0.076 mm), a material analogous to electrical tape, and attached to square Petri dishes. PDMS is cast into these molds and cured overnight at 70 °C (Fig. 3). After curing, the PDMS is removed to expose the flow and vacuum layers. A 1.0 mm hole punch is used to create extraction ports in the flow layer. The channel is masked with laser-cut 3M Scotchcal Marking Film, precisely matching the dimensions of the flow channel. Plasma etching (3 minutes of corona plasma) is applied to both the PDMS-cast flow layer and one side of the 100 μm silicone membrane for bonding.

Fig. 3. Fabrication of the sperm racetrack with standard PDMS molding techniques. A. Fabrication process, from design through molding to bonding. The device was designed in AutoCAD. A computer-controlled laser cutter is used to cut molds out of tape, and then the molds are stuck to the bottom of petri dishes. PDMS is poured into the molds and cured, and then the PDMS layers are each bonded to a silicone membrane that is sandwiched between them. B. Fully bonded device seen from above (quarter for scale). The vacuum port is visible at the top portion of the image.

Fig. 3

Open in a new tab

After flow layer and membrane bonding, a 1.0 mm access port is punched through both the PDMS and silicone membrane to establish vacuum layer access. The vacuum layer is masked with laser-cut marking film, and both the vacuum layer and the opposing membrane surface undergo an additional 3 minute corona etching cycle for final bonding. The assembled device is placed in a 70 °C oven overnight to ensure sealing. The final device is shown in Fig. 3.

Clinical sample acquisition

All semen samples used in this experiment were obtained from adult males visiting the Utah Center for Reproductive Medicine. All samples were consented under an IRB-approved protocol at the University of Utah by Kenneth Aston in the Department of Andrology.

Race protocol

The racetrack attempts to replicate the counterflow of the human oviduct. Preliminary experiments helped determine that a counterflow rate of 0.092 μL min−1 induces rheotaxis in sperm. With a channel height of 0.076 mm and a channel width of 2 mm, this translates to a counterflow velocity of 10.08 μm s−1. Given an average swimming speed of 50 μm s−1,33 and subtracting the counterflow to give a net velocity of 39.92 μm s−1, sperm should traverse the 6.5 cm channel in approximately 27 minutes.

To prepare the device, we sealed the extraction and injection ports with 1.0 mm optical fiber segments. We connected the counterflow port via 1.0 mm PTFE tubing (Masterflex, microbore PTFE, 0.22″ ID, 0.042″ OD) to a syringe pump equipped with a 100 μL gastight glass syringe (model 1710 TLL, PTFE Luer Lock). We routed the waste port to a collection container using identical tubing. We then linked the vacuum layer to a 30 mL syringe (Global Medical Products, model 1202543, 30 CC Luer Lock) to generate vacuum pressure via the syringe lock.

To establish a physiologically suitable environment, we fill the device with human tubal fluid (HTF) supplemented with 1% bovine serum albumin (BSA) (Sigma Aldrich, MR-070-D), which reduces sperm adhesion to channel surfaces. After ensuring a bubble-free channel, we set the syringe pump to a constant counterflow of 0.092 μL min−1. We introduce 2.5 μL of semen sample through the injection port using a 10 μL micropipette, after which we reseal the port with optical fiber. Sperm are allowed to swim for 30 minutes, and the sperm race is monitored via a Nikon Eclipse TE300 Inverted Microscope with a 20× objective (Fig. 4).

Fig. 4. Experimental procedure. Media is injected from the left side to induce counterflow, and sperm is introduced to the device via the 1st port to the left of the waste port. Negative pressure is applied through the vacuum port using a 30 cc syringe. Sperm are filmed at the end of the 30 minute race. Each extraction port is blocked with a 1.0 mm optical fiber. The sorted sperm are extracted through infusion of HTF + 1% BSA via a syringe pump into the right port of a targeted segment. The extracted solution of sperm and media is collected through the left port of the targeted segment. Sperm are collected in turn from each segment after the race.

Fig. 4

Open in a new tab

After a run of the sperm racetrack, we proceed to sperm separation and collection. We record each device segment using an AmScope 4K Series 2160P Camera, along with pre-experiment control videos. We perform sequential extraction by introducing HTF + 1% BSA into each segment's first port, displacing sperm through the subsequent port into a cryotube (Fig. 4). Extraction proceeds from the waste segment through segment 6. We place a 2 μL aliquot from each segment on individual glass slides and image using the AmScope camera. Sperm counts are obtained via a Makler counting chamber. Extracted aliquots and control samples are stored at −80 °C for future analyses.

Some aspects of this device and its operation better model features of the female reproductive tract than others. The solution in which the sperm swim, HTF, is a close model of the solution found in real human fallopian tubes,34 and the average flow of HTF in the system resembles the flow in fallopian tubes as well.28 On the other hand, the human fallopian tube epithelium is a complex structure with many folds and is hardly modelled by the simple, linear tube used here. That said, the 10 cm length and 2 mm cross sectional area of the device are similar to those of the fallopian tube.35 Overall, the goal of this device's design is not to replicate all of the features of the fallopian tube, but to present a swimming challenge to the sperm that tests the ability of sperm to swim in conditions roughly similar to the fallopian tube.

Swimming characterization with CASA

All videos are analyzed using the Computer Assisted Sperm Analyzer (CASA) plugin in ImageJ v1.51k.36 CASA measures several aspects of sperm swimming through motion tracking. It measures several parameters, such as the sperm's curvilinear velocity (VCL), straight-line velocity (VSL), and average path velocity (VAP) (Fig. 5). All velocities are reported in μm s−1. The VCL describes the point-to-point velocity of the sperm per second. The VSL is calculated using the straight-line distance between the first point and the last point of a sperm track, and represents the straight-line distance that the sperm traveled. The VAP represents a smoothed version of the VCL, ignoring the side-to-side movements of the sperm head. CASA also provides derived statistics on sperm linearity, wobble level (sperm head movement side-to-side calculated by VAP/VCL), and the percent of sperm that are motile. Sperm linearity describes the amount of curvature in a sperm's swimming path, and is determined by dividing VSL/VAP.

Fig. 5. CASA sperm tracking statistics derive from the path traveled by the sperm. Here, the curved line represents the curvilinear (exact) path of the sperm, and the gentle arc is the average path. The direct statistics of curvilinear velocity (VCL, velocity along the raw path), average path velocity (VAP, velocity along the smoothed path), and straight line velocity (VSL, velocity directly from the beginning to the end of the path) are calculated based on the sperm's travel along these paths, and the derived statistics linearity (LIN, the straightness of the swimming path) and wobble (WOB, the side-to-side movement of the sperm head) are calculated from the direct statistics.

Fig. 5

Open in a new tab

All CASA analyses require some amount of filtering to discriminate sperm from debris. We chose filtering parameters based on visual tests of preliminary sperm video footage, checking that sperm were tracked and debris was not. Minimum sperm size was reduced to 0 pixels due to sperm occasionally being nearly lost during long track runs, while maximum sperm size was set to 600 pixels, which corresponds to a generous maximum circular radius of 43 pixels. Minimum track length was kept at 1 frame so as not to bias sperm detection based on track length. Maximum sperm velocity between frames was set to 15 pixels to prevent mis-tracking events where a sperm appears to jump across the view because two sperm that should be independently tracked are being tracked together. We set minimum VSL, VAP, and VCL scores for motile sperm at standardized parameters.36 Other parameters are either not related to filtering or were set to be fully permissive. To better distinguish sperm from debris, we added three new motion tracking algorithms to the CASA plugin. The first problem was an inability to discriminate between motionless debris and sperm. We filtered tracked objects by a minimum net distance traveled by the sperm in pixels over the whole video. This removes most stationary debris from the analysis. To avoid tracking slow-moving debris in videos that include counter current, we filtered by a minimum net distance traveled in pixels per frame of the video. Finally, we also added an option to print an associated ID for each individual tracked sperm. This allows other programs to use the output of CASA for per-sperm track analysis. This updated version of CASA, which we call CASA2, is available at https://github.com/jgbaldwinbrown/CASA2. This optimized plugin was run on videos thresholded to only include sperm heads. The parameters used in all analyses are provided in Table 1.

Table 1. CASA parameters used in all analyses.

Code Feature Units Value
A Minimum sperm size Pixels 0
B Maximum sperm size Pixels 600
C Minimum track length Frames 1
D Maximum sperm velocity between frames Pixels 15
E Minimum VSL for motile μm s−1 0.11
F Minimum VAP for motile μm s−1 0.11
G Minimum VCL for motile μm s−1 0.11
H Minimum net distance traveled Pixels 50
I Minimum net distance traveled Pixels per frame 2
J Low VAP speed μm s−1 0.11
K Maximum percentage of path with zero VAP % 100
L Maximum percentage of path with low VAP % 100
M Low VAP speed 2 μm s−1 0.11
N Low VCL speed μm s−1 0.11
O High WOB % VAP/VCL 100
P High LIN % VSL/VAP 100
Q High WOB two % VAP/VCL 100
R High LIN two % VSL/VAP 100
S Frame rate Frames per second 26
T Microns per 100 pixels μm per pixel 189.5
Open in a new tab

Measuring sperm distribution along the channel

We investigated the distribution of sperm after the race using Makler counting. We loaded 2 μL of the extracted aliquots onto a Makler Sperm Counting Chamber. The number of motile and non-motile sperm in each row of 10 squares was counted. These numbers correspond to millions of sperm per mL of sample.

Sperm vitality assays

We measured sperm vitality using a standard vitality assay based on staining dead sperm. We added equal volumes of sperm medium and eosin–nigrosin stain to a slide, mixed, and smeared. We then fixed with cytoseal and covered with a cover slip. We counted either all of the sperm we could find on a slide or at least 200 sperm, whichever came first. Nigrosin is a counterstain that does not enter cells, while eosin is a stain permeable to the membranes of dead cells but not living cells. We counted unstained (alive) cells and stained (dead) cells and reported these counts.

Statistical methods

All data is expressed as a mean ± standard deviation (SD). We used a linear model to determine the relationship between segment and sperm swimming parameters. The specific swimming characteristics tested included VSL, VAP, VCL, linearity, and wobble level. All linear models were of the form statistic ∼ segment, where statistic is VSL, VAP, or similar, and segment is the number-coded segment between two ports from which the data was collected. Segments are numbered starting immediately after the waste port, and start at index 0. All statistics were based only on the values in segments 2, 3, and 4, because no sperm swam into segments 5 and 6, and segments 0 and 1 were immediately adjacent to the sperm injection port (E1) and were filled with sperm that did not have rheotaxis induced. Although segments 0 and 1 are not very informative, we still present these in figures for completeness.

For sperm vitality, we tested for statistical differences using a generalized linear model with a logit linker function for binomial data. To test for a difference in vitality between the original semen sample and sperm extracted from the device, we built the following linear model using R's lme4 package:glmer(alive ∼ indevice + (1|Name), family = binomial)where alive is set to 1 for a living sperm and 0 for a dead sperm, “indevice” refers to a sperm from the device vs. a sperm from the original semen sample, and “(1|Name)” refers to the random effect of which semen sample the sperm was drawn from. We tested to see if the “indevice” factor had a significant effect. We also performed a conservative power analysis of the same data using R's pwr.chisq.test function with the following parameters:pwr. chisq. test(w = NULL, N = 2152, df = 1, sig. level = 0.5, power = 0.8)For testing the effect of position in the device on sperm vitality, we subsetted the data to only include sperm from the device and then fit the following model:glmer(alive ∼ segment + (1|Name), family = binomial)where segment is the numerically-coded position in the device (waste = 0, segment 1 = 1, segment 2 = 2, etc.). We then tested for an effect of segment.

Results

To test the sperm racetrack, we ran sperm samples from three males. Two of these samples were run once each on separate devices and the third sample was run three times on separate devices to test the reproducibility of the device. We tracked 1569 sperm across these six runs, or 262 sperm per run on average. After running the racetrack, we extracted sperm from each segment of the channel. We counted the number of motile and non-motile sperm in each segment using a Makler sperm counter (Fig. 6). Some sperm occupied the later segments of the device, indicating that sperm successfully swam upstream from the injection port. The number of sperm per segment decreased sequentially as we moved away from the injection port. Sperm swam as far as the fourth segment, leaving two more segments available for longer runs. One open question here is how non-motile sperm wound up in later segments from which they could be extracted. Possibilities include accidental spreading of sperm into the channel during initial injection, accidental mixing of channel segments during extraction, and loss of motility after swimming into a later segment. Our data does not distinguish between these hypotheses, but informal observations of the device immediately after injection of sperm seemed to show few non-motile sperm in the later channels, making it unlikely that they were mixed along the channel during sperm injection. We believe that non-motile sperm in the later channels are primarily due to loss of motility during the run. The design of the extraction protocol does not put a physical barrier between different channel segments. There are two factors that may lead to potential mixing between segments. First, eddy-like non-laminar flow during extraction can potentially lead to some sperm mixing. Second, sperm can swim in an undirected way after counterflow is stopped, which can lead to potential mixing. However, the fastest-swimming sperm found in the farther segments in the device are also the fastest sperm outside of the device post-extraction. These results indicate that potential mixing does not interfere with the ability to isolate the fastest swimming sperm. Overall, this counting data confirmed that the device induced swimming through rheotaxis, and that sperm actively swam into sequential segments of the channel.

Fig. 6. Sperm count after the race decreases as you look farther along the channel. Sperm were extracted from the device post-race, and sperm from each segment of the main channel were counted using a Makler. X-Axis: segment number, where 1 is the segment immediately adjacent to the sperm input port and 2+ are the segments between extraction ports. Y-Axis: millions of sperm per milliliter in the extracted aliquot. Non-motile sperm are those that did not move at all in the Makler, and motile are those that moved even slightly. The lower image shows the structure of the racetrack device, with each segment number in the plot lined up with its position in the device.

Fig. 6

Open in a new tab

Sperm that reach further segments of the racetrack should be better swimmers than those in earlier segments, reflecting variation in swim speed between sperm in a sample. That is, sperm in distal segments are expected to have a higher average swimming velocity than those found in proximal segments. After each run, we compared the straight-line velocity (VSL), the average path velocity (VAP), and the curvilinear velocity (VCL) of sperm in each segment of the channel (Fig. 5). We first measured in-device VSL, VAP, and VCL of each sperm using video analysis (Fig. 7). Our analysis revealed a significant correlation between distance traveled and VCL (p = 0.27 VAP; p = 0.70 VSL; p = 0.036 VCL; sperm tracked = 288; fdr-corrected p = 0.324 VAP; p = 0.700 VSL; p = 0.054 VCL). These results confirmed that the sperm racetrack successfully separated sperm by swim speed, at least according to VCL. We were surprised that the distance traveled was not significantly correlated with VAP and VSL. While VSL is only weakly correlated with VCL (r = 0.26), VAP is strongly correlated with VCL (r = 0.85). We also tested for a difference between the VCL of sperm in segment 2 vs. segment 4 by t-test, but this was insignificant as well (p = 0.155). We suspect the lack of significance in VSL and especially VAP is due to the small sample size in our test dataset.

Fig. 7. Left: Velocities of sperm in the sperm racetrack show that sperm that travel farther swim faster. A.–C. VAP, VSL, and VCL, each measured from videos taken of sperm swimming in the transparent device during a run. The overall upward trend as sperm travel along the device indicates that faster sperm traveled farther than slower sperm. Also included are the “pre-sort” sperm, a separately measured set of sperm from the original semen sample. The line of best fit here only corresponds to segments 2, 3, and 4 because segments 0 and 1 are immediately adjacent to the sperm injection port and were filled with sperm that did not travel up the channel by rheotaxis. Right: Velocities of sperm in extracted aliquots also show that sperm that travel farther swim faster. D.–F. VAP, VSL, and VCL, each measured from videos taken of sperm extracted in aliquots from each segment of the device. The overall upward trend as sperm travel along the device indicates that faster sperm traveled farther than slower sperm. Also included are the “pre-sort” sperm, a separately measured set of sperm from the original semen sample. The line of best fit here only corresponds to segments 2, 3, and 4 because segments 0 and 1 are immediately adjacent to the sperm injection port and were filled with sperm that did not travel up the channel by rheotaxis. Segments 0 and 1 are included for completeness.

Fig. 7

Open in a new tab

We wanted to know if the sorting of sperm observed in-device was faithfully maintained after extracting sperm from the device. We measured VAP, VSL, and VCL for each sperm after extraction using video analysis. We correlated extraction segment and swim speed. As with the in-device sperm, the extracted sperm showed a significant increase in all three velocity parameters as we move away from the injection port (Fig. 7, p = 6.14 × 10−8 VAP; p = 2.67 × 10−7 VSL; p = 6.69 × 10−7 VCL; sperm tracked = 36; fdr-corrected p = 3.68 × 10−7 VAP; p = 8.01 × 10−7 VSL; p = 1.338 × 10−6 VCL). We also tested for a difference between VCL in segment 2 vs. segment 4 with a t-test. This was significant (p = 1.71 × 10−5). This confirmed that the extraction procedure did not re-mix sperm, and faithfully maintained the sorting of sperm by swim speed seen in-device. Unlike the in-device measurements discussed above, these measurements of extracted sperm were significant on all three of our metrics, increasing our confidence in the device's ability to separate sperm by swim speed.

There are at least two ways sperm may swim farther. They may do so either by swimming faster or by swimming in a more linear fashion. To discriminate between these two possibilities, we measured the linearity (LIN) of each sperm from the same videos as in the velocity analyses (Fig. 8). Linearity was insignificantly correlated with position in the channel both in-device and post-extraction (p = 0.63 in-device (288 sperm), p = 0.48 extracted (36 sperm), fdr-corrected p = 0.63 in device, 0.63 extracted). These results show that swim speed, rather than linearity, is a dominant factor in winning the race.

Fig. 8. Sperm swimming linearity and head wobble do not contribute to distance traveled. A. Linearity (LIN) of sperm measured within the channel during a run shows that sperm that travel farther do not swim significantly straighter. B. Head wobble (WOB) of sperm in the channel shows that head displacement is not associated with farther travel along the channel. Axes and colors match those of Fig. 7. C. and D. same as A. and B, but for sperm extracted from the channel after the run. The line of best fit here only corresponds to segments 2, 3, and 4 because segments 0 and 1 are immediately adjacent to the sperm injection port and were filled with sperm that did not travel up the channel by rheotaxis. Segments 0 and 1 are included for completeness.

Fig. 8

Open in a new tab

In addition to linearity, sperm that move their heads side-to-side more vigorously have also been hypothesized to swim faster.37 We measured the wobble (WOB, side-to-side head movement) of each sperm from the same videos as in the velocity analyses (Fig. 8). Wobble was insignificantly correlated with sperm position in-device and slightly positively correlated with position post-extraction (p = 0.443 for in-device (288 sperm), and p = 0.00051 for extracted, (36 sperm); fdr-corrected p = 0.63 in device, 0.0020 extracted). These trend differences indicate that swim speed, rather than wobble, is a dominant factor in winning the race.

In summary, we show that the sperm racetrack effectively separates sperm based on swim speed during rheotaxis, and that the re-extraction procedure faithfully maintains the separation of sperm by swim speed.

Finally, we wanted to make sure that the device did not have a negative effect on the vitality of the sperm that passed through it. We thus tested the effect of the device on sperm vitality. We performed nine additional runs of the device looking only at sperm vitality using an eosin–nigrosin stain. We ran sperm from six individuals through the device. No device was reused; each run of the racetrack was with a newly manufactured device. Four sperm samples were run once each, one sperm sample was run twice, and another one was run three times, for a total of nine runs of the device. We counted a total of 9645 sperm across 53 slides with an average of 185 sperm counted per slide, and an average of 1072 sperm counted per racetrack run. Raw counts are included in supplementary data 1. A generalized linear model showed that the vitality of sperm from the device is significantly higher than the vitality of sperm in the original semen sample (p = 5.46 × 10−6, 80% power to detect a difference of 0.028). These results show that the sperm extracted from the selected segments of the device have significantly higher vitality than the input sample. We also wondered if the device might separate sperm according to health, such that sperm that travel farther along the device are more likely to still be alive when the vitality assay is performed. We found a significant positive effect of position in the device on sperm vitality (p = 0.00985, Fig. 9). This indicates that in the process of separating sperm according to their swimming ability, we also separate them according to overall vitality.

Fig. 9. The sperm racetrack does not reduce vitality, and selecting for further distance traveled also selects for high vitality. The X-axis here represents the type of sample, where “pure” is the original semen sample and “waste” or a number represent the segment from the device from which the aliquot was extracted. “Sample” is the particular semen sample measured. The colored points represent the sperm vitality (living sperm over total) from an eosin–nigrosin vitality assay. The black line is the overall vitality based on all counted sperm for a particular aliquot type, and the blue fill is the 95% confidence interval by the Clopper–Pearson method.

Fig. 9

Open in a new tab

Discussion

We designed, fabricated, and tested the sperm racetrack, a biologically-inspired microfluidic platform that systematically sorts sperm in a semen sample by swimming ability, then allows faithful extraction of the sorted sperm for further use. In keeping with this, we find a strong and significant correlation between all of our velocity measures and the distance traversed. Sperm that swim faster in the sperm racetrack do not wobble their heads more or swim straighter than slow sperm.

Although we directly measure several swimming parameters from video of the sperm in motion, more parameters are known to influence the swimming success of sperm. For example, persistence length (the distance that sperm travel before changing direction) has been associated with overall better swimming,12 but we do not measure that here. Other sperm phenotypes that are associated with better motility are sperm vitality and DNA integrity.27 Future studies could contrast swimming distance with persistence length and other sperm features to better characterize the features of high-quality sperm.

The racetrack itself is modular enough that new features could be added to it in the future. For example, one advantage of the acoustofluidic approach in Chinnasamy et al. 2018 (ref. 12) is the ability to continuously collect sperm from an outlet at one end of the device throughout a run. The sperm racetrack could be modified to similarly continuously collect sperm that reach the “finish line”, allowing for more flexible collection of sperm into finer-grained aliquots.

We originally designed this device to enable the detection of segregation distorters that operate by manipulating sperm swim speed in humans. To detect distorters, we plan to separate sperm by swim speed and then Illumina sequence the separated aliquots of sperm.38 If a distorter is present, we expect the distorting haplotype to be overrepresented in the winning segment of the sperm racetrack as compared to the starting pool. Sequencing separated fast and slow sperm from many independent individuals may open the door to the discovery of such selfish chromosomes in human populations. While the discovery of the mouse t-locus was serendipitous, the sperm racetrack may enable a systematic screen for such elements in human genomes.

Going beyond the study of distorters, the racetrack may be useful for studying the biology of sperm competition in humans. Sperm competition influences mating success in many organisms,39 but most studies quantify sperm competition by measuring the paternity of offspring, and not by directly measuring sperm performance. Beyond scientific interest, there is also general public interest in the concept of sperm competition (https://www.spermracing.com/). Because the sperm racetrack is optically transparent, staining different sperm samples with different dyes and tracking them optically provides a feasible path to study sperm competition directly.

Perhaps the most widely applicable use for the sperm racetrack is in the facilitation of assisted reproductive technologies (ART). Widely-used techniques for enriching high-quality sperm for ART, such as the swim-up method and density gradient centrifugation,40 are susceptible to cell damage and sperm dysfunction, and may inhibit the efficacy of ART.41,42 Existing swimming-based devices devices13,15–17,20–22 sort motile sperm from non-motile sperm with minimal damage, but do not allow separation and extraction based on swim speed. The sperm racetrack can directly isolate the fastest sperm in a sample to improve the success of ART. Finally, it has long been hypothesized that the mutational load of sperm may correlate with swimming ability, such that sperm carrying a high mutational load are less likely to fertilize an egg.43 The sperm racetrack may thus provide a method for identifying faster sperm that carry lower mutational loads, and are thus more appropriate for ART.

To conclude, we have designed, manufactured, and tested the sperm racetrack, a device that separates sperm based on swim speed and allows extraction of fast and slow sperm. This platform opens the door to investigating phenomena such as segregation distortion and sperm competition, and also for use in assisted reproductive technologies.

Ethical statement

All experiments were performed in accordance with The Common Rule (45 CFR 46), and experiments were reviewed and approved by the Institutional Review Board at the University of Utah (IRB_00093004). Informed consent was obtained from all human participants of this study (IRB_00012049).

Author contributions

Conceptualization: JGB, NP, BG, US. Resources: NP, BG, KA. Data curation: JGB. Software: JGB. Formal analysis: JGB, LH. Supervision: NP, BG. Funding acquisition: NP. Validation: JGB, LH. Visualization: JGB, LH. Investigation: JGB, LH, NP, US, SN, AU, BG. Writing – original draft: JGB, NP, LH. Writing – review & editing: JGB, NP, LH, BG. Methodology: JGB, LH, US, AU, SN, BG, NP. Project administration: NP.

Conflicts of interest

The microfluidic device described here is protected under provisional US patent U-8401 “ MICROFLUIDIC DEVICE AND METHOD FOR TESTING SPERM”.

Supplementary Material

LC-026-D6LC00399K-s001

Acknowledgments

We acknowledge NIH grants R01GM141422 and R35GM156267, awarded to Nitin Phadnis, for funding this research. We also acknowledge Lauren Hulse and Nitin Phadnis for Hulse's undergraduate thesis,44 which formed the initial basis of this work.

Data availability

All CASA results data is included in supplementary information (SI). All video data is available upon request owing to the large file sizes. All software is available at https://github.com/jgbaldwinbrown/jgbutils and https://github.com/jgbaldwinbrown/CASA2.

Supplementary information: supplementary data table 1 contains counts of living and dead sperm for the vitality assay above. See DOI: https://doi.org/10.1039/d6lc00399k.

Notes and references

  1. Verspoor R. L. Price T. A. R. Wedell N. Selfish genetic elements and male fertility. Philos. Trans. R. Soc., B. 2020;375(1813):20200067. doi: 10.1098/rstb.2020.0067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Phadnis N. Orr H. A. A single gene causes both male sterility and segregation distortion in Drosophila hybrids. Science. 2009;323(5912):376–379. doi: 10.1126/science.1163934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Winkler L. Lindholm A. K. A meiotic driver alters sperm form and function in house mice: a possible example of spite. Chromosome Res. 2022;30(2–3):151–164. doi: 10.1007/s10577-022-09695-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Tessler S. Carey J. E. Olds-Clarke P. Mouse sperm motility affected by factors in the T/t complex. J. Exp. Zool. 1981;217(2):277–285. doi: 10.1002/jez.1402170214. [DOI] [PubMed] [Google Scholar]
  5. Chesley P. Dunn L. C. The inheritance of taillessness (anury) in the house mouse. Genetics. 1936;21(5):525–536. doi: 10.1093/genetics/21.5.525. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Styrna J. Klein J. Evidence for two regions in the mouse t complex controlling transmission ratios. Genet. Res. 1981;38(3):315–325. doi: 10.1017/S0016672300020632. [DOI] [PubMed] [Google Scholar]
  7. Lee M. Mouse t haplotypes. Annu. Rev. Genet. 1985;19:179–208. doi: 10.1146/annurev.ge.19.120185.001143. [DOI] [PubMed] [Google Scholar]
  8. Olds-Clarke P. Models for male infertility: the t haplotypes. Rev. Reprod. 1997;2(3):157–164. doi: 10.1530/ror.0.0020157. [DOI] [PubMed] [Google Scholar]
  9. Karavolos S. Sperm DNA fragmentation. Semin. Reprod. Med. 2021;39(5-06):194–199. doi: 10.1055/s-0041-1736261. [DOI] [PubMed] [Google Scholar]
  10. Rodprasert W. Toppari J. Virtanen H. E. Environmental toxicants and male fertility. Best Pract. Res. Clin. Obstet. Gynaecol. 2023;86(102298):102298. doi: 10.1016/j.bpobgyn.2022.102298. [DOI] [PubMed] [Google Scholar]
  11. Sciorio R. Greco P. F. Greco E. Tramontano L. Elshaer F. M. Fleming S. Potential effects of environmental toxicants on sperm quality and potential risk for fertility in humans. Front. Endocrinol. 2025;16:1545593. doi: 10.3389/fendo.2025.1545593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chinnasamy T. Kingsley J. L. Inci F. et al., Guidance and self-sorting of active swimmers: 3D periodic arrays increase persistence length of human sperm selecting for the fittest. Adv. Sci. 2018;5(2):1700531. doi: 10.1002/advs.201700531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cho B. S. Schuster T. G. Zhu X. Chang D. Smith G. D. Takayama S. Passively driven integrated microfluidic system for separation of motile sperm. Anal. Chem. 2003;75(7):1671–1675. doi: 10.1021/ac020579e. [DOI] [PubMed] [Google Scholar]
  14. El-Sherry T. M. Abdel-Ghani M. A. Abdel Hafez H. K. Abdelgawad M. Rheotaxis of sperm in fertile and infertile men. Syst. Biol. Reprod. Med. 2023;69(1):57–63. doi: 10.1080/19396368.2022.2141154. [DOI] [PubMed] [Google Scholar]
  15. Huang H. Y. Wu T. L. Huang H. R. et al., Isolation of motile spermatozoa with a microfluidic chip having a surface-modified microchannel. J. Lab. Autom. 2014;19(1):91–99. doi: 10.1177/2211068213486650. [DOI] [PubMed] [Google Scholar]
  16. Jafek A. Feng H. Brady H. et al., An automated instrument for intrauterine insemination sperm preparation. Sci. Rep. 2020;10(1):21385. doi: 10.1038/s41598-020-78390-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Knowlton S. M. Sadasivam M. Tasoglu S. Microfluidics for sperm research. Trends Biotechnol. 2015;33(4):221–229. doi: 10.1016/j.tibtech.2015.01.005. [DOI] [PubMed] [Google Scholar]
  18. Nosrati R. Graham P. J. Zhang B. et al., Microfluidics for sperm analysis and selection. Nat. Rev. Urol. 2017;14(12):707–730. doi: 10.1038/nrurol.2017.175. [DOI] [PubMed] [Google Scholar]
  19. Misko V. R. Baraban L. Makarov D. et al., Selecting active matter according to motility in an acoustofluidic setup: self-propelled particles and sperm cells. Soft Matter. 2023;19(44):8635–8648. doi: 10.1039/D3SM01214J. [DOI] [PubMed] [Google Scholar]
  20. Samuel R. Feng H. Jafek A. Despain D. Jenkins T. Gale B. Microfluidic-based sperm sorting & analysis for treatment of male infertility. Transl. Androl. Urol. 2018;7(Suppl 3):S336–S347. doi: 10.21037/tau.2018.05.08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Sarbandi I. R. Lesani A. Moghimi Zand M. Nosrati R. Rheotaxis-based sperm separation using a biomimicry microfluidic device. Sci. Rep. 2021;11(1):18327. doi: 10.1038/s41598-021-97602-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Sharma S. Kabir M. A. Asghar W. Selection of healthy sperm based on positive rheotaxis using a microfluidic device. Analyst. 2022;147(8):1589–1597. doi: 10.1039/D1AN02311J. [DOI] [PubMed] [Google Scholar]
  23. Suarez S. S. Wu M. Microfluidic devices for the study of sperm migration. Mol. Hum. Reprod. 2017;23(4):227–234. doi: 10.1093/molehr/gaw039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Bouloorchi Tabalvandani M. Javadizadeh S. Badieirostami M. Bio-inspired progressive motile sperm separation using joint rheotaxis and boundary-following behavior. Lab Chip. 2024;24(6):1636–1647. doi: 10.1039/D3LC00893B. [DOI] [PubMed] [Google Scholar]
  25. Yaghoobi M. Abdelhady A. Favakeh A. et al., Faster sperm selected by rheotaxis leads to superior early embryonic development in vitro. Lab Chip. 2024;24(2):210–223. doi: 10.1039/D3LC00737E. [DOI] [PubMed] [Google Scholar]
  26. Yazdan Parast F. O'Bryan M. K. Nosrati R. Sperm syringe: 3D sorting platform for assisted reproduction. Adv. Mater. Technol. 2022;7(9):2101291. doi: 10.1002/admt.202101291. [DOI] [Google Scholar]
  27. Nosrati R. Vollmer M. Eamer L. et al., Rapid selection of sperm with high DNA integrity. Lab Chip. 2014;14(6):1142–1150. doi: 10.1039/C3LC51254A. [DOI] [PubMed] [Google Scholar]
  28. Hunter R. H. F. Components of oviduct physiology in eutherian mammals. Biol. Rev. Cambridge Philos. Soc. 2012;87(1):244–255. doi: 10.1111/j.1469-185X.2011.00196.x. [DOI] [PubMed] [Google Scholar]
  29. Zhang Z. Liu J. Meriano J. et al., Human sperm rheotaxis: a passive physical process. Sci. Rep. 2016;6(1):23553. doi: 10.1038/srep23553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kantsler V. Dunkel J. Blayney M. Goldstein R. E. Rheotaxis facilitates upstream navigation of mammalian sperm cells. eLife. 2014;3:e02403. doi: 10.7554/eLife.02403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kang J. H. Kim Y. C. Park J. K. Analysis of pressure-driven air bubble elimination in a microfluidic device. Lab Chip. 2008;8(1):176–178. doi: 10.1039/B712672G. [DOI] [PubMed] [Google Scholar]
  32. Schetz J. A. and Fuhs A. E., Fundamentals of fluid mechanics, 1999, https://books.google.com/books?hl=en&lr=&id=YCSSolzuu9IC&oi=fnd&pg=PR13&dq=Joseph+A.+Schetz,+Allen+E.+Fuhs.+Fundamentals+of+fluid+mechanics.&ots=LQBX4KZKPK&sig=-0my_BNghokcdjXaZpi3l8haoRM [Google Scholar]
  33. Carlson B. M., The human body: linking structure and function, 2025, https://books.google.com/books?hl=en&lr=&id=U8gmEQAAQBAJ&oi=fnd&pg=PP1&dq=Bruce+M.+Carlson.+The+Human+Body:+Linking+Structure+and+Function.+Academic+Press,+October+2018.+Google-Books-ID:+twUsDgAAQBAJ.&ots=z4u_5_bXsL&sig=zmLC2RVcv1HYlisrGIrLx-YX9KQ [Google Scholar]
  34. Tay J. I. Rutherford A. J. Killick S. R. Maguiness S. D. Partridge R. J. Leese H. J. Human tubal fluid: production, nutrient composition and response to adrenergic agents. Hum. Reprod. 1997;12(11):2451–2456. doi: 10.1093/humrep/12.11.2451. [DOI] [PubMed] [Google Scholar]
  35. Standring S. and Anatomy G. H. G., The anatomical basis of clinical practice, Gray's anatomy, 2008 [Google Scholar]
  36. Wilson-Leedy J. G. Ingermann R. L. Development of a novel CASA system based on open source software for characterization of zebrafish sperm motility parameters. Theriogenology. 2007;67(3):661–672. doi: 10.1016/j.theriogenology.2006.10.003. [DOI] [PubMed] [Google Scholar]
  37. Wolf J. P. Feneux D. Ducot B. Rodrigues D. Jouannet P. Influence of sperm movement parameters on human sperm–oolemma fusion. Reproduction. 1995;105(2):185–192. doi: 10.1530/jrf.0.1050185. [DOI] [PubMed] [Google Scholar]
  38. Shen R. Fan J. B. Campbell D. et al., High-throughput SNP genotyping on universal bead arrays. Mutat. Res. 2005;573(1–2):70–82. doi: 10.1016/j.mrfmmm.2004.07.022. [DOI] [PubMed] [Google Scholar]
  39. Wigby S. Chapman T. Sperm competition. Curr. Biol. 2004;14(3):R100–R102. doi: 10.1016/j.cub.2004.01.013. [DOI] [PubMed] [Google Scholar]
  40. Gupta S. and Kumar A., The Human Semen, in Basics of Human Andrology, Springer Singapore, 2017, pp. 163 –170 [Google Scholar]
  41. Zini A. Finelli A. Phang D. Jarvi K. Influence of semen processing technique on human sperm DNA integrity. Urology. 2000;56(6):1081–1084. doi: 10.1016/S0090-4295(00)00770-6. [DOI] [PubMed] [Google Scholar]
  42. Henkel R. R. Schill W. B. Sperm preparation for ART. Reprod. Biol. Endocrinol. 2003;1(1):108. doi: 10.1186/1477-7827-1-108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Walsh C. B. McGuigan K. Do slower movers have lower reproductive success and higher mutation load? Evol. Lett. 2018;2(6):590–598. doi: 10.1002/evl3.87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Hulse L. and Phadnis N., The sperm racetrack: A novel microfluidic method for motility-based sperm separation, in RANGE: Undergraduate Research Journal, University of Utah, 2025 [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

LC-026-D6LC00399K-s001
LC-026-D6LC00399K-s001.xlsx (12.7KB, xlsx)

Data Availability Statement

All CASA results data is included in supplementary information (SI). All video data is available upon request owing to the large file sizes. All software is available at https://github.com/jgbaldwinbrown/jgbutils and https://github.com/jgbaldwinbrown/CASA2.

Supplementary information: supplementary data table 1 contains counts of living and dead sperm for the vitality assay above. See DOI: https://doi.org/10.1039/d6lc00399k.

Articles from Lab on a Chip are provided here courtesy of Royal Society of Chemistry

RESOURCES