Isolation of sperm cells from leukocytes by the MDDS device
We utilized the MDDS device32 to isolate sperm cells from leukocytes, and Fig. 1a shows the channel configuration of the MDDS device and a schematic diagram of the sperm cell isolation process in the MDDS device. The MDDS device is a new type of spiral microfluidic device that our group recently developed32, which is composed of two sequentially connected spiral channels having different dimensions. The first spiral channel was designed to have a relatively smaller channel dimension than the second spiral channel to generate a stronger inertial lift force to focus all the input particles into the inner wall side of the channel more effectively (sample focusing); despite the smaller channel dimension, the lift force could not be strong enough for the particles smaller than a specific size, resulting in no focusing or their focusing in the middle of the channel instead of the inner wall side. In the second spiral channel, due to the increased channel dimension, the magnitude of the lift force applied to the particles decreases, and particles move to their equilibrium locations determined by the balance between inertial lift and Dean drag forces, resulting in their separation by size (sample separation). Additionally, to achieve higher separation resolution, the second spiral channel was designed to have a trapezoidal cross-section where a stronger Dean vortex is generated at the outer half of the channel, which leads to more effective expulsion of small particles toward the outer wall without affecting the focusing position of large particles near to the inner wall, resulting in a greater difference between their equilibrium positions33,34.
Overview of the semen preparation process by the MDDS device. (a) Channel configuration (green: the first spiral channel with smaller dimension, yellow: the second spiral channel with larger trapezoidal dimension) and schematic diagram of the operation process; the first spiral channel has a rectangular cross-section with 800 μm in width and 60 μm in height, and the second spiral channel was designed to have trapezoidal cross-section for the effective particle separation with 800 μm in width and 80 and 120 μm in height for the inner wall side and the outer wall side, respectively. Trajectories of DAPI-stained sperm cells (represented in green color for better recognition) and 10-μm red-fluorescence beads (mimicking leukocytes, represented in red color) at (b) the S-shaped transition region and (c) the bifurcation region of the MDDS device under the optimal flow rate condition, 2.0 mL/min (scale bar: 200 μm). (d) Recoveries of sperm cells, 10-μm beads, and debris from the outer wall outlet depending on various input flow rate conditions. IW, inner wall; OW, outer wall; WBC, white blood cell or leukocyte.
To isolate the sperm cell from leukocytes (or white blood cells) and seminal fluids, we utilized the same MDDS device previously used for leukocyte separation from erythrocytes in human peripheral blood, the details of which can be found elsewhere32. The human sperm cell is composed of an ellipsoid head (~ 5 µm in length, ~ 3 µm in width) and a long flexible tail (30–49 µm in length), and in some previous works26,36,37, the human sperm cell was assumed to behave as a rotating sphere, with an effective diameter of ~ 5 μm. To investigate the effective size of the human sperm cell and its behavior in the MDDS device, we performed experiments with various fluorescent beads in the diameter range of 2–10 μm as well as sperm and PBMCs and monitored their trajectories at various flow rates in the device, which has the same channel dimensions and configuration with the original device but has 5 outlets for more detailed output analysis (Supplementary Fig. S1a). From these results, we found that the sperm actually behave like spheres in the diameter range of 3–6 μm in the outlet bifurcation region of the MDDS device, and their best focusing and separation from PBMCs can be achieved at a flow rate of 2 mL/min in the MDDS device. Also, behaviors of PBMCs and 10-μm beads at the end of the second channel in the MDDS device were comparable, although leukocyte size significantly varies depending on the subtype (Supplementary Fig. S1b,c)26,38. For further experiments, 10-μm beads were used to mimic leukocytes (the leukocyte concentration in normal semen is < 1%, so the addition of beads helps visualize the separation effect). To verify our observation of bead and cell behavior, we mixed sperm cells and 10-µm beads, as shown in Fig. 1b,c the trajectories of sperm cells (green) and 10-μm beads (red) show that both sperm cells and 10-μm beads formed their own focused bands in the different locations of the first channel. 10-μm beads were tightly focused into the inner wall side while sperm cells were less tightly focused, forming a band in the middle of the channel closer to the outer wall. Sperm cells achieved a lower focusing behavior as expected in the first spiral channel, but we found that the initial focusing in the first spiral channel greatly reduced the particle dispersion and helped the 10-μm beads and sperm cells move to their own equilibrium positions with tightly focused bands in the second spiral channel, with sperm focusing on the outer wall side and 10-μm beads focusing at the inner wall side. This resulted in increased separation efficiency compared with the conventional single spiral channel device (Fig. 1a–c vs. Supplementary Fig. S2a,b). As shown in Supplementary Fig. S3, we can obtain highly purified sperm cells by removing 10-μm beads representing leukocytes from the outer wall outlet of the MDDS device. Figure 1d shows the recoveries of sperm cells, 10-μm beads, and debris from the outer wall outlet under various input flow rate conditions. With a flow rate of 2.0 mL/min, sperm cell recovery was > 90%, while the removal rate of 10-μm beads was > 98%. In the case of other debris, which is defined as a separate population having a lower size/density compared to the sperm cells in flow cytometry analysis (Supplementary Fig. S4), the resulting lift force in the spiral channel is not strong enough to focus them to a certain equilibrium position within a focused band, leading to a more even distribution in each outlet. Despite the relatively lower focusing behavior, the flow cytometry analysis of Fig. 1d shows that debris recovery from the outer wall outlet is significantly lower (~ 40% recovery) than the inner wall outlet, which represents that the device also works for debris removal although it is not good enough.
Sample-dependent performance variation of the MDDS device
Seminal fluid is composed of secretions from male reproductive organs, including the testes, seminal vesicles, prostate, and bulbourethral gland. Therefore, semen’s fluidic properties (e.g., viscosity) are significantly influenced by individual differences, temporary health conditions, and even ingested foods and water intake39. To demonstrate how the change of seminal fluidic properties affects the separation performance of the MDDS device, we observed focusing of the sperm cells and beads from clinical (raw) semen samples (not washed seminal cells or washed semen samples) with various viscosity and dilution conditions in the MDDS device. The device used here has the same channel dimensions and configuration as the original device but has 5 outlets for more detailed output analysis (Fig. 2a–c, Supplementary Fig. S1a). Based on the results that the sperm have similar trajectories with spheres in the diameter range of 3–6 μm, we spiked 6-μm beads into the diluted semen sample to indirectly observe the movement of the sperm cells instead of staining the sperm cells, which could affect cell viability and morphology37,40. 10-μm beads were added to mimic leukocytes32,35, and modified human tubal fluid (mHTF) containing 5-mg/mL BSA was used as a buffer solution for sample dilution. The semen viscosity was classified based on World Health Organization (WHO) guidelines by its thread length (≤ 2 cm for low viscosity, 2–4 cm for intermediate viscosity, and > 4 cm for high viscosity)41. Semen hyperviscosity is not a rare event, an estimated 12–29% occurrence has been reported in men42. Therefore, high-viscosity semen should be addressable by a device to be relevant for ART. As shown in the trajectories and the recovery graphs of Fig. 2a–c, sperm cells, and 6-μm beads had similar trajectories regardless of the viscosity and dilution conditions, where the sperm cells were focused to the right-half side in the second spiral channel of the MDDS device so that the majority of them were collected from Outlet #4 and #5. On the other hand, the donor-specific viscosity condition affected the focal positions of 10-μm beads in the MDDS device. As shown in Fig. 2a–c, the trajectories of 10-μm beads moved from the inner wall side to the outer wall side as the semen viscosity increased. Many factors can cause increased semen viscosity, but the exact changes in composition leading to high semen viscosity have not been explicitly identified, and the viscosity may not be homogenous throughout the sample, potentially leading to the formation of an irregular gel-like matrix. However, based on the observation of the microbead behavior, we can assume that the components leading to increased viscosity of the seminal fluids were focused into the inner wall side, preventing the 10-μm beads from achieving their expected equilibrium positions and forcing them towards the outer wall side. Viscosity effects can be reduced by increasing semen dilution mitigating the influence on inertial focusing of particles in the MDDS device. As shown in Fig. 2a–c, better inertial focusing of 10-μm beads was achieved in the higher sample dilution condition (right figures) compared to the lower dilution condition (left figures), although the dilution conditions we tested are still not sufficient to bring back the original trajectory of 10-μm beads and remove them from sperm cells in the samples with an intermediate viscosity and a high viscosity.
Sample-dependent device performance. (a–c) Trajectories of 6-μm (green) and 10-μm (red) beads spiked in semen samples having different viscosity conditions (low-, intermediate-, and high-viscosity condition, respectively) with the low and high dilution conditions under the optimal flow rate condition, 2.0 mL/min (scale bar: 200 μm), and graphs showing corresponding recoveries of sperm cells and 6- and 10-μm beads from each outlet; 10-μm beads were used to mimic leukocytes, and 6-μm beads were used to indirectly observe the movement of the sperm cells without staining the sperm cells themselves; viscosity of semen was determined based on WHO guidelines by the thread length when dripping the semen sample (≤ 2 cm for low viscosity, 2–4 cm for intermediate viscosity, and > 4 cm for high viscosity)41; mHTF containing 5-mg/mL BSA was used as a buffer solution for sample dilution.
Sample-independent semen preparation by a recirculation platform
Although the viscosity effects can be mitigated by increasing semen dilution, the higher dilution of samples has the drawback of a low-abundant sperm cell output which is not acceptable for ARTs. To avoid this drawback, we recirculated the diluted semen sample to the MDDS device, removing the viscosity-inducing components by recirculation of the outer side sample. Previously, our group developed a recirculation platform based on a check-valve system, where a unidirectional flow is achieved by an internal membrane32. A cell solution output from the MDDS separation can be extracted back into the input syringe by the withdrawal motion of a syringe pump and processed again through the same MDDS device repeatedly. This mechanism allows the preparation of samples in large volumes, e.g., 50 mL, to be processed and reduced to a highly purified and concentrated output of 1–2 mL through a series of recirculation. In addition to further purification and concentration, we can minimize sample dependency of separation performance by the recirculation process. As discussed in the previous section, 10-μm beads (representing leukocytes) in the high-viscosity sample were not properly removed from sperm cells during the first circulation process, which might be caused by focusing of the highly viscous, gel-like portions of the semen to the inner wall side. In the recirculation process, a cell solution output from the outer wall outlet is recirculated to the MDDS device, while the output from the inner wall outlet is collected in a waste reservoir. Therefore, from the recirculation process, we can achieve gradual disruption and removal of any seminal fluid material accumulating on the inner wall side so that the viscosity effect can be mitigated, and adequate inertial focusing and removal of leukocytes can be achieved eventually in the remaining cycles.
To increase throughput, we utilized a quad-version device containing four MDDS channels in parallel (Fig. 3a,b). As the sperm cells are collected from the outer wall outlet, the channel configuration of the 3D-printed connector was designed to recirculate and re-process the outer wall output (Fig. 3c and Supplementary Fig. S5). The 3D-printed connector enables direct connection of the MDDS device with syringes (for input and output reservoirs) and check-valves with minimizing undesirable tubing-related dead volumes.
Sample-independent semen preparation process by the recirculation platform. (a) Channel configuration and (b) a photo of the quad-version MDDS device (scale bar: 2 cm). (c) A schematic diagram of the check-valve-based recirculation platform. Recoveries of sperm cells, 10-μm beads, and debris from the outer wall outlet with the 4 cycles of recirculation for the (d) low- and (e) high-viscosity semen samples under the optimal flow rate condition, 2.0 × 4 = 8.0 mL/min. Trajectories of 6-μm (green) and 10-μm (red) beads spiked in the high-viscosity semen sample during the (f) first and (g) forth recirculation cycles (scale bar: 200 μm). (h) Sperm concentration factor compared to the diluted semen sample prior to the sperm isolation. Comparison of operational performances with DGC; (i) recovery of the sperm cell, (j) portion of the motile sperm, and (k) portion of the dead sperm. The values in graphs of (h–k) are expressed as the mean ± SD (n = 2–9). WBC, white blood cell; MDDS device, multi-dimensional double spiral device; DGC, density gradient centrifugation.
As shown in the Fig. 3d, for the low-viscosity semen sample (with 50 × dilution), 10-μm beads were effectively removed in the first cycle, allowing for recovery of highly purified sperm cells obtained from a total of four cycles of recirculation within 15 min (> 99.95% of 10-μm bead removal and ~ 82% of sperm cell recovery under the optimal flow rate condition, 2.0 × 4 = 8.0 mL/min). On the other hand, in the case of the high-viscosity semen sample (with 50 × dilution), the majority of 10-μm beads (~ 85%) was collected with sperm cells from the outer wall outlet due to the confounding semen secretions in the first separation cycle (Fig. 3e,f). However, the inertial focusing of 10-μm beads returned to the expected behavior as recirculation proceeded due to the removal of the viscous material from the inner wall side outlet in the previous cycles (Fig. 3g). As a result, from the four cycles of recirculation, > 98% of 10-μm beads were removed while keeping sperm cells with a slightly higher recovery rate (~ 87%). Although the removal rate of 10-μm beads is relatively lower for the high-viscosity sample than the low-viscosity sample, the results showed that the recirculation platform could be used as a standardized semen preparation to overcome the variation of fluidic properties caused by the heterogeneity of semen samples.
With each recirculation-separation cycle, sperm cells are recovered into approximately half the input volume, increasing sperm concentration. Considering the variation in the number of sperm cells depending on individual donors, which is in the range of 15–250 × 106 cells/mL for normozoospermic men43, we calculated the concentration factor (ratio of the sperm number in the output after separation to the sperm number in the input before separation) instead of the sperm number itself. The results show that sperm cells in the output of ~ 3 mL were approximately tenfold concentrated compared to the input sample of 50 mL within 15 min from 4 recirculation cycles by the recirculation platform having one quad-version MDDS device (1 × MDDS platform) (Fig. 3h). We also tested another recirculation platform having two quad-version MDDS devices (2 × MDDS platform), i.e., eight MDDS devices, where we can achieve higher throughput (faster operation), but the available number of recirculation cycles is reduced from 4 to 3 due to the increased dead volume compared to the 1 × MDDS platform. As a result, due to fewer recirculation cycles, the 2 × MDDS platform brings relatively lower purification and decreased concentrated factor (approximately fivefold), although the operation time was significantly reduced to < 8.5 min, which is much more rapid than the conventional methods (~ 1 h).
The MDDS operation was compared with the DGC method, which is the most standard semen preparation method for ART, in terms of (1) sperm recovery, (2) motility change, and (3) cell viability (Fig. 3i–k). The results from the two MDDS platforms (1 × and 2 × MDDS platforms) were combined into a single plot labeled “MDDS” because there was no significant difference between them in terms of recovery and performance. As shown in Fig. 3i, MDDS shows much higher sperm cell recovery rates overall (approximately 2.7 times higher) than DGC. DGC utilizes sperm density affected by its motility and morphology to isolate just the mature sperm with high motility from a population, whereas MDDS performs sperm isolation in a passive manner by size-based separation utilizing inertial focusing in the device. Therefore, DGC results in low cell number recovery, but the recovered cells are classified as mainly viable (DAPI exclusion) and motile (Fig. 3j,k). In contrast, MDDS provides a high sperm cell recovery with no significant change in overall sperm cell motility and viability compared to the starting sample. The results are in line with the analysis in the previous works that showed no damage induced by the inertial microfluidic separation despite its high flow rate nature26,27.
