RBMOnline - Vol 7. No 1. 75 81 Reproductive BioMedicine Online; www.rbmonline.com/article/847 on web 16 April 2003 Article Isolation of motile spermatozoa from semen samples using microfluidics Dr Schuster completed his medical school training at Loyola University, Chicago Stritch School of Medicine in 1998. He is currently in his fifth year of urology residency training at the University of Michigan. He recently completed a year of basic science research in male infertility studying sperm sorting techniques as well as methods for freezing oligozoospermic samples with cryoloops. Dr Timothy Schuster Timothy G Schuster 1, Brenda Cho 2, Laura M Keller 3, Shuichi Takayama 2,5, Gary D Smith 1,3,4,6 1 Department of Urology, 2 Department of Biomedical Engineering, 3 Department of Obstetrics and Gynecology, 4 Department of Physiology, University of Michigan, Ann Arbor, MI 48109, USA 5 Correspondence: e-mail:takayama@umich.edu 6 Correspondence: e-mail: smithgd@umich.edu Abstract A microfluidic device was designed with two parallel laminar flow channels where non-motile spermatozoa and debris would flow along their initial streamlines and exit one outlet, whereas motile spermatozoa had an opportunity to swim into a parallel stream and exit a separate outlet. Motile sperm samples were prepared with density gradient separation (n = 5). Sperm motility was assessed the following day after exposing aliquots to polydimethylsiloxane (PDMS) used to construct the device. There was no difference in sperm motility when compared with unexposed aliquots (P > 0.05). Unprocessed semen samples (n = 10) were placed in wider channels and sperm motility and strict morphology were assessed from sorted outlets. Sperm motility increased from 44 ± 4.5% to 98 ± 0.4% (P < 0.05) and morphology increased from 10 ± 1.05% to 22 ± 3.3% (P < 0.05) following processing. Finally, density gradient prepared samples (n = 6) containing 5 10 6 motile spermatozoa/ml and 50 10 6 round immature germ cells/ml were sorted and assessed in a similar fashion. The ratio of motile spermatozoa to round immature germ cells in the wide inlet (1:10) was significantly improved in the thin outlet (33:1) (P < 0.05). This microfluidic device provides a novel method for isolating motile, morphologically normal spermatozoa from semen samples without centrifugation. This technology may prove useful in isolating motile spermatozoa from oligozoospermic samples, even with high amounts of non-motile gamete and/or non-gamete cell contamination. A movie sequence showing streaming and sorting of spermatozoa may be purchased for viewing on the internet at www.rbmonline.com/article/847 (free to web subscribers). Keywords: gravitational pumping, laminar flow, microfluidics, sperm motility, sperm processing Introduction Several methods for isolation of motile spermatozoa from semen samples for use in assisted reproduction are commonly utilized. Repetitive centrifugation and resuspension of the pellet in media allows for removal of seminal plasma from the sample, but fails to select for motile spermatozoa. Selection of highly motile spermatozoa from the pellet can be achieved by allowing spermatozoa to swim up into overlying media. However, motile spermatozoa can be damaged by reactive oxygen species when pelleted with senescent spermatozoa and neutrophils, making swim-up from a pellet undesirable in samples with large numbers of these cells (Aitken and Clarkson, 1988). To avoid this, one can perform a direct swimup by overlaying an uncentrifuged semen sample with media and allowing motile spermatozoa to swim into the top layer. Although both techniques result in a population of highly motile spermatozoa, a significant number of motile spermatozoa are not recovered with these methods, a problem magnified in oligozoospermic samples. Recovery rates of motile spermatozoa from initial samples can vary, with reports ranging from 0.8 to 50% for direct swim-up (Englert et al., 1992; Smith et al., 1995). Density gradient separation is another common technique used to process semen samples. Unlike sperm washing, this technique selectively isolates motile spermatozoa from neutrophils and debris. However, density gradient separation has been reported to have potential detrimental effects on sperm DNA integrity (Zini et al., 2000). Additionally, similar to the alternative methods, yield of recovered spermatozoa with progressive motility in oligozoospermic samples is as low 75
as 18% (Florence et al., 1992). With the advent of intracytoplasmic sperm injection (ICSI), men with severe oligozoospermia are now offered the opportunity to reproduce. In addition to low sperm counts, semen samples from these patients often have large amounts of cellular debris. Consequently, swim-up from a pellet is not a viable option to prepare these semen samples. Additionally, due to the low yield of motile spermatozoa with the other isolation methods discussed, the number of viable spermatozoa available after processing for use in ICSI is limited and occasionally prohibitive for this group of men. Because of this, new sperm isolation methods are required for this patient population. The field of microfluidics is a new area of biomedical engineering concerned with the microscopic flow of fluids. At the microscopic level, streams of fluid will maintain laminar rather than turbulent flow. Because viscous forces dominate over inertial forces in these small flows, two separate streams of fluid with laminar flow can be placed parallel to each other with the two streams mixing only by diffusion despite there being no physical barrier between them. Multiple laminar flow systems have previously been utilized to perform diffusionbased separation (Brody and Yager, 1997), to detect molecules (Wiegl and Yager, 1999), to fabricate microstructures in capillaries (Kenis et al., 1999), and to pattern cells and their environments (Takayama et al., 2001). It was thought that this unique feature of microscopic fluid flow could be utilized to separate motile spermatozoa from the remainder of the semen sample (Figure 1). Theoretically, a stream of semen could be flowed in parallel to a stream of media. Given the nature of laminar flow, only motile spermatozoa could cross from semen into the parallel stream and subsequently be sorted from the initial stream when the two fluids diverge. Based on this hypothesis, there were three objectives. First, a device was designed with two microscopic channels that would flow parallel to each other and eventually diverge. Second, overnight sperm motility was assessed after exposure to the material used to construct the channels to verify that the material was non-toxic to spermatozoa. Finally, the hypothesis was tested with fresh semen samples as well as samples with added germ cells. Materials and methods The microfluidic sperm sorting channel was designed to provide a laminar flow sorting system where non-motile spermatozoa and debris would flow along their initial streamlines and exit one outlet whereas motile spermatozoa would have a higher probability of deviating into the parallel stream and exiting through a different outlet. The widths of the streams were designed for a ratio of 1:3 with the intention of placing the unsorted samples in the thinner inlet channel. Additionally, a microfluidic gravity and capillary force driven pump was designed to maintain a steady flow rate of ~8 nl/s regardless of the fluid level in the reservoir. Channels were made using photolithography for fabrication of the silicon wafer master with the design shown in Figure 2 and replica moulding for preparing the polydimethylsiloxane (PDMS) channels from the master (Duffy et al., 1998; Whitesides et al., 2001). The channels and a glass slide were plasma oxidized for 5 min using a Plasma Prep II TM Chamber. This allowed the channels to seal onto the slide and to make the inner surface of the channels hydrophilic, which improved liquid flow inside the channels enough that no additional bovine serum albumin (BSA) was needed for most semen samples. Semen samples were obtained from men undergoing evaluation for infertility after a minimum of 3 days of 76 Figure 1. Two streams of fluid with laminar flow can be placed in parallel and will only mix by diffusion because of surface tension of the fluid, despite no physical barrier dividing the streams. The two fluid streams are depicted by dashed and solid lines. Hypothetically, the unique properties of microfluidics would allow spermatozoa in the original sample to swim into the parallel stream of media and be sorted when the streams diverged while non-motile spermatozoa, round cells and debris would stay in the initial stream. Figure 2. Photograph of microfluidic sperm sorter. A United States penny is shown for size comparison. Schematic drawing of channel design. The width of the thin channels is 100 µm, the wide channels 300 µm, and the common mid channel = 500 µm. Depth of all channels is 50 µm.
Figure 3. (A) Still image from the video of sperm sample entering channel at the inlet junction (this portion of the channel corresponds to the junction of inlet streams in Figure 2). Motile spermatozoa, depicted by white arrows, can be seen starting to swim out of their initial streamline and spreading throughout the width of the channel. The majority of the non-motile spermatozoa, however, are positioned in the initial streamline, which corresponds to the upper stream in this image. (B) Still image from the video of spermatozoa being sorted at the outlet junction (this portion of the channel corresponds to the stream bifurcation in Figure 2). At the outlet bifurcation, the motile spermatozoa depicted by white arrows are evenly distributed throughout the width of the wide channel. The majority of non-motile spermatozoa and debris continue to stay in the initial streamline, corresponding to the upper stream in this image. The bright white material seen in the chamber is tiny air bubbles trapped in the moulding used to construct the device. Scale: the width of the thinnest channel is 100 µm, that of the thickest channel is 500 µm. abstinence. After liquefaction, a semen analysis was performed by a single individual and consisted of assessment of semen volume, ph, viscosity, liquefaction, sperm count, sperm motility, sperm agglutination, strict sperm morphology and cell contamination. Semen samples containing spermatozoa with forward progression were selected for experimental use. Approval for utilizing semen samples for all experiments was obtained from the University of Michigan Institutional Review Board (IRB). For overnight survival, semen samples were prepared by density gradient separation using Isolate (Irvine Scientific, Santa Ana, CA, USA). The supernatant was removed after centrifugation at 300 g for 20 min and the pellet resuspended in 500 µl of HEPES-buffered human tubal fluid (Irvine Scientific) with 0.2% BSA (processing medium; PM). Initial sperm motility was assessed in each sample (n = 5) by evaluating 200 spermatozoa. From each sample, three aliquots of 200 µl were pipetted into Eppendorf tubes for treatment. Treatment groups consisted of group 1 (untreated); group 2 (30 min exposure to PDMS); and group 3 (30 min exposure to latex). After treatment, samples were incubated at 37 C overnight. Two hundred spermatozoa from each group were assessed the following day for motility. Microfluidic sperm sorting tests were conducted at room temperature using fresh semen samples (n = 10). Initial testing of the device was undertaken with semen samples processed by density gradient separation (data not shown), which demonstrated the ability to separate motile spermatozoa from the remainder of the sample regardless of loading the initial sample in the wide or thin inlets (Figure 3). Attempts at loading the unprocessed semen samples into the thin inlet were unsuccessful because debris in the samples obstructed the channels shortly after flow began. All semen samples were thereafter placed in the wider inlet. A 40 µl aliquot of each sample was pipetted into the wide inlet while 40 µl PM was pipetted into the thin inlet. Two semen samples were diluted 1:1 with PM prior to placement in the inlet in order to decrease the viscosity of the sample. Two hundred spermatozoa from the wide inlet and thin outlet were assessed for sperm motility. Additionally, at least 100 spermatozoa from the wide inlet and thin outlet were assessed for strict sperm morphology by a single individual blinded as to which well the samples were obtained from. To simulate semen samples containing large amounts of debris, a stock solution of round immature germ cells was prepared from a semen sample containing an extensive number of these cells. The stock solution was prepared by mixing the semen sample with 4 ml of PM, centrifugation of the solution at 300 g for 6 min, and removal of the supernatant. A swim-up was then performed by overlaying the pellet with 1 ml of PM and incubating the solution at 37 C for 2 h. The supernatant was removed and the pellet resuspended in 0.5 ml of PM. The final concentration of 50 10 6 germ cells/ml was verified using a Makler counting chamber (Sefi-Medical Instruments, Haifa, Israel). Aliquots of the prepared sample were frozen in liquid nitrogen until needed. Additional semen samples (n = 6) were prepared by density gradient separation as described above. Prepared samples were diluted with PM to 5 10 6 motile spermatozoa/ml and concentrations verified in duplicate using a Makler counting chamber. A 10 µl aliquot of the prepared sperm sample and thawed germ cells was added to 30 µl of PM. A 40 µl aliquot of each solution was placed into the wide inlet of the microfluidic sorting chamber while 40 µl of PM was simultaneously placed into the thin inlet. The initial 200 spermatozoa flowing in the thin outlet were assessed for motility. Additionally, the number of germ cells was counted in the fluid containing the initial 200 spermatozoa. 77
Figure 4. Mean initial and overnight sperm motility with and without exposure to polydimethylsiloxane (PDMS) and latex (n = 5). Columns with different case letters are significantly different (P < 0.05). Error bars represent standard error. Figure 6. Comparison of strict sperm morphology from the initial semen sample (wide inlet) and spermatozoa located in the thin (sorted) outlet (n = 10). At least 100 spermatozoa were evaluated from each group. Columns with different case letters are significantly different (P < 0.05). Error bars represent standard error. 78 Figure 5. Comparison of per cent sperm motility from the initial semen sample (wide inlet) and spermatozoa located in the thin (sorted) outlet (n = 10). Two hundred spermatozoa were evaluated in each group for each sample. Columns with different case letters are significantly different (P < 0.05). Error bars represent standard error. The Wilcoxon signed rank test was used for non-parametric comparison between groups. A P-value <0.05 was considered significant. Results Mean initial sperm motility (n = 5) for overnight survival experiments was 75 ± 5.6% (mean ± SE). Overnight sperm motility for group 1 (untreated) was 69 ± 4.6%; group 2 (30 min exposure to PDMS) was 64 ± 3.9%; and group 3 (30 min exposure to latex) was 0% (Figure 4). Compared with the initial sperm motility, overnight sperm motility was significantly lower for all groups (P < 0.05); however, there was no difference in sperm motility between spermatozoa with and without exposure to PDMS (P > 0.05). As postulated, motile spermatozoa deviated from their initial stream into the parallel stream and were sorted using the microfluidic sperm sorting device. Mean sperm motility of the fresh semen samples (n = 10) placed in the large inlet well for the microfluidic sperm sorting experiments was significantly lower (44 ± 4.5%) compared with sorted sperm motility in the thin outlet well (98 ± 0.4%; P < 0.05) (Figure 5). Strict sperm morphology in the initial samples (10 ± 1.1%) was also significantly lower compared with sorted sperm morphology in the thin outlet well (22 ± 3.3%) (Figure 6). Figure 7. Image of a sample prepared with 5 10 6 motile spermatozoa/ml and 50 10 6 germ cells/ml placed in the wide inlet is shown in (A). The thin outlet was evaluated for the motility of the initial 200 spermatozoa and number of germ cells accompanying them. An image of the thin outlet is shown in (B). Scale bar = 0.00156 mm. For experiments simulating debris within samples using round immature germ cells, the wide channel became clogged with debris approximately 5 min after fluid flow began, making it impossible to obtain enough volume in the thin outlet to quantify accurately with a pipette. However, mean sperm motility of the initial 200 spermatozoa in the thin outlet was 98 ± 0.8%. Mean number of germ cells within the sorted outlet with the initial 200 spermatozoa was 6. The ratio of motile spermatozoa to germ cells in the thin outlet (33:1) compared favourably to the initial solution placed in the wide inlet (1:10) (P < 0.05; Table 1, Figure 7). Discussion Microfluidics in assisted reproduction has only recently been described for use with oocytes and embryos (Beebe et al., 2002). As has been demonstrated, a highly motile subpopulation of spermatozoa can be isolated from semen using microfluidics, including samples with large amounts of
Table 1. Data from sorted samples (n = 6) prepared with 5 10 6 motile spermatozoa/ml and 50 10 6 germ cells/ml. Sample Ratio of motile Sperm motility Number of Ratio motile spermatozoa to of initial 200 germ cells spermatozoa: germ cells in inlet spermatozoa (%) in outlet germ cells in outlet A 1:10 96 3 64:1 B 1:10 99 2 100:1 C 1:10 95 6 32:1 D 1:10 97 15 12:1 E 1:10 99 6 34:1 F 1:10 100 1 200:1 Mean 1:10 98 6 33:1 debris. It has previously been demonstrated that approximately 40% of motile spermatozoa can be recovered using the microfluidic chamber dimensions described in these experiments (Cho et al., 2003). Sorting motile spermatozoa from semen samples using microfluidics offers several advantages over current isolation techniques. With the exception of direct swim-up from a semen sample, most sperm processing techniques, including density gradient separation and semen washing, require centrifugation, which has been reported to cause sub-lethal damage to the spermatozoa (Alvarez et al., 1993; Smith et al., 1995). Using microfluidics for motile sperm isolation eliminates the need for centrifugation. Additionally, the reported detrimental effects of reactive oxygen species from centrifugal pelleting of unselected spermatozoa, debris, and neutrophils (Mortimer, 1991) is eliminated using microfluidics. Experiments are underway to directly compare yields of motile spermatozoa using microfluidics, density gradient separation and swim-up techniques. A significant number of morphologically normal motile spermatozoa are lost with currently available sperm processing techniques, which presents a challenge to laboratories when processing oligozoospermic samples. Unfortunately, with swim-up or density gradient separation, the ability to recover motile spermatozoa lost with processing is limited. However, with microfluidic sperm sorting, if there are insufficient motile spermatozoa after sorting, more motile spermatozoa could be recovered from fluid in the unsorted outlet by placing it back into the inlet well and reprocessing the sample. Similar to reports using density gradient separation or swim-up techniques (Englert et al., 1992; Smith et al., 1995; Prakash et al., 1998; Van Der Zwalmen et al., 1991), motile spermatozoa isolated using microfluidics show a significant improvement in morphology compared with the original sample. This is not surprising, given that all sperm isolation techniques select for highly motile spermatozoa, which have improved morphology compared with spermatozoa with poor motility (Katz et al., 1982). The fertilizing capability of the recovered spermatozoa has not been tested; however, given the non-toxic material used in constructing the device and lack of centrifugation, it is unlikely to be compromised. Experiments are underway to verify this. It has been reported that density gradient separation selects for spermatozoa without chromatin and nuclear DNA anomalies when compared with the initial sample and spermatozoa separated by the swim-up technique (Sakkas et al., 2000). Despite this, other research has shown improved DNA integrity when using the swim-up technique compared with density gradient separation despite comparable sperm motility parameters with each processing technique (Zini et al., 2000). These differences may be due to the various methods used to assess for DNA damage in each study. Future studies will need to assess if motile spermatozoa isolated by microfluidics have improved DNA integrity compared with the original samples. Although the microfluidic sperm sorter can produce a population of highly motile, morphologically normal spermatozoa from a semen sample, several important aspects in the design of the device need to be addressed. Currently, only 40 50 µl of fluid can be loaded into each inlet well, which is well below the normal volume for human ejaculates. At the flow rate in the current device only 10 20 µl of semen can be sorted in 30 min. Several ways to address this are currently underway, including designing deeper channels and placing multiple channels in parallel. The efficiency of isolating motile spermatozoa from a semen sample using microfluidics is dependent on several variables. The rate of flow of the fluid streams, the duration of contact between parallel streams, and sperm motility parameters will all affect efficiency. Additionally, the width of the fluid streams can impact efficiency, since the chance of swimming into the parallel stream prior to stream divergence is random. The wider the initial stream containing semen, the lower the chance the spermatozoa located on the side opposite of parallel stream contact have of swimming into the parallel stream prior to divergence. These principles were used to design the channels used in the current experiments including the 1:3 ratio between the parallel stream diameters. Ideally, semen samples would flow through the thinner stream, allowing a greater likelihood that motile spermatozoa could swim into the wider stream and be sorted. Although the spermatozoa recovered in these experiments were highly motile and morphologically better than in the original samples when loaded in the wider channel, efficiency would be improved by placing the semen into thinner inlet channels. As noted, however, when semen was placed in the thin inlet, debris and clumps of spermatozoa obstructed the channel. Designing deeper and slightly wider inlet channels should circumvent this problem in the future. Additionally, the magnitude of motile force generated by a spermatozoon will determine if it crosses 79
80 the laminar flow boundary. Increasing the duration of contact between the streams would theoretically improve efficiency for spermatozoa with poor forward progression. This can be achieved by lengthening the common channel and/or decreasing the flow rates of the fluid streams. Further evaluation of these modifications is underway. A limitation of microfluidic sperm sorting is the reliance on forward sperm progression to isolate motile spermatozoa. Obviously, spermatozoa with poor forward progression would not be sorted as efficiently as those with rapid forward progression. Under normal physiological conditions, after incubation in the female genital tract, spermatozoa undergo capacitation, a requirement to acquire the ability to fertilize an oocyte (Yanagimachi, 1992). Sperm motility changes during capacitation from linear movement to an increased flagellar beat amplitude with decreased forward progression, termed hyperactivation. These changes are under control of calcium and can be activated in vitro with the addition of dibutyryl cyclic AMP, phosphodiesterase inhibitors, including caffeine and pentoxifylline, or by incubation of the spermatozoa with bicarbonate ions at 37 C (Barkay et al., 1977; Boatman and Bavister, 1984; Pang et al., 1993). Given the increase in flagellar movement coupled with the decrease in forward progression, it is difficult to predict if microfluidic sperm sorting efficiency will be improved by using hyperactivated spermatozoa. Testing is currently underway to explore this possibility. When two microscopic streams of fluid flow in parallel, the viscosity of the fluid within each stream needs to be similar or the streams will have different flow rates and the diameter of the more viscous stream will be compressed by the less viscous stream. Although semen samples used in the experiments flowed well, two of the samples did have increased viscosity, requiring a 1:1 dilution with PM prior to use. If viscous semen samples were to be processed for clinical use with microfluidics, the semen volume would be doubled, compounding the volume limitations that currently exist for the device. In addition to isolation of motile spermatozoa from a semen sample for use in assisted reproduction, several other possible applications exist for sperm processing with microfluidics. Motile spermatozoa could be isolated from semen samples prior to cryopreservation, which has been shown to improve thawed sperm parameters (Perez-Sanchez et al., 1994). Additionally, although controversial, it has been suggested that spermatozoa containing the Y chromosome have increased forward progression compared with X-bearing counterparts (Ericsson, 1994). Isolation of a Y-bearing sperm population for ICSI would be useful to prevent the transmission of X-linked diseases (Seidel and Johnson, 1999). Several microfluidic prototypes have been designed to evaluate this concept further by selectively isolating spermatozoa with the best forward progression. Finally, free antisperm antibodies and spermatozoa with large numbers of bound antisperm antibodies could be removed from semen by mixing the sample with immunobeads prior to placing the sample in microfluidic channels. Unbound spermatozoa could swim into the parallel stream while those bound to the immunobeads would remain in the initial streamline with non-motile spermatozoa and debris. This hypothesis is being tested currently. Beyond the area of assisted reproductive technologies, using microfluidics to sort motile from non-motile organisms has many potential applications. Hamster spermatozoa are often used in toxicology experiments to assess the effect on motility (Bavister and Andrews, 1988; Stewart-Savage and Bavister, 1988). In the future, microfluid chambers may be useful in assessing toxicity in a hamster sperm motility assay. So far as is known, this is the first demonstration that a population of highly motile, morphologically normal spermatozoa can be isolated from a semen sample by interstreamline crossing using a microfluidic device. This technique eliminates the need for centrifugation, which can be of benefit to all processed semen samples. Additionally, this technique may be essential for severely oligoasthenozoospermic samples in providing efficient means of isolating motile spermatozoa for assisted reproductive technology where current methodologies have suboptimal results. Current studies are focused on optimization of device design and extended applications for its use in assisted reproductive technology and beyond. References Aitken RJ, Clarkson JS 1988 Significance of reactive oxygen species and antioxidants in defining the efficacy of sperm preparation techniques. Journal of Andrology 6, 367 376. Alvarez JG, Lasso JL, Blasco L et al. 1993 Centrifugation of human spermatozoa induces sublethal damage; separation of human spermatozoa from seminal plasma by a dextran swim-up procedure without centrifugation extends their motile lifetime. Human Reproduction 7, 1087 1092. Barkay J, Zuckerman H, Sklan D, Gordon S 1977 Effect of caffeine on increasing the motility of frozen human sperm. Fertility and Sterility 28, 175 177. Bavister BD, Andrews JC 1988 A rapid sperm motility bioassay procedure for quality-control testing of water and culture media. Journal of in Vitro Fertilization and Embryo Transfer 2, 67 75. Beebe D, Wheeler M, Zeringue H et al. 2002 Microfluidic technology for assisted reproduction. Theriogenology 57, 125 135. Boatman DE, Bavister BD 1984 Stimulation of rhesus monkey sperm capacitation by cyclic nucleotide mediators. Journal of Reproduction and Fertility 71, 357 366. Brody JP, Yager P 1997 Diffusion-based extraction in a microfabricated device. Sensors and Actuators, Physical 58, 13 18. Cho B, Schuster TG, Smith GD, Takayama S 2003 Passively driven integrated microfluidic system for separation of motile sperm. Analytical Chemistry 75, 1671 1675. Duffy DC, McDonald JC, Schueller OJ et al. 1998 Rapid prototyping of microfluidic systems in polydimethylsiloxane. Analytical Chemistry 70, 4974 4984. Englert Y, Van den Bergh M, Rodesch C et al. 1992 Comparative auto-controlled study between swim-up and Percoll preparation of fresh semen samples for in-vitro fertilization. Human Reproduction 3, 399 402. Ericsson RJ 1994 Sex selection via albumin columns: 20 years of results. Human Reproduction 10, 1787 1788. Florence LHN, Liu DY, Baker HWG 1992 Comparison of Percoll, mini-percoll and swim-up methods for sperm preparation from abnormal semen samples. Human Reproduction 2, 261 266. Katz DF, Diel L, Overstreet JW 1982 Differences in the movement
of morphologically normal and abnormal human seminal spermatozoa. Biology of Reproduction 26, 566 570. Kenis PJA, Ismagilov RF, Whitesides GM 1999 Microfabrication inside capillaries using multiphase laminar flow patterning. Science 285, 83 85. Mortimer D 1991 Sperm preparation techniques and iatrogenic failures of in-vitro fertilization. Human Reproduction 2, 173 176. Pang SC, Chan PJ, Lu A 1993 Effects of pentoxifylline on sperm motility and hyperactivation in normozoospermic and normokinetic semen. Fertility and Sterility 60, 336 443. Perez-Sanchez F, Cooper TG, Yeung CH et al. 1994 Improvement in quality of cryopreserved human spermatozoa by swim-up before freezing. International Journal of Andrology 17, 115 120. Prakash P, Leykin L, Chen Z et al. 1998 Preparation by differential gradient centrifugation is better than swim-up in selecting sperm with normal morphology (strict criteria). Fertility and Sterility 69, 722 726. Sakkas D, Manicardi GC, Tomlinson M et al. 2000 The use of two density gradient centrifugation techniques and the swim-up method to separate spermatozoa with chromatin and nuclear DNA anomalies. Human Reproduction 5, 1112 1116. Seidel GE, Johnson LA 1999 Sexing mammalian sperm overview. Theriogenology 52, 1267 1272. Smith S, Hosid S, Scott L 1995 Use of post-separation sperm parameters to determine the method of choice for sperm preparation for assisted reproductive technology. Fertility and Sterility 63, 591 597. Stewart-Savage J, Bavister BD 1988 Deterioration of stored culture media as monitored by a sperm motility assay. Journal of in Vitro Fertilization and Embryo Transfer 2, 76 80. Takayama S, Ostuni E, LeDuc P et al. 2001 Laminar flows: subcellular positioning of small molecules. Nature 411, 1016. Van Der Zwalmen P, Bertin-Segal G, Geerts L et al. 1991 Sperm morphology and IVF pregnancy rate: comparison between Percoll gradient centrifugation and swim-up procedures. Human Reproduction 4, 581 588. Whitesides GM, Ostuni E, Takayama S et al. 2001 Soft lithography for biology. Annual Review of Biomedical Engineering 3, 335 373. Wiegl BH and Yager P 1999 Microfluidic diffusion-based separation and detection. Science 283, 346 347. Yanagimachi R 1992 Mammalian fertilization. In: Knobil E et al. (eds) The Physiology of Reproduction. Raven Press, New York, pp. 189 317. Zini A, Finelli A, Phang D et al. 2000 Influence of semen processing technique on human sperm DNA integrity. Urology 6, 1081 1084. Received 15 January 2003; refereed 7 February 2003; accepted 12 March 2003. 81