A Novel On-Chip Method for Differential Extraction of Sperm in Forensic Cases

Abstract

One out of every six American women has been the victim of a sexual assault in their lifetime. However, the DNA casework backlog continues to increase outpacing the nation's capacity since DNA evidence processing in sexual assault casework remains a bottleneck due to laborious and time-consuming differential extraction of victim's and perpetrator's cells. Additionally, a significant amount (60-90%) of male DNA evidence may be lost with existing procedures. Here, a microfluidic method is developed that selectively captures sperm using a unique oligosaccharide sequence (Sialyl-Lewis^X), a major carbohydrate ligand for sperm-egg binding. This method is validated with forensic mock samples dating back to 2003, resulting in 70-92% sperm capture efficiency and a 60-92% reduction in epithelial fraction. Captured sperm are then lysed on-chip and sperm DNA is isolated. This method reduces assay-time from 8 h to 80 min, providing an inexpensive alternative to current differential extraction techniques, accelerating identification of suspects and advancing public safety.

Keywords: DNA casework backlog; bioinspired materials; differential extraction; forensic cases; microfluidics.

Figure 1

Workflow of on‐chip differential extraction. i) In practice, samples are collected using a swab or cotton gauze in a forensic scene, where a mixture of semen and epithelial cells are majorly present on the victim's body and/or garments at the crime scene. ii) After collection, samples are simply introduced into the device using single‐step pipetting and incubated for an hour at room temperature. The channels are then washed and sperm cells are specifically captured, while epithelial cells are removed due to their larger size and lack of an adhesion molecule on the channel surface. iii) The captured sperm are treated with a lysis buffer on‐chip, and sperm DNA is collected into a tube for potential forensic downstream genomic analyses.

Figure 2

Evaluation of SLeX binding kinetics and binding locations on sperm head. a) SLeX structure was extracted from a protein complex defined in the Protein Data Bank (PDB ID: 3PVD) and visualized in silico. Computational analysis revealed the molecular surface of the SLeX agent for sperm binding using VMD's built‐in SURF tool. b) β1–4 galactosyltransferase 1 (B4GALT1) was extracted from human M340H‐beta‐1,4‐galactosyltransferase‐1 (M340H‐B4GAL‐T1, PDB ID: 4EE3) and visualized in silico. This enzyme‐receptor on the sperm plasma membrane plays a key role in sperm‐egg binding. B4GAL‐T1‐SLeX interactions were then computed using AutoDock Vina, and the analyses revealed at least nine unique locations for seventeen potential binding modes for SLeX binding to B4GALT1. c) At these docking sites, strong binding was observed with the affinity energies ranging from −9.0 to −11.6 kcal mol⁻¹. d) We further observed that SLeX molecules capture sperm cells with different morphologies (i.e., normal, condensed acrosome, abnormal middle‐piece, large head, double heads, double tails, small head, and tail‐less) on‐chip. These experimental findings confirmed our results observed in silico, indicating that SLeX targets sperm head and its binding is independent of distinct sperm morphologies. Scale bars (black lines) represent 10 µm.

Figure 3

Evaluation of surface chemistry and microfluidic chip parameters for sperm capture. a) Glass surfaces were decorated with SLeX agent using a layer‐by‐layer surface chemistry approach. Capture efficiency was evaluated by varying three parameters: i) concentration of mediator molecule (i.e., 4‐Aminobenzoic acid hydrazide: 4‐ABAH) and bovine serum albumin (BSA), ii) SLeX concentration, and iii) channel height. b) Various 4‐ABAH (0.25 mg mL⁻¹ and 2 mg mL⁻¹) and BSA concentrations (0% and 3%) were examined, and sperm capture efficiency was calculated at each concentration. In these experiments, 50 µm high microchannels were modified with a fixed SLeX concentration (0.1 mg mL⁻¹). Here, 0.25 mg mL⁻¹ of 4‐ABAH provided higher capture efficiency than 2 mg mL⁻¹ of 4‐ABAH but it was not statistically different (n = 3–4, p > 0.05). Further, this might be due to potential steric hindrance for SLeX immobilization to the surface. Further, BSA blocking did not significantly affect the sperm capture efficiency (n = 3–4, p > 0.05) in these experimental sets. c) Different SLeX concentrations ranging from 0.1 to 0.5 mg mL⁻¹ were used to evaluate sperm capture. The 50 µm high microchannels were modified with the optimized 4‐ABAH (0.25 mg mL⁻¹) and BSA (3%) concentrations. Here, 0.5 mg mL⁻¹ of SLeX concentration provided higher capture efficiency compared to the other groups. Further, the capture efficiency in 0.5 mg mL⁻¹ of SLeX was not statistically different than the other SLeX concentrations (n = 4, p > 0.05). d) Two channel heights (50 µm and 80 µm) were evaluated in terms of sperm capture efficiency. The microchannels were decorated with the optimized 4‐ABAH (0.25 mg mL⁻¹), BSA (3%), and SLeX (0.5 mg mL⁻¹) concentrations. We observed that 50 µm high channel heights resulted in higher capture efficiency than an 80 µm high channel. e) Surface functionalization and SLeX binding on the modified channels were confirmed by Fourier Transform Infrared‐Attenuated Total Reflectance (FTIR‐ATR) measurements. At the fingerprint region of SLeX (900 cm⁻¹ to 1280 cm⁻¹), we observed C—H wagging at 962 cm⁻¹, C—O stretching/bending at 1025 and 1082 cm⁻¹, C—O stretching at 1134 cm⁻¹, and C—N stretching at 1181 cm⁻¹. Due to characteristic absorption region of glass between ≈682 cm⁻¹ and ≈1200 cm⁻¹ (e.g., Si—OH bending and Si—O—Si stretching), the signal intensities of peaks in this region reduced. We also observed absorption peaks at 3500, 2850–2950, 2500–3000, 1733, and 1630–1690 cm⁻¹ caused by O—H stretching, C—H bending, O—H stretching, C=O (ester) stretching, and C=O (amide) stretching vibrations of SLeX molecule, respectively. By performing contact angle measurements, we evaluated hydrophilicity properties after surface modification (inset figure). The contact angle value of the bare glass surface altered from 48.5° ± 4.3 to 13.7° ± 3.1 after SLeX modification. According to the FTIR‐ATR and contact angle measurements, SLeX molecule was successfully immobilized to the microchannel surface. f) Spatial distribution of cell capture was analyzed on‐chip by imaging the entire microchannel surface through a tiling function of the microscope with an automated x–y stage. Sperm counts before and after the washing step were plotted through horizontal and vertical directions. Before the washing step, a homogenous cell distribution was observed in a horizontal direction, whereas sperm cell count increased in the middle of the channels on the vertical axis. The cell count was altered in the horizontal direction after the washing step and most of the sperm close to the inlet washed away from the channel surface. On the other hand, the distribution trend at the vertical axis did not change after the washing step. For statistical analysis, we used one‐way ANOVA with Tukey's post hoc test for multiple comparisons with the statistical significance threshold set at 0.05 (p < 0.05). Data is represented with average value ± standard deviation (n = 3–4).

Figure 4

Evaluation of nonspecific sperm cell binding (control) and limit of detection. a) The microchannels without surface chemistry were used as a control set. Nonspecific sperm cell binding was assessed with high (750–1800 sperm per channel) and low (100–300 sperm per channel) cell numbers. Only a limited number of sperm (275 ± 96 cells) remained in the channels when we applied 1742 ± 239 cells into the microchannels. Sperm samples with a high cell number were significantly removed from the channel surfaces in the absence of surface chemistry (n = 4, p < 0.05). In addition, some sperm (186 ± 97 cells) remained when we introduced 285 ± 111 cells to the microchannels (n = 4, p > 0.05). These results demonstrated that the bare glass surface itself has ≈200 nonspecific binding points over the sperm count range. b) We also evaluated the detection capability of microchannels modified with surface chemistry. Most sperm (748 ± 9 cells) were captured when we applied 798 ± 9 cells into the microchannels (n = 3, p > 0.05). In low cell count experiments, we observed that 116 ± 17 sperm were captured on‐chip when we introduced 134 ± 19 cells to the microchannels (n = 3, p > 0.05). As demonstrated in the plot, the microchannels modified with surface chemistry efficiently captured sperm in both high and low cell numbers with ≈94% and ≈86% efficiency, respectively. c) Further, cell numbers were converted into percentage of sperm remaining in microchannels after washing step. Higher ratios of sperm remained on the surface chemistry applied channels than that of control surfaces without surface chemistry (n = 3–4, p < 0.05). We also observed that ≈200 bindings were mainly due to sperm‐glass surface interactions in control surfaces (n = 4, p < 0.05). In these experiments, we introduced 15 µL of samples with high and low sperm counts into the channels. d,e) We evaluated the limit of detection parameter for the microchannels by applying multiple cell concentrations varying from ≈20 to ≈8000 cells per channel. The microchannels captured down to ≈20 sperm cells per channel with a capture efficiency of 75.4 ± 1.5% (n = 3, p < 0.05), and the capture efficiency increased up to 93.6 ± 3.0% at higher cell counts (up to ≈8000 cells per channel), indicating that the microchannels were able to handle a broad range of cell numbers and the capture capability of chips was independent of high cell numbers introduced into the microchannels. f) Limit of detection parameter was further analyzed through a nonlinear fitting function. The curve had a linearity of 0.94 and 0.87 for R ² (Coefficient of determination: COD) and adjusted R ², respectively. For statistical analysis, we used one‐way ANOVA with Tukey's post hoc test for multiple comparisons with the statistical significance threshold set at 0.05 (p < 0.05). Horizontal brackets and asterics demonstrate statistically significant differences between groups. Data is represented with average value ± standard deviation (n = 3–4).

Figure 5

Specificity experiments and validation of microfluidic chips with forensic mock samples. a) Specificity of SLeX was tested with a heterogeneous cell population consisting of a male's sperm and buccal epithelial cells collected from a female's inner cheeks. Two sets of microfluidic chips were prepared: i) all surface chemistry steps including SLeX and ii) all surface modifications without SLeX. b) SLeX‐modified surfaces provided 91.1 ± 3.1% of capture efficiency, whereas sperm cells drastically washed away from the surfaces without SLeX (n = 5, p < 0.05). In addition, SLeX provided high specificity to capture sperm (≈91%) compared to epithelial cells (≈7% and ≈1%) in both experimental sets (n = 5, p < 0.05). There was no significant binding of epithelial cells observed in the microchannels with SLeX and without SLeX (n = 5, p > 0.05). c) Microphotography was performed before and after the washing steps on microchannels with SLeX. Black arrows represent epithelial cells (ECs) in the microchannels. Scale bars represent 50 µm. d) Simulated forensic samples (noncasework samples) were obtained from the Broward Sheriff's Office Forensic Laboratory. Five different mock samples were introduced into the microchannels modified with SLeX, and the numbers of sperm were then counted before and after the wash steps. We observed various numbers of sperm ranging from ≈300 to ≈745 cells, and most of the sperm cells were captured in the microchannels. e) Mock samples provided high capture efficiencies, spanning from ≈70% to ≈92%. f) The mock samples were collected, using either cotton swab or cotton gauze, on different dates and they consisted of different cell content and concentrations. The details were presented in the table. Data was represented with average value ± standard deviation (n = 3). For statistical analysis, we used one‐way ANOVA with Tukey's post hoc test for multiple comparisons with the statistical significance threshold set at 0.05 (p < 0.05). Horizontal brackets demonstrate statistically significant differences between groups. Data is represented with average value ± standard deviation (n = 5).