A method for the detection of hCG β in spent embryo culture medium based on multicolor fluorescence detection from microfluidic droplets

Abstract

The evaluation of embryo quality via human chorionic gonadotropin beta (hCG β) and other proteins secreted by embryos in a spent embryo culture medium (SECM) receives a close review in the field of assisted reproduction. However, accurate and quantitative detection of these trace proteins is still a challenge. In this study, a highly sensitive protein detection method using microfluidic droplets and multicolor fluorescence detection was developed and used to detect hCG β secreted by embryos in SECM. β-Galactosidase (β-Gal) was used to label hCG β and can catalyze the conversion of nonfluorescent substrate fluorescein di-β-d-galactopyranoside to produce fluorescein to amplify the signal strength. Compared with previous studies, the proposed method requires only a simple microfluidic chip and can eliminate false-positive signals generated by free β-Gal through simultaneous detection of fluorescence, which can ensure the accuracy of the results. The lower detection limit of hCG β was 0.1 pg/ml. Using the developed method, hCG β in SECM was successfully detected; the hCG β secreted by top-quality blastocysts was significantly higher than that of non-top-quality blastocysts and embryos that do not develop into blastocysts. The proposed method can be used to detect secretory proteins from embryos in SECM and has application value in the screening of other biomarkers.

I. INTRODUCTION

Infertility is a highly prevalent global issue estimated to affect about 9% of reproductive-aged couples.¹ Assisted reproductive technology (ART) has become an effective method to treat infertility.^2,3 Despite the widening application of ART, there are still urgent problems associated with it, such as low implantation rates and high multiple pregnancy rates.⁴ One reason for such problems is that methods that use embryo morphology alone may be unable to select the embryos with the greatest potential for transplantation.⁵ A variety of proteins closely related to embryo development and implantation are secreted by embryos cultured in vitro.⁶ The quantitative detection of secretory proteins from spent embryo culture medium (SECM) is of high research interest for evaluating embryo quality and for the selection of embryos with greater potential for transplantation.^7,8

Human chorionic gonadotropin beta (hCG β) is an embryonic secretory protein that regulates the immune response of the endometrium during syncytiotrophoblast differentiation.⁹ hCG β is secreted at multiple stages of embryonic development and is considered to be a biomarker for evaluating embryo quality.⁶ Although related research has been reported as early as 1984,¹⁰ only a few articles have reported the relationship between hCG β concentration in SECM and embryo quality or its clinical use.¹¹ The volume of the embryo culture medium in ART is only 10–50 μl,¹² and the concentration of the protein secreted by embryos in the culture medium may be <1 pg/ml; so it is difficult to accurately detect them by common detection methods, e.g., enzyme-linked immunosorbent assay (ELISA)¹³ and Luminex.⁸ Therefore, the lack of effective detection methods may be the cause of slow developments of related research in ART.¹⁴

Detection of trace amounts of protein in biological samples remains a challenge because no signal amplification methods exist for proteins that are comparable to nucleic acid detection methods.¹⁵ Rissin et al.¹⁶ developed a reliable quantitative detection technique for trace proteins called single-molecule enzyme-linked immunosorbent assay (digital ELISA). The detection limit of this method can reach the fg/ml level, and only a few μl of sample are required. There are at least two factors that allow the low detection limits of this method:¹⁷ (1) Amplification: the target protein is labeled by β-galactosidase (β-Gal), and then β-Gal catalyzes hundreds of nonfluorescent substrate molecules to produce highly fluorescent resorufin molecules per second; and (2) Extremely small reaction volumes: Microwell arrays are used as microreactors for enzymatic reactions and each well requires only a femtoliter-volume, as the fluorescent products produced by a single enzyme-labeled immunocomplex are restricted to the reaction chamber, so they can reach detectable concentrations.

Microdroplets are commonly used as microreactors and can be used to perform digital ELISAs. The need for mechanical fabrication of femtoliter wells places inherent limits on the scalability and flexibility of ultrasensitive diagnostic assays; however, this could be overcome by microdroplet-based methods.¹⁸ Using different chip structures, Shim et al.¹⁸ and Guan et al.¹⁹ realized a single-molecule-counting immunoassay based on microfluidic droplet technology. More recently, Tian et al.²⁰ developed a digital analysis method using droplet-based microfluidics for the indirect detection of AFP and ultrasensitive detection of β-Gal; both β-Gal and human alpha-fetoprotein (AFP) were detected at fM levels.

However, fluorescence imaging was used in all of the above studies. On the one hand, millions of droplets or microwells must be imaged for digital analysis methods, which require a powerful and expensive fluorescence microscope and image analysis software. On the other hand, complex chip structures are required for droplet-based methods to capture droplets stably for imaging. More importantly, for indirect detection using droplets, it is difficult to exclude false-positive signals generated by free enzymes in fluorescence imaging.

Herein, we propose a hCG β detection method using a simple droplet-based microfluidic chip and a multicolor fluorescence detector (MFD). Fluorescent magnetic beads (FMBs) were used to capture hCG β in samples, and hCG β was labeled using β-Gal according to the steps in Scheme 1 to form a complex with FMBs. The FMB complex and nonfluorescent substrate fluorescein di-β-d-galactopyranoside (FDG) were simultaneously encapsulated in droplets. After incubation, the fluorescence of droplets was detected using an MFD. FMB and enzyme-induced fluorescence products have different excitation and emission wavelengths and can be detected by an MFD. The fluorescence signal of FMB was used to calculate the number of droplets containing the FBM complex, which is an important parameter to calculate the ratio of activated beads. Also, we considered that droplets without FMB fluorescence signals but with enzyme-induced fluorescence signals were false-positives. The detection limit of this method for hCG β was 0.1 pg/ml. Using this method, we quantified hCG β secreted by single embryos in SECM.

SCHEME 1.

Diagram of hCG β detection by a microfluidic droplet and multicolor fluorescence detection.

II. MATERIALS AND METHODS

A. Materials and reagents

Streptavidin-conjugated β-galactosidase (SA-β-Gal), fluorescein di-β-d-galactopyranoside (FDG), fluorescein, N-hydroxysuccinimide (NHS), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), sodium phosphate dibasic, and sodium phosphate monobasic dihydrate were purchased from Sigma-Aldrich (St. Louis, USA). Carboxyl modified 6 μm diameter flash red fluorescent magnetic beads (FMBs) were purchased from Bangs Laboratories (Fishers, USA). hCG β monoclonal antibody (capture Ab) and biotin-labeled hCG β monoclonal antibody (biotin-Ab) were purchased from Novus Biologicals (Littleton, USA). Fluorinated oil containing 2% perfluoropolyether-poly(ethylene glycol) (PFPE-PEG) was purchased from MorSic (Foshan, China). hCG β was purchased from R&D Systems (Minneapolis, USA). Quinn’s Advantage Cleavage Medium and Quinn's Advantage Blastocyst Medium were purchased from SAGE BioPharma (Bedminster, USA).

B. Microfluidic chip fabrication

Simple microfluidic droplet generation and detection chips were designed and fabricated for this study. The chip structure is shown in Fig. 1. Briefly, the main channel width of the droplet generation chip [Fig. 1(a)] was 400 μm; the width and length of the throat position were 50 μm and 100 μm, respectively; and the depth of the chip was 50 μm. The curved structure was designed to improve the mixing degree of liquid in the droplet. The detection chip [Fig. 1(b)] had a simple cross structure with a channel width and depth of 100 μm and 50 μm, respectively. The channel was etched on the poly-(dimethylsiloxane) (PDMS) using soft photolithography; PDMS was then irreversibly bonded to the glass substrate.

FIG. 1.

Structure of the microfluidic chip. (a) Structure of the droplet generation chip; (b) structure of the detection chip.

C. Microfluidic droplet generation procedure

The droplet generation chip as shown in Fig. 1(a) had an FMB complex inlet, an FDG inlet, an oil inlet, and an outlet channel. FDG and FMB complexes were injected from their respective inlets with an equivalent flow velocity of 1 μl/min. Oil was injected from the oil inlet at a flow rate of 6 μl/min. Droplets were generated at the throat position at an approximate frequency of 1000 Hz and a diameter of 40 μm.

D. Multicolor fluorescence detector

The MFD used in this study has previously been reported.²¹ Briefly, 488 and 638 nm lasers were used as two excitation lights for samples. Four colors of fluorescence, including green (510−530 nm), yellow (560−590 nm), red (650−690 nm), and near-infrared (700−740 nm) were detected simultaneously. The detection frequency of the detector was 100 kHZ.

E. Embryo culture and culture medium collection

Conventional in vitro fertilization (IVF) was performed after the oocyte was obtained; fertilization was evaluated 17 ± 1 h later. Fertilized zygotes were transferred to Quinn’s Advantage Cleavage Medium for 3 days, after which the morphological score was assigned. Cleaved embryos were transferred to Quinn's Advantage Blastocyst Medium and cultured independently to the blastocyst stage, and then the blastocyst was graded. A total of 30 μl blastocyst culture medium was used for each embryo, based on the clinical requirements of our center.

F. Embryo morphology score

Cleaved embryos were evaluated based on three aspects: the number of blastomeres, the degree of fragmentation, and the uniformity of the blastomere size. A cleaved embryo was defined as a top-quality embryo if it met either of the following two conditions: (1) The number of blastomeres ≥7 and ≤10, and the degree of fragmentation <20%; and (2) embryos with six uniform size blastomeres and the degree of fragmentation <10%. The morphological scoring criteria of blastocysts were based on the Gardner system.²² Scores were obtained for three aspects: the expansion status of blastocoel (1–6), inner cell mass (A–C), and trophectoderm cell (A–C). Only the blastocysts at stage 3–6 with inner cell quality and trophoblast cell scores of A or B were defined as high-quality blastocysts.

G. Sample collection

A total of 30 SECMs derived from 30 patients undergoing in vitro fertilization (IVF) at our center were collected from May to July 2019 for inclusion in this study. After the blastocyst was transplanted or frozen, 30 μl SECM was collected and stored at −80 °C until further analysis. All the blastocysts grown in SECM were formed on day 5 and developed from top-quality cleavage embryos formed by two pronuclear zygotes. This study was approved by the Research Ethics Committee of Shenzhen Zhongshan Urology Hospital (Approval No.: SZZSECHU-20180021), and informed consent was obtained from all couples prior to the study.

H. Detection of hCG β

The experimental process of detecting hCG β in the samples is shown in Scheme 1. First, capture antibodies were connected to the FMB according to the manufacturer's instructions to form an FMB-Ab complex. Approximately 100 000 FMB-Ab complexes were mixed with 30 μl samples and reacted at room temperature for 30 min. hCG β in the samples was captured to form the complex FMB-Ab-hCG β. After the reaction, the complexes were washed three times with a phosphate-buffered saline (PBS) buffer (10 mM Na₂HPO₄, 2 mM NaH₂PO₄, 20% Tween 20, pH 7.4). Then, 100 μl biotin-Ab was added to the reaction system, which was incubated at room temperature for 30 min and then washed three times with the PBS buffer to remove unreacted biotin-Ab. Finally, 100 μl of 1 pM SA-β-Gal was added into the reaction system, and the reaction was conducted at room temperature for 30 min. In this reaction, hCG β was labeled by β-Gal and the final FMB complex (FMB-Ab-hCG β-Ab-biotin-SA-β-Gal) was formed. The FMB complex was washed three times and re-suspended with 30 μl of PBS buffer. The FMB complex and FDG, as two water phases, form droplets according to the conditions of the microfluidic droplet generation procedure. The droplets were collected in a capillary and incubated at 37 °C for 4 h. After incubation, the droplets were introduced into the detection chip through the droplet inlet as shown in Fig. 1(b). Meanwhile, fluorine oil was introduced into the chip through the oil inlet as shown in Fig. 1(b). The added fluorine oil can dilute the density of the droplets and focus the droplets into the middle of the detection channel. The fluorescence signal of FMB and enzyme-induced fluorescence products in droplets were detected by MFD.

III. RESULTS AND DISCUSSION

A. Performance of MFD

In this study, the enzyme-induced fluorescence product was fluorescein,²³ whose excitation and emission wavelengths are 490 and 525 nm, respectively. The excitation and emission wavelengths of the FMB were 660 and 690 nm, respectively. Therefore, only two of the four fluorescence detection channels were used in this study (green: 510−530 nm; red: 650−690 nm). The performance of the two channels was evaluated (Fig. S1 in the supplementary material). The green channel was used to detect the fluorescence of fluorescein generated by the enzymatic reaction. The detection limit was as low as 10⁻¹²mol/ml when fluorescein was wrapped in a 40 μm diameter droplet, and the coefficient of variation (CV) of the 10-min peak height of droplets was only 0.0210 (Table S1 in the supplementary material), indicating that the green channel had a good detection sensitivity and detection stability. The red channel was used to detect the fluorescence signal of FMBs. When FMBs were wrapped in droplets for detection, the average peak height of the fluorescence signal was 1.6384 V, which is much greater than three times that of noise. The red channel had good stability, and the CV of the 10-min peak height was 0.0513 (Table S1 in the supplementary material). All the above results indicate that the MFD used in this study performs well.

B. Detection of hCG β

1. Optimization of the reaction conditions

The concentration of FDG is an important parameter affecting detection. If the concentration of FDG is low, the number of positive signals is low, which affects the sensitivity of quantitative detection. Different concentrations of FDG were evaluated [Fig. S2(a) in the supplementary material]. When the concentration of FDG was less than 200 μmol/ml, the proportion of positive FMB increased with an increase in FDG concentration. When the concentration of FDG was more than 200 μmol/ml, the proportion of positive FMB no longer changed. Therefore, 200 μmol/ml was used as the experimental concentration of FDG. Limited by the sensitivity of MFD, a signal can only be detected when the concentration of fluorescein produced by the enzymatic reaction in the droplet reaches a certain level. Therefore, the reaction time between the FMB complex and FDG in the droplet is a key factor affecting the experiment. The effect of reaction time on positive signals was examined in this study. As shown in Fig. S2(b) in the supplementary material, the proportion of positive FMB increased with the increase in reaction time until 4 h but did not change after 4 h. Therefore, the reaction time of the FMB complex and FDG in the droplet was set to 4 h.

2. Interpretation of fluorescence signal

The optimized experimental conditions were used to detect hCG β with different concentrations, and the fluorescence signal is shown in Fig. 2(a). For the green channel, a signal greater than 0.15 V was defined as a positive signal, while for the red channel, a signal greater than 1.5 V was defined as a positive signal. When a droplet was detected, four combinations of fluorescence signals may appear: (1) Positive signals appear in the red and green channels; (2) positive signals appear in the red channel but not in the green channel; (3) there are no positive signals in the red and green channels; and (4) positive signals appear in the green channel but not in the red channel. If the first case occurs [Fig. 2(b)], this indicates that there is an FMB in the droplet and the enzyme is attached to the FMB; we defined this as a positive FMB. In the second case [Fig. 2(c)], there is also an FMB in the droplet. The difference is that there is no enzyme on the FMB, so we defined it as a negative FMB. The third and fourth cases indicate that there is no FMB in the droplet, and the signal of the fourth case [Fig. 2(d)] was defined as a false-positive signal. The number of FMBs was obtained by calculating the number of red channel positive signals. The recovery rate of FMB was approximately 10%. The proportion of positive FMB increased with the increase in concentration, which indicates that the proportion of positive FMB had a certain correlation with the concentration. The CV values of samples with different concentrations that were detected three times were acceptable (Table S2 in the supplementary material), indicating that the detection method has good stability.

FIG. 2.

(a) Spectrogram of hCG β with different concentrations. The concentrations used for three spectrograms were 0, 0.1, and 1.0 pg/ml from the bottom to the top. Signals marked with * indicate false-positives. Only part of the spectrum is shown here, as the original spectrum is too large to display. The spectrum shown here accounts for approximately 1/10 of the original spectrum. (b), (c), and (d) are typical spectra of positive FMB, negative FMB, and false-positive signals, respectively. In all figures, the signal of the red channel is represented by a red line, and the signal of the green channel is represented by a green line.

False-positive signals were observed in the experiment. As shown in Fig. 2(a), the signal marked with * indicates a false-positive signal. A false-positive signal may be caused by unknown particles of impurity that adsorb on β-Gal or free β-Gal wrapped in the droplet. The ratio of the false-positive signal to the green channel positive signal had a large CV in three experiments (Table S2 in the supplementary material), indicating that the probability of false-positive signals fluctuated greatly. If the interference of these false-positive signals is not excluded, the accuracy of detection could be affected; the proposed method can accurately exclude these false-positive signals.

Positive signals and false-positive signals existed simultaneously in the control samples [Fig. 2(a)]. The number of positive signals did not increase with the increase in false-positive signals (Fig. S3 in the supplementary material); therefore, we speculated that the main reason for the production of the positive signal in the control group was not due to FMB and free β-Gal being wrapped in a droplet simultaneously. The production of these positive signals may be related to the non-specific adsorption of FMB. These positive signals were deducted as background in subsequent quantitative calculations to improve their accuracy.

3. Quantification of hCG β

Previous studies have shown that when the proportion of positive FMB is between 0% and 70%, the average enzyme per bead (AEB) can be calculated according to its Poisson distribution. The formula is as follows:²⁴

$$\mathrm{AEB}=\text{-}\ln [1-{\mathrm{f}}_{\mathrm{on}}],$$

where f_on is the ratio of positive FMB to the total FMB tested.

In this study, when the concentration of hCG β was 0.1–10 pg/ml, the range of f_on was 1.3%–61%; therefore, AEB within this concentration range can be calculated by the formula mentioned above. More importantly, concentrations within this range are well linear with AEB. The standard curve with R²= 0.9925 is shown in Fig. S4 in the supplementary material, which can be used for quantitative analysis of hCG β.

C. Detection of hCG β in SECM

To evaluate the relationship between hCG β secreted by embryos and embryo quality, hCG β in SECM was detected by the method designed in this study. SECM was divided into three groups according to the category of blastocysts growing in it, which were with the top-quality blastocyst group (TQ group), non-top-quality blastocyst group (NTQ group), and non-blastocyst formation group (NON group). There was no significant difference in the basic conditions of the patients in the three groups (Table S3 in the supplementary material). Typical spectra are shown in Fig. 3(a). The concentration of hCG β in the three groups of the culture medium was 1.4733 ± 0.2736 pg/ml (TQ), 0.7731 ± 0.2311 pg/ml (NTQ), and 0.8383 ± 0.2920 pg/ml (NON). The concentration of hCG β in the TQ group was significantly higher than that in the NTQ group and NON group [Fig. 3(b)], which is consistent with previous results²⁵ indicating that the ability of the embryo to secrete hCG β was related to embryo quality. A limitation arose in the present study, small sample size, and the relationship between hCG β concentration and pregnancy outcome cannot be discussed. However, the experimental results show that the detection method developed in this study can be used to detect the protein secreted by embryos in SECM.

FIG. 3.

(a) Typical spectra of SECM in different groups. The three typical spectrograms from the bottom to the top belong to the TQ group, NTQ group, and NON group, respectively. The signals marked with * represent false-positives. The signal of the red channel is represented by a red line, and the signal of the green channel is represented by a green line. Only part of the spectrum is shown here, as the original spectrum is too large to display. The spectrum accounts for approximately 1/10 of the original spectrum. (b) Concentration distribution of hCG β in SECM of groups TQ, NTQ, and NON. ** indicates a significant difference between the two groups (p < 0.01). NS indicates no significant difference between the two groups (p > 0.05). p values were calculated by chi-square test.

Another standard curve (R²= 0.9922) was used to investigate the effect of false-positive signals on clinical test results; the curve did not exclude false-positive signals (Fig. S4 in the supplementary material). The standard curve exhibited a higher slope and a higher intercept than the standard curve that excluded false-positives. More importantly, each point on the standard curve exhibited a greater standard deviation, especially at lower concentrations, which may affect the accuracy of detection. Using this standard curve, the concentrations of hCG β in the samples were recalculated (false-positive signals were not excluded in the calculation). As shown in Table S4 in the supplementary material, 21 of the 30 samples had a lower hCG β concentration than when the false-positive signal was excluded; another 8 samples had a higher hCG β concentration, and only one sample had the same hCG β concentration. Since the difference in hCG β concentration between TQ, NTQ, and NON groups is large, the false-positive signal does not affect the determination of the difference between the groups. However, the presence of false-positive signals can lead to a change in the sequence of hCG β concentrations of different samples within the same group, which means that if hCG β concentration is used to select better clinical embryos, an inaccurate selection may occur due to the inclusion of false-positive signals.

IV. CONCLUSION

We have presented a method for quantitative protein detection based on microfluidic droplets and an MFD. This method does not need a microfluidic chip with a complex structure. The use of MFD makes the statistical analysis of data simpler and more intuitive. More importantly, the interference of false-positive signals was eliminated by double fluorescence simultaneous detection to ensure the accuracy of test results. Using this method, the hCG β secreted by embryos in SECM was measured. The detection limit was as low as 0.1 pg/ml, and only 30 μl samples were needed. Compared with non-top-quality blastocysts and embryos that did not develop into a blastocyst, top-quality blastocysts secreted more hCG β; therefore, hCG β could be useful as a biomarker to evaluate embryo quality. This study provides an alternative method for the detection of secretory proteins in SECM. This method also has certain application value in the study of biomarker screening in biological samples.

SUPPLEMENTARY MATERIAL

See the supplementary material for details of performance of MFD (Fig. S1 and Table S1), optimization of the reaction conditions (Fig. S2), the relationship between the number of positive signals and the number of false-positive signals in the control experiment (Fig. S3), the standard curve (Fig. S4), CV of hCG β samples detected (Table S2), basic conditions of the patients (Table S3), and the concentration of hCGβ in the sample (Table S4).