Hydrophilic-treated plastic plates for wide-range analysis of Giemsa-stained red blood cells and automated Plasmodium infection rate counting

Plasmodium falciparum strain 3D7 was cultured as previously described [11]. Blood from a healthy donor was collected in BD Microtainer Tubes containing K2E (K₂EDTA) (BD Biosciences, Franklin Lakes, NJ, USA). Giemsa microscopy was also performed as described earlier [11]. For automated counting of infected parasites, the parasite-infected RBCs were stained with SYTO21 (Thermo Fisher Scientific, Waltham, MA, USA) at a final concentration of 5 µM for 10 min. Bright field and fluorescence images of parasite-infected RBCs stained with SYTO21 were acquired using an inverted fluorescence microscope (DM1L, Leica Microsystems, Wetzlar, Germany) with a digital camera (MC120 HD, Leica Microsystems) and analysed using MetaMorph Offline software (ver. 7.8, Molecular Devices, Sunnyvale, CA, USA).

Plastic plates (25 × 75 mm) made of cyclic olefin copolymer (COC, Shin-Etsu Polymer Co. Ltd., Tokyo, Japan) were rendered hydrophilic by reactive ion-etching treatment using a SAMCO RIE system (SAMCO Inc., Tokyo, Japan) [11‐13]. The treatment was performed for 15 min. The effect of reactive ion etching on the plate surface was examined by measuring the contact angle of water on the plate surface using a contact angle meter (Kyowa Interface Science Co. Ltd., Saitama, Japan). The contact angle of the plates was 18.2°, and the effect on the plates lasted for more than 2 weeks.

Because the adhesiveness of cells to plastic plates treated with reactive ions is increased, the hydrophilic-treated COC plates could be used for preparation of RBC monolayers. As described in Methods, reactive ion-etching treatment was performed on slide-glass shaped COC plates (Fig. 1a). 3 ml of P. falciparum-infected RBCs (1% parasitaemia) diluted to 1% of haematocrit with RPMI1640 medium were dropped onto the hydrophilic-treated COC plates. The plates were then left undisturbed for 10 min to allow the RBCs to settle and adhere onto the plate (Fig. 1b, e). Non-adherent RBCs were removed by rinsing the plate for 10 s with a medium containing 10% ethanol (v/v), which also facilitated the drying of adherent RBCs in subsequent steps, and the RBCs then formed a monolayer on the plate surface (Fig. 1c, f). RBCs detached from the plate if higher concentrations of ethanol were used in the rinsing medium. The monolayer, which covered the entire plate, was immediately dried using a hair dryer (Fig. 1d) to avoid disrupting the morphology of the RBCs or the parasites. The dried plates were stained by the conventional Giemsa staining method. Figure 1g shows a hydrophilic-treated COC plate after Giemsa staining. The RBCs were stained uniformly across the plate, which was significantly different from that observed using the conventional technique, in which a thin blood smear is stained. Most of the RBCs were attached to the plates because of immediate drying using a hair dryer. However, if they were not dried immediately, the RBCs were removed by methanol treatment. A hair dryer with a powerful motor should be used to apply air from directly above the plate.

When the plates were observed microscopically, a monolayer of non-aggregated RBCs was visualized all over the plate surface (Fig. 2A, B). Parasite-infected RBCs were observed using an oil immersion lens (Fig. 2C). The contact angle of water on the hydrophilic-treated COC plates was also measured. The concept of the contact angle is described in the Additional file 1: Figure S1). Figure 2F shows a Giemsa-stained plate with a contact angle of 18.2° (Fig. 2A–C, F); in contrast, the contact angle of the untreated plate was 111.9°. The RBCs formed a monolayer and were sparsely distributed across the hydrophilic-treated plate, which enabled parasite identification and accurate enumeration. Next, plates with a contact angle of 38.6° (Fig. 2D) or 31.0° (Fig. 2E) using weaker reactive ion etching were prepared, and these plates were used for Giemsa microscopy. There was a dense population of RBCs on the plate, with a contact angle of 38.6°, which was not suitable for microscopic analysis (i.e., most RBCs were aggregated). However, cell distribution on the plate with a contact angle of 31.0° was well suited for Giemsa microscopy. Giemsa staining on the whole surface of a glass slide was also performed as described in Fig. 1b–d. The contact angle of water on the conventional glass slide was 7.8°. The RBCs were aggregated and did not form a monolayer suitable for Giemsa microscopy on the slide. Additionally, diluting the RBCs to 0.1% haematocrit did not reduce aggregation.

Next, whether a uniform monolayer formed on the hydrophilic-treated plate when whole blood was used as the sample was evaluated. In this study, blood from a healthy donor was diluted 100-fold with RPMI 1640 medium followed by staining with Giemsa solution, as described in Fig. 1. Figure 3a shows the hydrophilic-treated COC plate after Giemsa staining, which is similar to the plate applied for cultured Plasmodium-infected RBCs (see Fig. 1g). Microscopic analysis with 1000× oil immersion revealed that RBCs formed a uniform monolayer on the plate and Giemsa-stained leukocytes formed a monolayer (Fig. 3b). These results indicate that RBCs in whole blood form a uniform monolayer on the plate and can be used for Giemsa staining. Therefore, the method may be suitable for use with Giemsa microscopy to diagnose patients with malaria.

Determination of the parasite infection rates is essential for malaria diagnosis as well as for research, but counting the number of RBCs and infected parasites by microscopic examination is laborious and time-consuming. Hence, a method for automated infection rate counting on hydrophilic-treated COC plates was developed. RBCs diluted to 1% hematocrit with RPMI1640 medium were added to the hydrophilic-treated COC plate with frame(s) composed of plastic, allowing 10 min for the RBCs to settle and adhere to the COC plate (Fig. 4a, b). Plastic frames of various sizes can be attached with double-sided tape and nail polish, depending on the experimental purpose (Fig. 4c). Non-adherent RBCs were removed by rinsing with medium, and the remaining RBCs formed a monolayer on the plate surface.

Figure 5a (left panel) shows a bright field image of an RBC monolayer on the COC plate. RBCs were sufficiently dispersed for accurate counting. The dimensions of the RBCs were very similar (approximately 7 µm in diameter), and the colour of the centre of the RBCs appeared white in the picture taken with a digital microscopic camera. These features were useful for automated RBC counting with a software tool. RBCs were specifically detected by integrated morphometry analysis (total area > 25 and 0.8 < shape factor < 1.0) in MetaMorph Offline software (ver. 7.8, Molecular Devices, Sunnyvale, CA, USA) (Fig. 5a, right panel). Concerning the five bright field images, the number of RBCs was counted manually and automatically to compare the values obtained by the two methods (Fig. 5b). The number of RBCs was not significantly different between the two counting methods (p = 0.48). The fluorescence signal intensity of the reticulocytes stained with SYTO21 (a dye used for staining the parasites, see below) was much weaker and larger in size than that of the parasites. Because the signals corresponding to reticulocytes were omitted in the analysis using MetaMorph software, reticulocytes did not show false-positive results. Howell–Jolly bodies were not observed in this study. It required less than 1 min to perform the automated counting. Taken together, these data indicate that accurate automated RBC counting can be performed easily and quickly with the hydrophilic-treated COC plate surfaces and MetaMorph Offline software.

Next, the automated counting of Plasmodium-infected RBCs on hydrophilic-treated COC plates was attempted. Since, unlike parasites, RBCs do not have nuclei, it is possible to detect parasite-infected RBCs by nuclear staining. SYTO21, a cell-permeant green fluorescent dye, was used for staining of the parasite nucleic acids. The parasite-infected RBCs, suspended in the medium containing SYTO21, were added to hydrophilic-treated plates, followed by staining for 10 min and removal of non-adherent RBCs. Non-synchronized parasites, in which more than 60% of the parasites were in the ring stage and whose fluorescence signals were weaker than those of parasites in other stages, were used in this study. Figure 6a (upper left panel) shows a representative bright field image of the parasite-infected RBCs. The RBCs formed a monolayer, which was then available for automated counting. Figure 6a (upper right panel) presents a fluorescence image of the same field. Infected Plasmodium nuclei stained with SYTO21 were detected (arrows). The signals were not identified in uninfected RBCs (Fig. 6a, lower panels).

Automated counting of the detected fluorescence signals was possible using the MetaMorph Offline software, and the automated counting was performed with Integrate Morphometry Analysis (total area > 10, and 0.4 < shape factor < 1.0). When the data obtained were combined with the number of RBCs automatically counted as described in Fig. 5, the infection rate could be calculated easily \(\left[ {{\text{infection rate }}\left( \% \right) = {\text{number of fluorescence signals}}/{\text{number of RBCs}} \times 100} \right]\). Eleven Plasmodium-infected RBC samples with various infection rates (from 0.10 to 2.15%) were analysed by Giemsa microscopy and the automated counting technique to compare the two counting methods (Fig. 4b). The infection rates calculated by these two methods exhibited a positive correlation (R² = 0.97). These results indicate that automated counting of the parasite infection rate is possible without preparation of a thin blood smear, Giemsa staining, or microscopic observation.