2.1.

$\beta$-Hematin Synthesis

$\beta$-hematin was synthesized using an acid precipitation method documented previously.²⁶ Specifically, nitrogen gas was constantly pumped into 0.1 M sodium hydroxide for 15 min to deoxygenize the solution. Then 50 mg hemin chloride (Sigma Aldrich 51280) was dissolved in 10 ml deoxygenized sodium hydroxide solution and was kept away from light. About 0.4 ml propionic acid was added drop by drop over a 20-min frame with the solution stirred at 200 rpm. The mixture was annealed for 18 h at 70°C. The amorphous aggregates were removed by washing with 0.1 M sodium bicarbonate alternating with water wash for three times. Then, the sample was washed with high performance liquid chromatography (HPLC) grade methanol (Sigma Aldrich 34860) alternating with water applied three more times. The solid products were dried over phosphorus pentoxide (Sigma Aldrich 431419) overnight to produce $\beta$-hematin crystal.

2.2.

Scanning Electron Microscopy

Scanning electron microscopy (SEM) of $\beta$-hematin was performed on a Zeiss Supra 55 VP system. $\beta$-hematin crystal suspensions were dispersed in ethanol and later dried on a silicon wafer. The specimen was mounted on aluminum stubs and coated with platinum for SEM imaging at an electron beam voltage of 3 kV and a beam aperture of $30\mu \mathrm{m}$.

2.3.

Time-Resolved Transient Absorption Microscopy

The setup of a time-resolved TA microscope is shown in Fig. S1 in the Supplementary Material. Our lab-built TA imaging system used a femtosecond pulsed laser (Spectra-Physics Insight) operating at 80 MHz with two synchronized outputs. The pump beam was modulated at $\sim 2.5\mathrm{MHz}$ using an acoustic-optical modulator (Isomet 1205-C). The probe beam went through a delay-tuning stage and combined with the pump beam. The combined beams were directed to a $60\times$, 1.2 numerical aperture (NA) water immersing objective (Olympus UPlanApo/IR). After passing through the specimen, the transmitted probe beam was collected using a 1.4-NA oil condenser (Olympus U-AAC), filtered with a bandpass filter, and detected by a photodiode (Hamamatsu 3994). The modulated TA signal was first amplified by a lab-built resonant circuit and then demodulated by a lock-in amplifier (Zurich Instrument HF2LI) with a time constant set to $7\mu \mathrm{s}$. To record the time-resolved TA signal, a delay-tuning stage was applied to tune the temporal delay between the pump pulse and the probe pulse with 66.7 fs per step. The pump–probe delay is scanned in a frame-by-frame manner to generate a three-dimensional data cube ($x-y-\tau$). A $200\times 200\times 120$ ($x-y-\tau$) data cube requires 48 s to acquire with a pixel dwell time set to $10\mu \mathrm{s}$. No photodamage was observed for RBC after data acquisition. Data were processed using ImageJ and analyzed using Origin software.

2.4.

Spontaneous Raman Spectroscopy

Spontaneous Raman spectroscopy of synthesized $\beta$-hematin was performed on a Horiba confocal Raman microscope (Horiba Scientific, Labram HR Evolution). A continuous wave laser at 785 nm was used for excitation. The laser was focused by a $40\times$ air objective with a laser power of 35 mW on the sample. Each Raman spectral acquisition was accumulated for 10 s. The pinhole size was set to $50\mu \mathrm{m}$ to reject the out-of-focus background. A $600\text{-}\text{grooves}/\mathrm{mm}$ grating was utilized to disperse the light and generate a Raman spectrum. Crystals of $\beta$-hematin were screened through transmission imaging to determine the point of interest for Raman spectroscopy.

2.5.

P. falciparum Culture and Sample Preparation

P. falciparum 3D7 strain was cultured by standard methods.²⁷ The parasites were kept in cell culture flasks (Corning 430825) containing blood (New York blood center) and Roswell Park Memorial Institute (RPMI) 1640 medium (Thermo Fisher 11875093) supplemented with 28 mM ${\mathrm{NaHCO}}_3$, 25 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) buffer solution, $25\mu \mathrm{g}/\mathrm{mL}$ gentamicin, and 0.5% AlbuMAX II (Life Technologies 11021-045). The culture flasks were kept in a cell culture incubator at 37°C. Culture medium was changed on a daily basis. P. falciparum 3D7 was diluted about twice per week to keep the parasitemia below 5%. During subculture, P. falciparum culture was synchronized using sorbitol.²⁸ The culture was then maintained as described above. P. falciparum fixation is done by immersing in 100% methanol (Sigma Aldrich 322415) for 4 s.

2.6.

Spectral Phasor Analysis

Time-resolved TA signals at each pixel were mapped into a two-dimensional (2-D) phasor space with the coordinates $g$ and $s$, which are the real and imaginary parts of the time-resolved TA signal after Fourier transformation, respectively.^24,29 The coordinates $g$ and $s$ can be expressed as $\mathrm{g}(\omega )=\int I(t)\times \cos (\omega t)\mathrm{d}t/\int |I(t)|\mathrm{d}t$, and $\mathrm{s}(\omega )=\int I(t)\times \sin (\omega t)\mathrm{d}t/\int |I(t)|\mathrm{d}t$, where $I(t)$ is the time-resolved TA signal and $\omega$ is the given frequency, which is set to 1.0 here. Clusters of hemozoin and RBCs in the phasor plot were separated manually, as shown in Fig. 3(c). Spectral phasor analysis was performed using an ImageJ plugin software.³⁰ To quantitatively analyze the TA decay rates of each cluster, the retrieved TA decay curves were approximated by fitting with a biexponential decay function as shown below: $y=\mathrm{A}1\times \exp (-t/{\tau }_1)+\mathrm{A}2\times \exp (-t/{\tau }_2)+y0$, where $t$ is the pump–probe delay, A1 is the amplitude, and ${\tau }_i$ are the decay time constants. The fast decay time constant ${\tau }_1$ was fixed at 0.38 ps as the system response time determined by the laser pulse duration. The TA decay rate was retrieved by fitting the slow decay time constant ${\tau }_2$.

3.1.

Time-Resolved Transient Absorption Imaging to Differentiate $\beta$-Hematin and Heme in Red Blood Cells

TA microscopy is shown to be able to map heme and heme-related products, such as imaging of heme dynamics in Caenorhabditis elegans²² and detection of glycated hemoglobin (HbA1c) in single RBCs.²⁴ Since hemozoin is synthesized from the polymerization of heme dimer,³¹ we tested whether a time-resolved TA signal can be used to differentiate the artificial hemozoin, $\beta$-hematin, from the heme in RBC.

3.1.1.

Characterization of $\beta$-hematin

We synthesized $\beta$-hematin from hemin chloride using an acid precipitation protocol.²⁶ SEM confirmed the synthesized $\beta$-hematin with the desired flake shape and with a size ranging from 0.5 to $2.0\mu \mathrm{m}$, similar to natural hemozoin [Fig. 1(a)].³² Spontaneous Raman scattering spectroscopy was performed to confirm the chemical composition. As shown in Fig. 1(b), the characteristic peak at $750{\mathrm{cm}}^{-1}$, which is assigned to the full pyrrole ring breathing modes ${\nu }_{15}$, indicates the $\beta$-hematin structure in the synthesized crystals.^33,34 Notably, the peak pattern from 1300 to $1600{\mathrm{cm}}^{-1}$ resembles the Raman spectrum of hemin [Fig. 1(b)],³³ indicating the remnant of hemin. However, the hemin remnant, which is likely in a powder format, did not confuse the experimental result later because individual $\beta$-hematin crystals were mixed with RBCs instead of hemin.

Fig. 1

TA microscopy spectroscopically differentiates $\beta$-hematin from hemoglobin in RBCs. (a) SEM image of synthesized $\beta$-hematin. (b) Raman spectrum of $\beta$-hematin: laser wavelength: 785 nm; laser power: 35 mW on the sample; dwell time: 10 s; objective: $40\times$ air; pinhole size: $50\mu \mathrm{m}$, grating: $600\mathrm{l}/\mathrm{mm}$. (c) Time-resolved pump–probe images of $\beta$-hematin and RBC mixture. Left panel: $\text{pump}/\text{probe}=520/680\mathrm{nm}$; right panel: $\text{pump}/\text{probe}=680/520\mathrm{nm}$; all images are at zero-time delay, dynamic range: 0.25 to 0.73; graphs are the decay curves of the $\beta$-hematin and RBCs indicated by the arrows (green and red, respectively); laser power: 10 mW for each beam; pixel dwell time: $10\mu \mathrm{s}$; $60\times$ water objective. (d) Phasor analysis outputs of the images in (c). (c)–(d) Left panel: $\text{pump}/\text{probe}=520/680\mathrm{nm}$; right panel: $\text{pump}/\text{probe}=680/520\mathrm{nm}$. (c)–(d) Top panel: retrieved images from phasor analysis ($\beta$-hematin is artificially colored green and RBCs in red); bottom panel: graphs are retrieved pump–probe spectra from phasor analysis.

3.2.

Time-Resolved Transient Absorption Imaging of Hemozoin within Infected Red Blood Cells

3.2.1.

Transient absorption signal of hemozoin and its dependence on excitation laser wavelength

Having proved that TA signals could separate $\beta$-hematin from RBCs, we next tested whether TA imaging could identify natural hemozoin within infected RBCs. We first optimized the hemozoin imaging condition with different pump–probe laser wavelength combinations. The optimization was performed under the constraint of our laser with one tunable beam ranging from 680 to 1300 nm and the other beam fixed at 1040 nm or 520 nm through frequency doubling. Since the absorbance difference of hemozoin and hemoglobin is largest at 680 nm,³⁵ we hypothesize that hemozoin gives a stronger TA signal when pumped at 680 nm compared with that at 520 or 1040 nm.³⁶ Indeed, we found that the $680/520\mathrm{nm}$ pump/probe combination obtained a higher signal-to-noise ratio (SNR) than the reversed combination (see Fig. S3 in the Supplementary Material). Later, we optimized the probe wavelength with the pump beam set to 680 nm. From the representative images, 1040-nm probing resulted in higher hemozoin signal normalized with the signals from the RBC [Fig. 2(a)], indicating that 1040 nm is a better probe wavelength. Finally, to confirm that 680 nm was the optimal pump wavelength, we carried out a pump wavelength dependency study. We fixed the probe beam at 1040 nm and compared SNRs generated from pump wavelengths ranging from 680 to 800 nm, with 10 nm increments [Fig. 2(b)]. The result shows that a pump wavelength of 680 nm optimizes the SNR. We conclude that the $680/1040\mathrm{nm}$ pump/probe combination is the optimal condition for in-situ TA imaging of hemozoin.

Fig. 2

Hemozoin TA signal dependence on excitation laser wavelength. (a) TA images of hemozoin using 680/1040 and $680/520\mathrm{nm}$ as the pump/probe wavelengths, respectively. Sample: RBCs infected with P. falciparum 3D7 strain; laser power: 10 mW for each beam; scale bar: $25\mu \mathrm{m}$; pixel dwell time: $10\mu \mathrm{s}$. Left panel: $\text{pump}/\text{probe}=680/1040\mathrm{nm}$; dynamic range: 0.1 to 1.8; right panel: $\text{pump}/\text{probe}=680/520\mathrm{nm}$; dynamic range: 0.1 to 0.7. (b) SNR of TA signal using different pump wavelengths. Sample was a single hemozoin crystal from RBCs infected with P. falciparum 3D7. Probe beam wavelength: 1040 nm; laser power: 10 mW for each beam, pixel dwell time: $10\mu \mathrm{s}$; SNR was calculated by first determining the intensity difference of a fixed area of hemozoin and an equal area of the RBC, then dividing the intensity difference by the standard deviation of the RBC.

3.2.2.

Time-resolved transient absorption imaging to identify natural hemozoin in 48-h infected red blood cells

With the optimized imaging condition, we then tested whether time-resolved TA imaging is able to differentiate hemozoin from hemoglobin in RBCs infected with late-stage P. falciparum. From the TA images at zero-time pump–probe delay, intracellular hemozoin crystal is clearly identified with a higher TA intensity [Fig. 3(a), left panel]. By plotting the pump–probe decay curves from the hemoglobin region of RBCs and hemozoin, which is indicated by the arrow in the left panel of Fig. 3(a), we found both showing TA decay patterns, but hemozoin differed substantially from RBC with around nine times larger TA intensity [Fig. 3(a), right panel]. We further applied phasor analysis to generate maps of hemozoin and RBC based on their TA decay rates [Fig. 3(b), left panel]. In the phasor plot from the uninfected RBCs dataset, we observed only one RBC main group, as shown in the red rectangle of Fig. 3(c). In the phasor plot from infected RBCs, the hemozoin group located lower than the main RBCs group was marked with a green rectangle [Fig. 3(c)]. The retrieved spectra revealed that the hemozoin crystal decayed faster than hemoglobin, with the TA decay time constants of $1.22\pm 0.12$ and $1.70\pm 0.11\mathrm{ps}$, respectively [Fig. 3(b), right panel]. We observed that the hemozoin mapping result is consistent with the higher TA intensity region in the zero-time pump–probe delay TA image [Fig. 3(b), left panel]. Collectively, these data demonstrated that even single-shot TA at zero temporal delay is able to in situ identify hemozoin from RBCs in late stages of malarial infection. In the next section, we used this simplified and faster approach for large area mapping of intracellular hemozoin in different infection stages.

Fig. 3

Time-resolved TA imaging and phasor analysis differentiates hemozoin from hemoglobin in RBCs. (a) Raw TA images and decay curves of malaria-infected RBCs. Sample: RBCs infected with P. falciparum; $\text{pump}/\text{probe}=680/1040\mathrm{nm}$; laser power: 10 mW for each beam. Left panel: images at zero-time delay; scale bar: $25\mu \mathrm{m}$; image display range: 0.02 to 0.5; RBC and hemozoin selected for generating decay curves were indicated with arrows. Right panel: raw TA decay curves of RBC and hemozoin; an area of about 100 pixels from both indicated objects were selected to plot the decay curves. The curves were smoothed by averaging among five nearby data points. (b) Mapping result retrieved from phasor analysis of data in (a). Left panel: retrieved images; red: RBC; green: hemozoin; scale bar: $25\mu \mathrm{m}$. Right panel: retrieved decay curves of hemozoin and RBCs; intensity was normalized to 1.0. (c) Phasor plot retrieved from phasor analysis of data in (a). Dots within red box: RBCs; dots within green box: hemozoin.

3.3.

In Situ Quantitation of Hemozoin within Different Infection Stages of Red Blood Cells

Having demonstrated the capability of a TA microscopy to in situ differentiate hemozoin from hemoglobin, we further explored how early in the malarial infection stage TA imaging could detect hemozoin. To this end, we tightly synchronized parasites (6-h window) and made fixed smears at 0, 12, 24, 36, and 48 h postinfection. We found that in the 0 h of infection, no intracellular hemozoin was detected. Instead, only small individual hemozoin crystals dispersing in the medium were observed [Fig. 4(a), 0 to 6 h]. These extracellular crystals were likely the remnants from the synchronization process.²⁸ We observed intracellular hemozoin crystal for cells after 12 h infection [Fig. 4(a), 2 to 18 h]. As infection proceeded, both crystal size and number increased [Fig. 4(a)]. Notably, contrary to the hemozoin, for which the TA signal increased with infection time, the hemoglobin signal continuously decreased as the infection progressed [Fig. 4(a)]. This observation demonstrates the power of TA imaging to visualize both hemoglobin degradation and hemozoin accumulation over the 48-h lifecycle of P. falciparum.

Fig. 4

Quantitative analysis of hemozoin in RBCs at different stages of malarial infection. (a) Representative single-shot TA images of infected RBCs. Sample: RBCs infected with P. falciparum 3D7 0, 12, 24, 36, and 48 h after invasion; since individual P. falciparum differs during synchronization and infection process, there is a range of 6 h at the beginning of synchronization, which expands to 12 h starting from 48 h after synchronization. Consequently, the time points were denoted as 0 to 6, 12 to 18, 24 to 30, 36 to 42 and 48 to 54 h; image was taken at zero-time delay; pump/probe: $680/1040\mathrm{nm}$; laser power: 30 mW for each beam; scale bar: $20\mu \mathrm{m}$; pixel dwell time: $10\mu \mathrm{s}$; (b) $7\times 7$ large area mapping of samples in (a). Images were artificially colored red; representative hemozoin crystals were indicated by arrows; scale bar: $100\mu \mathrm{m}$; displaying ranges were adjusted to standardize the RBCs: 0 to 6 h = 0.03 to 0.23, 12 to 18 h = 0.02 to 0.2, 24 to 30 h = 0.01 to 0.34, 36 to 42 h = 0.03 to 0.48, 48 to 54 h = 0.03 to 0.23. (c) The graph of the number of hemozoin crystals at different time points after infection; crystal number was normalized to 1000 RBCs. (d) The graph of the size of hemozoin crystal at different time points after infection. Hemozoin crystal size was determined by applying a threshold to filter out RBC background and then the total number of pixels of the hemozoin is converted into square micrometers.

We further performed large area mapping of thousands of RBCs in order to quantitatively analyze the growth of hemozoin in the rarely infected RBCs. With the estimated seeding parasitemia as 0.1%, and at least one hemozoin crystal produced by each parasite, we decided to sample at least 1000 cells to capture the rare event of hemozoin crystal formation in each image. Given that each single frame covers about 25 cells on average, a $7\times 7$ large area mapping covers more than 1000 cells. The 0- to 6-h sample contains fewer than 1000 cells due to low cell density, but this does not affect the quantification result because there was no intracellular hemozoin observed [Fig. 4(b)]. However, 12 h after infection, hemozoin could be detected and quantified [Fig. 4(b)]. Since the number of infected RBCs did change over the next 48 h, the number of hemozoin crystals only increased slightly over the course of infection [Fig. 4(c)]. In contrast, the crystal size increased substantially over time, reflecting the growth of hemozoin in each infected cell during mid-infection and late infection stages [Fig. 4(d)]. After 36 h, the average crystal size reached a plateau. Meanwhile, the number of hemozoin crystals started to increase again at 48 h, indicating the maturation and rupture of the first-generation P. falciparum and the invasion of new RBCs by the second generation [Fig. 4(c)]. These data collectively show the potential of using TA microscopy to quantitate intracellular hemozoin growth in size and number over a full lifecycle of P. falciparum infection.