Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2022 Mar 18;94(12):4988–4996. doi: 10.1021/acs.analchem.1c04754

Looking Inside Non-Destructively: Label-Free, Raman-Based Visualization of Intracellular Coxiella burnetii

Nancy Unger †,‡, Simone Eiserloh †,‡, Frauke Nowak ∥, Sara Zuchantke ∥, Elisabeth Liebler-Tenorio ∥, Katharina Sobotta ⊥, Christiane Schnee ∥, Christian Berens ∥, Ute Neugebauer †,‡,§,*
PMCID: PMC8974703  PMID: 35302749

Abstract

graphic file with name ac1c04754_0006.jpg

The life cycle of intracellular pathogens is often complex and can include different morphoforms. Treatment of intracellular infections and unperturbed studying of the pathogen inside the host cell are frequently challenging. Here, we present a Raman-based, label-free, non-invasive, and non-destructive method to localize, visualize, and even quantify intracellular bacteria in 3D within intact host cells in a Coxiella burnetii infection model. C. burnetii is a zoonotic obligate intracellular pathogen that causes infections in ruminant livestock and humans with an acute disease known as Q fever. Using statistical data analysis, no isolation is necessary to gain detailed information on the intracellular pathogen’s metabolic state. High-quality false color image stacks with diffraction-limited spatial resolution enable a 3D spatially resolved single host cell analysis that shows excellent agreement with results from transmission electron microscopy. Quantitative analysis at different time points post infection allows to follow the infection cycle with the transition from the large cell variant (LCV) to the small cell variant (SCV) at around day 6 and a gradual change in the lipid composition during vacuole maturation. Spectral characteristics of intracellular LCV and SCV reveal a higher lipid content of the metabolically active LCV.

Introduction

Intracellular pathogens are well adapted to withstand conditions inside the host cell and are often difficult to treat. Their life cycle is frequently complex and includes different morphoforms. Unperturbed studying of the intracellular pathogen in its niche remains difficult due to the limited access for direct characterization. Current invasive imaging methods to study the infection process in vitro comprise fluorescence-based imaging and electron microscopy, which require either preceding staining steps or destructive sample preparation. Most metabolic, genomic, and transcriptomic studies require purification of the bacteria from the host cell or at least destruction of the cell, so that single-cell-resolved analysis is often difficult to achieve.1,2

New, label-free and non-destructive methods that provide unprecedented insights into intact cells and tissue could help better understand pathogenesis. Raman spectroscopy-based methods hold high potential as they can reveal molecular differences by investigating inelastically scattered light in a label-free, non-invasive, and non-destructive manner.3 In combination with multivariate data analysis, Raman-derived chemical information can be used to generate false-color images, resolving cell compartments such as cytoplasm, nucleus, and lipid droplets.4 Raman spectroscopic fingerprints can serve to differentiate bacterial species and strains5 and have successfully been applied to visualize intracellular infections.6 Recent advances in Raman scanning and analysis methods enabled quantitative insights into subcellular distributions of organelles and lipid droplets7 and confocal imaging of living zebrafish.8 Here, we advance diffraction-limited 3D confocal Raman imaging not only to visualize and quantify bacterial morphoforms and lipids in high-quality false color Raman images but also to use the specific Raman fingerprint of intracellular bacteria to gain insights into their life cycle.

As model organism, we use the obligate intracellular, Gram-negative bacterium Coxiella burnetii that infects not only domestic livestock, such as sheep, goats, and cattle, but also many wild animal species.9,10 Following inhalation of contaminated aerosols, C. burnetii can cause the zoonotic disease Q fever in humans.1 Diagnosis of acute Q fever is difficult and relies on serology using immunofluorescence assays and enzyme-linked immunosorbent assays, as well as polymerase chain reaction-based tests.9,11C. burnetii has a biphasic life cycle with two distinct developmental stages: a metabolically active large cell variant (LCV), which can exceed 0.5 μm in diameter and 1 μm in length,12 and a spore-like small cell variant (SCV).13,14 SCVs have a diameter of only ∼0.3 μm and show high resistance toward environmental stressors. After infection of the host cell by the SCV, the bacteria reside within a phagosome-derived vacuole, called Coxiella-containing vacuole (CCV) or parasitophorous vacuole. The initial parasitophorous vacuoles are formed by homotypic fusion of multiple C. burnetii-containing vacuoles.15 The conversion from SCV to the metabolically active LCV takes place inside this vacuole within the first 16 h post infection (p.i.). After day 1 p.i., LCVs are found almost exclusively in differently sized vacuoles.15,16 After a lag phase, active replication of the LCVs in large and spacious vacuoles becomes evident at day 2 p.i.15 Exponential growth with doubling times of 20 to 45 h17 and in vitro condition of 15 to 37 h18 is reported until roughly 1 week p.i. Afterward, the morphogenesis of LCVs to SCVs occurs, and 2 weeks post infection, the vacuole is nearly homogeneously filled with SCVs, and the stationary phase of the infection is reached. After differentiation, C. burnetii is no longer able to replicate.19

Our new Raman-based imaging algorithm revealed biochemical changes during intracellular bacterial development in a cell culture model with diffraction-limited resolution in a time-resolved manner. Localization and quantitative analysis of the two morphoforms of C. burnetii and biochemical insights into lipid compositions can help characterize the infection process.

Materials and Methods

Sample Preparation for Isolated C. burnetii

Buffalo green monkey kidney (BGM) cells from the collection of cell lines in Veterinary Medicine at the Friedrich-Loeffler-Institut were used as host cells for Coxiella infection. BGM cells have been used routinely to propagate and study C. burnetii for many years and proven to be an adequate model, for instance, in terms of gene expression compared to in vivo experiments.20C. burnetii RSA439 Nine Mile phase II21 was obtained from Dr. Katja Mertens-Scholz from the Institute of Bacterial Infections and Zoonoses of the Friedrich-Loeffler-Institut. SCVs and LCVs were isolated, as described by Coleman et al.15 BGM cells were seeded in 75 cm2 cell culture flasks and incubated in Eagle’s modified essential medium (EMEM) (Lonza) at 37 °C and 5% CO2. 24 h later, EMEM was exchanged with UMDCK medium (Lonza), and cells were infected with C. burnetii RSA439 Nine Mile phase II at a multiplicity of infection (MOI) of 100. UMDCK medium was renewed 1 day post infection (dpi). The cells were then incubated for 4 days at 37 °C and 5% CO2 for the isolation of LCVs. For the isolation of SCVs, infected cells were incubated for 7 days at 37 °C and 5% CO2 and then for 21 days at room temperature with flask caps tightened. The bacteria were isolated by centrifugation at 15,000g/15 min/4 °C, followed by resuspending the pellet in 10 mL sterile water to lyse the cells. The lysed cells were centrifuged at 1000g/5 min/4 °C to remove the cell debris. The supernatant was again centrifuged at 15,000g/15 min/4 °C to harvest the bacteria.

As control, the same isolation procedure was performed with uninfected BGM cells. The fragments from this isolation were analyzed as control samples. The isolation procedure was repeated four times with independent biological samples, in the following called batches.

Prior to Raman analysis, the isolates were washed once in 200 μL phosphate-buffered saline and twice in 200 μL sterilized water and centrifuged at 14,500g/30 min at room temperature. After resuspension, a droplet of 10 μL was placed on the middle of a CaF2-slide (Crystal GmbH) and allowed to dry at room temperature.

Sample Preparation for C. burnetii-Infected BGM Cells

CaF2 microscope slides (9 × 11 × 1.5 mm3, crystal) were incubated overnight in 70% ethanol, rinsed with EMEM (Lonza), and placed in 24-well cell culture plates. BGM cells were seeded on the slides and incubated in EMEM at 37 °C and 5% CO2. 24 h later, EMEM was exchanged with UMDCK medium (Lonza), and cells were infected with C. burnetii RSA439 Nine Mile phase II at a MOI of 25. UMDCK medium was renewed 1 day post infection (dpi). Cells were chemically fixed at 1, 4, 6, 8, and 12 dpi in 2 mL of 4% formaldehyde overnight and stored in PBS at 4 °C until spectral analysis. For spectral analysis, CaF2 slides were transferred into small Petri dishes (diameter 3.5 cm) filled with PBS.

Raman Measurements

An upright confocal Raman microscope (WITec alpha300 R) equipped with a 600 g/mm (BLZ = 500 nm) grating (set at a spectral center of 2200 cm–1) and a DV401_BV CCD camera was used to record the Raman data. A frequency doubled Nd:YAG-laser (532 nm) giving 15 mW in the sample plane served for Raman excitation.

Image scans (6 × 6 μm2 with 69 × 69 points per line) of the dried isolated Coxiella morphoforms (and the fragments from uninfected BGM cells) were recorded with 1 s integration time per pixel at different sample positions. The Raman signal was collected through a 100× objective (NA 0.75, Zeiss) and guided through a 100 μm fiber.

Infected BGM cells were analyzed in aqueous medium (PBS) with a 60× water immersion objective (NA 1, Nikon). Individual cells with CCVs were selected using the white light image. In order to achieve the highest spatial resolution in z, a 25 μm collection fiber served as a pinhole. Raman maps were recorded with a step size of 0.2 μm covering an area of 20 × 20 μm2. Six layers were recorded with a spacing of 0.5 μm in z (three layers above the focal plane and two below). Integration time per pixel was 1 s. Our Raman system provides diffraction-limited resolution.22 Using Abbe’s formula, we obtain a lateral resolution of dxy = λ/2 NA = 532 nm/2 = 266–300 nm. This resolution was proven in other experiments outside the scope of this manuscript (data not shown). The axial resolution is estimated with Abbe’s formula to be dz = 2 λ/NA2 = 1064 nm to 1 μm, which agrees with the information given on the manufacturer’s homepage.

Analysis of Raman Data

Gnu R with the packages “hyperSpec” (Beleites, Sergo, 2018, http://hyperspec.r-forge.r-project.org hyperSpec), “MASS” (Venables, Ripley, 2002), “Spikes” (Ryabchykov, 2016), “pls” (2019, https://CRAN.R-project.org/package=pls Pls), and “unmixR” (McManus, 2013) were used to pre-process and analyze the Raman data. Prior to analysis, cosmic spikes were removed, and the silent region between 1800 and 2500 cm–1 was cut out.

Analysis of Isolated Morphoforms

Raman spectra of isolated samples were baseline corrected using a third-order polynomial and vector-normalized. For further analysis, the data set was split into four parts (one part being one independent batch). Three batches were used for training and one batch for prediction using leave-one-batch-out cross validation. Partial least squares regression (PLS-R) with the training data was set up with three groups: isolated LCV, isolated SCV, and fragments from uninfected BGM cells. The optimal number of PLS components was chosen based on the cross-validation results (score plots). PLS-components were used to train a linear discriminate analysis (LDA) model. Spectra of isolated SCV and LCV of the left-out batch were fed into this model, and their assignment to the trained groups (SCV, LCV, and BGM) was predicted. Raman spectra predicted as SCV and LCV were averaged to yield the characteristic Raman spectrum with standard deviation of the respective morphoform. The difference spectrum of the developmental forms with standard deviation was calculated by subtracting the mean spectrum of LCVs from that of SCVs.

Multivariate Data Analysis of Infected BGM Cells

Each of the six z-planes of the spectral maps was first treated individually using principal component analysis (PCA) to reduce noise and dimension. The optimal number of PCs was chosen based on the spectral information in the loadings. Therefore, the number of PCs can be different for the individual maps. The N-FINDR algorithm23 was used for spectral unmixing to reveal the most different spectra, called endmember, within the data set with the aim to identify the different chemical components within the image stack. These chemical components are LCV, SCV, DNA (e.g., host cell nucleus), cytoplasm of the host cell, lipid (e.g., in lipid droplets), and PBS/water of the surrounding medium. Depending on the time point p.i. and the imaged region (20 × 20 × 6 μm3), not all components could be identified within a z-stack. The optimal number of endmembers was selected based on the expected number of biological compartments of the analyzed area in connection with the subjective assessment of the results, so that no inexplicable splitting of these compartments occurs due to the division of one endmember into two or more. For most measurements, the analysis revealed four endmembers. In some cases, three endmembers were enough to represent the full spectral characteristic of the recorded section. These pure endmember spectra can be used to generate false color abundance maps. The color of each voxel in the six-layer image stack was calculated as a superposition of the involved endmember spectra.

In order to extract characteristic Raman spectra of intracellular components (e.g., LCV, SCV, or lipid droplets), Raman spectra of all pixels of an image stack that contained the respective endmember component to at least 80% were extracted with software Mathematica (version 10.2). Polynomial baseline correction was performed after extraction.

For quantitative analysis, each vacuole was analyzed separately. Only the voxels of the CCV contained in the measured image were chosen as region of interest (ROI), and the number of spectra inside the ROI was set to 100% (each pixel in the six z-planes contributed one spectrum). Bacterial occupation of the vacuole was estimated based on the sum of the number of spectra inside the ROI assigned to SCV and LCV.

An algorithm based on the Michelson contrast was used to calculate the relative ratio (SCV – LCV)/(SCV + LCV) in order to compare SCV and LCV quantities. The resulting scale (−1 to +1) was converted to 0% to 100% for visualization of the relative filling level.

Results and Discussion

Raman Spectroscopic Fingerprint of the Two Morphoforms of C. burnetii

The metabolically active LCV and the spore-like SCV of C. burnetii were extracted from their host cells, and the isolated, enriched morphoforms were characterized by means of Raman spectroscopy (Figure S1). To exclude the spectral contributions from contaminating host cell fragments in the spectra of the Coxiella morphoforms, LDA was performed using Raman spectra from different isolation days (Figure S2). Raman mean spectra of SCV and LCV morphoforms (Figure 1A) were computed from the Raman spectra predicted with the LDA model to be SCV and LCV, respectively. The computed difference spectrum (SCV minus LCV, Figure 1C) reveals the same spectral characteristics as the linear discriminant (LD) coefficient, separating LCV and SCV in the LDA model (Figures 1B and S2).

Figure 1.

Figure 1

Open in a new tab

Raman mean spectra of C. burnetii morphoforms and 3D visualization of C. burnetii vacuole sections inside intact BGM cells at 4 dpi. (A) Mean Raman spectra of isolated SCV (isolated 28 dpi) and LCV (isolated 4 dpi). (B) Raman difference spectra computed by subtracting LCV spectra from spectra of SCV. Red: computed difference using spectra of isolated Coxiella morphoforms (SCV, 28 dpi minus LCV, 4 dpi); gray: computed difference using Raman spectra extracted from intracellular pathogens during whole cell analysis (SCV, 12 dpi minus LCV, 4 dpi). These difference spectra show characteristics similar to the LD2 coefficient displayed in the inset (see also Supporting Information S2). Spectral features of an SCV correlate to positive values of the LD 2 coefficient, while LCV spectral features assume negative values. The red spectrum depicts the computed difference spectrum between SCV and LCV using Raman spectra predicted to be SCV and LCV, respectively. (C) Stack of false color Raman images of three C. burnetii vacuoles inside three intact BGM cells 8 days post infection generated using N-FINDR analysis (see Figure 3A for white light image of the section). (D) The four endmember spectra can be assigned to PBS/water (cyan), SCV of C. burnetii (violet), LCV of C. burnetii (orange), and lipids (dark green). The PBS endmember (cyan) is located in the spaces between the three vacuoles (in panel C) and indicates the presence of medium which surrounds the host cells. The lipid endmember (green) is found as droplets in close vicinity to the vacuoles and also inside as the result of digestion. The SCV endmember is found in two out of the three vacuoles. These two vacuoles are almost homogeneously filled with SCV. The third vacuole contains mainly LCV, indicating non-synchronous infection. (E) Mean spectrum (and standard deviation) from all pixels in the 3D stack in panel (C), in which the LCV (or SCV) endmember spectrum is at least 80% present, respectively. The inset shows the 3D visualization of the pixels fulfilling this criterion and, thus, the abundance of LCV (in orange) and SCV (in violet) inside the vacuoles of the host cells in panel (C). Similar analyses were performed for time points 1, 4, 6, and 12 days p.i. (Figure S3).

Raman features, that are more prominent in the LCV form than in the SCV form, include Raman bands at 720 cm–1 (C–H in-plane vibration), 1090 cm–1 (phosphodioxy group and C–C stretch), 1300 cm–1 (CH2 in plane twist vibrations in lipids), 1440 cm–1 (C–H deformation), and 1655 cm–1 (amide I and C=C). This points to chemical components that show higher relative abundance in the metabolically active LCV, in particular lipids and proteins. The Raman difference spectrum (Figure 1C) and the loading coefficient of LD2 show hardly any positive bands, only one around 1677/75 cm–1. This vibrational band has been assigned not only to condensed chromatin structures24 but also to peptidoglycan.25 The spectral findings agree with previously published studies on the chemical composition of SCV and LCV. Condensed chromatin and a thicker peptidoglycan layer have been reported for the SCV form,19 whereas the metabolically active LCV form shows a higher content of phospholipids26 and an increased level of gene expression.19 In connection with the negative band at 1655 cm–1, the positive band around 1677/75 cm–1 might also indicate a higher relative abundance of beta-sheet-containing proteins in SCV compared to LCV, which could have more proteins with alpha-helix secondary structure. Previous studies analyzing the protein composition of the SCV and LCV developmental forms have found different protein profiles for both variants (e.g., ref (13) and (27)); however, no systematic study is available, which analyzes the overall secondary structure of the proteins present in the two different morphoforms.

Label-Free Visualization of C. burnetii within Intact Infected Cells in 3D

False color Raman images of infected host cells provide detailed chemical information with diffraction-limited spatial resolution (in our case: ∼300 nm in xy, as estimated using the Abbe resolution limit) and enable the visualization and characterization of the bacteria directly within intact host cells, as exemplarily shown in Figure 1C (and Figure S3). No prior treatment or staining had to be performed as Raman spectroscopy directly probes intrinsic molecular properties. With N-FINDR analysis, endmember spectra were extracted from the Raman data and assigned to components inside the intact host cell: cytoplasm of the host cell, nucleic acids, lipids, and the two morphoforms LCV and SCV of C. burnetii (Figures 1D and S3). The assignment is based on the spectral characteristics (Figure S4) and supported by the spatial distribution of the endmember spectra in the map (e.g., Figure 1C).

The endmember assigned to the cytoplasm captures the shape of the eukaryotic host cell. In 3D false color Raman stacks, the CCV seems to stick out of the otherwise rather flat cell like the yolk in a fried egg (e.g., Figure S3A). The nucleic acid/DNA endmember with the characteristic phosphodiester backbone and nucleic acid base vibrations shows high abundance at the position of the host cell nucleus. With lower intensity, it is also found inside the parasitophorous vacuole, showing the presence of bacterial nucleic acids. Lipids with intense and sharp Raman bands (Figure S4) are found near the membrane of the parasitophorous vacuole in the form of small lipid droplets and also inside the vacuoles. The presence of lipid Raman spectra near and in the CCV is in agreement with the active recruitment of host cell lipids by C. burnetii.28

Endmember spectra assigned to C. burnetii were found inside the vacuole. For LCV and SCV, the average Raman spectra of the isolated bacteria (Figure 1A) were used as reference spectra to differentiate the two morphoforms in an automated manner directly within the intact, infected host cells. At day 4 post infection, spectral features are characteristic for LCVs (Figure S3D); at day 12 p.i., they resemble those for SCVs (Figure S3H). This correlates with transmission electron microscopy (TEM) findings in these cells (Figure S5) and agrees with the published life cycle of C. burnetii.19 Computed averaged Raman spectra of all intracellular LCV and all SCV (Figure 1D) show excellent agreement with the spectra of the isolated, physically extracted bacteria (Figure 1A). Computed difference spectra (SCV minus LCV) using spectral data from the isolated bacteria and the intracellular bacteria are almost identical (Figure 1B), demonstrating that the spectral characteristics of the different morphoforms of C. burnetii can be recognized in a non-destructive, label-free manner directly inside intact host cells.

Furthermore, cellular debris taken up by the CCV can be identified in Raman false color images and is also clearly seen in TEM micrographs at early infection time points (Figure 2).

Figure 2.

Figure 2

Open in a new tab

Visualization of cellular debris in Raman false color images and TEM micrographs at early infection time points. Left: False color Raman images of a z-stack, showing a CCV at 1 dpi. Orange arrows indicate C. burnetii assigned to LCV. Black arrows point to “particles” with typical spectral features of DNA and cytoplasm that could originate from cell debris (taken up by the Coxiella-containing parasitophorous vacuole). (B) TEM micrograph of a different cell at 2 dpi. Cellular debris and membrane fragments (examples indicated by long arrows) are present between LCVs (examples indicated by short arrows) in a CCV. Bar = 3 μm, both in Raman and TEM images.

Following the Infection Cycle within Intact Host Cells and Quantitative Insights

Growth phase information of the intracellular pathogen can be extracted directly from infected host cells. Representative white light images of the cells analyzed at different days post infection (1, 4, 6, 8, and 12 dpi) are shown in Figure 3A. False color Raman images (Figure 3B) and 3D visualization of SCVs and LCVs inside their host cells were used to qualitatively and quantitatively analyze the intracellular development of C. burnetii (Figures 3, Table 1, and Figure S3). Analysis was performed at the single host cell level to reveal cell-to-cell and vacuole-to-vacuole variations. Findings were confirmed by TEM of infected cells at selected days post inoculation.

Figure 3.

Figure 3

Open in a new tab

Quantitative analysis of the relative frequency of LCV and SCV during the infection process. Typical images are shown for 1, 4, 6, 8, and 12 days post infection (dpi). (A) White light images of the infected BGM cells at different dpi. The red square indicates the CCV section investigated by means of spectroscopy. (B) False color Raman images from one z-plane highlighted in panel (A). The color code is the same throughout the manuscript: cyan: PBS/water, red: DNA/nucleus, orange: LCV of C. burnetii, violet: SCV of C. burnetii, dark green: lipids, and light green: cytoplasm of the host cell. (C, D) 3D visualization of the LCV and SCV distribution inside the CCV. Pixels that contain the respective endmember for SCV and LCV to at least 80% were assigned to contain C. burnetii in the respective morphoform. (E) The extracted Raman spectra show the transition from LCV to SCV. N-FINDR analysis of 6 dpi always revealed two different bacterial endmembers in one vacuole. In their spectral appearance, they do not represent 100% the pure forms as found at 4 and 12 dpi and also contain mixed spectral properties. Nevertheless, the two endmembers can always be assigned to one of the two morphoforms due to their spectral characteristics at 1440 cm–1 and in the range from 2800 to 3100 cm–1. For clarity in the presentation, we have assigned these spectra to SCV or LCV and refrained from introducing another state. (F) Relative filling of CCV analyzed against the post-infection time. Shown is the percentage of pixels within the imaged CCV, which could be assigned to C. burnetii (both LCV and SCV) using endmember spectra. (G) Evaluation of the relative quantity of SCV compared to LCV over time. Shown is the computed ratio of pixels assigned to SCV and LCV during N-FINDR analysis. Data shown in panels (E,F) was obtained from three different host cells per time point yielding in total 21 CCV from five time points: 1 dpi: four vacuoles from three cells, 4 dpi: five vacuoles from three cells, 6 dpi: three vacuoles from three cells, 8 dpi: eight vacuoles from three cells, and 12 dpi: five vacuoles from three cells.

Table 1. Quantitative Insights Into Intracellular Development of C. burnetii in BGM Cells Using Single-Vacuole Analysis.

time point (dpi) cell vacuole number dimension of CCV (μm × μm)a relative degree of occupation (%)b relative amount of LCV in the vacuole (%)c relative amount of SCV in the vacuole (%)d
1 1   14 × 13.5 53.75 53.75 not detected
  2   9 × 7 45.52 45.52 not detected
  3 1 10.5 × 11 29.13 29.13 not detected
    2 11 × 10 8.43 8.43 not detected
4 1   27.5 × 26.5 14.27 14.27 not detected
  2   35 × 31.5 27.4 27.4 not detected
  3 1 21.5 × 21.5 33.42 33.42 not detected
    2 11 × 18 42.71 42.71 not detected
    3 23 × 19.5 16.86 16.86 not detected
6 1   21 × 22.5 61.38 12.36 49.02
  2   22 × 24 65.26 47.8 17.46
  3   16.5 × 16 18.99 18.99 not detected
8 1   17 × 18 84.66 11.36 76.3
  2 1 21 × 21 36.7 0.01 36.69
    2 52 × 32 1.11 0.88 0.23
  3 1 22.5 × 23 39.12 not detected 39.12
    2 25.5 × 26 38.64 36.47 2.17
    3 25.5 × 17.5 35.97 4.00 31.97
12 1   27 × 29 23.16 not detected 23.16
  2   37.5 × 41.5 42.54 not detected 42.54
  3   40 × 32.5 42.82 not detected 42.82
Open in a new tab
a

Dimensions of the CCV were estimated from the white light images using Project5 software from Witec. Most of the CCVs had slightly oval shape. The numbers give the length and width in μm.

b

Relative degree of occupation of the CCV was determined from the false color Raman images by calculating the space occupied by C. burnetii (LCV and SCV) within the CCV. Averaged values for each time point are depicted in Figure 2E.

c

Relative amount of LCV was calculated from the size of the whole CCV and the relative content of LCV in the vacuole section analyzed by Raman spectroscopy (using pixels assigned to LCV).

d

Relative amount of SCV was calculated from the size of the whole CCV and the relative content of SCV in the vacuole section analyzed by Raman spectroscopy (using pixels assigned to SCV).

At day 1 p.i., the CCV contains only very few bacteria and is mainly filled with water (see endmember distribution in Figures 3C, S3A and TEM in Figure S5A). Size and relative degree of occupation of the CCV differ for different vacuoles (Table 1). While some CCV are still rather small (∼8 μm in diameter), others extend over 13 μm in diameter. Similar variations have also been reported in the literature29 and can be explained by asynchronous infection. Based on the spectral characteristics (Figure S3B), bacteria at this time point can be exclusively assigned to the LCV morphotype, indicating that most of the SCVs have already converted into LCV. This corresponds with TEM observations at 2 dpi (Figure S5A). It is also in agreement with previously published temporal analysis performed with TEM, where at 16 h and 1 day p.i., the parasitophorous vacuoles were found to almost exclusively contain LCVs with only an occasional presence of SCV.15

During the course of infection, CCVs increase in size as seen in the white light images and reflected in the false color Raman images and stacks (Figure 3). With increasing size, CCVs push other components and compartments of the host cell toward the cytoplasmic membrane. Starting from 6 dpi, neither host cell nucleus nor host cell cytoplasm is detectable in the cell sections analyzed, but CCVs cover almost the full field-of-view (20 × 20 μm2). At later time points, not the entire CCV but only sections of it are imaged in the fixed-sized Raman maps. Vacuoles from different host cells are captured in a single Raman stack due to high host cell density (e.g., at 8 dpi, Figure 3).

In addition to the increase in size of the CCVs, the number of bacteria inside the vacuoles increases with progressing infection time, as revealed by their increased relative occupation calculated from the Raman data (Figure 3F, Table 1, Figure S3) and seen in TEM images (Figure S5B,C). This is in particular visible when comparing time points 4, 6, and 8 dpi and indicates multiplication of bacteria inside the vacuole. Vacuole size and relative occupation by bacteria vary from cell to cell and vacuole to vacuole, but a steady and significant increase in vacuole size and bacterial occupation is observed from day 1 to day 4 p.i. An average doubling time of 34 h can be estimated from our data (Table 1), which is in agreement with the literature values.18 At 4 dpi (Figures 3, 3D), bacteria in the vacuole are exclusively assigned to LCV based on their spectral characteristics and confirmed by TEM (Figure S5B). Starting from 6 dpi, a second bacterial endmember spectrum is found in the Raman maps, which can be assigned to SCV (Figure 3D). At 6 dpi and 8 dpi, spectral endmember analysis and TEM images reveal co-existence of LCVs and SCVs within the same vacuole (Figures 1C, 3 and Figures S3 and S5C). Raman spectra reveal a transition phase between LCV and SCV at day 6 p.i. (Figure 3E). However, for subsequent analysis, individual spectra were assigned to either LCV or SCV morphoform using the spectral properties at 1300, 1440, and around 2800–3050 cm–1. At these time points, maximal occupation of the vacuole is reached (Figure 3F), and the vacuoles are tightly packed with bacteria. At 12 dpi (Figures 3, S3), only SCVs are detected, and vacuoles are less densely filled with relative occupation levels ranging from around 23 to 43%. The decrease at this late time point can be explained with the conversion of the larger LCVs to the smaller SCVs. An occupation ratio of less than 50%, as shown in Figure 3F, is in good agreement with SCVs exhibiting less than half the size of LCVs.30

The Raman data were used to quantify the relative content of SCVs and LCVs in the CCVs. Figure 3G depicts the relative increase in SCVs from 0% at the first two observation time points (1, 4 dpi) to >50% around day 8 p.i. and finally to 100% SCV content at day 12 p.i. A high variation of the SCV–LCV ratio is observed at days 6 and 8 p.i. (Figure 3G), which can be explained with non-synchronized initiation and development of individual vacuoles, resulting also in a highly divergent CCV morphology in TEM images (Figure S5C).

The results that were extracted from spectroscopic data are in excellent agreement with literature reports of the replication cycle of C. burnetii(15,19) and demonstrate that label-free and non-destructive Raman spectroscopic analysis of intact infected host cells is a powerful visualization tool.

Insights into Lipid Metabolism

Lipids are not only nutrients but also involved in cellular signaling, organization, and dynamics of membrane microdomains and membrane trafficking.31 Thus, lipids are an important target of pathogens and play key roles in host–pathogen interactions.32

We have used the N-FINDR algorithm to extract lipid spectra from the 3D Raman images (Figures 1C, S3, S4). Lipid spectra derived from Coxiella-infected cells show typical Raman bands for bacterial membrane lipids (e.g., 720, 875, and 3010 cm–1 for phosphatidylcholine and phosphatidylethanolamine), cholesterol esters (C–C stretching at 1068 cm–1 and C=O stretching at 1750 cm–1 and C–H stretching at 3010 cm–1), and fatty acids (in particular palmitoleic and oleic acid), as can be identified by comparison with reference spectra.33

Unsupervised clustering using PCA revealed that the lipid composition differs at different time points post infection, while it is very similar within different cells at the same time point p.i., as can be seen in the PCA score plot (Figure 4A). Positive bands in the PC3 coefficient are found at 711, 878, 1067, 1126, 1440, and 2848 cm–1 (Figure 4B). These wavenumbers can be assigned to vibrations of phosphatidylethanolamine, phosphatidylcholine, and cholesterol ester and indicate a relative higher content of these components at 4 dpi than that at 12 dpi. The PC4 coefficient shows positive bands at 890, 1081, 1272, 1302, 1435, 1655, 1750, and 3010 cm–1 (Figure 4B), which indicate a higher relative abundance of palmitoleic acid and oleic acid at 12 dpi than that at the earlier time points.

Figure 4.

Figure 4

Open in a new tab

PCA of Raman spectra from lipid droplets observed at different time points post infection within C. burnetii-infected BGM cells. Lipid droplets were mainly found around the Coxiella-containing-vacuole. (A) PCA score plot color-coding infection time point. Raman spectra were extracted from several infected cells (4 dpi: n = 1 cell; 8 dpi: n = 3 cells; and 12 dpi: n = 3 cells). Each time point forms a dense cloud: Raman spectra of lipids extracted from infected cells at 4 dpi are found in the lower right part of the diagram (positive PC3 values and negative PC4 values), while Raman spectra of lipids from cells at 12 dpi are found in the upper left part (negative PC3 and positive PC4 values). Raman spectra of lipids from cells at 8 dpi are found in between. (B) Loading coefficients of PC3 and PC4 highlighting the spectral features, which change during the infection process.

Supervised LDA was used to analyze the same lipid spectra, and the results are depicted in Figure S6. The same Raman bands were found to be important to separate 4 dpi and later time points (LD 2), as well as to separate 12 dpi and earlier time points (LD 1). The positive bands in the LD 2 coefficient can be correlated with typical Raman bands of phosphatidylethanolamine and phosphatidylcholine, the negative features with Raman bands for fatty acids.

These findings are in agreement with a previous lipid analysis using thin layer chromatography of cell extracts,26 in which a change in amount and ratio of phosphatidylcholine, phosphatidylethanolamine, and palmitoleic acid was observed during the life cycle of C. burnetii from day 4 to day 14 post infection. The higher content of membrane lipids (phosphatidylethanolamine and phosphatidylcholine) at day 4 correlates with the active metabolism of the LCVs and the high need of membrane lipids during their replication. Fatty acids are increased at later time points (12 dpi), when only the SCV form is present.26

Conclusions

Our new Raman-based imaging algorithm enables precise localization, visualization, and quantification of intracellular pathogens in 3D within intact, chemically fixed host cells. The spectral information provides insights into biochemical changes of C. burnetii during intracellular bacterial development non-destructively without labeling and at different time points post infection over the time course of the infectious cycle with diffraction-limited resolution. It is not only possible to identify and characterize the two morphoforms including their transitions from SCV to LCV and back. It is also possible to reveal qualitative and quantitative information during the infection process. This includes not only alterations in the lipid composition in and around the CCV but also the visualization of single host cell heterogeneity, caused by, for example, asynchronous infections.

Acknowledgments

Financial support by the Leibniz Association via the Leibniz Science Campus (LSC) InfectoOptics (W8/2018), the BMBF via the CSCC (FKZ 01EO1502), and the EU via IMAGE-IN within Horizon 2020 research and innovation programme (grant agreement no 861122) and FLI internal funding are highly acknowledged. We furthermore acknowledge the “Thüringer Innovationszentrum für Medizintechnik-Lösungen (ThIMEDOP) (FKZ IZN 2018 0002).” We thank Dr. Katja Mertens-Scholz for providing the C. burnetii strain.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.1c04754.

  • Sample preparation for EM of intracellular pathogens (PDF)

Author Contributions

U.N., C.B., and C.S. designed research; F.N., S.Z., and K.S. prepared samples; N.U. and S.E. performed Raman measurements; E.L.T. performed electron microscopy; N.U. analyzed data; N.U., U.N., C.S., and C.B. discussed results and wrote the manuscript, and all authors read, modified, and approved the final version of the manuscript.

The authors declare no competing financial interest.

Supplementary Material

ac1c04754_si_001.pdf (1.3MB, pdf)

References

  1. Qiu J.; Luo Z.-Q. Legionella and Coxiella effectors: strength in diversity and activity. Nat. Rev. Microbiol. 2017, 15, 591–605. 10.1038/nrmicro.2017.67. [DOI] [PubMed] [Google Scholar]
  2. Yang Y.; Hu M.; Yu K.; Zeng X.; Liu X. Mass spectrometry-based proteomic approaches to study pathogenic bacteria-host interactions. Protein Cell 2015, 6, 265–274. 10.1007/s13238-015-0136-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. a Butler H. J.; Ashton L.; Bird B.; Cinque G.; Curtis K.; Dorney J.; Esmonde-White K.; Fullwood N. J.; Gardner B.; Martin-Hirsch P. L.; et al. Using Raman spectroscopy to characterize biological materials. Nat. Protoc. 2016, 11, 664–687. 10.1038/nprot.2016.036. [DOI] [PubMed] [Google Scholar]; b Krafft C.; Schie I. W.; Meyer T.; Schmitt M.; Popp J. Developments in spontaneous and coherent Raman scattering microscopic imaging for biomedical applications. Chem. Soc. Rev. 2016, 45, 1819–1849. 10.1039/C5CS00564G. [DOI] [PubMed] [Google Scholar]; c Durrant B.; Trappett M.; Shipp D.; Notingher I. Recent developments in spontaneous Raman imaging of living biological cells. Curr. Opin. Chem. Biol. 2019, 51, 138–145. 10.1016/j.cbpa.2019.06.004. [DOI] [PubMed] [Google Scholar]
  4. a van Manen H.-J.; Kraan Y. M.; Roos D.; Otto C. Single-cell Raman and fluorescence microscopy reveal the association of lipid bodies with phagosomes in leukocytes. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10159–10164. 10.1073/pnas.0502746102. [DOI] [PMC free article] [PubMed] [Google Scholar]; b Matthäus C.; Chernenko T.; Newmark J. A.; Warner C. M.; Diem M. Label-free detection of mitochondrial distribution in cells by nonresonant Raman microspectroscopy. Biophys. J. 2007, 93, 668–673. 10.1529/biophysj.106.102061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Stöckel S.; Kirchhoff J.; Neugebauer U.; Rösch P.; Popp J. The application of Raman spectroscopy for the detection and identification of microorganisms. J. Raman Spectrosc. 2016, 47, 89–109. 10.1002/jrs.4844. [DOI] [Google Scholar]
  6. Große C.; Bergner N.; Dellith J.; Heller R.; Bauer M.; Mellmann A.; Popp J.; Neugebauer U. Label-Free Imaging and Spectroscopic Analysis of Intracellular Bacterial Infections. Anal. Chem. 2015, 87, 2137–2142. 10.1021/ac503316s. [DOI] [PubMed] [Google Scholar]
  7. Kallepitis C.; Bergholt M. S.; Mazo M. M.; Leonardo V.; Skaalure S. C.; Maynard S. A.; Stevens M. M. Quantitative volumetric Raman imaging of three dimensional cell cultures. Nat. Commun. 2017, 8, 14843. 10.1038/ncomms14843. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Høgset H.; Horgan C. C.; Armstrong J. P. K.; Bergholt M. S.; Torraca V.; Chen Q.; Keane T. J.; Bugeon L.; Dallman M. J.; Mostowy S.; et al. In vivo biomolecular imaging of zebrafish embryos using confocal Raman spectroscopy. Nat. Commun. 2020, 11, 6172. 10.1038/s41467-020-19827-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. van Schaik E. J.; Chen C.; Mertens K.; Weber M. M.; Samuel J. E. Molecular pathogenesis of the obligate intracellular bacterium Coxiella burnetii. Nat. Rev. Microbiol. 2013, 11, 561–573. 10.1038/nrmicro3049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. a Eldin C.; Mélenotte C.; Mediannikov O.; Ghigo E.; Million M.; Edouard S.; Mege J.-L.; Maurin M.; Raoult D. From Q Fever to Coxiella burnetii Infection: a Paradigm Change. Clin. Microbiol. Rev. 2017, 30, 115–190. 10.1128/cmr.00045-16. [DOI] [PMC free article] [PubMed] [Google Scholar]; b The European Union One Health 2018 Zoonoses Report. EFSA J. 2019, 17, e05926 10.2903/j.efsa.2019.5926. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Anderson A.; Bijlmer H.; Fournier P. E.; Graves S.; Hartzell J.; Kersh G. J.; Limonard G.; Marrie T. J.; Massung R. F.; McQuiston J. H.; et al. Diagnosis and management of Q fever--United States, 2013: recommendations from CDC and the Q Fever Working Group. MMWR Recommendations and Reports, 2013; Vol. 62 ( (3), ), pp 1–30. [PubMed]
  12. a Minnick M. F.; Raghavan R. Developmental Biology of Coxiella burnetii. Adv. Exp. Med. Biol. 2012, 984, 231–248. 10.1007/978-94-007-4315-1_12. [DOI] [PubMed] [Google Scholar]; b Moffatt J. H.; Newton P.; Newton H. J. Coxiella burnetii: turning hostility into a home. Cell. Microbiol. 2015, 17, 621–631. 10.1111/cmi.12432. [DOI] [PubMed] [Google Scholar]
  13. Coleman S. A.; Fischer E. R.; Cockrell D. C.; Voth D. E.; Howe D.; Mead D. J.; Samuel J. E.; Heinzen R. A. Proteome and antigen profiling of Coxiella burnetii developmental forms. Infect. Immun. 2007, 75, 290–298. 10.1128/iai.00883-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Heinzen R. A.; Hackstadt T.; Samuel J. E. Developmental biology of Coxiella burnetii. Trends Microbiol. 1999, 7, 149–154. 10.1016/S0966-842X(99)01475-4. [DOI] [PubMed] [Google Scholar]
  15. Coleman S. A.; Fischer E. R.; Howe D.; Mead D. J.; Heinzen R. A. Temporal analysis of Coxiella burnetii morphological differentiation. J. Bacteriol. 2004, 186, 7344–7352. 10.1128/jb.186.21.7344-7352.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Howe D.; Melnicáková J.; Barák I.; Heinzen R. A. Maturation of the Coxiella burnetii parasitophorous vacuole requires bacterial protein synthesis but not replication. Cell. Microbiol. 2003, 5, 469–480. 10.1046/j.1462-5822.2003.00293.x. [DOI] [PubMed] [Google Scholar]
  17. Gürtler L.; Bauerfeind U.; Blümel J.; Burger R.; Drosten C.; Gröner A.; Heiden M.; Hildebrandt M.; Jansen B.; Offergeld R.; et al. Coxiella burnetii - Erreger des Q (query)-Fiebers. Bundesgesundheitsblatt-Gesundheitsforschung-Gesundheitsschutz 2013, 56, 1178–1190. 10.1007/s00103-013-1816-0. [DOI] [PubMed] [Google Scholar]
  18. Pechstein J.; Schulze-Luehrmann J.; Lührmann A. Coxiella burnetii as a useful tool to investigate bacteria-friendly host cell compartments. Int. J. Med. Microbiol. 2018, 308, 77–83. 10.1016/j.ijmm.2017.09.010. [DOI] [PubMed] [Google Scholar]
  19. Sandoz K. M.; Popham D. L.; Beare P. A.; Sturdevant D. E.; Hansen B.; Nair V.; Heinzen R. A. Transcriptional Profiling of Coxiella burnetii Reveals Extensive Cell Wall Remodeling in the Small Cell Variant Developmental Form. PLoS One 2016, 11, e0149957 10.1371/journal.pone.0149957. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kuley R.; Bossers-deVries R.; Smith H. E.; Smits M. A.; Roest H. I. J.; Bossers A. Major differential gene regulation in Coxiella burnetii between in vivo and in vitro cultivation models. BMC Genomics 2015, 16, 953. 10.1186/s12864-015-2143-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. a Williams J. C.; Peacock M. G.; McCaul T. F. Immunological and biological characterization of Coxiella burnetii, phases I and II, separated from host components. Infect. Immun. 1981, 32, 840–851. 10.1128/iai.32.2.840-851.1981. [DOI] [PMC free article] [PubMed] [Google Scholar]; b Millar J. A.; Beare P. A.; Moses A. S.; Martens C. A.; Heinzen R. A.; Raghavan R. Whole-Genome Sequence of Coxiella burnetii Nine Mile RSA439 (Phase II, Clone 4), a Laboratory Workhorse Strain. Genome Announc. 2017, 5, e00471 10.1128/genomeA.00471-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. WITec . WITec alpha300 Series: Modular Confocal Raman Microscopy Systems, 2022. https://www.witec.de/assets/Literature/Files/WITec-alpha300-Brochure.pdf (accessed Jan 19, 2022).
  23. Winter M.N-FINDR: an algorithm for fast autonomous spectral end-member determination in hyperspectral data; SPIE, 1999. [Google Scholar]
  24. Jahmani M. Y.; Hammadeh M. E.; Al Smadi M. A.; Baller M. K. Label-Free Evaluation of Chromatin Condensation in Human Normal Morphology Sperm Using Raman Spectroscopy. Reprod. Sci. 2021, 28, 2527–2539. 10.1007/s43032-021-00494-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Assmann C.; Kirchhoff J.; Beleites C.; Hey J.; Kostudis S.; Pfister W.; Schlattmann P.; Popp J.; Neugebauer U. Identification of vancomycin interaction with Enterococcus faecalis within 30 min of interaction time using Raman spectroscopy. Anal. Bioanal. Chem. 2015, 407, 8343–8352. 10.1007/s00216-015-8912-y. [DOI] [PubMed] [Google Scholar]
  26. Stead C. M.; Cockrell D. C.; Beare P. A.; Miller H. E.; Heinzen R. A. A Coxiella burnetii phospholipase A homolog pldA is required for optimal growth in macrophages and developmental form lipid remodeling. BMC Microbiol. 2018, 18, 33. 10.1186/s12866-018-1181-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. a Heinzen R. A.; Howe D.; Mallavia L. P.; Rockey D. D.; Hackstadt T. Developmentally regulated synthesis of an unusually small, basic peptide by Coxiella burnetii. Mol. Microbiol. 1996, 22, 9–19. 10.1111/j.1365-2958.1996.tb02651.x. [DOI] [PubMed] [Google Scholar]; b Varghees S.; Kiss K.; Frans G.; Braha O.; Samuel J. E. Cloning and porin activity of the major outer membrane protein P1 from Coxiella burnetii. Infect. Immun. 2002, 70, 6741–6750. 10.1128/IAI.70.12.6741-6750.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Allen P. E.; Martinez J. J. Modulation of Host Lipid Pathways by Pathogenic Intracellular Bacteria. Pathogens 2020, 9, 614. 10.3390/pathogens9080614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Schulze-Luehrmann J.; Eckart R. A.; Ölke M.; Saftig P.; Liebler-Tenorio E.; Lührmann A. LAMP proteins account for the maturation delay during the establishment of theCoxiella burnetii-containing vacuole. Cell. Microbiol. 2016, 18, 181–194. 10.1111/cmi.12494. [DOI] [PubMed] [Google Scholar]
  30. a Amano K.; Williams J. C.; McCaul T. F.; Peacock M. G. Biochemical and immunological properties of Coxiella burnetii cell wall and peptidoglycan-protein complex fractions. J. Bacteriol. 1984, 160, 982–988. 10.1128/jb.160.3.982-988.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]; b McCaul T. F.; Banerjee-Bhatnagar N.; Williams J. C. Antigenic differences between Coxiella burnetii cells revealed by postembedding immunoelectron microscopy and immunoblotting. Infect. Immun. 1991, 59, 3243–3253. 10.1128/iai.59.9.3243-3253.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. van der Meer-Janssen Y. P. M.; van Galen J.; Batenburg J. J.; Helms J. B. Lipids in host-pathogen interactions: Pathogens exploit the complexity of the host cell lipidome. Prog. Lipid Res. 2010, 49, 1–26. 10.1016/j.plipres.2009.07.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. a Samanta D.; Mulye M.; Clemente T. M.; Justis A. V.; Gilk S. D. Manipulation of Host Cholesterol by Obligate Intracellular Bacteria. Front. Cell. Infect. Microbiol. 2017, 7, 165. 10.3389/fcimb.2017.00165. [DOI] [PMC free article] [PubMed] [Google Scholar]; b Kohler L. J.; Roy C. R. Biogenesis of the lysosome-derived vacuole containing Coxiella burnetii. Microbes Infect. 2015, 17, 766–771. 10.1016/j.micinf.2015.08.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Czamara K.; Majzner K.; Pacia M. Z.; Kochan K.; Kaczor A.; Baranska M. Raman spectroscopy of lipids: a review. J. Raman Spectrosc. 2015, 46, 4–20. 10.1002/jrs.4607. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ac1c04754_si_001.pdf (1.3MB, pdf)

Articles from Analytical Chemistry are provided here courtesy of American Chemical Society

RESOURCES