Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Evaluation of changes to the Rickettsia rickettsii transcriptome during mammalian infection

  • Sean P. Riley ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Vector-Borne Disease Laboratories, Department of Pathobiological Sciences, Louisiana State University School of Veterinary Medicine, Baton Rouge, Louisiana, United States of America


  • Ludovic Pruneau,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Writing – review & editing

    Affiliation Vector-Borne Disease Laboratories, Department of Pathobiological Sciences, Louisiana State University School of Veterinary Medicine, Baton Rouge, Louisiana, United States of America

  • Juan J. Martinez

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Writing – review & editing

    Affiliation Vector-Borne Disease Laboratories, Department of Pathobiological Sciences, Louisiana State University School of Veterinary Medicine, Baton Rouge, Louisiana, United States of America

Evaluation of changes to the Rickettsia rickettsii transcriptome during mammalian infection

  • Sean P. Riley, 
  • Ludovic Pruneau, 
  • Juan J. Martinez


The lifecycle of Rickettsia rickettsii includes infection of both mammalian and arthropod hosts, with each environment presenting distinct challenges to survival. As such, these pathogens likely have distinctive transcriptional strategies for infection of each host. Herein, we report the utilization of next generation sequencing (RNAseq) and bioinformatic analysis techniques to examine the global transcriptional profile of R. rickettsii within an infected animal, and to compare that data to transcription in tissue culture. The results demonstrate substantial R. rickettsii transcriptional alteration in vivo, such that the bacteria are considerably altered from cell culture. Identification of significant transcriptional changes and validation of RNAseq by quantitative PCR are described with particular emphasis on known antigens and suspected virulence factors. Together, these results suggest that transcriptional regulation of a distinct cohort of genes may contribute to successful mammalian infection.


Rickettsia rickettsii is the causative agent of Rocky Mountain spotted fever (RMSF), the most severe tick transmitted infection in the western hemisphere with human case fatality rates comparable to the most dangerous of all North American bacterial diseases [1]. This bacterium survives in nature through an endozoonotic cycle of infection of mammalian and tick hosts. As such, the bacteria are faced with diverse challenges for infection of and transition between individual hosts. This feature of zoonotic infection suggests that the bacteria have different pathogenic strategies for each host, and, therefore, likely experience significant transcriptional variations throughout the life cycle. This concept is supported in other bacterial pathogens whereby synchronization of gene expression is critical for successful infection [2]. Herein, we examine the hypothesis that R. rickettsii undergoes distinct transcriptional changes in vivo in a mammalian model of infection by comparing the global transcriptome of bacteria from laboratory culture and during fatal infection of M. musculus.

Rickettsia species transcriptional profiles have been examined in response to temperature changes, iron starvation, and time in culture [39]. These studies demonstrated a limited response to temperature change, but significant transcriptional changes when R. rickettsii are exposed to a blood meal while within the transmitting tick [7]. In depth examination of R. rickettsii within the tick established that the genes encoding for the Type IV Secretion System (T4SS), potentially secreted proteins, iron acquisition proteins, and surface exposed proteins are all regulated in response to blood. In contrast, analysis of R. conorii eschar samples indicated a significant decrease in translation machinery and upregulation of transcripts associated with stress [10]. This in vivo analysis of R. conorii differs from the current study in the method of Rickettsia transmission (tick versus needle inoculation) and tissue source (eschar versus morbidly infected liver). Additionally, the authors note that R. conorii in eschars may be undergoing host immune clearance in the inoculation eschar [10]. Together, these transcriptomic studies establish that Rickettsia species, and particularly R. rickettsii, have the ability to regulate transcription in response to environmental variations.

When developing an experimental strategy to expand upon these previous transcriptional analyses, we considered multiple material sources. Our first goal was to characterize the R. rickettsii transcriptome within the infected mammal; therefore, we chose to analyze R. rickettsii within the liver of moribund C3H/HeN mice, because this sample contains multiple infected host cells types with the largest rickettsial load currently established [11]. The highest ratio of R. rickettsii to M. musculus RNA was predicted to produce the highest number of sequencing reads that could be attributed to the bacteria, and thus the most complete analysis of rickettsial transcription within an infected animal.

The in vivo transcriptome data necessarily needed to be compared to another condition to identify transcriptional changes. There are multiple in vitro culture conditions that could serve as a comparison to in vivo samples. After considering multiple conditions, we reasoned that the best comparison would be to R. rickettsii from infected Vero cells. This condition was chosen because much of the current understanding of R. rickettsii biology was generated through Vero cell culture. First, Ellison [5] describes a global transcriptional analysis of Vero cell propagated R. rickettsii under different environmental conditions. By utilizing Vero cell cultured R. rickettsii in the current experimental setup, we would be able to more accurately compare our findings to that published data. Additionally, Ammerman [12] is the current standard for in vitro propagation of R. rickettsii. We reasoned that most past and future in vitro propagations would utilize this culture method, so this RNA would be the most viable for comparison to in vivo transcription.

Transcriptional analysis of surface-exposed and secreted proteins of obligate intracellular Rickettsia are of particular importance because of the essential functions mediated by these proteins, including host cell invasion, escape from the phagolysosome, intracellular motility, and manipulation of the host response to infection [1316]. Experiments designed to identify surface proteins as defined by exposure to surface biotinylation have identified approximately 70 likely surface exposed rickettsial proteins [9, 17]. Of these proteins, the surface cell antigen (Sca) family of autotransporters have been individually localized to the bacterial surface [9, 1821]. Secretion of Sca proteins is dependent upon the Sec-dependent Type V Secretion, but Rickettsia also encode for Sec-Tol, Tat, Type I, and Type IV Secretion Systems [22]. As such, R. rickettsii is likely to encode for a plethora of surface and secreted proteins, and demonstration of transcriptional regulation of these potential virulence factors will aid future analyses of R. rickettsii pathogenesis.

Herein, we utilized next-generation sequencing technologies to compare quantities of and changes to R. rickettsii RNA in vivo during mouse infection and in vitro during tissue culture. The transcriptional alterations observed in vivo indicate that the bacterium is actively adapting to the mammalian host, and that our current understanding of in vitro R. rickettsii RNA and/or protein content is different than is found within the infected animal. We report changes to genes encoding for rickettsial surface proteins, Type IV Secretion System, and ribosomal subunits. The results of this study will serve as a basis for analyzing the pathobiology of R. rickettsii infection within the mammalian host.

Materials and methods

Cell lines and RNA samples

African green monkey kidney cells (Vero, ATCC) were cultured and maintained under standard conditions as described previously [19]. Briefly, 150cm2 flasks of confluent Vero cells were infected at 1 bacteria per host cell. The bacteria were allowed to infect at 34°C 5%CO2 for five days with the media being changed at day 3. R. rickettsii were purified from Vero cells by needle lysis and centrifugation over a sucrose cushion [12]. Total RNA was isolated twice from two separate R. rickettsii Sheila Smith propagations for a total of four replicates (2 technical replicates for each of 2 biological replicates). These samples are subsequently referred to as “in vitro. Similarly, total RNA was isolated from liver tissues of C3H/HeN mice at five days post infection with 1x107 intravenous R. rickettsii [11]. These mice were moribund and were euthanized prior to succumbing to R. rickettsii infection. RNA was isolated twice from two separate livers for a total of four replicates (2 technical replicates for each of 2 biological replicates). These samples will subsequently be referred to as “in vivo.”

RNA isolation, DNAse treatment, and enrichment of prokaryotic mRNA

Total RNA was isolated from the eight total samples using Purelink RNA Mini Kit (Ambion). DNA was removed from the RNA purification using Ambion Turbo DNAse according to manufacturer’s instructions. Removal of DNA contamination was verified by sca1 PCR as previously described [11]. Quality control was performed on a Fragment Analyzer (Advanced Analytical) to determine the RNA quality number (RQN). All RNA samples had a RQN > 7.6 (7.6–9.8). After confirmation of RNA structural integrity, 5μg RNA was subjected to MicrobeExpress and MicrobEnrich (Invirogen) protocols to remove bacterial rRNA and mammalian RNA, respectively.

cDNA synthesis and sequencing

cDNA libraries were constructed using the Ion Total RNA-Seq Kit v2 (Life). Templates for RNASeq were prepared with the Ion PI Hi-Q OT2 200 Kit (Life) using the Ion OneTouch 2 System for Ion Proton System semiconductor sequencing. These templates were sequenced using the Ion PI Hi-QSequencing 200 Kit and Ion PI Chip v2, on the Ion Proton Sequencer. Preliminary analysis of the ensuing sequencing data pertaining to quality of the run, read lengths, and coverage was performed using the Torrent Suite software. Total reads, usable reads, and mean read length for each sample are indicated in S1 Table.

Sequence analysis

The sequence reads from the four in vivo samples were first scanned for adapters and quality filtered using cutadapt with Partek Flow [23]. Next, reads from Rickettsia rickettsii were identified using a modification of the two-step method, whereby samples were first downloaded from NCBI on 08/2016, then aligned to Mus musculus GRCm38.p4 with tophat2, then the unaligned fastq reads were aligned to R. rickettsii Sheila Smith (NC_009882.1), using TMAP[24]. The four in vitro samples were aligned to the R. rickettsia genome using the SeqManNGEN portion of DNASTAR [25]. To determine the number of reads that were mapped to R. rickettsii genes, htseq was used to determine gene counts. The htseq counts for each gene were converted to RPKM using the formula: (number of reads)/[(gene size/1,000)x(total reads/1,000,000)] [26], and these values were averaged for the four in vivo sequencing reactions. DESeq2 was used for differential gene expression analysis by contrasting the in vitro and in vivo groups [27]. DEseq utilizes an adjusted value of significance (padj) which is described in “Statistical analyses.” A summary of the analysis pipeline is diagramed in S1 Fig.

qRT-PCR validation

R. rickettsii gDNA loads in mouse organs were quantified by TaqMan quantitative PCR (qPCR). Total chromosomal DNA from liver tissue (in vivo) and purified R. rickettsii (in vitro) was purified using PureLink Genomic DNA mini kit (Invitrogen) and normalized to 50μg/ml. TaqMan qPCR reactions were run as previously described [11] on the LightCycler 480 II system (Roche). PCR reactions were run as follows: 58°C for 3min, 95°C for 10min, followed by 50 cycles of 95°C for 15s and 58°C for 60s. Standard curves were generated using the plasmids pEC3 (sca1) and pYC82 (actin). gDNA content in organs are expressed as the number of R. rickettsii sca1 copies per murine M. musculus or C. aethiops actin gene copies.

Changes to the transcriptional content of specific genes were determined by quantitative reverse transcription PCR (qRT-PCR). 200ng of DNA-fee RNA was subjected to Superscript Vilo (Invitrogen) reverse transcription. The quantity of each cDNA was determined by qPCR using the primers described in S2 Table. PCR reactions were run as follows: 95°C for 10 min, followed by 50 cycles of 95°C for 30s, 63°C for 30s, and 72°C for 60s followed by melting curve. R. rickettsii gDNA was used for calculation of standard curve. qPCR-derived fold change values are expressed as:

Identification of open reading frames with signal sequences

To identify sec-dependent signal peptides the R. rickettsii protein reference sequence (GCF_000018225.1_ASM1822v1) was input into the signal sequence prediction algorithms: 1) Phobius [28] 2) PRED-TAT [29] 3) SignalP4.1 [30] and 4) LipoP [31]. Proteins categorized as “Secreted Proteins” were identified by 3 of the 4 servers (S3 Table).

Sequence deposition

The 8 sequence output files and metadata are available from NCBI Sequence Read Archive under the accession number PRJNA382417. A summary of all RNAseq data is found in S4 Table.

Animal research

Animal experiments were conducted in accordance with protocols approved by the Institutional Biological and Recombinant DNA Safety Committee (IBRDSC) and Institutional Animal Care and Use Committee (IACUC) at the Louisiana State University School of Veterinary Medicine. The University has a file with the Office of Laboratory Animal Welfare (OLAW), an approved Assurance Statement (#A3612-01).

Statistical analyses

Correlation between qPCR and RNAseq results was performed by Shapiro-Wilk Test for Normality and Pearson analysis of correlation in Graphpad Prism. “padj” values for fold change significance are computed in the DESeq2 package. This value uses the used Benjamin-Hochberg correction method. Padj values of <0.01 are considered significant [32].



The central hypothesis considered herein is that R. rickettsii utilizes genetic regulation to employ a specific transcriptional program for infection of the mammalian host. Accordingly, the experimental objective was to query R. rickettsii for transcriptional differences between bacteria derived from common cell culture and within the infected animal, called “in vitro” and “in vivo” respectively. To generate in vitro RNA samples, we purified R. rickettsii Sheila Smith from routine Vero cell culture at the time of maximum rickettsial burden. Bacteria at this stage of in vitro culture were the inoculum source for infection of mammals, and can therefore be considered at pre-infection transcriptional state. Additionally, transcriptional responses of R. rickettsii to environmental stimuli in Vero cell culture have been examined previously [5].

As a comparison, we purified in vivo whole RNA from within R. rickettsii infected M. musculus liver at the point where the animals were considered mortally infected [11]. This sample was chosen because it represents the highest rickettsial burden throughout mouse infection. This high bacterial burden would produce the highest ratio of Rickettsia to host RNA for technical purposes, and would also represent the highest number of input Rickettsia for the most heterogeneous R. rickettsii population. We performed four total RNA extractions on two different biological samples for each experimental group. These eight total RNA samples were subjected to reverse transcription and deep sequencing to determine the quantity and arrangement of RNA in each sample.

The most immediate challenge in analyzing the transcriptome of obligate intracellular bacteria is the pervasiveness of host RNA in the in vivo environment. This overabundance of host transcript masks the available Rickettsia-specific sequencing information. In order to overcome this obstacle without artificially amplifying rickettsial RNA, we used RNAseq with extensive bioinformatics analysis. This technology is based on direct cDNA sequencing and allows the detection and quantification of RNA from very small amounts of cellular materials [33]. Importantly, this method does not rely on non-specific amplification of cDNA, as is required to get sufficient cDNA for chip-based Rickettsia transcriptomics [8]. To maximize the prevalence of rickettsial mRNA we first performed depletion of eukaryotic RNA and bacterial ribosomal RNA before reverse transcription and deep sequencing. Of the 94–78 million reads per sample, the majority were discarded during quality control, leaving approximately 15 million usable output reads. Even after depletion of eukaryotic RNA, only 0.81–6.68% of in vivo sequencing reads mapped to the R. rickettsii genome. Reads that mapped to rRNA and tRNA were not analyzed, thereby leaving an average coverage depth of 19–128 reads per gene of the R. rickettsii protein coding sequence. As such, the results described herein represent a vigorous assessment of the transcriptional program of R. rickettsii within the infected mammal.

The 1345 predicted protein-encoding open reading frames of R. rickettsii Sheila Smith were analyzed for total in vivo cDNA quantity and expression changes. The average expression value of individual genes was calculated for the four in vivo samples, and is expressed as average reads per kilobase per million reads (RPKM), a value that normalizes for gene length. Expression changes to individual genes were calculated in the DEseq2 software package, and are expressed as fold change of in vivo samples as compared to in vitro. Fold change values are associated with an adjusted value of statistical significance (padj) where p<0.01 is considered significant. In all, 235 genes were significantly upregulated in vivo and 264 genes were downregulated in vivo. These values represent 17% upregulated genes and 20% downregulated genes, thus demonstrating substantial transcriptional differences between in vitro and in vivo Rickettsia rickettsii.

Validation of RNAseq by quantitative RT-PCR

16 genes were chosen for individual analysis in order to validate the RNAseq data by an independent experimental method. To this aim, both RNA and gDNA were re-isolated from the RNAseq source material. The isolated RNA samples were subjected to DNAase treatment, reverse transcription, and quantitative PCR (qRT-PCR) to determine cDNA concentration for each of the 16 genes. In order to control for rickettsial nucleotide concentration in the PCR reaction, we utilized gDNA copies present in the biological sample. The concentration of R. rickettsii sca1 copies per murine actin gDNA was determined in each sample, with a sca1/actin ratio of 3.27 for the in vivo infected liver sample and 78.32 in the in vitro sample. By adding equal amounts of cDNA and dividing the calculated cDNA concentration by the gDNA ratio, we are able to properly control for changes to the quantity of input Rickettsia RNA. This data transformation allows for accurate determination of the fold change in expression of each of the tested genes despite the contaminating host RNA. The qRT-PCR data for the 16 analyzed genes was compared to the fold change values obtained by RNAseq (Fig 1). Logically, if the two data acquisition mechanisms produce consistent results, then the two data sets should have a strong linear correlation. The results from each quantification method were compared and demonstrate a significant correlation with p = 0.0015. With these consistent results, we can extrapolate that the RNAseq data faithfully identifies changes to transcription.

Fig 1. Comparison of transcriptional fold changes as determined by RNAseq and quantitative RT-PCR.

In vivo / in vitro fold change values as determined by RNAseq (Y-axis) and qRT-PCR (X-axis). Pearson analysis of correlation p = 0.0015.

Genes encoding for secreted proteins

As R. rickettsii is an obligate intracellular pathogen, many necessary pathogenic functions occur at the bacterial surface. While several functions have been attributed to surface proteins, nothing is known about changes to the bacterial surface during the R. rickettsii lifecycle. Accordingly, the R. rickettsii Sheila Smith genome was queried by four signal sequence prediction algorithms (LipoP, Phobius, PRED-TAT, SignalP4.1) for the presence of a secretion signal. In all, 59 genes were predicted by 3 or more of the prediction algorithms to possess a secretion signal. This analytical method identified genes encoding for the well-known surface proteins OmpA, OmpB, Sca1, Sca2, 17kDa antigen, Adr1, and Adr2, but failed to identify some proteins previously identified by surface biotinylation [17].

The genes with predicted signal sequences were analyzed for in vivo RNA quantity (RPKM) and fold change of in vivo with relation to in vitro (Fig 2, S3 Table). In all, 12 genes were upregulated in vivo and 26 genes were significantly downregulated in vivo. As expected, the gene encoding for the ubiquitous surface protein ompB was the most frequently detected transcript, with high expression in vivo and insignificant expression change across the two conditions. By comparison, ompA, sca1, 17kDa antigen (A1G_07075), adr1, and adr2 were all significantly downregulated in vivo. Upon identifying these differences in RNA content, we conclude that the antigen profile of R. rickettsii in vivo will likely be significantly different than has been observed for cultured bacteria. The most impressively upregulated gene, ostA (A1G_05645), encodes for an outer membrane barrel that contributes to LPS biosynthesis and inorganic solvent tolerance [34]. Interestingly, 9 of the 21 LPS genes involved in LPS biosynthesis and transport are significantly regulated (S5 Table), suggesting that R. rickettsii may alter the quantity or chemical structure of LPS.

Fig 2. Analysis of R. rickettsii genes with predicted sec-dependent signal sequences.

Four signal sequence prediction programs were utilized to identify 59 genes encoding for secreted or surface exposed R. rickettsii proteins. The quantity of transcript is expressed as average reads per kilobase per million reads (RPKM) for in vivo sequencing reactions. Additionally, each gene is graphed for transcriptional changes as expressed as fold change of in vivo over in vitro. Colored bars transcriptional change padj<0.01.

R. rickettsii also encodes for 14 proteins containing Ankyrin repeat domains. The Ank domain containing proteins typically interact with eukaryotic proteins or DNA, and are therefore assumed to be secreted through Type I or IV Secretion Systems [35]. Indeed, Ank proteins from Orientia tsutsugamushi and R. typhi have been found within the host cytoplasm [36, 37]. The R. rickettsii ank genes were variably regulated (Fig 3, S5 Table). Analysis of total in vivo expression shows that both A1G_02960 and A1G_00070 were highly expressed (RPKM >1000). Conversely, A1G_01255 and A1G_00065 have extremely low expression in vivo.

Fig 3. Changes to genes encoding for Ankyrin repeat (Ank) domain proteins.

Transcription of 14 genes encoding for Ank domain proteins are analyzed for average in vivo expression (RPKM) and changes in vivo in relation to in vitro. Colored bars padj<0.01.

Type IV Secretion System

All Rickettsiales encode for a Type IV Secretion System (T4SS), consisting of a polypeptide complex that translocates proteins into the extracellular environment or directly into a host cell. Studies have established that T4SS are essential to pathogenesis in various bacterial species [38]. The genetic makeup of the Rickettsia T4SS has been thoroughly investigated, whereby the genetic complexity and conservation suggests that this system is essential for completion of the Rickettsia lifecycle [3941]. Some of the genes encoding for T4SS are duplicated or scattered thereby producing many identically annotated genes [39]. Upregulation of the T4SS genes on one of the major operons in R. rickettsii was reported in the infected arthropod, while another major operon was downregulated [7]. In our case, the A1G_02205–02240 region (virB9, virB8, virB9, virB10, virB11, virD11) was significantly downregulated in vivo and A1G_00815–00840 was not significantly changed (Fig 4, S5 Table).

Fig 4. Analysis of genes encoding for Type IV Secretion System.

The 18 annotated virB genes are analyzed for total in vivo transcription (RPKM) and changes to transcription in vivo as compared to in vitro. Colored bars padj<0.01.

In vivo suppression of genes encoding for ribosomal proteins

Possibly the most conspicuous transcriptional change observed in vivo is the consistent and significant down regulation of the genes encoding for the large and small ribosomal subunits. It is pertinent to note that all RNA samples were treated to remove bacterial ribosomal RNA (rRNA) prior to sequencing. However, rRNA depletion does not affect the concentration of mRNA nucleotides that encode for the protein components of the ribosome. 28 of the 47 genes encoding for ribosomal subunit proteins (rpl, rpm, rps) were significantly downregulated, with no genes upregulated in vivo (Fig 5, S5 Table). This is a striking downregulation of the translational apparatus, and mirrors the results found in R. conorii inoculation eschars [10]. This finding suggests that R. rickettsii is potentially reducing its translational capacity in vivo, and is consistent with a stringent response [42].

Fig 5. Downregulation of genes encoding for large (rpl / rpm) and small (rps) ribosomal subunits.

Genes encoding for protein subunits of the large ribosomal subunit (rpl, rpm) and small ribosomal subunit (rps) graphed for changes in transcription in vivo in relation to in vitro. 28 of 47 genes are significantly repressed in vivo. Colored bars padj<0.01.


Herein, we describe the use of RNAseq technology to ascertain quantity of and changes to R. rickettsii RNA content in vivo compared to in vitro. The in vitro RNA samples are prepared from R. rickettsii in routine Vero cell culture at the point of maximal bacterial burden. The transcriptome of those bacterial samples is compared to in vivo R. rickettsii derived from infected murine liver at the point of maximum morbidity. As such, comparison of these samples establishes adjustments made by the bacterium in response to the challenges of mammalian infection. The RNAseq data was validated by qRT-PCR to establish the validity of the transcriptomic data (Fig 1). RNAseq analysis of changes to the transcriptome demonstrated significantly altered transcription in 37% of the genes of the R. rickettsii genome. Importantly, we demonstrate changes to the quantity of transcripts encoding for secreted proteins (Fig 2), Ank domain containing proteins (Fig 3), the Type IV secretion apparatus (Fig 4), and translation machinery (Fig 5). Together, these results demonstrate that R. rickettsii actively adapts to its environment, and the quantity of many known or potential virulence factors are significantly changed in vivo.

The high bacterial load in the liver in this mouse model of R. rickettsii infection is different than occurs in the well-known guinea pig model of infection. R. rickettsii- associated liver pathology in human cases has been described in the past, with specific mention of mononuclear hepatitis [4345]. Liver pathology has also been noted for infection with the closely related bacterium, R. conorii [46]. Additionally, rickettsial infection of cells of mononuclear and hepatocyte cells has been extensively examined in the past [4749]. As such, the infected cell types and lesions observed in the mouse model of R. rickettsii infection are observed in human infections.

Since the vast majority of analyses into the molecular mechanisms of obligate intracellular bacteria pathogenesis are performed in a tissue culture setting, a relevant question pertains to the makeup of the bacteria within the infected animal. Herein, we establish that the transcriptional makeup of the bacteria in vivo is different than has been traditionally analyzed in vitro. An example of this phenomenon is described in Fig 2 where many genes encoding for secreted and surface proteins are regulated in vivo. This group of proteins includes the defined antigens on the R. rickettsii surface, including OmpA and OmpB [5053]. The importance of defining in vivo expression of these surface proteins is exemplified by complications stemming from vaccination against the Borrelia burgdorferi OspA antigen which is entirely repressed within the infected mammal [5456]. Of the major R. rickettsii surface antigens, ompA, sca1, adr1, adr2, and 17kDa antigen (A1G_07075) are all downregulated in vivo (Fig 2, S3 Table). However, despite decreased in vivo expression, there is substantial transcript present for all of these antigens, suggesting that targeting the encoded proteins still has potential for preventative efficacy. In contrast, the most prevalent transcript encoding for secreted proteins is the major surface antigen ompB. This gene is expressed at very high levels both in vivo and in vitro, and therefore remains a primary target for future attempts at preventative vaccination. Based on the expression levels demonstrated in Fig 2, there are additional genes with high in vivo expression levels (>1000 RPKM), including ostA, A1G_02215, pal, and A1G_02060. We posit that these and other highly expressed secreted proteins merit further investigation into the potential for contributing to R. rickettsii pathogenesis and for advancing efforts at preventative therapeutics.

Many of the R. rickettsii genes that catalyze the biosynthesis or transport of lipopolysaccharide (LPS) are actively regulated (S5 Table). Regulation of the quantity or makeup of LPS in mammalian pathogens is well documented [57]. In fact, the intracellular bacterium Coxiella burnetii actively regulates LPS content as an immune evasion strategy [5860]. R. rickettsii contains a polysaccharide-rich electron-translucent region surrounding the bacterial membrane [61]. This LPS-rich area has been found to vary in its physical properties, but the source of this variation is unknown. As such, the transcriptional changes in LPS biosynthesis and transport suggests that R. rickettsii LPS structure may vary between cell culture, within the infected mammal, and in arthropod hosts [62]. Exploration of the consequences of this phenomenon may illuminate rickettsial pathobiology.

The R. rickettsii Ank domain containing proteins are also presumed to be secreted. Multiple roles for Ank proteins within the mammalian host cytoplasm have been defined [35]. The R. rickettsii ank genes are variably regulated with no discernable expression pattern (Fig 3). However, A1G_01760 is a particularly interesting gene, because it encodes for a Rickettsia ankyrin-repeat protein 1 (RARP-1) homolog with a predicted N-terminal Sec secretion signal and C-terminal Ank domain [63]. The very high level of in vivo transcription (average RPKM = 763) suggests that this protein is readily transcribed/translated, and should be found in abundance outside of the host cell. A1G_02960 and A1G_00070 are also upregulated and highly expressed in vivo, but lack a probable Sec secretion signal. These high levels of expression suggest a significant in vivo function if these two proteins are secreted into the host cell.

All Rickettsia species encode for a Type IV Secretion System (T4SS). T4SS have been demonstrated to be essential in various other pathogens [38]. The genes encoding for the Rickettsia rickettsii T4SS macromolecular structure (virB) are found in a series of scattered genetic loci with multiple gene duplications [40]. There are two major loci encoding for VirB proteins, and the locus containing virB4 and five vir6 genes is not regulated (Fig 4). In contrast, the virB9, virB8, virB9, virB10, virB11, virD11 (A1G_02205–02240) locus is significantly repressed in vivo (Fig 4). These findings suggest that the transmembrane Type IV secretion apparatus is present, but may be underrepresented within the infected mammal.

As demonstrated in Fig 5, genes encoding for translational machinery are significantly downregulated in vivo. It is again important to note that rRNA depletion has no effect on the presence of mRNA encoding for ribosomal proteins. The downregulation of genes encoding for ribosomal proteins is similar to a previous report describing translation machinery downregulation R. conorii within inoculation eschars [10]. Though R. conorii eschar sample and R. rickettsii liver samples are quite different, both analyses reached the same conclusion, thereby suggesting that this may be a widespread phenomenon.

In contrast to the shared observation of ribosomal mRNA downregulation, in vivo analysis of R. conorii eschar samples and R. rickettsii liver produced quite different results. The upregulated transcripts reported in R. conorii did not correlate with the present study. 22 of the 31 reported upregulated R. conorii transcripts could be definitively assigned to a R. rickettsii homolog. Only 14 of the 22 shared genes were upregulated in in vivo in both R. conorii and R. rickettsii. Taken together, these results suggest that while both in vivo analyses found a decrease in translational machinery, the R. conorii global transcriptome in an inoculation eschar differs from R. rickettsii in the infected liver. This is not unexpected as Rickettsia in human eschar and murine liver are experiencing different challenges. Though Rickettsia are frequently cultured from inoculation eschars [6466], it is clear that these bacteria are located in a zone of limited fluid perfusion within significantly necrotic dermis and epidermis. In contrast, liver derived R. rickettsii are infecting a heterogeneous population of mononuclear cells (Kupffer cells), endothelial cells, and hepatocytes [11]. As such, care must be taken when comparing the only two transcriptional analysis of Rickettsia species in a mammal.

In summary, we have utilized RNAseq technologies to define the in vivo transcriptome of R. rickettsii, the causative agent of Rocky Mountain spotted fever. The global RNA profile in vivo is vastly different than is found in routine tissue culture. This data further advances the suitability of those genes that are highly expressed or upregulated in vivo as research topics. We have described regulation of RNAs encoding for secreted proteins, LPS production, Type IV secretion, and translational machinery. The observed regulation correlates with our hypothesis that R. rickettsii utilizes transcriptional alterations during pathogenesis in the mammalian host. Investigation into the molecular mechanisms of transcriptional regulation will greatly increase our understanding of the pathogenesis of these severe human infections, and experimental disruption of regulation may contribute to development of interruptive therapeutics.

Supporting information

S1 Fig. Diagram of programs used to analyze RNAseq reads to determine transcriptional changes.


S1 Table. Summary of quantity, quality, and length of sequencing reads prior to bioinformatic analysis.


S2 Table. Genes targeted for qPCR validation, primer sequences, and calculated fold changes (log2(in vivo / in vitro)).


S3 Table. RNA content (RPKM), transcriptional changes, and signal sequence prediction results.


S4 Table. In vivo expression (RPKM), fold change (log2(in vivo / in vitro)), and significance (padj) for all R. rickettsii protein encoding genes.


S5 Table. RNA content and transcriptional changes of genes encoding for LPS biosynthesis/transport, Ank-domain containing proteins, Type IV Secretion System, and ribosomal proteins.



These studies were aided by the Gene Lab unit of the Center for Experimental Infectious Disease and Research at the Louisiana State University School of Veterinary Medicine. We would like to particularly acknowledge the efforts of Vladimir Chouljenko and John Caskey in completion of this study. The authors would like to thank Pedro Curto, Isaura Simões, Abigail Fish, and Daniel Garza for helpful suggestions in completion of this study and preparation of this manuscript.


  1. 1. Alvarez-Hernandez G, Roldan JF, Milan NS, Lash RR, Behravesh CB, Paddock CD. Rocky Mountain spotted fever in Mexico: past, present, and future. Lancet Infect Dis. 2017;(17):30173–1. pmid:28365226.
  2. 2. Toledo-Arana A, Dussurget O, Nikitas G, Sesto N, Guet-Revillet H, Balestrino D, et al. The Listeria transcriptional landscape from saprophytism to virulence. Nature. 2009;459(7249):950–6. pmid:19448609.
  3. 3. Audia JP, Patton MC, Winkler HH. DNA microarray analysis of the heat shock transcriptome of the obligate intracytoplasmic pathogen Rickettsia prowazekii. Applied and environmental microbiology. 2008;74(24):7809–12. pmid:18952868; PubMed Central PMCID: PMCPMC2607166.
  4. 4. Dreher-Lesnick SM, Ceraul SM, Rahman MS, Azad AF. Genome-wide screen for temperature-regulated genes of the obligate intracellular bacterium, Rickettsia typhi. BMC microbiology. 2008;8:61. pmid:18412961; PubMed Central PMCID: PMCPMC2335108.
  5. 5. Ellison DW, Clark TR, Sturdevant DE, Virtaneva K, Hackstadt T. Limited transcriptional responses of Rickettsia rickettsii exposed to environmental stimuli. PloS one. 2009;4(5):e5612. pmid:19440298; PubMed Central PMCID: PMCPMC2680988.
  6. 6. Galletti MF, Fujita A, Nishiyama MY Jr., Malossi CD, Pinter A, Soares JF, et al. Natural blood feeding and temperature shift modulate the global transcriptional profile of Rickettsia rickettsii infecting its tick vector. PloS one. 2013;8(10):e77388. pmid:24155949; PubMed Central PMCID: PMCPMC3796454.
  7. 7. Galletti MF, Fujita A, Rosa RD, Martins LA, Soares HS, Labruna MB, et al. Virulence genes of Rickettsia rickettsii are differentially modulated by either temperature upshift or blood-feeding in tick midgut and salivary glands. Parasit Vectors. 2016;9(1):331. pmid:27287539; PubMed Central PMCID: PMCPMC4902979.
  8. 8. La MV, Francois P, Rovery C, Robineau S, Barbry P, Schrenzel J, et al. Development of a method for recovering rickettsial RNA from infected cells to analyze gene expression profiling of obligate intracellular bacteria. J Microbiol Methods. 2007;71(3):292–7. Epub 2007/10/30. S0167-7012(07)00325-9 [pii] pmid:17964675.
  9. 9. Sears KT, Ceraul SM, Gillespie JJ, Allen ED Jr., Popov VL, Ammerman NC, et al. Surface proteome analysis and characterization of surface cell antigen (Sca) or autotransporter family of Rickettsia typhi. PLoS pathogens. 2012;8(8):e1002856. pmid:22912578; PubMed Central PMCID: PMCPMC3415449.
  10. 10. Renesto P, Rovery C, Schrenzel J, Leroy Q, Huyghe A, Li W, et al. Rickettsia conorii transcriptional response within inoculation eschar. PloS one. 2008;3(11):e3681. Epub 2008/11/11. pmid:18997861.
  11. 11. Riley SP, Cardwell MM, Chan YG, del Piero F, Martinez JJ. Failure of a heterologous recombinant Sca5/OmpB protein based vaccine to elicit effective protective immunity against Rickettsia rickettsii infections in C3H/HeN mice. Pathog Dis. 2015;73(9):ftv101. Epub 2015 Oct 29. PubMed Central PMCID: PMC PMC4732028 pmid:26519448
  12. 12. Ammerman NC, Beier-Sexton M, Azad AF. Laboratory maintenance of Rickettsia rickettsii. Curr Protoc Microbiol. 2008;Chapter 3:Unit 3A 5. Epub 2008/11/20. pmid:19016440.
  13. 13. Chan YG, Riley SP, Martinez JJ. Adherence to and invasion of host cells by spotted Fever group rickettsia species. Frontiers in microbiology. 2010;78(5):1895–904. Epub 2010/02/22. PubMed PMID: 21687751.
  14. 14. Haglund CM, Choe JE, Skau CT, Kovar DR, Welch MD. Rickettsia Sca2 is a bacterial formin-like mediator of actin-based motility. Nat Cell Biol. 2010;12(11):1057–63. Epub 2010/10/26. ncb2109 [pii] pmid:20972427.
  15. 15. Renesto P, Dehoux P, Gouin E, Touqui L, Cossart P, Raoult D. Identification and characterization of a phospholipase D-superfamily gene in rickettsiae. J Infect Dis. 2003;188(9):1276–83. Epub 2003/11/01. JID30594 [pii] pmid:14593584.
  16. 16. Walker DH, Valbuena GA, Olano JP. Pathogenic mechanisms of diseases caused by Rickettsia. Annals of the New York Academy of Sciences. 2003;990:1–11. Epub 2003/07/16. pmid:12860594.
  17. 17. Gong W, Xiong X, Qi Y, Jiao J, Duan C, Wen B. Identification of novel surface-exposed proteins of Rickettsia rickettsii by affinity purification and proteomics. PloS one. 2014;9(6):e100253. Epub 2014/06/21. pmid:24950252; PubMed Central PMCID: PMC4065002.
  18. 18. Cardwell MM, Martinez JJ. The Sca2 autotransporter protein from Rickettsia conorii is sufficient to mediate adherence to and invasion of cultured mammalian cells. Infection and immunity. 2009;77(12):5272–80. Epub 2009/10/05. IAI.00201-09 [pii] pmid:19805531.
  19. 19. Chan YG, Cardwell MM, Hermanas TM, Uchiyama T, Martinez JJ. Rickettsial outer-membrane protein B (rOmpB) mediates bacterial invasion through Ku70 in an actin, c-Cbl, clathrin and caveolin 2-dependent manner. Cellular microbiology. 2009;11(4):629–44. Epub 2009/01/07. CMI1279 [pii] pmid:19134120.
  20. 20. Dasch GA, Samms JR, Williams JC. Partial purification and characterization of the major species-specific protein antigens of Rickettsia typhi and Rickettsia prowazekii identified by rocket immunoelectrophoresis. Infection and immunity. 1981;31(1):276–88. Epub 1981/01/01. pmid:6783537.
  21. 21. Riley SP, Goh KC, Hermanas TM, Cardwell MM, Chan YG, Martinez JJ. The Rickettsia conorii autotransporter protein Sca1 promotes adherence to nonphagocytic mammalian cells. Infection and immunity. 2010;78(5):1895–904. Epub 2010/02/24. IAI.01165-09 [pii] pmid:20176791.
  22. 22. Gillespie JJ, Kaur SJ, Rahman MS, Rennoll-Bankert K, Sears KT, Beier-Sexton M, et al. Secretome of obligate intracellular Rickettsia. FEMS Microbiol Rev. 2015;39(1):47–80. pmid:25168200; PubMed Central PMCID: PMCPMC4344940.
  23. 23. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnetjournal. 2011;17(1):10–2.
  24. 24. Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 2013;14(4):R36. pmid:23618408; PubMed Central PMCID: PMCPMC4053844.
  25. 25. SeqManNGEN. 13 ed. Madison, WI USA: DNASTAR Lastergene; 2016.
  26. 26. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods. 2008;5(7):621–8. pmid:18516045.
  27. 27. Anders S, Pyl PT, Huber W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31(2):166–9. pmid:25260700; PubMed Central PMCID: PMCPMC4287950.
  28. 28. Kall L, Krogh A, Sonnhammer EL. Advantages of combined transmembrane topology and signal peptide prediction—the Phobius web server. Nucleic acids research. 2007;35(Web Server issue):W429–32. pmid:17483518; PubMed Central PMCID: PMCPMC1933244.
  29. 29. Bagos PG, Nikolaou EP, Liakopoulos TD, Tsirigos KD. Combined prediction of Tat and Sec signal peptides with hidden Markov models. Bioinformatics. 2010;26(22):2811–7. pmid:20847219.
  30. 30. Petersen TN, Brunak S, von Heijne G, Nielsen H. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat Methods. 2011;8(10):785–6. pmid:21959131.
  31. 31. Juncker AS, Willenbrock H, Von Heijne G, Brunak S, Nielsen H, Krogh A. Prediction of lipoprotein signal peptides in Gram-negative bacteria. Protein Sci. 2003;12(8):1652–62. pmid:12876315; PubMed Central PMCID: PMCPMC2323952.
  32. 32. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. pmid:25516281; PubMed Central PMCID: PMCPMC4302049.
  33. 33. Ozsolak F, Milos PM. Transcriptome profiling using single-molecule direct RNA sequencing. Methods Mol Biol. 2011;733:51–61. pmid:21431762; PubMed Central PMCID: PMCPMC3272358.
  34. 34. Bishop RE. Structural biology: Lipopolysaccharide rolls out the barrel. Nature. 2014;511(7507):37–8. pmid:24990738; PubMed Central PMCID: PMCPMC5007118.
  35. 35. Al-Khodor S, Price CT, Kalia A, Abu Kwaik Y. Functional diversity of ankyrin repeats in microbial proteins. Trends in microbiology. 2010;18(3):132–9. pmid:19962898; PubMed Central PMCID: PMCPMC2834824.
  36. 36. Beyer AR, Rodino KG, VieBrock L, Green RS, Tegels BK, Oliver LD Jr., et al. Orientia tsutsugamushi Ank9 is a multifunctional effector that utilizes a novel GRIP-like Golgi localization domain for Golgi-to-endoplasmic reticulum trafficking and interacts with host COPB2. Cellular microbiology. 2017. pmid:28103630.
  37. 37. Rennoll-Bankert KE, Rahman MS, Guillotte ML, Lehman SS, Beier-Sexton M, Gillespie JJ, et al. RalF-Mediated Activation of Arf6 Controls Rickettsia typhi Invasion by Co-Opting Phosphoinositol Metabolism. Infection and immunity. 2016;84(12):3496–506. pmid:27698019; PubMed Central PMCID: PMCPMC5116726.
  38. 38. Byndloss MX, Rivera-Chavez F, Tsolis RM, Baumler AJ. How bacterial pathogens use type III and type IV secretion systems to facilitate their transmission. Current opinion in microbiology. 2016;35:1–7. pmid:27621139.
  39. 39. Gillespie JJ, Brayton KA, Williams KP, Diaz MA, Brown WC, Azad AF, et al. Phylogenomics reveals a diverse Rickettsiales type IV secretion system. Infection and immunity. 2010;78(5):1809–23. pmid:20176788; PubMed Central PMCID: PMCPMC2863512.
  40. 40. Gillespie JJ, Phan IQ, Driscoll TP, Guillotte ML, Lehman SS, Rennoll-Bankert KE, et al. The Rickettsia type IV secretion system: unrealized complexity mired by gene family expansion. Pathog Dis. 2016;74(6). pmid:27307105.
  41. 41. Gillespie JJ, Phan IQ, Scheib H, Subramanian S, Edwards TE, Lehman SS, et al. Structural Insight into How Bacteria Prevent Interference between Multiple Divergent Type IV Secretion Systems. mBio. 2015;6(6):e01867–15. pmid:26646013; PubMed Central PMCID: PMCPMC4676284.
  42. 42. Boutte CC, Crosson S. Bacterial lifestyle shapes stringent response activation. Trends in microbiology. 2013;21(4):174–80. pmid:23419217; PubMed Central PMCID: PMCPMC4238387.
  43. 43. Adams JS, Walker DH. The liver in Rocky Mountain spotted fever. Am J Clin Pathol. 1981;75(2):156–61. pmid:6781327.
  44. 44. Arguello AP, Hun L, Rivera P, Taylor L. A fatal urban case of rocky mountain spotted fever presenting an eschar in San Jose, Costa Rica. The American journal of tropical medicine and hygiene. 2012;87(2):345–8. pmid:22855769; PubMed Central PMCID: PMCPMC3414575.
  45. 45. Jackson MD, Kirkman C, Bradford WD, Walker DH. Rocky mountain spotted fever: hepatic lesions in childhood cases. Pediatr Pathol. 1986;5(3–4):379–88. pmid:3537996.
  46. 46. Walker DH, Gear JH. Correlation of the distribution of Rickettsia conorii, microscopic lesions, and clinical features in South African tick bite fever. The American journal of tropical medicine and hygiene. 1985;34(2):361–71. pmid:3885774.
  47. 47. Feng HM, Walker DH. Mechanisms of intracellular killing of Rickettsia conorii in infected human endothelial cells, hepatocytes, and macrophages. Infection and immunity. 2000;68(12):6729–36. Epub 2000/11/18. pmid:11083788.
  48. 48. Kenyon RH, Ascher MS, Kishimoto RA, Pedersen CE Jr. In vitro guinea pig leukocyte reactions to Rickettsia rickettsii. Infection and immunity. 1977;18(3):840–6. pmid:412792; PubMed Central PMCID: PMCPMC421310.
  49. 49. Radulovic S, Price PW, Beier MS, Gaywee J, Macaluso JA, Azad A. Rickettsia-macrophage interactions: host cell responses to Rickettsia akari and Rickettsia typhi. Infection and immunity. 2002;70(5):2576–82. Epub 2002/04/16. pmid:11953398; PubMed Central PMCID: PMC127898.
  50. 50. Bourgeois AL, and G.A. Dasch The species-specific surface protein antigen of Rickettsia typhi: immunogenicity and protective efficacy in guinea pigs. In: Anacker WBaRL, editor. Rickettsiae and Rickettsial Diseases. New York: Academic Press; 1981. p. 71–80.
  51. 51. Dasch GA. Isolation of species-specific protein antigens of Rickettsia typhi and Rickettsia prowazekii for immunodiagnosis and immunoprophylaxis. J Clin Microbiol. 1981;14(3):333–41. Epub 1981/09/01. pmid:6793620.
  52. 52. Feng HM, Whitworth T, Olano JP, Popov VL, Walker DH. Fc-dependent polyclonal antibodies and antibodies to outer membrane proteins A and B, but not to lipopolysaccharide, protect SCID mice against fatal Rickettsia conorii infection. Infection and immunity. 2004;72(4):2222–8. Epub 2004/03/25. pmid:15039346.
  53. 53. Li H, Lenz B, Walker DH. Protective monoclonal antibodies recognize heat-labile epitopes on surface proteins of spotted fever group rickettsiae. Infection and immunity. 1988;56(10):2587–93. Epub 1988/10/01. pmid:2458318.
  54. 54. de Silva AM, Telford SR, 3rd, Brunet LR, Barthold SW, Fikrig E. Borrelia burgdorferi OspA is an arthropod-specific transmission-blocking Lyme disease vaccine. J Exp Med. 1996;183(1):271–5. pmid:8551231; PubMed Central PMCID: PMCPMC2192397.
  55. 55. Fingerle V, Hauser U, Liegl G, Petko B, Preac-Mursic V, Wilske B. Expression of outer surface proteins A and C of Borrelia burgdorferi in Ixodes ricinus. J Clin Microbiol. 1995;33(7):1867–9. pmid:7665661; PubMed Central PMCID: PMCPMC228287.
  56. 56. Schwan TG, Piesman J, Golde WT, Dolan MC, Rosa PA. Induction of an outer surface protein on Borrelia burgdorferi during tick feeding. Proceedings of the National Academy of Sciences of the United States of America. 1995;92(7):2909–13. pmid:7708747; PubMed Central PMCID: PMCPMC42328.
  57. 57. Maldonado RF, Sa-Correia I, Valvano MA. Lipopolysaccharide modification in Gram-negative bacteria during chronic infection. FEMS Microbiol Rev. 2016;40(4):480–93. pmid:27075488; PubMed Central PMCID: PMCPMC4931227.
  58. 58. Kuley R, Bossers-deVries R, Smith HE, Smits MA, Roest HI, Bossers A. Major differential gene regulation in Coxiella burnetii between in vivo and in vitro cultivation models. BMC Genomics. 2015;16:953. pmid:26572556; PubMed Central PMCID: PMCPMC4647677.
  59. 59. Moos A, Hackstadt T. Comparative virulence of intra- and interstrain lipopolysaccharide variants of Coxiella burnetii in the guinea pig model. Infection and immunity. 1987;55(5):1144–50. pmid:3570458; PubMed Central PMCID: PMCPMC260482.
  60. 60. Shannon JG, Howe D, Heinzen RA. Virulent Coxiella burnetii does not activate human dendritic cells: role of lipopolysaccharide as a shielding molecule. Proceedings of the National Academy of Sciences of the United States of America. 2005;102(24):8722–7. pmid:15939879; PubMed Central PMCID: PMCPMC1150828.
  61. 61. Silverman DJ, Wisseman CL Jr., Waddell AD, Jones M. External layers of Rickettsia prowazekii and Rickettsia rickettsii: occurrence of a slime layer. Infection and immunity. 1978;22(1):233–46. Epub 1978/10/01. 83297. pmid:83297
  62. 62. Verhoeve VI, Jirakanwisal K, Utsuki T, Macaluso KR. Differential Rickettsial Transcription in Bloodfeeding and Non-Bloodfeeding Arthropod Hosts. PloS one. 2016;11(9):e0163769. pmid:27662479; PubMed Central PMCID: PMCPMC5035074.
  63. 63. Kaur SJ, Rahman MS, Ammerman NC, Beier-Sexton M, Ceraul SM, Gillespie JJ, et al. TolC-dependent secretion of an ankyrin repeat-containing protein of Rickettsia typhi. Journal of bacteriology. 2012;194(18):4920–32. pmid:22773786; PubMed Central PMCID: PMCPMC3430354.
  64. 64. Fournier PE, Xeridat B, Raoult D. Isolation of a rickettsia related to Astrakhan fever rickettsia from a patient in Chad. Annals of the New York Academy of Sciences. 2003;990:152–7. pmid:12860619.
  65. 65. Paddock CD, Koss T, Eremeeva ME, Dasch GA, Zaki SR, Sumner JW. Isolation of Rickettsia akari from eschars of patients with rickettsialpox. The American journal of tropical medicine and hygiene. 2006;75(4):732–8. pmid:17038703.
  66. 66. Paddock CD, Sumner JW, Comer JA, Zaki SR, Goldsmith CS, Goddard J, et al. Rickettsia parkeri: a newly recognized cause of spotted fever rickettsiosis in the United States. Clin Infect Dis. 2004;38(6):805–11. pmid:14999622.