Simultaneously Monitoring Immune Response and Microbial Infections during Pregnancy through Plasma cfRNA Sequencing.
Because cfRNA is derived from apoptotic bodies and exosomes, it is highly degraded and of low quantity. This makes it challenging to prepare sequencing libraries from cfRNA. The standard poly-A-based library preparation method is not applicable because of the low integrity. Instead, random primers are used for the reverse transcription. One downside of random priming-based amplification is that it amplifies the highly abundant but undesirable ribosomal RNA (rRNA). As a result, an additional rRNA removal step is usually required for random priming-based library preparation. Several RNA-seq library preparation methods have recently been proposed for low-quantity RNA samples, including RNase H depletion of rRNA, exome capture of target cDNA, and CRISPR to remove ribosomal cDNA. However, these methods have not been applied to cfRNA samples and their relative merits have not been carefully assessed.
In this study, we first systematically assessed different RNA-seq library preparation methods for cfRNA samples. We then characterized the dynamic changes of the plasma transcriptome and microbiome during pregnancy. Finally, we did a proof-of-principle case study using cfRNA sequencing to monitor infectious disease during pregnancy.
Materials and Methods
Blood samples from pregnant women were collected from Stanford hospital under an Institutional Review Board-approved protocol. The blood samples were collected into EDTA-coated Vacutainer Tubes. The blood sample were centrifuged at 1600g for 10 min at 4 [degrees]C as soon as possible, usually within 30 min to 2 h of collection, and the plasma was centrifuged at 16000g for 10 min at 4 [degrees]C to remove residual cells.
The cfRNA was extracted from 1 mL of plasma using a Plasma/Serum Circulating RNA and Exosomal Purification kit (Norgen, cat 42800). The DNA residual in the cfRNA was then digested using Baseline-ZERO[TM] DNase (Epicentre). The cfRNA was then purified using an RNA Clean & Concentrator-5 kit (Zymo), yielding 10 [micro]L of cfRNA per sample.
COMPARISON OF cfRNA-seq LIBRARY PREPARATION
We first evaluated 4 methods (ScriptSeq, RNase H, Exome, and Clontech) for preparing sequencing libraries from cfRNA. The ScriptSeq method uses random primers to synthesize cDNA from total cfRNA (Illumina ScriptSeq v2 kit). The RNase H method depletes rRNA (NEBNext[R] rRNA Depletion Kit) from the total cfRNA using RNase H before the RNA library preparation (Illumina ScriptSeq v2 kit). The Clontech method removes ribosomal cDNA after reverse transcription using CRISPR technology (Clontech SMARTer Stranded Total RNA-Seq Kit-Pico). The Exome method captures the exome regions of the cDNA library using probe hybridization (Roche NimbleGen SeqCap EZ Library SR). Multiple biological replicates of plasma samples (1 mL per sample) from healthy donors were used for each method. Eight plasma samples were used for each of the ScriptSeq, Exome, and Clontech methods, and 2 samples were used for the Rnase H method. The input cfRNA for each of these libraries was isolated using the same cfRNA isolation methods described above, using 5 [micro]L of cfRNA each. See more details in in the Data Supplement that accompanies the online version of this article at http:// www.clinchem.org/content/vol63/issue11.
QUANTIFICATION OF HUMAN TRANSCRIPTOME IN cfRNA
For each library, more than 10 million sequencing reads (2 X 75 bp) were generated using an Illumina NextSeq sequencer. Reads were aligned to hg19 human transcriptome using the STAR (4) aligner. The read counts for each gene were calculated using the htseq-count (5) tool. The number of reads mapped to each gene was normalized with the total number of reads mapped to human transcriptome.
QUANTIFICATION OF MICROBIOME IN cfRNA
Reads in the cfRNA libraries that failed to align to human transcriptome were extracted and aligned to the bacteriophage [phi] X 174, which is used as a control sequence in the Illumina platform. The remaining reads were aligned against a custom database (6) of potential pathogens that encompasses viruses, bacteria, and fungi in the NCBI database using BLAST (2.2.28+). Alignments were required to have an identity of at least 90% across 90% of the bases of the query. For high-confidence reads that aligned to multiple microbes, a customized Python script was used to find the lowest common ancestor in the taxonomical tree for each such read.
DETECTION OF PATHOGENIC VIRUSES FROM cfRNA
The virus species belonging to known human pathogens were included in the analysis (see Table S1 in the online Data Supplement). For each sample, the number of reads that was uniquely aligned to these virus species were counted. In this study, each patient had multiple cfRNA samples collected from different time points. The prevalence was calculated at the patient level. For each patient, if there was at least one read aligned to a certain pathogenic species in any of her samples, this virus was counted as detected in this patient.
QUANTIFICATION OF MICROBIOME FROM cfDNA
Plasma cfDNA libraries were prepared for selected samples with viral infections as a validation for cfRNA microbiome results. The method used for cfDNA sequencing library preparation and microbiome analysis was the same method described previously (6).
CALCULATION OF THE QC METRICS OF RNA-seq LIBRARIES
The QC metrics, including the rRNA rate, estimated library size, and strand-specificity for each sample, was calculated using the RNA-SeQC program (7).
IDENTIFY cfRNA GENES WITH VARIATIONS IN ABUNDANCE DURING PREGNANCY
In total, 204 plasma samples (see Table S2 in the online Data Supplement) were collected from 60 pregnant women at 4 time points: first trimester (T1: week 1 to week 13), second trimester (T2: week 13 to week 23), third trimester ([T.sub.3]: week 23 to week 40), and postpartum (P: within 24 h after delivery). The cfRNA-seq libraries were made using the ScriptSeq v2 kit. The libraries with poor sample quality (due to RNA degradation, DNA contamination, or blood coagulation) were removed from the subsequent statistical analysis (see the online Data Supplement). Samples from women with preterm birth were also removed from the analysis. A generalized linear model likelihood ratio test was used to determine the genes with variations in abundance between any pair of time points (T1, T2, [T.sub.3], P). The generalized linear model test was implemented using the edgeR version 3.12.0 package in R, using the glmFit() and glmLRT() functions.
To validate the genes with temporal variations identified by RNA-seq, qPCR was performed on the same sets of plasma samples using the Fluidigm Biomark system. See more details in the in the online Data Supplement.
ASSESSING DIFFERENT METHODS FOR cfRNA-seq
We evaluated the 4 cfRNA-seq library preparation methods (ScriptSeq, RNase H, Exome and Clontech) by comparing the rRNA rate (the fraction of reads mapped to rRNA among all reads mapped to human genome), microbiome fraction (the fraction of reads aligned to microbial organisms among all aligned reads), estimated library size (the number of expected fragments based on the total reads and duplication rate assuming a Poisson distribution), and the number of genes detected (Fig. 1 and Table S3 in the online Data Supplement). All of the values discussed in the following section were the average value among the biological replicates.
The rRNA rate is as high as 0.66 when preparing the RNA-seq library directly from total cfRNA (ScriptSeq). The other 3 methods all include an rRNA removal step that results in reduced rRNA rates of different levels (0.24 for RNase H, 0.05 for Exome, and 0.10 for Clontech). The ScriptSeq, Exome, and Clontech methods can recover approximately 9000 protein-coding genes, whereas the RNase H method can only detect approximately 3700 genes. The ScriptSeq and Clontech methods are also able to recover a broad spectrum of noncoding RNAs, including lincRNA, snRNA, miscRNA, etc. In addition to the dominant reads of human origin, a small fraction ofreads was mapped to microbial genomes (0.02 for ScriptSeq, 0.16 for RNase H, 0.18 for Clontech, and 0.009 for Exome libraries). For these methods, the estimated library size (0.64 million for ScriptSeq, 0.31 million for Clontech, 0.29 million for Exome, and 0.06 million for RNase H) is very small, which correlates to the scarce amount of cfRNA in plasma.
Based on the above analysis, the RNase H method has low sensitivity for cfRNA-seq library preparation. The decrease in the number of genes detected and the estimated library size in the RNase H method implies that a large number of mRNA transcripts were lost during the RNase H depletion step, either due to nonspecific hybridization of the ssDNA or RNA degradation. Both the Exome and Clontech methods were effective in the removal of rRNA and a comparable number of mRNA transcripts was recovered. The Clontech method is a better option if the microbiome information is of interest. However, because of the small number of unique molecules in the cfRNA-seq library, it becomes possible to achieve sequencing saturation at an affordable sequencing depth even without rRNA depletion. After comparing the cost of the reagents for adding the rRNA depletion step with the cost for the extra depth of sequencing, we found that there is no significant difference between the ScriptSeq and the Clontech methods in terms of the performance and cost. We chose to use the ScriptSeq method for the subsequent pregnant women's study.
THE DYNAMICS OF cfRNA REFLECT THE IMMUNE RESPONSE DURING PREGNANCY
We systematically examined the temporal variation of human plasma transcriptome during pregnancy. In total, 39 genes (Table S4 in the online Data Supplement) were identified with variations in the abundance among the 4 time points (false discovery rate <0.05). We applied hierarchical clustering to these genes and separated them into 4 major clusters based on their temporal trends (Fig. 2A).
For genes in cluster 1, the abundance of genes increased from the first trimester to the third trimester and remained at a higher level than during the first trimester after delivery. These genes are specifically expressed in leukocytes and participate in the immune modulation of pregnancy. Two major types of immune response were identified based on gene ontology and pathway analysis: regulation of inflammatory process and increased abundance of endogenous antimicrobials. Genes involved in inflammatory regulation include annexin (ANXA1)  and S100 calcium binding protein (S100A8, S100A9, and S100P). Increased expression of annexin is one of the mechanisms by which glucocorticoids inhibit inflammation (8). S100A8/A9 is also known to be involved in the migration of neutrophil migration to inflammatory sites (9).
A number of antimicrobial proteins and peptides showed an increased cfRNA abundance during pregnancy, including bactericidal/permeability-increasing protein (BPI) and peptidoglycan recognition protein (PGLYRP1). These are cytotoxic peptides or proteins synthesized mostly in neutrophils, which participate in innate immune response to fight against bacteria, fungi, and viruses.
Genes in cluster 2 are placenta-specific and only detectable during pregnancy. These genes participate in hormonal regulation (CSH1, CSH2, and CGA), immune regulation (PSG1), and tissue remodeling (PAPPA). Most of these genes were reported in our previous study (2), and the current results independently reproduced those findings.
Genes in cluster 3 decrease in abundance during the first trimester. Some of these genes (ANK1 and ALAS2) are specific to erythrocytes (red blood cells). It is known that the blood volume increases more than the red blood cell mass during pregnancy, which results in a relative anemia. This might explain the decreased abundance of these erythrocyte-related genes in the plasma at the beginning of pregnancy. There are also several genes that have lower abundance levels in the third trimester. These genes are not enriched for a specific functional group or tissue of origin; their relevance to pregnancy needs investigation.
To prove that these temporal variations in abundance were not caused by RNA-seq technical artifacts, we selected a subset of the genes related to the immune response and pregnancy hormones for qPCR validation. The qPCR was performed on the same pregnant sample set used for cfRNA-seq analysis. The results showed consistent temporal trends in qPCR compared to the original RNA-seq analysis (Fig. 2B; Fig. S2 in the online Data Supplement).
THE COMPOSITION AND TEMPORAL VARIATIONS OF PLASMA MICROBIOME DURING PREGNANCY
After examining the transcriptome, we analyzed the microbiome from plasma cfRNA-seq. Microbial reads in the cfRNA library can be from the blood microbiota or from contamination introduced during sample preparation. Negative controls (water) were prepared with the ScriptSeq method from the cfRNA extraction step. The microbial species detected in the negative controls were identified as contaminants and removed from the following microbiome analysis (see online Data Supplement). The microbial reads from the plasma covered all four superkingdoms. Fig. 3A displays the microbial read distribution. The percentage of each microbial group is the average value across all the samples.
Overall, the composition of the plasma microbiome remained relatively stable during pregnancy for both bacteria and viruses (Fig. 3B, C), attesting to the well-orchestrated balance between the immune system and the microbiome. We performed a statistical test (ANOVA) to find microbial genera that exhibited a change in abundance across the 4 time points (T1, T2, T3, and P) of pregnancy. Ureaplasma showed an increased average abundance and higher prevalence at postpartum (Benjamini-Hochberg adjusted P-value 0.0384, Fig. 3D, E). Two species of Ureaplasma--Ureaplasma parvum and Ureaplasma urealyticum--were detected in the cfRNA samples. Both species showed higher abundance and prevalence postpartum (see Fig. S3 in the online Data Supplement), and the majority of the reads in the Ureaplasma genus were aligned to Ureaplasma parvum. In a previous study, Ureaplasma was found in the blood stream of women with postpartum fever (10). The increased level of Ureaplasma at postpartum might be a reflection of microbial translocation into the maternal blood stream during delivery or an indication of Ureaplasma infection after delivery. Ureaplasma has an important clinical relevance with pregnancy complications. A recent study reported that the abundance of Ureaplasma in the vaginal microbiome is associated with preterm birth (11).
MONITORING VIRAL INFECTION DURING PREGNANCY
Viral infections in pregnancy are major causes of maternal and fetal morbidity and mortality. Despite the increased risk to pregnant women due to their immune suppressed state from a broad spectrum of viruses, only a small number of species are tested for in current routine pregnancy screening. We hypothesized that plasma cfRNA-seq can be used as an unbiased screening test for viral infection during pregnancy. A number of common pathogenic viruses were detected in our cfRNA samples, including Adenoviridae, Papillomaviridae, Hepesiviridae. Parvoviridae, and Polyomaviridae (Fig. 4A; Table S5 in the online Data Supplement). In particular, human parvovirus B19 (B19V) was detected from one pregnant plasma sample (patient 10039) using cfRNA-seq. The viral load was very high at the first trimester, and it gradually decreased in the later stages of pregnancy. The temporal change of B19V load in this patient was also validated from cfDNA sequencing data (Fig. 4B). B19V is known to attack erythroid progenitor cells and can trigger an acute cessation of erythrocyte production (12). Based on the cell-free transcriptomic data of the same patient, the abundance of erythrocyte-specific genes (ALAS2 and GYPA) plunged at the time of strong B19V infection (Fig. 4C). Parvovirus B19 virus infection was not diagnosed for this pregnant woman according to her clinical record. The estimated incidence of parvovirus B19 virus infection in pregnancy is 1%-5%. Although B19V infection can be asymptomatic during pregnancy, 3%-14% of infections lead to intrauterine fetal deaths (13). Fortunately, this woman recovered from the infection without clinical complications and there were no adverse outcomes for the fetus.
THE IMMUNE RESPONSE DURING PREGNANCY FROM cfRNA IS CONSISTENT WITH DATA FROM PREVIOUS MEASUREMENTS
There has been long-standing interest in the role of the maternal immune system during pregnancy, and it is clear from a variety of lines of evidence that pregnancy acts as a complicated modulator of the immune system. From the perspective of cell count, absolute neutrophil counts are increased during pregnancy (14). Because neutrophils are a major site for the synthesis of antimicrobial peptides, this might contribute to the increase of antimicrobial gene transcripts in plasma cfRNA during pregnancy. However, it is also known that placental trophoblasts synthesize antimicrobial peptides; this illustrates the value of measuring cfRNA levels as an approach that integrates the overall contribution of antimicrobial peptides across multiple cell types and tissues. From the perspective of the whole blood transcriptome, the gene expression levels ofMMP8, ANXA1, S100A8, and S100P have been previously shown to be up-regulated during pregnancy (15, 16) and some of these immune-related genes have been demonstrated as biomarkers for pregnancy complications. One recent study showed that women with preeclamptic pregnancy have aberrantly high levels of annexin protein in their blood (8). Evidence from another pathophysiological study suggests that inflammatory proteins S100A8 and S100A9 play a role in recurrent early pregnancy loss (17).
SAMPLE QUALIY CONTROL
In the study of temporal changes in plasma transcriptome and microbiome, two-thirds of the cfRNA-seq libraries did not pass the quality filtering owing to RNA degradation, DNA contamination, or blood coagulation. To increase the successful rate of sample processing in clinical applications, the method can be improved from the following aspects: (a) to avoid RNA degradation after the cfRNA extraction, better RNase-free treatment for the reagents and equipment can be implemented; (b) to avoid DNA contamination, the condition for DNA digestion in the cfRNA isolation step can be optimized; and (c) to avoid blood coagulation, the blood samples can be processed within 2 h of collection.
THE COMPOSITION OF BLOOD MICROBIOTA
Microbial nucleic acids in blood are normally detected using targeted methods such as RT-PCR and pan-viral or pan-microbial microarrays (18). To understand the systematic picture of microbial composition, unbiased deep sequencing is a possible solution. Recently, several efforts have been made toward the characterization of the microbiota in blood using 16S sequencing (19) or cfDNA sequencing (20). Although there are differences between the genomic and transcriptomic measurements in terms of microbial abundance, the composition of bacteria observed in our study is consistent with that observed in the previous studies: the top phyla are Proteobacteira, Firmicutes and Actinobacteria (21).
Compared to our understanding of the bacterial microbiome, our knowledge about human fungal microbiota ("mycobiome") lags behind. Our method also surveyed the mycobiome in plasma. In this cohort of samples, all of the nonhuman eukaryotic reads in the cfRNA libraries were aligned to fungi. Three fungi species were detected: Candida albicans, Encephalitozoon hellem, and Podospora anserina. Candida albicans is a common member of human gut flora (22). Encephalitozoon hellem can cause keratoconjunctivitis in immunecompromised patients (23).
Recent studies have investigated temporal variation of the human microbiome during pregnancy at multiple body sites (vagina, distal gut, saliva, and tooth gum) (11). Based on these studies, the microbiome at most body sites is stable during pregnancy, except for the vagina. However, the dynamics of plasma microbiome during pregnancy have not been explored before. From the results of our study, we conclude that the plasma microbiome, which integrates signals from many human body sites, is also relatively stable during pregnancy.
MONITORING INFECTION DURING PREGNANCY
In the current clinical practice, only the following pathogens are routinely tested for in pregnant women: Rubella, Hepatitis B, Hepatitis C, HIV, Tuberculosis syphilis, Chlamydia, and Group B Streptococcus. Other pathogenic species that can cause fetal defects are not routinely tested, such as human parvovirus B19 virus, herpes virus, adenovirus, and enteroviruses. Unbiased microbiome sequencing could potentially be used as a more comprehensive screening test. Shotgun sequencing of the cfDNA in blood has been used to monitor viral infection in organ transplant patients (6,20). However, this DNA-based method lacks the capability to detect RNA viral pathogens such as norovirus, hepatitis C virus (HCV), and zika virus. In principle, sequencing of cfRNA in plasma could detect both DNA and RNA viruses. Here, we demonstrate that asymptomatic viral infections could be detected using unbiased plasma cfRNA-seq. Our proof-of-principle study presents the potential use of plasma cfRNA-seq as a screening method for infectious diseases during pregnancy, especially for the asymptomatic and uncommon viruses.
In conclusion, we assessed the relative merits of different RNA-seq methods for cfRNA. We demonstrated that cfRNA-seq is able to monitor the pregnancy-related immune response and the blood-borne microbial infections in pregnant women.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contribution to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: R.J. Wong and D.K. Stevenson, March of Dimes.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or final approval of manuscript.
Acknowledgments: We thank Norma Neff, Jennifer Okamoto, and Gary Mantalas for assistance with sample preparation and sequencing. We also thank Marina Sirota, Daniel DiGiulio, and Brice Gaudilliere for helpful discussions.
(1.) Fan HC, Blumenfeld YJ, Chitkara U, Hudgins L, Quake SR. Noninvasive diagnosis of fetal aneuploidy by shotgun sequencing DNA from maternal blood. Proc Natl Acad Sci USA2008;105:16266-71.
(2.) Koh W, Pan W, Gawad C, Fan HC, Kerchner GA, Wyss-Coray T, et al. Noninvasive in vivo monitoring of tissue-specific global gene expression in humans. Proc Natl Acad Sci USA 2014;111:7361-6.
(3.) Robinson DP, Klein SL. Pregnancy and pregnancy-associated hormones alter immune responses and disease pathogenesis. Horm Behav 2012;62:263-71.
(4.) Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29:15-21.
(5.) Anders S, Pyl PT, HuberW. HTSeq-a Pythonframework to work with high-throughput sequencing data. Bioinformatics 2015;31:166-9.
(6.) DeVlaminck I, Martin L, Kertesz M, Patel K, Kowarsk yM, Strehl C, et al. Noninvasive monitoring of infection and rejection after lung transplantation. Proc Natl Acad Sci USA 2015;112:13336-41.
(7.) DeLuca DS, Levin JZ, Sivachenko A, Fennell T, Nazaire M-D, Williams C, et al. RNA-SeQC: RNA-seq metrics for quality control and process optimization. Bioinformatics 2012;28:1530-2.
(8.) Perucci LO, Carneiro FS, Ferreira CN, Sugimoto MA, Soriani FM, Martins GG, et al. Annexin A1 is increased in the plasma of preeclamptic women. PLoS One 2015; 10:e0138475.
(9.) Ryckman C, Vandal K, Rouleau P, Talbot M, Tessier PA. Proinflammatory activities of S100: proteins S100A8, S100A9, and S100A8/A9 induce neutrophil chemotaxis and adhesion.J Immunol 2003;170:3233-42.
(10.) Chaim W, Horowitz S, David JB, Ingel F, Evinson B, Mazor M. Ureaplasma urealyticum in the development of postpartum endometritis. Eur J Obstet Gynecol Reprod Biol 2003;109:145-8.
(11.) DiGiulio DB, Callahan BJ, McMurdie PJ, Costello EK, Lyell DJ, Robaczewska A, et al. Temporal and spatial variation of the human microbiota during pregnancy. Proc Natl Acad Sci 2015;112:11060-5.
(12.) Luo Y, Lou S, Deng X, Liu Z, Li Y, Kleiboeker S, et al. Parvo-virus B19 infection of human primary erythroid progenitor cells triggers ATR-Chk1 signaling, which promotes B19 virus replication. J Virol 2011;85:8046-55.
(13.) de Jong EP, de Haan TR, Kroes ACM, Beersma MFC, Oepkes D, Walther FJ. Parvovirus B19 infection in pregnancy. J Clin Virol 2006;36:1-7.
(14.) Pitkin RM. Platelet and leukocyte counts in pregnancy. JAMA 1979;242:2696-8.
(15.) Heung MMS, Jin S, Tsui NBY, Ding C, Leung TY, Lau TK, et al. Placenta-derived fetal specific mRNA is more readily detectable in maternal plasma than in whole blood. PLoS One 2009;4:e5858.
(16.) Mittal A, Pachter L, Nelson JL, Kjaergaard H, Smed MK, Gildengorin VL, et al. Pregnancy-induced changes in systemic gene expression among healthy women and women with rheumatoid arthritis. PLoS One 2015;10: e0145204.
(17.) Nair RR, Khanna A, Singh K. Role of inflammatory proteins S100A8 and S100A9 in pathophysiology of recurrent early pregnancy loss. Placenta 2013;34:824-7.
(18.) Leung RK-K, Wu Y-K. Circulating microbial RNA and health. Sci Rep 2015;5:16814.
(19.) Paisse S, Valle C, Servant F, Courtney M, Burcelin R, Amar J, Lelouvier B. Comprehensive description of blood microbiome from healthy donors assessed by 16S targeted metagenomic sequencing. Transfusion 2016;56:1138-47.
(20.) De Vlaminck I, Khush KK, Strehl C, Kohli B, Luikart H, Neff NF, et al. XTemporal response of the human virome to immunosuppression and antiviral therapy. Cell 2013;155:1178-87.
(21.) Potgieter M, Bester J, Kell DB, Pretorius E. The dormant blood microbiome in chronic, inflammatory diseases. FEMS Microbiol Rev 2015;39:567-91.
(22.) Kumamoto CA. Inflammation and gastrointestinal Candida colonization. Curr Opin Microbiol 2011;14: 386-91.
(23.) Didier ES, Didier PJ, Friedberg DN, Stenson SM, Orenstein JM, Vee RW, et al. Isolation and characterization of a new human microsporidian, encephalitozoon hellem (n.sp.), from three AIDS patients with keratoconjunctivitis. J Infect Dis 1991;163:617-21.
Wenying Pan,  Thuy T.M. Ngo,  Joan Camunas-Soler,  Chun-Xiao Song,  Mark Kowarsky,  Yair J. Blumenfeld,  Ronald J. Wong, [3, 4] Gary M. Shaw, [3, 4] David K. Stevenson, [3, 4] and Stephen R. Quake  *
 Departments of Bioengineering and Applied Physics, Stanford University, and Chan Zuckerberg Biohub, Stanford, CA;  Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Stanford University, Stanford, CA;  March of Dimes Prematurity Research Center, Stanford University School of Medicine, Stanford, CA;  Department of Pediatrics, Stanford University School of Medicine, Stanford, CA.
* Address correspondence to this author at: James H Clark Center E300, 318 Campus Drive, Stanford CA94305. Fax +1 650-7245473; e-mail email@example.com.
The sequencing data has been uploaded to SRA. (BioProject:PRJNA400333; SRA Study: SRP116273).
Received March 12, 2017; accepted August 4, 2017.
Previously published online at DOI: 10.1373/clinchem.2017.273888
 Nonstandard abbreviations: cfDNA, cell-free DNA; cfRNA, cell-free RNA; B19V, human parvovirus B19 virus; rRNA, ribosomal RNA.
 Human genes: ANXA1, annexin A1; S700AS, S100 calcium binding protein A8; S100A9, S100 calcium binding protein A9; S100P, S100 calcium binding protein P; BPI, bactericidal/permeability-increasing protein; PGLYRP1, peptidoglycan recognition protein 1; CSH1, chorionic somatomammotropin hormone 1; CSH2, chorionic somatomammotropin hormone 2; CGA, glycoprotein hormones, alpha polypeptide; PSG1, pregnancy specific beta-1-glycoprotein 1; PAPPA, pappalysin 1; ALAS2, 5'-aminolevulinate synthase 2; GYPA, glycophorin A (MNS blood group); ANK1, ankyrin 1.
Caption: Fig. 1. Comparison of QC metrics from different RNA-seq methods on cfRNA samples. rRNA rate (the fraction of reads mapped to rRNA among all the reads mapped to human genome) (A). Estimated library size (the number of expected fragments based on the total reads and duplication rate assuming a Poisson distribution) (B). Microbiome fraction (the fraction of reads aligned to microbial organisms among all the aligned reads) (C). The number of protein coding genes detected (D). The number of lincRNA genes detected (E). The number of snRNA genes detected (F). The number of miscRNA detected (G).
Caption: Fig. 2. Human genes displaying temporal changes during pregnancy in cfRNA. The genes with variations in abundance were grouped into 4 clusters based on their temporal patterns (A). The "tissue of origin" color bar annotates the origin of each gene based on tissue specificity. The "Gene function" color bar annotates the function of each gene. qPCR validation of a subset of the genes with variations in abundance (B).
Caption: Fig. 3. The plasma microbiome during pregnancy. The fraction of reads aligned to different superkindoms of microbiome (Ai). The distribution of bacterial reads at phylum level (Aii). The distribution of viral reads (Aiii). The composition of bacteria at four time points of pregnancy (1st, 2nd, 3rd and postpartum) (B). The composition of viruses at four different time points of pregnancy(1st, 2nd, 3rd and postpartum). The reads count represents the average value across all the samples at that time point (C). The abundance of Ureaplasma over the time course of pregnancy. The reads count represents the average value across all the samples at that time point (D). The prevalence of Ureaplasma at four different time points of pregnancy (E). All the value in Fig. 3A is the sample average of all the pregnant women's cfRNA samples passed sample quality filtering. The composition in Fig. 3B and C is the sample average after removing the patient (10039) with severe viral infection.
Caption: Fig. 4. Detection of pathogenic viruses in plasma. The prevalence of detectable pathogenic viruses among our pregnant women cohort at patient level (A). The abundance of Human parvovirus B19 virus in 1 patient (10039) during pregnancy from both cfRNA and cfDNA (B). The abundance of human erythrocyte- specific genes in cfRNA during pregnancy for the same patient with Human parvovirus B19 virus infection (C).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Molecular Diagnostics and Genetics|
|Author:||Pan, Wenying; Ngo, Thuy T.M.; Camunas-Soler, Joan; Song, Chun-Xiao; Kowarsky, Mark; Blumenfeld, Yair|
|Date:||Nov 1, 2017|
|Previous Article:||Potential for Monitoring Gut Microbiota for Diagnosing Infections and Graft-versus-Host Disease in Cancer and Stem Cell Transplant Patients.|
|Next Article:||Evaluation of Lipoprotein(a) Electrophoretic and Immunoassay Methods in Discriminating Risk of Calcific Aortic Valve Disease and Incident Coronary...|