A cross-sectional study was conducted by searching our clinical database and bank of biological samples, including patients in the following groups: (1) normal pregnancy (n=34) and (2) acute pyelonephritis (n=15). Patients with multiple gestations and fetal anomalies were excluded. The normal pregnant control group consisted of women who were not in labor, without obstetrical, medical, or surgical complications of pregnancy, and had blood samples collected within the same gestational age window as patients with pyelonephritis. Pyelonephritis was diagnosed in the presence of fever (temperature ≥38°C), clinical signs of an upper urinary tract infection (e.g., flank pain, costovertebral angle tenderness), pyuria, and a positive urine culture for microorganisms.

All patients provided written informed consent for the collection and use of samples for research purposes under the protocols approved by the Institutional Review Boards of Wayne State University and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Department of Health and Human Services.

Maternal peripheral venous blood was collected at the time of routine clinical blood draw into PAXgene blood RNA tubes (PreAnalytiX GmbH, distributed by Becton Dickinson Company, Franklin Lakes, NJ, USA). Blood tubes were maintained initially at room temperature for 24 h and then frozen at –70°C until further processing. Blood lysates were reduced to pellets by centrifugation, washed, and resuspended in buffer. Proteins were removed by Proteinase K digestion, and cellular debris were removed by centrifugation through a PAXgene Shredder spin column (PreAnalytiX GmbH). RNA was semi-precipitated with ethanol and selectively bound to the silica membrane of a PAXgene spin column (PreAnalytiX GmbH). The membrane was treated with DNase I to remove any residual DNA and washed, and the purified total RNA was eluted in nuclease-free water. Purified total RNA was quantified by UV spectrophotometry using a DropSense96 Microplate Spectrophotometer (Trinean, Micronic North America LLC, McMurray, PA, USA), and RNA purity was assessed based on the A260/A280 and A260/A230 ratios. An aliquot of the RNA was assessed using the RNA 6000 Nano Assay for the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The electrophoretogram, RNA integrity number, and the ratio of the 28S/18S RNA bands were examined to determine the overall quality of the RNA.

Peripheral blood samples were profiled using Affymetrix GeneChip HG-U133 Plus 2.0 arrays. Briefly, RNA was amplified using the Ovation RNA Amplification System V2 (NuGEN Technologies, San Carlos, CA, USA). cDNA was synthesized using the Ovation buffer mix, first-strand enzyme mix, and first-strand primer mix with 5 μL (∼20 ng) of total RNA in specified thermal cycler protocols according to the manufacturer’s instructions. Amplification and purification of the generated cDNA was performed by combining SPIA Buffer Mix, Enzyme Mix, and nuclease-free water with the products of the second-strand cDNA synthesis reactions in prespecified thermal cycler programs. The optical densities of the amplified cDNA products were obtained to demonstrate product yield and verified purity. Fragmentation and labeling were done using the FL-Ovation cDNA Biotin Module V2 (NuGEN Technologies). In the primary step, a combined chemical and enzymatic fragmentation process was used to produce cDNA products in the 50- to 100-base pair range. Fragmented cDNA products were then biotin-labeled using the Encore Biotin Module (NuGEN Technologies). All reactions were carried out according to the manufacturer’s protocols. Amplified, fragmented, and biotin-labeled cDNAs were used for hybridization cocktail assembly, and then hybridized to the Affymetrix GeneChip HG-U133 Plus 2.0 arrays according to the Affymetrix standard protocol.

A subset of differentially expressed genes (n=56) was selected for validation in an extended set of samples (pyelonephritis cases, n=19; controls, n=59) using the Biomark^™ high-throughput real-time quantitative reverse transcription-polymerase chain reaction (qRT-PCR) system (Fluidigm, San Francisco, CA, USA) based on their rank in the list of all differentially expressed genes as well as biological plausibility. Briefly, the Invitrogen Superscript III First-Strand Synthesis System (Invitrogen, Life Technologies, Carlsbad, CA, USA) was used to generate complementary DNA. Pre-amplification procedures included combining 1.25 μL cDNA with 2.5 μL TaqMan PreAMP Mastermix and 1.25 μL pooled assay mix. The reaction was performed with a thermal cycler for one cycle at 95°C for 10 min and 14 cycles at 95°C for 15 s and 60°C for 4 min. After cycling, the reaction was diluted 1:5 by ddH₂O to a final volume of 25 μL. The Fluidigm 96.96 Dynamic Array chip was used to perform the next step qRT-PCR assays. The 96.96 array chip was primed in an integrated fluidic circuit (IFC) controller with control fluid. After priming, 2.5-μL 20× TaqMan gene expression assays (Applied Biosystems) was mixed with a 2.5-μL 2× assay loading reagent (Fluidigm) and loaded into the assay inlet on the 96.96 array chip. A total of 2.25 μL preamplified cDNA was mixed with 2.5 μL TaqMan Universal PCR master mix (Applied Biosystems) and 0.25 μL 20× sample loading reagent (Fluidigm) and loaded into the sample inlet on the chip. The chip was returned to the IFC controller for loading. After loading, the chip was placed in the Biomark System to run the reactions. The cycle threshold (Ct) value of each reaction was obtained with the Fluidigm RT-PCR analysis software.

A total of 1309 probe sets corresponding to 983 unique genes demonstrated a differential expression between the two groups (q-value <0.1; fold change >1.5). A total of 457 genes were upregulated and 526 genes were downregulated in acute pyelonephritis. Table 2 shows the top 100 probe sets with differential expression between the study groups ranked by P-value.

A volcano plot (Figure 1A) displays the differential expression of all the annotated probe sets on the microarray as effect size vs. significance of expression change. An unsupervised principal component analysis (PCA)-based visualization of the microarray data (using all probes on the array, Figure 1B) revealed the between-group differences and no outlier samples.

Figure 1

Volcano plot and three dimensional principal component analysis (PCA) plot.

(A) The volcano plot shows probability values of all probes in the microarray plotted against the fold change. In this figure, the log (base 10) of the false discovery rate-adjusted probability values are plotted against the log (base 2) of fold change between patients with pyelonephritis and normal controls. On the Y-axis, values higher than the gray-line threshold represent significant probes with an adjusted probability value of <0.1. On the X-axis, values outside the red lines represent fold change >1.5. (B) The three-dimensional PCA plot demonstrates a degree of segregation between women with acute pyelonephritis and normal controls. Blue dots indicate individual samples from the normal control group, whereas black dots represent individual samples from the acute pyelonephritis group.

To gain further insight into the biology of the differences in the transcriptome of whole blood between pregnant women with acute pyelonephritis and controls, gene ontology analysis was used. Significant enrichment of 300 biological processes (Table 3) was found in acute pyelonephritis, including the innate immune response, signal transduction, regulation of cytokine production, regulation of adaptive immune response, immunoglobulin (Ig)-mediated immune response, T-cell immunity, B- and T-cell differentiation, positive regulation of leukocyte activation and proliferation, positive regulation of T-cell receptor signaling pathway, and blood coagulation. Moreover, gene ontology analysis revealed that 63 molecular functions were associated with differentially expressed genes in acute pyelonephritis (Table 4).

Pathway analysis of differentially expressed genes was undertaken with an overrepresentation method and the SPIA method. Using the overrepresentation method, three KEGG pathways were significantly impacted (q-value <0.1) in the comparison between the study groups (Table 5): (1) “primary immunodeficiency”, (2) “hematopoietic cell lineage”, and (3) “T-cell receptor signaling pathway”. SPIA identified four pathways that were significantly impacted (Table 6). Three of these pathways had not been identified by the overrepresentation method (the “Jak-STAT signaling pathway”, “cytokine-cytokine receptor interaction”, and “complement and coagulation cascade”) (Table 6, Figure 2).

Figure 2

Two-dimensional plot illustrates the relationship between the two types of evidence considered by SPIA.

The X-axis shows the overrepresentation evidence (–log [P-value]), whereas the Y-axis shows the perturbation evidence (–log [perturbation P-value]). Each pathway is represented by a point. Pathways above the oblique red line (red dots) are significant at 5% after Bonferroni correction; pathways (blue dots and red dots) above the oblique blue line are significant at 5% after false discovery rate correction. The vertical and horizontal thresholds represent the same corrections for the two types of evidence considered individually.

qRT-PCR was performed on an extended set of samples (normal controls, n=59; acute pyelonephritis, n=19) to validate microarray results. Of the 56 genes selected for testing from the differentially expressed gene list in the microarray, we confirmed differential expression (in terms of both direction and significance) in 48 (85.7%) of the tested genes by qRT-PCR (Table 7).

We report for the first time the transcriptome of maternal whole blood in acute pyelonephritis in pregnancy. The main findings are the following: (1) There was a gene expression signature consistent with a systemic maternal inflammatory response; (2) the transcriptome of peripheral WBCs in pyelonephritis was similar to that reported after intravenous endotoxin administration to nonpregnant individuals [14]. In both conditions, there was upregulation of genes involved in innate immune responses and downregulation of those involved in lymphocyte function. (3) We observed an upregulated expression of genes involved in the induction of apoptosis and downregulation of those with anti-apoptotic properties.

The innate limb of the immune response represents the first line of defense against bacterial invasion of the urinary tract [91, 119]. Urinary epithelial cells are the first to enter into contact with microorganisms. Bacterial attachment can trigger exfoliation of bacteria-laden epithelial cells, reconstitution of the urothelium, and an inflammatory response [47, 85–87, 143]. Microorganisms and their products are recognized by pattern recognition receptors, and this leads to the production of chemokines, cytokines, and antimicrobial peptides, and the generation of an acute inflammatory response [47, 85–87, 91, 119, 143]. Neutrophils play an important role in host defense in the urinary tract, and they appear in the bladder and kidney within hours of transurethral inoculation with uropathogenic E. coli [38, 48]. Disruption of neutrophil chemotaxis in animals with a gene deletion for the interleukin (IL) 8 receptor homologue [38] or depletion of neutrophils with a granulocyte-specific antibody [48] can lead to an increased bacterial burden in the bladder and kidney [38, 48] as well as bacteremia [38].

In addition to the local host defense in the urinary tract, acute pyelonephritis is also associated with a systemic inflammatory response that is characterized by fever, increased serum concentrations of cytokines (IL-6, IL-8) and acute phase reactant proteins, and an elevated WBC and neutrophil count [68]. Consistent with this, we found in the current study that the median WBC count of patients with pyelonephritis was higher than that of pregnant women with a normal pregnancy outcome (Table 1). We had previously reported that the maternal serum concentrations of a panel of chemokines and cytokines are also higher in this condition than in normal pregnant women [19]. Indeed, the median maternal serum concentrations of IL-8, TNF-α, IL-6, IL-7, IL-10, and interferon (IFN) γ were higher in pregnant women with acute pyelonephritis than in gestational age-matched normal pregnant women [19]. Moreover, acute pyelonephritis in pregnancy was found to be associated with higher median maternal serum concentrations of the pro-inflammatory chemokine CXCL-10 (also known as IP-10) [41] and higher median maternal plasma concentrations of the pro-inflammatory adipokine resistin [72] than in normal pregnant women. Resistin concentrations are considered an index of the severity of sepsis and a prognostic factor for survival in critically ill nonpregnant patients [55, 72, 122]. Of interest, median maternal plasma concentrations of the anti-inflammatory adipokine adiponectin [70] were lower in patients with pyelonephritis than in normal pregnant women. We have also reported the activation of the complement system, a component of the innate immune system, in acute pyelonephritis in pregnancy, as the median plasma concentrations of complement fragment Bb [118] and complement C5a [117] are higher in pregnant patients with acute pyelonephritis than in normal pregnant women. Complement C5a is a potent chemoattractant for neutrophils, and can upregulate the activating IgG Fc receptors and downregulate the inhibitory IgG Fc receptors on leukocytes, linking the complement system and IgG Fc receptor effector pathways [112]. This is consistent with the observations made in the present study, in which we observed an upregulation of FCGR1A and FCGR1B expression in pregnant women with acute pyelonephritis. These genes encode for the high-affinity IgG Fc receptor (CD64), which is expressed on neutrophils and other myeloid cells and is involved in the binding of IgG1 and IgG3 [94]. Consistent with this finding, Naccasha et al. [88] reported higher expression of CD64 on granulocyte and monocyte surfaces (determined by flow cytometry as median mean channel brightness) in pregnant women with pyelonephritis than in normal pregnant women. Neutrophil CD64 has emerged as a biomarker for the diagnosis of bacterial infection [22, 64, 94], because resting neutrophils express very low levels of CD64, whereas the expression of CD64 is upregulated in the context of acute bacterial infections [94]. Using flow cytometry analysis of whole blood, a CD64 index (ratio of mean fluorescent intensity of the cell to the beads) of >1.66 in hospitalized patients had a 100% sensitivity and a 95% specificity in the identification of sepsis, defined as the combination of bacteremia and clinical signs of infection [39]. Meta-analysis of 13 studies showed a pooled sensitivity of 79% [95% confidence interval (CI), 70%–86%)] and a pooled specificity of 91% (95% CI, 85%–95%) in the diagnosis of bacterial infection [22]. No studies have addressed the value of this marker in the assessment of pyelonephritis during pregnancy.

This is the first report of the transcriptome of whole-blood cells in pregnant women with pyelonephritis. This snapshot of the global mRNA expression of peripheral blood leukocytes and subsequent pathway analyses revealed interesting features of the systemic inflammatory response to bacterial infection in human pregnancy. We found 1309 probe sets corresponding to 983 unique genes differentially expressed in pregnant women with pyelonephritis, of which 457 genes were upregulated and 526 were downregulated. Gene ontology analysis indicated that these findings were associated with 63 molecular functions enriched in leukocytes, many of them strongly related to the innate and adaptive immune responses (e.g., “C-C chemokine receptor activity”, “complement receptor activity”, MHC class I protein binding”, “immunoglobulin receptor activity”, “T-cell receptor binding”, “chemokine binding”, “scavenger receptor activity”, “peptide antigen binding”, “cytokine receptor binding”) (Table 4). In accordance, several of the most enriched biological processes in pyelonephritis during pregnancy are related to immune responses (e.g., “positive regulation of leukocyte activation”, “innate immune response”, “activation of immune response” “regulation of lymphocyte activation”, “inflammatory response”). Of interest, biological processes related to apoptosis (i.e., “positive regulation of apoptosis”, “positive regulation of cell death”) are also among the most enriched processes (Table 3).

To identify pathways significantly impacted in pyelonephritis during pregnancy, we applied two pathway analysis methods. The overrepresentation analysis method identified three KEGG pathways significantly impacted in pyelonephritis (“primary immunodeficiency”, “hematopoietic cell lineage”, “T-cell receptor signaling pathway”) (Table 5). The SPIA method, which takes also into account the gene-gene signaling interactions as well as the magnitude and direction of gene expression changes besides differential expression [31, 128], identified one pathway in common with the overrepresentation method (“T-cell receptor signaling pathway”) and three pathways that the overrepresentation method could not identify (the “Jak-STAT signaling pathway”, “cytokine-cytokine receptor interaction”, and “complement and coagulation cascade”) (Table 6). Importantly, all of these impacted pathways are related to immune responses, suggesting that the systemic inflammatory response elicited in pyelonephritis in pregnancy has a characteristic gene expression signature in peripheral blood leukocytes.

To get further insight into the cellular pathways of systemic inflammation in pyelonephritis in pregnancy, we compared transcriptomic changes in our study to those documented in systemic inflammation in response to bacterial endotoxin in nonpregnant individuals (see below).

The transcriptome of peripheral blood leukocytes of nonpregnant volunteers has been studied after the administration of a single dose of bacterial endotoxin [14, 123]. Calvano et al. [14] reported dysregulation of 3714 genes in whole-blood cells at 2, 4, and 9 h after endotoxin administration and noted that gene expression returned to baseline by 24 h after endotoxin injection. The endotoxin quickly and transiently activated genes involved in the innate immune response, and after an initial pro-inflammatory phase, a self-limiting counter-regulatory response followed, with eventual resolution of gene expression changes within a day of endotoxin administration. Specifically, there was an increased expression of pro-inflammatory cytokines and chemokines (e.g., IL1A, IL1B, IL8, TNF) and NFκB family transcription factors within 2–4 h of endotoxin treatment [14]. There was upregulation of transcription factors critical in both the initiation and the containment of an innate immune response [e.g., signal transducer and activators of transcription (STAT) genes, suppressor of cytokine signaling 3, SOCS3], which was observed within 4–6 h of endotoxin administration. There was also increased expression of genes encoding membrane-bound and secreted proteins that limit the inflammatory response (e.g., IL1R2, IL1RAP, IL10) [14].

Similar observations were made by Talwar et al. [123], who investigated temporal gene expression changes in peripheral blood mononuclear cells and whole-blood cells from nonpregnant volunteers after a single dose of endotoxin. An upregulation of genes associated with pattern recognition receptors, intracellular signaling, cell mobility, and defense function was reported. The largest change in gene expression occurred 6 h after endotoxin treatment, with changes returning to baseline within 24 h [123]. Collectively, these results suggest that leukocyte response to bacterial products include a short pro-inflammatory phase followed by a counter-regulatory phase and resolution of inflammation [14, 123].

As the microarray data set of whole-blood leukocytes after endotoxin administration was available online from the study of Calvano et al. [14], we compared such data set with our findings (this comparison included only those genes from the study of Calvano et al. [14] that were differentially expressed at three to five time points after endotoxin administration). We found that 296 of the 983 genes in our study changed in the same direction as that of the Calvano et al. study [14] (Table 1).

Differentially expressed genes involved in the innate immune response in both studies were mainly upregulated (Table 8). Among the functions of the proteins encoded by these genes, the following groups emerged: (a) cell adhesion and cell-cell signaling (CD44, CLEC4D, CLEC4E, CLEC5A, ICAM1), (b) activation and/or differentiation of macrophages (CEBPD), (c) inflammasome priming (CASP1, CASP4, CASP5), (d) activation of the nuclear factor (NF) κB and mitogen-activated protein kinase (MAPK) pathways (IL18R1, IL18RAP, IRAK2, IRAK3), (e) cellular binding to particles and immune complexes that have activated complement (CR1), (f) phagocytosis and antibody-dependent cell-mediated cytotoxicity (FCAR, FCGR1A, FCGR1B, MARCO), and (g) breakdown of extracellular matrix and type IV and V collagens (MMP9). These results suggest that acute pyelonephritis during pregnancy elicits a host response similar to that induced by intravenous bacterial endotoxin in nonpregnant volunteers.

SPIA pathway analysis of the commonly differentially expressed genes in pyelonephritis and in the endotoxin-induced model of acute bacterial infection revealed five impacted pathways in both conditions: (1) “T-cell receptor signaling pathway”, (2) “natural killer cell mediated cytotoxicity”, (3) “cytokine-cytokine receptor interaction”, (4) “RIG-I-like receptor signaling pathway”, and (5) “complement and coagulation cascades”. We observed that the pathways inhibited in pyelonephritis included those generally implicated in adaptive immune responses. This is consistent with the findings of Talwar et al. [123], who described the downregulation of T lymphocyte-associated genes after endotoxin administration. Moreover, differentially expressed genes involved in lymphocyte functions common in our study and in the study reported by Calvano et al. [14] were mainly downregulated (Table 9).

The decreased expression of these genes and the encoded proteins may result in impairment of (a) T-cell recognition of antigens displayed by antigen presenting cells (CD3D, CD3E, CD8A, CD8B, CD247), (b) T-cell chemotaxis and migration to inflamed tissues (CXCR3), (c) T-helper lineage development (GATA3, STAT4, TBX21), (d) T-cell activation, proliferation, development, signal transduction, survival, and cytokine production (CCR7, CD6, CD28, DPP4, IL23A, IL23R, IL2RB, IL5RA, IL9R, ITK, LCK, PRKCQ, TCF7, ZAP70), (e) generation and long-term maintenance of T cell immunity (CD27), (f) T-cell-mediated cytotoxicity (GNLY, GZMA), and (g) regulation of B-cell activation, V(D)J recombination, and Ig synthesis (CCR7, IL7R).

These observations are consistent with the findings derived from transcriptome analysis of patients with sepsis [65, 125]. By analyzing multiple microarray data sets, Lindig et al. [65] found that the upregulated gene ontology categories in patients with sepsis include those related to innate immune responses, whereas the downregulated categories include those related to adaptive immune responses. The systematic review of the transcriptomic data of 12 studies by Tang et al. [125] revealed reduced expression of genes associated with immune response in lymphocytes (e.g., CCR7, CD28, CXCR3, IL2RB, IL7R) and upregulation of genes limiting inflammatory responses (e.g., SOCS1, SOCS3), which also changed in the same direction in our study.

Contrary to what was originally believed, sepsis does not represent a steady and uncontrolled systemic pro-inflammatory response [6, 51]. Some patients have a state consistent with immune suppression [51, 53, 140], which has been termed “immunoparalysis” [3, 53], characterized by a decreased Th1-like response [3, 51]. The initial pro-inflammatory state in sepsis is followed by a compensatory anti-inflammatory state, or in some cases, both pro-inflammatory and anti-inflammatory responses can occur at the onset of sepsis [51, 140]. The composition and direction of this complex systemic host response to infection depends on the load and virulence of the pathogen, the genetic characteristics of the host, and coexisting illnesses [6, 140].

Of importance, the majority of deaths occur in patients with sepsis who are immunosuppressed [53], and the prevention of immunosuppression improves the survival rate in animal models of sepsis [51]. The occurrence of the immunosuppression state has been attributed to the following[3, 11, 44, 51, 53, 98, 103, 140]: (1) an adaptive compensatory response characterized by increased expression of anti-inflammatory mediators (e.g., IL1RN, SOCS1, SOCS3) and the activation of T regulatory cells and myeloid derived suppressor cells, (2) apoptosis of T and B lymphocytes due to activation of Fas-Fas-ligand system and caspases as well as decreased expression and/or function of anti-apoptotic molecules (e.g., Bcl-2), (3) anti-inflammatory responses in phagocytic cells induced by the uptake of apoptotic immune cells, and (4) activation of a neuroinflammatory reflex through vagal nerve stimulation, which leads to acetylcholine secretion by a subset of T-helper lymphocytes and the subsequent suppression of proinflammatory cytokine release from acetylcholine receptor expressing macrophages.

Patients with sepsis may have a complex set of immunologic defects attributable to the cross-talk between specialized cells in the immune system that coordinate microbial eradication. For example, T lymphocytes play a pivotal role in the initial response to microbial infection because they produce IFN-γ, which activates macrophages [52, 53], and reduced Th1 function due to apoptotic cell death of T lymphocytes in sepsis leads to dampened cytokine production [52]. Indeed, the anti-inflammatory responses in sepsis lead to enhanced susceptibility to secondary infections [3, 6].

We found upregulation of IL1RN (IL-1 receptor antagonist) and suppressors of cytokine signaling (SOCS1, SOCS3) in women with pyelonephritis in pregnancy – these findings are consistent with a compensatory anti-inflammatory response. Moreover, we found upregulation of FAS (CD95/Fas cell surface death receptor), which plays a central role in the apoptosis of lymphocytes in sepsis and in the pathogenesis of ARDS [36], a severe complication of pyelonephritis in pregnancy [5, 16, 17, 23–27, 33, 45, 49, 56, 66, 97, 116, 136, 147].

Consistent with the observation of increased apoptosis of lymphocytes in sepsis [3, 44, 51, 53, 98], we found upregulation of pro-apoptotic genes (CFLAR, PDCD1LG2) as well as downregulation of anti-apoptotic genes (BCL2, FAIM3) in cases of acute pyelonephritis (Table 10). Moreover, the expression of CD36 (thrombospondin receptor), which is involved in the phagocytosis of apoptotic cells, was increased in pyelonephritis during pregnancy. We also found that the median absolute lymphocyte count of patients with pyelonephritis was significantly lower than that of in controls in both the microarray (P<0.05) and qRT-PCR (P=0.001) populations in the current report (Table 1).

In contrast, the median absolute neutrophil count was higher in patients with pyelonephritis than in controls (P<0.001, Table 1). This finding is consistent with the well-known phenomenon of delayed apoptosis of neutrophils in sepsis [8, 96, 124]. Previous studies also reported a higher median maternal plasma concentration of visfatin [71], a pro-inflammatory adipokine that promotes delayed neutrophil apoptosis as well as a lower median maternal plasma concentration of TRAIL [18], one of the mediators responsible for neutrophil apoptosis, in patients with pyelonephritis in pregnancy than in controls.

There were 22 differentially expressed genes in our study that changed in the opposite direction from that reported by Calvano et al. [14]. In addition, there were 665 differentially expressed genes in our study that did not change in expression after bacterial endotoxin administration [14]. Enrichment analyses showed that these 665 genes play a role in “immune response”, “immune effector process”, “inflammatory response”, “lymphocyte activation”, and “response to viruses”, among other biological processes. These results suggest that, although the pathways fundamentally impacted in pyelonephritis during pregnancy and endotoxin challenge in nonpregnant women are similar, the increased susceptibility of pregnant women to microbial products [46, 81, 84, 101] may have a well-defined molecular basis, and such differences (qualitative and quantitative) may be responsible for the difference in nature of the immune response to microbial products or infection in pregnant women. This requires further study of nonpregnant patients with pyelonephritis.

The administration of a single intravenous dose of endotoxin has been used as a model to study the systemic effects of a microbial product (endotoxin) and assumed by many to represent a state similar to systemic infection. However, this experimental approach cannot be equated with intravenous administration of live bacteria or actual systemic infection in humans. Therefore, the comparison of our results in pregnant women with pyelonephritis to those reported by Calvano et al. [14] should be interpreted with caution. The findings reported herein can be considered to reflect real bacterial infection in pregnant women.

There are technical differences between our study and that of Calvano et al. [14]. (1) methods used in blood collection, leukocyte lysis, storage, and RNA isolation were not the same. Whole-blood samples were collected into PAX gene tubes in our study, whereas leukocyte separation by centrifugation was undertaken by Calvano et al. before RNA isolation. (2) Our study was cross-sectional, whereas blood was collected at several time points [before (0 h) and at 2, 4, 6, 9, and 24 h after endotoxin infusion] in the study of Calvano et al. [14]. (3) Our population had several bacteria identified in urine and/or blood, whereas nonpregnant volunteers in the study of Calvano et al. received a single dose of endotoxin [14]. (4) We adjusted our gene expression results as a function of WBC count, but this was not performed by Calvano et al. [14].

The top three significantly upregulated probe sets in our study correspond to transcripts encoded by the FAM20A (family with sequence similarity 20, member A) gene. FAM20A belongs to an evolutionarily conserved family of three proteins (FAM20A, FAM20B, and FAM20C) secreted by hematopoietic cells [89]. FAM20A is a glycoprotein with high expression levels in hematopoietic tissues [89], especially in those cells committed to the granulocytic lineage [89]. Another gene involved in hematopoiesis and upregulated in pyelonephritis was ETV7 (encoding for ETS variant 7). The expression of ETV7 in normal and leukemic hematopoietic cells suggests an important role for this gene in normal hematopoietic development as well as in oncogenesis [15, 59]. As gene expression in our data was adjusted for WBC count, the results are not simply the reflection of a higher WBC count in patients with pyelonephritis but probably reflect enhanced hematopoiesis that involves the upregulation of these genes in response to acute microbial infection.

This is the first study to characterize the transcriptome of whole blood in pregnant women with an acute episode of infection. A limitation of this study was that the differentially expressed genes identified reflect the changes in the total intracellular mRNA in whole blood. However, it is not possible to attribute these changes to a particular population of leukocytes or reticulocytes. Meanwhile, if we had attempted to separate WBCs before RNA isolation, artifacts derived from cell separation procedures may have been introduced.