Molecular Mechanism Underlying Persistent Induction of LCN2 by Lipopolysaccharide in Kidney Fibroblasts

The neutrophil gelatinase-associated lipocalin 2 (LCN2) is a critical inflammatory mediator persistently induced during endotoxemia, contributing to tubular damage and kidney failure. The intracellular process responsible for persistent induction of LCN2 by bacterial endotoxin Lipopolysaccharide (LPS) is not well understood. Using primary kidney fibroblasts, we observed that LPS-induced LCN2 expression requires a coupled circuit involving an early transient phase of AP-1 path and a late persistent phase of C/EBPδ path, both of which are dependent upon the interleukin 1 receptor associated kinase 1 (IRAK-1). Using immunoprecipitation analysis we observed transient binding of AP-1 to the promoters of both TNFα and C/ebpδ. On the other hand, we only observed persistent binding of C/EBPδ to its own promoter but not on TNFα. Blockage of new protein synthesis using cyclohexamide significantly reduced the expression of C/EBPδ as well as LCN2. By chromatin immunoprecipitation analyses, we demonstrated that LPS recruited C/EBPδ to the Lcn2 promoter in WT, but not IRAK-1 deficient fibroblasts. A differential equation-based computational model captured the dynamic circuit leading to the persistent induction of LCN2. In vivo, we observed elevated levels of LCN2 in kidneys harvested from LPS-injected WT mice as compared to IRAK-1 deficient mice. Taken together, this study has identified an integrated intracellular network involved in the persistent induction of LCN2 by LPS.


Introduction
Initially identified in neutrophils as a gelatinase-associated small protein [1], LCN2 was recognized as an innate defense molecule by sequestering and depleting iron-containing siderophores and curbing bacterial growth [2]. Recent studies reveal that LCN2 can be potently induced by inflammatory stimulants and widely expressed in vital organs and tissues such as kidney, heart, and brain [3,4,5,6]. Elevated level of LCN2 in kidney is a well-recognized marker for both chronic kidney diseases, reflecting the extent of kidney damage [7,8]. LCN2 is not only a marker, but also a key contributor of kidney disease, as reflected by the alleviation of chronic kidney damage in mice with Lcn2 gene deletion [6].
Bacterial endotoxin (lipopolysaccharide-LPS) is a potent inducer of LCN2 [9,10,11]. Elevated levels of endotoxin are seen in both acute and chronic conditions. Acute endotoxemia leads to septic shock and multi-organ damages including acute kidney failure [12]. In contrast, chronic endotoxemia associated with obesity, aging, and other adverse health conditions can lead to persistent inflammatory complications and chronic diseases [11,13,14]. LCN2 levels persist during both acute and chronic phases of kidney damage, and may serve as a key propagating factor for chronic kidney disease [6]. However, the molecular mechanism underlying the persistent expression of LCN2 is not well understood.
Studies using other inflammatory stimulants such as IL-17 and IL-1b indicate that the induction of LCN2 requires multiple transcription factors including AP-1 and C/EBPd [9,15]. AP-1 activation by LPS is transient and responsible for the transient induction of pro-inflammatory mediators [16]. On a separate report, C/EBPd was implicated in propagating the persistent activation of TLR4 pathway [17].
Based on these observations, we examined the molecular circuit underlying the persistent induction of LCN2 by LPS in kidney fibroblasts. We identified that IRAK-1, a key TLR4 intracellular signaling component, is involved in coordinating both the transient activation of AP-1, as well as the persistent activation of C/EBPd, which are collectively responsible for the sustained expression of LCN2. We further performed a differential equation-based computational modeling, which complemented our experimental data and revealed a dynamic intracellular signaling circuit responsible for the persistent induction of LCN2 by LPS.

Materials and Methods
Reagents LPS (Escherichia coli 0111:B4) and Cycloheximide were obtained from Sigma Aldrich. Anti-LCN2 antibody was purchased from R&D Systems. Anti-IkBa antibody was purchased from Cell Signaling. Anti-Lamin B (ab-16048) was purchased from Abcam.

Western Blot Analysis
Isolation of whole cell, cytoplasmic, and nuclear lysates were performed as described previously [19]. Briefly, fibroblasts were lysed on ice using lysis buffer (10 mM HEPES, pH 7.9, 1.5 mM MgCl2, 10% Triton X-100, 10 mM KCl, 0.5 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 ug/mL leupeptin, 1 ug/mL pepstatin) for 30 minutes. Samples were centrifuged at 5000 rpm for 10 minutes at 4uC. The supernatants were saved as the cytoplasmic extract and the pelleted nuclei were re-suspend and lysed in a high salt buffer (20 mM HEPES, pH 7.9, 1.5 mM MgCl2, 0.4 M NaCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) on ice for 30 minutes. The samples were then centrifuged at 12,000 RPM for 20 minutes at 4uC and supernatant kept as the nuclear extract. Protein samples were analyzed by Western Blot as described before. Images were quantified with Fujifilm Multi Gauge software normalized against b actin, GAPDH, or Lamin B.

Chromatin Immunoprecipitation (ChIP) assays
Protein and chromatin from treated cells were cross-linked using 1% formaldehyde solution in complete media for 15 minutes with gentle rocking every 3 minutes at room temperature. Cells were washed twice with cold PBS and treated with a glycine solution for 5 minutes to stop the cross-linking reaction. The cells were then lysed in an ice cold buffer containing SDS and protease inhibitor cocktail. The samples were then subject to sonication in an ice water bath to shear the chromatin for 6 minutes (30 seconds on and 30 seconds off). The sheared chromatin was processed using Chip-IT Express Kit #53008 from Active Motif TM . The immuneprecipitated chromatin was analyzed by PCR using primer pairs that span the promoter regions of either Lcn2, C/ebpd, and Tnfa.

In vivo LPS injection
Female IRAK-1 2/2 and WT C57/BL/6 mice of 12 weeks old received intraperitoneal injections with 25 mg/kg of LPS. Six hours after the injection, mice were sacrificed and the whole kidney tissue was harvested for subsequent mRNA analysis by real time RT-PCR and protein analyses by Western blot.

Computational analysis
A differential-equation based computational modeling is performed as previously described [20]. A detailed description of the modeling approach is provided in the supplementary document accompanying this manuscript.

Statistical Analysis
Data significance was determined using the student t-test. Pvalues less than 0.05 were considered statistically significant as indicated by an asterisk (*).

LPS induces persistent expression of LCN2 in kidney fibroblasts in an IRAK-1 dependent fashion
We first compared the expression profiles of selected inflammatory genes in fibroblasts stimulated with LPS. As shown in Figure 1A, the induction of Tnfa message by LPS was transient, peaked at the 2 hour time point post challenge, and returned to basal values at the 16 hour time point. In contrast, the expression of Lcn2 mRNA persisted throughout the course of LPS treatment (Fig. 1B). Because IRAK-1 is one of the key downstream components of TLR4 pathway, we evaluated whether IRAK-1 is involved in both the transient induction of Tnfa and persistent induction of Lcn2 message. As shown in Figure 1A and 1B, the inductions of both Tnfa and Lcn2 were significantly reduced in IRAK-1 deficient kidney fibroblasts. We also confirmed the induction of LCN2 protein in wild type (WT) cells by LPS (Fig. 1C).

LPS induces transient activation of AP-1 and persistent induction of C/EBPd in kidney fibroblasts
To determine the potential mechanism for the persistent induction of LCN2 in kidney fibroblasts by LPS, we examined the activation status of relevant transcription factors including NFkB, AP-1 and C/EBPd. As shown in Figure 2A, LPS treatment caused no apparent decrease in the levels of Ikba in WT fibroblasts, indicating a lack of robust NFkB activation by LPS in kidney fibroblasts. Further supporting this claim, we observed that the residue nuclear p65/RelA remained constant following LPS treatment (Fig. 2B). In contrast, we observed a rapid yet transient induction of AP-1/c-Jun. Nuclear levels of AP-1/c-Jun was induced by LPS, reached its peak level by 60 to 120 minutes, and decreased at 240 minutes after LPS challenge (Fig. 2C). C/ EBPd, on the other hand, was elevated after 24 hr of LPS  Fig. 2D and 2E). Given the fact that Western blot may not give rise to quantitative result, we performed RT-PCR analysis of C/ebpd expression induced by LPS. As shown in Figure 2F, LPS significantly induced the expression of C/ebpd, and such induction is dependent on IRAK-1.
Based on above observation and previous studies, we propose that a coupled step-wise process may lead to the persistent activation and induction of C/EBPd. LPS may first induce an initial C/EPBd expression primarily through a transient activation of AP-1 and residue NFkB in the nucleus [21]. Subsequently, newly synthesized C/EBPd can auto-induce its own expression and contribute to a steady induction of C/EBPd [17] and its downstream target gene LCN2. To test this hypothesis, we performed Chromatin immunoprecipitation (ChIP) analysis to determine the recruitment of c-Jun and C/EBPd to the promoter of C/ebpd. As shown in Figure 3, LPS transiently increased the recruitment of c-Jun to the promoter of C/ebpd ( Figure 3A) and Tnfa ( Figure 3B), peaking at 4 hr for C/ebpd and 1 hr for Tnfa promoter, and dropping off at the 8 hr time point. In contrast, increased recruitment of C/EBPd at the 8 hr time point was evident on the promoter of C/ebpd, but not Tnfa.
Both De novo synthesis and subsequent activation of C/ EBPd is required for the expression of LCN2 To further confirm that de novo synthesis of C/EBPd is involved in LPS-induced LCN2 expression, we applied cycloheximide (CHX), a known protein synthesis inhibitor, to the culture medium. As shown in Figure 4A, CHX significantly reduced LPS-mediated induction of Lcn2 message, suggesting that new protein synthesis is required for the induction of LCN2 gene expression. In contrast, CHX has no inhibitory effect on LPSmediated expression of C/ebpd message ( Figure 4B). This is in agreement with a previous finding that the expression of C/EBPd and its subsequent transcriptional activation are two separate events controlled by two distinct upstream processes [21]. To further confirm the role of C/EBPd in LPS induced expression of Lcn2, we examined the recruitment of C/EBPd to the promoter of Lcn2 following LPS challenge. As shown in Figure 4C, LPS dramatically prolonged the elevated recruitment of C/EBPd to the Lcn2 promoter in WT cells.
The lack of inducible negative regulators contributes to prolonged expression of LCN2 in kidney fibroblasts upon LPS stimulation Through the classical NFkB pathway, LPS is known to induce negative regulators that subsequently turn off the expression of pro-inflammatory genes. Among the negative regulators, SOCS1 suppresses the upstream LPS signaling pathway by inhibiting IRAK-1 [22]. On the other hand, ATF3 is shown to compete and suppress the activity of C/EBPd [17,23]. Given the fact that we did not observe noticeable activation of the NFkB pathway, we hypothesize that LPS may not be able to induce related negative regulators in kidney fibroblasts. Indeed, we demonstrated that LPS failed to induce the expression of Socs 1 and Atf 3 (Fig. 5A). As a consequence, we observed that there was no apparent degradation of IRAK-1 protein in kidney fibroblasts following a 4 hr LPS treatment (Fig. 5B). Taken together, our data indicate that LPS treatment in kidney fibroblasts failed to induce negative regulators that modulate the persistent activation of C/EBPd, leading to a sustained induction of its downstream gene Lcn2.

In vivo induction of LCN2 expression in an IRAK-1 dependent fashion
To test the in vivo relevance of our finding, we examined the expression of LCN2 in kidney tissues from WT and IRAK-1 deficient mice following LPS challenge. As shown in Figure 6, LPS injection induced significant induction of Lcn2 messages as well as LCN2 proteins in kidney tissues harvested from LPS-injected WT mice. In contrast, the levels of Lcn2 messages were significantly lower in kidneys harvested from IRAK-1 deficient mice injected with LPS.

Computational analyses of the molecular circuit responsible for the persistent induction of LCN2 in kidney fibroblasts
We further fit our experimental data into a differential equation based computation model (Fig. 7). This model takes into consideration of the positive feedback loop involving C/EBPd that helps to maintain a sustained expression of LCN2.
This model is based on a coarse-grained TLR4-signaling network that highlights the core pathways controlling the induction of LCN2. As is shown in Figure 7A, this signaling circuit is composed of both an early-phase module with transient response and a late-phase module with persistent response. Each module can be viewed as an information processor that receives inputs from upstream signals, and generate outputs that regulate downstream response. The two modules are connected. The transcription of C/EBPd is dependent on AP-1/c-Jun, and serves as the core of the late-phase module.  Table 1 lists the equations in the computational model. Each ODE describes the change of a molecular species in the relative level as a function of time. Parameters such as mRNA degradation rates are estimated based on our data or from existing literatures. Other parameters such as activating threshold k and relative weights K are estimated using genetic algorithm of least square regression (MATLABH 2009b, the MathWorks, Natick, MI). All parameters are listed in Table 2.
As is shown in Figure 7B, our computational model based on the proposed signaling circuitry fits well with the experimental data. The model reveals that the essential requirement of the biphasic dynamics of C/EBPd and LCN2 is the proper timing of the two consecutive modules. The immediate yet transient induction in c-Jun explains the first pulse in C/EBPd, when the early-phase module dominates the dynamics of the system. As soon as c-Jun decays, the late-phase module relays the signal and an auto-regulation of C/EBPd gives rise to the persistent induction of both C/EBPd and LCN2. IKKe plays an important role in ensuring the correct timing of the second module. Experimentally, it was shown that IKKe activates the activity of C/EBPd, therefore the auto-regulation of C/EBPd is IKKe-dependent (21).
Taken together, our computational simulation based on the intracellular signaling circuitry identified in this report fitted well with the experimental data, capturing the dynamic intracellular circuits that give rise to transient induction of TNFa and persistent induction of LCN2.

Discussion
In this report, we have identified that LPS can induce a persistent expression of LCN2 in kidney fibroblasts. The intracellular signaling circuit responsible for the persistent induction of LCN2 involves dynamic interaction among AP-1 and C/EBPd. An early phase of a transient AP-1 activation helps to usher in a late phase of C/EBPd activation. IRAK-1 is responsible for the propagation of both phases necessary for the persistent induction of LCN2.
Our study shed light on the dynamic and functional circuitry within cells when challenged with bacterial endotoxin. Although empirical knowledge reckons that integrated intracellular circuits instead of linear cascades are more capable of interpreting and responding properly to environmental cues including LPS, almost all existing studies have adopted conventional approaches in defining the signaling cascades downstream of LPS and TLR4 receptors [24]. To this regard, a large number of signaling molecules and pathways have been discovered that can be activated by LPS [25]. However, how these signals are properly integrated and give rise to distinct cellular responses is not clear. Wild-type kidney fibroblasts were untreated (DMSO) or pretreated with cyclohexamide 5 ug/mL for 1 hour. Immediately following pretreatment, the cells were stimulated with or without LPS for 4 hours. Transcripts were quantitated using qRT-PCR and standardized using Gapdh as the internal loading control. Each experiment was performed in triplicate. *P,0.05 (C) The recruitment of C/EBPd to the promoter of LCN2 upon LPS stimulation depends upon IRAK-1. Nuclear lysates were subject to ChIP analysis using a C/EBPd specific antibody. The arrow points to the specific amplification signal for C/ebpd promoter. The * sign indicates a non-specific amplification. doi:10.1371/journal.pone.0034633.g004 This work builds upon conventional information uncovered in the area of LPS signaling within host cells, and reveals an integrated circuit that couples a transient activation of AP-1 with a sustained activation of C/EBPd.
Our effort is made possible by the integrated approach combining experimental tests and computational analyses. Taken into consideration of the experimental data, we have employed a differential equation-based approach to analyze time-dependent activation of AP-1 and C/EBPd, as well as expression of Tnfa and Lcn2. As a consequence, genes primarily under the control of AP-1 such as Tnfa experience a transient pattern of expression. The expression of C/ebpd, on the other hand, is initially activated by AP-1, and subsequently sustained by its own auto-activation [17,21]. Due to the transient nature of AP-1 activation, we observed a transient dip in the expression of C/ebpd (around the 4 hr time point). Subsequent self-activation of C/EBPd then sustains the continued expression of C/ebpd and its downstream gene Lcn2. A detailed computational analysis can be seen in the supplementary material.
It is likely that there will be significant perturbations as well as additional connections surrounding this circuit in different cell types. Consequently, this may allow distinct outcomes in different cell types in response to bacterial endotoxin. For example, the expression of C/EBPd in macrophages induced by LPS eventually subsides, due to a novel negative feedback initiated by ATF3 [17]. In contrast, we observed that LPS failed to induce Atf3 in kidney fibroblasts, which may explain the sustained expression and activation of C/EBPd. This may due to the fact that fibroblasts are less sensitive to the LPS challenge as compared to macrophages. Indeed, our recent study indicates that a much lower LPS dose (,50 pg/ml) induces a persistent activation of C/EBPd in macrophages [26], similar to the effect seen in this study using fibroblasts with a high dose LPS. Further supporting this claim, we observed that IRAK-1 is relatively stable in kidney fibroblasts treated with 100 ng/ml LPS. In contrast, it was observed that 100 ng/ml LPS causes IRAK-1 degradation in macrophages, which can dampen the signals leading to C/EBPd activation, leading to LPS tolerance [27]. Further studies employing systems biology approach are clearly warranted to clarify the differential sensing mechanisms of LPS signal strength in different cell types. On a separate note, the integrated systems approach will likely explain the complex phenotypes such as LPS priming and tolerance exhibited in cells treated with either a low or a high dose LPS.
Our findings bear significant clinical implication during the pathogenesis of chronic and persistent inflammatory diseases. Local tissues and fibroblasts encountering bacterial endotoxin may Wild-type kidney fibroblasts were either untreated or treated with 100 ng/mL LPS for either 2 or 4 hours. Whole cell lysates were harvested and analyzed by Western blot with IRAK-1 specific antibodies. doi:10.1371/journal.pone.0034633.g005 Figure 6. Loss of IRAK-1 causes decreased expression of Lcn2 mRNA and protein in vivo. WT and IRAK-1 2/2 C57/BL/6 female mice of 12 weeks old (6 each) were intraperitoneally injected with either 30 mg/kg of LPS or PBS for 6 hours. (A) Kidney tissues were extracted and subject to RNA extraction. Lcn2 transcripts were measured by qRT-PCR assays and standardized against their respective controls (mice receiving PBS injections). Data is depicted as three separate mice (both WT and IRAK-1 2/2 ). *P,0.05 (B) Protein lysates were extracted from the kidney tissues and subjected to Western blot. Blots were analyzed using LCN2 specific antibodies. The same blot was probed with b-actin as a loading control. doi:10.1371/journal.pone.0034633.g006 exhibit persistent activation of the C/EBPd circuit without a dampening mechanism. As a consequence, inflammatory mediators such as LCN2 may accumulate and lead to chronic tissue damages including kidney failure. Strategies targeting at this functional circuit may hold promise in designing therapeutic strategies to combat chronic inflammatory diseases.

Author Contributions
Performed the experiments: TG YF. Analyzed the data: TG LL. Contributed reagents/materials/analysis tools: YF JX. Wrote the paper: TG YF LL.  Table 1. Ordinary Differential Equations used in the model.

Variable
Name Differential Equations  Parameters such as mRNA degradation rates are estimated based on our data or from existing literatures. Values of other parameters such as activation/ inhibition threshold k and relative weights K are estimated to fit in the time course data by genetic algorithm of least square regression. Degradation rates are in the unit of/min. Activation/inhibition thresholds, nonlinearity coefficients and relative weights for multiple inputs are non-dimensionalized. doi:10.1371/journal.pone.0034633.t002