RNA-Seq and protein mass spectrometry in microdissected kidney tubules reveal signaling processes initiating lithium-induced nephrogenic diabetes insipidus

Abstract

Lithium salts, used for treating bipolar disorder, frequently induce nephrogenic diabetes insipidus (NDI) thereby limiting therapeutic success. NDI is associated with loss of expression of the gene coding for the molecular water channel, aquaporin-2, in the renal collecting duct (CD). Here, we use systems biology methods in a well-established rat model of lithium-induced NDI to identify signaling pathways activated at the onset of polyuria. Using single-tubule RNA-Seq, full transcriptomes were determined in microdissected cortical collecting ducts (CCDs) of rats after 72 hours without or with initiation of lithium chloride administration. Transcriptome-wide changes in mRNA abundances were mapped to gene sets associated with curated canonical signaling pathways, showing evidence for activation of NF-κB signaling with induction of genes coding for multiple chemokines and most components of the Major Histocompatibility Complex Class I antigen-presenting complex. Administration of anti-inflammatory doses of dexamethasone to lithium chloride-treated rats countered the loss of aquaporin-2. RNA-Seq also confirmed prior evidence of a shift from quiescence into the cell cycle with arrest. Time course studies demonstrated an early (12 hour) increase in multiple immediate early response genes including several transcription factors. Protein mass spectrometry in microdissected CCDs provided corroborative evidence and identified decreased abundance of several anti-oxidant proteins. Thus, in the context of prior observations, our study can be best explained by a model in which lithium increases ERK activation leading to induction of NF-κB signaling and an inflammatory-like response that represses Aqp2 transcription.

Keywords: NF-κB signaling; collecting duct; immediate early response.

Figure 1.. LiCl-treated rats display urine concentrating defect.

(A) The experimental design involved treating rats with 40 mmol/kg LiCl or 40 mmol/kg NaCl (Control) and collecting data in subsequent days. (B) Immunoblots for AQP2 in the whole kidney at 72- and 96-hr time points (n=3 per time point for Control and LiCl). For all immunoblots, loading was 20 μg/lane of total protein. Gly-AQP2, glycosylated aquaporin-2; non-Gly-AQP2, nonglycosylated AQP2. (C) Quantification of A. (D and E) Daily water intake and urine osmolality at 0, 24, 48, and 72 hrs (n=4 per time point for Control and LiCl). Error bars indicate standard error. ** P< 0.05, lithium versus control, t-test. Sprague-Dawley rats (S-D rats). Cortical collecting ducts (CCD). Cortical thick ascending limb (cTAL).

Figure 2.. Cell counting reveals minimal remodeling of cortical collecting ducts within 72 hrs of initiation of LiCl administration.

Cortical collecting ducts from control and LiCl-treated rats were dissected and immunolabeled for cell counting. (A) Samples are shown in 2D confocal, 3D projection, and cell counting views. In the 2D confocal and 3D projection image, blue shows DAPI labeling, red shows H⁺-ATPase labeling, and green shows AQP2 labeling. In the cell count image, pink dots are identified as intercalated cells, green dots are identified as principal cells, and yellow dots are hybrid cells. The scale bar represents 20 μm. Total cells per mm (B), principal cells per mm (C), and intercalated cells per mm (D) are shown for the three control and three LiCl-treated rats (Samples per rat = 5). Paired t-tests were used to evaluate differences in these values between control and LiCl-treated rates. Data for individual tubules in a given rat were not counted as separate replicates for statistical purposes but were averaged to get a single value for statistical comparisons. No significant differences were found for total cells per mm and intercalated cells per mm; however, there was a significant increase in the principal cells per mm (P = 0.006). See also Supplementary Figure 2.

Figure 3.. Mapped reads of Sgk1, Aqp2, Aqp3, and Fos, and volcano plots show log₂(TPM-LiCl/TPM-control) versus negative of log₁₀P value for paired t-tests for transcripts expressed in dissected cortical collecting duct (CCD, B and C) and cortical thick ascending limbs of Henle (cTAL, D and E).

(A)Visualization of mapped read distributions along gene bodies of 4 genes for a single pair of samples (lithium, blue; control, green). Vertical axis shows read counts. Map of exon/intron organization of each gene was shown on top of individual subpanel (exons were shown as red rectangles and introns as red lines connecting the exons). (B) Control versus control for microdissected CCDs, (C) LiCl versus control for microdissected CCDs, (D) Control versus control for microdissected cTALs, and (E) Lithium versus control for microdissected cTALs. Red points exceed thresholds for significance, P<0.05 and log₂(∣ LiCl/Control ∣) exceeds 99% confidence interval for control versus control distributions. 99% confidence interval 0.3689 in CCD, 0.5308 in cTAL are represented as vertical solid lines (see Methods).

Figure 4.. Time-courses of transcript abundances show group-specific patterns with immediate early response and a proliferative response.

(A) Transcription factors corresponding to ‘Immediate Early Genes’ were increased at earliest time point examined (12 hrs). (B-E) This pattern contrasts with that seen for gene groups associated with a proliferative response including “Cell Cycle Kinases”, “MCM Complex” and subunits of the “DNA Polymerase Complex” and “Cyclin Inhibitors”, which showed relatively little change early, but a progressive increase from 24 to 72 hrs. (F) The gene group “RNA Polymerase Complex” is included as a negative control.

Figure 5.. Time-courses of transcript abundances show group-specific patterns for “Inflammatory Signaling Transcription Factors”, “Transporters and Channels”, “ G-protein coupled receptors”, “Chemokines”, and “MHC Class.

(A) There was a steady increase in two transcription factors involved in inflammatory signaling, namely Nfkb2 and Relb. At 36 hrs and beyond, there were maintained decreases in transcripts that code for (B) “Transporters and Channels” and (C) “GPCRs”. These proteins play central functional roles in the collecting duct. (D) Many chemokines transcript such as Ccl2, Ccl20, Ccl5, Cxcl10, and Cxcl11, were greatly increased in both early and later timepoints. (E)Transcripts for MHC Class I showed more variable time course prior to moderately increased at 72 hrs.

Figure 6.. RT-qPCR analysis.

SYBR green fluorescence curves (left panels) and C_T values (right panels) for indicated target transcripts were shown. Control samples are shown in green curves and lithium samples are shown in orange curves. Data were collected from three pairs of control/lithium rats. (A) Atp1a1. (B) Ccl2. (C) Ccl20. (D) Cxcl10. (E) B2m.

Figure 7.. Administration of anti-inflammatory dose of the glucocorticoid dexamethasone reverses the loss of AQP2 protein caused by 72-hr LiCl administration.

(A) A diagram summarizes the experimental design. (B) Confocal immunofluorescence images from 3 different sets of rats including Control (NaCl), LiCl, and LiCl+Dexa group. Dexamethasone (Dexa, 3mg/kg, intraperitoneal) was administered 24 hrs before and 0, 24, 48 hrs after starting LiCl in LiCl+Dexa group. (C, D) Immunoblotting and quantification for AQP2 in the whole kidney (n=4 per group) at 72-hr time point for LiCl and LiCl+Dexa. (E, F) Immunoblotting and quantification for AQP2 in the cortex (n=4 per group) at 72-hr time point for LiCl and LiCl+Dexa. Gly-AQP2, glycosylated aquaporin-2; non-Gly-AQP2, nonglycosylated aquaporin-2. Scale bar is 20 μm. ** denotes P < 0.05.

Figure 8.. Quantitative proteomics of microdissected CCDs show response to administration of LiCl for 72 hrs at a protein level.

(A) A volcano plot displaying protein changes induced by LiCl in microdissected rat CCD segments. Data are from LC-MS/MS-based proteomics analysis of microdissected tubules from rats treated with either LiCl- or NaCl- (as a control group) supplemented diet for 72 hrs. Each point represents the mean value using label-free approach in 3 pairs of CCD samples. The x-axis specifies log₂ of abundance ratio (LiCl over control), and the y-axis specifies -log₁₀ of P-value obtained from moderated t-test (see Methods). 71 proteins which passed a significant threshold of P<0.05 are colored in red and labeled with gene symbol (or UniProt accession if no official gene symbol designated). (B) An overlay of zoomed MS1 spectra (control, blue; lithium, red) for one of the AQP2 peptide sequences, ‘GLEPDTDWEER’, shows a decreased peptide abundance in LiCl-treated CCD. M denotes the m/z of monoisotopic ion of this peptide sequence. [M+1], [M+2], [M+3] indicate a series of natural occurring isotope. (C) A representative MS2 spectrum for the same peptide sequence with matched b- and yfragmented ions annotated (purple peaks).