Complement 7 Is Up-Regulated in Human Early Diabetic Kidney Disease

Abstract

There is a temporal window from the time diabetes is diagnosed to the appearance of overt kidney disease during which time the disease progresses quietly without detection. Currently, there is no way to detect early diabetic nephropathy (EDN). Herein, we performed an unbiased assessment of gene-expression analysis of postmortem human kidneys to identify candidate genes that may contribute to EDN. We then studied one of the most promising differentially expressed genes in both kidney tissue and blood samples. Differential transcriptome analysis of EDN kidneys and matched nondiabetic controls showed alterations in five canonical pathways, and among them the complement pathway was the most significantly altered. One specific complement pathway gene, complement 7 (C7), was significantly elevated in EDN kidney. Real-time PCR confirmed more than a twofold increase of C7 expression in EDN kidneys compared with controls. Changes in C7 gene product level were confirmed by immunohistochemistry. C7 protein levels were elevated in proximal tubules of EDN kidneys. Serum C7 protein levels were also measured in EDN and control donors. C7 levels were significantly higher in EDN serum than control serum. This latter finding was independently confirmed in a second set of blood samples from a previously collected data set. Together, our data suggest that C7 is associated with EDN, and can be used as a molecular target for detection and/or treatment of EDN.

Over the past two decades, the incidence of end-stage renal disease in the United States has nearly doubled because of a steady increase in diabetic nephropathy (DN). In 1982, only 23% of new end-stage renal disease cases were attributed to diabetes, but by 1999, at 43%, DN became the leading cause of end-stage renal disease.¹ In 2015, the incidence of end-stage renal disease was >130,000, with prevalence >700,000 and an annual death rate of >100,000 individuals.² Despite progress made in delaying dialysis dependence, these trends are expected to increase, at least for the near future.

DN affects 15% to 25% of type 1 diabetics and 30% to 40% of type 2 diabetics.³ The detection of albumin in urine is often the first clinical indication of kidney disease. The natural progression of diabetic kidney disease has been closely examined in type 1 diabetic patients, in whom timing of disease onset is generally known. Investigation into type 1 diabetes has shown progressive decline in renal function over time (>3.5 mL/minute per year).⁴ However, for many years, the decline was not detected by standard laboratory testing because patients were deemed to have normal renal function if the estimated glomerular filtration rate was >60 mL/minute. Later in the disease process, individual patients can have differing rates of renal decline. Progressive renal decline precedes the onset of microalbuminuria and, as it continues, it increases the risk of proteinuria.⁴

Evidence from autopsy studies indicates that diabetic nephropathy remains relatively underdiagnosed,5, 6, 7 and diabetic kidney damage,⁸ often with no overt clinical findings, is usually confirmed during postmortem pathologic examination. In 2010, the Renal Pathology Society developed a new classification system for diabetic kidney disease.⁹ According to this classification: class I kidneys show increased thickness of glomerular basement membrane (≥395 nm in females and ≥430 nm in males), class II kidneys have mild (class IIa) to severe (class IIb) mesangial expansion, class III kidneys show at least one convincing Kimmelstiel-Wilson lesion, and in class IV kidneys show 50% of glomeruli have global glomerular sclerosis. Pathologic staging of diabetic kidney disease appears to be unrelated to clinical staging of chronic kidney disease, such as blood creatinine level.

Mechanisms underlying early diabetic nephropathy remain unknown, and there is a paucity of literature on renal molecular targets in early DN (EDN).10, 11, 12, 13, 14, 15, 16 Most differential gene expression studies have been performed in rodent diabetic models, and although rodent experiments provide important insights into disease mechanisms, they are limited because of the inherent differences between murine and human diabetic kidney disease. In the few human studies that have been reported in the literature, all have looked at kidneys with advanced diabetic kidney disease. There appears to be no study in the literature that has investigated gene expression in postmortem human kidneys with early diabetic nephropathy, otherwise scheduled for the kidney donor pool. Herein, we have compared transcriptome profiles of human postmortem early diabetic kidneys with those of nondiabetic kidneys. We then selected one of the most significantly increased EDN genes for further examination in postmortem human kidney and blood.

Materials and Methods

This study was approved by the Massachusetts General Hospital Institutional Review Board and abided by the guidelines set forth by the Declaration of Helsinki. No donor organs were obtained from executed prisoners or other institutionalized individuals at any national transplant center. Organ procurement offices were coordinated to retrieve organs with appropriate informed consents. All patient information was deidentified and was thus Health Insurance Portability and Accountability Act compliant.

Study Population and Sample Collection

Postmortem biological samples (blood and kidney tissue) from nondiabetics and individuals with EDN were obtained from the National Disease Research Interchange (Philadelphia, PA). Patient demographics have been provided in Table 1. Except for obesity, there was no significant difference in any of the parameters between study groups; there were more obese individuals in the EDN group than among nondiabetic controls. All diabetic donors had type 2 diabetes. Diabetes status was defined as hemoglobin A1c ≥6.5%. Proteinuria was identified by urine dipstick test, performed during the terminal hospital admission. These diagnoses were confirmed by careful review of premortem and postmortem clinical health records. There were stringent inclusion and exclusion criteria. Inclusion criteria required that: i) donors be between 17 and 80 years of age, ii) diabetic status be confirmed as noted above, and iii) the postmortem time be within 24 hours from cross-clamping in the operating room. The only exception with respect to hemoglobin A1c testing was if a donor had been prescribed and/or using glycemic agent(s), including but not limited to metformin and insulin, in which case repeated hemoglobin A1c testing was not mandatory. Exclusion criteria included active infection; malignancy; severe glomerulosclerosis, as determined in frozen operating room biopsy material; history of renal replacement therapy (hemodialysis or peritoneal dialysis at any time); and any known genetic renal condition, such as polycystic kidney disease. Pathologic staging of diabetic tissues was performed using the 2010 Renal Pathology Society classification system for diabetic kidney disease.⁹ From each postmortem nephrectomy sample, >100 glomeruli were reviewed in tissue sections. Pathologic classes II and III diabetic kidney disease were considered as EDN; class III kidneys were included because the renal function in these donors was preserved.

Table 1.

Summary of Patient Demographics

Sample type	Nondiabetic	Diabetic
Harvested kidneys
n (%)	18 (51)	17 (49)
Male sex, n (%)	8 (44)	6 (35)
Age in years, mean (range)	50 (17–75)	57 (34–78)
Hypertension, n (%)	9 (50)	17 (100)
Obese, n (%)	5 (28)	12 (71)
Tobacco use, n (%)	12 (67)	10 (59)
White, n (%)	14 (78)	12 (71)
Serum creatinine, mean (range), mg/dL∗	1.4 (0.4–4.5)	1.3 (0.5–2.3)
Proteinuria (≥1 + by UA), n %	4 (22)	7 (41)
Hemoglobin A1c, mean % (range)	5.4 (4.7–5.9)	8.2 (6.5–11.3)
eGFR (per MDRD), mean mL/minute	69.1	51.6
Stage I to IIIA eGFR, n %†	11 (61)	9 (53)
Serum complement C3, mg/dL	112	137
Serum complement C4, mg/dL	25	29
Total complement CH50, U/mL	51	57
Alternative complement AH50, U/mL	97	127
Kidney samples for microarray‡
n	4	4
Male sex, n	2	2
Age in years, mean (range)	52 (24–65)	64 (55–78)
Hypertension, n (%)	3 (75)	4 (100)
BMI in kg/m2, mean (range)	27 (26–29)	35 (22–43)
Tobacco use, n (%)	2 (50)	2 (50)
White, n (%)	3 (75)	3 (75)
Terminal serum creatinine, mean, mg/dL	0.9	1.4
Proteinuria (≥1 + by UA), %	50	50
Hemoglobin A1c, mean, %	5.4	9.5
eGFR, mean mL/minute	>60	>60
PRIMO blood samples
n	20	39
Male sex, n (%)	14 (70)	30 (77)
Age in years, mean (range)	66 (47–81)	66 (48–84)
Hypertension, n (%)	19 (95)	38 (97)
Obese, n (%)	6 (30)	24 (62)
Tobacco use, n (%)	11 (55)	23 (59)
White, n (%)	15 (75)	29 (74)
Serum creatinine, mean, mg/dL∗	2.1	2.1
Proteinuria (≥1 + by UA), n (%)	10 (50)	24 (62)
eGFR (per MDRD), mean mL/minute	33	34
Class I to IIIA or better eGFR, n (%)†	2 (10)	5 (13)

AH50, alternative pathway complement activity; BMI, body mass index; CH50, total complement activity; eGFR, estimated glomerular filtration rate; MDRD, Modification of Diet in Renal Disease; PRIMO, Paricalcitol Capsules Benefits Renal Failure Induced Cardiac Morbidity in Subjects with Chronic Kidney Disease Stage 3/4; UA, urinalysis.

^∗At time of admission.

^†Based on MDRD equations.

^‡Subset of harvested kidneys.

Isolation of Total RNA

For all real-time PCR studies, mRNA was extracted exclusively from kidney cortical tissue. Total RNA from flash-frozen kidney tissue was subjected to Trizol (Invitrogen, Carlsbad, CA) extraction and purified using an RNeasy mini kit. Total RNA extracted was quantitated using a NanoDrop Spectrophotometer (ND-2000; Thermo Scientific, Waltham, MA), and stored at −80°C until use. mRNA quality was confirmed by gel electrophoresis. Good quantities of high-quality RNA were obtained from all postmortem kidney samples.

Affymetrix Chip Hybridization and Expression Profiling

mRNA preparations were processed by the Beth Israel Deaconess Medical Center Genomics, Proteomics, Bioinformatics, and Systems Biology Center, per the standard Affymetrix protocol using the Affymetrix Human Transcriptome 2.0 GeneChip (Affymetrix, Inc., Santa Clara, CA), as previously described.¹⁷ Gene expression data have been deposited with the National Center for Biotechnology Information Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo; accession number GSE111154).

Gene Ontology, Pathway, and Interactive Network Analysis

To assess for potential molecular pathways underlying the early diabetes–specific gene expression signature and to more precisely understand the complex interactions between the differentially expressed genes, the 285 differentially expressed genes were further analyzed using the Ingenuity Pathway Analysis software package version 8.0 (Qiagen Bioinformatics, Redwood City, CA). P values were calculated on the basis of gene enrichments in biological pathways or networks using the Fisher exact test (P < 0.05). Gene ontology, functional category, canonical pathway, upstream regulator, and interactive network analyses were performed on the 285 differentially expressed genes. Interactive network analysis identified the complex interactions between the differentially expressed EDN-specific genes and the key focus hub genes (genes with the largest number of connections and anticipated to be the ones that ensure the stability and integrity of the network). Ingenuity Pathway Analysis also helped identify the canonical pathways with the statistically highest enrichment of EDN-specific genes.

cDNA Synthesis and Real-Time PCR

Protocol, as described in the literature,¹⁸ was used to perform real-time PCR for complement 7 (C7). Briefly, 1 μg of total cortical RNA from each specimen was reverse transcribed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA) with a Qiagen RNase inhibitor for 10 minutes at 25°C, 120 minutes at 37°C, and 5 minutes at 85°C in a GeneAmp PCR System 2700 thermal cycler (Applied Biosystems). The resulting cDNA was amplified using primers specific for the human C7 gene (UniGene Hs78065; chromosome 5: 40909497-40982939 on Build GRCh38; forward primer, 5′-TGTAAAACGACGGCCAGT-3′; reverse primer, 5′-CAGGAAACAGCTATGACC-3′; Applied Biosystems, Thermo Fisher Scientific, Waltham, MA) and the housekeeping β-actin gene (Applied Biosystems catalog number 4333762T; Thermo Fisher Scientific, Waltham, MA).

Enzyme-Linked Immunosorbent Assay

Serum component C7 protein levels from postmortem donors (Table 1) were measured using a commercially available quantitative enzyme-linked immunosorbent assay kit (catalog number 125964; Abcam, Cambridge, MA), as per manufacturer's instructions. The serum component C7 assay was specific for C7 and is not known to cross-react with any other complement component. The interassay CV was <12%.

Immunohistochemistry and Histology

Immunohistochemistry of formalin-fixed, paraffin-embedded kidney tissue sections was performed using a rabbit polyclonal IgG antibody to human C7 (catalog number HPA001465; Sigma, Billerica, MA; 1:200 dilution) and a goat secondary antibody staining kit (catalog number MP-7405; Vector Laboratories, Burlingame, CA). Scoring of immunohistochemical staining was performed as follows: 0, no staining; 1, low staining; 2, medium staining; and 3, high staining.

For light microscopic histology, kidney tissues were fixed in 10% buffered formalin, dehydrated, and embedded in paraffin by conventional techniques. Sections were stained with periodic acid-Schiff or hematoxylin-eosin. Ultrastructural images were captured using an FEI Morgagni Transmission Electron Microscope with an AMT 2K digital charge-coupled device camera (Advanced Microscopy Techniques, Woburn, MA). As per the protocol outlined in Dickersin's Diagnostic Electron Microscopy Text/Atlas, glomerular basement membrane thickness was measured using a rectangular grid. Tangential measurements and examination of paramesangial areas were avoided.

Additional Outcome Measures

Serum levels of complement component 3 (C3) and complement 4 (C4), total complement activity, alternative pathway complement activity, and urine protein and urine creatinine were measured by the Massachusetts General Hospital Clinical Resource Center. Creatinine measurements were also obtained from the primary hospital caring for the donor before death. An estimated glomerular function was determined using an online calculator, as established by the National Kidney Foundation (New York, NY; https://www.kidney.org/professionals/kdoqi/gfr_calculator, last accessed November 8, 2017).

Statistical Analysis

Demographic characteristic differences between nondiabetic donors and donors with EDN were compared using the χ² test, the Fisher exact test, the t-test, or the U-test, as deemed appropriate. Statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, Inc., La Jolla, CA). Differences between serum component C7 levels were analyzed using paired t-test. All reported P values are two sided, and P < 0.05 was regarded as statistically significant.

Results

Gene Expression Changes in EDN Kidney

Microarray analysis was performed on four EDN and four nondiabetic kidneys (patient demographics provided in Table 1). Ingenuity pathway analysis revealed top canonical pathways enriched among the EDN-specific differentially expressed genes. Compared with nondiabetic kidneys, EDN kidneys had 245 up-regulated genes and 44 down-regulated genes (Figure 1).

Figure 1.

Hierarchical cluster analysis of the top 80 genes in kidneys from donors with EDN and matched controls. Red indicates up-regulated genes; blue, down-regulated genes. Rows are genes, and columns are individual kidneys. ID, identification.

Gene Ontology and Pathway Enrichment Analysis

Compared with nondiabetic controls, EDN kidneys had overrepresentation of up-regulated genes in pathways linked to the complement system, hepatic fibrosis, phagosome formation, and acute phase response signaling. The complement system was identified as the canonical pathway with the statistically highest enrichment of EDN-specific genes, and was predicted to be activated because most of the genes in this pathway were increased in EDN kidneys. Quantitative PCR analysis of EDN kidneys showed a twofold to threefold higher level of C7 compared with kidneys from nondiabetic controls (Figure 2A).

Figure 2.

A: Real-time PCR for C7: C7 expression was measured in postmortem normal kidneys (ND) and early diabetic (ED) kidneys. Real-time PCR analysis shows that kidneys with EDN (white bar) have significantly higher levels of C7 gene expression (quantified as fold change in mRNA expression compared with the housekeeping β-actin gene) compared with matched nondiabetic control kidneys (black bar). Significance shown compared with ND. B: C7 immunohistochemistry in normal kidney and kidney with EDN. C7 is expressed in both renal tubular cells and endothelial cells. C: C7 protein levels in normal and EDN kidneys. C7 immunohistochemistry data were semiquantitated. Levels are significantly higher in EDN kidneys than nondiabetic kidneys. Closed bar: nondiabetic kidneys. Open bar: EDN kidneys. n = 4 (A, ND); n = 3 (A, ED kidneys, and C, EDN and nondiabetic kidneys). ^∗∗∗P < 0.005 versus ND. Original magnification, ×20 (B).

Component C7 Protein in EDN Kidney

Component C7 immunostaining was predominantly noted in the tubular segments (both cortical and medullary), with proximal tubules showing greater staining. Vessels in controls showed patchy mild staining of the media, whereas diabetics showed a similar but more intense pattern in the tubules. In addition, areas of hyalinosis in both glomeruli and vessels were intensely positive (Figure 2B). Semiquantification of immunohistochemical staining confirmed greater staining in EDN kidneys compared with nondiabetic controls (Figure 2C).

Component C7 Protein Level in EDN Donor Blood

Serum component C7 protein levels in EDN donors and nondiabetic controls were determined by enzyme-linked immunosorbent assay sandwich assay. C7 protein levels were significantly higher (P < 0.05) in the serum of donors with EDN (n = 8) than in nondiabetic controls (n = 8) (Figure 3A).

Figure 3.

A: Serum C7 levels in postmortem blood samples from nondiabetic and diabetic donors. Postmortem serum C7 levels were measured using enzyme-linked immunosorbent assay sandwich assay, and levels are significantly increased (P < 0.05) in donors with diabetic nephropathy (DN) compared with nondiabetic (ND) controls. B: Serum levels of C7 in patients enrolled in the Paricalcitol Capsules Benefits Renal Failure Induced Cardiac Morbidity in Subjects with Chronic Kidney Disease Stage 3/4 (PRIMO) trial. Serum C7 levels are significantly higher (P < 0.005) in patients with clinically identified diabetes compared with nondiabetic matched chronic kidney disease (CKD) controls. Closed dots indicate nondiabetic; open dots, diabetic. n = 8 (A, donor with DN and ND controls); n = 34 (B, patients with clinically identified diabetes); n = 18 (B, nondiabetic matched CKD controls).

Validation of Increased Serum C7 Protein Level in an Independent Cohort

The validity of elevated C7 level was tested in EDN blood by examining additional blood samples collected from an earlier clinical trial19, 20 (Table 1). Serum C7 protein levels were significantly elevated in patients with clinically identified diabetes compared with nondiabetic controls (Figure 3B); both groups had chronic kidney disease. To determine the specificity of C7 changes, levels of additional C3, C4, total complement activity, and alternative pathway complement activity were measured in diabetic and control sera. Levels of all these complements in nondiabetic and EDN samples were within normal limits (Table 1).

Discussion

Early diagnosis of EDN has the potential of improving the quality and duration of life in diabetic patients. Earlier, gene expression analysis studies were performed in animal models,21, 22 and the few human studies reported in the literature investigated gene expression in kidneys with advanced diabetic kidney disease.10, 23 This is the first study to perform differential gene expression analysis in well-characterized diabetic kidneys in the early stages of nephropathy when kidneys have undergone relatively little glomerulosclerosis or reduction in estimated glomerular filtration rate compared with matched nondiabetic kidneys. In the present study, pathologic class II and III kidney disease was considered as EDN; class III kidneys had at least one glomerulus with a nodular increase in mesangial matrix (Kimmelstiel-Wilson). The presence of unequivocal mesangial expansion, often with limited Kimmelstiel-Wilson nodules, was deemed the earliest time point when diabetic kidney disease could be identified.²⁴ Several families of genes were dysregulated in EDN kidneys, one of which was the complement system gene family. C7 expression was significantly increased in EDN kidneys. Compared with controls, component C7 protein levels were elevated in the blood of EDN donors. There was no alteration in the levels of any of the other complements tested (C3, C4, total complement activity, or alternative pathway complement activity) in the serum of EDN and control donors. Together, these data suggest that before any clinical manifestations of diabetic nephropathy, such as proteinuria or changes in glomerular filtration rate, C7 gene and gene product levels are elevated in postmortem human kidney and blood.

An association between complement activation and diabetic kidney disease has been shown in animals.25, 26, 27 In humans, complement activation has been noted in a variety of conditions, including Alzheimer disease,28, 29 fibromuscular disease,³⁰ and preeclampsia.31, 32 Complement activation has been hypothesized to be important in the initiation and development of peripheral vasculopathy,³³ diabetic neuropathy,³⁴ and proliferative diabetic retinopathy.35, 36 In one human study,¹⁰ gene expression profiling was done in cadaveric donor kidneys from two individuals with normal renal function and histologic characteristics and in two individuals with type 2 diabetes (both diabetic kidneys showed extensive glomerular hypertrophy, significant glomerulosclerosis, and interstitial fibrosis). Of the 12,000 genes on the chip, 96 were up-regulated and 519 were down-regulated. Authors reported down-regulation of vascular endothelial growth factor, nephrin (NPHS1), and TGFB1 genes in advanced diabetic kidney disease kidneys. It is unclear whether members of the complement family were investigated or not. Another human transcriptome analysis study performed in normal kidneys (n = 13) and kidneys with advanced diabetic kidney disease (n = 9) reported that the complement system (C3, CD55, and C1qA) was differentially expressed in diabetic kidney disease kidney, with a threefold increase in C3 expression.²³ Both these human studies used renal tissues from subjects with advanced diabetic kidney disease, where kidneys had already undergone extensive renal pathologic damage.

Three pathways are involved in the activation of the complement system.³⁷ The classic pathway, when activated by the complement 1 complex, cleaves complex components C4 and complement 2. The lectin pathway is activated by mannose-binding lectin and N-acetylglucosamine residue of ficolins. When this pathway is activated, it also cleaves C4 and complement 2. The alternate pathway is activated after activation of C3. Activation of all three pathways results in the formation of the membrane attack complex (MAC). MAC deposition has been reported in diabetic kidney.³⁸ C7 is a critical component of the terminal pathway of complement activation.³⁹ C7, along with complements 6, 8, and 9, is involved in MAC formation.⁴⁰ In this study, marked elevations were noted only in C7 levels in kidneys with EDN. Levels of C3 and C4 did not show any significant alterations in EDN kidneys, and total complement activity and alternative pathway complement activity levels were within normal limits; levels for complement 6, 8, and 9 were not examined. Future studies will investigate these latter complements in EDN kidneys. The membrane-bound C7 complex forms the transmembrane pore.⁴¹ C7 is expressed in endothelial cells as a trap for assembling MAC.42, 43 Alterations in MAC formation have been reported in several pathologic conditions. Vasil and Magro³³ found MAC deposition in skin biopsy specimens from diabetics with vascular lesions. MAC deposition has also been seen in glomerular injury, dermatomyositis microvasculature,44, 45 smooth muscle,⁴⁶ and mesangium of kidneys with more advanced diabetic disease.⁴⁷ In this study, tubular C7 deposition was observed. It is possible that the proximal tubule is an early target of injury.⁴⁸ It is not clear if there is juxtaglomerular feedback involving tubular C7 and mesangial MAC formation. Further studies will be required to identify a causal relationship between the two. Most diabetic donors experienced a fatal cardiovascular condition, stroke, or anoxia from presumed cardiac causes, whereas causes for death in nondiabetic donors included blunt head trauma and suicide from hanging. Whether these conditions affect complement regulation needs investigation.

These findings show that in the early stages of diabetes, there is increased expression of C7 and its gene product in EDN kidney. C7 is released into circulation and is reflected as high levels of C7 in blood. These findings are supported by other studies where vascular deposition of complement proteins has been reported. Falk³⁸ reported complement deposition in kidney from diabetics but not in kidneys from membranoproliferative renal disease. This study suggests that elevated C7 gene, and gene product, can be a biomarker for early diabetic nephropathy. Therapeutics based on this terminal complement may be used to treat diabetic patients in a precise and timely manner to prevent the development of clinically significant diabetic nephropathy. Complement activation has been suggested as a therapeutic measure in the progression of other diseases, including sepsis, age-related macular degeneration, Alzheimer disease, and atypical hemolytic-uremic syndrome.37, 49 Recently, complement inhibitors, such as the C5 antibody, have been approved by the US Food and Drug Administration for therapeutic use in paroxysmal nocturnal hemoglobinuria and atypical hemolytic uremic syndrome.50, 51

In conclusion, based on differential transcriptome microarray analysis of postmortem human EDN and nondiabetic kidney tissues, we report the identification of a novel target gene C7 with a strong signal in early diabetic nephropathy. C7 expression is significantly increased in early diabetic kidneys compared with nondiabetic kidneys. We further show a concurrent increase in C7 protein levels in the postmortem blood from EDN donors. Together, these data provide a novel approach to identify EDN before signs of the disease become clinically apparent. These findings may lead to the development of distinctive therapeutic strategy(s) in the early detection and/or treatment of EDN, and slow the progression of the disease. A noninvasive diagnostic method would allow for early detection of previously imperceptible renal damage, and may be more informative than hemoglobin A1c level, creatinine clearance, or urine protein quantification.