Novel Type of Renal Amyloidosis Derived from Apolipoprotein-CII

Abstract

Amyloidosis is characterized by extracellular deposition of misfolded proteins as insoluble fibrils. Most renal amyloidosis cases are Ig light chain, AA, or leukocyte chemotactic factor 2 amyloidosis, but rare hereditary forms can also involve the kidneys. Here, we describe the case of a 61-year-old woman who presented with nephrotic syndrome and renal impairment. Examination of the renal biopsy specimen revealed amyloidosis with predominant involvement of glomeruli and medullary interstitium. Proteomic analysis of Congo red–positive deposits detected large amounts of the Apo-CII protein. DNA sequencing of the APOC2 gene in the patient and one of her children detected a heterozygous c.206A→T transition, causing an E69V missense mutation. We also detected the mutant peptide in the proband’s renal amyloid deposits. Using proteomics, we identified seven additional elderly patients with Apo-CII–rich amyloid deposits, all of whom had kidney involvement and histologically exhibited nodular glomerular involvement. Although prior in vitro studies have shown that Apo-CII can form amyloid fibrils and that certain mutations in this protein promote amyloid fibrillogenesis, there are no reports of this type of amyloidosis in humans. We propose that this study reveals a new form of hereditary amyloidosis (AApoCII) that is derived from the Apo-CII protein and appears to manifest in the elderly and preferentially affect the kidneys.

Amyloidosis refers to a spectrum of rare diseases that are characterized by abnormal extracellular deposition of misfolded proteins in the form of insoluble fibrils. A total of 31 amyloid precursor proteins have been identified so far.¹ Renal involvement is common in AL, AA, and ALect2 amyloidosis. However, it may also occur in rare hereditary forms of AFib,² ATTR,³ AGel,⁴ ALys,⁵ AApoAI,⁶ AApoAII,⁷ and AApoAIV⁸ amyloidoses. Even with the availability of proteomics for amyloid typing, about 2% of cases of renal amyloidosis are currently unclassified.⁹

A 61-year-old white female with a history of hypertension and hypothyroidism was recently diagnosed with renal amyloidosis. Four months before, she was noted to have renal insufficiency (serum creatinine of 2.6 mg/dl) and nephrotic-range proteinuria. Serum and urine immunofixation (IFX) showed a monoclonal protein that could not be quantified. A renal biopsy showed amyloid deposits that were negative for Ig heavy and light chains by immunofluorescence (IF). A fat aspirate was negative for amyloid. Serum-free κ/λ light chain ratio was 3.9. Cardiac biomarkers were within normal range. Imaging studies revealed a normal liver and spleen on ultrasound, a negative bone survey, and a negative positron emission tomography scan, other than metatarsal uptake. The paraffin block was subsequently sent for amyloid typing by proteomics (laser microdissection [LMD], liquid chromatography [LC], and tandem mass spectrometry [MS/MS]; LMD-LC-MS/MS). The patient was hospitalized, due to peripheral edema, 2 weeks before her presentation to our hospital. Our workup revealed a hemoglobin of 11 g/dl, serum albumin of 2.6 g/dl, serum creatinine of 2.8 mg/dl, and 3.3 g/d of proteinuria. Serum IFX revealed a small IgG-λ M-protein whereas urine IFX was negative. Serum cholesterol, triglycerides, HDL, LDL, and non-HDL cholesterol were normal. There was no history of renal disease in the patient’s parents, six siblings, or two children (son aged 22 years; daughter aged 23 years).

The renal biopsy showed extensive glomerular and medullary interstitial involvement by acellular eosinophilic material that stained positive with Congo red (CR) and exhibited green birefringence under polarized light (Figure 1). LMD-LC-MS/MS performed on the CR-positive amyloid deposits detected Apo-CII and proteins commonly codeposited with amyloids of all types (ApoE, Apo-AIV, and SAP).¹⁰ We did not detect spectral evidence for other amyloid precursor proteins (Patient 1 in Figure 2A). The Apo-CII protein was also deposited in the entirety of the dissected glomeruli (Figure 2B). We performed germline DNA sequencing of all three exons of the APOC2 gene and found a heterozygous c.206A→T alteration in exon 3, which results in a glutamate to valine substitution at codon 69 (E69V) (Figure 3A). The same mutation was present in the patient’s son (Figure 3B) but not in her daughter (data not shown). We also detected the mutant peptide in the patient’s renal amyloid deposit (Figure 3C), further strengthening the case for the mutation’s pathogenicity.¹¹ We performed generalized Born implicit solvent molecular dynamics (isMD) simulations to assess the effect of the mutation on Apo-CII protein folding. The mutation is located in the linker region and it alters the native confirmation of the protein (Figure 3D), which could destabilize the protein and lead to amyloid formation.

Figure 1.

Light microscopy of AApoCII renal deposits (patient 1). (A) Hematoxylin and eosin staining of renal biopsy shows amorphous eosinophilic deposits in the glomeruli. (B) The glomerular deposits are CR-positive. (C) Apple-green birefringence of congophilic amyloid deposits using polarized light. (D) Extensive medullary interstitial involvement by CR-positive amyloid deposits. Original magnification, ×100 in A; ×200 in B and C; ×100 in D.

Figure 2.

Proteomic detection of Apo-CII in amyloid deposits from AApoC2 patients and from control patients with other renal diseases. (A) Scaffold display of the amyloid proteome from four patients with AApoCII (Patients 1–4) and from normal glomeruli from four healthy controls (Controls 1–4). Patient 1 represents the index case. Amyloid-related proteins are marked with stars and displayed at the top of the list (APOC2, blue star; other amyloid-associated proteins, yellow stars). The numbers displayed in the boxes in the vertical columns represent the total number of spectra matched to the listed protein. The colors of the boxes represent the probability that the spectra represent the identified protein (only spectra with 95% probability of a match to an identified protein [green boxes] are considered for diagnostic interpretation). In all patient samples there were abundant spectra for APOE, SAP, and APOAIV, providing biochemical evidence for the presence of amyloid in the microdissected sample. In addition, all patient samples contained many spectra corresponding to Apo-CII. The control kidney samples did not contain amyloid-associated proteins or Apo-CII. Structural (actin, vimentin, α-actinin-4) or serum/blood-related proteins (hemoglobin, serum albumin) were also detected in the specimens. (B) Portions of the Apo-CII sequence detected in the index sample are highlighted with bold black letters on yellow background. The first 22 amino acids in the sequence constitute the signal peptide, which is post-translationally clipped. (C) OAT stands for amyloid kidneys of other types (including AL, ALect2, AA, AH, AFib, AGel, AApoAIV, AApoAI, ATTR, and Aβ2M). NAGD represents glomeruli of patients with nonamyloid glomerular diseases (dense deposit disease, C3 GN, fibrillary GN, immunotactoid GN, cryoglobulinemic GN, pauci-immune crescentic GN, transplant glomerulopathy, diabetic glomerulosclerosis, glomerular thrombotic microangiopathy, and fibronectin glomerulopathy). Spectral counts from each replicate dissection were normalized to account for batch variation and plotted independently. P values were computed using Mann–Whitney rank sum test.

Figure 3.

APOC2 gene sequencing, mass spectrometry demonstration of corresponding amino acid sequence variant and molecular modeling of mutant Apo-CII. (A and B) APOC2 gene mutation in proband and kindred. An arrow marks the detected c.206A→T (p.E69V) mutation. (C) MS/MS spectrum of the mutant peptide detected in the CR-positive renal amyloid deposits of the proband. Values along the abscissa represent the mass-to-charge ratio (m/z) of the mutant peptide ions detected by mass spectrometry. The ordinate is the relative intensity of the peaks. The purple line at the top highlights the amino acid sequence of the mutant peptide oriented from left to right in the amino-terminus to the carboxy-terminus direction. The letters represent the individual amino acids (T, threonine; Y, tyrosine; L, leucine; P, proline; A, alanine; V, valine; D, aspartic acid; E, glutamic acid; and K, lysine). This sequence corresponds to the purple peaks in the MS/MS spectrum (b-ion series). The blue line at the top highlights the amino acid sequence of the mutant peptide oriented from left to right in the carboxy-terminus to the amino-terminus direction. This sequence corresponds to the blue peaks in the MS/MS spectrum (y-ion series). Arrows marks the amino acid sequence change corresponding to the genetic mutation. (D) Molecular dynamic trajectories of the wild-type (WT; gray) and mutant (gold) protein were assessed using radius of gyration. Mutant and WT protein conformational representatives with similar radii of gyration were shown. Mutant protein has different confirmation when compared with WT.

On the basis of our experience with this patient, we searched our amyloidosis data archive for cases that had abundant Apo-CII in the deposits. This archive contains clinical and proteomic data for a total of 10,167 amyloidosis cases, distributed between 17 subtypes and 30 different tissues, that were typed in our laboratory between 2008 and 2015. We identified seven additional cases, and all were initially reported as “amyloidosis, indeterminate type” because they showed both CR positivity and contained proteins commonly deposited with amyloids of all types¹⁰ but did not contain any recognized amyloid precursor proteins. Importantly, all of these cases were diagnosed from renal biopsies, suggesting that AApoCII primarily affects kidneys. Figure 2A shows the Apo-CII expression from four AApoCII cases and normal glomeruli from four healthy controls. Apo-CII was exclusively detected in the AApoCII glomeruli along with the expected universal amyloid tissue biomarkers.¹⁰ We also wanted to rule out the possibility that Apo-CII can be codeposited in renal amyloid deposits of other types or be present in the glomeruli of patients with nonamyloid glomerular diseases. For this, we utilized normalized MS/MS spectral counts to estimate the relative abundance of Apo-CII protein in the glomerular amyloid deposits of all AApoCII cases (n=8), CR-positive deposits in kidneys of other amyloid subtypes (n=532), glomeruli of patients with nonamyloid glomerular disease (n=49), and normal glomeruli of healthy individuals (n=10). As shown in Figure 2C, we detected Apo-CII protein only in amyloidosis cases of AApoCII type. In addition, we performed immunohistochemistry on paraffin sections of three kidney specimens from the AApoCII cases and on seven cases of renal amyloidosis of other types (AL-λ, n=2; AL-κ, n=1; AA, n=1; AApoAIV, n=3). The glomerular amyloid deposits in the AApoCII cases were all strongly positive for Apo-CII. None of the amyloid deposits in the other cases stained positively for Apo-CII (Supplemental Figure 1).

Our total cohort of AApoCII renal amyloidosis consisted of five females and three males, all elderly at diagnosis (median age 70 years, range 61–86). Patients presented with proteinuria with or without renal insufficiency (Table 1). The median 24-hour urine protein was 3.4 g/d (range 1.5–9.7). Proteinuria was in the nephrotic range in 67% of patients and 43% had full nephrotic syndrome. The median serum creatinine at diagnosis was 1.8 mg/dl (range 0.9–2.6) and 57% of patients had renal insufficiency (Table 1). Histologically, the glomeruli were affected in all cases. Whereas the index patient showed global glomerular involvement (Figure 1, A–C), the remaining seven cases exhibited a distinctive nodular glomerular involvement in which the mesangium was asymmetrically expanded by large rounded masses of amyloid that encroached on the capillary spaces (Supplemental Figure 2). This pattern can also be seen in AA amyloidosis.¹² The massive obliterative glomerular involvement that is characteristic of AFib was not seen in any case. In contrast to AL and AA amyloidosis, vascular involvement was absent (six cases) or minimal (two cases). In contrast to ALect2 amyloidosis, cortical interstitial involvement was absent (six cases) or minimal (two cases). Renal medulla was sampled in five cases, two of which showed medullary interstitial involvement (Figure 1D, Table 1). Prominent medullary interstitial involvement is also common in amyloidosis derived from other apolipoproteins (Apo-AI, Apo-AII, and Apo-AIV). Electron microscopy showed the typical ultrastructural appearance of amyloid with randomly oriented fibrils that measured 8–14 nm in thickness (Supplemental Figure 3). In all cases, amyloid deposits were negative for IgG, IgA, κ, and λ, as assessed by IF.

Table 1.

Clinical and pathologic characteristics of the patients with AApoCII

Age, yr	Gender	Serum Creatinine at Biopsya	24-h Urine Protein at Biopsya	Full Nephrotic Syndrome	No. of Glomeruli Sampled	Globally Sclerotic Glomeruli, %	Degree of Tubular Atrophy/Interstitial Fibrosis	Arteriosclerosis	Glomerular Involvementb	Nodular Glomerular Involvement	Vascular Involvementb	Cortical Interstitial Involvementb	Medullary Interstitial Involvementb	Tubular Basement Membrane Involvementb
61	F	2.3	6	Yes	9	11	Moderate	Marked	3+	No	No	1+	3+	1+
63	M	2.5	9.7	Yes	16	0	Moderate	Mild	2+	Yes	No	No	MNS	No
72	M	1.8	2	No	13	46	Moderate	Moderate	2+	Yes	No	1+	No	No
68	F	0.5	3.2	No	10	20	None	Moderate	2+	Yes	No	No	No	No
61	M	0.9	3.5	No	8	50	Mild	Mild	3+	Yes	No	No	MNS	No
86	F	2.6	NA	NA	8	38	Mild	Marked	3+	Yes	1+	No	MNS	No
75	F	NA	NA	Yes	25	0	Mild	Moderate	3+	Yes	1+	No	2+	No
73	F	1.1	1.5	No	17	12	Mild	None	2+	Yes	No	No	No	No

F, female; M, male; MNS, medulla not sampled; NA, not available.

^aSerum creatinine was represented as mg/dl. Measurements of 24-h urine total protein were represented as g/d. Clinical definitions of proteinuria, nephrotic-range proteinuria, and renal insufficiency are described in Concise Methods section.

^bScoring criteria for glomerular involvement, vascular involvement, cortical interstitial involvement, medullary interstitial involvement, and tubular basement membrane involvement are described in the Concise Methods section.

Apo-CII is a major constituent of chylomicrons, VLDL particles, and HDL particles. Apo-CII activates lipoprotein lipase,¹³ which hydrolyzes triglycerides into free fatty acids. Homozygous or compound heterozygous mutations in the APOC2 gene can cause hyperlipoproteinemia type IB,¹⁴ characterized by hypertriglyceridemia.¹⁵ In contrast, our index patient did not have abnormal levels of lipids. Previous in vitro studies have shown that lipid-free human Apo-CII can also form amyloid fibrils.^16–20 Researchers have shown that mutations introduced into lipid and lipoprotein lipase-binding portions of the Apo-CII protein can promote amyloid fibrillogenesis.^21–23 The E69V mutation discovered in this study is in the linker region, and isMD simulations suggest that the mutation can alter the conformation of the protein. Hence, it is possible that the E69V alteration mediates amyloid fibrillogenesis through destabilization of the mutant Apo-CII protein. Alternatively, the E69V mutation might alter the interaction between Apo-CII and lipids, sufficiently freeing the Apo-CII from the lipids such that amyloid formation can occur.

Concise Methods

The biopsy processing and diagnosis of renal amyloidosis in the eight cases was performed in other institutions. Evaluation of renal biopsies included staining of light-microscopy sections with hematoxylin and eosin, periodic acid–Schiff, Masson trichrome, Jones methenamine silver, and CR; and electron microscopy and standard IF staining for IgG, IgM, IgA, C3, C1q, albumin, fibrinogen, κ, and λ. The paraffin blocks were subsequently sent to the Mayo Clinic for amyloid typing by proteomics (LMD-LC-MS/MS). The extent of amyloid deposits in the kidney, tubular atrophy and interstitial fibrosis, and arteriosclerosis was evaluated by reviewing the CR and hematoxylin and eosin slides routinely prepared at the Mayo Clinic for specimens received for proteomic analysis. Glomerular amyloid deposits were scored using a semiquantitative scale: 0, absent; 1, amyloid deposits affecting <25% of the glomerulus; 2, amyloid deposits affecting 25%–50% of the glomerulus; and 3, amyloid deposits affecting >50% of the glomerulus. The extent of amyloid deposition in the cortical and medullary interstitium was scored as: 0, absent; 1, <25% of parenchyma involved; 2, 25%–50% of parenchyma involved; and 3, >50% of parenchyma involved. The extent of amyloid deposition in tubular basement membranes was scored as: 0, absent; 1, <25% of renal tubules involved; 2, 25%–50% of renal tubules involved; and 3, >50% of renal tubules involved. The extent of amyloid deposition in arteries was scored as: 0, absent; 1, mild; 2, moderate; and 3, severe. The following clinical definitions were used: proteinuria, >150 mg/d; nephrotic-range proteinuria, >3.0 g/d; nephrotic syndrome, nephrotic-range proteinuria, serum albumin <3.5 g/dl, and peripheral edema; and renal insufficiency, serum creatinine >1.2 mg/dl.

Immunohistochemistry for APOC2 was performed on paraffin sections of 11 kidney biopsy specimens involved by amyloidosis (AApoCII, n=3; AL-κ, n=1; AL-λ, n=2; AA, n=1; AApoA4, n=3). Tissue sections were deparaffinized, subjected to citrate buffer antigen retrieval, and incubated with rabbit polyclonal anti-APOC2 antibodies (Abcam, Inc., Cambridge, MA; 1:2500 dilution), followed by incubation in a biotin-free, polymeric horseradish peroxidase–linker antibody conjugate system (Leica Biosystems, Buffalo Grove, IL). Antibody localization was visualized using diaminobenzidine and the sections were counterstained with hematoxylin.

We utilized a previously established proteomics method for typing the amyloid deposits.²⁴ For each case, CR-stained, formalin-fixed, paraffin-embedded tissue sections were examined to confirm the presence of amyloid. The amyloid type was identified using an LMD-assisted LC-MS/MS method as previously described.²⁴ Multiple independent samples (replicates) were analyzed for each patient. Data for each patient was processed using a previously described bioinformatics pipeline and a patient-specific amyloid proteome profile was created.²⁵ For each patient sample, a pathologist reviewed the microdissection images to confirm that the amyloid deposits were included in the LC/MS-MS-analyzed tissue. After LC-MS/MS, the proteomic results, as displayed using Scaffold software (Proteome Software, Portland, OR), were reviewed (Figure 2A). The proteome profile was first scrutinized for the presence of proteins that are codeposited with amyloids of all types (Apo-E, SAP, and Apo-AIV)¹⁰ to verify that the basic biochemical signature of amyloid was detected. Next, the proteome was searched for potential amyloid proteins. In all eight cases, previously described canonical amyloid proteins were not identified in the proteome; rather, abundant spectra corresponding to Apo-CII were consistently detected across all replicates. This protein was previously shown to be amyloidogenic in animal models and in vitro studies.^{16,17,19–23} Thus, Apo-CII was considered a potential primary amyloid precursor protein in these cases.

Genetic evaluation for germline mutations in the APOC2 gene was performed using a blood sample. All three exons of the APOC2 gene were amplified using hybrid primers containing 20–22 bases of a gene-specific sequence and a universal sequencing primer sequence (19 or 23 bases for the forward and reverse primers) at the 5′ end. Amplified products were sequenced using universal sequencing primers, ABI BigDye terminators (Applied Biosystems, Foster City, CA), and capillary electrophoresis on an ABI 3730 sequencer. Data were analyzed using Mutation Surveyor (SoftGenetics, College Station, PA), configured to use corresponding reference sequences obtained from GenBank.

isMD simulations were carried out using NAMD²⁶ and the CHARMM22 with CMAP²⁷ force field. The PDB structure 1SOH²⁸ was utilized for our initial conformation. Initial mutant conformations were generated using PyMOL (The PyMOL Molecular Graphics System, version 1.5.0.3; Schrödinger, LLC, Cambridge, MA). We utilized an interaction cutoff of 15 Å with strength tapering (or switching) beginning at 12 Å, a simulation time step of 1 femtoseconds, conformations recorded every 2 picoseconds. Each initial conformation was used to generate five replicates and each was energy-minimized for 20,000 steps, followed by randomized temperature initialization (at 10 K; a standard approach that provides variation between the replicates), followed by heating to 300 K over 300 picoseconds via a Langevin thermostat. A further 12 nanoseconds of simulation trajectory was generated and the final 10 nanoseconds analyzed. Analysis was carried out using custom scripts, leveraging the Bio3D R package²⁹ and visual molecular dynamics.³⁰

Disclosures

None.