healthmedicinet

Stone extracts

It is not exactly clear what fraction of the total stone protein can be extracted
in a form usable for proteomic analysis. Our extraction method yielded 7.20 ?g protein/mg
stone powder for CaOx-Ia and 0.81 ?g/mg for CaOx-Id. The larger value is 3 times the
average extraction yield that we have previously reported for calcium oxalate monohydrate
(COM) stones using 9 M urea/1 mM DTT 5]. The increased amount of protein in CaOx-Ia was likely due to the presence of the
x-ray lucent material identified by micro CT. Note that most researchers, not using
micro CT, would likely be unable to detect the presence of such a non-mineral-rich
material, which, in this case, clung tightly to the surface of stone fragments. On
the other hand, protein extraction from specimen CaOx-Id yielded a value within the
range we found previously for different COM stone specimens 5], though on the low end of that range. The low yield of protein in this specimen might
be related to the tightly packed nature of this form of COM stone (Fig. 1d).

Fig. 1. Representative pieces from the two stone specimens used. a Fragment from CaOx-Ia, on mm-grid paper. b Micro CT slice through fragment in A, showing pure calcium oxalate monohydrate (COM).
c Two stones from specimen CaOx-Id; stone on right has shell broken off to reveal interior
core, and on left is shell (top) and core from another stone in this specimen. For
CaOx-Id, only the shell portions were collected for protein extraction. d Micro CT slice through CaOx-Id stone, showing pure COM

The total amount of protein in these types of stones, estimated from complete acid
hydrolysis and measurement of amino acids, has been reported to be about 17 ?g/mg
2], but of course, such an acid hydrolysis destroys protein identities. It is likely
that a considerable amount of stone protein is resistant to extraction by solubilization
methods that maintain protein primary structure, as has been previously reported 5]. In comparison, of the previous studies where proteomic analysis was used and extracted
protein yields reported 7], 9], 17], the average was 1.5 ?g/mg, probably less than 10 % of the total protein contained
within the stones.

Whether the extraction resistant proteins are 1) very low abundance components of
the matrix proteome trapped within the crystal matrix 15], 16] and thus not detected by conventional means, 2) chemically more hydrophilic and accordingly
more avidly bound to crystal surfaces, or 3) merely unextracted replicates of the
proteins identified below remains to be determined.

Label-free quantitative mass spectrometry

Using a label-free quantitative mass spectrometry platform, we identified and quantified
1,059 unique protein database (UniProt, http://www.uniprot.org) entries including splice variants or isoforms (809 unique gene names), in the two
human kidney stone samples with a false discovery rate (FDR) of ?0.2 %. These proteins
are listed in a table in Additional file 1, along with their UniProt identities, gene symbols, protein names, and abundances
22]. As illustrated in Fig. 2, 606 proteins were common to both stone types; 70 proteins were unique to CaOx-Ia,
while 383 proteins were detected only in the CaOx-Id stone powder. Specific peptide
information for all identified proteins and protein groups including protein coverage,
# of unique sequences, # of identified peptides, total # of identified sequences,
is available in the table found in Additional file 2.

Fig. 2. Venn diagrams showing the degree of overlap and exclusivity of proteins identified
and quantified in CaOx-Ia and CaOx-Id kidney stone powders

To account for potential bacterial proteins within the stone matrix, the MS data were
searched against Corynebacterium, Actinomyces, Lactobacillus jensenii, Streptococcus anginosus, and Staphylococcus epidermidis protein databases. No proteins were identified. This strongly suggests that there
is no significant bacterial contribution to the stone matrix proteome in the specimens
tested.

Since 2008, twelve papers have been published that described various analyses of human
kidney/bladder stone matrix proteomes or crystal-associated proteins 6]–17]. Compared to those earlier investigations, we found the kidney stone matrix proteome
to be larger and much more complex than observed previously. We identified and quantified
between 4–35 times as many proteins as found in the other human studies. More specifically,
we found 259 proteins that were identified in earlier studies (see the Human Kidney
Stone Matrix Proteome Database presented in Additional file 3), and also an additional 577 proteins not identified previously. This database also
contains the following information for each protein: isoelectric point (pI), # of
negatively charged residues (Asp?+?Glu), # of positively charged residues (Arg?+?Lys),
Neg/Pos Ratio, aliphatic index, GRAVY score (grand average of hydropathicity), molecular
class, biological process, cellular component, and function.

The increased protein identification rate that we achieved in this investigation may
be due to two factors. First, several previous studies used a demineralization approach
to extract proteins from powdered stones that included EDTA and/or SDS 7]–9], 11], 13]–17], both of which were subsequently removed by either centrifugation or dialysis steps.
Boonla et al. 6] used a commercially available lithium dodecyl sulfate/glycerol solution at 100 °C
to extract proteins. Our use of 8 M urea/10 mM DTT with sonication and two repeated
overnight extractions of the stone powder required no additional purification steps
where proteins might be lost. As mentioned earlier, this approach yielded, on average,
more protein/mg of stone powder than in previous studies. But that alone does not
account for the differences in the number of proteins detected. A second reason may
lie in the protein analysis itself. Several previous studies used 1D and/or 2D gel
electrophoresis and subsequent LC-MS/MS or MALDI-MS/MS to identify proteins in gel
plugs 6], 11], 13], 14], 16], 17]. Others used LC-MS/MS or MALDI-MS/MS of whole extracts 7]–10], 14]–17], but used LC gradient profiles of 60 min or less to separate the tryptic peptides.
Our use of a 190 min LC gradient combined with the rapid scanning features of the
Orbitrap Velos Pro mass spectrometer likely underlie our increased protein identification
rate. It should be noted that in a proteomic analysis of rat urinary melamine stone
matrix (where urea, thiourea, detergent and DTT were used to extract proteins, a 90 min
LC gradient and a high resolution Orbitrap mass spectrometer were used to separate
and analyze tryptic peptides), over 1,000 proteins were identified 24].

Interestingly, there are 204 proteins listed in the database in Additional file 3 that were identified in the previous twelve studies, but were not detected in our
stone samples. Nearly half of these are accounted for by 96 proteins (of 242) identified
by Jou et al. in uric acid stones 10]. The other 146 uric acid stone proteins they detected were also identified in our
CaOx stones. This substantial dissimilarity likely is due to the physicochemical differences
between these various stone types and the extraction methods and analytical approaches
used. Additionally, Merchant et al. analyzed the proteome of CaOx stones obtained
from human subjects and identified 158 proteins 14]. Forty-five of these were not detected in our CaOx stones, suggesting that even within
similar stone types, the protein composition may be patient-specific and thus differ
significantly.

A brief, non-statistical comparison of CaOx-Ia and CaOx-Id stone matrix protein composition
is presented below. At this point, it is important to restate that only two stone
matrices were analyzed in this study, by design, and that our intent was to 1) improve
and simplify the extraction of protein from the stone powder and 2) apply a novel,
more comprehensive LFQMS approach to identify and quantify as many proteins as possible.
The following data analysis is by no means intended to imply the existence of significant
differences between the unique CaOx stones. It is presented here to compare and contrast
the results of our new quantitative approach to previous stone matrix proteome studies
and not to make inferences on the pathophysiology of CaOx stone formation.

The 50 most abundant proteins from each stone are listed in Table 1. These correspond to many of the proteins reported in previous studies, in particular:
serum albumin, apolipoproteins, calgranulins, osteopontins (10-12-fold higher in CaOx-Id),
prothrombin, alpha and beta hemoglobin, neutrophil defensin I, complement proteins,
and alpha-1-antitrypsin. In contrast, proteins such as fatty acid synthase and numerous
cytokeratins were rarely observed in previous studies. Their absence from previous
studies may be accounted for by the fact that these proteins were not disclosed as
their occurrence is generally thought to be due to contamination from skin and ambient
environment. This may be true of the epidermal and cuticular keratins identified in
this study (KRT2, KRT10, KRT31, KRT77, KRT81 KRT84, KRT85), while others are well-established
components of epithelial cells throughout the kidney 25]–27] (KRT1, KRT3, KRT5, KRT6, KRT7, KRT8, KRT9, KRT13, KRT14, KRT17, KRT18, KRT19, KRT31,
KRT80).

Table 1. Fifty most-abundant proteins in kidney stones CaOx-Ia and CaOx-Id

When one examines the database in Additional file 3, of the 257 proteins reported in other papers that also were found in our study,
albumin, uromodulin (Tamm-Horsfall Protein, THP), calgranulin-A (Protein S100-A8),
and calgranulin-B (Protein S100-A9) were common to all studies. Albumin and THP were
similarly abundant in both our CaOx stones while calgranulin-A was nearly ten-fold
higher in CaOx-Ia. Lactotransferrin, osteopontin, and prothrombin were detected in
ten previous studies and Vitamin K-dependent protein Z was detected in nine. Twenty-two
immunoglobulin-related proteins and 28 complement-related proteins were identified
and quantified, and most were common to both of our CaOx specimens.

Note that many of the proteins listed are undoubtedly from blood or tissue, but this
is also true for proteins found in human urine 28]. Some work that has been done on crystallization of calcium oxalate crystals in vitro in human urine has suggested that proteins of blood and tissue are not adsorbed to
forming crystals, with the idea that proteins of the blood or tissue are not relevant
to stone formation 29]. But simple crystal formation is not the same as the formation of a stone, in which
the protein matrix plays an important role 30]. Moreover, injury to the renal papilla may be a normal part of stone formation 31], so the presence of blood and tissue proteins could well be a part of the formation
of a true stone.

Physicochemical analysis

Previous studies have considered the aggregate protein charge and its potential connection
to specific stone types 7], 10]. In a comparison of 4 different stone types, Canales et al. 7] failed to observe statistically significant differences between the number of acidic-
versus basic-fractionated matrix proteins. In our study (Table 2), the average pI of all proteins associated with the CaOx-Ia stone or the CaOx-Id
stone was similar (6.62 vs. 6.49), as was the average pI of all proteins unique to
CaOx-Ia or CaOx-Id stones (6.75 vs. 6.33). In comparison, a slightly more alkaline
average pI of 7.2 has been observed in uric acid stone matrix proteins 10]. Conclusive comparison of these and other properties requires additional samples
and further study. Nevertheless, negative amino acid/positive amino acid ratios, aliphatic
indices (a measure of protein stability) and GRAVY scores were similar across all
comparisons in the CaOx stones. The mean GRAVY scores, all considerably negative,
indicate overall protein hydrophilicity.

Table 2. Physicochemical characteristics of stone matrix proteomes

Bioinformatic analysis

Of the proteins listed in the Human Kidney Stone Matrix Proteome Database (Additional
file 3), 680 (82 %) have been detected in kidney cells or tissue (per the Human Proteome
Map), 154 (18 %) have not been detected in kidney, so their origin may be considered
“extra-renal”, 144 (17 %) are considered to be moderately or highly abundant in kidney,
248 (30 %) are considered to be in the “extracellular component” (kidney or otherwise),
and 75 (9 %) are considered to be “cytoskeletal and/or structural” proteins.

Proteins identified in the CaOx stones also represented a broad variety of molecular
classes (Table 3). The distribution of classes reflects the prevalence of “cellular proteins” in the
matrix, dominated by cytoskeletal associated proteins, structural proteins, transport/cargo
proteins, chaperone/heat shock proteins, and ribonucleoprotein/RNA binding proteins,
rather than an excess of urine- or plasma-related proteins. Many of these proteins
may stem from the considerable cellular components of stone matrices first observed
by Boyce in 1956 32]. Nevertheless, as Table 1 and Additional file 1 indicate, the individually most abundant proteins in the CaOx stone matrix are blood/plasma
derived – and presumably urine – proteins.

Table 3. Representation of CaOx stone proteins by molecular class (?5)

The following calcium binding proteins were detected: annexins A1, A2, A3, A4, A5,
A9, A10, and A11; calcyphosin; calmodulin; calsequestrin-2; cilaggrin-2; cucleobindin-1;
osteopontin; Profilaggrin; Protein S100-A2 (S100 calcium-binding protein A2); Protein
S100-A6 (Calcyclin); Protein S100-A7 (Psoriasin); Protein S100-A8 (Calgranulin-A);
Protein S100-A9 (Calgranulin-B); Protein S100-P (Migration-inducing gene 9 protein);
Protein S100-A11 (Calgizzarin); and Protein S100-A12 (Calgranulin-C), and these may
have implications in the mineralization process 33], 34]. Additionally, the following urinary proteins known to have the potential to modulate
crystal formation and retention 35]–37] were identified and quantified as prominent constituents of the stone matrix: Tamm-Horsfall
protein; osteopontin; ?-1 microglobulin; calprotectin (protein S100-A8 9); serum
albumin; prothrombin; inter-? trypsin inhibitor (heavy chains H1, H2, and 4); heparin
sulphate proteoglycan; bikunin; CD44, fetuin, and various collagens.

As in two previous studies 10], 14] where inflammatory, coagulation, cell adhesion, and acute-phase response pathways
were directly related to high abundance matrix proteins, we used pathway analysis
to predict with statistical confidence which pathways might be associated with the
matrix proteins identified and quantified in our CaOx stones. These results (79 unique
pathways) are presented in the Pathway Data found in Additional file 4. Some of the most statistically significant pathways included LXR/RXR activation,
coagulation system, acute phase response signaling, FXR/RXR activation, clathrin-mediated
endocytosis signaling, intrinsic prothrombin activation pathway, epithelial adherens
junction signaling/remodeling, extrinsic prothrombin activation, complement system,
and the production of nitric oxide and reactive oxygen species in macrophages. Analogous
observations were made via functional annotation clustering of the CaOx proteins presented
in the Functional Annotation Clusters found in Additional file 5, where inflammatory response and immune related functions were most notable and cytoskeletal
structural molecule activity, extracellular glycoprotein signaling, wound healing,
coagulation, and regulation of body fluid levels annotations were significantly represented.