Urine Proteomics Reveals Sex-Specific Response to Total Pancreatectomy With Islet Autotransplantation.

Urine samples were obtained from 22 participants pre- and post-TPIAT, resulting in 44 paired samples. The samples were analyzed by LC-MS/MS. Five samples were found to have less than 30% of the cumulative list of proteins (<743 proteins) and so these samples and their respective matching sample were thus removed from our dataset. This resulted in a complete dataset with 17 matched pre- and post-TPIAT sample pairs ( Table 1 ). Of these 17 participants, following surgery seven were classified as having full graft function, nine were classified as partial graft function, and one had islet graft failure. Additionally, 13 of the 17 participants were female and 4 were male, reflective of the fact that nearly 75% of the adults TPIAT patients are females. 14 Subjecting these samples to quantitative mass-spectrometry based proteomics identified over 2400 urinary proteins. To ensure high quality protein data, we applied additional filtering and included only proteins that were identified in at least 12 of the 17 the paired urine samples. This filtering resulted in 1106 quantified proteins ( Supplemental Table 1 ). To exclude potential batching and to investigate the largest variances in the dataset, we conducted an unsupervised principal component analysis (PCA) ( Fig. 1A ). Approximately half of the variance observed in all the samples were characterized by principal components (PC) 1 and 2. The urine samples all showed a clear pre-post TPIAT trajectory towards lower PC1 values except for two samples (see * on Fig. 1A ), with significantly larger differences for the samples from the female than the male participants. Additionally, the samples from all but one male (see + on Fig. 1A ) demonstrated distinct clustering with minor changes post-TPIAT. The scores plot showed a distinct clustering of samples based on sex, with a clear separation of the female samples from male samples pre-TPIAT which disappears post-TPIAT with the female samples more closely resembling the male samples ( Figs 1B , C ). Additionally, protein correlation matrices investigating protein change before and after TPIAT in the female and male cohorts ( Supplemental Fig. 1 ) show that the TPIAT related changes are more pronounced and distinct within the female cohort, compared to the male cohort which shows minimal changes of the urinary proteome following TPIAT. Focusing our analysis on proteins with a specific abundance in the samples from female or male participants, we identified 60 proteins with changed abundance comparing males to females in the pre-TPIAT samples ( Supplemental Fig. 2 and Supplemental Table 2A – C ), with pro-inflammatory and cell-cell adhesion molecules increased in females and anti-inflammatory proteins increased in males ( Fig. 1D ). However, we found only 9 proteins with a significant changed abundance between males and females in the post-TPIAT samples. These findings are consistent with the PCA in Figure 1 and the protein-protein correlation analysis ( Supplemental Fig. 1 ), demonstrating that the urinary proteomes from males and females are much more similar post-TPIAT. Analyzing the effect of TPIAT with respect to sex, we plotted the abundances of the 60 sex-specific proteins pre- and post-TPIAT for males and females, respectively ( Supplemental Fig. 3 ). The linear regression of the sex-specific proteins in the male cohort (male y = 0.92x + 1.8) demonstrated that the proteins are relatively stable for the samples post- vs pre-TPIAT with a slope close to a 1 and a minimal intercept. In contrast, the female cohort showed distinct differences post- vs pre-TPIAT as apparent from a slope and intercept deviating from 1 and 0, respectively (female y = 0.76x + 7.8). Effectively, this demonstrates a distinct impact of TPIAT on males and females. To identify urinary proteins with altered abundance post- vs. pre-TPIAT, we performed a sex-independent statistical analysis encompassing all 13 sample pairs (excluding samples with CKD). We identified 301 urinary proteins with a statistically significant induced change of abundance post- vs pre-TPIAT (FDR <0.05). Most of these proteins were related to cell-cell adhesion, stress response, glycolysis, adaptive immune response, and intracellular vacuoles ( Fig. 2D ). Proteins related to cell-cell adhesion, stress response, and glycolysis were mostly increased in the pre-TPIAT samples whereas proteins related to upregulation of the adaptive immune response and intracellular vacuoles were increased post-TPIAT. Additionally, due to the observed sex differences and independent trajectories post- vs pre-TPIAT, we decided to analyze the male and female TPIAT patients independently ( Fig. 2 ). In the urine proteomics dataset from the male cohort, not a single protein passed the significance threshold for post- vs pre-TIPAT ( Fig. 2A ). However, in the female cohort 434 urinary proteins showed statistically significant TPIAT-induced abundance differences ( Fig. 2B , Supplemental Table 3 ): 169 with increased and 265 with decreased abundances post-TPIAT. To investigate whether the lack of significant proteins in the smaller male cohort was due to lower statistical power, we correlated the fold-changes of all the significant proteins in the female samples to the fold-changes of the corresponding (non-significant) proteins in the male samples to determine if the fold changes correlated irrespective of the statistical significance. We found poor correlation (R = 0.22) between the fold changes of all the significant proteins in the female cohort to the corresponding fold changes in the male cohort ( Fig. 2C ), suggesting that the lower number of significant proteins in the male cohort was not explained by the smaller cohort size. Upon studying the clinical information for the two outlier samples from the scores plot (see * on Fig. 1 ), we found that these participants suffered from CKD based on eGFR prior to TPIAT. We found two additional subjects with CKD based on eGFR (n = 4). To investigate the effect of early-stage CKD on the urinary proteome, we identified urinary proteins with altered abundance in CKD samples compared to samples without CKD using the post- vs pre-TPIAT fold change of all urinary proteins. In the CKD samples, we identified 22 proteins with statistically significant change of abundance (FDR <0.05) in comparison to samples without CKD ( Fig. 3A ). Specifically, we found that the CKD samples had upregulation of proteins related to oxidative stress, glycolysis, and exocytosis ( Fig. 3B ). Additionally, the scores plot identified another marked outlier (see + on Fig. 1 ), who was the only urine sample from the male cohort that showed vast differences pre- versus post-TPIAT. Upon further investigation, this patient was found to have developed severe immunoglobulin A (IgA) nephropathy two years after TPIAT. Accordingly, the post-TPIAT sample from this patient had the highest normalized intensity, which is a proxy for urinary concentration, of IgA1 protein compared to the rest of the male samples. To investigate any potential early markers of IgA nephropathy, we included data from a study by Prikryl et al 37 comparing urine from IgA nephropathy patients to healthy controls. In the study, Prikryl et al identified 30 urinary proteins with significantly altered abundances (21 more and 9 less abundant proteins) in patients with IgA nephropathy compared to healthy controls, 37 with all 30 proteins detected in our study. We then compared the effect of these proteins on the variance observed in our loadings plot, investigating if the variation of the IgA nephropathy patient observed in PCA score (* in Fig. 4A ) was due to the 30 biomarkers found by Prikryl et al. Comparing the PC1 PCA loading plot, we found that the 21 more and 9 less abundant proteins had statistically significant different average PC1-coordinate value (combined P = 7.79 × 10 −3 ; more abundant proteins PC1 average coordinate: P = 2.83 × 10 −2 ; less abundant proteins PC1 average coordinate: P = −9.89 × 10 −3 ) with the proteins clustering into two unique groups based on directionality as determined by Prikryl et al ( Fig. 4B blue and red circles). Thus, suggesting that the 30 IgA nephropathy biomarkers were significantly contributing to the variance observed within our dataset from the patient with IgA nephropathy. Additionally, we found that those proteins upregulated in IgA nephropathy (including the 21 proteins from Prikryl et al combined with our proposed 10 biomarker candidates) were found to be involved in the humoral immune response, acute inflammatory response, glomerular filtration, and phagocytosis ( Fig. 4C ).

At the University of Minnesota Medical Center, samples were collected from 22 patients with CP prospectively enrolled in a TPIAT follow up study between September 2011 and February 2014. 20 The study protocol was reviewed and approved by the University of Minnesota Institutional Review Board (IRB) (protocol #0609M91887) and patient consent or parental consent and patient assent was obtained as age appropriate for all participants. Urine samples were collected within the week prior to TPIAT and 12–18 months post-surgery, yielding a total of 44 samples. Clinical information for each participant was obtained, including baseline demographics, pre- and post-operative kidney function, any history of diagnosed renal disease including chronic kidney disease or renal stones, islet cell yield, and islet cell graft function at 1-year post-TPIAT. Graft functions were defined as: 1) full graft (insulin independent at 1-year post-TPIAT), 2) partial graft (insulin dependence at 1-year post-TPIAT with C-peptide >0.6 ng/dl), and 3) failed graft (insulin dependence at 1-year post-TPIAT with C-peptide <0.6 ng/dl). In addition, creatinine samples were obtained at time of morning fasting metabolic blood work before and after TPIAT. Using this creatinine measurement, we retrospectively defined chronic kidney disease (CKD) as estimated glomerular filtration rate (eGFR) <90 mL/min/1.73m2. Urine samples and clinical history were collected under an IRB approved study at the University of Minnesota (IRB# 0609M91887). Patient consent or parental consent as age appropriate was obtained from all participants. Deidentified data and urine samples were then analyzed at Harvard Medical School under an IRB exempt protocol (#2018H0429). The urine samples were prepared for proteome analysis using the in-house developed MStern blotting approach as previously described. 18 , 21 Briefly explained, 150 μL urine (nominally ~15 μg protein) was mixed with dry urea and ammonium bicarbonate (ABC) to a final concentration of 8 M urea, 50 mM ABC pH 8.5 in a 96-well plate. Protein cysteine disulfide bonds were reduced by addition of dithiothreitol to a final concentration of 10 mM in 50 mM ABC and incubation for 20 min. Reduced disulfide bonds were alkylated by addition of iodoacetamide to a final concentration of 50 mM and incubation for 20 min. From each sample, a volume corresponding to nominally 15 μg of reduced and alkylated protein was transferred to a 96-well hydrophobic polyvinylidene difluoride (PVDF) membrane plate (MSIPS4510, Millipore, Burlington, Mass), which had been primed with 70% ethanol and conditioned with 8 M urea in 50 mM ABC pH 8.5. A 96-well plate adaptable vacuum manifold (Millipore) was used to apply a vacuum and facilitate liquid transfer through the PVDF membrane plate. After adsorption of the proteins onto the membrane, the membranes were washed with 50 mM ABC. Protein digestion was performed by adding 100 μg of 10 μg/mL sequencing grade modified trypsin (V5111, Promega, Madison, Wis) diluted in 5% acetonitrile, 50 mM ABC (nominal trypsin to protein ratio 1:15) and incubated for 2 hours at 37°C in a humidified incubator. The peptides were eluted from the PVDF membrane plate with 40% acetonitrile (ACN), 0.1% formic acid (FA). Subsequently, the elution solutions were pooled and dried in a vacuum concentrator. The dry urine product was stored at −20°C until liquid chromatography–mass spectrometry (LC-MS) analysis. The samples were analyzed using a microfluidic LC chip system coupled online to a high resolution/high accuracy Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, Mass). The samples were resuspended in 50 μL 5% ACN, 5% FA, and 3 μL was loaded onto a reversed-phase C18 analytical column (PicoChip 150 μm x 15 cm [New Objective, Littleton, Mass]) with 2 μL solvent A (0.1% FA). The peptides were eluted from the column by increasing the ratio of solvent B (0.1% FA in ACN) on a 70 min linear gradient from 7% buffer B (0.1% FA in ACN) to 25% buffer B, at a flow rate of 1000 nL/min a flowrate. The PicoChip with an integrated emitter was kept at 50°C and mounted directly in front of the inlet of the heated capillary of the Q Exactive mass spectrometer, which was operated in positive mode. The mass spectrometer was operated in data-dependent TOP10 (DDA) mode with the following settings: mass range 400–1000 Th; resolution for MS1 scan 70,000 @ 200 Th; lock mass: 445.120025 Th; resolution for MS2 scan 17,500 @ 200 Th; isolation width 1.6 Th; NCE 27; underfill ratio 1%; charge state exclusion: unassigned, 1, >6; dynamic exclusion 30 s. The urine raw files were searched with MaxQuant (v1.5.5.1, Max-Planck Institute for Biochemistry, Computational Systems Biochemistry, Martinsried, Germany) against a database containing the reviewed UniProt human reference proteome (downloaded: 2016-08-21; # of protein entries: 20,209). 22 , 23 Standard settings were employed in MaxQuant, with the match-between-runs feature enabled. A maximum of three tryptic missed cleavages were allowed, as recommended for the MStern protocol. Additionally, the following modifications were found to be abundant with the applied protocol, and were included in the search: carbamidomethylated cysteine residues (fixed), acetylation of proteins N-terminals (variable), oxidation of methionine (variable), and deamidation of asparagine and glutamine residues (variable). 24 , 25 Identified proteins and peptides were filtered to <1% false discovery rate (FDR) using the forward/reverse database search strategy in MaxQuant. The mass spectrometry raw data and protein databases have been deposited to the ProteomeXchange Consortium ( http://proteomecentral.proteomexchange.org ) via the PRIDE partner repository with data set identifier PXD028570. 26 , 27 To increase the quantitation accuracy, we employed additional filtering of the ≤1% FDR identified proteins in Perseus (v1.5.5.3, Max-Planck Institute for Biochemistry). 28 , 29 We removed samples (and the corresponding paired sample) with less than 30% of the cumulated list of proteins identified (<743 proteins) from further analysis. Additionally, proteins quantified with only one peptide unique to a protein group, and proteins with less than 70% valid values in either the pre or post condition were removed. Of note, the applied filtering strategy could potentially remove proteins with a specific abundance between males and female subjects. Therefore, we conducted an additional analysis of proteins with a sex-specific abundance, using a dataset filtered for proteins with less than 70% valid values in the samples from males or females. However, the lists of significant proteins between males and females were nearly identical regardless of the applied filtering scheme. Proteins with a statistically significant change of abundance were identified by paired two-tailed t-tests, corrected for multiple hypothesis testing by permutation-based FDR control using standard parameters in Perseus (250 randomizations, FDR <0.05, s0 = 0.1). Protein changes with FDR (q-value) <0.05 were considered statistically significant. For the purpose of conducting Principal Component Analysis, missing values were replaced with values from a normal distribution (width 0.3 and down shift 1.8), to simulate signals from low-abundance proteins. 30 To identify underlying biological themes and pathways of differentially expressed proteins, we performed a pathway enrichment analysis using both the Database for Annotation, Visualization and Integrated Discovery (DAVID) database and ClueGo plugin of Cytoscape (v3.8.2, The Cytoscape Consortium, San Diego Calif). 31 – 34 The urinary dataset constitutes by itself a sub-proteome, enriched in selected categories and pathways. To correct for this bias, we used all quantifiable urinary proteins in the present study as a background in the calculations. Additionally, we performed Pearson’s correlation analyses of the LFQ value differences post- vs pre-TPIAT in Rstudio (v1.3, RStudio PBC, Boston, Mass), using a previously developed script. 35 , 36 To ensure reliable correlations we required >50% valid values in both post- and pre-TPIAT samples ensuring at least 7 data points for the correlations, in addition to the mentioned filtering of quantifiable proteins.

In this study, we performed a hypothesis-generating proteomic analysis on urine from participants before and after TPIAT. We report novel findings including highly sex-specific changes that occur with TPIAT, showing pronounced sex differences in the urinary proteome before TPIAT but minimal differences following TPIAT, where after surgery the proteins with a sex-specific abundance in the female urine more closely resembled that of the male urine. We also found that following TPIAT, there was downregulation of inflammatory and adhesion proteins but upregulation of immunoglobulins and cathepsins. Finally, we were able to identify several tentative biomarkers which may be useful for early detection of IgA nephropathy. Overall, our data can provide a basis of the pathophysiology of sex differences observed in the patients undergoing TPIAT and potentially extended to the development of a more personalized selection of the TPIAT applicants, leading to the improved outcomes of the graft function.

Over the past two decades, TPIAT has been increasingly performed for refractory pain in the setting of CP. 38 However, the systemic molecular response to TPIAT remain poorly understood. We performed the first hypothesis-generating proteomics study of the molecular impact of TPIAT on the urinary proteome. We identified 434 urinary proteins with significantly changed protein abundance in the female cohort, of which 265 proteins were upregulated pre-TPIAT and 169 proteins were upregulated post-TPIAT. Our analysis demonstrated several novel findings, including (1) the response to TPIAT is highly sex-specific and these sex differences were diminished post-TPIAT, (2) the urine proteomic changes observed are dependent on kidney function, and (3) TPIAT specific changes of the urine proteome includes decreased abundance of urinary proinflammatory proteins. According to our analysis, the female and male samples showed clear separation based on principal component analysis ( Fig. 1 ), indicating i) clear sex-dependent differences in the urinary proteomes prior to TPIAT, and ii) clear changes between the female and male urinary composition in response to TPIAT. When examining sex differences with respect to TPIAT status, we found significant sex differences in the baseline pre-surgical samples. Our finding of the significant differences in the urinary proteome of females and males pre-TPIAT could represent sex specific responses to CP or normal variations of the proteomic profile present at baseline. Prior studies have demonstrated sex-specific differences in the proteomic profiles of healthy females compared to males. 39 , 40 In this context, it is most likely that the significant sex specific differences that we observed were due to normal variations of the proteomic profiles between sexes. However, it does appear that the response to TPIAT is highly sex dependent as we found that females had significant changes in their urinary proteome upon TPIAT, however males showed almost no TPIAT-induced change in their urinary proteome. While our study cohort was skewed, containing three times more female participants than males, these sex-specific changes did not appear to be due to smaller number of male participants as there was poor correlation in the TPIAT-induced protein fold change between the females and males ( Fig. 2C ). Interestingly, we identified that these sex differences were diminished following TPIAT. Based on the PCA plot as well as our linear regression analysis of the abundance of sex-specific proteins ( Fig. 1 and Supplemental Fig. 3 ), we identified a distinct difference in the females with TPIAT and that the abundance of sex-specific proteins decreased post-TPIAT to more closely resemble the urinary proteome of males. While several studies have identified sex differences in the proteomic profile of various bodily fluids, 39 – 41 our study identifies that these sex differences can diminish following TPIAT ( Fig. 1 ). These novel findings are not clearly understood, highlighting the need for future studies to investigate potential causes of why these changes occur. From a clinical perspective, one of the major differences after TPIAT is the removal of the chronically inflamed pancreas. To better understand the effect of removing the source of inflammation on the urinary proteome, we analyzed the effect that TPIAT has on different immune system- and inflammation-related proteins. We found that there were clear distinctions in the proteomic profile before compared to after TPIAT. Prior to surgery, there was significant upregulation of proteins related to cell-cell adhesion and proinflammatory proteins. Specifically, there were increased concentrations of actin proteins (actin-1, actin-related protein 2 and 3, and alpha-actinin-4), desmosome proteins (desmoglein 3, desmoplakin, envoplakin, junction plakoglobin, and plakophilin 1 and 3), S100 proteins, interleukin-18, annexin 1, extracellular matrix protein 1, and toll interacting protein in the pre-TPIAT samples. However, following surgery there were increased concentrations of proteins involved in the adaptive immune response and intracellular vacuoles, including cathepsin proteins, immunoglobulins, and complement proteins. The reduced concentrations of proinflammatory proteins after TPIAT likely reflects the removal of the inflamed pancreas following the surgery. Conversely, we observed decreased concentrations in cell-cell adhesion modules as well as the increased concentrations of immunoglobulins and cathepsin proteins in the post-TPIAT samples. We have previously reported a similar increase of most immunoglobulins in the serum post TPIAT. 18 While the mechanism of this is unclear, one hypothesis is that this may represent glomerular damage that occurs after the surgery. These post-TPIAT changes are significantly more pronounced in female patients. Prior studies have demonstrated increased urinary concentrations of immunoglobulins complement proteins, and cathepsins in diabetic nephropathy, raising concern that the observed findings could be from early stage diabetic renal disease. 42 – 46 While diabetic nephropathy was historically defined as proteinuria >300 mg/day, this definition is limited by the fact that significant glomerular damage has occurred by the time a patient develops overt albuminuria. Indeed, studies have demonstrated that some maladaptive changes have occurred within the kidney even at the time of diagnosis. 47 However, it should be noted that our findings of upregulation of proteins suggestive of glomerular damage post-TPIAT have several limitations. These include the fact that these samples were collected at only 12–18 months after surgery, that all participants were non-diabetic before TPIAT, and that the post-TPIAT HbA1c was <7% in all but one participant which indicates normoglycemia or near normoglycemia. The application of urinary proteomics for kidney diseases has long been established, as it has been shown to improve diagnostic accuracy and lead to early diagnosis for various kidney diseases. 48 – 50 Most available studies have focused on case-control analysis of the urinary proteome of patients with CKD to healthy controls or cohort studies in patients with CKD or diabetes mellitus. 51 Additionally, prior studies have limited their analysis of the urinary proteome on patient’s who carry a clinical diagnosis of CKD, potentially restricting the discovery of early biomarkers of kidney disease. 52 Given our finding of increased concentrations of proteins involved with glomerular damage post-TPIAT, we decided to compare the urinary proteome in CP patients with and without CKD based on pre-operative eGFR. Although many patients in our cohort had related renal disorders (including renal stones, renal cysts, and history of acute kidney injury), none of our study participants carried an official diagnosis of CKD. This may be because these participants had such mild alterations in their creatinine that were not clinically apparent. Despite lacking a formal diagnosis of CKD, we identified four participants who met criteria for early-stage CKD based on pre-operative eGFR. Comparing the urine proteome of the patients with early-stage CKD to those without, we demonstrated that the proteomic profile of patients with CKD is readily distinguishable from patients without CKD. Patients with CKD were found to have increased concentrations of proteins related to oxidative stress and glycolysis. Specific examples include peroxiredoxin (PRDX-1, PRDX-2, and PRDX-6), glutathione S-transferase pi 1 (GSTP1), and phosphoglycerate mutase 1 (PGAM1). These findings align with and expand upon prior studies, which have shown elevation of these proteins in different bodily fluids. 53 – 55 Upregulation of oxidative stress mediators as well as glycolysis proteins in patients with CKD may provide additional insight into the molecular changes that occur within the early stage of the disease. Through our proteomic analysis we also discovered one patient who was diagnosed with IgA nephropathy approximately 2 years after sample collection. Analyzing the concentration of previously identified biomarkers for IgA nephropathy within our data, we found that these biomarkers contributed significantly to the variance observed within our dataset from the patient with IgA nephropathy. This finding suggests the ability to detect IgA nephropathy in urine samples obtained at least two years prior to clinical diagnosis. Using this knowledge, we constructed a list of possible very early onset IgA nephropathy urinary biomarkers, including pancreatic secretory trypsin inhibitor, CD27 antigen, neutrophil defensin, and ganglioside GM2 activator, to mention a few ( Supplemental Table 4 ). As this list is based on one patient only, additional studies are needed to validate these markers, but our findings clearly demonstrate the feasibility of early detection of IgA nephropathy. There are several limitations to the present study that must be acknowledged. First, the study took place at a single center and as such samples were collected from a single institution, so results may not be generalizable. Second, given the relatively small sample size, the etiology of CP varied among study participants and may have impacted our results. Lastly, the sample size of the present study was relatively small and further studies with larger cohorts of patients to validate our findings may be needed.

Chronic pancreatitis (CP) is a progressive fibroinflammatory disease of the pancreas, in which there is a persistent injury to pancreatic parenchyma and eventual destruction of acinar and islet cells, leading to exocrine and endocrine pancreatic insufficiency. 1 There are various genetic, environmental, and other risk factors that are involved in the pathogenesis as well as variations in the natural history of CP. 1 Historically, alcohol use was thought to be the primary etiology of CP, however it is now clear that underlying genetic mutations and smoking are also important factors in the development of CP. 2 – 6 Chronic pancreatitis is estimated to affect approximately 50 per 100,000 individuals, with recent studies indicating an increasing incidence over the past decade. 7 , 8 Additionally, CP also represents a significant cause of morbidity and financial burden in the United States, with estimated annual costs of $638 million. 9 , 10 Patients with CP often suffer severe abdominal pain that can be difficult to treat and ultimately impact patients’ quality of life. 5 , 11 The management of pancreatitis-related pain includes pancreatic enzyme replacement therapy, analgesia, and more invasive options for those with refractory pain, such as endoscopic retrograde cholangiopancreatography (ERCP), nerve block procedures, and surgical decompression via pancreaticojejunostomy. 12 However, for 25% of patients undergoing such surgery, pain relief is temporary at best. 13 , 14 For patients with refractory pain, total pancreatectomy (TP) may be an option. Even though the procedure removes the fundamental cause of the pain, TP immediately results in iatrogenic type 1 diabetes due to a lack of endocrine pancreatic insulin secretion. The resulting diabetes can be quite difficult to manage and can be associated with significant morbidity. 15 Autologous islet cell transplantation has become a promising treatment for those patients undergoing TP. Total pancreatectomy with islet autotransplantation (TPIAT) involves isolating the islets from the resected pancreas and infusing them back into the liver where they remain while maintaining their functionality, ie, the natural responsiveness to blood sugar levels and insulin secretory capacity, which can reverse or reduce the risk for post-operative diabetes associated with TP. The first TPIAT surgery was performed at the University of Minnesota in 1977, and as of 2018, approximately 1000 patients have received the treatment. 16 , 17 A study of TPIAT patients from the period 1977 to 2011 reported that patient-survival one (five) year post-TPIAT was 96% (89%) in adults and 98% (98%) in children. Additionally, 90% of the patients had post-TPIAT C-peptide levels greater than 0.6 ng/mL blood, indicating a successful procedure and preserved insulin-producing capacities. Three years post-TPIAT, 30% were still insulin-independent (25% in adults and 55% in children) and an additional 33% retained partial insulin production function (35% in adults and 25% in children), strongly complementing the treatment of type-1 diabetes. Additionally, post-TPIAT, 85% of the patients had an improvement or absence of pain, making the procedure highly attractive as a last-resort treatment option for CP. 17 Most TPIAT-related publications so far have focused on the clinical aspects, likely due to the scarcity and nature of TPIAT. The systemic molecular response to TPIAT thereby remains poorly investigated. We recently published the first hypothesis-generating study of the effect of TPIAT on the serum proteome, where we demonstrated that several serum proteins that are known to be altered in various pancreatic diseases, including CP, return to normal levels post-TPIAT. 18 In the present study, we sought to investigate the effect of TPIAT on the urine proteome. Urine is an ultrafiltrate of blood from the kidneys and reflects changes in the entire body. Metabolites, proteins, and peptides are concentrated in urine by the kidneys, yet the urine proteome is significantly more manageable compared to the serum- and plasma proteome due to a reduced concentration range. 18 , 19 Our urine proteomics approach captured 1) sex-specific changes following TPIAT, 2) increased urinary concentrations of immunoglobulins, complement proteins, and cathepsins in post-TPIAT samples indicating glomerular damage, and 3) decreased proinflammatory protein abundances in post-TPIAT indicating reduced inflammation. Taken together, the urinary proteomics approach enabled us to capture the overall landscape of the biological mechanisms affected in CP and the impact of TPIAT.

Supplemental Figure S1. Correlation matrices for the urinary proteins, ordered based on the clustering of the pre-TPIAT females. The strong correlations seen in the pre-TPIAT female samples is not preserved in the post-TPIAT samples, nor found in the samples from the males. The figure highlights that the protein-protein correlations in the female samples pre-TPIAT are diminished post-TPIAT, and that the clustering is distinct to the females. Supplemental Figure S2. Analysis of sex-specific proteins in TPIAT participants. The log-transformed student’s t-test p value of each protein is plotted against the log-transformed fold change. Statistically significant proteins (FDR <0.05) highlighted with proteins significantly increased in females labeled in red ( ) and proteins significantly increased in males in blue ( ). Supplemental Figure S3. Scatterplot with linear regression of the average abundance of all sex-specific proteins post- vs. pre-TPIAT for males (blue) and females (red), respectively. Ribbon indicates regression confidence interval (0.95). Supplementary Table S1: List of quantifiable urinary proteins. Supplementary Table S2A-C: List of proteins with a sex-specific abundance in the (A) pre-, (B) post-TPIAT, and (C) healthy samples. Supplementary Table S3: List of proteins with a significant change of abundance (q-value<0.05) post- vs. pre-TPIAT in Females. Supplementary Table S4: List of tentative very early onset IgA nephropathy urinary biomarkers.