Two independent proteomic approaches provide a comprehensive analysis of the synovial fluid proteome response to Autologous Chondrocyte Implantation

Abstract

BackgroundAutologous chondrocyte implantation (ACI) has a failure rate of approximately 20%, but it is yet to be fully understood why. Biomarkers are needed that can pre-operatively predict in which patients it is likely to fail, so that alternative or individualised therapies can be offered. We previously used label-free quantitation (LF) with a dynamic range compression proteomic approach to assess the synovial fluid (SF) of ACI responders and non-responders. However, we were able to identify only a few differentially abundant proteins at baseline. In the present study, we built upon these previous findings by assessing higher-abundance proteins within this SF, providing a more global proteomic analysis on the basis of which more of the biology underlying ACI success or failure can be understood.MethodsIsobaric tagging for relative and absolute quantitation (iTRAQ) proteomic analysis was used to assess SF from ACI responders (mean Lysholm improvement of 33; n = 14) and non-responders (mean Lysholm decrease of 14; n = 13) at the two stages of surgery (cartilage harvest and chondrocyte implantation). Differentially abundant proteins in iTRAQ and combined iTRAQ and LF datasets were investigated using pathway and network analyses.ResultsiTRAQ proteomic analysis confirmed our previous finding that there is a marked proteomic shift in response to cartilage harvest (70 and 54 proteins demonstrating ≥ 2.0-fold change and p < 0.05 between stages I and II in responders and non-responders, respectively). Further, it highlighted 28 proteins that were differentially abundant between responders and non-responders to ACI, which were not found in the LF study, 16 of which were altered at baseline. The differential expression of two proteins (complement C1s subcomponent and matrix metalloproteinase 3) was confirmed biochemically. Combination of the iTRAQ and LF proteomic datasets generated in-depth SF proteome information that was used to generate interactome networks representing ACI success or failure. Functional pathways that are dysregulated in ACI non-responders were identified, including acute-phase response signalling.ConclusionsSeveral candidate biomarkers for baseline prediction of ACI outcome were identified. A holistic overview of the SF proteome in responders and non-responders to ACI has been profiled, providing a better understanding of the biological pathways underlying clinical outcome, particularly the differential response to cartilage harvest in non-responders.Electronic supplementary materialThe online version of this article (10.1186/s13075-018-1573-4) contains supplementary material, which is available to authorized users.

Increased synovitis and pro-inflammatory macrophage abundance are observed in the synovia of patients at risk of developing post-traumatic OA compared to those with established OA

“…The study parameters were then compared between patients within the early-OA group who had (n ​= ​9) and had not (n ​= ​6) undergone a cartilage harvest as part of their treatment ( supplementary Table 4 ), as this procedure has been shown to produce an inflammatory response in a subset of patients [ [29] , [30] , [31] ]. There was no difference in age, BMI or disease severity (WORMS) between groups, as expected due to the randomisation of patients to the treatment groups ( supplementary Table 4 ).…”

“…As part of their cell therapy treatment, some of the early-OA patients underwent a cartilage biopsy. It has been previously reported that a proportion of patients who undergo a cartilage harvest as part of their treatment have a marked proinflammatory and catabolic response to the procedure, detectable in their SF [ [29] , [30] , [31] ]. However, there was no difference in synovitis prevalence or severity in the cartilage harvest subgroup of early-OA patients, and therefore the procedure had no impact on synovitis.…”

Increased synovitis and pro-inflammatory macrophage abundance are observed in the synovia of patients at risk of developing post-traumatic OA compared to those with established OA

“…The study parameters were then compared between patients within the early-OA group who had (n ​= ​9) and had not (n ​= ​6) undergone a cartilage harvest as part of their treatment ( supplementary Table 4 ), as this procedure has been shown to produce an inflammatory response in a subset of patients [ [29] , [30] , [31] ]. There was no difference in age, BMI or disease severity (WORMS) between groups, as expected due to the randomisation of patients to the treatment groups ( supplementary Table 4 ).…”

“…As part of their cell therapy treatment, some of the early-OA patients underwent a cartilage biopsy. It has been previously reported that a proportion of patients who undergo a cartilage harvest as part of their treatment have a marked proinflammatory and catabolic response to the procedure, detectable in their SF [ [29] , [30] , [31] ]. However, there was no difference in synovitis prevalence or severity in the cartilage harvest subgroup of early-OA patients, and therefore the procedure had no impact on synovitis.…”

Proteomic Analyses of Autologous Chondrocyte Implantation Plasma Highlight Cartilage Acidic Protein 1 as a Candidate for Preclinical Screening

“…After the iTRAQ analysis, we were encouraged that the greatest number of protein differences were identified in response to cartilage harvest surgery (ie, at stage 2 vs stage 1) in the ACI nonresponders, as this replicated our previous findings in the synovial fluid of these same patients. 11,12 Moreover, several of these proteins related to acute-phase response signaling, a key biological pathway that we have associated with the synovial fluid of nonresponders, who have a poor clinical outcome to the treatment. 11 To examine this finding in more depth, we undertook an independent analysis of plasma samples (some matched to the iTRAQ analysis) with LF LC-MS/ MS to analyze each sample individually.…”

“…Bead-bound proteins were washed and then treated with 0.05% (wt/vol) RapiGest (Waters) in 25mM ammonium bicarbonate for 10 minutes at 80°C before reduction, alkylation, in situ protein digestion without removal of the beads, and acidification of trifluoroacetic acid, as described previously. 12 , 23 The peptide-containing supernatant fractions were frozen at −80° before LC-MS/MS.…”

Identification of Candidate Synovial Fluid Biomarkers for the Prediction of Patient Outcome After Microfracture or Osteotomy

“…The second, targeted approach was to assess a panel of SF and plasma proteins that we routinely test and that have been previously associated with OA 32,45 or that we have identified as changed in response to ACI. 11,12 This method meant that any candidate proteins identified from the proteomic analysis could be assessed in a larger cohort to determine whether they added any predictive value over and above other more commonly assessed OA proteins, baseline OA severity (as deemed by Kellgren-Lawrence score), and patient demographic characteristics, together contributing to the development of predictive models for patient outcome to microfracture or osteotomy surgery. Further, particularly for the microfracture cohort, there was a wide range of patient ages, which in itself could account for differential abundance of some of the measured proteins (eg, serum HA 19 ); therefore, this statistical approach could account for collinearity between parameters, such as between age and serum HA.…”

Increased synovitis and pro-inflammatory macrophage abundance are observed in the synovia of patients at risk of developing post-traumatic OA compared to those with established OA

“…The study parameters were then compared between patients within the early-OA group who had (n ​= ​9) and had not (n ​= ​6) undergone a cartilage harvest as part of their treatment ( supplementary Table 4 ), as this procedure has been shown to produce an inflammatory response in a subset of patients [ [29] , [30] , [31] ]. There was no difference in age, BMI or disease severity (WORMS) between groups, as expected due to the randomisation of patients to the treatment groups ( supplementary Table 4 ).…”

“…As part of their cell therapy treatment, some of the early-OA patients underwent a cartilage biopsy. It has been previously reported that a proportion of patients who undergo a cartilage harvest as part of their treatment have a marked proinflammatory and catabolic response to the procedure, detectable in their SF [ [29] , [30] , [31] ]. However, there was no difference in synovitis prevalence or severity in the cartilage harvest subgroup of early-OA patients, and therefore the procedure had no impact on synovitis.…”

Proteomic Analyses of Autologous Chondrocyte Implantation Plasma Highlight Cartilage Acidic Protein 1 as a Candidate for Preclinical Screening

“…After the iTRAQ analysis, we were encouraged that the greatest number of protein differences were identified in response to cartilage harvest surgery (ie, at stage 2 vs stage 1) in the ACI nonresponders, as this replicated our previous findings in the synovial fluid of these same patients. 11,12 Moreover, several of these proteins related to acute-phase response signaling, a key biological pathway that we have associated with the synovial fluid of nonresponders, who have a poor clinical outcome to the treatment. 11 To examine this finding in more depth, we undertook an independent analysis of plasma samples (some matched to the iTRAQ analysis) with LF LC-MS/ MS to analyze each sample individually.…”

“…Bead-bound proteins were washed and then treated with 0.05% (wt/vol) RapiGest (Waters) in 25mM ammonium bicarbonate for 10 minutes at 80°C before reduction, alkylation, in situ protein digestion without removal of the beads, and acidification of trifluoroacetic acid, as described previously. 12 , 23 The peptide-containing supernatant fractions were frozen at −80° before LC-MS/MS.…”

Identification of Candidate Synovial Fluid Biomarkers for the Prediction of Patient Outcome After Microfracture or Osteotomy

“…The second, targeted approach was to assess a panel of SF and plasma proteins that we routinely test and that have been previously associated with OA 32,45 or that we have identified as changed in response to ACI. 11,12 This method meant that any candidate proteins identified from the proteomic analysis could be assessed in a larger cohort to determine whether they added any predictive value over and above other more commonly assessed OA proteins, baseline OA severity (as deemed by Kellgren-Lawrence score), and patient demographic characteristics, together contributing to the development of predictive models for patient outcome to microfracture or osteotomy surgery. Further, particularly for the microfracture cohort, there was a wide range of patient ages, which in itself could account for differential abundance of some of the measured proteins (eg, serum HA 19 ); therefore, this statistical approach could account for collinearity between parameters, such as between age and serum HA.…”

Increased synovitis and pro-inflammatory macrophage abundance are observed in the synovia of patients at risk of developing post-traumatic OA compared to those with established OA

“…The study parameters were then compared between patients within the early-OA group who had (n ​= ​9) and had not (n ​= ​6) undergone a cartilage harvest as part of their treatment ( supplementary Table 4 ), as this procedure has been shown to produce an inflammatory response in a subset of patients [ [29] , [30] , [31] ]. There was no difference in age, BMI or disease severity (WORMS) between groups, as expected due to the randomisation of patients to the treatment groups ( supplementary Table 4 ).…”

“…As part of their cell therapy treatment, some of the early-OA patients underwent a cartilage biopsy. It has been previously reported that a proportion of patients who undergo a cartilage harvest as part of their treatment have a marked proinflammatory and catabolic response to the procedure, detectable in their SF [ [29] , [30] , [31] ]. However, there was no difference in synovitis prevalence or severity in the cartilage harvest subgroup of early-OA patients, and therefore the procedure had no impact on synovitis.…”

Proteomic Analyses of Autologous Chondrocyte Implantation Plasma Highlight Cartilage Acidic Protein 1 as a Candidate for Preclinical Screening

“…After the iTRAQ analysis, we were encouraged that the greatest number of protein differences were identified in response to cartilage harvest surgery (ie, at stage 2 vs stage 1) in the ACI nonresponders, as this replicated our previous findings in the synovial fluid of these same patients. 11,12 Moreover, several of these proteins related to acute-phase response signaling, a key biological pathway that we have associated with the synovial fluid of nonresponders, who have a poor clinical outcome to the treatment. 11 To examine this finding in more depth, we undertook an independent analysis of plasma samples (some matched to the iTRAQ analysis) with LF LC-MS/ MS to analyze each sample individually.…”

“…Bead-bound proteins were washed and then treated with 0.05% (wt/vol) RapiGest (Waters) in 25mM ammonium bicarbonate for 10 minutes at 80°C before reduction, alkylation, in situ protein digestion without removal of the beads, and acidification of trifluoroacetic acid, as described previously. 12 , 23 The peptide-containing supernatant fractions were frozen at −80° before LC-MS/MS.…”

Identification of Candidate Synovial Fluid Biomarkers for the Prediction of Patient Outcome After Microfracture or Osteotomy

“…The second, targeted approach was to assess a panel of SF and plasma proteins that we routinely test and that have been previously associated with OA 32,45 or that we have identified as changed in response to ACI. 11,12 This method meant that any candidate proteins identified from the proteomic analysis could be assessed in a larger cohort to determine whether they added any predictive value over and above other more commonly assessed OA proteins, baseline OA severity (as deemed by Kellgren-Lawrence score), and patient demographic characteristics, together contributing to the development of predictive models for patient outcome to microfracture or osteotomy surgery. Further, particularly for the microfracture cohort, there was a wide range of patient ages, which in itself could account for differential abundance of some of the measured proteins (eg, serum HA 19 ); therefore, this statistical approach could account for collinearity between parameters, such as between age and serum HA.…”