A proteomic investigation of protein adsorption to cerebral microdialysis membranes in surgically treated intracerebral hemorrhage patients

Background Cerebral microdialysis (CMD) is a minimally invasive technique for sampling the interstitial fluid in human brain tissue. CMD allows monitoring the metabolic state of tissue, as well as sampling macromolecules such as proteins and peptides. Recovery of proteins or peptides can be hampered by their adsorption to the CMD membrane as has been previously shown in-vitro, however, protein adsorption to CMD membranes has not been characterized following implantation in human brain tissue. Methods In this paper, we describe the pattern of proteins adsorbed to CMD membranes compared to that of the microdialysate and of cerebrospinal fluid (CSF). We retrieved CMD membranes from three surgically treated intracerebral hemorrhage (ICH) patients, and analyzed protein adsorption to the membranes using two-dimensional gel electrophoresis (2-DE) in combination with nano-liquid mass spectrometry. We compared the proteome profile of three compartments; the CMD membrane, the microdialysate and ventricular CSF collected at time of CMD removal. Results We found unique protein patterns for each of the three compartments, particularly in the molecular weight range of 10-35 kDa. Conclusion This study highlights the importance of analyzing the membranes in addition to the microdialysate when using CMD to sample proteins for biomarker investigation.


Introduction
Sampling, detecting and analyzing protein biomarkers associated with ongoing brain injury is of growing interest in clinical neuroscience research, both in the acute setting of e.g. traumatic brain injury or intracerebral hemorrhage (ICH) as well as in chronic neurodegenerative diseases. Such protein biomarkers may help clarify pathophysiological mechanisms of ongoing brain injury, or guide interventions and aid in prevention of further injury.
There are a number of different methods for sampling biomarkers for neuroproteomic analysis. One method is the collection of cerebrospinal fluid (CSF), either from an external ventricular drain (EVD) used to monitor intracranial pressure (ICP) or from a lumbar puncture. CSF has the advantage of being relatively easy to obtain but does not entirely reflect the interstitial fluid (1,2). Another method for sample collection, which does indeed sample the interstitial fluid, is minimally invasive monitoring of brain tissue by cerebral microdialysis (CMD) used since the 1990s for monitoring brain metabolism as part of multimodal monitoring in the neurocritical care setting (3)(4)(5). With the introduction of highmolecular-weight cut-off catheters (100 kDa) it has become possible to sample macromolecules from the interstitial fluid. Thus apart from providing information on brain metabolism, CMD can now also be used for biomarker discovery or determining concentration of specific proteins or peptides of interest.
When using CMD to sample macro molecules from interstitial fluid there are several technical factors to be aware of. The CMD technique involves inserting a thin dialysis catheter consisting of a semipermeable membrane into the brain tissue. The catheter is perfused with fluid which is subsequently collected and analyzed. The collected fluid, referred to as the microdialysate, is a reflection of the interstitial fluid. However, the relative recovery of molecules in the microdialysate is always a fraction of their true tissue concentration and is affected by a variety of factors. These factors include perfusate flow rate and composition; tissue temperature, pH and tortuosity; and protein molecular weight, three dimensional structure and hydrophobicity (6)(7)(8). Furthermore, CMD membrane characteristics such as material, length and diameter, and pore size also affect protein interaction with the membrane surface and thereby recovery (9). Proteins adsorb to solid surfaces (10) and CMD membranes are no exception. This adsorption may prevent proteins from passing over to the microdialysate, or may cause them to pass over in an unpredictable manner. Protein adsorption may also change over time as the surface gets saturated, which in turn also affects the diffusion of other particles or proteins across the membrane,(7) thereby affecting their relative recovery.
Although protein adsorption to microdialysis membranes has been shown in vitro (6,8,11) and in invivo in skin(12) it has not been characterized on microdialysis membranes implanted in human brain tissue.
By exploring which proteins adhere to the membrane the protein expression in the interstitial fluid can be characterized more accurately. If only the microdialysate is analyzed and not the membrane, valuable information may be overlooked and important biomarkers might evade detection or the tissue concentration of them may be greatly underestimated. In this study, we aimed to determine the difference in proteome profile of CSF, cerebral microdialysate, and on the CMD membrane by using two-dimensional gel electrophoresis (2-DE) to separate proteins followed by identification by liquid chromatography tandem mass spectrometry (LC-MS/MS).

Materials And Methods Patients
Patients undergoing acute neurosurgical treatment for ICH followed by subsequent treatment at the neurocritical care (NCC) unit, University Hospital, Linköping were prospectively recruited to this observational study.
Patients received dual microdialysis catheters with a molecular weight cut-off of 100 kDa, and membrane length 10 mm (71 High Cut-off Brain MD Catheter, M-dialysis AB, Solna, Sweden) at the time of surgical removal of ICH. One catheter was placed within 1 cm of the evacuated hematoma, in the perihemorrhagic zone (PHZ), and the other at a distance of at least 1 cm from the evacuated ICH in a non-eloquent area of seemingly normal brain cortex (SNX) on the ipsilateral side ( Fig. 1a-c).
Microdialysis sampling was initiated directly post-surgery and microdialysis vials were changed every two hours. The catheters were perfused with a commercially available 5% Human Albumin in a water solution containing the excipients sodium chloride, N-acetyl-DL-tryptophan and caprylic acid (Albunorm, Octapharma, Stockholm, Sweden), at a rate of 0.3 µL/min. The use of albumin in the perfusate, to counteract ultrafiltration and subsequent fluid loss to the tissue, has been clinical routine in our department since 2013. Microdialysate was collected at a time corresponding to 72 hours after ICH onset. The results of 2-DE of this microdialysate has been previously published (13).
Patients received an external ventricular drain (EVD) for ICP monitoring from which CSF was collected at a time point corresponding to removal of the microdialysis membranes, towards the end of the patient's NCC unit treatment period. The 2 first mL of aspired CSF were discarded and the following 2 mL were immediately sent to the laboratory where the sample was centrifuged for 10 min at 1800 x g at 4 °C, and the supernatant was subsequently stored at − 86 °C awaiting analysis.
Microdialysis catheter membranes were removed at the end of the NCC treatment period, cut into a polypropylene tube and stored initially at − 20 °C, then at − 86 °C, awaiting analysis. Three membranes from the perihemorrhagic zone (PHZ1, PHZ2, PHZ3) and two membranes from seemingly normal brain tissue (SNX1, SNX2) were used. The third SNX membrane was accidentally discarded in the NCC unit and thus not available for analysis.

Sample preparation Preparation of dialysate samples
The proteomic analysis of the microdialysate of the three patients has been previously published (13) along with a detailed description of the sample preparation. Briefly, 40 µL of each sample was applied onto an Albumin & IgG Depletion column (GE Healthcare, Uppsala, Sweden), before being desalted and lyophilized. Protein concentration was determined (14) before applying the sample to 2-DE.

Preparation of CSF samples
The CSF samples were applied to albumin and IgG depletion columns

Protein identification
Selected protein spots were excised from the gels using a home-made spot picker and destained. Protein identification for the low abundant proteins was performed using LC-MS/MS whereby the trypsinated peptides were dissolved in 6 µL of 0.1% formic acid (FA) and applied to a nano-flow HPLC system, EASY-nLC II (Thermo Scientific) in conjugation with the mass spectrometer, LTQ Orbitrap Velos Pro hybrid mass spectrometer (Thermo Scientific) with a nano-electrospray source as previously described (17).
Database searching was performed using MS-Fit search engine and MaxQuant version 1.5 with trypsin as digestion enzyme against a human taxonomy of the SwissProt and the NCBI databases. The following search parameters were used: maximum two missed cleavages; fragment ion mass tolerance 0.5 Da; parent ion mass tolerance 6 ppm; fixed modification-carbamidomethylation of cysteine; variable modifications -N-terminal acetylation and methionine oxidation. Data was filtered at 1% false discovery rate. Identifications were based on a minimum of two unique peptides.

Statistical analysis
Central tendency and dispersion of data are presented as mean and standard deviation when data are summarized; alternatively numbers are presented comprehensively for clarity.

Results
Highest numbers of protein spots were found on CMD membranes Microdialysis samples from six catheters, five CMD membranes and three CSF samples were available for proteomic analysis from three ICH patients. CMD membranes were implanted in each patient for Distinct differences in protein expression in CSF, microdialysate and CMD membrane Distinct groups of protein spots could be observed on the CMD membrane gels which were totally missing from both CSF and microdialysate gels (Figs. 2 and 4). These protein spots were chosen for further analysis and identified by mass spectrometry (Table 2, Fig. 4). Spots of interest were excised, 68 spots from membrane gel and 62 from the CSF-gel, destained, digested by trypsin and analyzed by mass spectrometry. Results of identified spots on the dialysate gels have been previously published (13). Some of these results are included in Table 2 of this current study for clarity. Twentyfive proteins from the membrane gel and 30 proteins from the CSF gel were identified. Results were dominated by high abundant plasma proteins such as, hemoglobin, apolipoproteins and haptoglobins, but there were also less abundant proteins specifically present only on the membrane and not in CSF or in microdialysate including calmodulin-like-4 isoform CRA_c, BRICK1, mitochondrial import inner membrane translocase subunit Tim8 A, transthyretin, S100-A9, and Aldo-keto-reductase family 1 member C2 (Table 2). Furthermore, there were distinct quantitative differences between the three compartments (Fig. 3).   Table 2 Identified proteins in samples extracted from CMD membrane (M), CSF (C) and dialysate (D).

Discussion
This study is the first to report on the pattern of protein adsorption to CMD membranes following implantation in human brain tissue. It is also the first to compare this pattern to that of CSF and microdialysate; two compartments commonly sampled for biomarker discovery. Our results offer insights into which proteins adsorb to the membranes and thus may not be present in the microdialysate or CSF samples.
In recent years CMD has been used as part of NCC monitoring and to study biomarkers of various neurological diseases in particular traumatic brain injury (18)(19)(20), gliomas (21,22), ischemic stroke(1), intracerebral hemorrhage(23) and subarachnoid hemorrhage (24,25). CMD sampling for subsequent proteomic biomarker analysis has advantages, such as capture of proteins near the site of origin in the extracellular space, and no effect of dilution as is the case in CSF or plasma (18). But there are well known challenges to the microdialysis method including its focal resolution, limited time resolution, small sample volume(26), a tendency to elicit an inflammatory response (27) and variations in relative recovery (28,29), which is in part due to protein adsorption. Such protein adsorption on CMD membranes implanted in patients has been poorly described to date.
Ultrafiltration is another technical issue to consider when using high-molecular-weight cut-off CMD catheters (100 kDa). Ultrafiltration means that fluid escapes from the perfusate into the sampling tissue thereby changing the environment immediately surrounding the CMD catheter (2,26,30). To counteract ultrafiltration colloids such as albumin can be added to the perfusate as was clinical routine in our department during the study period. However, albumin is known to adsorb to membrane surfaces and may in turn affect the recovery of other proteins. This was indirectly shown in a previous study using CMD in a NCC setting where the addition of albumin to the perfusate was found to be necessary in order to extract amyloid-β (Aβ) in the microdialysate. The authors argue that this was presumably due to Aβ adhering to the CMD membrane and tubing in the absence of albumin (31). It is plausible that the addition of albumin to the perfusate also in the present study has affected the pattern of protein adsorption to the CMD membrane, and use of a different perfusate could result in a different pattern. Therefore the pattern of protein adsorption to CMD membranes cannot be assumed to be the same in different study protocols as protein adsorption will be affected by a number of factors, of which perfusate composition is one. This is why we recommend analyzing membrane adsorbed proteins when sampling macromolecules using CMD.
Modifications can be made to either the membrane or the perfusate in order to increase relative recovery of macro molecules. Modifications to the perfusate include the addition of antibodies with high affinity for a specific compound of interest which can increase its relative recovery as has been done with cytokines (32). This method, however, is not relevant when the aim is to sample all proteins present. Previous in vitro and preclinical studies show decreased protein adhesion to both CMD membrane and tubing following catheter modifications using triblock copolymers such as Pluronics 127® (6,33). However, to date these modifications have only been used in vitro or in preclinical studies. Certain proteins of interest may display particular tendencies to adhere to surface materials, such as for example Aβ protein and modifications of membrane and tubing may be useful in studies focused on such particularly 'sticky' proteins. As this present study shows proteins have differing tendency to adsorb to the membrane. An awareness of this is recommended when using CMD for sampling macromolecules.
It has been suggested that adsorption of proteins to the membrane surface is the first step in a potential foreign body reaction elicited by catheter implantation (34). Such a foreign body reaction has not been fully characterized following implantation in human brain, however, several preclinical studies have shown an immediate traumatic response in brain tissue following CMD catheter implantation (12). Furthermore, one study using scanning electron microscopy demonstrated buildup of cellular debris on CMD membranes implanted in human brain tissue (30). This cellular debris is presumably preceded by protein adsorption to the membrane surface, and characterization of such protein adsorption can plausibly give more insights also into the foreign body reaction.
In a recent study of ICH patients, using a similar study design we found a difference in proteins expressed in the perihemorrhagic zone (PHZ) compared to seemingly normal cortex (SNX) (13). In light of the results of this present study it is likely that several proteins evaded detection by adsorbing to the CMD membranes. Plausibly, a paired catheter design involves a similar pattern of protein adsorption to both catheters, but this cannot be guaranteed. Factors such as pH, rate of diffusion, and brain tissue edema may differ locally in the tissue and thus protein adsorption to CMD membranes could be different in two catheters placed in different regions of the brain. This can only be adequately characterized by analyzing the membranes.
Limitations of the present study include the small sample size, which precludes any comparisons or correlation with clinical parameters such as outcome or medical or surgical complications.
Furthermore, the present study cannot determine the effect of time on protein adsorption, and future studies should aim to elucidate the time course of protein adsorption to CMD membranes in brain tissue as this may cause relative recovery to change over time.
The present study is the first to characterize protein adsorption to CMD membranes after implantation in human brain tissue. Previous studies have described protein adhesion to membranes in animal studies or in vitro studies (7,11). One such in vitro study described protein adsorption to two microdialysis membranes after dialysis of a sample matrix of ventricular CSF. The study showed that 50% of proteins found on the CMD membrane were not identified in the sample matrix pointing to methodological limitations. This study used on-surface tryptic digestion of proteins on the membrane prior to denaturation, followed by a bottom-up approach for protein identification using LC-MS/MS. Technical issues may explain the low overlap of proteins between the CMD membrane and the sample matrix. To avoid failure of proteins to desorb from the CMD membrane in the present study we employed a protocol previously determined by our group, used on membranes implanted in human muscle tissue (unpublished data), and calibrated for optimal protein elution.
By showing particular differences in the proteome profile of the three compartments, CMD membrane, microdialysate and CSF, our results emphasize the need to analyze protein adsorption to CMD membranes in addition to microdialysate protein content and highlight the need to interpret analysis results of proteins sampled by CMD with caution.

Conclusion
Information on protein content in interstitial fluid is inadequately described by analyzing only the microdialysate, since proteins present may adsorb to and be found only on the CMD membrane.
These proteins may play an important role in understanding the pathophysiology of the studied disease, in this case intracerebral hemorrhage (ICH), and failure to identify them in the dialysate samples may lead to incorrect conclusions. The adsorbed proteins may also alter the function of the membrane and thereby affect the dialysate composition and influence the concentration of the proteins in the microdialysate. When using CMD to sample proteins from human brain tissue we recommend analysis of protein adsorption to the membranes in addition to in the microdialysate samples.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Authors' contributions
LT, CZ, BG and NM wrote the manuscript. All authors critically revised the manuscript. BG prepared     Table 2.