Identification of chronic brain protein changes and protein targets of serum auto-antibodies after blast-mediated traumatic brain injury.

In addition to needing acute emergency management, blast-mediated traumatic brain injury (TBI) is also a chronic disorder with delayed-onset symptoms that manifest and progress over time. While the immediate consequences of acute blast injuries are readily apparent, chronic sequelae are harder to recognize. Indeed, the identification of individuals with mild-TBI or TBI-induced symptoms is greatly impaired in large part due to the lack of objective and robust biomarkers. The purpose of this study was to address these need by identifying candidates for serum-based biomarkers of blast TBI, and also to identify unique or differentially regulated protein expression in the thalamus in C57BL/6J mice exposed to blast using high throughput qualitative screens of protein expression. To identify thalamic proteins differentially or uniquely associated with blast exposure, we utilized an antibody-based affinity-capture strategy (referred to as "proteomics-based analysis of depletomes"; PAD) to deplete thalamic lysates from blast-treated mice of endogenous thalamic proteins also found in control mice. Analysis of this "depletome" detected 75 unique proteins, many with associations to the myelin sheath. To identify blast-associated proteins eliciting production of circulating autoantibodies, serum antibodies of blast-treated mice were immobilized, and their immunogens subsequently identified by proteomic analysis of proteins specifically captured following incubation with thalamic lysates (a variant of a strategy referred to as "proteomics-based expression library screening"; PELS). This analysis identified 46 blast-associated immunogenic proteins, including 6 shared in common with the PAD analysis (ALDOA, PHKB, HBA-A1, DPYSL2, SYN1, and CKB). These proteins and their autoantibodies are appropriate for further consideration as biomarkers of blast-mediated TBI.

Introduction

Blast-mediated traumatic brain injury (TBI) is a common condition among active and recently-active military personnel, and also affects civilian populations [1]. Blast-mediated TBI is a traumatic event that needs both acute and chronic management, and symptoms typically manifest and progress chronically [2]. Identification of individuals with mild TBI or TBI-induced symptoms is difficult for multiple reasons, including self-reporting of blast-exposure. In addition, improvements in protective armor have improved survivability in recent conflicts, which has resulted in an increased incidence of TBI [3]. Even if TBI is suspected based on the reported history, a confounding factor for symptom-based diagnosis is that individuals with TBI can present with a wide constellation of symptoms which include cognitive, behavioral, neuropsychological, motor and visual impairment [4, 5]. Many of these symptoms may not be immediately apparent and may only manifest months to years after the initial injury, or are diagnosed post-mortem [6, 7]. Thus, there is a significant unmet need for objective blood-based biomarkers for mild injuries that can be used to help confirm diagnosis.

A significant challenge to developing tools for diagnosis and therapeutic treatments for blast-mediated mild TBI is that the timing of neuronal loss and the pathways leading to neuronal impairment and degeneration are not well understood. The pathophysiology of blast-mediated TBI is precipitated by the interaction of a blast wave with neuronal tissue. Following this interaction, multiple mechanisms that lead to the death and dysfunction of neurons after exposure to TBI have been reported from several models [8, 9, 10]. These mechanisms include immediate [11, 12, 13] and delayed [14, 15] neuronal changes. In many models, classic markers of apoptotic cell death are not present [16]. Cellular mechanisms that are believed to lead to neuronal death and dysfunction after mild TBI or repetitive TBI include diffuse axonal injury and myelin damage, excitotoxicity, dysregulation of the neurovascular unit, inflammation, and oxidative stress [17, 18, 19, 20, 21]. Each of these processes has been individually implicated in contributing to neuronal death; the possibility of their complex combinatorial interactions seems likely, but remains largely untested.

In part because of the difficulties in symptom-based diagnosis and an incomplete knowledge of damage-inducing mechanisms, developing protein-based biomarkers in the body that can serve as a proxy for poorly-accessible organs such as the brain has been a focus of considerable research effort [22, 23, 24]. Many proteins that are expressed differentially or uniquely during pathogenesis of TBI have been identified. Among the most promising candidates for biomarkers of blast-mediated TBI are several which have proven useful for other forms of acquired brain injury, including: S-100β, neuron specific enolase (NSE), glial fibrillary acid protein (GFAP), myelin basic protein (MBP), and BDNF [25]. While these proteins have been implicated with varying levels of TBI, their utility in detecting patients with mild blast-mediated TBI has not been completely validated.

Here, we report the discovery of novel candidates for serum-based biomarkers of blast TBI in addition to discovering unique and upregulated proteins in the thalamus in C57BL/6J mice exposed to blast using novel, robust qualitative screens for rapid identification of proteins and analytes of interest. To achieve this, we first identified thalamic proteins differentially or uniquely associated with blast exposure. Next, we identified blast-associated proteins eliciting production of circulating autoantibodies. Finally, results of both approaches were compared to identify the small subset shared in common, i.e. proteins induced by blast that also solicit sustained autoantibody production. We performed these analyses using variations of proteomics-based approaches: proteomics-based analysis of depletomes (PAD) and proteomics-based expression library screening (PELS), which each utilize antibody-based affinity capture of proteins and their characterization using tandem mass spectrometry in different iterations (Figure 1). The results identify six proteins with properties appropriate for further consideration as biomarkers of blast-mediated TBI, as well as several others previously not known to be associated with blast-injury but potentially relevant to mechanisms of damage caused by blast-exposure. These proteins are: fructose-biphosphate aldolase A, phosphorylase b kinase regulatory subunit beta, alpha globin 1, dihydropyrimidinase-related protein 2, isoform 1b of synapsin 1, and creatine kinase B-type.

Figure 1

The PELS principle for generation of affinity-captured proteome/depletome used in this study.

Methods

Animals

All animal studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Iowa City Veterans Affairs Institutional Animal Care and Use Committee. The study utilized two experimental groups of mice, which both consisted entirely of male C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME), and were randomly assigned to one of the following treatment groups at 8 weeks of age: 1) mice were exposed to one single blast injury (referred to throughout as “TBI-mice”), or 2) mice were subjected to a sham-injury (“sham-mice”). A total of 92 mice were used in this study.

Blast injury induction

An enclosed blast chamber was used for the purpose of these studies to investigate the effect of primary blast exposure. One half of this tank was pressurized, with a 13-cm opening between the sides of the chamber, as described previously [11, 12]. A Mylar membrane (Mylar A, 0.00142 gauge; Country Plastics, Ames, IA, USA) was placed over the opening on the pressurized side of the chamber. The unpressurized side of the tank contained a padded polyvinyl chloride (PVC) protective restraint for positioning of an anesthetized mouse 30 cm from the Mylar membrane. Compressed air was pumped into the pressurized side of the tank to 20 psi, at which point the membrane ruptured and created a blast wave. Mice were anesthetized with a combination of ketamine (0.03 mg/g, intraperitoneal [IP]) and xylazine (0.005 mg/g, IP) and positioned within the blast chamber with the left side of the head and eye oriented toward the source of the blast wave. To ensure that the primary effect of the blast wave was at the level of the head, only the head of TBI-mice was exposed to the blast wave, with the rest of the body shielded. The head was allowed to move freely and was not fixed in position, although care was taken to ensure that the head did not impact hard surfaces, and was supported from major movement with thick foam placed behind the head. Immediately following exposure to the blast wave, TBI-mice were placed on a heating pad to facilitate recovery from general anesthesia and to prevent hypothermia. Xylazine anesthesia was reversed with yohimbine chloride (0.001 mg/g, IP) to speed the recovery from anesthesia. Control sham-mice used in this study were anesthetized and placed in the blast chamber, but did not receive a blast exposure. TBI-mice and sham-mice equally received analgesic via subcutaneous injection (0.1 mL/20 g body weight) of buprenorphine (0.003 mg/mL) immediately after the blast or sham-blast, respectively.

Sample collection

For studies of thalamic proteins, TBI-mice (n = 12; 4 weeks post blast) and sham-mice (n = 12; 4 weeks post sham injury) were humanely euthanized and thalamic regions micro-dissected. Homogenates from each subject group were pooled together and stored at -80 °C prior to analysis. For studies of circulating auto-antibodies, serum specimens were collected from cardiac punctures of TBI-mice (n = 12) and sham-mice (n = 12) 8 weeks following blast exposure. Serum specimens were pooled and polyclonal antibodies purified via Protein A affinity chromatography using HiTrap Protein A HP (1 ml) columns (GE Healthcare) per manufacturer guidelines.

Proteomics-based analysis of depletomes (PAD)

The term “depletome” refers to the complement of interesting molecules resident in a complex mixture, following selective depletion of irrelevant components. To derive the depletome of the thalamus from blast-exposed mice, bait polyclonal antibodies were generated in chickens (IgY) against proteins from pooled thalami of sham-mice (C57BL/6J Male mice, 8 weeks of age at the beginning of the study) using the services of a commercial vendor (Aves Labs, OR), and affinity purified using anti-chicken IgY polyclonal generated in goats. The bait IgY-polyclonal antibodies (titer assessed to be >1:10,000 in dot immunoblotting against 2 μg of the immunogen mixture) were then covalently coupled to Dynabeads M-280 Tosylactivated (Invitrogen/Life Technologies, CA) and HiTrap NHS-activated columns (1 ml; GE Healthcare Life Sciences) per manufacturer guidelines. The thalamus protein extracts from TBI-mice (complex mixture; 5 mg total protein in 5 mls of PBS [pH 7.4]) were reacted first with charged Dynabeads M-280 Tosylactivated and then passed through charged HiTrap NHS-activated columns per manufacturer guidelines. This process of selective depletion of confounding proteins from the complex mixture and the simultaneous enrichment for relevant proteins, resulted in a depletome constituted by proteins that were either differentially (i.e., produced in larger amounts in thalami of TBI-mice than in those of untreated mice, defined as an increase of 1 or more identified peptides compared to untreated mice) or uniquely expressed in thalami of TBI-mice 4 weeks post injury. Increases in protein peptides are commonly used for analysis of high-throughput, qualitative assays of protein expression [26]. The proteins comprising the depletome were processed and subjected to tandem mass spectrometry for identification. Protein identifications were linked to gene symbols for 75 proteins in the depletome; 2 peptides were excluded (IPI00987580, IPI00224605) because they linked to predicted pseudogenes.

Proteomics-based expression library screening (PELS)

The overall strategy followed a published PELS protocol [27], with variations [28] to identify host thalamus proteins shed in body fluids following blast-mediated injury. First, “bait” polyclonal antibodies (bait PAbs) were generated from the pooled sera of TBI-mice (8 weeks post blast) and were covalently coupled to HiTrap NHS-activated columns (1 ml; GE Healthcare Life Sciences) creating “charged columns”. Next, pooled thalamic protein extracts from TBI-mice (4 weeks post blast) containing the analytes of interest were subjected to immunoaffinity capture by passage through the charged columns. The captured proteins were then eluted and subjected to tandem mass spectrometry for identification. Elutions of the same extracts loaded on NHS columns charged with bait PAbs affinity purified from sera collected from untreated mice and on NHS columns without covalently coupled polyclonal antibodies, but quenched active groups (“uncharged”) served as controls for assessing both specificity of bait PAbs and nonspecific adsorption to the column matrix. Protein identifications were linked to gene symbols for 46 proteins identified by PELS; 3 peptides were excluded because they linked to predicted pseudogenes (IPI00987580, IPI00265107) or could not be linked to a gene (IPI00462809).

Tandem mass spectrometry

Tandem mass spectra were extracted by ABI Analyst version 2.0. All MS/MS samples were analyzed using Mascot (Matrix Science, London, UK; version 2.2.2). Mascot was set up to search the IPI-Mouse FASTA database assuming digestion enzyme trypsin. Protein identifications were accepted if they could be established at greater than 99.0% probability and contained at least two identified unique peptides. Proteins with single peptide hits were included if they exhibited high confidence based on low false discovery rates [29]. Relative protein abundance was estimated using the normalized total spectral counts [30]. Protein probabilities were assigned using the Protein Prophet algorithm [31]. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.

Functional annotation and pathway analysis

DAVID [32, 33] and WebGestalt [34, 35] were used to compare protein lists against a C57BL/6J mouse brain proteome [36]. Settings for functional annotation using DAVID utilized the gene symbols for the 75 proteins in the depletome for “gene list” the brain proteome list as “background”, and “Mus musculus” as species. Outputs, and their abbreviations used herein, for UP_KEYWORDS (keywords), GOTERM_CC_DIRECT (cellular compartment), GOTERM_MF_DIRECT (molecular function), GOTERM_BP_DIRECT (biological process), KEGG_PATHWAY (KEGG pathway), INTERPRO (protein domains) were compiled into spreadsheets with only terms surpassing statistical significance after multiple hypothesis testing performed by DAVID using the Benjamini-Hochberg method. Settings for WebGestalt utilized the same gene lists described above, with “mmusculus” for species and default settings for all other parameters. One of the 75 genes in the depletome (Atp5f1b, IPI00468481) was not recognized by DAVID or WebGestalt because of ortholog ambiguity.

Results

To identify thalamic proteins differentially or uniquely associated with blast exposure, we utilized the PAD strategy to identify a depletome of proteins over-represented in the thalamus of TBI-mice at 4 weeks after blast, compared to the thalamus of age-, sex-, and strain-matched sham-mice. This analysis identified 75 proteins (Table 1). To identify blast-associated proteins eliciting production of circulating autoantibodies, we utilized a variant of the PELS strategy to identify serum antibodies of TBI-mice at 8 weeks post blast that recognize thalamic proteins of TBI-mice at 4 weeks post blast. This analysis identified 46 blast-associated immunogenic proteins (Table 2). A comparison of the results in common to both approaches, i.e. proteins differentially/uniquely associated with blast exposure and those that elicit a sustained production of autoantibodies, identified six proteins (Table 3).

Table 1

Depletome proteins identified using PAD.

Identified proteins	Accession number	Molecular weight	Number of unique peptides in thalamus of untreated mouse	Number of unique peptides in depletome	UniProtKB	Gene Symbol
Gamma-enolase	IPI00331704	47 kDa	4	12	P17183	Eno2
Serum albumin	IPI00131695	69 kDa	2	9	P07724	Alb
Creatine kinase B-type	IPI00136703	43 kDa	4	8	Q04447	Ckb
Malate dehydrogenase, mitochondrial	IPI00323592	36 kDa	1	7	P08249	Mdh2
Heat shock cognate 71 kDa protein	IPI00323357	71 kDa	0	6	P63017	Hspa8
Dihydropyrimidinase-related protein 2	IPI00114375	62 kDa	2	6	O08553	Dpysl2
Isoform Ib of Synapsin-1	IPI00136372 (+1)	70 kDa	3	6	O88935	Syn1
14-3-3 protein gamma	IPI00230707	28 kDa	2	5	P61982	Ywhag
Isoform M2 of Pyruvate kinase isozymes M1/M2	IPI00407130 (+1)	58 kDa	0	5	P52480	Pkm
Aspartate aminotransferase, cytoplasmic	IPI00230204	46 kDa	0	5	P05201	Got1
Actin, cytoplasmic 1	IPI00110850 (+4)	42 kDa	2	4	P60710	Actb
Superoxide dismutase [Cu–Zn]	IPI00130589	16 kDa	2	4	P08228	Sod1
Alpha-enolase	IPI00462072 (+2)	47 kDa	0	4	P17182	Eno1
Tubulin beta-4 chain	IPI00109073 (+1)	50 kDa	0	4	Q9D6F9	Tubb4a
Isoform 1 of Alpha-synuclein	IPI00115157 (+1)	14 kDa	0	4	O55042	Snca
Tubulin alpha-1A chain	IPI00110753 (+3)	50 kDa	0	3	P68369	Tuba1a
Stathmin	IPI00551236	17 kDa	0	3	P54227	Stmn1
Malate dehydrogenase, cytoplasmic	IPI00336324	37 kDa	0	3	P14152	Mdh1
Isoform 1 of Microtubule-associated protein 1A	IPI00408909 (+1)	300 kDa	1	2	Q9QYR6	Map1a
Triosephosphate isomerase	IPI00467833 (+1)	32 kDa	1	2	P17751	Tpi1
SH3 domain-binding glutamic acid-rich-like protein 3	IPI00127358	10 kDa	1	2	Q91VW3	Sh3bgrl3
Isoform HuC-L of ELAV-like protein 3	IPI00122451	40 kDa	0	2	Q60900	Elavl3
Fructose-bisphosphate aldolase C	IPI00119458	39 kDa	0	2	P05063	Aldoc
2-iminobutanoate/2-iminopropanoate deaminase	IPI00130640	14 kDa	0	2	P52760	Rida
Peroxiredoxin-2	IPI00117910 (+1)	22 kDa	0	2	Q61171	Prdx2
Isoform 1 of Serine/threonine-protein phosphatase 2B catalytic subunit alpha isoform	IPI00121545 (+2)	59 kDa	0	2	P63328	Ppp3ca
Annexin A5	IPI00317309	36 kDa	0	2	P48036	Anxa5
Heat shock protein HSP 90-alpha	IPI00330804	85 kDa	0	2	P07901	Hsp90aa1
Elongation factor 1-alpha 2	IPI00119667	50 kDa	0	2	P62631	Eef1a2
Isoform Mitochondrial of Peroxiredoxin-5, mitochondrial	IPI00129517 (+3)	22 kDa	0	2	P99029	Prdx5
L-lactate dehydrogenase B chain	IPI00229510	37 kDa	0	2	P16125	Ldhb
Ras-related protein Rab-3A	IPI00122965	25 kDa	0	2	P63011	Rab3a
Ubiquitin carboxyl-terminal hydrolase isozyme L1	IPI00313962 (+1)	25 kDa	0	2	Q9R0P9	Uchl1
Fructose-bisphosphate aldolase A	IPI00221402	39 kDa	0	2	P05064	ALDOA
Alpha globin 1	IPI00845802	15 kDa	1	2	Q91VB8	Hba-a1
Cytochrome P450, family 2, subfamily c, polypeptide 68	IPI00405136 (+1)	56 kDa	0	1	Q8VCP4	Cyp2c68
Translationally-controlled tumor protein	IPI00129685	19 kDa	0	1	P63028	Tpt1
Proteasomal ubiquitin receptor ADRM1	IPI00331155	42 kDa	0	1	Q9JKV1	Adrm1
Isoform 1 of Serine/threonine-protein kinase SMG1	IPI00403352	410 kDa	0	1	Q8BKX6	Smg1
Isoform 2 of Neurogenic locus notch homolog protein 2	IPI00621767	243 kDa	0	1	O35516	Notch2
Isoform 4 of Myocyte-specific enhancer factor 2C	IPI00318314 (+1)	47 kDa	0	1	Q8CFN5	Mef2c
Parvalbumin alpha	IPI00230766	12 kDa	0	1	P32848	Pvalb
Neurogranin	IPI00380227	7 kDa	0	1	P60761	Nrgn
Polyadenylate-binding protein 1	IPI00124287 (+2)	71 kDa	0	1	P29341	Pabpc1
Histidine triad nucleotide-binding protein 1	IPI00108189 (+1)	14 kDa	0	1	P70349	Hint1
Isoform 1 of 60 kDa heat shock protein, mitochondrial	IPI00308885 (+1)	61 kDa	0	1	P63038	Hspd1
Isoform 1 of Microtubule-associated protein 6	IPI00115833	96 kDa	0	1	Q7TSJ2	Map6
Tubulin alpha-4A chain	IPI00117350	50 kDa	0	1	P68368	Tuba4a
Ras-related protein Rab-1A	IPI00114560 (+3)	23 kDa	0	1	P62821	Rab1a
V-type proton ATPase subunit B, brain isoform	IPI00119113	57 kDa	0	1	P62814	Atp6v1b2
Protein kinase C and casein kinase substrate in neurons protein 1	IPI00123613	51 kDa	0	1	Q61644	Pacsin1
Isoform IIb of Synapsin-2	IPI00134492 (+2)	52 kDa	0	1	Q64332	Syn2
Isoform 1 of Alpha-adducin	IPI00136000 (+4)	81 kDa	0	1	Q9QYC0	Add1
Profilin-1	IPI00224740 (+1)	15 kDa	0	1	P62962	Pfn1
Phosphoglycerate mutase 2	IPI00230706 (+1)	29 kDa	0	1	O70250	Pgam2
Rho GDP-dissociation inhibitor 1	IPI00322312	23 kDa	0	1	Q99PT1	Arhgdia
Rab GDP dissociation inhibitor alpha	IPI00323179	51 kDa	0	1	P50396	Gdi1
Isoform 1 of Structural maintenance of chromosomes protein 5	IPI00380203 (+1)	129 kDa	0	1	Q8CG46	Smc5
Amphiphysin	IPI00400180	75 kDa	0	1	Q7TQF7	Amph
ATP synthase subunit beta, mitochondrial	IPI00468481	56 kDa	0	1	P56480	Atp5f1b
Choline transporter-like protein 3	IPI00122287	73 kDa	0	1	Q921V7	Slc44a3
Ras-related protein Rab-11B	IPI00135869 (+5)	24 kDa	0	1	P46638	Rab11b
Ubiquitin-60S ribosomal protein L40	IPI00138892 (+11)	15 kDa	0	1	P62984	Uba52
Isoform 3 of NSFL1 cofactor p47	IPI00387232 (+2)	41 kDa	0	1	Q9CZ44	Nsfl1c
Mitogen-activated protein kinase 1	IPI00119663 (+1)	41 kDa	0	1	P63085	Mapk1
Neurocan core protein	IPI00135563 (+1)	137 kDa	0	1	P55066	Ncan
Tyrosine-protein phosphatase non-receptor type substrate 1	IPI00756790 (+3)	17 kDa	0	1	P97797	Sirpa
Vomeronasal 2, receptor 112	IPI00660717	98 kDa	0	1	L7N221	Vmn2r112
Adaptin ear-binding coat-associated protein 1	IPI00225533	30 kDa	0	1	Q9CR95	Necap1
Transcriptional activator protein Pur-beta	IPI00128867	34 kDa	0	1	O35295	Purb
Isoform 1 of D-aminoacyl-tRNA deacylase 1	IPI00133713	23 kDa	0	1	Q9DD18	Dtd1
Endophilin-A1	IPI00331110 (+3)	40 kDa	0	1	Q62420	Sh3gl2
Macrophage migration inhibitory factor	IPI00230427	13 kDa	0	1	P34884	Mif
14-3-3 protein epsilon	IPI00118384 (+1)	29 kDa	0	1	P62259	Ywhae
Phosphorylase b kinase regulatory subunit beta	IPI00380735	124 kDa	0	1	Q7TSH2	Phkb

Table 2

Identification of proteins shed into circulation after TBI that elicited the formation of autoantibodies in the serum of TBI-mice and identified with PELS.

Identified proteins	Accession number	Molecular weight	Number of unique peptides	UniProtKB	Gene symbol
Fructose-bisphosphate aldolase A	IPI00221402	39 kDa	11	P05064	ALDOA
40S ribosomal protein S19	IPI00113241 (+5)	16 kDa	7	Q9CZX8	Rps19
Alpha globin 1	IPI00845802	15 kDa	7	Q91VB8	Hba-a1
40S ribosomal protein S13	IPI00125901 (+2)	17 kDa	6	P62301	Rps13
60S ribosomal protein L27	IPI00122421 (+1)	16 kDa	6	P61358	Rpl27
Histone H2B type 1-F/J/L	IPI00114642 (+10)	14 kDa	5	P10853	Hist1h2bf
Hemoglobin subunit beta-1	IPI00762198 (+3)	16 kDa	5	P02088	Hbb-b1
40S ribosomal protein S15	IPI00319231 (+1)	17 kDa	4	P62843	Rps15
Isoform 2 of 40S ribosomal protein S24	IPI00402981 (+4)	15 kDa	4	P62849	Rps24
60S ribosomal protein L29	IPI00222548 (+9)	18 kDa	3	P47915	Rpl29
Histone H1.4	IPI00223714	22 kDa	3	P43274	Hist1h1e
60S ribosomal protein L13	IPI00224505 (+1)	24 kDa	3	P47963	Rpl13
60S ribosomal protein L6	IPI00313222 (+2)	34 kDa	3	P47911	Rpl6
40S ribosomal protein S10	IPI00112448	19 kDa	3	P63325	Rps10
40S ribosomal protein S18	IPI00317590 (+2)	18 kDa	3	P62270	Rps18
60S ribosomal protein L39	IPI00273974 (+4)	6 kDa	2	P62892	Rpl39
Isoform 1 of GRB10-interacting GYF protein 2	IPI00473912 (+1)	149 kDa	2	Q6Y7W8	Gigyf2
40S ribosomal protein S7	IPI00136984 (+1)	22 kDa	2	P62082	Rps7
40S ribosomal protein S29	IPI00222553	7 kDa	2	P62274	Rps29
60S ribosomal protein L35a	IPI00115902 (+1)	13 kDa	2	O55142	Rpl35a
Histone H2A.J	IPI00153400 (+13)	14 kDa	2	Q8R1M2	H2afj
Isoform Ib of Synapsin-1	IPI00136372 (+1)	70 kDa	2	O88935	Syn1
Isoform 2 of Heterogeneous nuclear ribonucleoprotein Q	IPI00406118 (+3)	63 kDa	1	Q7TMK9	Syncrip
Isoform 2 of Calcium-activated potassium channel subunit alpha-1	IPI00120643 (+12)	135 kDa	1	Q08460	Kcnma1
60S ribosomal protein L18	IPI00555113 (+1)	22 kDa	1	P35980	Rpl18
CCAAT/enhancer-binding protein beta	IPI00116613 (+3)	31 kDa	1	P28033	Cebpb
60S ribosomal protein L8	IPI00137787	28 kDa	1	P62918	Rpl8
60S ribosomal protein L35	IPI00263879 (+1)	15 kDa	1	Q6ZWV7	Rpl35
Histone H1.3	IPI00331597	22 kDa	1	P43277	Hist1h1d
Anti-colorectal carcinoma light chain	IPI00462809 (+7)	26 kDa	1	Q7TS98	N/A
Isoform 1 of Protein FAM126B	IPI00226426 (+1)	59 kDa	1	Q8C729	Fam126b
V-type proton ATPase subunit G 2	IPI00123817 (+1)	14 kDa	1	Q9WTT4	Atp6v1g2
Plexin-A2	IPI00137313	212 kDa	1	P70207	Plxna2
Nuclear receptor subfamily 1 group D member 2	IPI00119178	64 kDa	1	Q60674	Nr1d2
Serine/arginine-rich splicing factor 2	IPI00121135 (+2)	25 kDa	1	Q62093	Sfrs2
Keratin, type II cytoskeletal 6B	IPI00131366 (+7)	60 kDa	1	Q9Z331	Krt6b
Isoform 1 of Dynamin-1	IPI00272878 (+5)	98 kDa	1	P39053	Dnm1
60S ribosomal protein L24	IPI00323806 (+3)	18 kDa	1	Q8BP67	Rpl24
Isoform 1 of Syntaxin-binding protein 1	IPI00415402 (+2)	68 kDa	1	O08599	Stxbp1
Peptidyl-prolyl cis-trans isomerase A	IPI00554989	18 kDa	1	P17742	Ppia
Probable G-protein coupled receptor 158	IPI00465871	134 kDa	1	Q8C419	Gpr158
Succinyl-CoA ligase [GDP-forming] subunit alpha, mitochondrial	IPI00406442 (+1)	36 kDa	1	Q9WUM5	Suclg1
LYR motif-containing protein 4	IPI00169804	11 kDa	1	Q8K215	Lyrm4
Phosphorylase b kinase regulatory subunit beta	IPI00380735	124 kDa	1	Q7TSH2	Phkb
Dihydropyrimidinase-related protein 2	IPI00114375	62 kDa	1	O08553	Dpysl2
Creatine kinase B-type	IPI00136703	43 kDa	1	Q04447	Ckb

Table 3

Potential TBI biomarkers simultaneously identified with PELS and PAD.

Identified proteins	Accession number	Molecular weight	Number of unique peptides in thalamus of untreated mouse	Number of unique peptides in depletome	Number of unique peptides identified as immunogenic with PELS	UniProtKB	Gene symbol
Fructose-bisphosphate aldolase A	IPI00221402	39 kDa	0	2	11	P05064	ALDOA
Phosphorylase b kinase regulatory subunit beta	IPI00380735	124 kDa	0	1	1	Q7TSH2	Phkb
Alpha globin 1	IPI00845802	15 kDa	1	2	7	Q91VB8	Hba-a1
Dihydropyrimidinase-related protein 2	IPI00114375	62 kDa	2	6	1	O08553	Dpysl2
Isoform Ib of Synapsin-1	IPI00136372 (+1)	70 kDa	3	6	2	O88935	Syn1
Creatine kinase B-type	IPI00136703	43 kDa	4	8	1	Q04447	Ckb

To test whether these protein lists included an over-representation of any gene ontology terms, the DAVID (Table 4) and WebGestalt (Table 5) databases were utilized to compare our results to a previously published C57BL/6J mouse brain proteome [36]. Analysis of the PAD-identified depletome indicated a wide-range of over-represented ontology terms. From analysis with DAVID, the most statistically significant over-represented terms of the depletome were both in the “Cellular Component” category, “Myelin sheath” (27 members, P = 3.7E-24) and “Extracellular exosome” (41 members, P = 1.7E-10). From analysis with WebGestalt, the statistical significance levels were less pronounced and more closely clustered, with several of the top terms in the “Cellular Component” category involving synapse- or axon-related ontology terms (including, “Neuron projection, 17 members, P = 2.09E-5; and “Axon”, 9 members, P = 6.0E-4).

Table 4

Gene ontology term and pathway analysis using the DAVID database.

	Depleteome (74)
Keywords:	Acetylation (37) 1.3E-4
	Phosphoprotein (52) 1.9E-4
	Glycolysis (6) 1.0E-4
	Cytoplasm (35) 1.0E-3
	Nitration (5) 7.8E-3
	Methylation (14) 3.3E-2
Biological Process:	Glycolytic process (6) 6.8E-4
Cellular Component:	Myelin sheath (27) 3.7E-24
	Extracellular exosome (41) 1.7E-10
	Cytosol (28) 5.6E-6
	Cytoplasm (46) 4.9E-4
	Extracellular space (16) 9.0E-5
	Axon (11) 6.3E-4
	Terminal bouton (9) 9.5E-6
	Synaptic vesicle (8) 5.7E-4
	Neuron projection (10) 1.6E-2
	Neuronal cell body (11) 1.5E-2
	Axon terminus (5) 1.6E-2
	Protein complex (12) 3.3E-3
	Blood microparticle (5) 1.7E-2
Molecular Function:	MHC Class II protein complex binding (4) 2.1E-2
	Catalytic activity (12) 1.5E-2
KEGG Pathway:	Carbon metabolism (9) 2.3E-3
	Glycolysis/Gluconeogenesis (7) 2.3E-3
	Biosynthesis of antibiotics (10) 6.1E-3
	Biosynthesis of amino acids (7) 4.6E-3

Table 5

Gene ontology term and pathway analysis using the WebGestalt database.

	Depleteome (74)
Biological Process:	Multicellular organismal process (32) 1.0E-4
	Catabolic process (22) 4.0E-4
	Single-multicellular organism process (32) 1.0E-4
	Organonitrogen combound metabolic process (20) 5.0E-4
	Nucleobase-containing small molecular metabolic process (16) 3.0E-4
	Monosaccharide catabolic process (6) 3.0E-4
	Hexose catabolic process (6) 3.0E-4
	Generation of precursor metabolites and energy (10) 4.0E-1
	Glucose catabolic process (6) 3.0E-4
	Glycolysis (6) 1.0E-4
Cellular Component:	Synapse (13) 1.0E-4
	Synapse part (10) 6.0E-4
	Cell projection (21) 1.8E-5
	Neuron projection (17) 2.09E-5
	Axon (9) 6.0E-4
	Cytoplasm (51) 2.45E-5
	Cytosol (22) 9.56E-8
	Coated vesicle (9) 3.32E-5
	Clathrin-coated vesicle (7) 6.0E-4
	Synaptic vesicle (6) 7.0E-4
Molecular Function:	Binding (50) 4.9E-3
	Catalytic activity (36) 2.6E-3
	Cell surface binding (4) 8.0E-4
	Protein binding (40) 1.1E-3
	Enzyme binding (15) 2.6E-3
	Phosphoprotein binding (4) 5.8E-3
	Protein phosphorylated amino acid binding (3) 4.9E-3
	Malate dehydrogenase activity (2) 9.1E-0.03
	Phosphopyruvate hydratase activity (2) 4.9 E-3
	L-malate dehydrogenase activity (2) 2.6E-3
KEGG Pathway:	Glycolysis/Gluconeogenesis (6) 4.0E-4
	Pyruvate metabolism (4) 1.5E-2
	Neurotrophin signaling pathway (4) 4.9E-2
	Phagosome (4) 4.9E-2
	Oocyte meiosis (4) 4.9E-2
	Phenylalanine metabolism (2) 4.9E-2

The cognitive function of mice was tested with the Morris water maze, a measure of hippocampal dependent learning and memory, in order to demonstrate that bTBI was inducing damage in the brain. Our results demonstrate that blast exposure did not affect learning during the training period of the task (Supplemental Figure 1A), but did result in a significant decrease in memory retention in mice with TBI compared to sham-blast mice (Supplemental Figure 1B, P = 0.0021). The average speed of each mouse calculated during the probe-test was not significantly different between sham and TBI-mice (Supplemental Figure 1C). The range of distributions for each parameter demonstrated that TBI induction results in consistent phenotypes without significant outliers. Anti-GFAP staining was performed in the brain to evaluate the astrocytic response to blast exposure (Supplemental Figure 2). These results demonstrate that there is not significant GFAP reactivity in the brains of mice 4 weeks following blast exposure when compared to sham-mice. These results also demonstrate that the intra-animal response to blast is uniform.

Discussion

We have utilized novel proteomics-based qualitative approaches to identify candidates for serum-based biomarkers of TBI. We focused our study on the thalamus, as it is a major sensory relay station in the brain, and previous findings of both auditory and visual difficulties have been observed in humans following blast-mediated TBI [37, 38, 39, 40, 41, 42, 43, 44]. Thus, the thalamus may be a site particularly prone to damage and appropriate for developing biomarkers. Using a previously described mouse model for studying blast-induced mild TBI, our current analysis identified six proteins with properties appropriate for further consideration as biomarkers of blast-mediated TBI, as well as several others previously not known to be associated with blast-injury and which may be relevant to ongoing mechanistic studies of damage caused by blast-exposure.

If developed, biomarkers could be of particular use with patients in which overt traumatic blast injuries were not sustained, but mild or chronic TBI is suspected. The identification and routine use of biomarkers could help to particularly improve the quality of life of Warfighters and Veterans, for whom mild to moderate cases of TBI might otherwise go undiagnosed. Many Veterans who have been exposed to blast complain of sensory impairment chronically post injury, but lack clearly identifiable diagnoses. For example, many blast-exposed Veterans complain of vision problems, have no measurable deficits in the visual pathway, but none-the-less progress to develop chronic visual impairments. Biomarkers could lead to the earlier identification of such patients and also promote their better monitoring and possible treatment.

A history of exposure to a blast from an explosion is common among Veterans of recent military conflicts. Blast exposures account for nearly 75% of combat-related injuries – 50% of which result in a diagnosis of mild TBI [45, 46]. Blast exposure results in axonal damage in the brain [47], and individuals exposed to blast often report chronic dysfunction of sensory organs [48]. Sensory system dysfunction in humans increases with repetitive blast exposure [49], and is reported in greater numbers further from the time of injury [50]. Damage to the visual system is found in both humans and preclinical models exposed to blast [51], although the mechanism of injury has not been elucidated [37]. Damage to the visual system has been reported in a variety of rodent bTBI models [11, 52, 53], which replicates visual dysfunction observed in Veterans exposed to blast [54, 55]. Visual dysfunction can be observed immediately following injury [56], with deficits persisting months after the injury [57]. The visual system damage observed after blast exposure is similar to damage observed in weight drop models of TBI [52, 58, 59], suggesting similar mechanisms. These mechanisms are varied and include axonal damage, activation of microglia, tissue swelling, infiltration of immune cells, upregulation of chemoattractants, and neuroinflammation [8, 60, 61, 62, 63, 64, 65, 66, 67].

The current study was designed based on using damage to the eye as a surrogate for sensory impairment in general, as sensory impairment, particularly visual function, is diminished by blast and non-blast TBI exposure [41, 68]. Thus, we sought to identify protein changes at a time point after blast exposure (4 weeks) when visual function has a transient recovery from deficits observed 1 h to one week following bTBI [11], but subtle physiologic abnormalities to visual stimuli can none-the-less be detected [12]. The visual function in this model declines significantly below baseline by two months post injury, and remains suppressed until at least 10 months post bTBI [11]. Although these time frames might all be considered “acute” for humans, in the context of the comparatively shortened life-span of a mouse, it is reasonable to surmise that they may roughly correlate to the pertinent time frame for a human who has been exposed to blast, sustained a mild TBI injury, and is manifesting an early stage of chronic sensory pathology when additional diagnosis tools could be useful.

Among the autoantibodies that were detected, there were identifiable links with other diseases and damage to CNS tissues. Notably, a recent study of autoantibodies that increase at the subacute phase of human spinal cord injury [69] discovered elevated levels of autoantibodies of three proteins also identified by our analyses: alpha globin alpha (HBA-A1), alpha globin beta (HBA-B1), and peptidyl-prolyl cis-trans isomerase A (PPIA). Amongst these, HBA-A1 stands out because it was identified by both our depletome and PELS analyses. As well as functioning in red blood cells, HBA and HBB have also been detected in brain neurons and hypothesized to have a role in response to injury [70]. PPIA, also referred to as cyclophilin A, is the intracellular receptor for the immunosuppressant cyclosporin A, and functions physiologically as a chaperone or foldase [71]. Though speculative, Arevalo-Martin et al. have previously hypothesized that autoantibodies to PPIA may promote immunosenescence after spinal cord injury [69]. Our results also included a large signature for antibodies directed against ribosomal proteins. Although autoantibodies to three ribosomal proteins (commonly referred to as P0, P1, and P2; encoded by RPLP0, RPLP1, and RPLP2, respectively) have previously been associated with systemic lupus erythematosus [72], autoantibodies to these specific ribosomal proteins were not detected in our analysis. Also of relevance, our analysis did not detect autoantibodies to myelin-associated proteins such as myelin oligodendrocyte glycoprotein (MOG) or myelin basic protein (MBP) which are hallmarks of multiple sclerosis [73].

Our findings also have implications with respect to mechanisms of blast-mediated injury. In our analysis of the thalamic depletome, the most pronounced pathway perturbation was clearly to proteins with links to the myelin sheath. Perturbations involving myelin have previously been observed in differing mouse models of blast-induced TBI [74, 75, 76]. In this regard, the common identification of this pathway hints that although each of these models evokes blast damage in differing ways, that there are none-the-less similarities in some of the basic mechanisms of injury. As reviewed by Fehily and Fitzgerald [10], demyelination or dysmyelination can involve multiple mechanisms, including sensitivity of oligodendrocytes, oligodendrocyte progenitor cells, glutamate excitotoxicity, calcium overload, oxidative stress, and/or altered metabolism, among others.

A second pathway with potential mechanistic relevance, which was specifically identified by WebGestalt, was “L-malate dehydrogenase activity” (2 members, P = 2.6E-3). The malate-aspartate shuttle is a biochemical reaction pathway that is important in regenerating NADH from NAD(+) within the mitochondrial matrix [77]. It has previously been shown that disruption of the NAD(+) pathway results in mitochondrial damage and neuronal death. Analysis of models of moderate brain injury has shown presence of chronic mitochondrial dysfunction and reduced antioxidant activity [78], and that lack of efficient re-synthesis of NAD(+) controlled by NAMPT leads to tissue damage in ischemic tissue [78, 79]. We, and others, have previously shown that treatment with a neuroprotective compound that increases NAD salvage via NAMPT [80] reduces the severity of neuronal damage in blast-exposed mice [12, 13] and mice with fluid-percussion injury [81, 82].

Some proteins identified in our study were previously identified as upregulated in a proteomic analysis of optic nerves from mice subjected to repetitive blunt impact TBI [59]. These proteins included ALB, MAP1A, and TUBA4A. A down-regulation of DPYSL2 was observed following repetitive impact TBI in the optic nerve. While we did not observe a change in DPYSL2 in the thalamus, we did detect auto-antibodies to DPYSL2 in the serum. These models also shared increases in similar classes of proteins, including peroxiredoxins, heat shock proteins and ATPases. Taken together these data suggest that similar molecular mechanisms are responsible for cell death and dysfunction in blast and blunt-trauma mediated TBI.

Our current study had caveats that are important to note. Regarding the injury, it is relevant that brain injury comes in many forms and intensities. Our study focused on mild blast-mediated TBI, which will have a very different presentation and protein profile than moderate to severe TBIs, or penetrating and concussive-mediated TBIs. The mild nature of our injury may explain why classic markers of TBI such as GFAP, UCH-L1 and S-100β were not detected. We have characterized many of the physical features of our blast protocol (including a pressure/time profile, overpressure, positive phase and duration) [83], which will help the comparison of these studies to other models of blast-mediated and non-blast mediated TBI. It should be noted that the visual and cognitive dysfunction we have previously reported using our model [11, 12, 13, 84] is bilateral, suggesting a lack of effect from blunt impacts, and these same dysfunctions are frequently observed in Soldiers and Veterans exposed to blast injury [37, 38, 42, 43, 85, 86]. This study also only focused on a single brain region, whereas past studies have focused globally on the brain or brain-connected fluids [23]. While this does not limit the impact of our study in beginning to define the molecular events that lead to sensory damage after blast-mediated TBI, we do not know if other regions of the brain, or other CNS neurons, respond similarly to blast injury. Our study, performed at a single time point following injury, is also only a “snapshot” of the changes in brain proteins and it does not address how protein expression changes over time. Completion of additional replicates, whether by PELS or by other technologies, will be an important component of future work, in addition to analysis of expression change over time.

Regarding the modeling with mice, our study was conducted in a single strain and sex of mice. We selected the strain (C57BL/6J) because it represents the most widely utilized genetic background of mice. However, our choice of species, genetic background, and sex could all bias the results, and it will be important to extend future studies to females and also different strains and species. Also of relevance, the brain tissue and sera were pooled to analyze proteins that change after blast injury. While it is important to examine the expression profiles of these proteins and auto-antibodies in individual mice, it was beyond the current scope to perform replication studies of candidates within individual mice to assess individual variability of the findings. It should be noted that the method of inducing blast-mediated TBI that was used in this study results in uniform damage among mice exposed to blast. Previously published transcriptomic studies using this model demonstrates very little intra-animal variability in gene expression changes [87].

Finally, regarding our proteomic methodology, it is important to re-iterate that we only sought to identify proteins with increased abundance in response to blast. We did not attempt to identify proteins that might have decreased abundance in response to blast. As with all protein studies, there is the possibility that protein-protein interactions may exist, although we believe that the likelihood of this happening in this study is low, given the high confidence of the identification methods used, the fact that proteins identified in certain fractions were expected, and the fact that our fractionation and differentiation approaches are widely used and designed to reduce this possibility.

In summary, a combined PAD and PELS analysis identified six proteins with properties appropriate for further consideration as biomarkers of blast-mediated TBI (ALDOA, PHKB, HBA-A1, DPYSL2, SYN1, and CKB), as well as several others potentially relevant to the incompletely understood mechanisms of damage caused by blast-exposure. In our opinion, HBA-A1 stands out as a leading candidate for further testing as a biomarker, and cellular responses involving myelin seem likely to be of mechanistic importance, both of which we intend to test in our ongoing work.

Declarations

Author contribution statement

Matthew M. Harper, Manohar John: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Danielle Rudd, Kacie J. Meyer, Edwin Vázquez-Rosa, Min-Kyoo Shin, Kalyani Chaubey, Yeojung Koh, Lucy P. Evans, Michael G. Anderson, Indira T. Kudva: Analyzed and interpreted the data; Wrote the paper.

Anumantha G Kanthasamy, Vellareddy Anantharam: Conceived and designed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Andrew A. Pieper, Alexander G. Bassuk: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Laura Dutca: Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

Funding statement

This work was supported by the Department of Veterans Affairs (RX000952), Department of Defense (W81XWH-14-1-0583), and the Iowa City VA Center for the Prevention and Treatment of Visual Loss. AAP is supported by the Brockman Medical Research Foundation, Hamilton, Bermuda, the Elizabeth Ring Mather & William Gwinn Mather Fund, Cleveland, OH, USA, the S. Livingston Samuel Mather Trust, Cleveland, OH, USA, and Department of Veterans Affairs Merit Review 1IO1BX002444 to AAP. The contents of this manuscript do not represent the views of the U.S. Department of Veterans Affairs, Department of Defense, or the U.S. Government.

Competing interest statement

The authors declare no conflict of interest.

Additional information

No additional information is available for this paper.