The proteomic advantage: label-free quantification of proteins expressed in bovine milk during experimentally induced coliform mastitis

Article (PDF Available)inVeterinary Immunology and Immunopathology 138(4):252-66 · October 2010with90 Reads
DOI: 10.1016/j.vetimm.2010.10.004 · Source: PubMed
Abstract
Coliform mastitis remains a primary focus of dairy cattle disease research due in part to the lack of efficacious treatment options for the deleterious side effects of exposure to LPS, including profound intra-mammary inflammation. To facilitate new veterinary drug approvals, reliable biomarkers are needed to evaluate the efficacy of adjunctive therapies for the treatment of inflammation associated with coliform mastitis. Most attempts to characterize the host response to LPS, however, have been accomplished using ELISAs. Because a relatively limited number of bovine-specific antibodies are commercially available, reliance on antibodies can be very limiting for biomarker discovery. Conversely, proteomic approaches boast the capability to analyze an unlimited number of protein targets in a single experiment, independent of antibody availability. Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS), a widely used proteomic strategy for the identification of proteins in complex mixtures, has gained popularity as a means to characterize proteins in various bovine milk fractions, both under normal physiological conditions as well as during clinical mastitis. The biological complexity of bovine milk has, however, precluded the complete annotation of the bovine milk proteome. Conventional approaches to reducing sample complexity, including fractionation and the removal of high abundance proteins, has improved proteome coverage, but the dynamic range of proteins present, and abundance of a relatively small number of proteins, continues to hinder comparative proteomic analyses of bovine milk. Nonetheless, advances in both liquid chromatography and mass spectrometry instrumentation, including nano-flow liquid chromatography (nano-LC), nano-spray ionization, and faster scanning speeds and ionization efficiency of mass spectrometers, have improved analyses of complex samples. In the current paper, we review the proteomic approaches used to conduct comparative analyses of milk from healthy cows and cows with clinical mastitis, as well as proteins related to the host response that have been identified in mastitic milk. Additionally, we present data that suggests the potential utility of LC-MS/MS label-free quantification as an alternative to costly labeling strategies for the relative quantification of individual proteins in complex mixtures. Temporal expression patterns generated using spectral counts, an LC-MS/MS label-free quantification strategy, corresponded well with ELISA data for acute phase proteins with commercially available antibodies. Combined, the capability to identify low abundance proteins, and the potential to generate temporal expression profiles, indicate the advantages of using proteomics as a screening tool in biomarker discovery analyses to assess biologically relevant proteins modulated during disease, including previously uncharacterized targets.
Veterinary Immunology and Immunopathology 138 (2010) 252–266
Contents lists available at ScienceDirect
Veterinary Immunology and Immunopathology
journal homepage: www.elsevier.com/locate/vetimm
Research paper
The proteomic advantage: Label-free quantification of proteins
expressed in bovine milk during experimentally induced
coliform mastitis
Jamie L. Boehmer
a,
, Jeffrey A. DeGrasse
b
, Melinda A. McFarland
b
, Elizabeth A. Tall
a
,
Kevin J. Shefcheck
b
, Jeffrey L. Ward
a
, Douglas D. Bannerman
c,1
a
U.S. Food and Drug Administration Center for Veterinary Medicine, 8401 Muirkirk Road, Laurel, MD 20708, United States
b
U.S. Food and Drug Administration Center for Food Safety and Applied Nutrition, College Park, MD 20740, United States
c
Bovine Functional Genomics Laboratory, USDA-Agricultural Research Service, Beltsville, MD 20705, United States
article info
Keywords:
Bovine milk proteome
Liquid chromatography/tandem mass
spectrometry (LC–MS/MS)
Coliform mastitis
Label-free quantification
abstract
Coliform mastitis remains a primary focus of dairy cattle disease research due in part to
the lack of efficacious treatment options for the deleterious side effects of exposure to
LPS, including profound intra-mammary inflammation. To facilitate new veterinary drug
approvals, reliable biomarkers are needed to evaluate the efficacy of adjunctive therapies
for the treatment of inflammation associated with coliform mastitis. Most attempts to
characterize the host response to LPS, however, have been accomplished using ELISAs.
Because a relatively limited number of bovine-specific antibodies are commercially avail-
able, reliance on antibodies can be very limiting for biomarker discovery. Conversely,
proteomic approaches boast the capability to analyze an unlimited number of protein tar-
gets in a single experiment, independent of antibody availability. Liquid chromatography
coupled to tandem mass spectrometry (LC–MS/MS), a widely used proteomic strategy for
the identification of proteins in complex mixtures, has gained popularity as a means to
characterize proteins in various bovine milk fractions, both under normal physiological
conditions as well as during clinical mastitis. The biological complexity of bovine milk
has, however, precluded the complete annotation of the bovine milk proteome. Conven-
tional approaches to reducing sample complexity, including fractionation and the removal
of high abundance proteins, has improved proteome coverage, but the dynamic range of
proteins present, and abundance of a relatively small number of proteins, continues to
hinder comparative proteomic analyses of bovine milk. Nonetheless, advances in both liq-
uid chromatography and mass spectrometry instrumentation, including nano-flow liquid
chromatography (nano-LC), nano-spray ionization, and faster scanning speeds and ion-
ization efficiency of mass spectrometers, have improved analyses of complex samples.
In the current paper, we review the proteomic approaches used to conduct comparative
analyses of milk from healthy cows and cows with clinical mastitis, as well as proteins
related to the host response that have been identified in mastitic milk. Additionally, we
present data that suggests the potential utility of LC–MS/MS label-free quantification
as an alternative to costly labeling strategies for the relative quantification of individ-
ual proteins in complex mixtures. Temporal expression patterns generated using spectral
counts, an LC–MS/MS label-free quantification strategy, corresponded well with ELISA data
Corresponding author. Tel.: +1 301 210 4281; fax: +1 301 210 4685.
E-mail address: jamie.boehmer@fda.hhs.gov (J.L. Boehmer).
1
Present address: United States Department of Veterans Affairs, Office of Research Oversight, Washington, DC 20420, United States.
0165-2427/$ see front matter. Published by Elsevier B.V.
doi:10.1016/j.vetimm.2010.10.004
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 253
for acute phase proteins with commercially available antibodies. Combined, the capability
to identify low abundance proteins, and the potential to generate temporal expression pro-
files, indicate the advantages of using proteomics as a screening tool in biomarker discovery
analyses to assess biologically relevant proteins modulated during disease, including pre-
viously uncharacterized targets.
Published by Elsevier B.V.
1. Background
Mastitis caused by gram negative pathogens remains
a principal focus of veterinary research due to stag-
gering economic loss, the limited number of efficacious
treatment options, and the lack of accurate biomarkers
to evaluate the efficacy of new animal drugs proposed
as adjunctive therapies. The need to better understand
the host response to gram negative pathogens, and to
identify reliable biomarkers specific to coliform mastitis,
has led to several investigations into the soluble medi-
ators of innate immunity in the bovine mammary gland
(reviewed in Bannerman, 2009). Historically, characteri-
zation of the bovine innate immune response to LPS, and
the quantification of changes in mediators of inflamma-
tion present in bovine milk during coliform mastitis has,
however, been dominated by the use of ELISAs. While
ELISAs feature both accuracy and specificity, antibody-
based strategies are restricted by the ability to detect and
quantify only one protein at a time, and by a reliance on the
availability or development of species-specific antibodies.
ELISAs, therefore, have little application to the discovery of
novel inflammatory mediators, as currently only a limited
number of bovine-specific antibodies are commercially
available.
Proteomics, defined as the identification and character-
ization of all proteins within a cell or tissue (Colantonio
and Chan, 2005), boasts a significant advantage over ELISAs
in that proteomics involves the use of analytical method-
ologies, such as liquid chromatography (LC) and mass
spectrometry (MS), to isolate, identify, and characterize
proteins, and is not reliant on the use or availability of anti-
bodies. The use of proteomics is also much less restrictive
than ELISAs in that theoretically, an unlimited number of
proteins can be analyzed simultaneously in a given exper-
iment using proteomic strategies. Furthermore, advances
in soft ionization techniques in mass spectrometry, includ-
ing electro-spray ionization (ESI), nano-spray ionization,
and matrix-assisted laser desorption/ionization (MALDI),
have broadened the applications of mass spectrometry to
include the characterization of biopolymers such as intact
proteins and peptides (reviewed in Mann et al., 2001).
Previous studies have utilized proteomic strategies
in attempts to identify protein biomarkers of the host
response present in whey from bovine milk during mastitis
(Boehmer et al., 2008, 2010; Smolenski et al., 2007; Hogarth
et al., 2004). In most cases, however, a limited number of
low abundance proteins have been robustly identified in
comparative proteomic analyses of the bovine milk pro-
teome. Better characterization of extremely low abundance
proteins in bovine milk following infection with a gram
negative pathogen was of specific interest to our group,
because the assessment of the modulation in low abun-
dance proteins during disease could prove useful in the
establishment of biomarkers of coliform mastitis for use
both as diagnostic tools, and as indicators of drug efficacy.
Biomarker discovery in bovine milk has, however,
been hindered both by prominent proteomic bottlenecks,
as well as other experimental factors. The most signif-
icant factor that has precluded the identification of a
larger number of low abundance proteins related to host
response in milk, and one of the most common obsta-
cles in proteomic analyses, is the biological complexity of
the matrix. The analytical challenges associated with the
complexity of milk include protein proteolysis, the numer-
ous post-translational modifications (PTMs) that occur in
milk proteins including glycosylation, phosphorylation,
and disulfide bond formation, as well as the dynamic range
of proteins in milk (Gagnaire et al., 2009; O’Donnell et
al., 2004). The profound relative abundance of a limited
number of proteins in bovine milk is arguably the most
challenging aspect of the proteomic analysis of milk, as the
presence of abundant proteins often prohibits the robust
identification of low abundance components. Compara-
tive analyses of bovine milk are further confounded by the
dynamic range of proteins present in milk, because milk
collected from healthy cows is characterized by the abun-
dance of the caseins and whey proteins -lactoglobulin
and -lactalbumin, while milk collected from cows with
coliform mastitis is marked by the profound increase in
vascular-derived proteins, most notably serum albumin
(Fig. 1).
Conventional approaches to reducing sample complex-
ity prior to analysis, including the selective depletion of
high abundance proteins and fractionation of samples,
have not yet been effectively adapted to address the spe-
cific complexities of bovine milk. For example, attempts to
remove high abundance proteins, including casein deple-
tion by acid precipitation (Hogarth et al., 2004) and the
removal of immunoglobulins by immunoaffinity (Yamada
et al., 2002), have resulted in a rather drastic reduction
in the number of milk proteins identified when compared
to proteomic analyses of bovine milk that did not involve
removal of high abundance proteins (Boehmer et al., 2008;
Smolenski et al., 2007). An additional complication related
to the comparative analyses of bovine milk is that mul-
tiple strategies are required to reduce the complexity of
normal versus mastitic milk samples, as the dynamics of
protein abundance change during inflammation. Likewise,
it has been demonstrated that the removal of serum albu-
min, which would be necessary to reduce the complexity of
mastitic milk samples, can lead to the non-specific deple-
tion of low abundance proteins (Gundry et al., 2007), which
could interfere with biomarker discovery.
254 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
Fig. 1. Differential protein expression in whey from normal milk (A)
and whey from mastitic milk (B) profiled using 2-dimensional gel
electrophoresis (Boehmer et al., 2008). Whey from normal milk is char-
acterized by the abundance of the casein and whey proteins, while whey
from mastitic milk, in contrast, is characterized by the increased abun-
dance of serum albumin and other vascular-derived proteins including
serotransferrin and fibrinogen.
Independent of matrix complexity, another critical
aspect when considering biomarker discovery and the uti-
lization of biomarkers to evaluate disease progression, or
the efficacy of a drug treatment, is the evaluation of protein
changes during the course of clinical disease (Simpson et al.,
2009; Mueller et al., 2008; Old et al., 2005). Regrettably,
most prior studies of protein modulation in the bovine
milk proteome during disease focused solely on normal
versus mastitic milk, rather than changes in protein abun-
dance over the course of infection (Boehmer et al., 2008;
Smolenski et al., 2007; Hogarth et al., 2004).
In biomarker discovery, however, there is still no “gold
standard” for the accurate quantification of individual
proteins in complex biological samples using proteomic
strategies, especially for proteins present in low abundance
(Mueller et al., 2008). Several labeling strategies are avail-
able for the quantification of proteins in conjunction with
LC–MS/MS analyses, but labeling strategies can be cost
limiting, often require pair wise comparisons which can
be problematic when quantifying proteins that are only
present in a given physiological state, and labeling strate-
gies do not allow for retrospective quantification (Old et al.,
2005). Consequently, label-free methods for the relative
quantification of proteins in complex biological samples
have been investigated, including the use of ion intensity,
the number of unique peptides assigned to a given protein,
and spectral counts as measures of relative abundance for
individual proteins in a complex sample (Old et al., 2005;
Zybailov et al., 2005; Liu et al., 2004). Label-free meth-
ods have gained popularity primarily because there are no
associated costs, normalization can allow for comparisons
of protein abundance across a longitudinal set of samples,
and analyses can be conducted retrospectively (Mueller
et al., 2008). A recent study conducted by our group focused
on changes in the relative abundance of milk proteins over
the course of an experimentally induced coliform infection
using peptide counts, a label-free strategy, but the pro-
teins evaluated were mainly medium to high abundance
proteins (Boehmer et al., 2010).
In addition to expanding our knowledge of the bovine
milk proteome, an added objective was to further evalu-
ate the feasibility of using label-free LC–MS/MS strategies
as a screening tool to identify biologically relevant pro-
teins modulated during disease, especially low-abundance
proteins, and proteins for which there are no available anti-
bodies. To assess the validity of using spectral counts to
quantify changes in proteins for which no antibody has
been developed, we conducted a comparison of the expres-
sion of milk proteins determined using LC–MS/MS data
with expression profiles generated using an ELISA, simi-
lar to comparisons made in our previous studies (Boehmer
et al., 2010).
A second interest was the evaluation of samples col-
lected over the course of infection from several biological
replicates. Our aim was the discovery of a reproducible
biomarker or pattern of biomarkers that presented in
several biological replicates, as consistent patterns could
suggest both a response that was indicative of coliform
mastitis, as well as the time frame following infection dur-
ing which the potential biomarker or biomarkers could
be accurately monitored. Accordingly, we sought to deter-
mine if refinements in proteomic methodology, including
investigations into utilizing more advanced approaches
such as a mass spectrometer with a faster scanning speed
and the ability to trap ions, and the use of nano-flow liq-
uid chromatography in-line with nano-spray ionization,
could enable the identification and characterization of a
greater number of low abundance proteins when com-
pared to earlier analyses of whey from mastitic bovine
milk (Boehmer et al., 2008, 2010; Smolenski et al., 2007;
Hogarth et al., 2004). Following recent analyses detailed in
the current paper, the number of low abundance proteins
identified suggests that proteomics could lead to a more
thorough annotation of the bovine milk proteome. Addi-
tionally, the close correspondence of LC–MS/MS label-free
data and ELISA data was a positive indication that pro-
teomic strategies could serve as valuable screening tools
for biomarker discovery, as well as the establishment of
biomarkers specific to coliform mastitis.
2. Proteomic tools, strategies and challenges
Protein identification through the use of mass spec-
trometry can be divided into two main categories, referred
to as top-down and bottom-up. The primary distinguish-
ing features between the two main proteomic approaches
is the isolation and fragmentation of intact proteins using
mass spectrometry in a top-down approach, versus prote-
olytic digestion of mixtures of proteins, and the subsequent
separation and fragmentation of peptides, in bottom-up
proteomics. Identification of proteins in complex biological
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 255
mixtures using bottom-up proteomics is reliant upon the
measurement of the masses of the peptides that are gen-
erated following proteolytic cleavage of the proteins. The
mass of a peptide is determined using mass spectrometry,
and is based upon a mass: charge ratio (m/z). Charged pep-
tides are generated as a result of ionization, or the addition
of a proton to the peptide, which results in the conversion of
the peptide into an ion. The two most popular forms of ion-
ization used in bottom-up proteomic analyses are ESI and
MALDI. The primary features that distinguish MALDI from
ESI are the matrices used for ionization, the charge states of
the ions generated, and the actual mechanisms of ion for-
mation characteristic of each method (reviewed in Mann
et al., 2001). Similar in concept and ion formation to ESI,
nanospray is another ionization method that has become
very popular in proteomic analyses in recent years. The
primary distinctions between ESI and nanospray are the
significantly lower flow rates and smaller needle diameters
used for nano-spray. An advantage of nano-spray ioniza-
tion is that nano-spray sources can accommodate much
lower flow rates than ESI, down to fractions of microliters
per minute (Wilm and Mann, 1996). Additionally, droplet
formation occurs more readily using nano-spray, resulting
in increased ionization efficiency. The lower flow rates in
the nanospray technique also allow for a longer length of
analysis time, which leads to fewer missed peptides eluting
off the chromatographic column while the mass spectrom-
eter is engaged in MS/MS scans (Wilm and Mann, 1996).
Since the invention of soft ionization techniques,
bottom-up proteomic strategies including the use of LC
to separate peptides coupled with MS/MS for peptide
sequencing, a process commonly referred to as LC–MS/MS,
has become the most extensively applied bottom-up
proteomic approach for the identification of individual pro-
teins in complex mixtures. LC–MS/MS involves proteolytic
digestion of complex protein mixtures followed by the
separation of peptides using one- or two-dimensional LC,
and analysis of the peptides by MS/MS (reviewed in Mann
et al., 2001). An enzyme commonly used to cut proteins
into peptides is tryspin, which digests at the amino acid
residues arginine and lysine. Peptide mixtures are sepa-
rated, prior to introduction into the mass spectrometer
for mass analysis, by passage over a chromatographic col-
umn and subsequent separation based on either charge,
called ion exchange LC, or hydrophobicity, which is termed
reverse phase (RP) LC. The number of proteins identified
using LC–MS/MS is directly dependent on the efficiency
of peptide separation (reviewed in Mann et al., 2001).
In data-dependent acquisitions, the mass spectrometer is
programmed to scan the masses of ionized peptides and
to select anywhere from 3 to 10 most abundant peptides
for further fragmentation by collision-induced dissociation
(CID). An inert gas introduced into the collision cell of the
mass spectrometer during CID induces fragmentation of
the peptides, a process which results in the production of
a tandem mass spectrum. Peak lists distilled from tandem
mass spectra are used to query against an MS/MS spectra
database to determine peptide identity, and the sequenced
peptides are assigned to proteins for protein identification.
When the chromatographic separation of peptides is
poor, the potential for selection of a peptide from a low
abundance protein for CID will decrease, due to the fact that
peptides from dominant proteins will be in greater num-
bers in the sample and will be preferentially selected for
further fragmentation. Additionally, poorly resolved pep-
tides tend to co-elute off the chromatographic column into
the mass spectrometer, a phenomena which leads to CID of
more than one peptide at a time. The tandem mass spec-
trum that results from co-eluting peptides thus represents
the fragmentation of a peptide mixture, and will often fail
to yield a match when searched against a protein database,
or will lead to a false positive peptide assignment.
2.1. Biological complexity and proteomic bottlenecks
Proteomic strategies, though capable of analyzing
a theoretically unlimited number of proteins in a
single experiment, are not devoid of challenges. Post-
translational modifications (PTMs) of proteins in a given
proteome, and the dynamic range of proteins present in
the sample, are direct reflections of the complexity of the
biological matrix, and can pose significant roadblocks to
protein identification. Dynamic range is one of the most
prominent bottlenecks in proteomic experiments because
many biological samples, including milk, are characterized
by the presence of a select number of highly abundant pro-
teins that account for a large percentage of the total protein
concentration in the sample, and numerous low abundance
proteins that comprise a very small percentage of protein
concentration (Gagnaire et al., 2009; O’Donnell et al., 2004).
Given the fact that abundant proteins are often affiliated
with a variety of biological functions and pathways, and
thus rarely meet the specificity criteria necessary to be
termed a biomarker of disease, the removal or depletion
of abundant proteins is a common first step in proteomic
analyses aimed at biomarker discovery.
In some cases, however, removal of abundant proteins
from a complex matrix can also result in the non-selective
depletion of low abundance proteins, a consequence which
can cause the loss of potentially relevant diagnostic, clin-
ical, and biological information. Albumin, which accounts
for nearly 55% of the total protein concentration of plasma,
is often targeted for removal prior to proteomic analy-
sis. Investigation into the albuminome, or the number of
proteins that bind to, or are associated with, albumin in
plasma and are thus depleted along with the abundant
protein following the application of depletion strategies,
revealed that as many as 35 high and low abundance pro-
teins were bound to and removed along with albumin
following an affinity removal process (Gundry et al., 2007).
The increased concentration of albumin in milk during col-
iform mastitis, due to the breakdown of the blood–milk
barrier following exposure to LPS, presents a significant
analytical roadblock for the identification of low abun-
dance proteins. Abundant albumin peptides often mask the
detection of low abundance proteins, and albumin deple-
tion could result in the loss of low abundance proteins that
potentially bind to albumin in milk.
There are many strategies available commercially that
are designed to remove or deplete high abundance
proteins, most notably serum albumin, from biological
samples in order to enhance the detection of low abun-
256 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
dance proteins. Albumin removal strategies often involve
an analytical column or disc pre-packed with a form of
Cibacron Blue, a sulfonated triazine dye used for affin-
ity chromatography, immobilized onto a support matrix
(Angal and Dean, 1977). For the removal of several
high abundance proteins simultaneously, multiple affinity
removal system (MARS) affinity spin and chromatographic
columns are commercially available from Agilent Tech-
nologies. Low abundance protein enrichment strategies
are also commercially available, including a widely used
product from Bio-Rad Laboratories called ProteoMiner.
ProteoMiner columns consist of beads containing a highly
diverse library of hexapeptides, each with a specific protein
affinity, bound to chromatographic supports. The theory
behind ProteoMiner is that proteins in a given biological
sample, when passed over the beads, will bind to spe-
cific ligands, out-competing high abundance proteins, and
allowing excess proteins to wash off the column as flow-
through. Proteins bound to the beads on the ProteoMiner
sample column are eluted, and the resulting protein pool
is predicted to contain a more equivalent representation of
both high and low abundance proteins present in the given
sample. Nearly all of the depletion, removal, or enrich-
ment strategies that are available, however, have been
optimized, and are intended, for use with human serum
or plasma. Used according to the manufacturer’s recom-
mended protocol, the ProteoMiner enrichment strategy
was extremely effective at depleting serum albumin from
mastitic milk samples, but also resulted in the depletion of
several other milk proteins (Fig. 2). While the enrichment
for low abundance targets in mastitic milk was effective for
proteins between 10 and 15 kD, most of the spots in that
range were identified as proteolysis products of the casein
proteins (data not shown). In order to make effective use of
commercially available strategies for the reduction of sam-
ple complexity, optimization would have to be performed
for milk samples specifically, which might not be econom-
ically feasible for some studies, given the high cost of many
of the kits.
In addition to serum albumin, several proteins whose
presence in blood increases during the inflammatory
response such as complement, clotting factors, adhesion
molecules, and acute phase proteins increase in concen-
tration in milk during coliform mastitis, due to the well
characterized break-down of the blood–milk barrier. Many
of the proteins that leak into the milk from systemic cir-
culation are very large glycoproteins that become heavily
modified during the course of infection (Kjeldsen et al.,
2003; Soerensen et al., 1995), an aspect that further com-
plicates analytical challenges. Accurate identification of
modified proteins can require specialized sample prepara-
tion prior to analysis, the inclusion of numerous potential
variable modifications when querying peak lists against
protein databases, or the use of fragmentation strategies
other than CID, including electron-transfer dissociation
(ETD).
While sample complexity reduction strategies may
not be feasible approaches for enriching low abundance
targets in milk, proteomic capabilities, including both
instrumentation and fractionation options, continue to
advance. Utilizing the features of different instrument
Fig. 2. The two dimensional profiles of whey from mastitic bovine milk
following protein enrichment using ProteoMiner (BioRad Laboratories).
The abundance of the protein serum albumin, indicated in box 1 with the
white arrow, is clearly lower in the profile of proteins eluted off the beads
containing a highly diverse library of hexapeptides (A) when compared
to the profile of the proteins in the flow-through that did not bind to the
beads (B). Likewise, smaller proteins at the bottom of the gel, in box 2,
appeared to be more abundant in the protein pool eluted off the beads
(A) when compared to the column flow through (B). Also apparent is the
fact that the majority of the proteins appear in the flow through (B), as
opposed to beingenrichedby binding to the beads (A). The use of strategies
such as ProteoMiner, which was developed and optimized for use with
plasma and serum samples, may not be feasible or may require added
optimization steps, when used on complex biological samples such as
bovine milk.
systems, such as mass spectrometers with faster scan-
ning speeds and increased ion transmission capabilities,
in lieu of instruments with higher mass accuracy, could
lead to more protein identifications, as well as targets for
more focused analyses aimed at quantification or mass
accuracy. Nano-flow liquid chromatography in-line with a
mass spectrometer equipped with a nano-spray ionization
source could likewise result in more robust identification
of low abundance proteins, as nano-spray ionization has
demonstrated advantages over traditional ESI for protein
identification (Juraschek et al., 1998). Advances in LC and
nano-flow LC, including two-dimensional LC separation
strategies, could also drastically improve peptide separa-
tions and lead to additional protein identifications.
2.2. Quantification
An important criterion for the establishment of quality
biomarkers is reliable quantification. Accordingly, relative
and absolute quantification of changes in biomarkers in
biological matrices using proteomic strategies is a topic
that has garnered significant attention in recent years
(Simpson et al., 2009; Mueller et al., 2008; Fenselau,
2007; Roe and Griffin, 2006). Quantification methods can
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 257
be assigned to one of two broad categories: a labeling
approach that requires the incorporation of labels into
proteins or peptides prior to MS analysis, or the use of a
label-free method (Simpson et al., 2009).
The basis of most labeled quantification methods is the
theory that a labeled peptide will behave chemically in the
same fashion as its unlabeled counterpart, and will have
identical chromatographic and MS properties (Simpson
et al., 2009). The addition of a label, however, does result in
a mass difference between the two peptides, and thus rela-
tive abundance can be inferred by comparing the respective
signal intensities of the labeled and unlabeled peptides in
the same MS run (Simpson et al., 2009). Labels can be incor-
porated in several ways, with the most popular being either
metabolically, chemically, or enzymatically (Simpson et al.,
2009; Fenselau, 2007). A popular means of metabolically
incorporating a label is stable isotope labeling by amino
acids in cell culture, or SILAC. The primary advantage of
SILAC is that isotope labels can be incorporated into all
cellular proteins, because cells are cultured with nutrients
that are highly enriched with stable isotopes (Jiang and
English, 2002; Gu et al., 2002; Oda et al., 1999). Different
cell populations can then be combined following labeling in
culture, and the relative abundances of individual proteins
in the distinct cell populations determined by measuring
the isotope ratios of peptide pairs with mass spectrome-
try (Fenselau, 2007). Metabolic labeling is, however, only
applicable to the analysis of bacterial and eukaryotic cells,
and cannot be adapted to accommodate clinical or animal
studies.
Other labeling strategies that can be extended to a wider
variety of biological samples include proteolytic labeling
with O
18
, in which two atoms of O
18
are added to the
carboxylic acid group of every peptide in a protein pool
following proteolysis (Yao et al., 2003). Similar to SILAC,
relative protein concentrations are determined by mea-
suring isotope ratios of O
18
and O
16
labeled peptide pairs
with mass spectrometry. Stable isotopes can also be incor-
porated by chemical derivatization at either the protein
or peptide level. Two popular examples of labeling strate-
gies that are based on chemical reactions are isotope coded
affinity tags (ICAT; Gygi et al., 1999), and isobaric tags for
relative and absolute quantification (iTRAQ; Ross et al.,
2004). Both ICAT and iTRAQ are based on the derivatization
of the primary amine groups of proteins or peptides. Using
ICAT, samples are either labeled with an isotopically light
probe or an isotopically heavy version, combined, digested
with a protease (trypsin), and then labeled peptides are iso-
lated by avidin affinity chromatography and analyzed by
LC–MS (Gygi et al., 1999). Using iTRAQ, however, enables
users to comparatively analyze four or more different pro-
tein pools simultaneously, unlike most labeling strategies
that are limited to the analysis of two distinct protein pop-
ulations (Ross et al., 2004). With iTRAQ, the proteins or
peptides from different samples or treatments are deriva-
tized with different tags that all have the same total mass.
The labeled peptides are then pooled and analyzed using
LC–MS/MS. Peptides from the different protein pools can
be differentiated and relatively quantified following the
fragmentation of the attached tag which generates a low
molecular mass reporter ion (Ross et al., 2004). Because
of the number of differential protein pools that can be
simultaneously analyzed, iTRAQ is a very popular labeling
strategy commonly used in proteomic screens, including
the analysis of mastitis pathogens (Lippolis et al., 2009),
the bovine milk fat globular membrane (Reinhardt and
Lippolis, 2008), and bovine milk following in vivo LPS chal-
lenge (Danielsen et al., 2010).
In comparative proteomic analysis of normal versus
mastitic milk aimed at biomarker discovery, however,
labeling strategies are not always feasible for protein quan-
tification, depending on the design and focus of the study.
Labeling strategies can be extremely cost-limiting, which
leads to the analysis of only a small number of samples
from a limited number of biological replicates (Danielsen
et al., 2010), and the buffers that are required during the
labeling steps are not always amenable to maintaining all
of the proteins present in milk in solution. Additionally, the
majority of labeling strategies are based on the pair-wise
comparison of the relative intensities of peptides generated
from a target protein in two different physiological states
(i.e. healthy versus diseased). Since many of the target pro-
teins in comparative proteomic analyses of mastitic milk
are not present in normal or control samples of milk, but
only during a mastitis infection, accurately comparing the
abundance of a peptide that is present at one physiologi-
cal state but not the other is problematic. Because labeling
strategies are well adapted to targeted analyses but not dis-
covery screens, an interest in quantification without the
incorporation of labels has emerged in the field of pro-
teomics.
Label-free strategies are based on the correlation
between the abundance of a protein or peptide in a sam-
ple, and the MS signal (Simpson et al., 2009). One of the
most popular methods of label-free quantification is ion
intensity, determined using extracted ion chromatograms
(XIC), in which the number and intensity of selected pre-
cursor ions at a particular m/z are summed, and the peak
areas used as a measure of relative abundance (Old et al.,
2005). An alternative approach that is gaining popularity,
however, is spectral counting, or the number of MS/MS
spectra that contribute to the identification of a given pro-
tein (Zybailov et al., 2005; Liu et al., 2004). The theory
behind spectral counting in particular is that the abun-
dant proteins, when proteolytically digested, will yield
numerous copies of the same peptide. Furthermore, the
probability that abundant peptides will trigger multiple
MS/MS events is higher than the likelihood of repeatedly
sampling a peptide from a lower abundance protein. In
previous investigations into the accuracy and linearity of
spectral counts, the spectral counts for peptides from pro-
teins spiked into yeast samples at known concentrations
exhibited linearity over 2 orders of magnitude, and were
highly correlated to relative protein abundance (Liu et al.,
2004). The number of unique peptides identified for each
protein in a sample has likewise been used as a measure of
relative protein abundance (Liu et al., 2004). Similar to the
theory behind the utility of spectral counts as a measure of
relative protein abundance, the assumption with number
of unique peptide assignments is that more abundant pro-
teins will naturally have greater sequence coverage than
lower abundance proteins, based on the fact that a greater
258 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
number of peptides from abundant proteins will be avail-
able for MS/MS sampling in a given peptide pool.
The inherent drawbacks of label-free methods, how-
ever, include the assumption that the linearity of response
will be the same for each protein, which often does not
hold true because the chromatographic behavior of pep-
tides tend to vary (Simpson et al., 2009). Likewise, the
amino acid composition of every peptide differs, and as a
result, the ionization potential of each peptide generated in
a bottom-up proteomic experiment is unique. Thus, some
peptides will naturally ionize more efficiently than others,
regardless of abundance. Additionally, label-free strategies
are almost exclusively based on the identification of pep-
tides that are unique to a given protein. In the instance
of highly conserved proteins that share a specific func-
tional domain, such as the iron binding domain present in
both lactoferrin and transferrin, differentiation of peptides
belonging to closely related proteins can be challenging
unless the mass spectrometer employed in the analyses
offers increased sensitivity and high mass accuracy. Nor-
malization of XIC, and spectral or peptide counts, however,
can help account for differences in chromatography and
ionization potentials of peptides when using label-free
methods (Old et al., 2005). Finally, label-free quantifica-
tion does not require any extra sample processing and can
be performed retrospectively; two attributes which make
non-labeled quantification methods the continued focus
of development in biomarker discovery research (Simpson
et al., 2009).
3. Comparative proteomic analysis of whey from
bovine milk
3.1. Two-dimensional gels with MALDI-TOF mass
spectrometry
Some of the earliest comparative proteomic anal-
yses of normal versus mastitic whey from bovine
milk were accomplished using 2-dimensional gel elec-
trophoresis (2D-GE) followed by matrix-assisted laser
desorption/ionization time-of-flight (MALDI-TOF) mass
spectrometry (Smolenski et al., 2007; Hogarth et al., 2004),
or MALDI-TOF/TOF post-source decay (PSD; Boehmer et al.,
2008). The number of proteins identified in the studies that
utilized 2D-GE followed by MALDI-TOF mass spectrome-
try (MS) was thirty-one, sixteen of which were detected
only in whey from milk collected from cows with clini-
cal mastitis (Table 1). Though somewhat limited, results
of the 2D gel-based comparative proteomic analyses of
normal versus mastitic whey from milk did contribute
to our overall knowledge of protein modulation during
mastitis, and extended the number of proteins related to
host response identified in milk during a mastitis infec-
tion. Interestingly, the two studies that did not attempt
the removal of high abundance proteins prior to analy-
sis (Boehmer et al., 2008; Smolenski et al., 2007) resulted
in the identification of a greater overall number of pro-
teins, including a higher number of proteins related to
the host response, than the study that attempted the
Table 1
Summary of proteins detected in bovine milk using MALDI-TOF MS.
Swiss-Prot entry name Primary accession number Protein name Condition present
a
Reference
b
PIGR BOVIN P81265 Polymeric-immunoglobin receptor N, M 1, 3
ALBU
BOVIN P02769 Serum albumin N, M 1, 2, 3
CASA1
BOVIN P02662 -S1-casein N, M 1, 2, 3
CO3
BOVIN Q2UVX4 Complement C3 N, M 1
MFGM
BOVIN Q95114 Lactadherin N 1, 3
CASK
BOVIN P02668 -Casein N 1, 2, 3
CASB
BOVIN P02666 -Casein N 1, 2, 3
LACB
BOVIN P02754 -Lactoglobulin N 1, 2, 3
NPC2
BOVIN P79345 Epididymal secretory protein E1 precursor N 1
LALBA
BOVIN P00711 -Lactalbumin N 1, 2, 3
TTHY
BOVIN O46375 Transthyretin N, M 1
B2MG
BOVIN P01888 -2-Microglobulin N 1
FABPH
BOVIN P10790 Fatty acid-binding protein, heart N 1
A1AG
BOVIN Q3SZR3 -1-Acid glycoprotein N, M 1
TRFE
BOVIN Q29443 Serotransferrin M 1, 2, 3
TRFL
BOVIN P24627 Lactoferrin M 3
A1AT
BOVIN P34955 -1-Antitrypsin M 1
FETUA
BOVIN P12763 -2-HS-glycoprotein M 1
FIBB
BOVIN P02676 Fibrinogen -chain M 1
CO4
BOVIN P01030 Complement C4 precursor M 1
APOA1
BOVIN P15497 Apolipoprotein A-I precursor M 1
CTHL1
BOVIN P22226 Cyclic dodecapeptide precursor M 1
GLCM1
BOVIN P80195 Glyearn-1 N, M 3
CTHL4
BOVIN P33046 Indolicidin precursor M 1
CTHL3
BOVIN P19661 Bactenecin-7 precursor M 1
CTHL2
BOVIN P19660 Bactenecin-5 precursor M 1
APOA2
BOVIN P81644 Apolipoprotein A-II precursor M 1
ACTB
BOVIN P60712 Actin, cytoplasmic 1 M 3
NUCB1
BOVIN Q0P569 Nucleobindin M 3
APOC3
BOVIN P19035 Apolipoprotein C-III M 1
S10AC
BOVIN P79105 Protein S100-A12 M 1
a
Condition present; N = normal bovine milk; M = mastitic bovine milk.
b
References; 1 = Boehmer et al. (2008);2=Hogarth et al. (2004);3=Smolenski et al. (2007).
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 259
depletion of the highly abundant caseins (Hogarth et al.,
2004).
Other promising discoveries that resulted from the 2D-
GE of whey from bovine milk included the identification
of the acute phase protein (APP) -1-acid glycoprotein
(A1AG) in both normal and mastitic whey samples, and
the apparent higher relative abundance of A1AG in mas-
titic milk (Boehmer et al., 2008). Previous analyses of APP
expression in milk during bovine mastitis had only iden-
tified the acute phase proteins SAA, lipopolysaccharide
binding protein (LBP), and haptoglobin (Hiss et al., 2004;
Bannerman et al., 2004; Eckersall et al., 2001). In addition
to the identification of A1AG, proteins with antimicrobial
properties not identified in other comparative proteomic
analyses of bovine milk were discovered, including several
members of the cathelicidin family of cationic antimicro-
bial proteins (Boehmer et al., 2008). However, despite the
identification of proteins related to host response, includ-
ing previously unreported acute phase and antimicrobial
proteins, a very limited number of proteins were identified
using 2D-GE coupled with MALDI-TOF MS. Additionally,
though 2D-GE is a popular method for separation and
quantifying proteins in complex biological samples, the
electrophoresis approach and quantification of proteins
via densitometry alone is limited in sensitivity, and prob-
lematic for proteins that are insoluble or have either a
very high or low molecular weight (Old et al., 2005). Like-
wise, protein recovery from in-gel tryptic digestion can
be limited, which reduces the probability of robust pro-
tein identification using MALDI-TOF MS. Lastly, the poor
reproducibility of 2D gels, and the time required to gen-
erate 2-dimensional protein profiles do not lend well to
high throughput biomarker discovery or to the analysis of
protein modulation over the course of clinical mastitis.
3.2. LC–MS/MS of whey from normal and mastitic bovine
milk
In total, 71 proteins (Table 2) have been identified in
comparative analyses of whey from bovine milk using
bottom-up LC–MS/MS strategies (Danielsen et al., 2010;
Boehmer et al., 2008, 2010; Smolenski et al., 2007). The very
first LC–MS/MS-based proteomic analysis of whey, skim
milk, and the milk fat globular membrane from mastitic
bovine milk resulted in the identification of 24 defense-
related proteins, in addition to the abundant casein and
whey proteins (Smolenski et al., 2007). Of the 24 proteins
identified, however, only 6 proteins related to the host
response were identified in whey using direct LC–MS/MS.
Additionally, a major limitation of the study was the fact
that milk from only one cow with a naturally occurring case
of clinical mastitis was analyzed, and the study was not
comparative, as proteins were only identified in mastitic
milk and not in normal milk from the same cow. Despite
the lack of biological replicates and control milk samples,
however, the study marked the first comparative analyses
of several different mastitic milk fractions, and was the first
to report the identification of several low abundance milk
proteins (Smolenski et al., 2007).
In our first proteomic analyses of a longitudinal series
of bovine milk samples collected from 8 mid-lactation
cows before and over the course of experimental infection
with E. coli, a traditional bottom-up LC–MS/MS approach
was applied using an ultra pressure LC instrument (UPLC)
coupled to a quadrupole time-of-flight (Q-TOF) mass spec-
trometer (Boehmer et al., 2010). Similar to earlier analyses,
only 6 proteins related to host response were identified in
whey from milk following E. coli challenge, including trans-
ferrin, lactoferrin, and the cationic antimicrobial proteins
cathelicidins-1 and peptidoglycan recognition receptor
protein. Nonetheless, temporal expression of the proteins
identified in whey from milk before, and in the time points
following the experimental induction of coliform mastitis,
was evaluated using both the number of unique pep-
tides assigned to each identified protein, and the spectral
count, or the number of times each peptide triggered an
MS/MS event. The number of unique peptides identified
and the spectral counts were nearly identical for each pro-
tein identified which, in combination with the extremely
limited number of proteins identified, indicated that the
analytical methods lacked adequate sensitivity. Addition-
ally, there was a large gap in the number of peptides
identified for abundant whey proteins, and the number of
peptide assignments for the remaining proteins identified
(Boehmer et al., 2010).
Due to the inherent lack of sensitivity achieved using the
previously described instrument system (Boehmer et al.,
2010), whey from bovine milk before and after challenge
with E. coli was analyzed using one-dimensional nano-
LC followed by nano-spray tandem MS. A linear ion-trap
mass spectrometer was chosen in order to evaluate the
advantage of the fast scanning speed of the instrument,
and nano-spray was applied to potentially improve ioniza-
tion efficiency. Across 8 biological replicates, a total of 33
different proteins were detected in milk samples collected
from 4 or more biological replicates and at a minimum of 2
of the 5 time points analyzed (Table 3). Of the 33 proteins
identified, 12 proteins were detected across all time points,
while 21 were detected only following infection. In addi-
tion to the major whey and casein proteins, also detected
in all biological replicates across all experiments were
the proteins lactoferrin, lactophorin (glycam-1), osteo-
pontin, polymeric immunoglobulin receptor (PIGR), and
butyrophilin. An additional 33 proteins were identified by 2
or more peptides including lactoperoxidase, prothrombin,
apolipoprotein A-IV, alpha-1-acid glycoprotein, comple-
ment C4, xanthine dehydrogenase, and plasminogen, but
the proteins were detected in fewer than 4 biological repli-
cates and thus temporal expression was not evaluated
(Supplemental Table 1).
Unlike the previous analysis (Boehmer et al., 2010),
the number of spectral counts was greater for most pro-
teins than the number of unique peptides identified, which
indicated that the fast scanning speed of the linear ion
trap allowed for more accurate relative quantification of
a greater number of proteins (Supplemental Table 2).
Additionally, though the faster scanning speed of the
instrument resulted in lower resolution data, a greater
number of proteins were identified, even after low qual-
ity spectra were eliminated from the data set (Table 3).
As would be expected considering no depletion of high
abundance proteins was performed, higher spectral counts
260 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
Table 2
Summary of proteins identified in whey from normal or mastitic bovine milk using LC–MS/MS.
Swiss-Prot entry name Primary accession number Protein Condition present
a
Reference
b
ALBU
BOVIN P02769 Serum albumin N, M 1, 2, 3, 4
LACB
BOVIN P02754 -Lactoglobulin N, M 1, 2, 3, 4
LALBA
BOVIN P00711 -Lactalbumin N, M 1, 2, 3, 4
CASA1
BOVIN P02662 -S1-casein N, M 1, 2, 3, 4
CASA2
BOVIN P02663 -S2-casein N, M 1, 2, 3, 4
CASB
BOVIN P02666 -Casein N, M 1, 2, 3, 4
CASK
BOVIN P02668 -Casein N, M 1, 2, 3, 4
TRFL
BOVIN P24627 Lactoferrin N, M 1, 2, 3, 4
GLCM1
BOVIN P80195 Glycam-1 (lactophorin) N, M 1, 2, 4
OSTP
BOVIN P31096 Osteopontin N, M 1
PIGR
BOVIN P81265 Polymeric immunoglobulin receptor N, M 1, 4
BT1A1
BOVIN P18892 Butyrophylin N, M 1
PGRP
BOVIN Q8SPP7 Peptidoglycan recognition receptor protein M 1, 2, 3
CTHL1
BOVIN P22226 Cathelicidin-1 M 1, 2, 3
CTHL2
BOVIN P19660 Cathelicidin-2 M 1, 3
CTHL4
BOVIN P33046 Cathelicidin 4 M 1, 3
ACTB
BOVIN P60712 Actin, cytoplasmic-1 M 1, 4
TRFE
BOVIN Q29443 Transferrin M 1, 2, 3, 4
APOA1
BOVIN P15497 Apolipoprotein A1 M 1, 3
APOA2
BOVIN P81644 Apolipoprotein A2 M 1, 3
APOE
BOVIN Q03247 Apolipoprotein E M 1
APOC3
BOVIN P109035 Apolipoprotein C3 M 1
APOA4
BOVIN Q32PJ2 Apolipoprotein A-IV M 1, 3
APOH
BOVIN P17690 Beta-2-glycoprotein-1 M 1
VTDB
BOVIN Q3MHN5 Vitamin D binding protein M 1
CO3
BOVIN Q2UVX4 Complement C3 M 1, 3
CFAB
BOVIN P81187 Complement factor B M 1
CO4
BOVIN P01030 Complement C4 M 1, 3
Q693V9
BOVIN Q693V9 Complement component 3d M 3
FETUA
BOVIN P12763 Alpha-2-HS-glycoprotein M 1, 3
KNG1
BOVIN P01044 Kininogen 1 M 1, 3
KNG2
BOVIN P01045 Kininogen 2 M 1
B2MG
BOVIN P01888 Beta-2-microglobulin M 1
CLUS
BOVIN P17697 Clusterin M 1, 3
SAA
BOVIN P35541 Serum amyloid A M 1, 3
ITIH4
BOVIN Q3T052 Inter-alpha trypsin inhibitor heavy chain 4 M 1
A1AT
BOVIN P34955 -1-Antitrypsin M 3
A1AG
BOVIN Q3SZR3 Alpha-1-acid glycoprotein M 1
HPT
BOVIN Q2TBU0 Haptoglobin M 1, 3
ITIH1
BOVIN Q0VCM5 Inter-alpha-trypsin inhibitor heavy chain H1 M 1
FIBA
BOVIN P02672 Fibrinogen alpha chain M 1, 3
FIBB
BOVIN P02676 Fibrinogen beta chain M 1, 3
FIBG
BOVIN P12799 Fibrinogen gamma chain M 1, 3
PERL
BOVIN P80025 Lactoperoxidase M 1
PROF1
BOVIN P02584 Profilin-1 M 1
THRB
BOVIN P00735 Prothrombin M 1
A2MG
BOVIN Q7SIH1 Alpha-2-macroglobulin M 1, 3
HEMO
BOVIN Q3SZV7 Hemopexin M 1
NUCB1
BOVIN Q0P569 Nucleobindin-1 M 1
ANXA1
BOVIN P46193 Annexin A1 M 3
CH3L1
BOVIN P30922 Chitinase-3-like protein 1 M 1
LIPL
BOVIN P11151 Lipoprotein lipase M 1
XDH
BOVIN P80457 Xanthine dehydrogenase/oxidase M 1
PTGDS
BOVIN O02853 Prostaglandin-H2 d-isomerase M 1
PLMN
BOVIN P06868 Plasminogen M 1, 3
GELS
BOVIN Q3SX14 Gelsolin M 1
FETA
BOVIN Q3SZ57 Alpha-fetoprotein M 1
MFGM
BOVIN Q95114 Lactadherin M 1
S10A8
BOVIN P28782 Protein S100-A8 M 1
CAP1
BOVIN Q3SYV4 Adenyl cyclase-associated protein 1 M 1
ANT3
BOVIN P41361 Antithrombin-III M 1, 3
CRF
BOVIN Q3MHN5 Corticoliberin M 1
ENOA
BOVIN Q9XSJ4 Alpha-enolase M 1
S10A9
BOVIN P28783 Protein S100-A9 (Calgranulin B) M 3
Q5J801
BOVIN Q5J801 Endopin 2B M 3
PGK1
BOVIN Q3T0P6 Phosphoglycerate kinase-1 M 1
COF1
BOVIN Q5E9F7 Cofilin-1 M 1
G3P
BOVIN P10096 Glyceraldehyde-3-phosphate dehydrogenase M 1
B4GT1
BOVIN P08037 -1,4 Galactosyltransferase-1 M 1
FOLR1
BOVIN P02702 Folate receptor alpha M 1
FIT2 BOVIN A4IFN5 Fat storage-inducing transmembrane protein 2 M 1
a
Condition present; N = normal bovine milk; M = mastitic bovine milk.
b
References; 1 = Boehmer et al. (current paper); 2 = Boehmer et al. (2010);3=Danielsen et al. (2010);4=Smolenski et al. (2007).
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 261
Table 3
Summary of proteins detected following the proteomic analysis of normal and mastitic milk.
Swiss-Prot
entry name
Primary accession
number
Protein
a
Biological function
Proteins present before and after infection
ALBU
BOVIN P02769 Serum albumin Main plasma protein; binds Ca2+, Na+, K+, fatty acids
LACB
BOVIN P02754 -Lactoglobulin Major whey protein; binds and transports retinol
LALBA
BOVIN P00711 -Lactalbumin Involved in lactose synthesis; regulatory subunit of lactose synthase
enzyme
CASA1
BOVIN P02662 -S1-casein Major milk protein; transports calcium phosphate
CASA2
BOVIN P02663 -S2-casein Transports calcium phosphate; inhibits the growth of E. coli
CASB
BOVIN P02666 -Casein Acts as a macrophage activator and as a bradykinin-potentiating peptide
CASK
BOVIN P02668 -Casein Stabilizes micelle formation
TRFL
BOVIN P24627 Lactoferrin Iron-binding protein; has antimicrobial activity
GLCM1
BOVIN P80195 Glycam-1 (Lactophorin) Mediates trafficking of lymphocytes to lymph nodes
OSTP
BOVIN P31096 Osteopontin Enhances production of interferon-gamma and interleukin 12
PIGR
BOVIN P81265 Polymeric immunoglobulin
receptor
Receptor that binds polymeric IgA and IgM at the basolateral surface of
epithelial cells
BT1A1
BOVIN P18892 Butyrophylin Functions in the secretion of milk-fat droplets
Proteins present only after infection
PGRP
BOVIN Q8SPP7 Peptidoglycan recognition
receptor protein
Involved in innate immunity; microbicidal for gram (+) and gram ()
bacteria
CTHL1
BOVIN P22226 Cathelicidin-1 Potent microbicidal activity against E. coli
CTHL2
BOVIN P19660 Cathelicidin-2 Potent antimicrobial activity against E. coli
CTHL4
BOVIN P33046 Cathelicidin 4 Potent microbicidal activity against E. coli
ACTB
BOVIN P60712 Actin, Cytoplasmic-1 Involved in various types of cell motility
TRFE
BOVIN Q29443 Transferrin Iron-binding protein
APOA1
BOVIN P15497 Apolipoprotein A1 Transport protein; major plasma HDL protein
APOA2
BOVIN P81644 Apolipoprotein A2 Involved in lipid transport; stablizes HDL; has antimicrobial activity
VTDB
BOVIN Q3MHN5 Vitamin D binding protein Transport protein; also associates w/immunoglobulins on lymphocytes
CO3
BOVIN Q2UVX4 Complement C3 Major protein in complement activation
CFAB
BOVIN P81187 Complement factor B Part of the alternative complement pathway
FETUA
BOVIN P12763 Alpha-2-HS-glycoprotein Promotes endocytosis; has lymphocyte stimulating properties
KNG2
BOVIN P01045 Kininogen 2 Produces active peptide bradykinin; inflammatory mediator
B2MG
BOVIN P01888 Beta-2-microglobulin Beta chain of major histocompatibility complex class I molecules
CLUS
BOVIN P17697 Clusterin Function unclear: may be involved in programmed cell death
SAA
BOVIN P35541 Serum amyloid A Major acute phase protein
ITIH4
BOVIN Q3T052 Inter-alpha trypsin
inhibitor heavy chain 4
Involved in acute phase reactions
HPT
BOVIN Q2TBU0 Haptoglobin Combines w/free plasma hemoglobin; acute phase protein
FIBA
BOVIN P02672 Fibrinogen alpha chain Yields monomers that polymerize into fibrin; acts in platelet aggregation
FIBB
BOVIN P02676 Fibrinogen beta chain Yields monomers that polymerize into fibrin; acts in platelet aggregation
FIBG
BOVIN P12799 Fibrinogen gamma chain Yields monomers that polymerize into fibrin; acts in platelet aggregation
a
Proteins all identified in 4 or more biological replicates with 2 or more peptide assignments.
per unique peptide identified were detected for the more
abundant proteins, while the number of unique peptides
identified and the corresponding spectral counts tended
to be more equivalent for the lower abundance proteins
(Supplemental Table 2). Because spectral counts have been
shown to have higher technical reproducibility than pep-
tide counts (Zhang et al., 2006; Old et al., 2005), the
temporal expression patterns of proteins identified using
nano-LC followed by nano-spray tandem MS were evalu-
ated using spectral counts.
3.3. Relative quantification of proteins related to host
response identified in bovine milk
The use of label-free quantification strategies such as
number of unique peptide assignments, ion intensity, and
spectral counts for the quantification of relative protein
abundance has been reported previously (Mosley et al.,
2009; McFarland et al., 2008; Florens et al., 2006; Zybailov
et al., 2005; Liu et al., 2004). Most previous endeav-
ors have, however, utilized normalized spectral counts to
estimate changes in relative abundance of individual pro-
teins present in complex matrices (Mosley et al., 2009;
McFarland et al., 2008; Florens et al., 2006; Zybailov et al.,
2005). In our previous analysis in which we utilized num-
ber of unique peptide assignments to track changes in the
expression of proteins detected in whey from milk over the
course of a clinical mastitis infection (Boehmer et al., 2010),
normalization had little effect on the data due to the mini-
mal sensitivity of the analyses. For the data generated using
a linear ion trap, however, raw spectral counts were used to
track the temporal expression of proteins present in whey
from bovine milk. Normalization was based on digesting
the same amount of protein for each sample, and injecting
the same volume of digested protein onto the chromatog-
raphy column for analysis using mass spectrometry. The
rationale for normalizing by the amount of protein ana-
lyzed versus resulting spectral counts was based on the
observation that milk samples collected prior to infection
were less complex than later time points, primarily due
to the fact that proteins related to host response were
absent in the milk from healthy cows. The difference in
complexity between 0 h samples and time points follow-
ing infection was evidenced by fewer proteins identified,
262 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
Time following infection (hours) Time following infection (hours)
484236302418126
0
5
10
15
20
Complement C3
Apolipoprotein A1
Transferrin
012
0 6 12 18 24 30 36 42 48
Spectral counts
Spectral counts
0
2
4
6
8
10
12
14
Fibrinogen alpha spectral counts
Fibrinogen beta spectral counts
Fibrinogen gamma spectral counts
BA
Fig. 3. Temporal expression patterns of vascular derived proteins (mean spectral counts ± standard error) complement C3, apolipoprotein A1, and trans-
ferrin (A) and the three different chains of the blood coagulation protein fibrinogen (B). Temporal expression of all the vascular-derived proteins was in
accord with previous reports of cytokine expression, in particular IL-1, and TNF, which are known to induce vascular leak.
and a much lower number of total spectral counts overall
in milk collected from cows prior to inoculation with E. coli.
Thus, normalization based on the sum of spectral counts at
each time point would have resulted in an apparent wash-
out effect of increases in spectral counts between baseline
samples and later time points during clinical mastitis.
As was evident in earlier LC–MS/MS analyses (Boehmer
et al., 2010; Smolenski et al., 2007), proteins identi-
fied in whey from bovine milk at time points following
infection using nano-LC–MS/MS were predominantly vas-
cular derived, acute phase, antimicrobial, complement, or
related to immune response, and fell into categories that
could be broadly classified as secondary effects of cytokine
induction. The most abundant vascular-derived proteins
identified included complement factor C3, transferrin, and
apolipoprotein A3 (Fig. 3A). Somewhat less abundant were
the fibrinogen proteins that are known to be involved
in blood coagulation (Fig. 3B). The vascular-derived pro-
teins identified were all clearly biologically relevant, as
firmly established hallmarks of coliform mastitis include
the induction of cytokines IL-1 and TNF, which cause
fever, complement activation, hepatic production of acute
phase proteins (Suojala et al., 2008; Grönlund et al., 2005;
Dinarello, 1996), and the leakage of plasma proteins and
complement factors into the milk (Bannerman et al., 2004,
2008; Riollet et al., 2000; Shuster et al., 1997). Biologi-
cal relevance was further evidenced by the fact that the
majority of proteins present only in whey from milk col-
lected following infection exhibited the greatest increases
in total spectral counts between 18 h and 24 h after inoc-
ulation (Supplemental Table 2), which corresponds to
peak cytokine expression previously reported to occur
between 16 and 24 h during coliform mastitis (reviewed
in Bannerman, 2009).
Temporal expression determined using spectral counts
also corresponded well with increases in milk somatic cell
4842363024181260
Spectral Counts
0
2
4
6
8
10
12
14
16
18
Glycam-1
PIGR
Osteopontin
Time following infection (hours) Time following infection (hours)
4842363024181260
Spectral Counts
0
2
4
6
8
10
12
Milk somatic cell score
0
2
4
6
8
10
12
14
PGRP
Somatic cell scores
Cathelicidin-1
A
B
Fig. 4. Temporal expression patterns of antimicrobial proteins (mean spectral counts ± standard error) detected only in time points following inoculation
with E. coli (A) and glycam-1, PIGR, and osteopontin, proteins present in milk before (Time 0) and after infection (B). Antimicrobial proteins exhibited similar
expression patterns to milk somatic cell counts (MSCC), which are comprised primary of neutrophils during infections, and are an established source of
the proteins with antimicrobial properties. The correspondence of increases in MSCC to increases in the abundance of antimicrobial proteins indicates that
LC–MS/MS label-free data can accurately model clinical indications of coliform mastitis.
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 263
Time following infection (hours)
Spectral Counts
Milk Haptoglobin (mg/mL)
4842363024181260
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.2
0.4
0.6
0.8
Haptoglobin spectral counts
Haptoglobin ELISA
A
Time following infection
4842363024181260
Spectral Counts
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Milk Serum amyloid A (mg/mL)
0.00
0.02
0.04
0.06
0.08
SAA Spectral counts
SAA ELISA
B
Fig. 5. Comparison of temporal expression patterns of low abundance acute phase proteins determined using ELISA and total spectral counts (mean spectral
counts ± standard error) for (A) milk haptoglobin and (B) milk serum amyloid A. Though sensitivity levels differ, the correspondence of the overall patterns
exhibited by the LC–MS/MS data and the ELISA data indicates that spectral counts can be used as a screening tool to profile changes in biologically relevant
proteins without a reliance on antibodies.
counts (MSCC; Fig. 4A), which are comprised primarily of
PMN during intramammary infection (Paape et al., 1981),
and are also a well established source of antimicrobial pro-
teins including peptidoglycan recognition receptor protein
(PGRP), and members of the cathelicidin family of cationic
antimicrobial peptides (Tydell et al., 2006; Scocchi et al.,
1997). Though the identifications of the antimicrobial pep-
tides (AMP) were not as robust as more abundant proteins
(Supplemental Table 2), the current proteomic analysis
marks the first evaluation of the temporal expression of
AMP during coliform mastitis, and the first indication that
spectral counts can accurately model clinical signs of the
disease.
In contrast, some proteins identified, including glycam-
1, PIGR, and osteopontin, exhibited decreases in total
spectral counts at 18 h following infection (Fig. 4B). The
increased dynamic range of proteins present in the milk
samples at 18 h following infection due to the influx of
vascular-derived proteins and other proteins related to
host response is one possible explanation for the apparent
decrease in the detection of constitutively expressed milk
proteins. However, glycam-1, PIGR, and osteopontin are all
known to be subject to glycosylation and phosphorylation,
and thus heavy modifications at 18 h following infection is
a more plausible explanation for the apparent decrease in
all 3 proteins as was evidenced by a decline in their respec-
tive total spectral counts (Kjeldsen et al., 2003; Soerensen
et al., 1995).
3.4. LC–MS/MS label free quantification correlates with
ELISA data
To the authors’ knowledge, no other proteomic evalua-
tion of bovine milk has assessed the temporal expression
of proteins identified in milk over time, nor compared
LC–MS/MS data with ELISA data to assess potential accu-
racy. The accuracy of LC–MS/MS label-free strategies in
tracking changes in the relative abundance of proteins in
milk during clinical mastitis was, however, evaluated in our
earlier proteomic analysis of milk by comparison of ELISA
data to LC–MS/MS peptide count data for abundant milk
proteins (Boehmer et al., 2010). After changing instrument
systems, however, a greater number of low abundance pro-
teins related to host response were detected in whey from
mastitic milk (Table 3), allowing us to evaluate the accuracy
of LC–MS/MS label-free quantification of minor proteins
when compared with antibody-based strategies. Compar-
isons of ELISA data and total spectral counts for milk Hp
(Fig. 5A) and SAA (Fig. 5B) revealed trends in temporal
expression with very similar overall patterns. Biological
variability was apparent, but the peptides identified for
each protein were uniform among samples, and trends
in total spectral counts tracked both across the biologi-
cal replicates and with ELISA data (Supplemental Table 2).
The comparison of ELISA data from low abundance acute
phase proteins with total spectral counts indicated slight
advantages in sensitivity with ELISA compared to spec-
tral counts, in that the antibody-based analysis was able to
detect the presence of acute phase proteins in milk at ear-
lier time points than mass spectrometry. However, the fact
that peptides from relatively low abundance acute phase
proteins were detected in an extremely biologically com-
plex matrix using LC–MS/MS despite the lack of sample
fractionation and only a one-dimensional LC separation,
further establishes the feasibility of using spectral counts
to track changes in proteins related to host response, espe-
cially those for which no antibody or ELISA currently exists.
3.5. Temporal expression of proteins not previously
identified in bovine milk
Previous proteomic analyses of whey from mastitic
bovine milk (Danielsen et al., 2010; Boehmer et al., 2008,
2010; Smolenski et al., 2007; Hogarth et al., 2004) have
identified a number of low abundance proteins related to
the host response in bovine milk (Table 2). The comparative
analysis of whey from bovine milk following experimen-
tal induction of coliform mastitis accomplished using one
dimensional nano-LC and a linear ion trap mass spectrome-
ter, however, identified a small number of potentially novel
biomarkers of coliform mastitis not previously reported
including the acute phase protein ITIH4, the eicasonoid
264 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
Time following infection (hours)
4842363024181260
Spectral counts
0
1
2
3
4
5
6
7
Clusterin spectral counts
Kininogen-2 spectral counts
ITIH4 spectral counts
Fig. 6. The temporal expression of the previously uncharacterized pro-
teins clusterin, kininogen-2, and ITIH4 (mean spectral counts ± standard
error) detected in whey from mastitic milk. The roles of clusterin,
kininogen-2, and ITIH4 during coliform mastitis are currently not known,
but the results represent the advantage of using LC–MS/MS label-free
strategies to screen for biologically relevant candidate biomarkers, with-
out the use of antibodies or costly labeling reagents.
precursor kininogen-2, several apolipoproteins, and the
poorly characterized protein clusterin (Table 3). Similar
to the vascular-derived proteins, ITIH4, kininogen-2, and
clusterin all appeared to be biologically relevant, as their
patterns of modulation (Fig. 6) corresponded with previ-
ous reports of peak cytokine expression during coliform
mastitis (reviewed in Bannerman, 2009). Prior reports of
ITIH4 in cattle have been limited to isolation of the APP
from the serum of heifers with experimentally induced
summer mastitis (Pineiro et al., 2004). Furthermore, ITIH4
has never been identified in bovine milk, and has never
been associated with coliform mastitis in lactating dairy
cattle. The association of ITIH4 with innate immunity has,
however, been studied in models of acute inflammation
in swine (Gonzalez-Ramon et al., 2000), and ITIH4 was
recently reported to be a novel marker of acute ischemic
stroke in humans (Kashyap et al., 2009). Previous research
and similarity to a human homolog led to the classification
of ITIH4 as a plasma kallikrein-sensitive glycoprotein, but
the exact role and function of ITIH4 in the bovine mam-
mary gland during inflammation associated with coliform
mastitis is not yet clear (Nishimura, 2003).
Kininogen-2, on the other hand, belongs to the fam-
ily of plasma kallikreins, or serine proteases, known to
play key roles in blood coagulation (Furie and Furie, 1988),
fibrinolysis (Ichinose et al., 1984), activation of comple-
ment (Discipio, 1982), and the release of bradykinin from
its precursor, kininogen (Scharfstein et al., 2007). The
kinin peptides, including bradykinin, are potent media-
tors of vasodilation, pain, and udder edema during clinical
mastitis (Eshraghi et al., 1999). The relationship between
elevated levels of milk bradykinin and increased severity of
clinical symptoms of mastitis has been demonstrated, but
bradykinin levels in milk from cows with experimentally
induced coliform mastitis have not yet been investigated
(Eshraghi et al., 1999). Likewise, while the activation of
complement and the cleavage of complement proteins
into active peptides that further enhance the inflammatory
response have been documented during coliform masti-
tis (Suojala et al., 2008; Grönlund et al., 2005; Shuster
et al., 1997), less is known about the contributions of
the kallikrein–kinin system to inflammation in the bovine
mammary gland.
Apolipoproteins are typically associated with high
density lipoproteins (HDLs), but other roles for the
apolipoproteins during disease, and inflammation in par-
ticular, have been investigated (Burger and Dayer, 2002;
Cockerill et al., 2001; Bausserman et al., 1988). Though
some apolipoproteins have been identified in previous pro-
teomic analyses of bovine milk (Danielsen et al., 2010;
Boehmer et al., 2008, 2010; Smolenski et al., 2007), the
role of the apolipoproteins during inflammation, and col-
iform mastitis specifically, has not yet been determined.
Similarly, the function of clusterin during coliform mastitis
is entirely unclear; however, the involvement of clusterin
in several inflammatory diseases including myocarditis,
and glomerulopathy has been implicated from previous
research (Rosenberg et al., 2002; McLaughlin et al., 2000).
Results of prior research likewise indicate that clusterin
could possess anti-inflammatory properties (Rosenberg
et al., 2002; McLaughlin et al., 2000). In terms of the role of
clusterin in the bovine mammary gland, all that has been
inferred is that clusterin could be associated with mam-
mary gland involution, the clearance of cellular debris, and
apoptotic cell death (Jones and Jomary, 2002).
3.6. Temporal expression of proteins yields quantitative
and qualitative data
Variable response to experimental infection will always
be a factor when using a bovine in vivo challenge model
for biomarker discovery analyses. In some cases, however,
while biological variation may prevent accurate quantifi-
cation, the apparent deviation of proteins identified in
milk from one animal, compared to the rest of the bio-
logical replicates, could provide useful qualitative data. In
the comparative proteomic analysis of whey from bovine
milk that utilized one-dimensional nano-LC followed by
nano-spray tandem MS, although biological variability was
apparent across all cows, the biological replicates with
patterns of total spectral counts that most often did not
track with the rest of the subjects were cows 3 and
6(Supplemental Table 2). The most interesting aspect
regarding proteins identified in milk from cows 3 and 6
is the nearly complete absence of proteins related to host
response, other than vascular-derived proteins, in samples
collected from both cows following infection. Markedly
higher total spectral counts for serum albumin and lactofer-
rin in the baseline milk sample collected from cow 3, when
compared to other biological replicates, could indicate that
cow 3 had recently resolved a naturally occurring clini-
cal infection, and that despite otherwise normal clinical
parameters, milk levels of serum albumin and lactofer-
rin had not yet returned to baseline at the time of the
challenge. Cow 6, on the other hand, had baseline spec-
tral count values for serum albumin and transferrin that
were in accord with the other biological replicates, but
total spectral counts for vascular-derived proteins such as
complement C3, apolipoprotein A1, transferrin, and the fib-
J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266 265
rinogens that did not increase above baseline values until
48 h following infection. The indication, based on proteins
detected and total spectral count patterns, is that cow 6
was perhaps much slower to respond to challenge than
the other cows included on the study. While analytical
errors could explain the somewhat aberrant results for
cows 3 and 6, the results could also indicate that with fur-
ther refinements, including the use of both biological and
technical replicates and 2D-LC separations, biologically rel-
evant information of a more qualitative nature, including
variation in response to challenge or infection status, could
be obtained from proteomic milk protein expression pat-
terns.
4. Conclusion
Clearly, several challenges still remain regarding the
identification and accurate quantification of biomarkers
of host response in bovine milk following experimental
induction of coliform mastitis. While results of compara-
tive proteomic analyses have revealed promising candidate
biomarkers, biological complexity of bovine milk, both
before and following challenge, as well as the inherent vari-
ability apparent across biological replicates, has precluded
the establishment of a biomarker or pattern of biomark-
ers specific to coliform mastitis. Nonetheless, the results
of the proteomic analysis conducted to date have pro-
vided information that could prove useful in the design
and execution of future studies of inflammatory biomark-
ers in bovine milk, including potential candidates for more
focused follow-up analyses, and more in-depth knowl-
edge of the suitability of different methodologies and the
capability of different instrument systems. A significant
contribution of more recent analyses that included the
evaluation of a longitudinal set of milk samples collected
from 8 cows over the course of clinical infection, is the
more biologically robust identification of several relevant
proteins related to host response in the bovine mammary
gland detected following the use of nano-LC, nano-spray
ionization, and a linear ion trap mass spectrometer. The
proteins that stand out as logical candidates for follow-
up analyses include the various APP and AMP detected,
vascular-derived proteins such as the apolipoproteins, and
proteins previously unaffiliated with coliform mastitis
including clusterin, ITIH4 and kininogen-2, a bradykinin
precursor. Additionally, the evaluation of spectral counts
as a means of tracking temporal expression of milk pro-
teins during coliform mastitis demonstrated that despite
inherent flaws, biological variability, and the current lack
of supporting statistical analysis, spectral counts appear to
be an effective means of illustrating protein modulation
during clinical infections. Given the fact that a majority
of previous reports dealing with the expression of inflam-
matory mediators during coliform mastitis have used data
derived from ELISAs, and that there is currently only a
limited number of commercially available bovine-specific
antibodies, proteomic strategies designed to discover or
characterize novel or poorly understood proteins related
to host response afford clear advantages over traditional
antibody-based methods.
5. Conflict of interest statement
The authors of this manuscript have no financial, con-
tractual, or personal relationships with any other persons,
organizations, or entities that could influence or bias the
nature of the research presented.
Appendix A. Supplementary data
Supplementary data associated with this arti-
cle can be found, in the online version, at
doi:10.1016/j.vetimm.2010.10.004.
References
Angal, S., Dean, P.D.G., 1977. The effect of matrix on the binding of albumin
to immobilized cibacron blue. Biochem. J. 167, 301–303.
Bannerman, D.D., 2009. Pathogen-dependent induction of cytokines and
other soluble inflammatory mediators during intramammary infec-
tion of dairy cows. J. Anim. Sci. 87 (13 Suppl.), 10–25.
Bannerman, D.D., Kauf, A.C., Paape, M.J., Springer, H.R., Goff, J.P.,
2008. Comparison of Holstein and Jersey innate immune responses
to Escherichia coli intramammary infection. J. Dairy Sci. 91 (6),
2225–2235.
Bannerman, D.D., Paape, M.J., Lee, J.-W., Zhao, X., Hope, J.C., Rainard,
P., 2004. Escherichia coli and Staphylococcus aureus elicit differential
innate immune responses following intramammary infection. Clin.
Diagn. Lab Immunol. 11 (3), 463–472.
Bausserman, L.L., Bernier, D.N., McAdam, K.P.W.J., Herbert, P.N., 1988.
Serum amyloid A and high density lipoproteins during the acute phase
response. Eur. J. Clin. Invest. 18, 619–626.
Boehmer, J.L., Ward, J.L., Peters, R.R., Shefcheck, K.J., McFarland, M.A., Ban-
nerman, D.D., 2010. Proteomic analysis of the temporal expression of
bovine milk proteins during coliform mastitis and label-free relative
quantification. J. Dairy Sci. 93 (2), 593–603.
Boehmer, J.L., Bannerman, D.D., Shefcheck, K.J., Ward, J.L., 2008. Proteomic
analysis of differentially expressed proteins in bovine milk during
experimentally induced Escherichia coli mastitis. J. Dairy Sci. 91 (11),
4206–4218.
Burger, D., Dayer, J.-M., 2002. High-density lipoprotein-associated
apolipoprotein A-I: the missing link between infection and chronic
inflammation? Autoimmun. Rev. 1, 111–117.
Cockerill, G.W., Huehns, T.Y., Weerasinghe, A., Stocker, C., Lerch, P.G.,
Miller, N.E., Haskard, D.O., 2001. Elevation of plasma high-density
lipoprotein concentration reduces interleukin-1-induced expression
of E-selectin in an in vivo model of acute inflammation. Circulation
103, 108–112.
Colantonio, D.A., Chan, D.W., 2005. The clinical application of proteomics.
Clin. Chim. Acta 357, 151–158.
Danielsen, M., Codrea, M.C., Ingvartsen, K.L., Friggens, N.C., Bendixen, E.,
Røntved, C.M., 2010. Quantitative milk proteomics Host responses
to lipopolysaccharide-mediated inflammation of bovine mammary
gland. Proteomics (Epub ahead of print).
Dinarello, C.A., 1996. Biological basis for interleukin-1 in disease. Blood 87
(6), 2095–2147.
Discipio, R.G., 1982. The activation of the alternative pathway C3 conver-
tase by human plasma kallikrein. Immunology 45 (3), 587–595.
Eckersall, P.D., Young, F.J., McComb, C., Hogarth, C.J., Safi, S., Weber, A.,
McDonald, T., Nolan, A.M., Fitzpatrick, J.L., 2001. Acute phase proteins
in serum and milk from dairy cows with clinical mastitis. Vet. Rec.
148, 35–41.
Eshraghi, H.R., Zeitlin, I.J., Fitzpatrick, J.L., Ternent, H., Logue, D., 1999. The
release of bradykinin in bovine mastitis. Life Sci. 64 (18), 1675–1687.
Fenselau, C., 2007. A review of quantitative methods for proteomic studies.
J. Chromatogr. B. 855, 14–20.
Florens, L., Carozza, M.J., Swanson, S.K., Fournier, M., Coleman, M.K., Work-
man, J.L., Washburn, M.P., 2006. Analyzing chromatin remodeling
complexes using shotgun proteomics and normalized spectral abun-
dance factors. Methods 40 (4), 303–311.
Furie, B., Furie, B.C., 1988. The molecular basis of blood coagulation. Cell
53 (4), 505–518.
Gagnaire, V., Jardin, J., Jan, G., Lortal, S., 2009. Invited review: proteomics of
milk and bacteria used in fermented dairy products: from qualitative
to quantitative advances. J. Dairy Sci. 92 (3), 811–825.
266 J.L. Boehmer et al. / Veterinary Immunology and Immunopathology 138 (2010) 252–266
Gonzalez-Ramon, N., Hoebe, K., Alava, M.A., van Leengoed, L., Pineiro,
M., Carmona, S., Iturralde, M., Lampreave, F., Pineiro, A., 2000. Pig
MAP/ITIH4 and haptoglobin are interleukin-6-dependent acute-phase
plasma proteins in porcine primary cultured hepatocytes. Eur. J.
Biochem. 267 (6), 1878–1885.
Grönlund, U., Hallén Sandgren, C., Persson Waller, K., 2005. Haptoglobin
and serum amyloid A in milk from dairy cows with chronic sub-clinical
mastitis. Vet. Res. 36 (2), 191–198.
Gu, S., Pan, S., Bradbury, E.M., Chen, X., 2002. Use of deuterium-labeled
lysine for efficient protein identification and peptide de novo sequenc-
ing. Anal. Chem. 74 (22), 5774–5785.
Gundry, R.L., Fu, Q., Jelinek, C.A., Van Eyk, J.E., Cotter, R.J., 2007. Inves-
tigation of an albumin-enriched fraction of human serum and its
albuminome. Proteomics Clin. Appl. 1, 73–88.
Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H., Aebersold, R., 1999.
Quantitative analysis of complex protein mixtures using isotope-
coded affinity tags. Nat. Biotechnol. 17, 994–999.
Hiss, S., Mielenz, M., Bruckmaier, M., Sauerwein, H., 2004. Haptoglobin
concentrations in blood and milk after endotoxin challenge and
quantification of mammary Hp mRNA expression. J. Dairy Sci. 87,
3778–3784.
Hogarth, C.J., Fitzpatrick, J.L., Nolan, A.M., Young, F.J., Pitt, A., Eckersall, P.D.,
2004. Differential protein composition of bovine whey: a comparison
of whey from healthy animals and from those with clinical mastitis.
Proteomics 4 (7), 2094–2100.
Ichinose, A., Kisiel, W., Fujikawa, K., 1984. Proteolytic activation of tissue
plasminogen activator by plasma and tissue enzymes. FEBS Lett. 175
(2), 412–418.
Jiang, H., English, A.M., 2002. Quantitative analysis of the yeast proteome
by incorporation of isotopically labeled leucine. J. Proteome Res. 1 (4),
345–350.
Jones, S.E., Jomary, C., 2002. Clusterin. Int. J. Biochem. Cell Bio. 34, 427–431.
Juraschek, R., Dülks, T., karas, M., 1998. Nanoelectrospray more than
just a minimized-flow electrospray ionization source. J. Am. Soc. Mass
Spectrom. 10, 300–308.
Kashyap, R.S., Nayak, A.R., Deshpande, P.S., Kabra, D., Purohit, H.J., Taori,
G.M., Daginawala, H.F., 2009. Inter--trypsin inhibitor heavy chain
4 is a novel marker of acute ischemic stroke. Clin. Chim. Acta 402,
160–163.
Kjeldsen, F., Haselmann, K.F., Budnik, B.A., Sorensen, E.S., Zubarev, R.A.,
2003. Complete characterization of posttranslational modification
sites in the bovine milk protein PP3 by tandem mass spectrometry
with electron capture dissociation as the last stage. Anal. Chem. 75,
2355–2361.
Lippolis, J.D., Bayles, D.D., Reinhardt, T.A., 2009. Proteomic changes in
Escherichia coli when grown in fresk milk versus laboratory media.
J. Proteome Res. 8 (1), 149–158.
Liu, H., Sadygov, R.G., Yates III, J.R., 2004. A model for random sampling
and estimation of relative protein abundance in shotgun proteomics.
Anal. Chem. 76 (14), 4193–4201.
Mann, M., Hendrickson, R.C., Pandey, A., 2001. Analysis of proteins and
proteomes by mass spectrometry. Annu. Rev. Biochem. 70, 437–473.
McFarland, M.A., Ellis, C.E., Markey, S.P., Nussbaum, R.L., 2008. Proteomics
analysis identifies phosphorylation-dependent -synuclein protein
interactions. Mol. Cell. Prot. 7 (11), 2123–2137.
McLaughlin, L., Zhu, G., Mistry, M., Ley-Ebert, C., Stuart, W.D., Florio,
C.J., Groen, P.A., Witt, S.A., Kimball, T.R., Witte, D.P., Harmony, J.A.K.,
Aronow, B.J., 2000. Apolipoprotein J/clusterin limits the severity of
murine autoimmune myocarditis. J. Clin. Invest. 106, 1105–1113.
Mosley, A.L., Florens, L., Wen, Z., Washburn, M.P., 2009. A label free quan-
titative proteomic analysis of the Saccharomyces cerevisiae nucleus. J.
Proteomics 72 (1), 110–120.
Mueller, L.N., Brusniak, M.-Y., Mani, D.R., Aebersold, R., 2008. An assess-
ment of software solutions for the analysis of mass spectrometry
based quantitative proteomics data. J. Proteome Res. 7 (1), 51–61.
Nishimura, T., 2003. Expression of potential lymphocytes trafficking medi-
ator molecules in the mammary gland. Vet. Res. 34, 3–10.
Oda, Y., Huang, K., Cross, F.R., Cowburn, D., Chait, B.T., 1999. Accurate quan-
titation of protein expression and site-specific phosphorylation. Proc.
Natl. Acad. Sci. U.S.A. 96 (12), 6591–6596.
O’Donnell, R., Holland, J.W., Deeth, H.C., Alewood, P., 2004. Review: milk
proteomics. Int. Dairy J. 14, 1013–1023.
Old, W.M., Meyer-Arendt, K., Aveline-Wolf, L., Pierce, K.G., Mendoza, A.,
Sevinsky, J.R., Resing, K.A., Ahn, N.G., 2005. Comparison of label-free
methods for quantifying human proteins by shotgun proteomics. Mol.
Cell. Prot. 4 (10), 1487–1502.
Paape, M.J., Wergin, W.P., Guidry, A.J., Schultze, W.D., 1981. Phagocytic
defense of the ruminant mammary gland. Adv. Exp. Med. Biol. 137,
555–578.
Pineiro, M., Andres, M., Iturralde, M., Carmona, S., Hirvonen, J., Pyorala, S.,
Heegard, P.M.H., Tjornehoj, K., Lampreave, F., Pineiro, A., Alava, M.A.,
2004. ITIH4 (Inter-alpha-trypsin inhibitor heavy chain 4) is a new
acute-phase potein isolated from cattle during experimental infec-
tion. Infect. Immun. 72 (7), 3777–3782.
Reinhardt, T.A., Lippolis, J.D., 2008. Developmental changes in the milk fat
globule membrane proteome during the transition from colostrums
to milk. J. Dairy Sci. 91, 2307–2318.
Riollet, C., Rainard, P., Poutrel, B., 2000. Differential induction of
complement fragment C5a and inflammatory cytokines during intra-
mammary infections with Escherichia coli and Staphylococcus aureus.
Clin. Diagn. Lab. Immunol. (2), 161–167.
Roe, M.R., Griffin, T.J., 2006. Gel-free mass spectrometry-based high
throughput proteomics: tools for studying biological response of pro-
teins and proteomes. Proteomics 6, 4678–4687.
Rosenberg, M.E., Girton, R., Finkel, D., Chmielewski, D., Barrie III, A.,
Witte, D.P., Zhu, G., Bissler, J.J., Harmony 2, J.A.K., Aronow, B.J., 2002.
Apolipoprotein J/clusterin prevents a progressive glomerulopathy of
aging. Mol. Cell. Bio. 22, 1893–1902.
Ross, P.L., Huang, Y.N., Marchese, J.N., Williamson, B., Parker, K., Hat-
tan, S., Khainovski, N., Pillai, S., Dey, S., Daniels, S., Purkayastha, S.,
Juhasz, P., Martin, S., Bartlet-Jones, M., He, F., Jacobson, A., Pappin,
D.J., 2004. Multiplexed protein quantitation in Saccharomyces cere-
visiae using amine-reactive isobaric tagging reagents. Mol. Cell. Prot.
3, 1154–1169.
Scharfstein, J., Schmitz, V., Svensjo, E., Granato, A., Monteiro, A.C., 2007.
Kininogens coordinate adaptive immunity through the proteolytic
release of bradykinin, an endogenous danger signal driving dendritic
cell maturation. Scand. J. Immunol. 66 (2–3), 128–136.
Scocchi, M., Wang, S., Zanettia, M., 1997. Structural organization of the
bovine cathelicidin gene family and identification of a novel member.
FEBS Lett. 417 (3), 311–315.
Shuster, D.E., Kehrli, M.E., Rainard, P., Paape, M., 1997. Complement frag-
ment C5a and inflammatory cytokines in neutrophil recruitment
during intramammary infection with Escherichia coli. Infect. Immun.
65 (8), 3286–3292.
Simpson, K., Whetton, A.D., Dive, C., 2009. Quantitative mass
spectrometry-based techniques for clinical use: biomarker
identification and quantification. J. Chromatogr. B. 877 (13),
1240–1249.
Smolenski, G., Haines, S., Kwan, Y.-S., Bond, F., Farr, J., Davis, V.S.R., Stelwa-
gen, K., Wheeler, T.T., 2007. Characterization of host defense proteins
in milk using a proteomic approach. J. Proteome Res. 6 (1), 207–
215.
Soerensen, E.S., Hoejrup, P., Petersen, T.E., 1995. Posttranslational mod-
ifications of bovine osteopontin: identification of twenty-eight
phosphorylation and three O-glycosylation sites. Protein Sci. 4,
2040–2049.
Suojala, L., Orro, T., Järvinen, H., Saatsi, J., Pyörälä, S., 2008. Acute phase
response in two consecutive experimentally induced E. coli intra-
mammary infection in dairy cows. Acta Vet. Scand. 50 (June (13)),
18 (Epub.).
Tydell, C.C., Yuan, J., Tran, P., Selsted, M.E., 2006. Bovine peptidoglycan
recognition protein-S: antimicrobial activity, localization, secretion,
and binding properties. J. Immunol. 176 (2), 1154–1162.
Wilm, M., Mann, M., 1996. Analytical properties of the nanoelectrospray
ion source. Anal. Chem. 68 (1), 1–8.
Yamada, M., Murakami, K., Wallingford, J., Yuki, Y., 2002. Identification
of low-abundance proteins of bovine colostral and mature milk using
two-dimensional electrophoresis followed by microsequencing and
mass spectrometry. Electrophoresis 23 (7–8), 1153–1160.
Yao, X., Afonso, C., Fenselau, C., 2003. Dissection of proteolytic 18O
labeling: endoprotease-catalyzed 16O-to-18O exchange of truncated
peptide substrates. J. Proteome Res. 2, 147–152.
Zhang, B., VerBerkmoes, N.C., Langston, M.A., Uberbacher, E., Hettich,
R.L., Samatova, N.F., 2006. Detecting differential and correlated pro-
tein expression in label-free shotgun proteomics. J. Proteome Res. 5,
2909–2918.
Zybailov, B., Coleman, M.K., Florens, L., Washburn, M.P., 2005. Cor-
relation of relative abundance ratios derived from peptide ion
chromatograms and spectrum counting for quantitative proteomic
analysis using stable isotope labeling. Anal. Chem. 77 (19), 6218–
6224.
    • "Ideally, to maintain general mastitis screening capabilities, as with SCC, the marker should be a molecule , enzyme, or protein that is suitable for detection with enzymatic assays or immunoassay procedures and is released in milk as a result of inflammation within the mammary gland (Viguier et al., 2009). In keeping with this goal, dedicated biomarker discovery studies, carried out mainly in cows, have reported different proteins that are released in mastitic milk and might therefore have potential for IMI detection (Boehmer et al., 2010; Akerstedt et al., 2011; Ceciliani et al., 2012; Wheeler et al., 2012). Recent studies in sheep by our group have revealed that cathelicidins are among the most prominent and promising molecules for this purpose because they are released abundantly, specifically, and very early in milk following a microbial stimulus. "
    [Show abstract] [Hide abstract] ABSTRACT: Mastitis due to intramammary infections is one of the most detrimental diseases in dairy sheep farming, representing a major cause of reduced milk productions and quality losses. In particular, subclinical mastitis presents significant detection and control problems, and the availability of tools enabling its timely, sensitive, and specific detection is therefore crucial. We have previously demonstrated that cathelicidins, small proteins implicated in the innate immune defense of the host, are specifically released in milk of mastitic animals by both epithelial cells and neutrophils. Here, we describe the development of an ELISA for milk cathelicidin and assess its value against somatic cell counts (SCC) and bacteriological culture for detection of ewe mastitis. Evaluation of the cathelicidin ELISA was carried out on 705 half-udder milk samples from 3 sheep flocks enrolled in a project for improvement of mammary health. Cathelicidin was detected in 35.3% of milk samples (249/705), and its amount increased with rising SCC values. The cathelicidin-negative (n = 456) and cathelicidin-positive (n = 249) sample groups showed a clear separation in relation to SCC, with median values of 149,500 and 3,300,000 cells/mL, respectively. Upon bacteriological culture, 20.6% (145/705) of the milk samples showed microbial growth, with coagulase-negative staphylococci being by far the most frequent finding. A significant proportion of all bacteriologically positive milk samples were positive for cathelicidin (110/145, 75.9%). Given the lack of a reliable gold standard for defining the true disease status, sensitivity (Se) and specificity (Sp) of the cathelicidin ELISA were assessed by latent class analysis against 2 SCC thresholds and against bacteriological culture results. At an SCC threshold of 500,000 cells/mL, Se and Sp were 92.3 and 92.3% for cathelicidin ELISA, 89.0 and 94.9% for SCC, and 39.4 and 93.6% for bacteriological culture, respectively. At an SCC threshold of 1,000,000 cells/mL, Se and Sp were 93.3 and 91.9% for cathelicidin ELISA, 80.0 and 97.1% for SCC, and 39.4 and 93.5% for bacteriology, respectively. In view of the results obtained in this study, the measurement of cathelicidin in milk by ELISA can provide added Se while maintaining a high Sp and may therefore improve detection of subclinical mastitis.
    Full-text · Article · Jun 2016 · EuPA Open Proteomics
    M.F. AddisM.F. AddisV. TeddeV. TeddeS. DoreS. Dore+1more author...[...]
    • "Nowadays, hundreds of unique proteins can be identified in different fractions of bovine milk by mass spectrometry (Hettinga et al., 2011; Nissen et al., 2013). This makes a proteomics approach a valuable tool for discovery of novel biomarkers (Boehmer et al., 2010; Ferreira et al., 2013). Here, we use shotgun proteomics (NanoLC–MS/MS) to compare milk samples of cows with a good health history (high-resistant cows) to milk samples of cows with a poorer health history (low-resistant cows). "
    [Show abstract] [Hide abstract] ABSTRACT: The objective of this study was to identify and characterize potential biomarkers for disease resistance in bovine milk that can be used to indicate dairy cows at risk to develop future health problems. We selected high- and low-resistant cows i.e. cows that were less or more prone to develop diseases according to farmers’ experience and notifications in the disease registration data. The protein composition of milk serum samples of these high- and low-resistant cows were compared using NanoLC–MS/MS. In total 78 proteins were identified and quantified of which 13 were significantly more abundant in low-resistant cows than high-resistant cows. Quantification of one of these proteins, lactoferrin (LF), by ELISA in a new and much larger set of full fat milk samples confirmed higher LF levels in low- versus high-resistant cows. These high- and low-resistant cows were selected based on comprehensive disease registration and milk recording data, and absence of disease for at least 4 weeks. Relating the experienced diseases to LF levels in milk showed that lameness was associated with higher LF levels in milk. Analysis of the prognostic value of LF showed that low-resistant cows with higher LF levels in milk had a higher risk of being culled within one year after testing than high-resistant cows. In conclusion, LF in milk are higher in low-resistant cows, are associated with lameness and may be a prognostic marker for risk of premature culling.
    Full-text · Article · Apr 2016
    • "c W: Whey. response to mastitis [10,11,15,35]. This supports the notion that involution and the presence of mastitis pathogens trigger similar inflammatory responses in the bovine mammary gland. "
    [Show abstract] [Hide abstract] ABSTRACT: Changes of abundance that occur in the repertoire of low abundance milk proteins after cessation of milk removal have not been characterised. Skimmed milk and whey from cows sampled at day 0 and either day 3 or day 8 after drying off were subjected to three untargeted proteomics techniques; 2-D gel electrophoresis, GeLC-MS, and dimethyl isotopic labelling of tryptic peptides. The changes observed included 45 fragments of abundant milk proteins and 36 host-defence proteins, suggesting activation of proteolysis and inflammation. The findings form a basis for adding value to dairy production.
    Full-text · Article · Sep 2015
Show more