ABBS 2008,40(11): Expression, purification and mass spectrometric analysis of LIM mineralization protein-1 in human lung epithelial cells

Keywords        LIM
mineralization protein-1; mass spectrometry; A549 cells

LIM mineralization protein-1 (LMP-1) is involved in potentiation of
bone morphogenetic protein responsiveness and induction of bone formation [1–4]. LMP-1 belongs to a sub-family of LIM domain proteins to which
Enigma, ENH and ZASP/Cypher-1 also belong. The sequence of LMP-1 contains a
highly conserved N-terminal PDZ domain and three C-terminal LIM domains. The
LIM domain is a cysteine-rich structural motif composed of two special zinc
fingers that are joined by a two-amino acid spacer. These domains are highly
related (up to 70% sequence homology), whereas the middle portion between the
PDZ and LIM domains is more diverse. A central region of LMP-1 has been shown
to be required in various bone-forming variants of LMP-1. Although the precise
mechanism of action of LMP-1 remains elusive, a bone morphogenetic
protein-potentiating effect of LMP-1 through its interaction with an E3 ligase
(Smurf1) was proposed in our recent report [1]. Here, we describe a transient,
transfection-based mammalian expression system that provides the speed and
yield needed to meet the demands of proteomics.

The full-length LMP-1 gene encodes a protein of 457 amino acids. There
are two potential N-glycosylation sites, Asn-Lys-Thr and Asn-Arg-Thr, at amino
acid positions 113–116 and 257–259, respectively, in the polypeptide sequence deduced from the
LMP-1 cDNA. Several potential O-glycosylation sites within the central region
of LMP-1 polypeptide sequence are predicted using the NetOGlyc 3.1 prediction
algorithm [5]. The addition of carbohydrate
chains to the polypeptide backbone of a protein may have an impact on the structure, solubility, antigenicity, folding, secretion and stability of the protein
[6]. In this report, we describe the overexpression and biochemical
characterization of LMPTumor cell lines, such as human lung carcinoma cells (A549), are
extensively relied on for cancer investigations; yet, cultured cells in an in vitro environment differ
considerably in behavior compared
with those same cancer cells that proliferate
and form tumors in vivo. In this report, we used in vivo cells
for the transfection and purification of LMP-1 protein to determine whether the
overexpressed protein undergoes any post-translational glycosylation in a
mammalian expression system. In the event of finding such modification, we
intended to characterize the purified protein and its glyco-peptides. Further,
we proposed to determine the role of the carbohydrate moiety in the LMP-1 mode
of action. The molecular weights of p53 protein and the overexpressed
recombinant LMP-1 are similar, and the two proteins co-migrate with endogenous
LMPMost of the cell transfection studies routinely show a Western blot
signal as the sole proof of protein expression. However, in this report, we
chose to demonstrate conclusively both the identity of the LMP-1 protein by
internal sequencing and assess post-translational glycosylation by carbohydrate
analysis of LMP-1 hydrolysate. In order to determine more clearly the nature of
LMP-1 expressed in a mammalian system, we modified the standard
hexahistidine-tagged protein purification method to increase the yield of LMP

Materials and Methods

Chemicals, enzymes, plasmid
vector and the host strain

The restriction and modifying enzymes used in this work were purchased
from Promega (

Transfection of A549 cells

A549 cells (ATCC CCL-185) are type II alveolar epithelial cells from a human adenocarcinoma. A549 cells were
grown in Ham’s F12K medium (Gibco, Grand Island,

Preparation of nuclear and cytoplasmic protein fractions from A549
cells

The A549 cell pellets were resuspended for cell lysis in low-salt buffer
[

Purification of recombinant LMP-1 from cellular extracts

Size-exclusion chromatography of the cellular proteins was carried
out on a Sephacryl S-300 column connected to the AKTA FPLC System. The column
was pre-calibrated with known low- and high-molecular weight protein markers
from gel filtration calibration kits (Amersham Biosciences) in The A549 cell lysates were suspended in 10 ml ice-cold lysis buffer
(The proteins were applied onto a Ni++-affinity column (5 ml resin) previously equilibrated with 4 5 ml of
A HiTrap SP Sepharose (Fast Flow) cation exchange column (1 ml
HiTrap) was equilibrated in buffer A. Protein samples were syringe filtered
using a 0.2 mm membrane and applied onto the column in

Cloning of human LMP-1 cDNA
into TAT-HA vector and expression in Escherichia coli

The full-length cDNA for LMP-1 was cloned into TAT-HA vector. The
BL21 (DE3) competent cells (Novagen,

SDS-PAGE and Western blotting

SDS-PAGE was performed using 10% gels according to Laemmli [7], and
the resolved proteins were transferred from the gel onto a nitrocellulose
membrane at 50 volts (constant) for 2 h. The membranes were blocked with 25 ml
5% milk protein for 1 h at room temperature. Membranes were incubated with
anti-LMP-1 antibody at a dilution of 1:5000 [5 ml/25 ml Tris-buffered saline containing 0.1% Tween 20 (TBST)] gently
shaken for 2 h at room temperature. Membranes were washed with 25 ml TBST for 5
min. The washes were repeated twice. Membranes were incubated with goat
anti-rabbit IgG linked to horse radish peroxidase (NEF 812; NEN,

Sugar composition analysis

Sugar compositions were determined by dissolving the purified
protein (100 mg) in 20 ml distilled water in a test tube to which

In-gel digestion of LMP-1 by
trypsin

Electrophoresis of a protein sample eliminates impurities and shows
the purity of the isolated protein; immobilization of the protein in the gel
allows reduction, alkylation and washing of the protein. SDS-PAGE gels were
stained with 0.25% Coomassie brilliant blue in 45% methanol and 10% acetic acid
and destained in 35% methanol with 10% acetic acid. The protein bands
corresponding to a positive signal on the Western blots were sliced from the
gel, soaked in 50% methanol with

Preparation of peptide samples
for mass spectrometry analysis

Peptide samples were purified and concentrated using a ZipTip
(Millipore) which has C18 resin
fixed at its end. The resin was rinsed according to the manufacturer抯 instructions with 10 ml 0.1% TFA and 50% ACN. Peptides were eluted in 10 ml 50% ACN in 0.1 % TFA. A 0.5 ml volume of the concentrated peptide-containing sample was mixed
with a saturated solution of a-cyano-4-hydroxycinnamic
acid (0.5 ml). Each sample (0.5 ml) was spotted on the mass spectrometer sample plate [13]. The mass
spectrometer determines the mass of the peptides and the sequence (by
collision-induced dissociation). From the masses of the peptide fragments,
sequence data were determined by comparison with known sequences or by manual
interpretation.

Separation of peptides by HPLC

After trypsin digestion, the mixture (85%) of LMP-1 peptides was
separated by capillary reversed-phase HPLC [14]. The peptide fragments were
separated by small-bore reversed-phase HPLC on a Vydac C18 column (

Protein identification and
amino acid sequence analysis

In order to increase sequence coverage of LMP-1, aliquots of HPLC
fractions of the digest also were analyzed by MALDI-time of flight
(TOF)/TOF-MS/MS using a model 4700 Proteomics Analyzer (Applied Biosystems).
For each fraction, a MS spectrum was initially collected. For post-source decay
analysis, the HPLC-purified peptide was subjected to ion generation by
post-source decay [15]. A MALDI-post-source decay TOF spectrum was recorded
using a-cyano-4-hydroxycinnamic acid as a matrix;
acquisition was at 27.5 kV under continuous extraction conditions; reflector
voltage was stepped from 30 to 1.27 kV, and the spectrum was constructed using
the FASTTM method from Bruker-Daltonic (

Database searches for protein
identification

Monoisotopic peptide masses obtained from mass spectra were searched
against the SWISS-PROT, NCBInr and MSDB databases using the MASCOT search
program. The following parameters were used in the searches: mammalian, human,
MS/MS Ion Search, protein mass of 50 kDa, trypsin digest with two missed
cleavages and fragment ion mass tolerance of 75 ppm. The resulting protein hits
were scored using a probability based Mowse score. The score is –10´log(P), where P is the probability that the observed match is a random
event.

Results

The overall goal of the present study was to determine whether LMP-1
overexpressed in mammalian cells undergoes post-translational glycosylation. In
addition, we desired an optimum protocol for the purification of overexpressed LMP-1
protein that is folded in the chaperone-environment provided by mammalian
systems. These needs prompted us to purify at least 500 mg protein required for the determination of carbohydrate composition
and internal peptide sequencing for LMP-1 protein.

Expression of
hexahistidine-tagged LMP-1

Based on initial Western blot analysis of total cell lysates, the
transfection levels were considered satisfactory to pursue purification and
analysis of overexpressed LMP-1. The yield of total protein (cytoplasmic plus
nuclear fractions) was approximately 0.2 mg/106 cells. Since the LMP-1 was
detected both in nuclear and cytoplasmic fractions in Western blots, we
combined both fractions to purify overexpressed LMP-1.

Purification of recombinant
LMP-1 protein

The presence of a hexahistidine tag in the recombinant protein
allowed use of Ni++-affinity resins for
purification of LMP-1. However, when the cell lysate was directly applied to Ni++-affinity resins, the abundance of
non-specific proteins competing with the desired protein for binding did not
permit the effective use of the metal affinity resin. In addition, the high
concentration of proteins in these samples (over 10 mg/ml) promoted
non-specific aggregation and poor performance of the resin. To circumvent this
problem, we chose to first perform molecular exclusion chromatography to select
desired protein fractions based on molecular size. The addition of this step
eliminated about 60%–80% of the unwanted
proteins before employing Ni++-affinity
selection of the desired protein. LMP-1 contains a high proportion of proline
and cysteins (10.1% and 4.2% by weight, respectively) that are known to
contribute to a tendency for protein aggregation at neutral pH, especially in
concentrated samples. Initial attempts to purify LMP-1 protein by Ni++-affinity selection in native conditions did
not yield satisfactory results due to aggregation. Performing Ni++-affinity chromatography in the presence of The purified protein was seen as a predominantly single band at
molecular size 53 kDa in Coomassie-stained SDS-PAGE gels and Western blots [Fig.
2(A,B)]. Another band running at approximately 33 kDa was also found to be
immunoreactive to LMP-1 antibody. We presume that this band originates from
partial degradation of full length LMP-1. The tumor marker protein, p53, and
the recombinant LMP-1 co-elute from the Sephacryl S-300 molecular exclusion
column. However, the purification strategy that we employed with the
hexahistidine tag ensures that the co-migrating p53 is eliminated in the wash
buffers of the Ni++-affinity column.
The final recombinant LMP-1 preparation was predominantly homogeneous, although
minor impurities were still visible. These additional bands resolved away from
LMP

Carbohydrate analysis of
purified LMP-1 protein

Sequence analysis of LMP-1 showed several potential sites for N- and
O-glycosylation (Fig. 3). Treatment of an aliquot of
purified LMP-1 at 37 ºC overnight with endoglycosidase peptide Nglycosidase F (2
units) (Boehringer Ingelheim,

MALDI-MS analysis of LMP-1
protein band

Using 10% (W/V) total polyacrylamide gel concentration
in SDS-PAGE analysis, the LMP-1 band was well resolved from other proteins and
well suited for excision, digestion by trypsin and sequencing. The minimal gel
area containing the intensely stained band of approximately 53 kDa was excised
and destained for trypsin digestion using a trypsin in-gel digestion kit [Fig.
2(A)]. Trypsin cleaves after lysine and arginine residues (except if they
are N-terminal to proline), therefore producing specific fingerprints for
specific proteins. The purified LMP-1 did not yield any N-terminal sequence
using the Edman degradation technique most likely due to a blocked N-terminus.
We therefore digested the protein in-gel with trypsin. Approximately 15% of the
extracted peptides were microdesalted using C18 ZipTip column for MALDI-TOF
MS analysis. Resulting peptide digests were then analyzed by MALDI-MS.

The MALDI spectrum obtained for the 15% portion of the tryptic
digestion contained several molecular ions that closely matched the expected
tryptic peptide mass map of a theoretical digestion of LMP-1. The remaining 85%
of the extract from tryptic digestion was fractionated by capillary reverse
phase HPLC. The peptide mixture was resolved on HPLC. The column and the
running conditions were optimized to overcome the high aggregation propensity.
The peptides showed wide variation in their relative amounts due to the
inherent propensity of LMP-1 peptides for aggregation due to high content of
proline and cysteine in LMP-1 protein. The peptide profile obtained is shown in
Fig. 4.

A post-source decay MALDI-MS analysis was performed on some of the
matching molecular ions. Indeed, the product ion spectrum of many ions
confirmed the identity of specific HPLC purified-tryptic fragments of LMP-1. In
addition, a large number (up to 40 samples) of HPLC fractions (that were confirmed
to be pure by isocratic HPLC runs) were analyzed by MALDI-TOF MS. We subjected
them to Edman degradation and obtained amino acid sequences that matched the
LMP-1 sequence. Based on m/z estimation, there were many other peptide matches
corresponding to different regions of the LMP-1 polypeptide (Table 2).
The internal sequences obtained by Edman degradation corresponded to different
regions of LMP-1. If we assume an average mass of 1500 daltons per site of glycosylation [16], then the observed mass of the LMP-1 protein/peptide should be
higher by the same measure. However, the mass peaks of peptides that were
predicted to be N-glycosylated were
also present in the fingerprint analysis of the LMP-1 protein. This suggests that the potential glycosylation
sites are not glycosylated. Representative profiles for hydrosylates of LMP-1
expressed in A549 cells and LMP-1 expressed in E. coli are shown in Fig.
5. Only trace amounts of carbohydrate were found in LMP-1 hydrolysates
prepared from A549 and E. coli cells as quantitated by comparison to a
standard set of monosaccharides. This observation supported the finding that
carbohydrate analysis of protein contains no significant amounts of sugars.
However, some of the peptides did not correspond to any region in LMP-1 and
perhaps originated from a minor impurity.

Discussion

Recombinant expression of protein factors has become a powerful tool
for a variety of applications ranging from basic research to human therapy. Cultured
mammalian cells have become the dominant system for the production of
recombinant mammalian proteins for clinical applications because of their
capacity for proper protein folding, assembly and post-translational
modification. The main drawback of the expression of heterologous proteins in
non-mammalian hosts (e.g. bacteria, yeast, baculovirus) is that the resulting
recombinant proteins often display poor functional and structural properties
due to a lack of proper folding and/or post-translational modifications.
High-level mammalian recombinant protein production mostly relies on the
establishment of stably expressing cell lines. This procedure is not only labor
intensive and time consuming, but it also precludes the expression of proteins
with biological activities that interfere with cell growth. Although expression
levels from transiently transfected cells are presently low, the short time
span of a few days between DNA delivery and protein harvest makes this approach
very appealing.

A549 cells derived from a human lung adenocarcinoma are not fully representative of normal human respiratory epithelium but have been a quick
and useful in vitro model for protein expression studies [17–20]. The aim of this study was to confirm that LMP-1 protein is expressed
upon plasmid-mediated transformation of mammalian cells by purifying and
characterizing the identity of the protein. In this report, we purified the
recombinant proteins to homogeneity using size fractionation of proteins prior
to metal affinity chromatography to improve efficiency of affinity resin
followed by identification of tryptic fragments of purified protein.

The second aim was to find out whether the expressed protein
undergoes any post-translational modification. The A549 cell system has all the
eukaryotic protein processing capabilities. It is generally accepted that A549
cells can fold, modify, traffic and assemble newly synthesized polypeptides to
produce highly authentic, soluble end products. However, it is equally possible
that A549 protein processing pathways are not necessarily equivalent to those
of normal cells. If we had determined the presence of carbohydrate on the LMP-1
polypeptide in A549 cells, we intended to further characterize the structural
and functional role of the carbohydrate moiety in LMP-1 purified from normal
human cells. The present study provides
conclusive evidence that a full-length LMP-1 is indeed expressed in A549 cells,
and milligram quantities of protein can be obtained from mammalian cell
cultures even though the level of transfection efficiency could be low. We also
determined that the protein did not contain carbohydrate, as chemical analysis
showed little or no N-acetyl glucosamine or N-acetyl galactosamine. These
results were also consistent with our previous observation that the treatment
of purified LMP-1 with endoglycosidase peptide N-glycosidase F for enzymatic
deglycosylation did not reduce the molecular size of LMP-1. The observation
that the LMP-1 protein does not undergo post-translational glycosylation
implies that more efficient expression systems, such as bacteria or yeast, may
be pursued for further studies of the mode of action of LMP-1 through
protein-protein interactions.

Acknowledgements

This work was
performed at the Atlanta Veterans Affairs (VA)

References

 1   Sangadala S, Boden SD, Viggeswarapu M, Liu Y,
Titus L. LIM mineralization protein-1 potentiates bone morphogenetic protein responsiveness
via a novel interaction with Smurf1 resulting in decreased ubiquitination of
Smads. J Biol Chem 2006, 281: 17212–17219

 2   Boden SD, Liu Y, Hair GA, Helms JA, Hu D,
Racine M, Nanes MS et al. LMP-1, a LIM domain protein, mediates BMP-6
effects on bone formation. Endocrinology 1998, 139: 5125–5134

 3   Boden SD, Titus L, Hair G, Liu Y,
Viggeswarapu M, Nanes MS, Baranowski C. Lumbar spine fusion by local gene
therapy with a cDNA encoding a novel osteoinductive protein (LMP-1). Spine
1998, 23: 2486–2492

 4   Viggeswarapu M, Boden SD, Liu Y, Hair GA,
Louis-Ugbo J, Murakami H, Kim HS et al. Adenoviral delivery of LIM
mineralization protein-1 induces new-bone formation in vitro and in
vivo. J Bone Joint Surg Am 2001, 83: 364–376

 5   Julenius K, Mølgaard A, Gupta R, Brunak S.
Prediction, conservation analysis and structural characterization of
mammalian mucin-type O-glycosylation sites. Glycobiology 2005, 15: 153–164

 6   Narhi LO, Arakawa T, Aoki KH, Elmore R, Rohde
MF, Boone T, Strickland TW. The effect of carbohydrate on the
structure and stability of erythropoietin. J Biol Chem 1991, 266: 23022–23026

 7   Laemmli UK. Cleavage of
structural proteins during the assembly of the head of bacteriophage T4.
Nature 1970, 227: 680–685

 8   Yasuno S, Kokubo K, Kamei M. New method for determining the sugar composition of glycoproteins,
glycolipids and oligosaccharides by high performance liquid chromatography.
Biosci Biotechnol Biochem 1999, 63: 1353–1359

 9   Sangadala S, Azadi P, Carlson R, Adang MJ.
Carbohydrate analyses of Manduca sexta aminopeptidase N, co-purifying
neutral lipids and their functional interactions with Bacillus thuringiensis
Cry10  Bernardo K, Fleer S, Pakulat N, Krut O, Hünger F, Krönke M. Identification
of Stapylococcus aureus exotoxins by combined sodium dodecyl sulfate gel
electrophoresis and matrix-assisted laser desorption/ionization-time of flight
mass spectrometry. Proteomics 2002, 2: 740–746

11  Wilkins JC, Homer KA, Beighton D. Altered protein expression of Streptococcus
oralis cultured at low pH revealed by two-dimensional gel electrophoresis.
Appl Environ Microbiol 2001, 67: 3396–3405

12  Winters MS, Day RA. Detecting protein-protein
interactions in the intact cell of Bacillus subtilis (ATCC 6633).
J Bacteriol 2003, 185: 4268–4275

13  Trémoulet F, Duché O, Namane A, Martinie B, Labadie JC. A proteomic
study of Escherichia coli O157:H7 NCTC 12900 cultivated in biofilm or in
planktonic growth mode. FEMS Microbiol Lett 2002, 215: 7–14

14  Hubálek F, 15  Chaurand P, Luetzenkirchen F, Spengler B. Peptide and protein
identification by matrix-assisted laser desorption ionization (MALDI) and
MALDI-post-source decay time-of-flight mass spectrometry. J Am Soc Mass
Spectrom 1999, 10: 91–103

16  Jaikaran AS, Kennedy TD, Dratewka-Kos E, Lane BG. Covalently bonded
and adventitious glycans in germin. J Biol Chem 1990, 265: 12503–12512

17  Allen CB, White CW. Glucose modulates cell death due to normobaric
hyperoxia by maintaining cellular ATP. Am J Physiol 1998, 274: L159–L164

18  Kazzaz JA, Xu J, Palaia TA, Mantell L, Fein AM, Horowitz S.
Cellular oxygen injury toxicity. Oxidant injury without
apoptosis. J Biol Chem 1996, 271: 15182–15186

19  Lee PJ, Alam J, Wiegand GW, Choi AM. Overexpression of heme
oxygenase20  Wong HR, Ryan M, Wispé JR. Stress response decreases NF-kB translocation and increases I-kBa expression in A549 cells. J Clin Invest 1997, 97: 2423–2428