REVIEW ARTICLE
Sherlock Holmes and the proteome ) a detective story
Pier Giorgio Righetti
1
and Egisto Boschetti
2
1 Department of Chemistry Materials and Chemical Engineering ‘Giulio Natta’, Polytechnic of Milano, Milan, Italy
2 Ciphergen Biosystems, Fremont, CA, USA
Introduction
The word ‘detective’ originates from the Latin ‘detego’
(detexi, detectum, detegere), i.e. to find out, to discover
(in fact, to remove the teges or tegmen, in English
slang the cover, therefore to uncover!). Modern prote-
ome analysis is a very complex ‘detective story’, which
might baffle even the most famous investigator, Sher-
lock Holmes [1]. The reason is that, in any proteome,
a few proteins dominate the landscape and often oblit-
erate the signal of the rare ones, so that, when the
police reach the scene of the crime, the thin thread of
evidence remains hidden. In addition, proteomes of
any origin can be extremely complex, impervious to
even the most sophisticated analytical tools. For
instance, according to Anderson et al. [2,3], the human
plasma should contain most, if not all, human pro-
teins, as well as proteins derived from viruses, bacteria
and fungi. Also, numerous post-translationally modi-
fied forms of each protein are present, along with,
possibly, millions of distinct clonal immunoglobulin
sequences. To this intrinsic complexity, one can add
the enormous dynamic range, encompassing some 10
orders of magnitude between the least abundant (e.g.

Tel: +39 022399 3016
E-mail: [email protected]
Note
This lecture was delivered at the 7th Siena
Meeting ‘From Genome to Proteome: Back
to the Future’, September 3–7, 2006, Siena,
Italy.
(Received 5 October, revised 26 November
2006, accepted 13 December 2006)
doi:10.1111/j.1742-4658.2007.05648.x
The performance of a hexapeptide ligand library in capturing the ‘hidden
proteome’ is illustrated and evaluated. This library, insolubilized on an
organic polymer and available under the trade name ‘Equalizer Bead Tech-
nology’, acts by capturing all components of a given proteome, by concen-
trating rare and very rare proteins, and simultaneously diluting the
abundant ones. This results in a proteome of ‘normalized’ relative abun-
dances, amenable to analysis by MS and any other analytical tool. Exam-
ples are given of analysis of human urine and serum, as well as cell and
tissue lysates, such as Escherichia coli and Saccharomyces cerevisiae
extracts. Another important application is impurity tracking and polishing
of recombinant DNA products, especially biopharmaceuticals meant for
human consumption.
FEBS Journal 274 (2007) 897–905 ª 2007 The Authors Journal compilation ª 2007 FEBS 897
strongly alkaline proteins are poorly represented
on classical two-dimensional electrophoresis [6], and
highly hydrophobic proteins cannot be properly solubi-
lized and consequently not analyzed and ⁄ or identified
at all. Electrophoresis-based methods on their own
(still the most commonly used to date) are neither
appropriate for polypeptides of mass lower than

out this task. The approach that is gaining momentum,
especially in analysis of biological fluids, such as
plasma, sera, cerebrospinal fluid, urine, is sequential or
simultaneous immunoaffinity depletion of the most
abundant proteins present in the samples [13]. How-
ever, even this approach may not be good enough to
gain access to the ‘deep proteome’. Although depletion
of the nine most abundant proteins represents the
removal of as much as 90% of the overall protein
content, the vast number of serum proteins that
comprise the remaining 10% remain dilute, and the
improvement in detecting rare proteins might be quite
disappointing. In fact, Echan et al. [14], using a
commercial column for removal of the top six most
abundant proteins, reported: ‘many of the moderate
and low-intensity protein spots that were detected on
the depleted sample gels were actually detectable on
the unfractionated sample gel’. Another major draw-
back of such immuno-subtraction methods appears to
be co-depletion. As reported by Shen et al. [15], during
depletion of human serum albumin, another 815
species (not including this protein) were co-depleted.
When capturing IgGs, another 2091 species (not inclu-
ding IgG) were co-depleted, among which 56% were
antibody sequences and the other 44% included
low-abundance cytokines and related proteins. Para-
doxically, in the sera thus subtracted from just these
two major proteins, only 1391 free proteins could be
detected. Ironically, most of the newly discovered spe-
cies were found in the two fractions that had to be dis-

until a sequence of the desired length is produced (this
process is detailed in Lam et al. [16]).
The ligands are represented throughout the beads’
porous structure and can achieve an amount of
 15 pmol per bead of the same hexapeptide distri-
buted throughout the core of the pearl. This amounts
to a ligand density of  40–60 lmolÆ mL
)1
bead volume
Sherlock Holmes and the proteome P. G. Righetti and E. Boschetti
898 FEBS Journal 274 (2007) 897–905 ª 2007 The Authors Journal compilation ª 2007 FEBS
(average bead diameter  60 lm). As a result of the
nonrandomized combinatorial hexapeptide construc-
tion, each bead has many copies of a single, unique
ligand, and each bead has a different ligand from every
other bead. Considering that, for the synthesis of a
protein, many amino acids are used, the resulting
library contains a population of linear hexapeptides
amounting to millions of different ligands. Such a vast
population of baits means that, in principle, every
protein present in a complex proteome (be it a bio-
logical fluid or a tissue or cell lysate of any origin)
potentially has a bead partner carrying the peptide
ligand with which it is able to interact under the
well-known affinity chromatography mechanism. As
demonstrated in another article [17], each bead cap-
tures a different dominant protein and co-adsorbs a
small amount of a very few other species. The principle
has been used to identify the hexapeptide ligand struc-
ture specific to selected proteins [18,19]. It should be

fundamental properties has recently been published
[20], together with reviews describing the basic con-
cepts [9,21,22]. The mechanism of action of the Equal-
izer Beads is illustrated in Fig. 1. Rather than acting in
depletion methods, or by selecting a given population
of species, via any possible prefractionation tool, the
beads are meant to adsorb just about any component
of the proteome under analysis, but in a very unusual
way. As shown in the lower left graph (Fig. 1), the
relative abundance of proteins is such that a few are
present in a large excess, whereas the vast majority are
present at a concentration often considerably below
the detection limit. As, in principle, each protein
species has the same number of baits available on the
adsorbing pearls, the species present in vast excess
quickly saturate their ligand, leaving the remainder
unbound in solution. In contrast, rare and very rare
species keep being adsorbed to their respective ligand,
thus being depleted (or very nearly so) from the
Fig. 1. Illustration of the mechanism of action of Equalizer Beads. Bottom panel: relative protein abundances in a generic proteome (left) ver-
sus ‘normalized’ protein abundances after treatment with the hexapeptide ligand library (right). Upper right: adsorbed proteins can be eluted
en bloc, or with sequential treatments of increasing strength.
P. G. Righetti and E. Boschetti Sherlock Holmes and the proteome
FEBS Journal 274 (2007) 897–905 ª 2007 The Authors Journal compilation ª 2007 FEBS 899
solution. This results in ‘normalization’ of the relative
abundance ratios (lower right panel, Fig. 1), rendering
the vast majority of proteins amenable to further ana-
lysis and identification by MS or any other appropriate
tool.
The technique described is not yet commercially

In contrast, the first eluate (in 2.2 m thiourea, 7 m urea
and 4% CHAPS) allowed identification of 334 unique
protein species, and the second eluate (in 9 m urea
titrated to pH 3.8 with 5% acetic acid) an additional
148 species. By eliminating the redundancies and
counting all the species detected, we arrived at a total
of 471 unique protein species in urine [23]. This com-
pares quite favourably with the best data available in
the literature so far, which were obtained using much
more complex technologies and experimental proto-
cols, such as the data of Pieper et al. [24], who repor-
ted 150 unique protein annotations (obtained by
extensive sample prefractionation and two-dimensional
map analysis). However, in the most recent report [25],
1543 proteins were identified in urine samples obtained
from 10 healthy donors, using highly sophisticated
methodology involving analysis of the tryptic digests
via a linear ion trap-Fourier transform (LTQ-FT) and
a linear ion trap-orbitrap (LTQ-Orbitrap) mass spec-
trometers.
A similar approach was adopted, exploiting our pep-
tide library beads, for a large-scale proteomic study of
human blood serum. After ‘equalizing’ sera on the
hexameric peptide baits, analysis by liquid chromato-
graphy of trypsin hydrolyzates coupled with high-
resolution MS resulted in the identification of 3869 or
1559 proteins, depending on how the 95% confidence
was estimated. In either case, the analysis showed that
ligand beads were able to capture a large number of
proteins in a single operation [26]. To determine what

should also work, in principle, in the case of cell and
tissue extracts from any origin. In fact, cell and tissue
lysates should also exhibit a similar disparity in protein
concentration ranges to that found in body fluids. It is
a fact that, when a total cell extract is examined, for
instance, by two-dimensional maps, the most intense
spots are those from cytoskeletal proteins and house-
keeping proteins. Here also rare and very rare proteins
Sherlock Holmes and the proteome P. G. Righetti and E. Boschetti
900 FEBS Journal 274 (2007) 897–905 ª 2007 The Authors Journal compilation ª 2007 FEBS
cannot be brought to the forefront. As an example,
Fig. 4A,B shows SDS ⁄ PAGE profiles of Escherichia
coli and Saccharomyces cerevisiae extract, respectively,
before and after treatment with Equalizer Beads. In
both cases, it can be appreciated that a much larger
number of bands is visible over the entire trace, inclu-
ding lower molecular mass protein ⁄ peptides that often
escape detection by conventional means. In the partic-
ular case of E. coli, bands were cut out from the elec-
trophoresis gel (Fig. 4A, lane b) to identify proteins by
in-gel digestion followed by liquid chromatography-
MS ⁄ MS analysis. The protein identity of several of
them was reported by Thulasiraman et al. [20]. All
these proteins were of low abundance. For instance, on
the basis of previous work, ADP-l-glycero-b-manno-
heptose-6-epimerase is present at  220 copies per cell;
another five enzymes listed (NADH–quinone oxidore-
ductase chain C ⁄ D; tagatose-6-phosphate kinase, gat-Z;
glutamate-1-semialdehyde 2,1-aminomutase; glycine ace-
tyltransferase; galactitol-1-phosphate-5-dehydrogenase)

28
17
14
M.wt Stds
Pooled Elution
3
rd
wash
2
nd
wash
2
nd
wash
1
st
wash
1sth elution
FT
Serum
Fig. 2. Analysis of human serum proteins before and after Equalizer
Bead treatment. One-dimensional SDS ⁄ PAGE profiles. Staining
with colloidal Coomassie Blue. Lanes 1)4 refer to control serum
(untreated), flow through (FT) after bead treatment, followed by
two washing steps, respectively. Lanes 5)8 refer to first elution
en block (with 6
M guanidine hydrochloride, pH 6.0) followed by
two washing steps, and finally SDS ⁄ PAGE of all pooled eluates,
respectively. Lane 9: SDS profile of molecular mass standards.
Fig. 3. Analysis of mouse serum proteins by SDS ⁄ PAGE. Lanes: 1,

of the selected antibodies. In conclusion, all detection
methods for host cell proteins have a challenging
problem, namely, how to deal with very low concen-
trations of contaminating proteins present in ‘pure’
biopharmaceuticals after separation ⁄ purification with
current processing techniques.
Aware of these limitations, we have used the Equal-
izer Bead library to track these very low level impurit-
ies, and have already reported a couple of most
promising applications [29,30]. We give here an exam-
ple of such an impurity ‘amplification’, as applied to
purified monoclonal antibodies produced in hybridoma
cells. Figure 5A shows a two-dimensional map of
control monoclonal antibodies, purified with a merca-
pto-ethyl-pyridine resin [31,32], where very few con-
taminants are visible. After treatment with Equalizer
Beads (Fig. 5B), a large number of new spots appear.
Most of them were excised, digested and subjected to
liquid chromatography-MS ⁄ MS analysis. Two classes
of ‘contaminants’ could be detected: (a) mouse hybri-
doma proteins and culture broth proteins (notably
BSA along with its fragments); (b) a large number of
fragments of the monoclonal antibodies produced.
This seems to be a general trend with all recombinant
DNA products we have analysed so far. It should be
emphasized here that the unique ability of Equalizer
Beads to track and concentrate such impurities is a
process that could be (the necessary changes having
been made) compared with PCR for amplification of
nucleic acid fragments, allowing the detection of pro-

ity of the target protein but also for the qualitative
and quantitative presence of traces of impurity.
Equalizer Beads can be applied here for two proces-
ses: in the first instance, for tracking and concentrating
such impurities, so as to render them amenable to
identification by MS and other analytical techniques
(in this case, a small amount of beads is incubated
with large sample volumes and quantities); by the same
token, if now the beads are in excess over the sample
amount, the beads will also remove such impurities
and thus would be the ideal final ‘polishing’ step for
such biopharmaceuticals [29].
A panacea?
It is intrinsic to human nature to try to overemphasize
the importance of any innovation, with claims often
vastly exceeding what can be achieved in practice
with any novel concept or methodology; in daily use,
such innovations rarely meet the expectations. Science
grows by small increments, quantum jumps being rare
events. Panaceas existed only in legends and the dreams
of sorcerers and healers, and they were scorned in the
famous comedy of Molie
`
re, Le Malade Imaginaire,
where the candidate physicians would advocate only a
single remedy for any possible disease, and hardly a
mild one at that (clysterium practicare, postea salassare,
infinem purgare). We will thus briefly highlight the major
advantages as well as the limitations of the present
approach. The advantages are at least twofold: while

ism is rather delicate (it encompasses all types of bonds
250-
150-
100-
75-
50-
37-
25-
20-
15-
10-
M
r
(kDa)
M
r
(kDa)
3 pI 10
A
250-
150-
100-
75-
50-
37-
25-
20-
15-
10-
3 pI 10

protein with unexpected behaviour: apolipoprotein J
(Apo J), which is greatly enriched compared with all
other serum components, rendering it the most abun-
dant component after equalization. Apo J possesses a
large number of binding sites for several components,
suggesting that it may recognize more than one hexa-
peptide ligand, thus saturating an abnormal number of
sites in a larger bead population compared with other
‘well-behaved’ proteins. Close examination of two-
dimensional maps suggests that a few other proteins
might exhibit similar behaviour, although to what
extent this abnormal behaviour will affect the total
proteome of a tissue is yet to be investigated.
Conclusions
We briefly summarize here the major points worth
considering when using the heaxapeptide combinatorial
library in any proteome analysis. Here is what can be
accomplished with this method: (a) amplification,
detection and identification of protein traces, partic-
ularly in various biological fluids and extracts, detec-
tion of host cell protein in recombinant pure proteins;
(b) identification of specific ligands for protein; (c) pol-
ishing step in downstream processing; (d) discovery of
biomarkers of diagnostic interest; (e) protein–protein
interaction studies.
Acknowledgements
PGR is supported by grants from the European Com-
munity (Allergy card), by PRIN 2006 (MIUR, Rome)
and by Fondazione Cariplo. We thank providers of
biological fluids, such as E. coli extracts (Dr S. Lin)

9 Righetti PG, Castagna A, Antonioli P & Boschetti E
(2005) Prefractionation techniques in proteome analysis:
the mining tools of the third millennium. Electrophoresis
26, 297–319.
10 Garbis S, Lubec G & Fountoulakis M (2005) Limita-
tions of current proteomics technologies. J Chromatogr
A 1077, 1–18.
11 Vlahou A & Fountoulakis M (2005) Proteomic appro-
aches in the search for disease biomarkers. J Chromatogr
B 814 , 11–19.
12 Huber LA, Pfaller K & Vietor I (2003) Organelle pro-
teomics: implications for subcellular fractionation in
proteomics. Circ Res 92, 962–968.
13 Pieper R, Gatlin CL, Makusky AJ, Russo PS, Schatz CR,
Miller SS, Su Q, McGrath AM, Estock MA, Parmar PP,
et al. (2003) The human serum proteome: display of
nearly 3700 chromatographically separated protein
spots on two-dimensional electrophoresis gels and
identification of 325 distinct proteins. Proteomics 3,
1345–1364.
14 Echan LA, Tang HY, Ali-Khan N, Lee KB & Speicher
DW (2005) Depletion of multiple high-abundance
Sherlock Holmes and the proteome P. G. Righetti and E. Boschetti
904 FEBS Journal 274 (2007) 897–905 ª 2007 The Authors Journal compilation ª 2007 FEBS
proteins improves the protein profiling capacities of
human serum and plasma. Proteomics 5, 3292–3303.
15 Shen Y, Kim J, Strittmatter E, : F, Jacobs JM,
CampDG, IIFang R, Tolie
´
N, Moore RJ & Smith RD

tic proteome’. Proteomics 6, 3980–3992.
23 Castagna A, Cecconi D, Sennels L, Rappsilber J, Guer-
rier L, Fortis F, Boschetti E, Lomas L & Righetti PG
(2005) Exploring the hidden human urinary proteome
via ligand library beads. J Proteome Res 4, 1917–1930.
24 Pieper R, Gatlin CL, McGrath AM, Makusky AJ,
Mondal M, Seonarain M, Field E, Schatz CR, Estock
MA, Ahmed N, et al. (2004) Characterization of the
human urinary proteome: a method for high-resolution
display of urinary proteins on two-dimensional electro-
phoresis gels with a yield of nearly 1400 distinct protein
spots. Proteomics 4, 1159–1174.
25 Adachi J, Kumar C, Zhang Y, Olsen JV & Mann M
(2006) The human urinary proteome contains more than
1500 proteins including a large proportion of mem-
branes proteins. Genome Biol 7, R80–R90.
26 Sennels L, Salek M, Lomas L, Boschetti E, Righetti PG
& Rappsilber J (2007) Proteomic analysis of human
blood serum using peptide library beads. Mol Cell
Proteomics in press.
27 States DJ, Omenn GS, Blackwell TW, Fermin D, Eng
J, Speicher DW & Hanash SM (2006) Challenges in
deriving high-confidence protein identifications from
data gathered by a HUPO plasma proteome collabora-
tive study. Nat Biotechnol 24, 333–338.
28 Lathrop JT, Hayes TK, Carrick K & Hammond DJ
(2005) Rarity gives a charm: evaluation of trace proteins
in plasma and serum. Expert Rev Proteomics 2, 393–
406.
29 Fortis F, Guerrier L, Righetti PG, Antonioli P &