Purification and characterization of the cysteine
proteinases in the latex of Vasconcellea spp.
Tina Kyndt
1,2
, Els J. M. Van Damme
1
, Jozef Van Beeumen
3
and Godelieve Gheysen
1,2
1 Department of Molecular Biotechnology, Faculty of Bioscience Engineering, Ghent University, Belgium
2 Institute for Plant Biotechnology for Developing Countries (IPBO), Ghent University, Belgium
3 Laboratory of Protein Biochemistry and Protein Engineering, Ghent, Belgium
Articulated laticifers, containing a milky latex, are pre-
sent in all organs of members of the small plant family
Caricaceae [1]. The two economically most important
genera of this family are the commonly grown tropical
species Carica papaya and the group of highland papa-
yas (Vasconcellea spp.), of which many are locally tol-
erated and ⁄ or semicultivated for their fruit production.
Although they used to be classified in one single genus
(Carica), recent phenetic and phylogenetic results [2]
have shown a clear separation between C. papaya and
the 21 species of Vasconcellea, confirming their classifi-
cation into two separate genera [3].
Experimental evidence has shown that latex gener-
ally contributes to protecting the plant against pre-
dators [4,5] in both a mechanical (by wound
coagulation) and chemical (by the presence of toxic
substances) manner. While proteinase inhibitors (PIs)
are generally believed to actively contribute to plant

the characteristic features of all known papain-class cysteine proteinases,
and a phylogenetic analysis revealed the existence of several papain and
chymopapain homologues in these species. Ion-exchange chromatography
and gel filtration procedures were applied on latex of V. · heilbornii in
order to characterize its cysteine proteinases at the protein level. Five major
protein fractions (VXH-I–VXH-V) revealing very high amidase activities
(between 7.5 and 23.3 nkatÆmg protein
)1
) were isolated. After further purifi-
cation, three of them were N-terminally sequenced. The observed microhet-
erogeneity in the N-terminal and cDNA sequences reveals the presence of
several distinct cysteine proteinase isoforms in the latex of Vasconcellea spp.
Abbreviations
AA, amino acid; BAPNA, a-N-benzoyl-
L-arginine 4-nitroanilide; MP, maximum parsimony; PI, proteinase inhibitor; VXH,
Vasconcellea · heilbornii.
FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS 451
the total latex [3], at a molar concentration that
probably exceeds 1 mm [7]. Cysteine proteinases are
proteolytic enzymes that depend upon a cysteine resi-
due for activity. Within this group of enzymes, at
least seven different evolutionary origins have been
determined, allocating them to seven clans, each con-
sisting of several related families [8].
Because of their economic importance in the bever-
age, food and pharmaceutical industries, constituents
of the latex of C. papaya have been investigated thor-
oughly. The four cysteine proteinases present in the
latex of C. papaya, namely papain (EC 3.4.22.2),
chymopapain (EC 3.4.22.6), caricain (formerly known

not consistent about the level of proteolytic activity,
probably due to varying experimental conditions, they
confirm the potential of Vasconcellea spp. for commer-
cial proteinase production.
In this study, we report on the identification and
characterization of cDNA sequences coding for cys-
teine proteinases in the latex of V. stipulata and
V. · heilbornii. Based on these sequences, the evolution
of the cysteine proteinases within the Caricaceae was
investigated. Furthermore, proteolytic enzymes were
purified from the latex of V. · heilbornii and the N-ter-
minal amino acid sequence characterized.
Results and Discussion
Amidase activity of total latex
Proteolytic activity, measured as amidase activity per
milligram dried latex using the BAPNA (a-N-benzoyl-
l-arginine 4-nitroanilide) substrate, was evaluated by
Scheldeman et al . [23] for several Vasconcellea spe-
cies. They reported that Vasconcellea cundinamarcensis,
V. stipulata and V. · heilbornii latex show a proteolytic
activity that is approximately 4–13 times higher than
the papaya reference. Two factors might play a role in
the higher activity observed in Vasconcellea spp.: (1) a
higher protein content in their latex; and (2) the pres-
ence of other cysteine proteinases or isoforms in the
latex. To investigate these two hypotheses, the protein
concentration per milligram of dried latex, as well as
the amidase activity per milligram of protein, was
measured for three species (Table 1). The results show
that the protein concentration in Vasconcellea latex is

)
Proteolytic activity
(nkatÆmg latex
)1
)
C. papaya 26.69 6.50 0.17
V. monoica 36.31 8.09 0.29
V. stipulata 40.57 12.98 0.53
V. · heilbornii 31.75 8.65 0.27
Cysteine proteinases of Vasconcellea spp. T. Kyndt et al.
452 FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS
the protein content and activation of proteolytic
enzymes in its latex [24]. While the fruits used in this
experiment were all tapped for the first time, we are
not aware of the frequency of tapping in other studies.
These and probably several other, as yet unknown,
factors affecting the activity and concentration of latex
proteins complicate comparisons with previous studies.
Our results (Table 1), obtained using equal condi-
tions for all plants, clearly show that the higher pro-
teolytic activity is only to a certain extent due to a
higher protein concentration in latex of Vasconcellea
fruit. The presence of other, possibly more proteolyti-
cally active enzymes will be evaluated in the following
experiments.
cDNAs coding for cysteine proteinases in
V.
·
heilbornii and V. stipulata
Using a PCR approach with a cysteine proteinase

single papain in papaya [18]. Another cysteine protein-
ase, CC-23, was purified from V. cundinamarcensis
latex and its corresponding DNA fragment cloned by
Pereira et al. [20]. The N-terminal sequence appeared
to be different from the N-terminal sequences reported
for CC-I to CC-IV. The authors suggested the existence
of six to seven cysteine proteinases in latex from V. cun-
dinamarcensis, some of which may be isoforms, as in the
case of chymopapain. For V. cundinamarcensis, one
incomplete DNA-sequence (no stop codon) called
CC-23, and five amino acid sequences were traced in the
database: CC-Ia, CC-Ib, CC-II, CC-III, CC-IV. CC-II
and CC-IV were only N-terminally sequenced [17].
Although CC-Ia (213 amino acids), CC-Ib (213 amino
acids) and CC-III (214 amino acids) have a calculated
mass corresponding to mass spectrometric results
[18,19], they might be incomplete at the 3¢ end.
Papaya proteinases are naturally synthesized with
N-terminal signal and pro-peptides. The pro-regions
aid the folding of the mature enzymes and act as
selective high affinity inhibitors to prevent inappropri-
ate proteolysis within the plant [25]. The enzymes are
present in the latex as inactive precursors and are
activated in response to wounding of the plant [26].
Similar to other known cysteine proteinases from Car-
icaceae, the deduced protein sequences of VXH-A, -B,
-C and -D, and VS-A and -B, were predicted to con-
tain a signal peptide. For most of them, this signal
peptide contains 26 amino acids. However, the signal
peptide prediction software SignalP predicts VXH-B to

acids according to their position in mature papain).
One putative N-glycosylation site occurs in the prore-
gion of VS-A and VXH-A:
115
NWS
117
. The glycan in
the propeptide might aid in the protection against deg-
radation, or in targeting or maturation of the enzyme,
as was shown before in cathepsin C, which also belongs
to the papain family of cysteine proteinases [27].
The overall similarity at the amino acid level
between all known cysteine proteinases from Carica-
ceae is 73%. Tables 2 and 3 show the percentage
sequence similarity between the obtained cDNA
T. Kyndt et al. Cysteine proteinases of Vasconcellea spp.
FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS 453
Fig. 1. Alignment of translated cDNAs of V. stipulata (VS), V. · heilbornii (VXH), V. cundinamarcensis (CC) and C. papaya. The transition
between signal peptide and proregion is indicated by a red arrow. The black arrow shows the beginning of the mature enzyme. Cysteine res-
idues involved in disulfide bridges are indicated with blue stars. Yellow stars show amino acids which are important for proteolytic activity.
Cysteine proteinases of Vasconcellea spp. T. Kyndt et al.
454 FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS
sequences and all known cysteine proteinases from the
Caricaceae. VXH-A and VS-A are almost identical
(99%) at the amino acid and nucleotide level. A single
amino acid substitution is present at position 122 of
VXH-A, where a serine is replaced by a proline in
VS-A. As is the case for CC-Ia, CC-Ib [18], CC23 [20],
CCIII [19], and papain [11], VS-A, VS-B and VXH-A
lack the insertion of four amino acids between position

major clusters. Cluster I combines the five chymo-
papain isoforms of C. papaya with VXH-C, VS-B,
CC-III and CC23. These results confirm the hypothesis
of Walraevens et al. [17] that CC-III is a chymopapain
homologue and predict VXH-C and VS-B to be the
corresponding genes in V. · heilbornii and V. stipulata,
respectively. In addition, this suggests CC23 to be a
Table 3. Percentage pairwise sequence similarity between the cys-
teine proteinase cDNA sequences of V. stipulata and V. · heilbornii.
Values higher than 85% are indicated in bold.
VS-A VS-B VXH-A VXH-B VXH-C VXH-D
VS-A 100.0
VS-B 66.1 100.0
VXH-A 99.7 67.3 100.0
VXH-B 91.0 64.7 91.3 100.0
VXH-C 63.8 86.5 64.4 63.0 100.0
VXH-D 95.7 65.9 96.0 91.4 63.0 100.0
Fig. 2. The 50% Majority Rule Consensus
tree of three Maximum parsimonious trees
of cysteine proteinases of V. stipulata (VS),
V. · heilbornii (VXH), V. cundinamarcensis
(CC) and C. papaya. (Tree length ¼ 498,
consistency index ¼ 0.8193, retention
index ¼ 0.8811). Bootstrap values are indi-
cated above the branches.
Cysteine proteinases of Vasconcellea spp. T. Kyndt et al.
456 FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS
paralogue of CC-III, possibly also coding for a chymo-
papain-like enzyme.
Cluster II contains papain, next to VXH-A, VXH-B,

being involved in the hybrid origin of V. · heilbornii
[28], and (2) their reported recent speciation [2]. Since
no caricain or glycyl endopeptidase homologues have
yet been found in V. cundinamarcensis, V. stipulata
and V. · heilbornii, it is not possible to draw conclu-
sions about their evolution.
Fractionation of the proteinases
from V.
·
heilbornii
In an attempt to purify the proteinases from V. · heil-
bornii, latex was collected and subjected to ion
exchange chromatography. Figure 3 displays a typical
elution profile from the Mono S 5 ⁄ 50 GL column of
the total dialysed soluble fraction of the latex of
V. · heilbornii. In general, five peaks with apparent
microheterogeneity can be distinguished: VXH-I,
VXH-II, VXH-III, VXH-IV and VXH-V, all showing
amidase activity (Table 4). Amidase activity was inhib-
ited by the addition of the cysteine proteinase inhibitor
E-64. This observation clearly showed that the proteo-
lytic activity was due to cysteine proteinases solely.
Specific amidase activity of VXH-I toVXH-V ranged
from 7.5 to 23.3 nkatÆmg protein
)1
, which is 4.5- to
14-fold higher than the activity of chymopapain from
papaya latex (1.68 nkatÆmg
)1
) [28]. The activities

T. Kyndt et al. Cysteine proteinases of Vasconcellea spp.
FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS 457
fruits all year round, our purification was hampered
by the limited amount of starting material available.
Therefore, fractions VXH-I–VXH-V from the first ion
exchange chromatography were pooled, and rechroma-
tographed on a gel filtration column in an attempt to
remove the smaller proteins. Size-exclusion chromato-
graphy yielded essentially one large peak (data not
shown) which was divided into pools A and B.
Whereas the later fractions (pool B) revealed two pro-
tein bands of 27 and 30 kDa after SDS ⁄ PAGE, the
earlier fractions (pool A) showed an extra protein
band of higher molecular weight (33 kDa) (Fig. 4).
SDS ⁄ PAGE results confirm that the smaller contamin-
ating proteins have been removed after gel filtration.
The size of the polypeptides present in pools A and B
is equivalent to or slightly larger than the molecular
mass reported for papain (23 kDa), chymopapain
(27 kDa) [30], caricain (24 kDa) [11], CC-IV (28 kDa)
[29] and CC23 (23 kDa) [20]. Subsequently, the pro-
teins in pools A and B were re-fractionated using ion-
exchange chromatography (Fig. 5A,B). Pool A yielded
three major peaks. Comparison of the elution profiles
in Figs 3 and 5A,B suggests that these peaks corres-
pond to VXH-I, VXH-III and the second part of
VXH-IV (VXH-VIb). Pool B contained only two
peaks (Fig. 5B), corresponding to VXH-III and the
earlier part of VXH-IV (VXH-IVa). Apparently peaks
VXH-II and VXH-V are not present in pools A and B

family. Sequencing of VXH-I and VXH-III yielded
two signals of equal intensity at positions 7 and 17,
respectively, suggesting that these pools might contain
different isoforms. The N-terminal sequences of VXH-
IVa and VXH-IVb reveal different amino acids at
positions 9, 18 and 20, confirming that peak VXH-IV
holds at least two different cysteine proteinases.
All N-terminal sequences show between 65 and
100% similarity, with an average of 80%. Such a high
degree of homology makes it difficult to decide which
form of VXH corresponds to which papaya or V. cun-
dinamarcensis proteinase. The identical N-terminal
sequence of VXH-IVb, CC-III and CC-IV suggests
that these might be homologous proteins, but complete
sequencing is necessary to confirm this result. Based
on the 100% identity between the deduced amino acid
sequence of VXH-C and the N-terminal sequence of
VXH-I (Fig. 1) we can assume that cDNA clone
VXH-C encodes VXH-I.
Table 4. Amidase activity of pooled fractions VXH-I–VXH-V, meas-
ured by BAPNA degradation. nkat: amount of enzyme that hydro-
lyses 1 nmol BAPNA per second.
Pool nkatÆmg protein
)1
VXH-I 23.3
VXH-II 12.5
VXH-III 22.1
VXH-IV 15.0
VXH-V 7.5
Fig. 4. SDS ⁄ PAGE electrophoresis of pool A and pool B from the

Unripe fruits of plants grown in the greenhouse were the
source of latex used in this study. Latex was collected by
making several incisions into the surface of the unripe fruit
using a sharp blade. An equal volume of 100 mm thio-
ureum was added to avoid oxidation, as recommended by
Azarkan et al. [31], and the latex was stored at )20 °Cin
the dark, until use.
Electrophoresis
Protein samples were electrophoresed in 15% polyacryl-
amide denaturing gels (SDS ⁄ PAGE) [32] after boiling for
5 min at 95 °C. Electrophoresis was performed for 90 min
at 150 V. Gels were stained with 0.1% Coomassie blue
A
B
L
L
Fig. 5. Second ion exchange chromatography
after size-selection of pool B from the latex of
V. · heilbornii. Columns: mono S 5 ⁄ 50 GL;
buffer: 50 m
M sodium acetate, pH 5.0; flow
rate: 2 mLÆmin
)1
; gradient: 0–1 M NaCl,
pH 5.0. The triangles show absorbance (A
280
)
of each fraction. The black line indicates the
conductivity. (A) Fractionation of pool A;
(B) fractionation of pool B.

RNA was extracted using the Qiagen Rneasy Plant Mini
Kit (Qiagen GmbH, Hilden, Germany) from unripe fruits.
The RNA was transformed into double-stranded cDNA by
LD PCR with the Creator SMART
TM
cDNA Library Con-
struction kit (BD Biosciences, Heidelberg, Germany). Dur-
ing this procedure, adapters are ligated to the 5¢ and 3¢ end
of the cDNAs. The 3¢ ends of cysteine proteinases were spe-
cifically amplified using an internal cysteine proteinase
specific primer (CyPr: 5¢-AAGGAGCYGTNACTCCT
GTAA-3¢), derived from a central conserved region present
in all known cysteine proteinases from Caricaceae, and a
primer complementary to the previously built-in 3¢ adapt-
ers. The total PCR volume of 20 lL consisted of 2 lL10·
diluted cDNA (from LD PCR), 2 lL CyPr-primer (10 lm),
2 lL3¢ adapter primer (10 lm), 2 lL Pfx amplification buf-
fer (Invitrogen, Paisley, UK), 2 lL dNTPs (5 mm), 0.4 lL
Pfx polymerase, 0.4 lL MgSO
4
(50 mm) and 9.2 lL water.
The PCR programme involved an initial denaturation for
4 min at 95 °C, followed by 30 cycles of 30 s at 95 °C, 30 s
at 55 °C and 90 s at 68 °C, and a terminal extension of
10 min at 68 °C.
PCR products were separated on 1% agarose 0.5· TAE
(20 mm Tris-Acetate, 0.5 mm EDTA) gels and purified
using the QIAquick Gel Extraction Kit (Qiagen). As the
high-fidelity Pfx polymerase was used in the PCR reaction,
terminal 3¢ A-ends had to be added to the PCR products

group. The consistency index and retention index were calcu-
lated. Support for the different clades was tested by
bootstrap analysis (100 replicates using heuristic search,
random sequence addition and TBR branch-swapping).
Fig. 6. N-Terminal amino acid sequences of proteinases from
V. · heilbornii (VXH) and V. stipulata (VS), compared with the
known sequences from V. cundinamarcensis (CC) and C. papaya
(Cp). In the case of double signals, both amino acids are specified.
Cysteine proteinases of Vasconcellea spp. T. Kyndt et al.
460 FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS
Chromatographic procedures
The latex of V. · heilbornii was exhaustively dialyzed
against cold H
2
O, centrifuged for 10 min at 3000 g, and the
supernatant filtered through a Whatman paper. The filtered
solution was chromatographed on an AKTA fplc system
(Amersham Biosciences, Uppsala, Sweden) using a Mono S
5 ⁄ 50 GL column (5 · 50 mm) previously equilibrated with
50 mm sodium acetate, pH 5.0. The column was eluted
using a linear gradient of 0–1 m NaCl in 50 mm sodium
acetate (pH 5). Protein concentration and amidase activity
of all fractions was measured. SDS ⁄ PAGE of fractions with
high activity revealed the polypeptide composition. As
many of smaller contaminating polypeptides were observed,
apart from bands of the expected mass (22–30 kDa), fur-
ther purification was performed by gel filtration on a Seph-
acryl S-200 column (3 · 70 cm). The column was run in
20 mm 1,3-diamino propane, pH 11. Ion-exchange chroma-
tography of two pools (A and B) that contained protein

Romero-Motochi JP, Scheldeman X, Goetghebeur P,
Van Damme P & Gheysen G (2005) Molecular phylo-
geny and evolution of Caricaceae based on rDNA inter-
nal transcribed spacer (ITS) and chloroplast sequence
data. Mol Phyl Evol 37, 442–459.
3 Badillo VM (2000) Carica L. vs. Vasconcella St. Hil.
(Caricaceae): con la rehabilitacio
´
n de este u´ ltimo. Erns-
tia 10, 74–79.
4 El Moussaoui A, Nijs M, Paul C, Wintjens R, Vincen-
telli J, Azarkan M & Looze Y (2001) Revisiting the
enzymes stored in the laticifers of Carica papaya in the
context of their possible participation in the plant
defence mechanism. Cell Mol Life Sci 58, 556–570.
5 Konno K, Hirayama C, Nakamura M, Tateishi K,
Tamura Y, Hattori M & Kohno K (2004) Papain pro-
tects papaya trees from herbivorous insects: role of
cysteine proteases in latex. Plant J 37, 370–378.
6 Baldwin IT & Preston CA (1999) The eco-physiological
complexity of plant responses to plant herbivores.
Planta 208, 137–245.
7 Oberg KA, Ruysschaert J, Azarkan M, Smolders N,
Zerhouni S, Wintjens R, Amrani A & Looze Y (1998)
Papaya glutamine cyclase, a plant enzyme highly resis-
tant to proteolysis, adopts an all b conformation. Eu J
Biochem 258, 214–222.
8 Barrett AJ & Rawlings ND (2001) Evolutionary lines of
cysteine peptidases. Biol Chem Hoppe-Seyler 382, 727–
733.

Carica papaya prochymopapain isoforms in Escherichia
coli. Plant Sci 145, 41–47.
17 Walraevens V, Jaziri M, Van Beeumen J, Schnek AG,
Kleinschmidt T & Looze Y (1993) Isolation and charac-
terization of the cysteine-proteinases from the latex of
Carica candamarcensis Hook. Biol Chem Hoppe-Seyler
374, 501–506.
18 Walraevens V, Vandermeers-Piret M, Vandermeers A,
Gourlet P & Robberecht P (1999) Isolation and charac-
terization of the CCI papain-like cysteine proteinases
from the latex of Carica candamarcensis Hook. Biol
Chem Hoppe-Seyler 380, 485–488.
19 Jaziri M, Kleinschmidt T, Walraevens V, Schnek AG &
Looze Y (1994) Primary structure of CC-III, the glyco-
sylated cysteine proteinase from the latex of Carica
candamarcensis Hook. Biol Chem Hoppe-Seyler 375,
379–385.
20 Pereira MT, Lopes MTP, Meira WO & Salas CE (2001)
Purification of a cystein proteinase from Carica canda-
marcensis L. & cloning of a genomic putative fragment
coding for this enzyme. Prot Expr Purif 22, 249–257.
21 Baeza G, Correa D & Salas C (1990) Proteolytic
enzymes in Carica candamarcensis. J Sci Food Agric 51,
1–9.
22 Dhuique-Mayer C, Caro Y, Pina M, Ruales J, Dornier
M & Graille J (2001) Biocatalytic properties of lipase
in crude latex from Babaco fruit (Carica pentagona).
Biotechn Lett 23, 1021–1024.
23 Scheldeman X, Van Damme P & Romero-Motochi JP
(2002) Highland Papayas in Southern Ecuador: need for

31 Azarkan M, El Moussaoui A, Van Wuytswinkel D, De-
hon G & Looze Y (2003) Fractionation and purification
of the enzymes stored in the latex of Carica papaya.
J Chrom B 790, 229–238.
32 Laemmli UK (1970) Cleavage of structural proteins
during the assembly of the head of bacteriophage T4.
Nature 227, 680–685.
33 Hall TA (1999) BioEdit: a user-friendly biological
sequence alignment, ed. and analysis program for
Windows 95 ⁄ 98 ⁄ NT. Nucleic Acids Symp 41, 95–98.
34 Swofford DL (1998) paup* 4.0: Phylogenetic Analysis
Using Parsimony, Beta Version 4.0b10. Sinauer Associ-
ates, Sunderland, MA.
Cysteine proteinases of Vasconcellea spp. T. Kyndt et al.
462 FEBS Journal 274 (2007) 451–462 ª 2006 The Authors Journal compilation ª 2006 FEBS