Porphyrins, chlorins, bacteriochlorins and other closely related tetrapyrrolic
macrocycles occur widely in nature with significant roles in various biological
processes. Protoporphyrin IX (PPIX), the immediate precursor of heme; chlorophylls
which play important roles in photosynthesis and vitamin B12, are
representatives of those molecules.
Many tetrapyrrolic macrocycles are effective photosensitizers in Photodynamic
Therapy (PDT) (Maiya, 2000; Miah,
2002a; Castano et al., 2004; Tjahjono,
2006). Tetrapyrrole molecules are also fluorescent compounds, thus its imaging
capability can be applied in photodynamic detection (Shahbazi-Gahrouei
and Khodamoradi, 2007; Miah, 2002b). Moreover, a
photo insectiside activities has also been reported for tetrapyrrole derivatives
(El-Tayeb et al., 2011).
PDT is a treatment technique for cancer and for certain benign conditions which
use a photosensitizing drug and light at a specific wavelength to produce reactive
oxygen species in cells (Castano et al., 2004).
This two-part procedure makes it possible to avoid the toxicity which associated
with drug-based therapy, because only the light-activated sensitizer will yield
a cytotoxic effect to target cell. However, it is necessary to consider the
long-term dark toxicity of those compounds.
Prior to labor-intensive and expensive laboratory toxicity assays, a toxicity
predictions by (Quantitative) Structure-activity Relationships ((Q)SARs) models
have been used to support in hypothesis and prioritizing further experimental
studies. A large number of relationships have been reported that the biological
effect, mathematically determined as the output of the model, was defined in
relation to some chemical parameters, identified as the model inputs (Cronin
and Livingstone, 2004).
Recently, we have focused our interest on 1-hydroxyethyl derivative of tetrapyrrolic
macrocycles bearing carboxylic acid groups. Tetrapyrrolic macrocycles with conjugated
substituents bearing carboxylic acid group have red-shift absorption compared
to non-substituted tetrapyrrolic macrocycles. The red-most absorption
is ideal criteria for PDT as the light can penetrate deep into tissues (Eriksson
and Eriksson, 2011). Naghavi and Baygi (2009) have
developed a model that is able to predict the depth of tissue necrosis during
PDT. The present of 1-hydroxyethyl substituent is expected to increase the hydrophilicity,
an advantage when the drug is formulated as parenteral dosage form.
In the present study, we investigated the influence of carboxylic acid group
and 1-hydroxyethyl side chain (substituted vinyl groups) on the toxic potency
of porphyrin, chlorin and bacteriochlorin. The aim of this study was to predict
the toxicities of some tetrapyrrolic macrocycles bearing carboxylic acid groups,
i.e., ecotoxicity and one for human toxicological endpoints. The results of
the study may serve as a useful considerant in the development of more effective
MATERIALS AND METHODS
ECOSAR: ECOSAR is an easy-to-use computer program which are developed
and routinely applied by the US EPA for predicting aquatic toxicity to fish,
aquatic invertebrates (daphnids) and green algae. The programs latest
version (v.1.00a February, 2009) is freely available from the EPA website (ECOSAR,
2009). The input data is the SMILES notation of the substance and the related
log Kow value. If the experimental log Kow is not available,
the log Kow is then calculated by ECOSAR using the sub-routines Kow
Win and or CLOGP. However, predicted values will be eliminated from the dataset
if ECOSAR has indicated that the log Kow of the substance was too
high for correct prediction (log Kow cut off).
TOXTREE: Toxtree was developed by IDEA consult. Ltd. (Sofia, Bulgaria).
It has been made available as a free download (Toxtree, 2010).
The current version (v.2.2.0 October, 2010) includes decision trees for predicting
Cramer rules, Cramer rules with extensions, Verhaar scheme, START biodegradability,
eye irritation and corrosion, structure alerts for the in vivo micronucleus
assay in rodents, Michael acceptors, Benigni/Bossa rulebase (for mutagenicity
and carcinogenicity), skin irritation/skin corrosion, cytochrome P450-mediated
drug metabolism, skin sensitisation alerts and Kroes TTC decision tree. The
main window of Toxtree is classified into three different areas: the compound
properties area, used to resume the available information on the current compound;
the compound structure diagram area which shows a picture of the current compound
and provides an easy way to navigate through the list of compounds in the currently
opened file and the classification area which provides access to the classification
output for the current compound.
To generate predictions with Toxtree, user-defined molecular structures can
be input as SMILES codes or using the built-in 2D structure diagram editor.
The software can also be used to perform batch processing of large numbers of
compounds by importing datasets of various file types.
Kow Win: The log Kow (Kow Win) program estimates the log
octanol/water partition coefficient of organic chemicals using an atom/fragment
contribution method developed at the Syracuse Research Corporation. The log
Kow values generated by Kow Win are then integrated into the ECOSAR
SMILES: The Simplified Molecular Input Line Entry System (SMILES) is
a chemical notation system used to represent a molecular structure by a linear
string of symbols. SMILES notations are comprised of atoms (designated by atomic
symbols), bonds, parentheses (used to show branching) and numbers (used to designate
ring opening and closing positions) (Weininger, 1988).
The PBT-Profiler program has been used to generate SMILES notations for all
investigated compounds that did not have known CAS numbers. Chemical structures
are drawn using the Structure Drawing Interface portion of the program and the
SMILES notation of the investigated compound is generated. The generated SMILES
notation is then input into the ECOSAR and Toxtree model.
RESULTS AND DISCUSSION
The structure of three free-base porphyrin, ten chlorin and two bacteriochlorin
(Fig. 1, Table 1) were inputted into the
ECOSAR and Toxtree software.
|| Chemical structure of (a) Porphyrin (b, c) Chlorin and (d)
|| Porphyrins, chlorins and bacteriochlorins studied
|1The structure (a, b, c and d) refer to Fig.
The studied porphyrins were include 7,12-diethenyl-3,8,13,17-tetramethyl-21H,23H-porphine-2,18-dipropanoic
acid (PPIX); 7-(1-hydroxyethyl)-12-ethenyl-3,8,13,17-tetramethyl-21H,23H-porphine-2,18-dipropanoic
acid (PPIX-1OH) and 7,12-bis(1-hydroxyethyl)-3,8,13,17-tetramethyl-21H,23H-porphine-2,18-dipropanoic
acid (PPIX-2OH). The studied chlorins were include chlorin e6, rhodochlorin,
purpurin 7, rhodin g7, pyropheophorbide a (pyrophe a), chlorin e4,
chlorin p6, isochlorin e4, pheophorbide a (Phe a) and
3-(1-hydroxyethyl) pheophorbide a or Phe a-OH while the studied bacteriochlorin
derivatives were bacteriopheophorbide a (BPhe a) and 3-(1-hydroxyethyl)-bacteriopheophorbide
a (BPhe a-OH).
The ECOSAR program was capable to generating LC50 or EC50
of the studied compounds. The results are presented in Table 2
All of the studied compounds may not be soluble enough to measure this predicted
effect. Nevertheless, these results can be reasonable to compare the toxic potential
of the test compounds each other. In examining the toxic potency data, several
observations are apparent. Firstly, comparing toxic potency for PPIX, PPIX-1OH
and PPIX-2OH, it demonstrated that toxic potency decrease with increasing the
number of hydroxyl substituents.
|| Baseline toxicity (neutral organic SAR) of some tetrapyrrolic
macrocycles as predicted by ECOSAR
|MW: Molecular weight, SAR: Structure-activity relationship,
Kow: Octanol-water partition coefficient
This is due to the hydroxyl groups the same reason was proposed for Phe a and
Phe a-OH, as well as for BPhe a and BPhe a-OH. Second, toxic potency of most
of chlorin derivatives were not significantly different, except rhodin g7
(less toxic). In rhodin g7 the aldehyde group contribute in decreasing
toxicity. Third, toxic potency of bacteriochlorin (i.e., BPhe a) was lower than
that of chlorin (i.e., Phe a). It is an advantageous because bacteriochlorin
are promising candidates for PDT as they display the red-most absorption (Qx)
at longer wavelengths and its absorptivity is also stronger compared to those
of corresponding chlorin.
ECOSAR requires an estimated or measured value of log Kow. Toxicity
values for new chemicals is then calculated by inserting the estimated Kow
into the regression equation and correcting the resultant value for the molecular
weight of the compound. The toxicity value for a substance increases as the
solubility of the compound in water increases (Table 2, Fig.
Based on the SMILES notation, each substance was allocated to a chemical class
as defined by ECOSAR. It might happen however, that due to its molecular structure,
a substance is assigned to more than one chemical class. This was allowed in
the current evaluation as it was attempted to get information on the prediction
capability with respect to the individual ECOSAR chemical classes and the corresponding
SARs. When an acid moiety was found in a molecule, ECOSAR multiplied the predicted
values which were calculated from the SARs of the specific class, e.g. neutral
organics, by a factor of 10. The chemical was then allocated to the same class
but with the acid moiety, e.g. pyrazoles/pyrrole acid (Reuschenbach
et al., 2008). Table 3 shows the results of toxicity
prediction in fish of a specific ECOSAR class.
Tetrapyrrolic macrocycles with fewer carboxylic acid groups showed higher toxic
potency than those with more carboxylic acid groups and greater toxicity was
observed with increasing molecular weight (Table 4, Fig.
3). These results were consistent with previous research (Frank
et al., 2009).
||Plot of log LC50 (fish, 96 h) against log Kow
for some tetrapyrrolic macrocycles as predicted by ECOSAR
|| Toxicity based on ECOSAR class as predicted by ECOSAR
|1Fish LC50 96 h is the dose (mg L-1)
required to kill half the members of fish after 96 h
||Baseline toxicity (neutral organic SAR) of some tetrapyrrolic
macrocycles with one, two and three carboxylic acid groups as predicted
|1Fish LC50 96 h (mg L-1)
required to kill half the members of fish after 96 h
||Predicted toxicity against molecular weight of some tetrapyrrolic
macrocycles with 1,2 and 3 carboxylic acid groups as predicted by ECOSAR
|| Toxtree plugins and classes
The enhanced strength of the electronic charge which induced by the increase
of number of carboxylic acid in a structure could impair uptake by a cellular
membrane, consequently reducing toxicity. Such evidences were observed when
investigating the impact of ionization on the toxicity of aliphatic carboxylic
acids (Seward and Schultz, 1999).
The majority of the data input into the ECOSAR models
database did not contain information regarding to the pH of test solutions.
Therefore, all solutions were assumed to have a pH of 7 and as a result, the
model was not able to accurately predict the toxicity of ionic compounds at
pH other than 7.
Toxtree is able to estimate toxic hazard by applying a decision tree approach.
The decision tree approaches and classes were then used to predict the toxicities
(Table 5, 6).
|| Toxic hazard classification as predicted by Toxtree
|The classification numbers refer to Table 5.*Structure
alerts for reactivity in Toxtree (START) biodegradability, #Structure
alerts for the in vivo micronucleus assay in rodents, Benigni/Bossa
rule base for mutagenicity and carcinogenicity, Cytochrome
P450-Mediated drug Metabolism, ∞Kroes threshold of toxicological
Cramer rules classify chemicals into three structural classes based on a decision
tree. Questions asses different features include structural features (functional
groups, ring substituents, etc.), propensity of reaction, natural occurrence
in body and in traditional foods and the logic of tree relies primarily on knowledge
of common metabolic pathways (Patlewicz et al., 2008).
As shown in Table 6, all of the studied compounds fell into
class 3, indicating that the compounds are substances that permit no strong
initial impression of safety and may even suggest a significant toxicity. This
result probably misclassified as porphyrins, chlorins and other closely related
macrocyclic tetrapyrrole occur widely in nature with significant roles in various
biological processes. In terms of the Cramer classification rules, the studied
tetrapyrrolic macrocycles are naturally present in the life system and therefore,
proposed to be Class 1 by virtue of a positive response to question 1. The list
of normal components in the life system could be extended within Toxtree to
include the compounds as another example. All of the test compounds fell into
class 3 by Cramer rules because a negative response to question 1 due to heterocyclic
rings with complex substituents in the compounds.
Potential mechanisms of toxic action were identified for these compounds through
application of the Verhaar scheme. A number of mechanisms have been identified
that can lead to aquatic toxicity, with the majority of industrial chemicals
exerting their toxic influence via two non-covalent mechanisms; polar narcosis
and non-polar narcosis. According to the Verhaar scheme, all of the compounds
fell into class 5 (Table 6). Compounds which cannot be classified
as belonging to classes 1, 2 or 3 and that are not known to be compounds acting
by a specific mechanism can only be classified as not possible to be classified
according to these rules. Verhaar scheme needs a number of improvements.
One possible scheme would be one which identifies reactive chemicals first,
using the wealth of knowledge currently available. In addition, sub-classes
are required within Verhaar classes 3 and 4 to reflect the number of differing
mechanisms of action that have been identified since the original Verhaar publication
in 1992 (Enoch et al., 2008).
START biodegradation and persistence is a compilation of structural alerts
for environmental persistence and biodegradability. These structural alerts
are molecular functional groups or substructures that are known to be linked
to the environmental persistence or biodegradability of chemicals. According
to START biodegradability, all of the compounds fell into class 2 because they
have two or more rings (Table 6).
Skin irritation and skin corrosion refer to localized toxic effects resulting
from a topical exposure of the skin to a substance. There is strong evidence
that chemicals which are corrosive to the skin should also be classified as
being corrosive to the eye, especially if the assessment is made from knowledge
of acidity and alkalinity. In particular, in the EU and OECD classification
schemes, chemicals that have been found to be corrosive to the skin are automatically
considered to be corrosive to the eye. Toxtree implements the BfR rules (The
German Federal Institute for Risk Assessment) for predicting skin/eye irritation
and corrosion. The system is based on the combined use of two predictive approaches:
exclusion rules based on physicochemical cut-off values to identify chemicals
that do not exhibit a certain hazard (e.g., skin irritation/corrosion) and inclusion
rules based on structural alerts to identify chemicals that do show a particular
toxic potential. According to eye irritation/corrosion and skin irritation/corrosion
methods, all of the compounds were classified as not skin corrosion R34 or R35
and not corrosive to skin, respectively (Table 6). These results
indicated that the compounds were likely safe as photosensitizer for skin cancer
Structure alerts for the micronucleus assay in rodent resulted in a classification
of class 1 for all compounds (Table 6). It means that all
of compounds have at least one positive alerts for the micronucleus assay. This
plugin provides a list of structural alerts for a preliminary screening of potential
in vivo mutagens. These structural alerts are molecular functional groups
or substructures that are known to be linked to a positive in vivo micronucleus
assay. Molecular functional groups or substructures that are known to be linked
to a positive in vivo micronucleus assay are heterocyclic polycyclic
aromatic hydrocarbons and molecules which can form non-covalent interactions
with protein or DNA. Such interaction, as in the case of DNA intercalation or
groove binding, is potentially genotoxic. The negative in vitro results
in genotoxicity testing are usually considered sufficient to indicate lack of
mutagenicity, whereas a positive result is not considered sufficient to indicate
that the chemical represents a mutagenic hazard (i.e.,. it could be a false
positive). The micronucleus test in rodents is the most widely used as a follow-up
to positive in vitro mutagenicity results.
A study published by Benigni et al. (2010) showed
analyses and considerations relative to the role of the in vivo micronucleus
assay in the pre-screening of chemical carcinogenicity. The positive predictivity
(i.e., probability for a chemical with structure alert to be positive) by this
method is 0.33. As shown in Table 6, application of the Benigni-Bossa
method showed that all of the compounds had structural alerts for genotoxic
carcinogenicity. Those need to be verificated through experimental assay.
Michael acceptors resulted in a classification of class 2 for all compounds
except for purpurin 7 and rhodin g7 (class 1). Purpurin 7 and rhodin
g7 have structural alerts for Michael acceptors (Table
6) due to their α-β-unsaturated aldehyde or ketone. Michael type
addition can lead to the molecular initiating event of protein alteration which
results in the formation of covalent adduct at a soft electro (nucleo) philic
center without expulsion of a leaving group in the molecule. Electrophiles acting
in this manner are typically organic molecules that contain olefinic π-bonds
polarized by a neighboring electron-withdrawing substituent. Schultz
et al. (2007) have observed that α-β-unsaturated carbonyl
||Acetylenic-substituted derivatives are more reactive than
their corresponding olefinic-substituted analogues
||Vinyl-substituted derivatives are more reactive than their vinylene-substituted
||Methyl substitution on a vinyl C atom diminishes reactivity with methyl
substitution on the α-C atom which resulting in a larger reduction
||Vinyl-substituted derivatives are more reactive than aldehydes
||Having an additional unsaturated groups increases reactivity
Moreover, reactive toxicity (the irreversible interaction of a xenobiotic chemical
with endogenous molecules including specific sites on proteins) has been identified
as the major gap to predict key hazards such as skin sensitization accurately
(Schultz et al., 2007).
Toxtree can identify potential mechanism of toxic action for skin sensitization.
It was observed that all compounds have alerts for skin sensitization by Michael
type addition, except for BPhe a-OH (no skin sensitization alerts identified).
These results were different to those of Michael acceptors method. It is possible
because the structure alerts also include alerts for precursors of Michael reaction
Cytochrome P450-mediated drug metabolism resulted in positive for all site
of metabolism except for PPIX-1OH (negative in SMARTCyp.Rank3.sites) and for
PPIX-2OH (negative in SMARTCyp.Rank2.sites). SMARTCyp is an in silico
method that predicts the sites of metabolism for drug metabolites mediated by
cytochrome P450 3A4 isoform. The idea behind SMARTCyp is that activation energies
of CYPs reacting with ligand fragments computed by quantum chemical methods
are the best possible reference for the reactivity of a fragment. Those results
indicated that the compounds were likely easily metabolized.
TTC is a pragmatic risk assessment tool which is based on the principle of
establishing a human exposure threshold value for all chemicals, below which
there is a very low probability of an appreciable risk to human health. Kroes
TTC decision tree resulted in negligible risk for all compounds (Table
The correlation between similarity of structure and similarity of biological
response is the basic to making prediction on toxicological properties. Toxic
estimation software can reduce animal testing, time and cost. ECOSAR predict
the aquatic toxicity based on Structure-Activity Relationship (SAR) while Toxtree
predicts different type of toxicological hazard and modes of action by applying
decision tree approaches and SAR. Thus, both packages can be used for initial
toxicity assessments of designed or existing molecules. The ECOSAR prediction
showed that tetrapyrrolic macrocycles with more carboxylic acid groups or hydroxyl
groups showed lower toxic potency than those with fewer carboxylic acid groups
or hydroxyl groups. Bacteriochlorin has lower toxic potency than that of chlorin.
Based on the prediction by Toxtree, the studied compounds could be categorized
as negligible risk (by Kroes TTC) and not corrosive to skin. Most compounds
are likely easily metabolized. However, they were classified as persistent chemical
and they showed, at least, one positive alerts for micronucleus assay.
This research was supported in part by KK ITB Research Grant 2010 and Competitive
Research Grant 2010 from the Ministry of National Education, Indonesia.