Metabolomic Coverage of Chemical-Group-Submetabolome Analysis: Group Classification and Four-Channel Chemical Isotope Labeling LC-MS

Chemical
isotope labeling (CIL) liquid chromatography mass spectrometry
(LC-MS) is a powerful technique for in-depth metabolome analysis with
high quantification accuracy. Unlike conventional LC-MS, it analyzes
chemical-group-based submetabolomes and uses the combined results
to represent the whole metabolome. Due to analysis time and cost constraint,
not all submetabolomes can be profiled and thus knowledge of chemical
group classification is important in guiding submetabolome selection.
Herein we report a study of determining the distribution of functional
groups of compounds in a database and then examine how well we can
experimentally analyze the major chemical groups in two representative
samples (i.e., human plasma and yeast). We developed a computer algorithm
to classify chemical structures according to their functional groups.
After removing lipids which are targeted molecules in lipidomic analysis,
inorganic species and other molecules that are unique to drug, food,
plant, and environmental origins, five groups (i.e., amine, phenol,
hydroxyl, carboxyl, and carbonyl) are found to be the dominant classes.
In the databases of MCID (2683 filtered metabolites), HMDB (5506),
KEGG (11598), YMDB (1107), and ECMDB (1462), 94.7%, 85.7%, 86.4%,
85.7%, and 95.8% of the filtered metabolites belong to one or more
of the five groups, respectively. These groups can be analyzed in
four-channel CIL LC-MS where hydroxyls (H), amines and phenols (A),
carboxyls (C), and carbonyls or ketones/aldehydes (K) are separately
profiled as individual channels using dansyl and DmPA labeling reagents.
A total of 7431 peak pairs were detected with 6109 unique-mass pairs
from plasma, while 5629 pairs with 4955 unique-mass pairs were detected
in yeast. Compared to group distributions of database compounds, hydroxyl-containing
metabolites were severely underdetected, which might indicate that
the current method is less than optimal for analyzing this group of
metabolites. As a result, the overall experimental coverage is likely
significantly lower than the database-derived coverage. In short,
this study has shown that high metabolome coverage is theoretically
attainable by analyzing only the H, A, C, and K submetabolomes and
the group classification information should be helpful in guiding
future analytical method development and choices of submetabolomes
to be analyzed.

化学同位素标记（Chemical Isotope Labeling，CIL）液相色谱-质谱联用（Liquid Chromatography Mass Spectrometry，LC-MS）是一种可实现深度代谢组分析且定量精度优异的强有力技术。与传统LC-MS不同，该技术通过分析基于化学基团的亚代谢组，并整合各亚代谢组的分析结果以表征完整代谢组。受限于分析时间与实验成本，无法对全部亚代谢组进行表征，因此掌握化学基团分类信息对指导亚代谢组的筛选至关重要。本研究旨在明确数据库中化合物的官能团分布情况，并考察可通过实验表征人类血浆与酵母两种代表性样本中主要化学基团的效果。我们开发了一款计算机算法，可根据官能团对化学结构进行分类。在剔除脂质组学分析的靶标分子脂质、无机物以及药物、食品、植物与环境来源的特有分子后，我们发现胺类、酚类、羟基类、羧基类与羰基类这五类官能团为优势类别。在MCID（2683个过滤后代谢物）、HMDB（5506个）、KEGG（11598个）、YMDB（1107个）与ECMDB（1462个）数据库中，经过滤后的代谢物分别有94.7%、85.7%、86.4%、85.7%与95.8%属于上述五类官能团中的一种或多种。上述五类官能团可通过四通道CIL LC-MS进行分析：借助丹磺酰与DmPA标记试剂，可分别在四个通道中表征羟基类（H）、胺类与酚类（A）、羧基类（C）以及羰基类/酮类/醛类（K）代谢物。血浆样本中共检测到7431个峰对，对应6109个唯一质量对；而酵母样本中则检测到5629个峰对，对应4955个唯一质量对。与数据库化合物的官能团分布相比，含羟基代谢物的检出率严重偏低，这表明当前方法在分析该类代谢物时并非最优方案。因此，整体实验覆盖度大概率显著低于基于数据库预测的覆盖度。综上，本研究表明，仅需对H、A、C、K四类亚代谢组进行分析，即可在理论上实现较高的代谢组覆盖度；而官能团分类信息将有助于指导未来分析方法的开发以及待分析亚代谢组的选择。