Changeover from signalling to energy-provisioning lipids during transition from colostrum to mature milk in the giant panda (Ailuropoda melanoleuca)

Among the large placental mammals, ursids give birth to the most altricial neonates with the lowest neonatal:maternal body mass ratios. This is particularly exemplified by giant pandas. To examine whether there is compensation for the provision of developmentally important nutrients that other species groups may provide in utero, we examined changes in the lipids of colostrum and milk with time after birth in giant pandas. Lipids that are developmental signals or signal precursors, and those that are fundamental to nervous system construction, such as docosahexaenoic acid (DHA) and phosphatidylserines, appear early and then fall dramatically in concentration to a baseline at 20–30 days. The dynamics of lysophosphatidic acid and eicosanoids display similar patterns, but with progressive differences between mothers. Triglycerides occur at relatively low levels initially and increase in concentration until a plateau is reached at about 30 days. These patterns indicate an early provision of signalling lipids and their precursors, particularly lipids crucial to brain, retinal and central nervous system development, followed by a changeover to lipids for energy metabolism. Thus, in giant pandas, and possibly in all bears, lactation is adapted to provisioning a highly altricial neonate to a degree that suggests equivalence to an extension of gestation.


Contents
Panel S1. Detailed structural determination of lipids -details additional to Materials and Methods explained in main text.      Table S2. Reproductive histories and milk sampling times of giant pandas from whom samples were collected.

References for Supplementary Information.
Note -all the data obtained in this study are available from the corresponding author.

Panel 1. Detailed structural determination of lipids -details additional to Materials and Methods explained in main text
Following the procedures detailed in the main text, the pre-processed lipid-chromatography-mass spectrometry (LC-MS) data were searched by accurate mass (±3 ppm) against the Lipid Metabolites  Table S1, selection of the correct adduct forms in different ESI modes for accurate mass search of each lipid class can avoid confusion in putative annotation. Glycerolphospholipids with a single acyl tail showed slightly longer retention times than those with the same polar head but double acyl tails, which may be explained by their slight increase of hydrophilicity due to the presence of an exposed hydroxyl group in the glycerol backbone structure.
Further lipid structure elucidation was performed using collision-induced dissociation multiple stage mass spectrometry (MS n ) fragmentation in a data-dependent scan mode. Different lipid classes could be identified by their characteristic MS 2 fragmentation patterns representing the loss of the specific polar heads (Table S1). The fragment ions produced by losing a water molecule equivalent could be included with the phospholipids in ESI-positive mode if the phospholipids have a free hydroxyl group in their backbone structures. MS 3 fragmentation, especially in ESI-negative mode, provided more details in structural information of the acyl tails. An example of this is given in Figure S4, depicting the MS 2 product ion of a PS that was further fragmented. The resulting MS 3 fragment ions indicate the number of carbons of each acyl tail with the corresponding degree of unsaturation (18:0/22:6). In addition, the relative intensity of the fatty acid ions might imply the position of the acyl tail (sn-1 > sn-2). Consequently, as shown in Figure S3, the acyl tail of 18:0 was assigned at sn-1 position and the 22:6 was at sn-2. More examples of MS n spectra for the other lipid classes are shown in Figure S4.
The MS n spectra of polar lipids obtained in this study accorded with the published data produced for Thermo Scientific mass spectrometers (LTQ ion trap or LTQ-Orbitrap) by the manufacturer. Analysis of MS n spectra for each lipid class or subclass revealed that the abundant polar lipids in giant panda milk possessed acyl chains comprising even numbers of carbons from 14 to 26 with the number of double bonds varying from 0 to 6. However, acyl chains with odd numbers of carbons were also found on both mono and diacyl glycerolphospholipids but were present in low relative abundance.
Compared to a single phospholipid species, a single triacylglyceride lipid species generated more fragment ions through the loss of heterogeneous fatty acid tails in the population of molecules (M-RCOO+H), indicating the presence of a more extensive range of isomers in the neutral than in the polar lipids.
Another explanation for the apparently low abundance of glycerolipids may lie in the interpretation of the low MS signal for this group of molecules. Glycerolipids provide low-abundance MS signals because of their poor ESI performance such that the intensity of their MS signals did not provide quite the same clarity in changes in their relative concentrations in comparison with the polar lipids, despite being the major lipids in mammalian milks including that of giant panda.
In the data-dependent scan mode, the MS n fragmentation was only triggered on those lipids with the highest intensities in the MS 1 scans. Interpretation of the MS n spectra suggests putative constructions shown in Figure S4. Characterisation of the lipids that did not require additional MS n fragmentation was based on the elemental composition prediction and supplemented by the specific retention times corresponding to the lipid classes (see Table S1). Without considering the number of isomers, 403 species of lipid were putatively annotated in different classes and used for the statistical analysis given in the main text. More lipid species were detected for PC and PS than for other polar lipids.  Co., Tokyo, Japan) and described in reference [2]. This specialist replacer contains little or none of the lipids we find natural to giant panda colostrum (this study), and is also known to contain levels of lactose in excess of those occurring in natural panda milks after day 10 post-partum [1, 3]. Figure S4. Mass spectra of the main lipids identified in giant panda milk. The first frame is an example of serial mass spectrometry analysis used here in the identification phosphatidylserine, followed by MS, MS 2 280  300  320  340  360  380  400  420  440  460  480  500  520  540  560  580  600  620  640  660  680  700  720  740  760  780  800  820  840  860  880  SM (d18:1/16:0) The formula is consistent with a phytosphingolipid which has an additional hydroxyl group in comparison with most mammalian sphingolipids. The additional group may be acetylated since the ion at m/z 687.5437 results from the loss of acetic acid. The MS 3 spectrum for the ion at 687.5437 produces a fragment at m/z 616.4716 which must result from the loss of the trimethyl ammonium head group and an extra carbon. The most diagnostic fragment is at m/z 449.3163 which results from the loss of palmitoyl from the amine functionality of the sphingolipid.  Table S2. Reproductive histories and milk sampling times of giant pandas from whom samples were collected. All mothers and cubs were born in captivity. Giant panda international studbook numbers as given. Parturition dates in bold indicate the beginning of the lactation period from which samples were collected. See also ref. [1].