Variability of Phenylalanine side chain conformations facilitates promiscuity of Fatty acid binding in Cockroach milk proteins

The pacific beetle cockroach, Diploptera punctata, is a viviparous cockroach that produces a milk-like substance to support the growing embryo with a brood sac. The structure of the in vivo grown crystals present in the gut of the embryo showed that the milk-derived crystals are heterogenous and are made of three proteins (called Lili-Mips). Multiple fatty acids could be modeled into the active site, and we hypothesized that each of the three isoforms of the protein bound to a different fatty acid. We previously reported that the recombinantly expressed Lili-Mip2 has a structure similar to the structure of the protein determined from in vivo crystals, and this single isoform also binds to several fatty acids. In this study, we aimed to probe the specificity and affinity of fatty acid binding and test the stability of different isoforms. We show that all the isoforms can bind to different fatty acids with very similar affinities, and the local abundance of a fatty acid determined bound fatty acid ratios. Lili-Mips’ thermostability is pH dependent, where stability is highest at acidic pH and declines as the pH increases to physiological levels near 7.0. The measurement of the pH in the gut lumen and the gut cells suggests that the pH in the gut is acidic and the pH inside the gut cells is closer to neutral pH. We propose that the protein has evolved to be highly stable in the acidic gut lumen and, when absorbed inside the gut cells, becomes less stable to enable the breakdown of the glycosylated lipo-protein complex to provide essential metabolites for survival and development of the embryo. The different orientations of Phe-98 and Phe-100 control the binding pocket volume and allow the binding of different chain-length fatty acids to bind with similar affinities.


73
The binding and delivery of hydrophobic ligands to different places are critical parts of 74 lipocalin function. Apolipoprotein D is a classic example of a lipid transporter [15]. The ligand 75 binding and release mechanism of different lipocalins has been investigated. Human retinoic acid 76 binding protein and tear lipocalin are known to release ligands at an acidic pH [16,17]. 77 Lipocalins show nano-to-micromolar binding affinities to various hydrophobic ligands [12,18]. 78 Many lipocalins bind to multiple ligands [18]. This is, in turn, determined by the ligand binding 79 site and the four loops at the entrance of the calyx [13]. The large barrel that houses the ligand 80 makes lipocalins amenable for protein engineering [8,13]. These four loops that connect 81 antiparallel beta sheets can be designed to target the lipocalin to bind to any molecule of choice -82 haptens, fatty acids, peptides, and so on. These modified lipocalins are called Anticalins [19]. In 83 tear lipocalin and Lili-Mip 2, Zinc is involved in crystal packing [10,13]. Zinc binds at the entrance 84 of the calyx and is responsible for the preferential binding of Lili-Mip 2 to palmitoleic acid rather 85 than oleic acid [10]. 86 The remarkable structural similarity of lipocalins warrants the conservation of some unique 87 sequential features. For instance, the overall RMS deviation values of Lili-Mip (PDB: 4NYQ) with 88 respect to Human tear Lipocalin(PDB: 1XKI) is 4.5 Å for 103 C-alpha atoms [8]. The disulfide 89 bonds and tryptophan at the base of the β-barrel are the most conserved [8,13,20,21]. The 90 tryptophan at the N terminus (W20 in Lili-Mip) covers the base of the 10 Å wide and 15 Å deep 91 ligand binding calyx [8,10]. Tryptophan at this position is known to be important for the structure 92 and preventing oxidation of retinol in β-lactoglobulin. It stabilizes the protein and plays a role in 93 binding [12,21]. The presence of this single tryptophan makes Lili-Mip, like many other members 94 of its family, amenable to binding studies through the measurement of intrinsic tryptophan 95 fluorescence [12,18].

96
Structures of Lili-Mip and Lili-Mip2 provide a molecular basis for fatty acid binding [8,10]. The 97 Glutamate(E38) at position 38 forms a kink in the ligand binding site in Lili-Mip2 (Figure 1). It is 98 held in position by histidine (H115) and tyrosine (Y40). One could hypothesize that it favors the 99 binding of Lili-Mip2 to omega-6 unsaturated fatty acids [10]. In tear lipocalin, charged residues 100 Glutamate 34, Histidine 84, and Lysine 114 are thought to be playing a role in ligand interaction 101 [13]. In Lili-Mip, the carboxylic acid moiety of the fatty acid is exposed to the surface, whereas 102 the acyl chain is buried [8,10]. This has been observed for another lipocalin-like retinol-binding 103 protein, RBP4 [22], that binds laurate, palmitate, oleate, and linoleate. Glutamate 38 and 104 Phenylalanine at 98 and 100 are conserved between these structures [8,10,22,23]. However, the 105 carboxylate group faces inwards and interacts with tyrosine and arginine in the proteins of the fatty 106 acid binding proteins (FABPs) family [24].

107
Previous work in our lab has shown that Lili-Mip2 is a thermophilic protein that does not 108 denature even at 100 °C [10]. Another lipocalin, β-lactoglobulin, is thermodynamically stable at 109 low pH. There is a difference of 16.6 °C in melting temperature (T m ) of β-lactoglobulin as the pH 110 changes from 7.5 to 1.0 [25]. The stability of lipocalins at low pH could be attributed to 111 glycosylation. This study aims to elucidate the stability and binding properties of recombinantly 112 expressed Lili-Mip1, 2, and 3 and compare these factors to the properties of the Lili-Mips purified 113 from the cockroach midgut. The results provide detailed information to engineer ligand binding 114 by mutating residues in the active site. Lili-Mip1, Lili-Mip2, and Lili-Mip3 were synthesized by GeneArt and cloned in pYES2-

119
CT vectors with a C-terminal Histidine tag [8,10]. Further, Lili-Mip1, 2, and 3 were sub-cloned 120 into the pYES2 vector without an affinity tag. All constructs were codon optimized for yeast 121 expression and had an N-terminal secretion signal (Ost1 or α-factor) for protein secretion into the  at 600 nm was measured, and the culture was diluted with new YPD media to an OD of 0.2/ml.

133
This was again grown till the OD reached 0.6-0.7, induced with 2% galactose, and incubated for 134 24 hours at 30 °C, 220 rpm. The supernatant was clarified by centrifugation at 4000 g for 10 135 minutes. The clarified media was concentrated using Sartorius Vivaflow200, with a 10kDa 136 molecular weight cut off filter to a final volume of 500ml. The media was buffer exchanged into 137 50mM sodium acetate at pH 5.0 and 10mM NaCl while concentrating the media. The buffer 7 138 exchanged media was purified using cation exchange chromatography on the SP FF column 139 (Cytiva). The Lili-Mip protein eluted between 100mM to 200mM of NaCl (on a gradient that 140 extended to 1.0M NaCl). The protein fractions were concentrated using a 10 kDa filter in an 141 Amicon spin filtration system. The concentrated protein was purified using the size exclusion       Lipocalins are known to be thermostable, such a large change in T m has not been observed so far.    (14,16,18) and degrees of unsaturation (0, 1, 2) were assayed. All tested fatty acids bind with 310 micromolar range affinities to the different Lili-Mips (Table 1). This data suggests that the length 311 of the acyl chain and the unsaturation does seem to significantly affect the affinity of binding. This 312 reiterates that the fatty acids the embryo receives almost exclusively depend on the fatty acid 313 composition of the milk-secreting cells.

315
We considered how an increase in pH could affect the binding affinity. Binding studies 316 with Lili-Mip1 against palmitic acid showed that Lili-Mip binds to the ligand with similar binding 317 affinities between pH 4.7 and 8.2. This also reaffirms that a change in thermal stability with pH is   that at a pH of 4.7 the fatty acid stays in the ligand binding pocket for a longer time than it does at 364 the pH inside the cells ( Figure 5). Interestingly as the pH increases to 9.1, probability density 365 plots begin to suggest that binding is tighter. This result, along with the lower stability of the 366 protein at physiological pH supports the idea that once absorbed into the cells, the protein can be 367 broken down, and the fatty acid, amino acids, and sugars can be used. The simulations show that 368 myristic acid binds farther away from W20 than palmitoleic acid. This is expected due to the 369 shorter length of the acyl chain. Similar to that observed for other lipocalins, the simulations also 370 suggest that the residues that change conformation the most are at the entrance of the binding