PHYSIOLOGICAL PH AND ACIDIC PHOSPHOLIPIDS CONTRIBUTE TO SUBSTRATE SPECIFICITY IN LIPIDATION OF ATG8

Yeast Atg8 and its mammalian homolog LC3 are ubiquitin-like proteins involved in autophagy, a primary pathway for degradation of cytosolic constituents in vacuoles/lysosomes. Whereas the lipid phosphatidylethanolamine (PE) was identified as the sole in vivo target of their conjugation reactions, in vitro studies showed that the same system can mediate the conjugation of these proteins with phosphatidylserine as efficiently as with PE. Here, we show that, in contrast to PE conjugation, the in vitro phosphatidylserine conjugation of Atg8 is markedly suppressed at physiological pH. Furthermore, the addition of acidic phospholipids to liposomes also results in the preferential formation of the Atg8-PE conjugate. We have successfully captured authentic thioester intermediates, allowing us to elucidate which step in the conjugation reaction is affected by these changes in pH and membrane lipid composition. We propose that these factors contribute to the selective formation of Atg8-PE in the cell.


EXPERIMENTAL PROCEDURES
Protein purification and liposome preparation-S. cerevisiae proteins Atg8 (truncated for the C-terminal arginine), Atg7, Atg3, and Atg4, were expressed in E. coli and purified as described previously (16). The Atg12-Atg5 conjugate was formed in E. coli and purified as reported previously (17).
Liposomes were also prepared as previously    (20). Therefore, we tried the NuPAGE system Immunoblotting analyses showed that the 85 and 55 kDa products both contained Atg8, and exclusively contained Atg7 and Atg3, respectively (Fig. 2B). Taken together, these results suggest that these products are thioester intermediates of the conjugation reaction, Atg8-Atg7 and Atg8-Atg3.

Atg8-PS formation is suppressed
We also analyzed reactions containing PE liposomes using this system (Fig. 3). We found that both of the 85 and 55 kDa products significantly decreased upon the completion of Atg8-PE formation, accompanied by a concomitant increase in free Atg3 (Fig. 3 (Fig. 4).
Consistent with the previous report, Atg8 was efficiently conjugated with PE as the PI concentration increased, saturating at about 10 mol% (Fig. 4A, left). In contrast, we found that PS conjugation was not at all stimulated by the addition of PI (Fig. 4A, right). This Atg8-Atg7 or the Atg8-Atg3 intermediates (Supplemental Fig. S1). We then examined the binding of the proteins to liposomes, by cosedimentation experiments (Fig. 5) (Fig. 5A, lanes 1-3). Similarly, no significant cosedimentation was observed when the proteins were mixed with PE liposome that did not contain PI (Fig. 5A,   lanes 4-6). However, the addition of PI to this liposome significantly increased the levels of Atg8-Atg7 and Atg8-Atg3 cosedimented with the liposome, especially for the latter (Fig. 5A,   lanes 7-9). In addition, the experiments without ATP clearly showed that Atg8, Atg7 and Atg3 alone did not bind to the PE liposome, irrespective of the presence of PI (Fig. 5A, lanes 10-15). These results suggest that acidic phospholipids specifically promote the recruitment of the thioester intermediates to the membrane, thereby facilitating the conjugation reaction of Atg8. We also showed that the binding of Atg8-Atg7 to PE liposomes containing PI depends on Atg3 (Fig. 5A, lanes 16-21; see Discussion).
The binding of the Atg proteins to PS liposomes was also examined. The thioester intermediates did bind to PS liposomes as substantially as PE liposomes containing PI, and this binding was not enhanced by PI addition (Fig. 5A, lanes 22-27). This insensitivity to PI is consistent with what we observed in the conjugation reaction (Fig. 4).
These results with the PS liposome could be However, neither PE nor PS conjugation occurred (Fig. 6, upper panels). We have reported that the Atg12-Atg5 conjugate is indispensable for Atg8-PE production in vivo (21) and that recombinant Atg12-Atg5 indeed stimulates it in vitro (17). Therefore, Atg12-Atg5 was included in the reaction under these conditions. As reported previously, the formation of both Atg8-PE and Atg8-PS was accelerated by Atg12-Atg5; however, the acceleration was more prominent for PE conjugation under these conditions (Fig. 6, lower panels). These results suggest that preferential formation of  Fig. S2). Our results also highlighted a unique aspect of Atg3: it is an enzyme whose activity is lowered under physiological conditions. In addition, observation of conjugate formation with the thioester intermediates clearly showed that the conjugase activity of Atg3 is rate-limiting (Fig. 3); Atg12-Atg5 drastically enhances this activity through a direct interaction with Atg3 (17). It is known that most autophagy-related proteins, including  pH. The pH of each reaction mixture was determined by a pH meter. After incubation for the indicated time periods, the reaction was stopped by mixing with SDS sample buffer, followed by urea-SDS-PAGE separation. The intensities of the CBB-stained protein bands were measured using the Scion Image Software and used to calculate conjugate formation efficiencies (%), in which the intensities of lipidated Atg8 were divided by intensities of total Atg8 protein.

DISCUSSION
D, The conjugate formation efficiencies after 120-min incubation shown in C are also presented, with the horizontal axis showing the pH of the reaction mixtures.  reactions performed at pH 7.0 as described in Fig. 1C were analyzed using the NuPAGE gel system as described in Fig 2A. The same samples were also analyzed using urea-SDS-PAGE to assess lipidation of Atg8. The double asterisk shows bands containing Atg8 that do not appear in the presence of ME. mol% PE or PS and the indicated and remaining concentrations of PI and PC, respectively, which were analyzed as described in Fig 1C. B and C, The PE and PS conjugation of Atg8 was similarly examined in the presence of various acidic phospholipids.