A chemogenetic platform for controlling plasma membrane signaling and synthetic signal oscillation

TL;DR: In this article, a bioorthogonal iK6DHFR/mDcTMP-based self-localizing ligand-induced protein translocation (SLIPT) approach was proposed.

Abstract: Chemogenetic methods that enable the rapid translocation of specific signaling proteins in living cells using small molecules are powerful tools for manipulating and interrogating intracellular signaling networks. However, existing techniques rely on chemically induced dimerization of two protein components and have certain limitations, such as a lack of reversibility, bioorthogonality, and usability. Here, by expanding our self-localizing ligand-induced protein translocation (SLIPT) approach, we have developed a versatile chemogenetic system for plasma membrane (PM)-targeted protein translocation. In this system, a novel engineered Escherichia coli dihydrofolate reductase in which a hexalysine (K6) sequence is inserted in a loop region (iK6DHFR) is used as a universal protein tag for PM-targeted SLIPT. Proteins of interest that are fused to the iK6DHFR tag can be specifically recruited from the cytoplasm to the PM within minutes by addition of a myristoyl-O_SCPLOWDC_SCPLOW-Cys-tethered trimethoprim ligand (mDcTMP). We demonstrated the broad applicability and robustness of this engineered protein-synthetic ligand pair as a tool for the conditional activation of various types of signaling molecules, including protein and lipid kinases, small GTPases, heterotrimeric G proteins, and second messengers. In combination with a competitor ligand and a culture-medium flow chamber, we further demonstrated the application of the system for chemically manipulating protein localization in a reversible and repeatable manner to generate synthetic signal oscillations in living cells. The present bioorthogonal iK6DHFR/mDcTMP-based SLIPT system affords rapid, reversible, and repeatable control of the PM recruitment of target proteins, offering a versatile and easy-to-use chemogenetic platform for chemical and synthetic biology applications.

Summary

INTRODUCTION

• Cellular functions are regulated by signaling networks involving proteins, lipids, and other second messenger molecules, and cells precisely coordinate biochemical signaling events in space and time in a fast and reversible manner. [1, 2].
• The ability to manipulate the (which was not certified by peer review) is the author/funder.
• Chemogenetic protein translocation systems are particularly attractive because small-molecule control is easy to perform in vitro, ex vivo, and in vivo to confer rapid temporal modulation.
• To address these issues, several new CID systems have been rationally constructed using two ligand-binding protein tags, including those based on Escherichia coli dihydrofolate reductase and the FKBP F36V mutant, [15, 16] SNAP-tag and (wild-type) FKBP, [17] and and HaloTag [16, [18] [19] [20] pairs.

Development of a loop-engineered eDHFR tag

• The inner leaflet of the PM serves as a platform for intracellular signaling networks, and almost all signaling pathways that determine cell physiology, such as growth, differentiation, phagocytosis, and migration, are initiated and modulated at the PM.
• The myr D C motif undergoes S-palmitoylation of the Cys residue by (which was not certified by peer review) is the author/funder.
• The copyright holder for this preprint this version posted March 16, 2021.
• This may explain the impaired PM-specificity of the EGFP-K6-eDHFR translocation observed previously.
• These results demonstrated that both loop-engineered eDHFR 69K6 and eDHFR 145K6 functioned as PM-specific SLIPT tags that can be fused to either the N-or C-termini of proteins.

Mechanistic characterization of the m D cTMP/eDHFR iK6 SLIPT system

• Next, the authors investigated which membrane lipids contributed to the PM-targeting specificity of the iK6 DHFR tag.
• The inner PM is highly anionic compared with other organelle membranes because of the presence of the negatively charged phospholipids, such as PS, [27, 28] and phosphatidylinositol 4-phosphate (PI4P) and phosphatidylinositol-4,5bisphosphate [PI(4,5)P2] [28] [29] [30] .
• On the basis of their previous work, [22] the authors performed depletion experiments of these anionic lipid species.
• As a result, EGFP-iK6 DHFR dissociated from the PM and was relocalized to the endomembrane .
• This fact raises the possibility that the m D cTMP/ iK6 DHFR complex, which contains a myristoyl lipid anchor and the K6 motif, might be able to localize to the PM without the S-palmitoylation of m D cTMP.

Chemogenetic activation of the Raf/ERK signaling pathway

• The authors applied the m D cTMP/ iK6 DHFR SLIPT system to manipulate intracellular signaling.
• The copyright holder for this preprint this version posted March 16, 2021.
• Such ERK activation was not observed when the authors performed the same experiment in the presence of PD184352, a MEK inhibitor , or using EGFP-iK6 DHFR lacking the cRaf protein .
• Furthermore, the above experiment showed that the loop-engineered iK6 DHFR functions as a PM-specific SLIPT tag even when it is inserted between two protein domains (in this case, EGFP and cRaf).

Chemogenetic control of heterotrimeric G proteins and second messenger signaling

• In particular, there are several Gα isoforms, including Gαq, Gαs, Gα12/13, and Gαi, and more than one of these isoforms are often activated simultaneously upon GPCR stimulation. [35, 36].
• Thus, it has been difficult to study the contribution of each isoform to cellular functions.
• In a pioneering work, Putyrski and Schultz have developed conditional Gα activation systems using the rapamycin CID technique. [37].
• Chemogenetic methods applicable to control heterotrimeric G proteins are still limited.
• Therefore, the applicability of the m D cTMP/ iK6 DHFR SLIPT system to control Gαq protein and Ca 2+ signaling was demonstrated.

Chemical induction of synthetic signal oscillations

• Recent single-cell analyses have revealed that several signaling proteins, including p53, ERK, and NF-kB, exhibited oscillatory dynamics to govern cell fate decisions. [3, 4, [41] [42] [43].
• To date, such oscillatory control of protein activity has been achieved only by optogenetics, [41, [44] [45] [46] and chemogenetic approaches for this purpose have not yet been well established.
• The copyright holder for this preprint this version posted March 16, 2021.
• First, the cells were treated with m D cTMP (10 µM) to induce the PM translocation of iK6 DHFR-EGFP .
• The cells were first treated with m D cTMP, followed by repeated TMP addition and washing cycles.

CONCLUSION

• The iK6 DHFR tag can be fused to either the N-terminus or C-terminus of proteins, and even between two proteins or protein domains, without the loss of its PM specificity, offering a high degree of flexibility in the construction of iK6 DHFR fusion proteins.
• The authors demonstrated the applicability of the m D cTMP/ iK6 DHFR SLIPT system to control a wide range of signaling proteins with different sizes (ca. 24 to 73 kDa) and physiological functions, including cRaf, Gαq, Gαs, RasGEF, PI3K, and Tiam1, verifying the versatility and general utility of the system.
• Therefore, the SLIPT-based synthetic signal oscillation system will be useful to reproduce the frequency of ERK activity pulses observed in cultured cells for biological research.
• It should also be noted that the signal duration time and frequency of signal activation can further be modulated by the timing of the TMP addition and washing steps.
• In the schematic illustration, the crystal structure of eDHFR complexed with methotrexate (PDB: 1RG7 [26] ) was used and the cartoon model was created with UCSF ChimeraX [47] (methotrexate is not shown).

Figures

Figure 3. Chemogenetic activation of Gαq signaling and Ca2+ oscillations. (a) Schematic 2 illustration of the experimental setup. (b) Representative time-lapse confocal fluorescence images 3 of a HeLa cell coexpressing mNG-iK6DHFR-Gαq and R-GECO. Images were taken before and 4 after the addition of mDcTMP (5 µM). Scale bar, 10 μm. For the time-lapse movie, see Movie S3. 5 (c) The time course of mNG-iK6DHFR-Gαq translocation and Ca2+ spikes observed in the cell is 6 shown in panel b. The normalized fluorescence intensities of mNG-iK6DHFR-Gαq (bottom) and 7 R-GECO (top) in the cytoplasm were plotted as a function of time. (d) Heatmaps depicting 8 mDcTMP-induced Ca2+ oscillations for 17 randomly selected cells coexpressing mNG-iK6DHFR-9 Gαq and R-GECO. 10 11 12

Figure 2. Chemogenetic activation of the Raf/ERK pathway. (a) Schematic illustration of the 2 experimental setup. (b) Confocal fluorescence images of HeLa cells coexpressing EGFP-3 iK6DHFR-cRaf and ERK-KTR-mKO were taken before (left) and 60 min after the addition of 4 mDcTMP (10 µM) (right). Scale bar, 20 μm. For the time-lapse movie, see Movie S2. (c) Time 5 course of EGFP- iK6DHFR-cRaf translocation and ERK activation. To evaluate EGFP- iK6DHFR-6 cRaf translocation (bottom), the normalized fluorescence intensity of EGFP-iK6DHFR-cRaf in the 7 cytoplasm was plotted as a function of time. To evaluate ERK activity (top), the normalized ratios 8 of the cytoplasmic fluorescence intensity to the nuclear fluorescence intensity (C/N ratios) of 9 ERK-KTR-mKO were plotted as a function of time. Data from control experiments using EGFP-10 iK6DHFR-cRaf in the presence of the MEK inhibitor PD184352 (Figure S7a) or EGFP-iK6DHFR 11 (lacking the cRaf protein) (Figure S7b) were plotted on the same graph. Data are presented as 12 the mean ± SD (n = 5 cells). 13 14

Figure 4. Chemical induction of synthetic ERK signal oscillation. (a) Schematic illustration of 2 the experimental setup. The washing procedure was carried out using a culture-medium flow 3 system as shown in Figure S13a. (b) Representative time-lapse confocal fluorescence images of 4 a HeLa cell coexpressing cRaf-mNG-iK6DHFR and mCh-ERK. Images were taken at the time 5 points indicated by the asterisks shown in panel c. mDcTMP and TMP were added at 6 concentrations of 10 and 50 µM, respectively. During the washing step (the blue bar in panel c), 7 fresh medium continuously flowed at a rate of 1 mL/min. Scale bar, 20 μm. For the time-lapse 8 movie, see Movie S8. (c) Time course of cRaf-mNG-iK6DHFR translocation and ERK activity. To 9 evaluate cRaf-mNG-iK6DHFR translocation (top), the normalized fluorescence intensity of cRaf-10 mNG-iK6DHFR in the cytoplasm was plotted as a function of time. To evaluate ERK activity 11 (bottom), the normalized ratios of the nuclear fluorescence intensity to the cytoplasmic 12 fluorescence intensity (N/C ratios) of mCh-ERK were plotted as a function of time. Data are 13 presented as the mean ± SD (n = 12 cells). 14 15

Figure 1. Development of a universal protein tag for PM-specific SLIPT. (a) Chemical structure 2 of mDcTMP. (b) Design of loop-engineered eDHFR constructs based on the topology of PM-3 anchored eDHFR. In the schematic illustration, the crystal structure of eDHFR complexed with 4 methotrexate (PDB: 1RG7[26]) was used and the cartoon model was created with UCSF 5 ChimeraX[47] (methotrexate is not shown). The structure is depicted in a manner such that eDHFR 6 is anchored on the putative PM by mDcTMP. The N- and C-termini are shown in magenta and 7 cyan, respectively. The K6 motif was inserted between D69 and D70 or between A145 and Q146 8 of wild-type eDHFR to generate eDHFR69K6 and eDHFR145K6, respectively. (c–f) mDcTMP-9 induced translocation of eDHFR69K6-EGFP (c), EGFP-eDHFR69K6 (d), eDHFR145K6-EGFP (e), 10 and EGFP-eDHFR145K6 (f). Confocal fluorescence images of HeLa cells expressing the indicated 11 construct were taken before (left) and 30 min after the addition of mDcTMP (10 µM) (right). Scale 12 bars, 10 µm. For the time-lapse movie of EGFP-eDHFR69K6 translocation, see Movie S1. (g) Time 13 course of PM translocation (quantification of data shown in c–f and Figure S2). The normalized 14 fluorescence intensity in the cytoplasm was plotted as a function of time. Data are presented as 15 the mean ± SD (n = 5 cells). 16 17 18

Full text

1
A chemogenetic platform for controlling plasma membrane signaling
1
and synthetic signal oscillation
2
3
Yuka Hatano,
1,#
Sachio Suzuki,
2,#
Akinobu Nakamura,
3,#
Tatsuyuki Yoshii,
1,4,#
Kyoko
4
Atsuta-Tsunoda,
1
Kazuhiro Aoki,
3,5,6
& Shinya Tsukiji
1,2
*
5
6
1
Department of Life Science and Applied Chemistry, Nagoya Institute of Technology, Gokiso-cho,
7
Showa-ku, Nagoya 466-8555, Japan
8
2
Department of Nanopharmaceutical Sciences, Nagoya Institute of Technology, Gokiso-cho, Showa-
9
ku, Nagoya 466-8555, Japan
10
3
Quantitative Biology Research Group, Exploratory Research Center on Life and Living Systems
11
(ExCELLS), National Institutes of Natural Sciences, 5-1 Higashiyama, Myodaiji-cho, Okazaki, Aichi
12
444-8787, Japan
13
4
PRESTO, Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 332-
14
0012, Japan
15
5
Division of Quantitative Biology, National Institute for Basic Biology, National Institutes of Natural
16
Sciences, 5-1 Higashiyama, Myodaiji-cho, Okazaki, Aichi 444-8787, Japan
17
6
Department of Basic Biology, Faculty of Life Science, SOKENDAI (The Graduate University for
18
Advanced Studies), 5-1 Higashiyama, Myodaiji-cho, Okazaki, Aichi 444-8787, Japan
19
20
#
These authors contributed equally to this work.
21
22
ORCID
23
Tatsuyuki Yoshii: 0000-0002-3465-4219
24
Kazuhiro Aoki: 0000-0001-7263-1555
25
Shinya Tsukiji: 0000-0002-1402-5773
26
27
*Correspondence should be addressed to S.T. (email: stsukiji@nitech.ac.jp)
28
29
30
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted March 16, 2021. ; https://doi.org/10.1101/2021.03.16.435568doi: bioRxiv preprint

2
ABSTRACT
1
Chemogenetic methods that enable the rapid translocation of specific signaling
2
proteins in living cells using small molecules are powerful tools for manipulating and
3
interrogating intracellular signaling networks. However, existing techniques rely on
4
chemically induced dimerization of two protein components and have certain
5
limitations, such as a lack of reversibility, bioorthogonality, and usability. Here, by
6
expanding our self-localizing ligand-induced protein translocation (SLIPT)
7
approach, we have developed a versatile chemogenetic system for plasma membrane
8
(PM)-targeted protein translocation. In this system, a novel engineered Escherichia
9
coli dihydrofolate reductase in which a hexalysine (K6) sequence is inserted in a loop
10
region (
iK6
DHFR) is used as a universal protein tag for PM-targeted SLIPT. Proteins
11
of interest that are fused to the
iK6
DHFR tag can be specifically recruited from the
12
cytoplasm to the PM within minutes by addition of a myristoyl-D-Cys-tethered
13
trimethoprim ligand (m
D
cTMP). We demonstrated the broad applicability and
14
robustness of this engineered protein–synthetic ligand pair as a tool for the
15
conditional activation of various types of signaling molecules, including protein and
16
lipid kinases, small GTPases, heterotrimeric G proteins, and second messengers. In
17
combination with a competitor ligand and a culture-medium flow chamber, we
18
further demonstrated the application of the system for chemically manipulating
19
protein localization in a reversible and repeatable manner to generate synthetic
20
signal oscillations in living cells. The present bioorthogonal
iK6
DHFR/m
D
cTMP-
21
based SLIPT system affords rapid, reversible, and repeatable control of the PM
22
recruitment of target proteins, offering a versatile and easy-to-use chemogenetic
23
platform for chemical and synthetic biology applications.
24
25
26
INTRODUCTION
27
Cellular functions are regulated by signaling networks involving proteins, lipids, and
28
other second messenger molecules, and cells precisely coordinate biochemical signaling
29
events in space and time in a fast and reversible manner.
[1,2]
Recently, single-cell imaging
30
experiments have also revealed that several signaling proteins, such as the extracellular
31
signal-regulated kinase (ERK), show oscillatory activation dynamics and the frequency
32
of the activation pulse determines the cellular responses.
[3,4]
The ability to manipulate the
33
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted March 16, 2021. ; https://doi.org/10.1101/2021.03.16.435568doi: bioRxiv preprint

3
activity of signaling molecules and pathways in a rapid, reversible, and repeatable manner
1
would therefore be useful for understanding the relationships between signaling input
2
dynamics and cellular outputs, and ultimately for engineering synthetic cellular behaviors.
3
Methods that enable the rapid localization control or translocation of specific proteins
4
in living cells offer a powerful means for modulating signaling activities. Consequently,
5
several chemogenetic
[5–7]
and optogenetic
[8,9]
approaches have been developed as tools for
6
controlling protein translocation. Chemogenetic protein translocation systems are
7
particularly attractive because small-molecule control is easy to perform in vitro, ex vivo,
8
and in vivo to confer rapid temporal modulation. In addition, unlike most optogenetic
9
approaches, chemogenetic methods can be used together with fluorescent reporters of
10
various colors, enabling experiments that combine user-defined perturbation with
11
simultaneous visualization of the subcellular dynamics of multiple signaling activities in
12
the same living cell (“experimental multiplexing”).
[10]
13
The most widely used method for chemogenetic protein translocation control relies
14
on a chemically induced dimerization (CID) system using the small-molecule rapamycin,
15
which induces the heterodimerization of the FK506-binding protein (FKBP) and the
16
FKBP-rapamycin-binding (FRB) domain.
[11–13]
The rapamycin CID system has been
17
proven to be a versatile tool to control the translocation of a wide range of signaling
18
proteins with fast kinetics. However, the rapamycin CID is essentially irreversible.
[14]
19
Rapamycin also binds and interferes with the functions of endogenous FKBP and the
20
mammalian target of rapamycin (mTOR), leading to undesired biological effects. To
21
address these issues, several new CID systems have been rationally constructed using two
22
ligand-binding protein tags, including those based on Escherichia coli dihydrofolate
23
reductase (eDHFR) and the FKBP F36V mutant,
[15,16]
SNAP-tag and (wild-type)
24
FKBP,
[17]
and eDHFR and HaloTag
[16,18–20]
pairs. In these systems, chimeric molecules
25
consisting of two small-molecule ligands for each protein tag are used as chemical
26
dimerizers. These systems allow reversible protein translocation by the combined use of
27
the chemical dimerizer and a free competitor ligand. However, the protein tag-based CID
28
systems require the careful adjustment of the chemical dimerizer concentrations because
29
excess chemical dimerizer will bind to both the protein components, competitively
30
interfering with the protein heterodimerization. In addition, the chemical dimerizer for
31
the SNAP-tag/FKBP CID system lacks biorthogonality because it also binds to
32
endogenous FKBP.
[17]
Moreover, as a common limitation shared by all CID tools,
33
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted March 16, 2021. ; https://doi.org/10.1101/2021.03.16.435568doi: bioRxiv preprint

4
dimerization-dependent methods require the coexpression of two proteins with
1
appropriate expression levels and stoichiometry to control the target protein, which is
2
challenging in practice. Therefore, the establishment of chemogenetic protein
3
translocation methods that overcome these limitations is highly desirable.
4
Here we present a versatile, single protein component, chemogenetic protein
5
translocation system that can be used for manipulating diverse signaling processes at the
6
plasma membrane (PM) based on our self-localizing ligand-induced protein translocation
7
(SLIPT) strategy.
[21–24]
In this system, a novel engineered eDHFR in which a hexalysine
8
(K6) sequence is inserted in a loop region is used as a universal protein tag for PM-
9
targeted SLIPT. Proteins of interest that are fused to the loop-engineered eDHFR can be
10
rapidly and specifically recruited to the PM upon addition of myristoyl-D-Cys-tethered
11
trimethoprim (m
D
cTMP), a previously developed self-localizing ligand (SL). We show
12
the broad applicability and robustness of this bioorthogonal, loop-engineered
13
eDHFR/m
D
cTMP-based SLIPT system for the conditional activation of various types of
14
signaling molecules, including protein and lipid kinases, small GTPases, heterotrimeric
15
G proteins, and second messengers, such as Ca
2+
and cAMP. We further demonstrated
16
that the combined use of the present tool and a culture-medium flow system can enable
17
the chemical control of protein localization in a reversible and repeatable manner to
18
produce synthetic signal oscillations in living cells, expanding the repertoire of
19
chemogenetic tools for manipulating intracellular signaling dynamics.
20
21
22
RESULTS AND DISCUSSION
23
Development of a loop-engineered eDHFR tag
24
The inner leaflet of the PM serves as a platform for intracellular signaling networks, and
25
almost all signaling pathways that determine cell physiology, such as growth,
26
differentiation, phagocytosis, and migration, are initiated and modulated at the PM.
27
Therefore, chemogenetic methods capable of recruiting signaling proteins to the PM are
28
particularly important. Using a bioorthogonal small-molecule trimethoprim (TMP) and
29
eDHFR pair,
[25]
we have previously developed a PM-targeted SLIPT system (Figure
30
S1a).
[22,23]
In this system, a TMP ligand is conjugated via a flexible linker to a designer
31
myristoyl-D-Cys (myr
D
C) lipopeptide motif to form m
D
cTMP, which is used as an SL
32
(Figure 1a).
[23]
The myr
D
C motif undergoes S-palmitoylation of the Cys residue by
33
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted March 16, 2021. ; https://doi.org/10.1101/2021.03.16.435568doi: bioRxiv preprint

5
palmitoyl acyltransferases in cells, which localizes the motif to the PM and the Golgi. In
1
conjunction, an eDHFR variant containing a hexalysine (K6) sequence at the N-terminus
2
(K6-eDHFR) is used as a protein tag for fusion to proteins of interest.
[22]
Non-engineered
3
(wild-type) eDHFR is recruited, not only to the PM, but also undesirably to the Golgi by
4
m
D
cTMP (Figure S1b and Figure S2a). However, K6-eDHFR is translocated
5
preferentially to the PM by m
D
cTMP because the PM localization of the m
D
cTMP/K6-
6
eDHFR complex is enhanced by electrostatic interactions between the cationic K6 tag
7
and the negatively charged phospholipid phosphatidylserine (PS) present at the inner
8
PM.
[22]
Consequently, by fusing a protein of interest to the C-terminus of the K6-eDHFR
9
tag, the resulting protein can be recruited specifically to the PM upon addition of
10
m
D
cTMP (Figure S1c and Figure S2b). This m
D
cTMP/K6-eDHFR-based SLIPT system
11
has been used as a tool for conditional PM-specific protein translocation in live cultured
12
cells and a nematode (Caenorhabditis elegans).
[22,23]
However, when the K6-eDHFR tag
13
was fused to the C-terminus of a protein, such as enhanced green fluorescent protein
14
(EGFP) (EGFP-K6-eDHFR), the protein was translocated to the Golgi in addition to the
15
PM (Figure S1d and Figure S2c), implying that the PM specificity of K6-eDHFR was
16
insufficient. This result indicated that K6-eDHFR can be used as a PM-specific tag only
17
when it is fused to the N-terminus of protein targets, limiting the application of the system.
18
In this work, we aimed to engineer a universal SLIPT tag that can be fused to either
19
the N-terminus or C-terminus of proteins or even between two proteins or protein domains,
20
while retaining the PM-targeting specificity. To this end, we reconsidered the molecular
21
orientation of the m
D
cTMP/eDHFR complex when it is anchored to the inner surface of
22
the PM. As shown in Figure 1b, in the crystal structure of eDHFR (PDB: 1RG7
[26]
), the
23
N- and C-termini of eDHFR are located at the opposite sides of the ligand-binding pocket.
24
In the previous K6-eDHFR tag, the K6 motif was linked to the N-terminus of eDHFR via
25
a 16 amino acid linker such that the K6 sequence could access the PM.
[22]
Therefore, it is
26
reasonable to consider that when a protein is fused to the N-terminus of K6-eDHFR, the
27
access of the K6 tag to the PM will be sterically hindered. This may explain the impaired
28
PM-specificity of the EGFP-K6-eDHFR translocation observed previously. To address
29
this issue, we sought to introduce the K6 sequence to a different site of the eDHFR. On
30
the basis of the model shown in Figure 1b, we focused on two loop regions, Ser63–Val72
31
and Ser135–Ser150. These loops are positioned near the ligand-binding pocket and
32
closely face the inner surface of the PM when the m
D
cTMP/eDHFR complex is in the
33
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted March 16, 2021. ; https://doi.org/10.1101/2021.03.16.435568doi: bioRxiv preprint