A viral ADP-ribosyltransferase attaches RNA chains to host proteins

Abstract

The mechanisms by which viruses hijack the genetic machinery of the cells they infect are of current interest. When bacteriophage T4 infects Escherichia coli, it uses three different adenosine diphosphate (ADP)-ribosyltransferases (ARTs) to reprogram the transcriptional and translational apparatus of the host by ADP-ribosylation using nicotinamide adenine dinucleotide (NAD) as a substrate1,2. NAD has previously been identified as a 5′ modification of cellular RNAs3–5. Here we report that the T4 ART ModB accepts not only NAD but also NAD-capped RNA (NAD–RNA) as a substrate and attaches entire RNA chains to acceptor proteins in an ‘RNAylation’ reaction. ModB specifically RNAylates the ribosomal proteins rS1 and rL2 at defined Arg residues, and selected E. coli and T4 phage RNAs are linked to rS1 in vivo. T4 phages that express an inactive mutant of ModB have a decreased burst size and slowed lysis of E. coli. Our findings reveal a distinct biological role for NAD–RNA, namely the activation of the RNA for enzymatic transfer to proteins. The attachment of specific RNAs to ribosomal proteins might provide a strategy for the phage to modulate the host’s translation machinery. This work reveals a direct connection between RNA modification and post-translational protein modification. ARTs have important roles far beyond viral infections⁶, so RNAylation may have far-reaching implications.

Full text

1054 | Nature | Vol 620 | 31 August 2023
Article
A viral ADP-ribosyltransferase attaches RNA
chains to host proteins
Maik Wolfram-Schauerte1,2,7, Nadiia Pozhydaieva1, Julia Grawenhoff2, Luisa M. Welp3,4,
Ivan Silbern3,4, Alexander Wulf3, Franziska A. Billau2, Timo Glatter1, Henning Urlaub3,4,5,
Andres Jäschke2 ✉ & Katharina Höfer1,2,6,7 ✉
The mechanisms by which viruses hijack the genetic machinery of the cells they infect
are of current interest. When bacteriophage T4 infects Escherichiacoli, it uses three
dierent adenosine diphosphate (ADP)-ribosyltransferases (ARTs) to reprogram the
transcriptional and translational apparatus of the host by ADP-ribosylation using
nicotinamide adenine dinucleotide (NAD) as a substrate1,2. NAD has previously been
identied as a 5 modication of cellular RNAs3–5. Here we report that the T4 ART
ModB accepts not only NAD but also NAD-capped RNA (NAD–RNA) as a substrate and
attaches entire RNA chains to acceptor proteins in an ‘RNAylation’ reaction. ModB
specically RNAylates the ribosomal proteins rS1 and rL2 at dened Arg residues, and
selected E.coli and T4 phage RNAs are linked to rS1 invivo. T4 phages that express an
inactive mutant of ModB have a decreased burst size and slowed lysis of E.coli. Our
ndings reveal a distinct biological role for NAD–RNA, namely the activation of the
RNA for enzymatic transfer to proteins. The attachment of specic RNAs to ribosomal
proteins might provide a strategy for the phage to modulate the host’s translation
machinery. This work reveals a direct connection between RNA modication and
post-translational protein modication. ARTs have important roles far beyond viral
infections6, so RNAylation may have far-reaching implications.
ARTs catalyse the transfer of one or multiple ADP–ribose (ADPr) units
from NAD to target proteins
7
. Bacterial and archaeal ARTs act as tox-
ins and are involved in host defence or drug-resistance mechanisms8,
whereas eukaryotic ARTs have roles in distinct processes ranging from
DNA damage repair to macrophage activation and stress response
9
.
Viruses use ARTs as weapons to reprogram the host’s gene-expression
system6. Mechanistically, a nucleophilic residue of the target protein
(usually Arg, Glu, Asp, Ser or Cys) attacks the glycosidic carbon atom
in the nicotinamide riboside moiety of NAD, forming a covalent bond
as N-, O- or S-glycoside
7
(Fig.1a). As the adenosine moiety of NAD is
not involved in this reaction, we speculated that elongation of the
adenosine to long RNA chains (by means of regular 5–3 phospho-
diester bonds) might be tolerated by ARTs, potentially leading to the
formation of covalent RNA–protein conjugates (Fig.1b). RNAs that
have a 5-NAD cap have previously been found in bacteria (including
E.coli
3,10,11
), archaea
12,13
and eukaryotes
5,14–19
, with NAD–RNA concen-
trations ranging from 1.9 to 7.4fmolµg−1 RNA16. This modification
was observed in different types of RNA, including mRNA and small
regulatory RNA (sRNA)
20
. However, little is known about the biological
functions of this RNA cap21.
The infection cycle of bacteriophage T4 relies on the sequen-
tial expression of early, middle and late phage genes that are tran-
scribed by E.coli RNA polymerase (RNAP)22. For the specific temporal
reprogramming of the E.coli transcriptional and translational appara-
tus, the T4 phage uses 3 ARTs that modify more than 30 host proteins.
Upon infection, one of these ARTs, Alt, is injected into the bacterium
with the phage DNA and immediately ADP-ribosylates E.coli RNAP
at different residues, which is thought to result in the preferential
transcription of phage genes from early promoters23,24. Two early
phage genes encode the ARTs ModA25 and ModB1,26. ModA completes
the ADP-ribosylation of RNAP, whereas ModB is thought to modify
the host protein rS1 (refs. 1,26). However, it is still not known how
ADP-ribosylation changes the properties of the target proteins, or
whether other proteins are also modified during T4 infection.
ModB catalyses RNAylation invitro
To test our idea that ARTs may accept NAD–RNAs as substrates, we
purified Alt, ModA and ModB. We incubated them with either a syn-
thetic, site-specific 32P-labelled 5-NAD–RNA 8-base oligonucleotide
(8-mer) or a 3-fluorophore-labelled 5-NAD–RNA 10-mer to test
for either self-modification or the modification of target proteins.
Whereas both Alt and ModA showed only a small amount of target
RNAylation (Extended Data Fig.1a), ModB rapidly RNAylated its known
ADP-ribosylation target protein, rS1, without detectable self-RNAylation
(Fig.2a and Extended Data Fig.1b). By contrast, ModB-mediated
https://doi.org/10.1038/s41586-023-06429-2
Received: 4 June 2021
Accepted: 12 July 2023
Published online: 16 August 2023
Open access
Check for updates
1Max Planck Institute for Terrestrial Microbiology, Marburg, Germany. 2Institute of Pharmacy and Molecular Biotechnology, Heidelberg University, Heidelberg, Germany. 3Bioanalytical Mass
Spectrometry, Max Planck Institute for Multidisciplinary Sciences, Göttingen, Germany. 4Department of Clinical Chemistry, University Medical Center, Göttingen, Germany. 5Cluster of
Excellence “Multiscale Bioimaging: from Molecular Machines to Networks of Excitable Cells” (MBExC), Georg-August-University, Göttingen, Germany. 6Center for Synthetic Microbiology
(SYNMIKRO), Philipps-Universität Marburg, Marburg, Germany. 7These authors contributed equally: Maik Wolfram-Schauerte, Katharina Höfer. ✉e-mail: jaeschke@uni-hd.de; Katharina.Hoefer@
synmikro.mpi-marburg.mpg.de
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Nature | Vol 620 | 31 August 2023 | 1055
ADP-ribosylation in the presence of 32P-NAD resulted in the modifica-
tion of both proteins (ModB and rS1) with similar intensity (Fig.2b and
Extended Data Fig.1c). No signal was evident when either ModB or
rS1 was missing, or when a 5-
32
P-monophosphate–RNA (5-
32
P–RNA)
of the same sequence was used as a substrate for ModB (Extended
Data Fig.1d). Moreover, a mutated active site (R73A,G74A) of ModB
also prevented the RNAylation of rS1 (ref. 1) (Extended Data Fig.2a,b).
This mutation similarly affected both the ADP-ribosylation and the
RNAylation activity of ModB.
RNAylation follows an ADP-ribosylation-like
mechanism
ModB-catalysed RNAylation of rS1 was strongly inhibited by the ART
inhibitor 3-methoxybenzamide (3-MB)
27
, which is thought to mimic
the nicotinamide moiety (Extended Data Fig.2c), confirming an
ADP-ribosylation-like mechanism. Moreover, RNAylated rS1 proteins
that carry a
32
P-labelled ADPr moiety were treated with the ribonuclease
(RNase) T1 to determine whether the RNA and the protein are covalently
linked (Extended Data Fig.2d). This treatment would remove the
32
P
label if the RNA were non-covalently bound to rS1 or covalently linked
at any position other than the 5-terminal positions. The 32P-rS1 signal
did not disappear after treatment with T1, but it disappeared entirely
after treatment with trypsin, which breaks down rS1 (Extended Data
Fig.2e). Collectively, these data indicate that the RNA is covalently
linked to rS1 at its 5 end, as shown in Fig.1b.
RNAylation assays using short linear or hairpin-forming NAD–RNAs
(Fig.2c and Extended Data Fig.3a) revealed that ModB has a preference
for unstructured NAD–RNAs as a substrate, although it also accepted
longer, biologically relevant NAD-capped RNAs as substrates, such as a
NAD-capped Qβ RNA fragment of around 100 nucleotides
28
(Fig.2d and
Extended Data Fig.3b). RNAylation with NAD-capped 100-nucleotide
RNA caused the modified rS1 protein to migrate with an apparent mass
of 100kDa (Fig.2e). Treatment of the RNAylated protein with nuclease
P1, which hydrolyses 3–5 phosphodiester bonds but does not attack
the pyrophosphate bond of the 5-ADPr, reversed this shift, and the
32P-labelled product migrated in a similar way to unmodified rS1 or ADPr–
rS1 (Fig.2e), confirming the proposed nature of the covalent linkage.
To exclude the possibility that ModB removes only the nicotinamide
moiety from the NAD–RNA by hydrolysis, thereby generating a highly
reactive ribosyl moiety that could (through its masked aldehyde group)
spontaneously react with nucleophiles in its vicinity29, we prepared
ADPr-modified RNA and tested it as a substrate for ModB. No modifica-
tion could be detected (Extended Data Fig.3c), providing no support
for spontaneous RNAylation.
To exclude the degradation of RNA during RNAylation, we supplied
ModB with an NAD–RNA 10-mer that carried a fluorescent dye (Cy5) at
the 3 terminus (Extended Data Figs.2a and3a). The time-course analysis
of the RNAylation indicates that intact oligonucleotide chains were
attached to rS1 for a variety of NAD-capped RNAs (Extended Data Fig.3a).
ModB modifies Arg residues in rS1
To identify the amino acid residues in protein rS1 to which RNA chains
are covalently linked during RNAylation, we used tools developed to
analyse protein ADP-ribosylation.
N
N
N
N
NH2
O
OHOH
O
P
O
O
O–
P
O
O–
O
O
OHOH
N
O
H2N
N
N
N
N
NH2
O
OHOH
O
P
O
O
O–
P
O
O–
O
O
OHO
O–
O
ART H
N
HNH
NH
H
–O
O
ART
N
N
N
N
NH2
O
OHOH
O
P
O
O
O–
P
O
O–
O
O
OHOH
NH
H
N
NH
N
N
N
N
NH2
O
OHO
O
P
O
O
O–
P
O
O–
O
O
OHOH
N
O
H2N
N
N
N
N
NH2
O
OHO
O
P
O
O
O–
P
O
O–
O
O
OHO
O–
O
ART H
N
HNH
NH
H
–O
O
ART
Acceptor
protein
N
N
N
N
NH2
O
OHO
O
P
O
O
O–
P
O
O–
O
O
OHOH
NH
H
N
NH
ab
NAD–RNANAD
RNAylated protein ADP-ribosylated protein
Acceptor
protein
Acceptor
protein
Acceptor
protein
+
+
+
Fig. 1 | Mechanisms of ADP-ribosylation and proposed RNAylation. a, The
mechanis m of ADP-ribosy lation for Arg. I nitially, the N-glycosi dic bond betw een
the ribos e and nicotinami de is destabili zed by a Glu residue of an A RT. This
leads to the for mation of an oxocar benium ion of ADP r, with nicotinamide a s
the leavin g group. This elec trophilic ion is at tacked by a nucle ophilic Arg
residue of th e acceptor prote in after Glu-med iated proton abs traction, le ading
to the formati on of an N-glycosi dic bond50. b, Our proposed RNAylation-reaction
mechanis m. In a similar way to AD P-ribosylatio n in the presence o f NAD, we
propose th at ARTs might use NAD –RNA to catalys e an RNAylation re action,
thereby covale ntly attach ing an RNA to an acc eptor protein.Red, n icotinamide
riboside o f NAD and NAD -RNA; blue,cat alytic residue s of the ART; purple,
nucleophilic Arg residue of the acceptor protein.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1056 | Nature | Vol 620 | 31 August 2023
Article
The radioactive signal of
32
P-RNAylated protein rS1 and
32
P–ADP-
ribosylated rS1 did not change after treatment with HgCl2 (which cleaves
S-glycosides at Cys residues), NH2OH (which hydrolyses O-glycosides
at Asp and Glu) (Extended Data Fig.4a) or recombinant enzyme ARH3
(which hydrolyses O-ADPr glycosides specifically at Ser residues)
(Extended Data Fig.4b), although it was efficiently removed by treat-
ment with human ARH1 (Fig.3a,b and Extended Data Fig.4c,d). These
findings indicate that the main products of ModB-catalysed RNAylation
are linked as N-glycosides by Arg residues (Extended Data Fig.4c,d).
To establish that ModB-mediated ADP-ribosylation or RNAyla-
tion also occurs at Arg residues invivo, we isolated genomically
His-tagged rS1 from non-infected or T4-infected E.coli. Analysis
using liquid chromatography with tandem mass spectrometry
(LC–MS/MS) confirmed that there was specific modification of Arg
residues in rS1 with ADPr. These ADPr modifications were present
only in the T4-infected sample (Extended Data Table1 and Supple-
mentary Table1). R139 was identified as a modifiedresidue, as con-
firmed by site-directed mutagenesisto Lys or Ala; rS1(R139K) and
rS1(R139A) mutants were expressed in T4-infected E.coli, purified
and analysed, revealing that these mutations prevent modifica-
tion at those positions (Extended Data Table2 and Supplementary
Table2).
rS1–
8-mer
012510 20 30 60
Time (min) 012510 20 30 60
a
d
b
Coomassie Radioactivity
012510 20 30 60
Time (min) 012510 20 30 60
Coomassie Radioactivity
c
015103060 120 Ref 015103060 120 015103060 120
kDa
70
100
180/130
rS1
70
100
180/130
*
Time (min)
Coomassie Radioactivity Merge
kDa
rS1
Coomassie Radioactivity Merge
Re
fR
ef
015103060 120 Ref 015103060 120 015103060 120
Re
fR
ef
006060
+P1 –P1
RNAylated rS1
e
70
100
180/130
kDa
Time (min) 006060
+P1 –P1
RNAylated rS1
006060
+P1 –P1
RNAylated rS1
*
*
*
0306090 120
0
0.25
0.50
0.75
1.00
Incubation time (min)
Relative conversion
5′ blunt
2-nt 5′ overhang
3′ overhang
Linear
70
100
kDa
rS1–
ADPr 70
100
kDa
N
N
N
N
NH2
O
OHO
O
P
O
O
O
P
O
O
O
O
OHOH
rS1
NH
H
N
NH
RNAylated (100-nt) protein rS1
Nuclease P1
NMPs
N
N
N
N
NH2
O
OHOH
O
P
O
O
O
P
O
O
O
O
OHOH
rS1
NH
H
N
NH
ADP-ribosylated protein rS1
Time (min)
rS1 + 32P–NAD–8mer + ModB rS1 + 32P–NAD–8mer + ModB
rS1 + 32P–NAD + ModB rS1 + 32P–NAD + ModB
rS1 + 5′-NAD–100nt–RNA + ModB rS1 + 5′-NAD–100nt–RNA + ModB rS1 + 5′-NAD–100nt–RNA + ModB
rS1 + 5′-P–100nt–RNA + ModB rS1 + 5′-P–100nt–RNA + ModB rS1 + 5′-P–100nt–RNA + ModB
*
*
*
*
Fig. 2 | Post-tran slational p rotein modi ficati on of rS1 by ModB invitr o.
a, Time cours e of the RNAylation o f rS1 by ModB (n=3). SDS–polyac rylamide
gel elect rophoresis (SDS–PAGE) gels ar e shown for rS1+32P–NAD –8-mer+ModB.
Complete ge ls and a reaction s chematic are show n in Extende d Data Fig.1b.
b, Time cours e of the ADP-ribo sylation of rS1 by Mo dB (n=3), showing
rS1+32P–NAD+Mo dB. Complete ge ls and a reactio n schematic are s hown in
Extende d Data Fig.1c. rS1 R NAylation (a) and ADP-ribosylation (b) are indicate d
by the acquisi tion of a radioac tive signal overlapp ing with the Co omassie stai n.
c, The role of R NA secondar y structure on R NAylation reac tion. Four dif ferent
3 Cy5-label led NAD-c apped RNAs we re tested, inclu ding a linear 10-mer
NAD-c apped RNA a nd three struc tured NAD- capped RN As with a 3 overhang , a
dinucleot ide 5 overhang or a blunt e nd. SDS–PAGE analysis is sh own in Extend ed
Data Fig.3a . Relative conversio n refers to the inten sity of the RN Aylated rS1 band
relative to the m aximal RNAylat ion intensit y observed am ong all four tests .
Data poin ts represen t mean±s.d. valu es based on qua ntifica tion of fluo rescence
Cy5 signals (n=3 biologically in dependent re plicates). d,Invitro kine tics of the
RNAylation o f rS1 by ModB using 5-NAD–100 -nucleotide (100 -nt) RNA as the
substrat e (top), analysed by SDS–PAGE. The pink a sterisk indic ates shifte d
RNAylated rS1; the blue asterisk indicates ADP-ribosylated rS1.ADP-ribosylated
rS1 serve s as a reference (Ref ). The mass of 10 0nucleotide s is around 30kDa;
RNAylated r S1 has a mass of arou nd 100kDa (70kDa fr om rS1, 30kDa fro m RNA).
5-P–100nt RN A was used as a neg ative control (b ottom, n=2).The two ban ds
above the 100 kD a band are denote d 180/130. e, Th e nuclease P1 bre aks down
RNAylated pro tein rS1. The c ovalently att ached 100-nu cleotide-lon g RNA
results in a s hift of the RNAylat ed protein rS1 (which h as a mass of around
100kDa) in SDS–PAGE. Nu clease P1 cle aves the phospho diester bond, r esulting
in degrada tion of the att ached RNA i nto mononucle otides. Nuc lease P1 conver ts
RNAylated r S1 into ADP-ribo sylated rS1 (mass o f around 70kDa), which can b e
seen by the pre sence of a down shifted prote in band in the SDS–PAGE gel (n=1).
Red, ribose moiety of RNAylated/ADP-ribosylated protein; NMPs, nucleoside
monophosphates; radioactivity symbol indicates site of 32P-label; pacman
symboliz es nuclease P 1. The pink an d blue asterisk s are the same as in d.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Nature | Vol 620 | 31 August 2023 | 1057
01510203060
Time (min)
012510 20 30 60
Coomassie Radioactivity Merge
01510203060
Time (min) Time (min)
ARH1
ARH1
D55A, D56A
01510203060
Time (min)
012510 20 30 60
Coomassie Radioactivity Merge
01510203060
ARH1
Time (min) Time (min)
a
b
m/z
m/z
100
50
Relative intensity
Relative intensity
MS1
cd
e
- 70
- 100
- 70
- 100
- 70
- 100
- 70
- 100
ARH1
D55A, D56A
Ribose–H2O
N
N
O
NH2
N
N
N
N
NH2
C′
A′
N
N
N
N
NH2
O
O
O
OH
O
P
HO
O
ON
N
NH2
O
OHOPHO
O
OH
P
O
OP
O
OH OH
O
OH OH
HN
NH
NH
ADPr
ADPr–A′
Loss of CH2N2
H+
H+
1,116.1440
z = 3
1,116.4781
z = 3
1,115.8094
z = 3
1,116.8121
z = 3
1,117.1469
z = 3
1,117.4823
z = 3 1,117.8137
z = 3
1,115 1,116 1,117 1,118 1,119
200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 2,200 2,400
MI:C′
MI:A′
MI:CMP
MI:AMP
MI:ribose-H2O+HPO3
MI:ADPribose+HPO3
MI:ADPribose
b2
a2
b3 VDVR#2+
PVRD#2+
a5
b5
b6
a6
a7
b7
b8 b9
DVRPVR#
RPVRDT#
[M+2H]#2+
[M+2H]2+
[M+2H]*-H2O2+
[M+2H]*-H2ONH3
2+
[M+2H]#-H2O2+
b20#2+
DVRPVRDTLH*-H2O+HPO3
b14#2+
y1
y2
y3
y4
y5
y6
y7 y8
y20*-H2O2+
y20*-2H2O2+
y19*+HPO3
3+
y19*-H2O2+
y19*-2H2O2+
y13*-2H2O2+
y112+ y19#2+
y20#2+
y16#2+
y13#2+
y18#2+
y15#2+
y19***+HPO3
3+
y19*-H2O3+
y19**+HPO3
2+ y14#2+
y15#2+
y13#2+ y19#2+
b2 b5 b6 b7 b8 b9
a2 a5 a6 a7
b14 b20
##
y20 y19 y18 y16 y15 y14 y13 y11 y8 y7 y6 y5 y4 y3 y2 y1
###### #
**
*
**
**
AFLPGSL VDVRPVRDTLHL EGK
#
*
*
**
***
kDa
kD
a
b3
Fig. 3 | Identification of RNAylation sites of rS1. a,b, Specific removal of
ADP-ribo sylation and RN Aylation by ARH1 (n=3). Sch ematics of the re action
are shown in Ex tended Data F ig.4c,d. Enzy me kinetic s of ARH1 in the pre sence
of ADP-ribosylated (a) or RNAylated (b) prote in rS1 were analyse d by SDS–
PAGE. Mutation o f the cataly tically impor tant residue s D55 and D56 abo lished
the removal of A DP-ribosylat ion and RNAylatio n. c-e, Tandem MS-based
identif ication of R NAylated rS1 pept ide. c, The MS/ MS fragmen t ion spectr um
(spectru m ID: 23723) of RN Aylated rS1 pepti de AFLPGSLVDVRP VRDTLHLEGK
carry ing ADPr plus c ytidine mono phosphate and a 3 ph osphate group. T he
spectr um shows marker ions (M I)of adenine (A) and cy tosine (C), adenosin e
monophosphate (AMP), cytidine monophosphate (CMP), ribose–H2O and
ADPr. The prec ursor ion ([M+2H ]2+) and fragment i ons y13–y16, y18–y20, b14
and b20 show a spe cific loss o f mass of 42.0217 98 Da (#), which can be explaine d
by the loss of CH 2N2 at the modif ied Arg31. Pr ecursor ions, y1 3, y19 and y20 are
shifted by t he mass of ribo se–H2O (*). The spec trum also shows pre cursor ions
and y19 being shif ted by ADPr wi th (**) and without (* **) the loss of ad enine.
Blue, MI; red, pre cursor ions,int ernal fragm ent ions, b-ty pe fragmen t; green,
y-type fr agment ions . d, Isotopic pea k pattern of the p recursor ion as d etected
in the MS prec ursor ion scan for t he MS/MS spe ctrum shown i n c. e, Sequence
and RNA add uct represe ntation of the R NAylated pepti de shown in c and d,
including an notations of u nshifted fr agment ions an d fragment io ns showing
arginine lo ss (#), as well as ribose–H 2O (*), ADPr (**) and ADPr–adenine (* **). The
fragme ntation produ cts of the ADP r+CMP+3-phosphate adduc t observed i n
the MS/M S spectrum s hown in c are indica ted in the struc ture by light blue
(mass loss) and dark b lue (mass adduct s) lines.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1058 | Nature | Vol 620 | 31 August 2023
Article
LC–MS/MS analysis verifies RNAylation
The LC–MS/MS analysis above did not show unambiguously that the
modification of rS1 was derived from RNAylated or ADP-ribosylated
rS1. We therefore optimized LC–MS/MS to detect the covalent attach-
ment of RNA to rS1. For this analysis, invitro RNAylated, truncated
rS1 protein was subjected to an RNase A/T1 and tryptic digest. The
obtained mixture was directly subjected to LC–MS/MS analysis, and MS
data were evaluated using the RNPXL software tool30, on the assump-
tion that the RNAylated rS1 peptide still has a trinucleotide (ADPr–
cytidine) attached. The LC–MS/MS analysis this time showed the cova-
lent attachment of a trinucleotide (ADPr–cytidine) to an rS1 peptide
encompassing amino acid positions 129–150. Strikingly, the precursor
mass ([M+3H]
3+
with a mass-to-charge ratio (m/z)=1,115.81, expected
molecular mass=3,344.41Da) plus the gas-phase b- and y-type
fragmentation pattern, which shows the characteristic neutral loss of
CH2N2 (derived from a modified Arg31) or ribose, ADPr or ADPr-A adducts,
revealed that the RNA is attached by an N-glycosidic bond to R139 and/or
R142 (Fig.3c–e, Extended Data Fig.5 and Supplementary Table3). We
could not unambiguously assign the modified Arg because of the low
intensity of the respective fragment ions and the occurrence of mixed
spectra containing ion fragments of the same peptide species modified
at different sites (Fig.3c–e).
rS1 is RNAylated and ADP-ribosylated invivo
To distinguish quantitatively between ADP-ribosylation and RNAylation
invivo, we used immunoblotting with an antibody-like ADPr-binding
reagent (pan-ADPr) that specifically recognizes ADP-ribosylated pro-
teins but detects RNAylated proteins only after treatment with nuclease
Modication level (%)
RNAylated protein
Nuclease P1
ADP-ribosylated protein
NMPs
Pan-ADPr
1
70
55
40
25
15
180
130
100
kDa
(1) rS1–ADPr
(2) rS1–ADPr + nuclease P1
(3) rS1–RNA
(4) rS1–RNA + nuclease P1
ModB–ADPr
rS1–ADPr
b
(1) rS1 + T4
(2) rS1 + T4 + ARH1
(3) rS1 + T4 + nuclease P1
(4) Reference rS1–ADPr
Pan-ADPr
Pan-ADPr
d
ModB
35
15
25
kDa
40
55
70
erS1
Coomassie Radioactivity Merge
c
ADP-ribosylationRNAylation
acpP
gadY
rob
mcaS
glmY
oxyS
a-gt
rnaC
ipIII
T4 E. coli
sRNA mRNA
12345 6
rS1 architecture
0
1
2
12345 6
rS1 domain
Relative RNAylation
acpP
gcvB
malQ
gadY
aspA rob
ipIII
–5
0
5
log2-transformed fold change
Genome
T4 phage
E. coli
RNAI spike-in
–5 0510
log2-transformed mean expression
20
40
60
80
0
ADPr–rS1RNA–rS1
70
100
kDa
a
N
N
N
N
NH2
O
OHO
O
P
O
O
O–
P
O
O–
O
O
OHOH
NH
H
N
NH
N
N
N
N
NH2
O
OHOH
O
P
O
O
O–
P
O
O–
O
O
OHOH
NH
H
N
NH
234Lane
12 34
rS1–ADPr
RNase E NudC* BSA rS1 RNase E NudC* BSA rS1 RNase E NudC* BSA
+–+–+–+–+–+–+–+–+–+–+–+–
Fig. 4 | Invivo characterization of ADP-ribosylation and RNAylation.
a, Quantif ication of the R NAylation of rS1 u sing a nuclease P 1 digest and
western bl ot analysis.Gree n circle represen ts the protein. b, Quantif ication
of rS1 RNAylat ion invivo based on bi ological tri plicates (n=3). Data are s hown
as mean (g rey bar)and individua l data points . Complete blot s and intensit y
normaliz ation are shown in E xtended Dat a Fig.6b. c, Identif ication of R NA
substrat es of ModB using R NAylomeSeq . The MA plot show s data for one of
three biolo gical replica tes (n=3). Further deta ils are given in Ext ended Data
Fig.6c,d. d, Quant ificatio n of the RNAylation o f rS1. Modif ication of rS1
domains 1–6 (n=2 biol ogically inde pendent repli cates; black lin es show the
mea n). e, SDS–PAGE analysis of the RNAyla tion of protein rS1 , RNase E, inac tive
NudC muta nt (NudC*: V1 57A, E174A, E177A, E178A) and b ovine serum alb umin
(BSA) by ModB (n=2 bi ologically ind ependent re plicates).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Nature | Vol 620 | 31 August 2023 | 1059
P1 (Fig.4a and Extended Data Fig.6a). rS1 was expressed in non-infected
or T4-infected E.coli, affinity-purified and its ADP-ribosylation was ana-
lysed with pan-ADPr. We found extensive ADP-ribosylation of rS1 only in
the T4-infected sample. After treatment with nuclease P1, the pan-ADPr
signal intensity of the rS1 band increased (Fig.4b and Extended Data
Fig.6b), indicating RNAylation of rS1. Thus rS1 was found to be both
ADP-ribosylated and RNAylated invivo, with RNAylation accounting for
around 30% of the modifications. It remained unclear, however, whether
the two modifications are mutually exclusive or can occur simultane-
ously in the same molecule at different sites. Moreover, the signal for
ADPr disappeared after ARH1 treatment, further confirming the nature
of the RNA–protein linkage (Fig.4b and Extended Data Fig.6b). We found
that the ADP-ribosylation and RNAylation of rS1 occur in parallel invivo.
ModB RNAylates proteins with selected RNAs
To identify the RNAs linked to rS1 by ModB during infection by the
T4 phage, we developed an RNAylomeSeq approach (Extended
Data Fig.6c) in which genomically His-tagged rS1 was isolated from
T4-infected E.coli and captured on Ni-NTA beads. In a similar way to NAD
captureSeq
32
, RNA was reverse-transcribed ‘on-bead’ and the result-
ing cDNA was amplified by PCR and analysed using next-generation
sequencing.
We applied this workflow to E.coli treated with wild-type (WT) T4
phage. As a negative control, we used CRISPR–Cas9 technology to
generate a T4 phage that expressed the catalytically inactive mutant
ModB(R73A,G74A) (ref. 33). We compared the abundance of reads
mapped to individual RNA species and identified specific E.coli and T4
phage RNAs enriched in WT T4 phage samples (Fig.4c, Extended Data
Fig.6d,e, Supplementary Table4 and Supplementary Fig.3). Several
of the E.coli transcripts (mRNAs and sRNAs) have been reported to be
5-NAD-capped in E.coli3,34, including RNAs of the genes acpP, glmY, mcaS,
oxyS, aspA and rob, which makes them suitable substrates for ModB. We
also identified phage transcripts, such as ipIII (internal head protein III),
that were enriched in our datasets (Fig.4c, Extended Data Fig.6d,e and
Supplementary Table4). The enriched RNAs do not share any common
features apart from adenosine (+1A) at the transcription start site, which
is crucial for the biosynthesis of NAD-capped RNAs invivo35.
ModB RNAylates OB-fold proteins
To understand how ModB identifies its target proteins, we ana-
lysed the structural features of known target proteins. rS1 contains
oligonucleotide-binding (OB)-fold domains28. One structural vari-
ant of OB folds is the S1 domain, which is present in rS1 in six copies
that vary in sequence (Extended Data Fig.7a). RNAylated R139 and
R142 are located in domain2 of rS1. We speculated that the S1 domain
might be important for substrate recognition by ModB. To charac-
terize the specificity of ModB for different S1 domains, we cloned,
expressed and purified each S1 domain of rS1 (D1–D6) and tested them
in an RNAylation assay (Fig.4d and Extended Data Fig.7b). In agree-
ment with the mass spectrometry (MS) data (Extended Data Table1
and Supplementary Table1), we detected strong RNAylation signals
for rS1 D2 and D6, whereas rS1 D1, D3, D4 and D5 were modified to a
lesser extent. Multiple sequence alignment of rS1 D2 and D6, and the S1
domain of E.coli PNPase, revealed that these S1 domains share an Arg
residue as part of the loop that connects strands 3 and 4 of the β-barrel
36
(Extended Data Fig.7c). This loop is packed on the top of the β-barrel
and might therefore be accessible to ModB. For rS1 D2, the residues
R139 and R142 are the sites of RNAylation identified by MS (Fig.3e–g
and Supplementary Tables1–3). Mutation analysis confirmed that the
RNAylation level of D2 is significantly reduced if R139 is replaced by Ala
or Lys (Extended Data Fig.8a,b). E.coli RNaseE also has an S1 domain
in its active site with an Arg in the loop between strands 3 and 4. In the
RNAylation invitro assays, RNaseE was modified by ModB, whereas
control proteins without the S1 domain (such as BSA and the NudC
inactive mutant) were not. These data suggest that OB folds such as S1
domains with an embedded Arg are RNAylation target motifs (Fig.4e).
rL2 is a target for RNAylation by ModB
To discover additional RNAylation target proteins of ModB, a cell
lysate, prepared from exponentially growing E.coli, was incubated
with purified ModB and an NAD–10-mer RNA with a fluorescent 3 Cy5
label (Fig.5a and Extended Data Fig.8c). We approximated the cellular
conditions with respect to the presence of proteins, nucleic acids and
various small molecules, including NAD37.
Kinetic analysis of the ModB activity in these lysates showed that
several E.coli proteins were RNAylated (Extended Data Figs.8c and 9a),
including rS1 (which migrates in a similar way to an RNAylated rS1 we
added as a marker) and a protein with a mass of around 35kDa. Notably,
this pattern was not observed in the presence of 5-monophosphorylated
RNA–Cy5. We also characterized the simultaneous ADP-ribosylation
in the same lysates showing different patterns of ADP-ribosylation
targets and RNAylation targets of ModB (Extended Data Fig.9b). In
E.coli, NAD–RNA concentrations amount to around 5µM (ref. 4),
compared with an approximately 700-fold excess of NAD (2.6mM;
ref. 37). To simulate this molar excess of NAD over NAD–RNA in the
lysate assay, we added NAD to our lysates. This showed that with a
700-fold excess of NAD, RNAylation still occurs with an efficiency
of approximately 67% (Extended Data Fig.9c). We then assessed the
intensity of ModB relative to E.coli proteins by proteomics, which
revealed that a 100-fold dilution, relative to our standard assay con-
ditions, may resemble relative ModB intensity during infection38
(Extended Data Table3). In lysates with ModB concentrations closer
to those in cellular conditions, similar ADP-ribosylation and RNAyla-
tion patterns were observed as under standard conditions (Extended
Data Fig.9d).
These results indicate that in cellular conditions in which NAD is
much more abundant than NAD–RNA, ModB RNAylates specific target
proteins (Extended Data Figs.8c and 9c). Because ModB was previ-
ously assumed to preferentially ADP-ribosylate proteins involved in
translation1, we monitored the RNAylation patterns of isolated E.coli
ribosomes (Fig.5a) and observed a similar pattern to that for the lysates
(Extended Data Figs.8c and 9).
To identify the RNAylated proteins, we RNAylated the E.coli ribo-
some with a 40-nucleotide-long NAD–RNA, resulting in a gel shift of
RNAylated ribosomal proteins. MS analysis of the isolated gel band
identified the ribosomal protein L2 (rL2) as a target for RNAylation by
ModB (Extended Data Fig.10a,b). rL2 is a protein with a mass of around
35kDa and is probably the target observed in the lysates (Extended Data
Figs.8c and 9). It is evolutionarily highly conserved and is required for
the association of the 30S and 50S subunits, involved in tRNA binding
to both the A and P sites, and important for peptidyltransferase activ-
ity39. Similar to rS1, PNPase and RNaseE, rL2 contains an RNA-binding
domain that is homologous to the OB fold
40
. Invitro RNAylation assays
found that about 80% of the rL2 was RNAylated by ModB in the presence
of NAD–RNA (Extended Data Fig.10c). Invitro RNAylation sites of rL2
were identified using the LC–MS/MS approach, including an MS data
search with RNPxl, as described above. Trinucleotides (ADPr–C) were
found to be attached to R217 and R221 (Extended Data Fig.10d–g and
Supplementary Table6). R221 is located close (11Å away) to H229, which
is indispensable for ribosomal peptidyltransferase activity
39
. Future
studies will reveal whether the RNAylation of rL2 and rS1 influences
the translation efficiency of the ribosome (Fig.5b).
ModB is important for phage infection
To investigate the functional role of ModB during phage infection,
we compared the phenotypes of WT T4 and T4 ModB(R73A,G74A).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1060 | Nature | Vol 620 | 31 August 2023
Article
We observed that the burst size (the number of virions released per
infected E.coli cell) of T4 ModB(R73A,G74A) was decreased fourfold
by 50min after infection (15±3 progeny per cell) compared with WT T4
(60±32 progeny per cell) (Fig.5c). By 140min after infection, phages
produced by WT T4 (6.6×10
5
±1.3×10
5
progeny) significantly exceeded
the number of progeny from T4 ModB(R73A,G74A) (5.5×10
4
±3.1×10
4
)
(Fig.5c and Supplementary Fig.4a). At 140min after infection, a 12-fold
decrease in the progeny number compared with the WT T4 phage was
observed for T4 ModB(R73A,G74A). Thus, ModB inactivation notice-
ably affects phage propagation properties.
We also observed a delay in lysis of approximately 20min for the
E.coli culture grown in the presence of the mutant phages (Fig.5d).
To determine whether ModB affects the infection cycle at the intra-
or extracellular stage of infection, we measured the kinetics of
phage adsorption to the cell (Fig.5e). We observed a significantly
lower adsorption rate for mutant phages. At 8min after infection,
around 61.3±7.3% of the T4 ModB(R73A,G74A) mutants successfully
entered E.coli, compared with 85.3±2.4% for WT T4 phages (Fig.5e
and Supplementary Fig.4b). These results indicate that phages are
generated in the presence of inactivated ModB are less effective
in the first stages of the infection, namely the attachment to, and
penetration of, the host. This finding is consistent with the delayed
host lysis.
Discussion
Most of the interactions between RNA and proteins are non-covalent
41
,
but there are some exceptions
42
. These include the peptidyl–tRNA
intermediates in protein biosynthesis43 (which are esters) and the
adenoviral VPg proteins that form a phosphodiester bond (by means
of a tyrosine OH group) with a nucleotide, which is then used to initiate
transcription
44,45
. Here we show that an ART can attach NAD-capped
RNAs to target proteins post-transcriptionally through the forma-
tion of glycosidic bonds. This finding represents a distinct biological
function of the NAD cap on RNAs in bacteria, namely the activation
of the RNA for enzymatic transfer to an acceptor protein. We discov-
ered that the RNAylation of target proteins (a previously undescribed
post-translational protein modification) has a role in the infection of
the bacterium E.coli by bacteriophage T4. We discovered that ModB
is a target-specific ART that RNAylates proteins that are part of the
translational apparatus. We found that rS1 and rL2 are RNAylated at
specific Arg residues in their RNA-binding regions. Moreover, we identi-
fied predominantly E.coli transcripts that are linked to rS1 during T4
phage infection. Inactivation of ModB caused a delay in bacterial lysis
during phage infection and decreased the number of progeny released.
It remains unclear how the mutation of ModB (a non-capsid protein)
will affect phage adsorption to the host cell. Precisely defining phage
composition and architecture in future studies might help to explain
this phenomenon.
Our findings introduce a molecular mechanism by which the T4
phage targets the translational machinery of its host and indicate
that RNAylation might have a role in bacteriophage pathogenicity.
It remains to be determined, however, whether ADP-ribosylation
or RNAylation is the more important function of ModB. The T4
mutant ModB(R73A,G74A) abolished not only RNAylation but also
ADP-ribosylation activity. This makes it difficult to determine whether
the observed effects on T4 infection are due to RNAylation specifically
or to the loss of ADP-ribosylation activity.
ModB was known to be an enzyme that uses NAD as a substrate to
ADP-ribosylate host proteins during T4 infection. During this study, it
became clear that ModB accepts not only NAD as a substrate, but also
NAD–RNA. Enzymes typically have high specificity for their substrates
and tolerate only limited chemical modifications. It was therefore sur-
prising that ModB tolerates the attachment of a bulky RNA chain to
the 3 OH group of NAD (NAD–RNA) for the modification of a specific
subset of target proteins. Remarkably, all four of the proteins (rS1, rL2,
RNase E and PNPase) identified here as RNAylation targets of ModB are
well known to interact with RNA. We therefore assume that both the
100
70
55
35
25
kDa
70S ribosomes
NAD–10-mer–Cy5
ModB +–+
+++
–
+
++––
––++
Fluorescence (Cy5)
rL2
rS1
E. coli lysate
Coomassie Merge
050 100 150 200
101
103
105
107
Time (min)
Phages per ml
T4 ModB R73A, G74A
T4 WT
050 100 150 200 250
0
0.5
1.0
1.5
Optical density
at 600 nm
0102
030
0
20
40
60
80
100
Phage adsorption rate (%)
a
c
b
rL2
rS1
50S
30S
de
+–+
+++
–
+
++––
––++
+–+
+++
–
+
++––
––++
Time (min) Time (min)
Fig. 5 | RNAylat ion of the ribo some and phe notype of a Mo dB mutant T4
phage. a, Character ization of Mod B substrate sp ecificit y. RNAylation of two
ribosom al proteins (rS1 and rL 2) in cell lysates a nd 70S ribosom e assemblie s
(n=3). b, Illustration of th e RNAylated protei ns rS1 and rL2 in th e context of
the 70S ribo some, based o n the cryo-ele ctron micros copy structur e of the
hiberna ting 70S E.coli ribosom e (PDB: 6H4N)51. c–e, Charact erization of t he
T4 ModB R73A,G74A mutant p henotype , showing the burs t size (c), E.coli lysis
(d) and phage adsor ption (e) of WT T4 phages and T4 ModB(R 73A,G74A) (n=3
biologic ally independ ent replicate s for each). Data point s with error bar s
represen t mean±s.d. Grey do tted boxes indic ate time point s used for
assess ing statist ical signif icance in the c ase of burst size (c, 140mi n after
infectio n; two-sided Stu dent’s t-test, P=0.0015 at Psig nif<0.05) and phage
adsorption (e, 8min after infec tion; t-test, two- sided, P=0.029 at Psignif<0.05)
but indica te the delayed lysis w ithout a stat istical test i n d. Statistic al tests are
shown in Suppl ementary F ig.4.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Nature | Vol 620 | 31 August 2023 | 1061
ability of ModB to accept NAD–RNA as a substrate and the RNA affin-
ity of the target protein determine RNAylation specificity. We did not
succeed in generating a mutant of ModB that only ADP-ribosylates
or RNAylates. RNAylation occurs by an ADP-ribosylation-like mecha-
nism that involves the same catalytic residues as ADP-ribosylation,
but the RNA affinity of the target protein might determine RNAylation
specificity.
We considered why a phage ART would attach specific RNAs to
proteins involved in translation. When a T4 phage infects E.coli it
aims to reprogram the host ribosome to translate its mRNAs
46
. One
way to achieve this may be a controlled shutdown of ribosomes that
do not participate in the translation of T4 mRNAs. The discovery of
crucial ribosomal proteins, rS1 and rL2, as RNAylation targets leads
us speculate that RNAylation might impair their functionality, such
as modulating peptidyltransferase activity. The fact that mostly
E.coli transcripts are linked to rS1 invivo suggests that undesired
host gene-expression events are stopped by RNAylation. In this
way, the phage might exploit RNAylation to inactivate distinct host
ribosomes.
Future studies could show whether ribosomes that translate E.coli
transcripts are blocked by RNAylation. This proposed mechanism would
enable the phage to regulate the activity of the ribosome throughout
the infection cycle and to stop the translation of host proteins.
Why only one of the three known T4 ARTs carries out efficient RNAyla-
tion is not understood. ModA and ModB both contain characteristic
features of Arg-specific ARTs, such as the active-site motif R-S-EXE
1
.
Differences in substrate specificity are therefore probably due to
sequence differences (ModA and ModB are 25% identical and have
47% homologous amino acids)1.
ARTs are not limited to phages. ADP-ribosylated proteins have been
detected in hosts following infection by various viruses, including
influenza, coronaviruses and HIV. As well as viruses using ARTs as
weapons, the mammalian antiviral defence system uses host ARTs to
inactivate viral proteins. Moreover, mammalian ARTs and poly-(ADPr)
polymerases are regulators of critical cellular pathways and are known
to interact with RNA47. Thus ARTs might catalyse RNAylation reactions
in different organisms, making RNAylation a phenomenon of broad
biological relevance.
Finally, RNAylation may be considered as both a post-translational
protein modification and a post-transcriptional RNA modification.
Our findings challenge the established views of how RNAs and pro
-
teins interact with each other. The discovery of these previously unde-
scribed RNA–protein conjugates comes at a time when the structural
and functional boundaries between different classes of biopolymer
are becoming increasingly blurred48,49.
Online content
Any methods, additional references, Nature Portfolio reporting summa-
ries, source data, extended data, supplementary information, acknowl-
edgements, peer review information; details of author contributions
and competing interests; and statements of data and code availability
are available at https://doi.org/10.1038/s41586-023-06429-2.
1. Tiemann, B. etal. ModA and ModB, two ADP-ribosyltransferases encoded by bacteriophage
T4: catalytic properties and mutation analysis. J. Bacteriol. 186, 7262–7272 (2004).
2. Koch, T., Raudonikiene, A., Wilkens, K. & Rüger, W. Over expression, puriication, and
characterization of the ADP-ribosyltransferase (gpAlt) of bacteriophage T4:
ADP-ribosylation of E. coli RNA polymerase modulates T4 “early” transcription. Gene
Expr. 4, 253–264 (1995).
3. Cahová, H., Winz, M. L., Höfer, K., Nübel, G. & Jäschke, A. NAD captureSeq indicates NAD
as a bacterial cap for a subset of regulatory RNAs. Nature 519, 374–377 (2015).
4. Chen, Y. G., Kowtoniuk, W. E., Agarwal, I., Shen, Y. & Liu, D. R. LC/MS analysis of cellular
RNA reveals NAD-linked RNA. Nat. Chem. Biol. 5, 879–881 (2009).
5. Jiao, X. etal. 5 End nicotinamide adenine dinucleotide cap in human cells promotes RNA
decay through DXO-mediated deNADding. Cell 168, 1015–1027(2017).
6. Fehr, A. R. etal. The impact of PARPs and ADP-ribosylation on inlammation and host–
pathogen interactions. Genes Dev. 34, 341–359 (2020).
7. Cohen, M. S. & Chang, P. Insights into the biogenesis, function, and regulation of
ADP-ribosylation. Nat. Chem. Biol. 14, 236–243 (2018).
8. Simon, N. C., Aktories, K. & Barbieri, J. T. Novel bacterial ADP-ribosylating toxins: structure
and function. Nat. Rev. Microbiol. 12, 599–611 (2014).
9. Lüscher, B. etal. ADP-ribosylation, a multifaceted posttranslational modiication
involved in the control of cell physiology in health and disease. Chem. Rev. 118,
1092–1136 (2018).
10. Morales-Filloy, H. G. etal. The 5 NAD cap of RNAIII modulates toxin production in
Staphylococcus aureus isolates. J. Bacteriol. 202, e00591-19 (2020).
11. Frindert, J. etal. Identiication, biosynthesis, and decapping of NAD-capped RNAs in
B. subtilis. Cell Rep. 24, 1890–1901 (2018).
12. Ruiz-Larrabeiti, O. etal. NAD+ capping of RNA in Archaea and Mycobacteria. Preprint at
bioRxiv https://doi.org/10.1101/2021.12.14.472595 (2021).
13. Gomes-Filho, J. V. etal. Identiication of NAD-RNAs and ADPR-RNA decapping in the
archaeal model organisms Sulfolobus acidocaldarius and Haloferax volcanii. Preprint at
bioRxiv https://doi.org/10.1101/2022.11.02.514978 (2022).
14. Walters, R. W. etal. Identiication of NAD+ capped mRNAs in Saccharomyces cerevisiae.
Proc. Natl Acad. Sci. USA 114, 480–485 (2017).
15. Zhang, Y. etal. Extensive 5-surveillance guards against non-canonical NAD-caps of nuclear
mRNAs in yeast. Nat. Commun. 11, 5508 (2020).
16. Wang, J. etal. Quantifying the RNA cap epitranscriptome reveals novel caps in cellular
and viral RNA. Nucleic Acids Res. 47, e130 (2019).
17. Zhang, H. etal. NAD tagSeq reveals that NAD+-capped RNAs are mostly produced from
a large number of protein-coding genes in Arabidopsis. Proc. Natl Acad. Sci. USA 116,
12072–12077 (2019).
18. Wang, Y. etal. NAD+-capped RNAs are widespread in the Arabidopsis transcriptome and
can probably be translated. Proc. Natl Acad. Sci. USA 116, 12094–12102 (2019).
19. Dong, H. etal. NAD+-capped RNAs are widespread in rice (Oryza sativa) and
spatiotemporally modulated during development. Sci. China Life Sci. 65, 2121–2124
(2022).
20. Wolfram-Schauerte, M. & Höfer, K. NAD-capped RNAs – a redox cofactor meets RNA.
Trends Biochem. Sci. 48, 142–155 (2022).
21. Höfer, K. & Jäschke, A,. Epitranscriptomics: RNA modiications in Bacteria and
Archaea. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.RWR-0015-2017
(2018).
22. Miller, E. S. etal. Bacteriophage T4 genome. Microbiol. Mol. Biol. Rev. 67, 86–156
(2003).
23. Rohrer, H., Zillig, W. & Mailhammer, R. ADP-ribosylation of DNA-dependent RNA polymerase
of Escherichia coli by an NAD+: protein ADP-ribosyltransferase from bacteriophage T4.
Eur. J. Biochem. 60, 227–238 (1975).
24. Depping, R., Lohaus, C., Meyer, H. E. & Rüger, W. The mono-ADP-ribosyltransferases Alt
and ModB of bacteriophage T4: target proteins identiied. Biochem. Biophys. Res. Commun.
335, 1217–1223 (2005).
25. Skórko, R., Zillig, W., Rohrer, H., Fujiki, H. & Mailhammer, R. Puriication and properties of
the NAD+: protein ADP-ribosyltransferase responsible for the T4-phage-induced modiication
of the α subunit of DNA-dependent RNA polymerase of Escherichia coli. Eur. J. Biochem.
79, 55–66 (1977).
26. Tiemann, B., Depping, R. & Rüger, W. Overexpression, puriication, and partial
character ization of ADP-ribosyltransferases ModA and ModB of bacteriophage T4.
Gene Expr. 8, 187–196 (1999).
27. Rankin, P. W., Jacobson, E. L., Benjamin, R. C., Moss, J. & Jacobson, M. K. Quantitative
studies of inhibitors of ADP-ribosylation invitro and invivo. J. Biol. Chem. 264, 4312–4317
(1989).
28. Goelz, S. & Steitz, J. A. Escherichia coli ribosomal protein S1 recognizes two sites in
bacteriophage Qβ RNA. J. Biol. Chem. 252, 5177–5179 (1977).
29. Cervantes-Laurean, D., Jacobson, E. L. & Jacobson, M. K. Glycation and glycoxidation of
histones by ADP-ribose. J. Biol. Chem. 271, 10461–10469 (1996).
30. Kramer, K. etal. Photo-cross-linking and high-resolution mass spectrometry for
assignment of RNA-binding sites in RNA-binding proteins. Nat. Methods 11, 1064–1070
(2014).
31. Gehrig, P. M. etal. Gas-phase fragmentation of ADP-ribosylated peptides: arginine-speciic
side-chain losses and their implication in database searches. J. Am. Soc. Mass. Spectrom.
32, 157–168 (2021).
32. Winz, M. -L . etal. Capture and sequencing of NAD-capped RNA sequences with NAD
captureSeq. Nat. Protoc. 12, 122–149 (2017).
33. Tao, P., Wu, X., Tang, W.-C., Zhu, J. & Rao, V. Engineering of bacteriophage T4 genome
using CRISPR-Cas9. ACS Synth. Biol. 6, 1952–1961 (2017).
34. Zhang, H. etal. Use of NAD tagSeq II to identify growth phase-dependent alterations in
E. coli RNA NAD+ capping. Proc. Natl Acad. Sci. USA 118, e2026183118 (2021).
35. Bird, J. G. etal. The mechanism of RNA 5 capping with NAD+, NADH and desphospho-CoA.
Nature 535, 444–447 (2016).
36. Bycroft, M., Hubbard, T. J. P., Proctor, M., Freund, S. M. V. & Murzin, A. G. The solution
structure of the S1 RNA binding domain: a member of an ancient nucleic acid-binding
fold. Cell 88, 235–242 (1997).
37. Bennett, B. D. etal. Absolute metabolite concentrations and implied enzyme active site
occupancy in Escherichia coli. Nat. Chem. Biol. 5, 593–599 (2009).
38. Wolfram-Schauerte, M., Pozhydaieva, N., Viering, M., Glatter, T. & Höfer, K. Integrated
omics reveal time-resolved insights into T4 phage infection of E. coli on proteome and
transcriptome levels. Viruses 14, 2502 (2022).
39. Diedrich, G. etal. Ribosomal protein L2 is involved in the association of the ribosomal
subunits, tRNA binding to A and P sites and peptidyl transfer. EMBO J. 19, 5241–5250
(2000).
40. Nakagawa, A. etal. The three-dimensional structure of the RNA-binding domain of
ribosomal protein L2; a protein at the peptidyl transferase center of the ribosome. EMBO J.
18, 1459–1467 (1999).
41. Hent ze, M. W., Castello, A., Schwarzl, T. & Preiss, T. A brave new world of RNA-binding
proteins. Nat. Rev. Mol. Cell Biol. 19, 327–341 (2018).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1062 | Nature | Vol 620 | 31 August 2023
Article
42. Drygin, Y. F. Natural covalent complexes of nucleic acids and proteins: some comments
on practice and theory on the path from well-known complexes to new ones. Nucleic
Acids Res. 26, 4791–4796 (1998).
43. Ibba, M. & Söll, D. Aminoacyl-tRNA synthesis. Annu. Rev. Biochem. 69, 617–650 (2000).
44. Rekosh, D. M. K., Russell, W. C., Bellet, A. J. D. & Robinson, A. J. Identiication of a protein
linked to the ends of adenovirus DNA. Cell 11, 283–295 (1977).
45. Rothberg, P. G., Harris, T. J. R., Nomoto, A. & Wimmer, E. O4-(5-uridylyl)tyrosine is the bond
between the genome-linked protein and the RNA of poliovirus. Proc. Natl Acad. Sci. USA
75, 4868–4872 (1978).
46. Uzan, M. & Miller, E. S. Post-transcriptional control by bacteriophage T4: mRNA decay and
inhibition of translation initiation. Virol. J 7, 360 (2010).
47. Ke, Y., Zhang, J., Lv, X., Zeng, X. & Ba, X. Novel insights into PARPs in gene expression:
regulation of RNA metabolism. Cell. Mol. Life Sci. 76, 3283–3299 (2019).
48. Otten, E. G. etal. Ubiquitylation of lipopolysaccharide by RNF213 during bacterial infection.
Nature 594, 111–116 (2021).
49. Flynn, R. A. etal. Small RNAs are modiied with N-glycans and displayed on the surface of
living cells. Cell 184, 3109–3124 (2021).
50. Tsuge, H. etal. Structural basis of actin recognition and arginine ADP-ribosylation by
Clostridium perfringens ι-toxin. Proc Natl Acad. Sci. USA 105, 7399–7404 (2008).
51. Beckert, B. etal. Structure of a hibernating 100S ribosome reveals an inactive conformation
of the ribosomal protein S1. Nat. Microbiol. 3, 1115–1121 (2018).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afiliations.
Open Access This article is licensed under a Creative Commons Attribution
4.0 International License, which permits use, sharing, adaptation, distribution
and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons licence,
and indicate if changes were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your
intended use is not permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a copy of this licence,
visit http://creativecommons.org/licenses/by/4.0/.
© The Author(s) 2023
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Methods
General
Reagents were purchased from Sigma-Aldrich and used without further
purification. Oligonucleotides, DNA and RNA were purchased from
Integrated DNA Technologies (Supplementary Tables7–10). Concen-
trations of DNA and RNA were determined by measurements using
the NanoDrop ND-1000 spectrophotometer. Radioactively labelled
proteins and nucleic acids were visualized using storage phosphor
screens (GE Healthcare) and a Typhoon 9400 imager (GE Healthcare).
Uncropped gel and blot images are provided (Supplementary Fig.1).
Preparation of 5′ppp–RNA, 5′p–RNA and 5′-NAD–RNA by invitro
transcription
DNA templates for Qβ RNA (100-nucleotide RNA) and E. coliRNAI
were amplified by PCR (primer sequences are listed in Supplementary
Table9), and PCR products were analysed by 2% agarose gel electro-
phoresis and purified using the QIAquick PCR purification kit (QIA-
GEN). 5-Triphosphate (ppp) Qβ RNA and RNAI were synthesized by
invitro transcription in the presence of 1×transcription buffer (40mM
Tris, pH8.1, 1mM spermidine, 10mM MgCl2, 0.01% Triton X-100),
5%DMSO, 10mM DTT, 4mM of each NTP, 20µg T7 RNA polymerase
(2mgml−1, purified in our laboratory) and 200nM DNA template.
NAD–RNAI was made under similar conditions using 2mM ATP and
4mM NAD. The same conditions were applied for the synthesis of a
mixture of α-32P-labelled 5-NAD and pppQβ RNAs, except we used
2mM ATP, 80µCi
32
P-α-ATP and 4mM NAD instead of 4mM ATP. The
invitro transcription reactions were incubated at 37°C for 4h and
digested with DNaseI (Roche). RNA was purified by denaturing PAGE,
isopropanol-precipitated and resuspended in Millipore water. RNA
sequences are listed in Supplementary Table7.
To convert 5ppp–RNAs into 5-monophosphate–RNAs (5p–RNAs),
250pmol Qβ RNA was treated with 60U RNA 5-polyphosphatase
(Epicentre) in 1×polyphosphatase reaction buffer at 37°C for 70min.
Protein was removed from 5p–RNAs by phenol–chloroform extrac-
tion and residual phenol–chloroform was removed by three rounds
of diethyl ether extraction. 5p–RNAs were isopropanol precipitated
and resuspended in Millipore water.
5′-radiolabelling of 5′-monophosphate and NAD-capped RNAs
We treated 120pmol 5p-Qβ RNA or 6.25nmol 5p–RNA 8-mer (Sup-
plementary Table7) with 50U T4 polynucleotide kinase in 1×reaction
buffer B and 1,250µCi
32
P-γ-ATP. The reaction was incubated at 37°C for
2h. The resulting 5-32P-RNA 8-mer and 5-32P-Qβ RNA were separated
from residual protein by phenol–chloroform extraction. The remain-
ing
32
P-γ-ATP was removed by washing with three column volumes of
Millipore water and centrifugation in 10kDa (for Qβ RNA) or 3kDa
(for the 8-mer) Amicon filters (Merck Millipore) at 14,000rpm at 4°C
four times. RNA sequences are listed in Supplementary Table7. To
convert the purified 5-
32
P-RNAs into 5-
32
P-NAD-capped RNA, 800pmol
5-
32
P-RNA 8-mer or 30pmol 5-
32
P-Qβ RNA was incubated in 50mM
MgCl
2
in the presence of a spatula tip of nicotinamide mononucleo-
tide phosphorimidazolide, synthesized as described
52
, at 50°C for 2h.
RNAs were purified by washing with Millipore water and centrifugation
in 10kDa (Qβ RNAs) or 3kDa (8-mer) Amicon filters at 14,000 rpm
at 4°C four times. The concentrations of the 5-
32
P-RNAs were meas-
ured using a NanoDrop ND-1000 spectrophotometer and were used to
calculate the approximate concentrations of yielded 5-NAD-capped
32P-RNAs, assuming an approximate yield of the imidazolide reac-
tion of 50% (ref. 52). The 5-
32
P-ADPr–RNA 8-mer was synthesized by
incubating 8µM 5-32P-NAD–RNA 8-mer and 0.08µM ADP-ribosyl
cyclase CD38 (R&D Systems) in 1×degradation buffer at 37°C for 4h.
The reaction was purified by P/C/I-diethyl ether extraction and filtra-
tion through 3kDa filters and washing with four column volumes of
Millipore water.
Cloning of ADP-ribosyltransferases, ADP-ribose hydrolases and
target proteins
To amplify bacteriophage T4 genes modA (GeneID: 1258568; Uniprot:
P39421), modB (GeneID: 1258688; Uniprot: P39423) and alt (GeneID:
1258760; Uniprot: P12726), a single plaque from bacteriophage T4
revitalization was resuspended in Millipore water and used in a
‘plaque’ PCR, analogous to bacterial-colony PCR. The gene encoding
the ADP-ribosylhydrolase ARH1 (GeneID: 141; Uniprot: P54922) was
purchased from IDT as gBlocks and amplified by PCR. E.coli genes cod-
ing for rS1 (GeneID: 75205313; Uniprot: P0AG67), rL2 (GeneID: 947820;
Uniprot: P60422) and PNPase (GeneID: 947672; Uniprot: P05055) were
PCR-amplified from genomic DNA of E.coli K12, which was isolated
using a GenElute Bacterial Genomic DNA Kit (Sigma-Aldrich). Nucleo-
tide sequences are listed in Supplementary Table8. XhoI and NcoI
restriction sites were introduced during amplification using appropri-
ate primers (Supplementary Table9). The resulting PCR product was
digested with XhoI and NcoI (Thermo Fisher Scientific) and cloned into
the pET–28c vector (Merck Millipore). After Sanger sequencing, the
resulting plasmids were transformed into E.coli One Shot BL21 (DE3)
(Life Technologies). The ARH1 D55,56A, ModB(R73A) and rS1 mutants
were generated by site-directed mutagenesis using a procedure based
on the Phusion Site-Directed Mutagenesis Kit (Thermo Scientific). The
resulting plasmids were sequenced and transformed into E.coli One
Shot BL21 (DE3). All strains used and generated in this work are sum-
marized in Supplementary Table10.
Purification of rS1, rS1 domains and variants, rL2, the PNPase S1
domain, RNase E(1–529), Alt, NudC, NudC*(V157A, E174A, E177A,
E178A) and NudC(E178Q)
Isopropyl beta--thiogalactoside (IPTG)-induced E.coli One Shot BL21
(DE3) containing the respectiveplasmid (Supplementary Table10)
was cultured in LB medium at 37°C. Protein expression was induced
at an optical density at 600 nm (OD600) of 0.8, bacteria were collected
after centrifugation for 3h at 37°C and lysed by sonication (30s at 50%
power, five times) in HisTrap buffer A (50mM Tris-HCl, pH7.8, 1M NaCl,
1M urea, 5mM MgSO
4
, 5mM β-mercaptoethanol, 5% glycerol, 5mM
imidazole, one tablet per 500ml complete EDTA-free protease inhibitor
cocktail (Roche)). The lysate was cleared by centrifugation (37,500g
for 30min at 4°C) and the supernatant was applied to a 1ml Ni-NTA
HisTrap column (GE Healthcare). The protein was eluted with an imida-
zole gradient using an analogous gradient of HisTrap bufferB (HisTrap
buffer A with 500mM imidazole added) and analysed by SDS–PAGE.
Further protein purification was achieved by size-exclusion chro-
matography (SEC) through a Superdex200 10/300 GL column (GE
Healthcare) using SEC buffer containing 0.5M NaCl and 25mM Tris-HCl,
pH8. Fractions of interest were analysed by SDS–PAGE, pooled and
concentrated in Amicon Ultra-4 centrifugal filters (molecular weight
cut-off (MWCO) 10kDa with centrifugation at 2,000rpm and 4°C). Pro-
tein concentration was measured with a NanoDrop ND-1000 spectro-
photometer. Finally, proteins were stored in SEC buffer supplemented
with 50% glycerol at −20°C.
Purification of ARH1 and ARH1(D55A, D56A)
E.coli BL21 DE3 pET28-ARH1 and BL21-pET28-ARH1 D55A, D56A (Sup-
plementary Table10) were grown to an OD600=0.6 at 37°C and 175rpm.
Afterwards, bacteria were allowed to cool to room temperature for
30min. Expression was induced with 1mM IPTG, and bacteria were
finally grown overnight at room temperature while shaking at 175rpm.
Bacteria were collected by centrifugation and proteins were purified
in a similar way to rS1 variants.
Purification of ModA
E.coli BL21 DE3 pET28-ModA (Supplementary Table10) was grown
to an OD600=1 at 37°C with shaking at 175rpm. Protein expres-
sion was induced with 0.5mM IPTG and bacteria were collected by
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
centrifugation after 3h at 37°C. Pelleted bacteria were resuspended
in 50mM NaH2PO4, pH8, 300mM NaCl, 1mM DTT with one tablet per
500ml complete EDTA-free protease inhibitor cocktail (Roche) and
lysed by sonication (3×1min at 5% power). Lysates were centrifuged
at 3,000g at 4°C for 20min. Sediments were washed by resuspension
in 30ml 50mM Tris-HCl, pH7.5, 2mM EDTA, 100mM NaCl, 1M urea,
1mM DTT and one tablet EDTA-free protease inhibitor (Roche), and
centrifuged at 10,000g at 4°C for 20min. Pellets containing inclu-
sion bodies were resuspended in 40ml 100mM Tris, pH11.6, 8M urea,
transferred to 12–14kDa MWCO dialysis bags (Roth) and dialysed over-
night against 50mM NaH2PO4, 300mM NaCl. Protein solutions were
centrifuged at 20,000g at 4°C for 30min. Supernatants were batch
purified using disposable 10ml columns (Thermo Fisher Scientific)
packed with 2ml Ni-NTA agarose (Jena Bioscience) and equilibrated
with 10 column volumes of 50mM NaH
2
PO
4
(pH 8), 300mM NaCl. Pro-
teins were purified by washing the columns with 30column volumes
of 50mM NaH2PO4, 300mM NaCl, 15mM imidazole, eluted with 5ml
50mM NaH
2
PO
4
, 300mM NaCl, 300mM imidazole and concentrated
in Amicon (Merck Millipore) filters (MWCO 10kDa with centrifuga-
tion at 2,000rpm and 4°C). Finally proteins were purified by SEC, as
described for rS1.
Purification of ModB and ModB(R73A, G74A)
E.coli BL21 DE3 pET28–ModB and E.coli BL21 DE3 pET28–
ModB(R73A,G74A) (Supplementary Table10) were grown to OD600=2.0
at 37°C with shaking at 185rpm and cooled to 4°C while being shaken
at 160rpm for at least 30min. Protein expression was induced by the
addition of 1mM IPTG. The cultures were then incubated for 120min at
4°C, with shaking at 160rpm and bacteria were collected by centrifu-
gation (4,000rpm at 4°C for 25min). The ModB protein was purified
from the supernatant as described for rS1 variants.
Alphafold prediction of ModB structure
The Alphafold prediction of ModB structure was performed with Alpha-
Fold2.ipynb (v.1.3.0, https://colab.research.google.com/github/sokryp-
ton/ColabFold/blob/main/AlphaFold2.ipynb) with default parameters
(use_templates=false, use_amber=false; msa_mode=MMseqs2
(UniRef+Environmental), model_type=“AlphaFold2-ptm”, max_
msa=null, pair_mode=unpaired+paired, auto advanced settings).
The ModB protein sequence was retrieved from Uniprot (primary acces-
sion: P39423). The ModB structure prediction model from rank_1 was
further assessed using PyMol.
Invitro ADP-ribosylation and RNAylation of rS1 and rL2 with
32P-labelled NAD, NAD–8-mer, NAD–Qβ RNA or NAD–10-mer–Cy5
rS1 (0.3µM) was ADP-ribosylated in the presence of 0.25µCiµl−1
32
P-NAD or RNAylated in the presence of one of 0.6µM
32
P-NAD–8-mer,
0.03µM
32
P-NAD–Qβ RNA or 0.8 µM NAD–10-mer–Cy5 (Supplemen-
tary Table7) by 1.4µM ModB and in 1×transferase buffer (10mM
Mg(OAc)
2
, 22mM NH
4
Cl, 50mM Tris-acetate pH7.5, 1mM EDTA, 10mM
β-mercaptoethanol and 1% glycerol) at 15°C for at least 120min. Sam-
ples (5µl) were taken before the addition of ModB and after 1, 2, 5, 10, 30,
60 and 120min, and mixed with 5µl 2×Laemmli buffer to stop the reac-
tion. Reactions were assessed by 12% SDS–PAGE and gels were stained
in Instant Blue solution (Sigma-Aldrich) for 10min. Radioactive signals
were visualized using storage phosphor screens and a Typhoon 9400
imager. The intensity of the radioactive bands was quantified using
ImageQuant 5.2 (GE Healthcare). The RNAylation with NAD-capped
Cy5-labelled RNA was visualized with the ChemiDoc (Bio-Rad) Cy5
channel. Gels were then stained by Coomassie solution and imaged
using the same system. In some cases, stain-free imaging of proteins in
SDS gels was performed by 2,2,2-trichloroethanol (TCE) incorporated
in the gel. TCE binds to tryptophan residues of the proteins, which
enhances their fluorescence under ultraviolet light and thereby enables
their detection53.
rL2 was ADP-ribosylated or RNAylated at the same settings using
either 6.4µM NAD or 6.4µM NAD–8-mer as a substrate to modify 4.6µM
rL2 in the presence of 1.57µM ModB for 4h for LC–MS/MS measure-
ments. For shift assays, 538nM rL2 was RNAylated by 2.61µM ModB
in the presence of 6µM NAD–8-mer. 12% SDS–PA gels were fixed with
a solution of 40%ethanol and 10%acetic acid overnight and stained
using Flamingo fluorescent protein dye (Bio-Rad) for up to 6h and
imaged with the ChemiDoc (Bio-Rad). Signal intensity was quantified
in ImageLab (Bio-Rad). Where indicated, statistical tests were per-
formed using two-sided t-tests in R (v.4.2.2) implemented in the ggpubr
package (v.0.6.0) using a significance level of 0.05.
Invitro RNAylation of E. coli RNA polymerase with NAD–
10-mer–Cy5
We incubated 0.8µM NAD–10-mer–Cy5 (Supplementary Table7) with
0.5µM of protein E.coli RNA polymerase (New England Biolabs) and
3µM Alt or ModA in the presence of 1×transferase buffer at 15°C for
60min. Samples were taken before the addition of Alt or ModA and
after 60min incubation. The reactions were stopped by the addition
of 1volume of 2×Laemmli buffer. Reactions were analysed by 10% SDS–
PAGE with rS1 RNAylated by ModB with NAD–10-mer–Cy5 as a refer-
ence protein. RNAylated proteins were visualized using the ChemiDoc
(Bio-Rad) Cy5 channel. Afterwards, gels were stained in Coomassie
solution and imaged using the same system.
Analysis of protein rS1 self-RNAylation
In 20-µl reactions, 3.6µM
32
P-ADPr–8-mer (Supplementary Table7)
was incubated with either 2.6µM rS1, 3.9µM ModB or both 2.59µM rS1
and 3.9µM ModB in 1×transferase buffer. As a positive control, equal
amounts of protein rS1 and ModB were incubated with 0.6µM 32P-NAD–
8-mer. All reactions were incubated at 15°C for 60min. Samples were
taken before the addition of ModB or after 60 min, and reactions
were stopped by adding one volume of 2×Laemmli buffer. Reactions
were analysed by 12% SDS–PAGE and autoradiography imaging.
RNAylation of protein rS1 with Qβ RNA (100-nucleotide–RNA)
and specificity for the 5′-NAD cap
0.05µM
32
P-NAD–Qβ RNA, 0.15µM 5-
32
P-Qβ RNA or 0.15µM 5-
32
PPP-Qβ
RNA (Supplementary Table7) was incubated with 2.3µM rS1 and 1.4µM
ModB in the presence of 1×transferase buffer at 15°C for 60min.
Samples were taken before the addition of ModB and after 60 min,
and reactions were stopped by adding 1volume 2×Laemmli buffer.
Reactions were analysed by 10% SDS–PAGE, applying rS1–
32
P-ADPr in
1×Laemmli buffer as a reference, and subsequent autoradiography
imaging.
Preparation of RNAylated and ADP-ribosylated rS1 for enzymatic
treatments
ADP-ribosylation or RNAylation reactions were performed with
radio-labelled substrates, washed and equilibrated in 1×transferase or
1×degradation buffer for further enzymatic treatments. The reactions
were washed with four column volumes of the corresponding buffer by
centrifugation at 10,000g at 4°C in 10kDa Amicon (Merck Millipore)
filters. Proteins RNAylated with Cy5-labelled RNA were equilibrated
in the same buffers using Zeba Spin desalting columns (7kDa MWCO,
0.5ml) (Thermo Fisher Scientific) according to the manufacturer’s
instructions.
Nuclease P1 digest of protein rS1 RNAylated with 100-nucleotide–
RNA (rS1-100-nucleotide–RNA)
An rS1–100-nucleotide-RNA (
32
P) mixture (19µl) was equilibrated in
1×transferase buffer and incubated with either 1µl nuclease P1 or 1µl
Millipore water at 37°C for 60min. Samples were taken at the beginning
and after 60 min, and reactions were stopped by adding one volume of
2×Laemmli buffer. Reactions were analysed by 10% SDS–PAGE, applying
Content courtesy of Springer Nature, terms of use apply. Rights reserved

rS1–
32
P-ADPr in 1×Laemmli buffer as a reference, and subsequent auto-
radiography imaging.
Tryptic digest of 32P-labelled rS1–8-mer and rS1–ADPr
Mixtures (19µl) of both rS1 and rS1–8-mer (32P) and of rS1 and rS1–
ADPr (
32
P) in 1×degradation buffer were incubated with either 0.2µg
Trypsin (Sigma, EMS0004, mass-spectrometry grade) or Millipore
water as a negative control at 37°C. Samples were taken before the
addition of Trypsin/Millipore water and after 120 min. Reactions
were stopped by adding one volume 2×Laemmli buffer to sam-
ples and were analysed by 12% SDS–PAGE and autoradiography
imaging.
Chemical removal of ADP-ribosylation and RNAylation invitro
Aliquots from washed and equilibrated ADP-ribosylated (1µl) and
RNAylated (2µl) (32P) rS1 were treated with either 10mM HgCl2 or
500mM NH
2
OH (refs. 54,55) at 37°C for 1h. Reactions were stopped
by adding 2×Laemmli buffer and analysed by 12% SDS–PAGE.
Enzymatic removal of ADP-ribosylation and RNAylation invitro
Aliquots from washed and equilibrated (in 1×degradation buffer)
ADP-ribosylated (1µl) and RNAylated (2µl) rS1 (32P) were treated
with 0.5U endonucleaseP1 (Sigma-Aldrich)56 or 0.95µM ARH1 or
ARH3 (human recombinant, Enzo Life Science)
57
in the presence of
10mM Mg(OAc)
2
, 22mM NH
4
Cl, 50mM HEPES, 1mM EDTA, 10mM
β-mercaptoethanol and 1% (v/v) glycerol in a total volume of 20µl at
37°C for 1h. Enzymatic reactions were stopped by adding 2×Laemmli
buffer and analysed by 12% SDS–PAGE.
Inhibition of RNAylation and ADP-ribosylation with
3-methoxybenzamide
Reactions (20µl) of 1.4µM ModB and 2.3µM protein rS1 with either
1µM 32P-NAD–8-mer or 3µM 5′-32P–8-mer (Supplementary Table7)
were incubated in the presence of 2mM 3-MB (50mM stock in DMSO)
or the absence of the inhibitor (DMSO only) at 15°C (ref. 58). Samples
were taken before the addition of ModB and after 60 min. Reactions
were stopped by the addition of 1 volume 2×Laemmli buffer and
analysed by 12% SDS–PAGE.
Effect of RNA secondary structure on RNAylation efficiency
We incubated1.1µM NAD–RNA–Cy5 (linear, 5 overhang, 3 overhang
and blunt ends; Supplementary Table7) with 0.9µM rS1 and 0.4µM
ModB in 1×transferase buffer. Samples of 5µl were taken before the
addition of ModB protein and 2, 5, 10, 30, 60 and 120min after the
start of the reaction. The samples were directly mixed with one vol-
ume of 2×Laemmli buffer to stop the reaction. The conversion of the
substrates was analysed by 12% SDS–PAGE, following visualization
on ChemiDoc (Bio-Rad) in the Cy5 channel. The maximum observed
signal intensity of RNAylated rS1 protein was used to determine the
relative conversion for each of the analysed substrates at distinct
time points.
Culture of the E. coli B strain and infection with T4 phages
Precultures of E.coli B strain pTAC-rS1 (Supplementary Table10) were
incubated in LB medium with 100µgml−1 ampicillin at 37°C and 185rpm
overnight. For the main cultures, 150ml LB medium with 100µgml−1
ampicillin were inoculated with preculture to an OD600=0.1. At
OD600=0.4, protein expression was induced by the addition of 1mM
IPTG. At OD
600
=0.8, cultures were either infected with bacteriophage
T4 at a multiplicity of infection (MOI) of 10 (20ml phage solution)
(DSM 4505, Leibniz Institute DSMZ) or not infected by adding 20ml
LB medium instead (negative control). Cultures were incubated for
20min at 37°C with shaking at 240rpm. Bacteria were collected by
centrifugation at 4,000g at room temperature for 15min. Pellets were
stored at −80°C.
Purification of His-tagged rS1 from infected E. coli strain B
pTAC-rS1
Bacterial pellets were resuspended in 10ml bufferA and lysed via
sonication (1×5min, cycle 2 at 50% power). Lysates were centrifuged
at 37,500g at 4°C for 30min. The supernatant was filtered through
0.45-µm filters (Sarstedt). rS1 from bacteriophage T4-infected or
non-infected E.coli B strain was purified from the supernatant by
gravity Ni-NTA affinity chromatography. Ni-NTA agarose slurry (1ml,
Thermo Fisher Scientific) was added to a 10ml propylene column and
equilibrated in bufferA. The supernatant was loaded onto the column
twice. The column was washed with a mixture of 95% buffer A and 5%
buffer B containing 29.75mM imidazole. Protein was eluted from the
column with 10ml bufferB.
His-tagged-protein rS1 from T4-infected or uninfected E.coli B strain
pTAC-rS1 was washed with two filter volumes of 1×degradation buffer
(12.5mM Tris-HCl, pH7.5, 25mM NaCl, 25mM KCl, 5mM MgCl2) by
centrifugation in 10-kDa Amicon filters at 5,000g at 4°C and concen-
trated to a final volume of 120µl. The fractions were analysed by 12%
SDS–PAGE analysis and the gel was stained in Instant Blue solution for
10min and imaged immediately.
Purification of His-tagged rS1 and rL2 for LC–MS/MS analysis
E.coli B strain with endogenously His-tagged rS1 and E.coli B strain
expressing His-tagged rS1 WT, R139A or R139K were infected with T4
to an MOI of 5.0, as described above for 8min. 100ml culture was col-
lected and the pellet resuspended in 1.5ml Ni-NTA bufferA with 15mM
imidazole (50mM Tris-HCl, pH7.8, 1M NaCl, 1M urea, 5mM MgSO
4
,
5mM β-mercaptoethanol, 5% glycerol, 15mM imidazole, one tablet
per 500ml complete EDTA-free protease inhibitor cocktail (Roche)).
Cells were lysed by sonication (three times for 2min at 80% power)
and supernatant was cleared by centrifugation at 17,000g at 4°C for
30min. The supernatant was incubated with 75µl Ni-NTA magnetic
beads ( Jena Bioscience) equilibrated in Ni-NTA bufferA with 15mM
imidazole for 1h at 4°C. Magnetic beads were washed seven times
with 1ml Ni-NTA bufferA with 15mM imidazole and three times with
Ni-NTA buffer without imidazole but with 4M urea. Finally, protein was
eluted by addition of Ni-NTA elution buffer (50mM Tris-HCl, pH 7.8, 1M
NaCl, 1M Urea, 5mM MgSO4, 5mM β-mercaptoethanol, 5% glycerol,
300mM imidazole, one tablet per 500ml complete EDTA-free protease
inhibitor cocktail (Roche)). Protein was equilibrated in 1×transferase
buffer with Zeba columns (7kDa MWCO, 0.5ml) according to the manu-
facturer’s instructions, and protein was digested with trypsin in a 1:20
ratio (w/w) at 37°C for 3h. Peptides were C18-purified using 50mM
triethylamine-acetate (pH7.0) buffer in combination with 0–90% ace-
tonitrile and Chromabond C18 WP spin columns (20mg, Macherey
Nagel). Purified peptides were dissolved in HPLC-grade H2O and sub-
jected to LC–MS/MS analysis (see below).
Invitro RNAylated rS1 (D2) reactions in 1×transferase buffer were
directly digested (without further purification) with 1µg RNaseA
(Thermo Fisher Scientific) and 100U RNaseT1 (Thermo Fisher
Scientific) at 37°C for 1h, following tryptic digest at 37°C for 3h in
the same buffer with trypsin (Promega) in a 1:30 ratio (w/w) relative
to the total protein content per sample. Peptides were purified with
Chromabond C18 WP spin columns as described above and used for
LC–MS/MS analysis (see below).
Invitro RNAylation reactions of rL2 with NAD–8-mer and ADP-
ribosylation reactions were purified at similar settings to the pro-
teins from T4 phage-infected E.coli. Here, reactions (200µl) were
incubated with 100µl Ni-NTA beads equilibrated in 800µl Ni-NTA
bufferA with 10mM imidazole and 40U murine RNase inhibitor (New
England Biolabs) at 4°C for 1h. Beads were washed eight times with
1ml streptavidin wash buffer (50mM Tris-HCl, pH7.4, 8M urea) at
room temperature and protein was eluted with 130µl Ni-NTA elution
buffer. Purified proteins were rebuffered in 100mM NH
4
OAc using
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Zeba spin desalting columns (7 kDa MWCO, 0.5ml) according to the
manufacturer’s instructions. rL2 samples were dissolved in 4M urea in
50mM Tris-HCl (pH7.5) and incubated for 30min at room temperature,
followed by dilution to 1M urea with 50mM Tris-HCl (pH7.5). 10µg
RNaseA (Thermo Fisher Scientific) and 1kU RNaseT1 (Thermo Fisher
Scientific) were added, following incubation for 4h at 37°C. For protein
digestion, 0.5µg trypsin (Promega) was added to each sample and
digestion was performed overnight at 37°C. Samples were adjusted
to 1% acetonitrile (ACN) and to pH3 using formic acid. Samples were
cleaned up using C18 columns (Harvard Apparatus) according to the
manufacturer’s instructions.
LC–MS/MS analysis of His-tagged, invitro RNAylated rS1 and rL2
Cleaned-up rS1 and rL2 peptide samples were dissolved in 2% ACN,
0.05% trifluoroacetic acid and subjected to LC–MS/MS analysis using
an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific)
coupled to a Dionex Ultimate 3000 RSLCnano system. Peptides were
loaded on a Pepmap 300 C18 trap column (Thermo Fisher Scientific)
(flow rate, 10µlmin−1) in bufferA (0.1% (v/v) formic acid) and washed
for 3min with bufferA. Peptide separation was performed on an
in-house-packed C18 column (30cm; ReproSil-Pur 120Å, 1.9µm, C18
AQ; inner diameter, 75µm; flow rate 300nlmin−1) by applying a linear
gradient of bufferB (80% (v/v) ACN, 0.08% (v/v) formic acid). The
main column was equilibrated with 5% bufferB for 18s, the sample
was applied and the column was washed for 3min with 5% bufferB.
A linear gradient of 10–45% bufferB over 44min was applied to elute
peptides, followed by 4.8min washing at 90% bufferB and 6min at
5% bufferB. Eluting rS1 and rL2 peptides were analysed for 58min in
positive mode using a data-dependent top-20 acquisition method. The
resolution for MS1 and MS2 were set to 120,000 and 30,000 full-width
at half-maximum, respectively, and automatic gain control (AGC) tar-
gets were set to 10
6
(MS1) and 10
5
(MS2). The MS1 scan range was set
to m/z=350–1,600. Precursors were fragmented using 28% normal-
ized, higher-energy collision-induced dissociation fragmentation.
Other analysis parameters were set as follows: isolation width, 1.6 m/z;
dynamic exclusion, 9s; maximum injection times for MS1 and MS2,
60ms and 120ms, respectively.
For all measurements, the lock mass option (m/z 445.120025) was
used for internal calibration.
Analysis of invitro RNAylated rS1 and rL2 MS data
MS data were analysed and validated manually using the OpenMS pipe-
line RNPxl and OpenMS TOPPASViewer
30
. Precursor mass tolerance was
set to 6ppm. MS/MS mass tolerance was set to 20ppm. A neutral loss
of 42.021798Da (C
1
H
2
N
2
) at Arg residues was defined, as well as adducts
of ribose minus H
2
O (78.010565Da, C
5
H
2
O), ADP-ribose (541.06111Da,
C
15
H
2
1N
5
O
13
P
2
) and ADPr without adenine (485.97295Da; C
10
H
17
O
16
P
3
)
31
.
Results were filtered for a 1% false discovery rate on peptide spectrum
match level. Ion chromatograms for rS1 peptides were extracted and
visualized using Skyline (v.21.2.0.369)59.
LC–MS/MS analysis of His-tagged rS1 isolated from T4-phage-
infected E. coli
LC–MS/MS analysis of protein digests was performed on an Explo-
ris 480 mass spectrometer connected to an electrospray ion source
(Thermo Fisher Scientific). Peptide separation was done using the
Ultimate 3000 nanoLC-system (Thermo Fisher Scientific), equipped
with a packed-in-house C18 resin column (Magic C18 AQ 2.4µm,
Dr.Maisch). The peptides were eluted from a precolumn in backflush
mode with a gradient from 98% solventA (0.15% formic acid) and 2%
solventB (99.85% ACN, 0.15% formic acid) to 35% solventB over 40min
and 90min, respectively. The flow rate was set to 300nlmin
−1
. The
data-dependent acquisition mode for label-free quantification was set
to obtain one high-resolution MS scan at a resolution of 60,000 (m/z
of 200) with scanning range from 350 to 1,650 m/z. MS/MS scans were
acquired for the 20 most-intense ions (90min gradient) and for the
most-intense ions detected within 2s (cycle 1s, 40min gradient). To
increase the efficiency of MS/MS attempts, the charged-state screen-
ing mode was adjusted to exclude unassigned and singly charged
ions. The ion accumulation time was set to 25ms for MS and ‘auto’ for
MS/MS scans. The AGC was set to 300% for MS survey scans and 200%
for MS/MS scans.
Raw MS spectra were analysed using MaxQuant (v.1.6.17.0 and 2.0.3.0)
using a fasta database of the targets proteins and a set of common
contaminant proteins. The following search parameters were used: full
tryptic specificity required (cleavage after Lys or Arg residues); three
missed cleavages allowed; carbamidomethylation (C) set as a fixed
modification; and oxidation (M; +16Da), deamidation (N, Q; +1Da) and
ADP-ribosylation (K; +541Da) set as variable modifications. MaxQuant
was executed in the default setting. All MaxQuant parameters are listed
in Supplementary Tables1 and 2. The MS proteomics data have been
deposited with the ProteomeXchange Consortium by the PRIDE partner
repository under the dataset identifier PXD041714.
Generation of E. coli B strain with endogenously His-tagged rS1
The E.coli B strain with endogenously His-tagged rS1 was created by
homologous recombination of linear transforming DNA (tDNA) using
the pRET/ET plasmid in the E.coli B strain. The linear tDNA was gen-
erated by fusion PCR aligning four fragments: 156 base pairs (bp) of
the rpsA gene with an additional His-tag amplified from the pET28
rS1 vector (serving as the left homologous flank), a 70-bp fragment of
the native rpsA terminator, the Flp-flanked kanamycin cassette from
pKD4 and 140bp of the 3 flanking region of the rpsA gene (the right
homologous flank). The primers used are indicated in Supplementary
Table9. The subsequent procedure for recombination is based on the
protocol for the E.coli Gene Deletion Kit by RET/ET Recombination
(Gene Bridges). In brief, E.coli B strain containing the pRED/ET plasmid
was grown in LB medium supplemented with 100 µgml
−1
ampicillin at
30°C. At OD
600
=0.35, L-arabinose was added to 0.33% (w/v) to induce
expression of the RED/ET recombination system during growth at 37°C
for 1h. Next, 1.4ml culture was collected by centrifugation at 3,000g at
4°C for 1min, and cells were washed twice with 1ml cold 10% glycerol
and finally resuspended in 50µl 10% glycerol. Cells were electroporated
with 1µg tDNA using a MicroPulser Electroporator (Bio-Rad) at standard
settings (Ec1). Electroporated cells were immediately resuspended in
1ml prewarmed LB medium and incubated at 37°C with shaking at 600
rpm for 3h. Finally, cells were plated on kanamycin (30µgml
−1
) LB–agar
plates. Cells took 2 days to recover and grow. Successful recombination
was evaluated by Sanger sequencing and correct protein expression
was validated by pull-down and proteomics.
RNAylomeSeq
Cultures (100ml) of E.coli B strain with endogenously His-tagged rS1
(Supplementary Table10) in LB medium supplied with 1mM CaCl2, 1mM
MgCl
2
and 30µgml
−1
kanamycin were grown at 37°C in 250ml baffled
Erlenmeyer flasks to an OD60 of around 0.8. T4 phage WT or T4 phage
ModB(R73A,G74A) were added to an MOI of 5.0. For the uninfected
negative control, the same volumes of LB medium were added to the
cultures. Cultures were then incubated at 37°C for 8min and E.coli
was collected by centrifugation at 3,000g for 13min. Dried pellets
were stored at −80°C.
Pellets from the 100ml culture infected with either WT T4 phage, T4
phage ModB(R73A,G74A) or the uninfected control (LB) were resus-
pended in 2ml Ni-NTA wash buffer (10mM imidazole, 50mM Tris-HCl,
pH 7.5, 1M NaCl, 1M urea, 5mM MgSO4, 5mM β-mercaptoethanol, 5%
glycerol, pH 8.0, EDTA-free protease inhibitor (Roche, one tablet per
500ml)) on ice and lysed by sonication (6min, 50% power, 0.5s pulse).
The lysate was cleared from the cell debris by centrifugation at 21,000g
at 4°C for 30min. Supernatant (1.9ml), 50µl Ni-NTA agarose beads
( Jena Bioscience, equilibrated in Ni-NTA wash buffer), 80U murine
Content courtesy of Springer Nature, terms of use apply. Rights reserved

RNase inhibitor (New England Biolabs) and 4.72µg rS1 D2 RNAylated
with NAD-capped RNAI were combined and incubated at 4°C in a rotary
mixer for 30min. Entire samples were transferred to Mobicol mini spin
columns (MoBiTec). Beads were washed four times with 200µl Ni-NTA
wash buffer and subsequently eight times with 200µl streptavidin wash
buffer (50mM Tris-HCl, pH7.5, 8M urea). Beads were equilibrated in
standard ligation buffer (10mM MgCl
2
, 50mM Tris-HCl, pH7.4) and
blocked with bovine serum albumine (BSA) before 3 RNA-adapter
ligation at 4°C overnight in the presence of standard ligation buffer,
50mM β-mercaptoethanol, 0.05µgµl
−1
BSA, 15% (v/v) DMSO, 5µM
adenylated RNA-3-adapter, 0.5Uµl−1 T4 RNL1 (New England Biolabs)
and 10Uµl
−1
T4 RNL2, tr. K227Q (New England Biolabs). Protein was
rebound by the addition of NaCl to 1.5M and incubation at 20°C, with
shaking at 400rpm for 20min. Beads were subsequently washed six
times with streptavidin wash buffer and equilibrated in first strand
buffer (50mM Tris-HCl, pH8.3, 3mM MgCl2, 75mM KCl) and blocked
with BSA. Reverse transcription of protein-bound RNA was done in a
30-µl scale for 1h at 40°C using 10Uµl−1 Superscript IV Reverse Tran-
scriptase (Invitrogen) in the presence of 5µM RT primer, first strand
buffer, 25mM β-mercaptoethanol, 0.05µgµl
−1
BSA and 0.5mM dNTPs.
After incubation, NaCl was added to 1.5M and the solution was incu-
bated at 20°C, with shaking at 400rpm for 1h to rebind RNA–cDNA
hybrids. Beads were subsequently washed five times with 0.25×strepta-
vidin wash buffer (2M urea, 50mM Tris-HCl, pH7.5), equilibrated in
ExoI buffer (10mM Tris-HCl, pH7.9, 5mM β-mercaptoethanol, 10mM
MgCl
2
, 50mM NaCl) and blocked with BSA. Residual RT primer was
removed by ExoI digest with 1Uµl−1 E.coli ExoI (New England Biolabs)
in ExoI buffer at 37°C for at least 30min. Finally, beads were washed
with 200µl 0.25×streptavidin wash buffer five times and subsequently
with 200µl immobilization buffer (10mM Na-HEPES, pH7.2, 1M NaCl)
three times. cDNA was eluted by incubation of beads in 100µl 150mM
NaOH at 55°C for 25min and by washing with 100µl MQ water. Eluate
pH was neutralized by the addition of 0.05volumes of 3M NaOAc,
pH5.5. cDNA was removed from the residual protein by phenol–chlo-
roform extraction and precipitated with 2.5volumes of ethanol in the
presence of 0.3M NaOAc, pH5.5 overnight. Precipitated cDNA was
C-tailed using 1Uµl−1 TdT (Thermo Fisher) in the presence of 1.25mM
CTP and 1×TdT buffer at 37°C for 30min and subsequently inactivated
at 70°C for 10min. 5µM cDNA anchor (hybridization of forward and
reverse anchor, Supplementary Table 9) was ligated to C-tailed cDNA
in standard ligation buffer in the presence of 10µM ATP and 1.5Uµl
−1
T4 DNA Ligase (Thermo Fisher Scientific) at 4°C overnight. Ligation
reactions were inactivated at 70°C for 10min and cDNA was ethanol
precipitated.
For the preparation of the Illumina RNAylomeSeq library, cDNA was
amplified by PCR using 2U Phusion Polymerase (Thermo Fisher Scien-
tific) in the presence of 5% (v/v) DMSO, 200µM dNTPs and 2,500nM
New England Biolabs Next Universal and Index Primer each (Primer Set
1, New England Biolabs). PCR products were purified by native PAGE
and ethanol-precipitated. The double-stranded DNA (dsDNA) concen-
tration was determined using a Quantus fluorometer (Promega) and
library size was determined with the Bioanalyzer (Agilent). Equimolar
amounts of each library were sequenced on a MiniSeq system (Illumina)
using the MiniSeq High-Output Kit (150 cycles, Illumina) generating
20million 151-bp single-end reads.
Analysis of next-generation sequencing data
Next-generation sequencing (NGS) data were demultiplexed using
bcl2fastq (v.2.20.0, Illumina). Fastq files were assessed using FastQC
(v.0.11.9) and Illumina sequencing adapters were trimmed from reads
using cutadapt (v.1.18). Reads were aligned to a reference genome com-
posed of an E.coli K12 (U00096.3), bacteriophage T4 (NC_000866.4)
and RNAI (our design) with hisat2 (v.2.2.1). Primary alignments
were selected using samtools (v.1.7) and reads per genomic feature
were counted with featureCounts (v.2.0.1 from Subread package).
The resulting counts table was subjected to further analysis and data
visualization in R (v.4.1.2). Read counts were normalized to the overall
number of mapped reads per sample and to the respective read counts
for the RNAI spike-in as follows:
normreadcount=
readcount×readcount(RNAI )
∑readcount
ij
ij j
iij
,
,
,
Data visualization was done with a custom R script
60
and alignments
were manually inspected in Integrative Genomics Viewer (IGV v.2.4.9).
Hits were identified based on the following criteria: log2-transformed
fold change (LFC)≥1.5 comparing WT T4 and the T4 R73A,G74A mutant
and log
2
-normalized mean expression among WT and R73A,G74A sam-
ple of one replicate≥−0.5.
Quantitative PCR validation of NGS data
cDNAs from RNAylomeSeq were diluted 1:30 in Millipore water. Quanti-
tative PCR was performed on 1µl diluted cDNA in 10µl scale in technical
duplicates amplifying regions of 100–150bp with the iTaq Universal
SYBR Green Supermix (Bio-Rad), according to the manufacturer’s
instructions, using the primers indicated in Supplementary Table9.
The log
2
-transformed difference in cycle-threshold values for WT T4
and T4 R73A,G74A infected samples from corresponding replicates was
computed and an LFC≥1 was set as a threshold for cDNA enrichment.
Ribosome RNAylation and proteomic analysis of RNAylated
proteins
70S ribosomes (4.3µgµl
−1
) were RNAylated in transferase buffer in
the presence of either 1µM NAD–10-mer–Cy5 or 1µM NAD–40-mer–
Cy5 (Supplementary Table7) by 0.05µgµl−1 ModB at 15°C for 90min.
RNAylated and non-RNAylated control samples were analysed using 12%
SDS–PAGE. To identify RNAylated proteins, SDS–PAGE-separated pro-
tein bands were excised and proteins were digested in gel as described
previously
61
. LC-MS was carried out on an Exploris 480 mass spectrom-
eter connected to an Ultimate 3000 RSLCnano system with a Proflow
upgrade and a nanospray flex ion source (all Thermo Scientific). Peptide
mixtures were then analysed on the LC-MS system described above
with a peptide-separating gradient of 30min from 2% to 35% bufferB.
Peptide separation was performed on a reverse-phase HPLC column
(75µm×42cm) packed in-house with C18 resin (2.4µm, Dr.Maisch).
Peptides were ionized at 2.3kV spray voltage with a heated capillary
temperature at 275°C and funnel RF level at 40. MS survey scans were
acquired with a resolution of 120.000 at m/z 200 and full MS AGC
target of 300% with a maximal IT of 50ms. The mass range was set
to 350–1,650. Fragment spectra were acquired in data-dependent
acquisition mode with a quadrupole isolation window of m/z=1.5, an
AGC target value of 200% and a resolution of 15.000, and fragmenta-
tion was induced with a normalized higher-energy collision-induced
dissociation collision energy of 27%. MS raw data were searched with
SEQUEST embedded in Proteome Discoverer 2.2 (Thermo Scientific)
against a Uniprot E.coli protein database containing the bacteriophage
T4 protein ModB. Spectral counts were exported from Scaffold Viewer
and total spectral counts per sample were used to normalize spec-
tral counts for all other proteins by division in Microsoft Excel 2016
followed by calculation of the ratio of normalized spectral counts from
modified and unmodified bands.
RNAylation of proteins in E. coli lysates
A fresh pellet from 40ml E.coli B strain culture at an OD600 of around
0.8 was resuspended in 2ml transferase buffer (10mM Mg(OAc)2,
22mM NH4Cl, 50mM Tris-acetate, pH7.5, 1 mM EDTA, 10mM
2-mercaptoethanol, 1% glycerol). Cells were lysed by sonication
(3×2min at 50% power, 0.5s pulse) and the lysate was cleared from
the cell debris by centrifugation at 27,670g at 4°C for 30min. The
supernatant was used in RNAylation assays.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Lysate (100µl) was incubated in the presence of 0.93µM NAD–
10-mer–Cy5 (0.47µM with reference to the NAD-capped) or 0.93µM
P–10-mer–Cy5 (Supplementary Table7), 0.37U murine RNase inhibitor
(New England Biolabs) and 0.69µM ModB at 15°C. Samples of 10µl were
taken before the addition of ModB and after 2, 5, 10, 20, 30 and 60min,
and were immediately resuspended in one volume of 2×Laemmli buffer.
Samples were analysed by 12% SDS–PAGE applying the same reference
(rS1 RNAylated with NAD–10-mer–Cy5) to each gel. The Cy5 signal was
recorded using the Cy5 blot option of the ChemiDoc Imaging System
at a manual exposure of 90s. Gels were then stained in Coomassie
solution and imaged with the same system.
E.coli lysates with various concentrations of ModB were processed
and analysed by proteomics as described previously38.
Determination of NAD concentrations in E. coli lysates
A dilution series of E.coli cell lysate was prepared in PBS. NAD was
diluted in PBS starting from a 100mM stock creating NAD solutions
of 1,000nM to 3.125nM. The NAD solutions, the lysate dilution series
and a PBS blank were assessed for their NAD concentrations using the
NAD/NADH-Glo Assay (Promega), according to the manufacturer’s
instructions in triplicates. Luminescence measurements were carried
out on a Tecan plate reader (Spark) in a 384-well flat white plate. A lin-
ear fit (R
2
=0.9836) was performed for NAD concentrations between
400nM and 4nM with a linear correlation to intensity. The equation
was used to calculate NAD concentrations for the E.coli lysate as the
mean of the technical triplicates.
Western blotting
Proteins were separated by 10% SDS–PAGE and gels were equilibrated
in transfer buffer (25mM Tris, pH8.3, 192mM glycine, 20% (v/v) metha-
nol). Polyvinylidene difluoride membranes with a pore size of 0.2µm
(GE Healthcare) were activated in methanol for 1min and equilibrated in
transfer buffer. Proteins were transferred from gels to the membranes
in a semi-dry manner at 300mA for 1.5h, unless indicated differently.
After the transfer, membranes were dehydrated by soaking in metha-
nol and washed twice with TBS-Tween (TBS-T; 10mM Tris-HCl, pH7.5,
150mM NaCl, 0.05% (v/v) Tween20). Afterwards, 10ml blocking buffer
(5% (w/v) milk powder in TBS-T) were added to the membranes and incu-
bated at room temperature for 1h. To detect ADP-ribosylated proteins,
membranes were incubated with a 1:10,000 dilution of anti-pan-ADPr
binding reagent MABE1016 (Merck) in 10ml washing buffer (1% (w/v)
milk powder in TBS-T) at 4°C overnight
62
. Membranes were washed
and incubated with 10ml of a 1:10,000 dilution of the horseradish
peroxidase–goat-anti-rabbitIgG secondary antibody (Advansta) in
washing buffer at room temperature for 1h. Afterwards, membranes
were washed with PBS. ADP-ribosylated proteins were visualized by
chemiluminescence using the SignalFire ECL Reagent or the Signal-
Fire Elite ECL Reagent (Cell Signaling Technology), according to the
manufacturer’s instructions.
If proteins in SDS–PAGE gels needed to be visualized before blotting,
a TCE staining method53 was used. Resolving gels were supplemented
with 0.5% (v/v) TCE. For visualization, gels were activated by ultraviolet
transillumination (with a wavelength of 300nm) for 60s. Proteins then
showed fluorescence in the visible spectrum.
Quantification of RNAylation
rS1 proteins were isolated from E.coli strain B pTAC rS1 bacteria
(Supplementary Table10) that were either uninfected or infected with
bacteriophage T4. rS1 (1.5µM) was treated with 1µM ARH1 in the pres-
ence of 12.5mM Tris-HCl, pH7.5, 25mM NaCl, 25mM KCl and 5mM
MgCl
2
. Alternatively, rS1 (1.5µM) was treated with 0.5U endonuclease
P1 in 100mM Mg(OAc)
2
, 220mM NH
4
Cl, 500mM HEPES, pH7.5, 10mM
EDTA, 100mM β-mercaptoethanol and 10% glycerol. Digests were incu-
bated at 37°C for 2h. Afterwards, digests were precipitated by the
addition of nine volumes of ethanol and precipitated by centrifugation
(14,000pm) at 4°C for 1h. Protein pellets were resuspended in 10µl
1×Laemmli buffer and analysed by Western blotting. ADPr modifica-
tions were detected by the primary antibody MABE1016 (Merck) as
described above. The pan-ADPr signals for ADP-ribosylated rS1 were
normalized to the corresponding band intensities in the TCE stain.
Normalized intensities for untreated rS1 were then divided by the
intensity for P1-treated rS1 to yield the fractions of ADP-ribosylated
and RNAylated rS1 for the two modifications.
Phage mutagenesis
The CRISPR–Cas9 spacer plasmids were generated by introducing the
modB spacer sequence into the DS-SPCas plasmid (Addgene, 48645)
(Supplementary Table10). The modB-carrying vector pET28_ModB
was used as a donor DNA for homologous recombination in CRISPR–
Cas9-mediated mutagenesis. The pET28_ModB plasmid was modified
by site-directed mutagenesis, during which point mutations R73A and
G74A were exposed to modB. The R73A mutation led to the inactivation
of ADP-ribosyltransferase activity. The G74A mutation was located in
the protospacer adjacent motif and was required to prevent the cleav-
age of donor DNA by Cas9 nuclease. The applied primers are listed in
Supplementary Table9. The resulting plasmids were sequenced and
transformed into E.coli BL21 (DE3).
The CRISPR–Cas9-mediated mutagenesis was based on previous
work33. The DS_SPCas_ModB plasmid with the target spacer sequence
and the donor plasmid pET28a_ModB_R73A/G74A were co-transformed
into E.coli DH5α. The cells were further infected by bacteriophage T4
(1:10,000 phages:cells), and the plaque assay was done. The plates were
incubated overnight at 37°C and the resulting plaques were screened
for mutants. Single plaques were picked by sterile pipet tips and trans-
ferred into 200µl Pi–Mg buffer (26mM Na
2
HPO
4
, 68mM NaCl, 22mM
KH
2
PO
4
, 1mM MgSO
4
, pH7.5) supplied with 2µl CCl
3
H. The samples
were incubated at room temperature for 1h. Next, 2µl of the sample
was transferred to a new PCR tube and heated to 95°C for 10min. The
sample was further used for DNA amplification using PCR (primers used
are listed in Supplementary Table9). The amplified DNA was purified
by agarose gel electrophoresis and submitted for Sanger sequencing.
Plaque assay
The E.coli culture of interest was grown to an OD600 of around 0.8–1.0.
Next, 300µl of the culture was infected with 100µl of WT T4 phage
or T4 ModB(R73A,G74A) (Supplementary Table10) mutant, with
either defined or unknown MOI. The bacteria–phage suspension was
incubated at 37°C for 7min and subsequently transferred to 4ml LB
soft agar (0.75%), mixed and poured onto an LB-agar plate (1.5% LB
agar). The plates were incubated at 37°C overnight and validated the
following day.
Time course of T4-mediated lysis of E. coli
LB medium (100ml in 500-ml baffled flasks) was inoculated with E.coliB
culture overnight to OD
600
=0.1 and was then incubated at 37°C with
shaking at 180rpm until OD600=0.8 was reached. The culture was
cooled to room temperature and infected by either WT T4 phages or
T4 ModB(R73A,G74A) mutants (Supplementary Table10) to an MOI of
5. The culture was further incubated at room temperature with shaking
at 150rpm. Cell lysis was tracked by measuring the OD600 at different
times of infection (0–200min after infection). The experiment was
run in biological triplicates.
Burst-size determination
LB medium (100ml in 500-ml baffled flasks) was inoculated with
E.coliB culture overnight to OD600=0.1 and was then incubated at 37°C
with shaking at 180rpm until OD
600
=0.8 was reached, as above. The
culture was infected either by WT T4 phages or T4 ModB(R73A,G74A)
mutant (Supplementary Table10) to an MOI of 0.01 and further incu-
bated at 37°C without shaking.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

To determine the total number of infective centres, T0 (comprising
unadsorbed and already adsorbed phages), at 5min after infection,
100µl of infected culture was used to reinfect 300µl E.coli B cells
(OD600=1.0) with a subsequent plaque assay. The number of unad-
sorbed phages (U) was determined by transferring 1ml infected culture
to 50µl CCl3H. In this way, E.coli cells were disrupted, after which the
unadsorbed phages remained intact and were used for plaque assay.
T
0
−U, represents the number of initially infected centres. The number
of unadsorbed phages (U
xmin
) was continuously traced during infection
and used to calculate the number of T4-phage progeny (T4-phage prog-
eny=Uxmin/(T0−U5min). The time point at which the first increase in phage
number was observed was treated as the first burst time point and was
used to calculate the phage burst size (burst size=Uburst1/(T0−U5min)).
Data were plotted using OriginPro 2020b software. Error bars rep-
resent s.d. of the means for three biological replicates. For selected
time points, statistical tests were done as two-sided t-tests in R (v.4.2.2)
implemented in the ggpubr package (v.0.6.0) using a significance level
of 0.05.
Phage adsorption kinetics
LB medium (100ml in 500-ml baffled flasks) was inoculated with E.coli
B culture overnight to an OD
600
=0.1 and incubated at 37°C with shak-
ing at 180rpm until OD
600
=0.8 was reached, as above. The culture was
cooled to room temperature and infected by either WT T4 phages or
T4 ModB(R73A,G74A) mutants (Supplementary Table10) to an MOI
of 0.1. Immediately after infection, 100µl of the culture was used to
determine the number of total infective centres, T0, by plaque assay.
Then 100µl of the culture was taken at different time points of infec-
tion (0–25min after infection) and 5µl CCl3H was added to disrupt
E.coli cells. This suspension was subsequently used to determine the
number of unadsorbed phages (U
xmin
) by plaque assay. The calcula-
tion of the adsorption rate was performed as follows: adsorption
rate(%)=100%−(Uxmin/T×100%).
Data were plotted using OriginPro 2020b software. Error bars rep-
resent s.d. of the means for three biological replicates. For selected
time points, statistical tests were done as two-sided t-tests in R (v.4.2.2)
implemented in the ggpubr package (v.0.6.0) using a significance level
of 0.05.
Reporting summary
Further information on research design is available in theNature Port-
folio Reporting Summary linked to this article.
Data availability
The datasets generated and/or analysed during the current study are
available from the corresponding author on reasonable request. NGS
data are accessible via GEO record GSE214431. LC–MS/MS raw data for
the measurements of rS1 ADP-ribosylation invivo, in-gel digest and
estimation of ModB abundance have been deposited in PRIDE with the
accession code PXD041714. LC–MS/MS raw data for measurements of
invitro ADP-ribosylated and RNAylated rS1 and rL2 have been deposited
in PRIDE with the accession code PXD038910. Reference genomes for
E.coli (U00096.3) and T4 phage (NC_000866.4) were retrieved from
NCBI. Protein structures (2MFI, 2MFL, 2KHI, 5XQ5, 2KHJ, 7K00 and
6H4N) were downloaded from PDB using the indicated accession code
(https://www.rcsb.org/). E.coli K12 pan proteome (UP000000625) and
selected protein sequences were retrieved from Uniprot (https://www.
uniprot.org/).Supplementary information is available, including raw
gel and blot images.Source data are provided with this paper.
Code availability
The custom R code used to analyse the RNAylomeSeq data is publicly
available on Zenodo60.
52. Höfer, K., Abele, F., Schlotthauer, J. & Jäschke, A. Synthesis of 5-NAD-capped RNA.
Bioconjug. Chem. 27, 874–877 (2016).
53. Ladner, C. L., Yang, J., Turner, R. J. & Edwards, R. A. Visible luorescent detection of
proteins in polyacrylamide gels without staining. Anal. Biochem. 326, 13–20 (2004).
54. Hsia, J. A. etal. Amino acid-speciic ADP-ribosylation. Sensitivity to hydroxylamine of
[cysteine(ADP-ribose)]protein and [arginine(ADP-ribose)]protein linkages. J. Biol. Chem.
260, 16187–16191 (1985).
55. McDonald, L . J., Wainschel, L. A., Oppenheimer, N. J. & Moss, J. Amino acid-speciic
ADP-ribosylation: structural characterization and chemical differentiation of ADP-ribose-
cysteine adducts formed nonenzymatically and in a pertussis toxin-catalyzed reaction.
Biochemistry 31, 11881–11887 (1992).
56. Silberklang, M., Gillum, A. M. & Rajbhandary, U. L. The use of nuclease P1 in sequence
analysis of end group labeled RNA. Nucleic Acids Res. 4, 4091–4108 (1977).
57. Abplanalp, J. etal. Proteomic analyses identify ARH3 as a serine mono-ADP-ribosylhydrolase.
Nat. Commun. 8, 2055 (2017).
58. Banasik, M., Komura, H., Shimoyama, M. & Ueda, K. Speciic inhibitors of poly(ADP-ribose)
synthetase and mono(ADP-ribosyl)transferase. J. Biol. Chem. 267, 1569–1575 (1992).
59. MacLean, B. etal. Skyline: an open source document editor for creating and analyzing
targeted proteomics experiments. Bioinformatics 26, 966–968 (2010).
60. Wolfram-Schauerte, M. etal. RNAylomeSeq data analysis from “A viral ADP-ribosyltransferase
attaches RNA chains to host proteins” (v.1.0). Zenodo https://doi.org/10.5281/zenodo.
7977386 (2023).
61. Dell’Aquila, G. etal. Mobilization and cellular distribution of phosphate in the diatom
Phaeodactylum tricornutum. Front. Plant Sci. 11, 579 (2020).
62. Gibson, B. A., Conrad, L . B., Huang, D. & Kraus, W. L. Generation and characterization of
recombinant antibody-like ADP-ribose binding proteins. Biochemistry 56, 6305–6316
(2017).
63. Jumper, J. etal. Highly accurate protein structure prediction with AlphaFold. Nature 596,
583–589 (2021).
64. Qureshi, N. S., Bains, J. K., Sreeramulu, S., Schwalbe, H. & Fürtig, B. Conformational
switch in the ribosomal protein S1 guides unfolding of structured RNAs for translation
initiation. Nucleic Acids Res. 46, 10917–10929 (2018).
65. Di Tommaso, P. etal. T–Coffee: a web server for the multiple sequence alignment of
protein and RNA sequences using structural information and homology extension.
Nucleic Acids Res. 39, W13–W17 (2011).
66. Watson, Z. L. etal. Structure of the bacterial ribosome at 2 Å resolution. eLife 9, e60482
(2020).
Acknowledgements We thank N. Beumer, J. Hoff, S. J. Keding, J. Kahnt, J. Koch, N. Lichti,
P. Mann, N. Moskalchuk, M. Raabe, E. Tamerler, M. Viering and M. Weber for experimental
assistance. This project has received funding from the European Research Council under
the European Union’s Horizon 2020 research and innovation programme (grant 882789
RNACoenzyme, to A.J.) and from the German Research Council (DFG; project 439669440,
TRR319, project A02, to A.J.). M.W.-S. is supported by the Studienstiftung des Deutschen
Volkes and the Joachim Herz Stiftung. K.H. is supported by the Max Planck Society, Baden-
Württemberg Stiftung, Carl-Zeiss-Stiftung and the German Research Council (grant
DFG-SPP2330). H.U. is supported by the Max Planck Institute for Multidisciplinary Sciences
and by the German Research Council (grants DFG-SPP1935, DFG-SFB1286 and DFG-SFB1565
(project number 469281184)).
Author contributions K.H. and A.J. designed the study. K.H., M.W.-S., J.G., F.A.B. and N.P.
cloned, expressed, puriied and analysed the ARTs and their target proteins. K.H., I.S., L.M.W.,
A.W. and M.W. prepared samples for mass spectrometry. I.S., L.M.W., A.W. and H.U. developed
an LC–MS/MS pipeline to study ADP-ribosylation and RNAylation, and analysed the data. T.G.
performed mass spectrometry analysis of rS1. M.W.-S. developed the RNAylomeSeq pipeline
and analysed the data. N.P. created and characterized the ModB mutant phage. K.H., H.U. and
A.J. supervised the work. K.H., M.W.-S. and A.J. wrote the irst draft, and all authors contributed
to reviewing, editing and providing additional text for the manuscript.
Funding Open access funding provided by Max Planck Society.
Competing interests TheMax Planck Society and Heidelberg University are in the process of
applying for a patent(PCT/EP2021/071295)covering the RNAylation that lists K.H.and A.J.as
inventors. The remaining authors declare no competing interests.
Additional information
Supplementary information The online version contains supplementary material available at
https://doi.org/10.1038/s41586-023-06429-2.
Correspondence and requests for materials should be addressed to Andres Jäschke or
Katharina Höfer.
Peer review information Nature thanks the anonymous reviewers for their contribution to the
peer review of this work.
Reprints and permissions information is available at http://www.nature.com/reprints.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 1 | Se e next page for capti on.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 1 | AD P-ribosylat ion and RNAylat ion by T4 ARTs.
a, Characte risation of R NAylation of the RN A polymeras e (RNAP) by the ARTs
Alt or Mod A in the presenc e of NAD-10mer-Cy5 (1), additi onal 10 equivalen ts of
NAD (2) or in the ab sence of the res pective ART (3) (n=3). rS1 RNAylat ed with
NAD-10mer-Cy5 by M odB serve s as a reference (4). The RNA P is a well-esta blished
target prot ein of Alt and Mod A and was thus cho sen to asses s RNAylation by
Alt and Mod A. Alt sligh tly RNAylates th e RNAP invitro which i s abolished in th e
presenc e of 10 equivalents o f NAD relative to NA D-10mer-Cy5. Prot ein load is
visualis ed by Coomassi e staining and R NAylated protein is v isualised in th e
fluorescent Cy5 channel. b,c, Time course a nalysis of the Mod B-mediated
RNAylation ( b) and ADP-ribosyl ation (c) of rS1 analysed by S DS-PAGE (n=3
each). RNAylate d or ADP-ribosy lated protein is v isualised by rad ioactivit y scan
and protein lo ad confirm ed by Coomassi e staining. d, Ne gative controls fo r
RNAylation o f rS1 with ModB an alysed by SDS-PAGE. R NAylation assay w as
perform ed in the presen ce of 32P-RNA lack ing the NAD- cap (upper panel) an d in
the absen ce of either rS1 (- rS1) (middle p anel) or ModB (-ModB) (lowe r panel)
(n=3). In these exper iments, no RN Aylation was dete cted in the radi oactive
scan of the S DS-PAGE gel.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 2 | Se e next page for capti on.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 2 | Char acteris ation of the RN Aylation of protei n rS1
by ModB. a, RNAylation of rS1 in the pr esence of cat alytically a ctive ModB and
cataly tically inac tive ModB R73A , G74A with NAD-10mer-Cy5 (n=3). In additi on
to the cata lytically imp ortant res idue R73, we mutat ed G74 as well. Mutatio n
of G74 results in an alt ered PAM region, whi ch is importa nt for CRISPR-Ca s9
gene editin g of the T4 phage genome. b, AlphaFold prediction63 of th e structure
of ModB. Ac tive site residue s of the R-S-EXE motif 1 are highlig hted in red.
Corresp onding confi dence metric s are shown in Supplem entary Fig .2.
c, Inhibitio n of invitro R NAylation of prote in rS1 by ModB via A RT inhibitor 3-M B.
Reactio ns were perform ed with 32P-NAD -RNA 8mer (32P-N AD-8mer) as well a s
32P-RNA 8me r (negative contro l) (n=3). 3-MB reduces the y ield of RNAylate d rS1.
d, invitro di gest of RNAylate d and ADP-ribos ylated protein r S1 by RNase T1 .
Reactio ns performe d in the absence o f RNase T1 (-) serve as ne gative controls .
Protein rS1 A DP-ribosylat ed in the presen ce of 32P-NAD was appli ed as a
refere nce (S1-ADPr) (n=2). Upon T1 dige st, the 100nt-RNA at r S1 is shortene d,
and the molecular weig ht of RNAylated rS1 i s reduced. Thi s leads to similar
elec trophor etic mobil ity as for ADP-rib osylated rS1 . e, Tryptic digest of A DP-
ribosyla ted and RNAylate d protein rS1 (n=2). The prote in is degraded i n the
presenc e of trypsin, an d RNAylation and A DP-ribosylat ion signals are lo st. All
samples we re analysed by 12 % S DS-PAGE, protein was vi sualised by Coo massie
staining a nd RNAylation was a ssessed v ia a radioact ivity scan .
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 3 | Char acteris ation of ModB m ediated R NAylation
invitro. a, Analysis o f the role of RNA se condary str ucture on RNAyla tion
reaction. Four different 3′-Cy5-labelled NA D (NppA)-capp ed RNAs were te sted
including a lin ear (green) NA D-capp ed RNA (10mer) and thre e structure d RNAs
with eith er a 3′-overhang (blue), a 5′-overhang (red) or a blunt e nd (black) (n=3).
SDS-PAGE analyse s of the time cour se of RNAylation are s hown. Quantif ication
of the signa l intensitie s (Cy5 scan) relative to the re ference is shown in Fi g.2c.
ModB prefers l inear 5′-ends of NAD -capped R NAs. L = ladde r. b, Analysis of the
RNAylation d ependency o n the presence o f a 5′-NAD-cap of the R NA. 10 %
SDS-PAGE analysis of invitro RNAylat ion of the protein r S1 by ModB in the
presence of either 5′-NAD-capped (NAD-RNA), 5′-monophosphate- (5′-P-RNA)
or 5′-triphosphate-100nt-RNA (5′-PPP-RNA) (n=2). RNAylati on with radiola belled
RNA is det ected by radio activity s can and protein l oad visualise d by Coomassie
staining. Invitro RNAyla tion of rS1 is only o bserved in th e presence of NA D-RNA .
RNAylated r S1 cannot be det ected by Coom assie staini ng due to low sensit ivity.
c, Characte risation of A DPr-RNA (which is lack ing the nicotina mide moiety
compared to N AD-RNA) a s a substrate for Mo dB (n=2). As a positive contr ol,
NAD-8m er was applied. A ll reaction s were analysed by 12 % S DS-PAGE.
RNAylation with radiolabelled RNA is detected by radioactivity scan and
protein load v isualised by Co omassie sta ining. ADPr-RN A is not accepte d as a
substrat e for ModB-cat alysed RNAylat ion invitro.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 4 | Se e next page for capti on.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 4 | Spe cific re moval of RNAylation u sing chemi cal and
enzymatic treatments. a, Different ADP-rib ose-protein lin kages have been
shown to be ei ther stable or un stable in the pre sence of HgCl 2 and neutral
hydroxylamine (NH2OH), which repre sents a relat ively straight forward and fast
approach to id entify ADP-r ibosylation s ites. Treatment wi th NH2OH hydrolyses
linkages b etween glut amate/aspart ate and ADP-ribo se. HgCl2 specifically
cleave s thiol-glycos idic bonds. A DP-ribosylate d and RNAylated pro tein rS1
were treate d with NH2OH or Hg Cl2. The removal of AD Pr or RNA by thes e
chemi cals would re sult in a decrea se of the radioac tive signal of prot ein rS1.
All sample s were analysed by 1 2 % SDS-PAGE, staine d in Coomassie ( protein
loading con trol) and RNAylati on assesse d as radioact ivity. A decreas e of the
radioactive signal in comparison to the control (untreated) was not determined
(n=1). b, invitro time cour se of the stabili ty of rS1 ADP-ri bosylation in t he
presenc e of ARH3 analys ed by 12 % SDS-PAGE (n=1). The auto radiography sc an
is present ed. ARH3 did n ot remove the ADP-rib osylation. c-d, Re actio n
schem atics for t he removal of the AD P-ribosylatio n (c) and RNAylation (d) of
rS1 by ARH1 .
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 5 | Ion ch romatogra ms of unmodif ied rS1 an d invitro
RNAylated r S1 extracted fr om LC-MS/MS dat a. Extracted ion chromatograms
(XICs) for tripl y and quadruply cha rged precurso r ions (monoisotop ic masses
1115 .8096 and 837.1090, respe ctively). XICs were e xtracted us ing Skyline 59, an
open sourc e document edi tor for creating and a nalysing targe ted proteomic s
experime nts. The ma sses corres pond to an rS1 pep tide AFLPGS LVDVRPVRDTL
HLEGK with an a ttached A DPr-cytidine . Recombinant S1 d omain 2 was invitro
incubate d with ModB and on e of the following com ponents: a, no o ther
supplements, b, uncappe d RNA-8mer, c, NAD-R NA-8mer, d, NAD-R NA-8mer
treated w ith RNase A and T 1 (results in ADP r-cytidine addu cts). An elution pe ak
at 42.3min i s observab le in d and corresp onds to the pept ide modifie d with
ADPr-cyto sine. Spurio us intensitie s can be obser ved in c and might re present
a degradation product. a and b show only background/contaminant peaks. A
contamin ant peak at 40mi n can be also obs erved in d (consider t he difference
in the inten sity scale be tween d and a-c).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 6 | Se e next page for capti on.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 6 | Invivo characterisation of the RNAylation by Western
blot and RNAylomeSeq. a, Analysis of t he substrate sp ecificit y of the pan-
ADPr antibody. Invitro p repared ADP-ri bosylated or R NAylated protein r S1
was applied t o evaluate the spec ificity o f the antibody (n=3). The pa n-ADPr
antibod y detected A DP-ribosylat ed proteins rS1 an d ModB (lane 1). In con trast,
RNAylated r S1 is not detec ted by pan-ADPr (la ne 2). However, a signal for ADP-
ribosyla ted ModB was ob served due to s elf-ADP-ribosylati on in its express ion
host E. coli (la ne 2). b, Quantifica tion of RNAylatio n using the combi nation of
nucleas e P1 digest and det ection of prot ein-linked ADP-rib ose by Western bl ot.
Visualis ation of protein loa d by TCE stain. Removal of t he ADP-ribos e signal by
ARH1 tre atment. pan-A DPr signals for A DP-ribosylat ed rS1 were normal ised to
corresp onding band inte nsities in the TCE s tain. Normal ised intensi ties for
untreate d rS1 were then div ided by the intens ity for P1-treate d rS1 to yield the
fracti ons of ADP-ribo sylated and RN Aylated rS1 among t he two modif ications.
The corre sponding dot pl ot is shown in Fig.4b (n=3 biologi cally indepe ndent
rep lica tes). c, Sc hematic illustr ation of the RNAylo meSeq protoc ol:
Ident ific ation of R NAylated RNA s which are covalently a ttached to r S1 invivo.
Brief ly, endogenously His-t agged rS1 is i solated from T4 phage infec ted E. coli
with Ni-NTA bead s. A spike-in - rS1 dom ain 2 RNAylated wi th NAD-RN AI -
(RNAI spi ke-in) is added to the lys ate which is mean t to be enriched v ia the
RNAylomeS eq workf low. rS1 captured on Ni-N TA beads is intensi vely washed
with 8 M urea in o rder to remove RNA non -covalently bou nd to rS1. Similar to
NAD captureSeq32, an RNA 3′-adapter is ligate d to covalently linke d RNAs and
RNA is revers e transcrib ed “on-bead”. cDNA is t hen eluted by alka line digest of
RNA and an ad ditional adapt er is ligated to th e 3′-terminus of the cDN A. cDNA is
amplifi ed by PCR and sequ enced by next-genera tion single-e nd sequencin g
(Illumina). Impor tantly, the RNA I spike-in is not mea nt to be enriche d in any
sample but ra ther to be found in ea ch sample in simila r amounts. T hereby, read
counts ca n be normalise d to the RNAI co unts in each sa mple allowing for the ir
comparison. d, MA plot showin g RNAs enric hed in the T4 phage WT infec ted
sample com pared to T4 phage ModB R73A , G74A control identif ied by
RNAylomeS eq for replicate 2 (tot al of n=3 biologic al replicates). Read co unts
per sample h ave been normali sed to RNAI s pike-in read count s which serve s as
an enrichm ent control for eac h sample. Thu s, RNAI is not fou nd enriched
comparin g T4 WT and T4 ModB R73A, G74A. Mean expre ssion values ( T4 WT
and T4 ModB R73A, G74A condi tion) have been norm alised by Log2 (x-axis) for
each replic ate separate ly. T4 WT and T4 ModB R73A , G74A read counts were
compared v ia log2 fold change ( y-axis). Read coverage on id entified R NAylated
RNAs as a nalysed in IGV is exem plarily shown for acpP in t he lower panel
depicti ng reads in T4 WT sample s (green) vs. T4 ModB R7 3A, G74A samples
(red). RNAylomeS eq merely iden tifies 5’-termini of m RNAs or, if 200nt or
smaller, entire sR NA sequence s. This is due to t he application o f single-end
Illumina-Se q which automat ically only capt ures the 5’-end of the resp ective
read/transcript. e, Select ed hits of RNA s identif ied by RNAylomeS eq comparin g
T4 phage WT and T4 ModB R73A, G 74A. acpP was identif ied in all three re plicates.
However, some trans cripts were only de tected in one o r two replicate s.
Enrichme nts have been fur ther validate d on cDNA level by qPCR . +: enriched; −:
not enrich ed; (+): enriched , but Log2 fold chan ge <= 1; n.d.: not def ined.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 7 | RN Aylation of rS1 doma ins D1 – D6 and S1 moti f of
PNPase by Mo dB invitro. a, Illustrat ion of the rS1-motif s of rS1 based on cr ystal
struct ures (PDB) of domains 1 (2 MFI), 2 (2MFL), 4 (2KHI), 5 (5XQ5) and
6 (2KHJ) as well a s an NMR struc ture of domain 364. b, invitro RNAylatio n of S1
domains and PN Pase by ModB us ing a 32P-NAD-8 mer. ModB and S1 domains
(D1-6) are marked with bla ck arrows. RNAyla ted rS1 domains , characteri sed
by a shift comp ared to the non-mo dified prote ins, are highligh ted with red
arrows. n=2 of bio logically ind ependent rep licates. Rea ctions were ana lysed
by 16 % Tricine-SDS-PAGE, sta ined in Coomas sie and RNAylatio n recorded by
autoradiography imaging (Radioactivity). c, Local alignme nt of rS1 D2 and
D6 as well as the S 1 domain of PNPase u sing T-coffee expres so65. R139 of D2
(highligh ted with an arrow) is c onserved in PN Pase and D6.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extended Data Fig. 8 | Characterisation and identification of RNAylation
target pr oteins of Mod B. a, Analysis of the invitro RNAylation o f rS1 domain 2
and its mut ants R139A and R 139K by 16 % Tricine-SDS-PAGE. A n inactive Nud C
mutant (Nu dC*; V157A, E174A, E1 77A, E178A) was used a s a negative cont rol
(n=3). Radioactiv ity indicate s RNAylation, Co omassie sc an visualise s protein
load. b, Quantific ation of relative in tensities of R NAylation of rS1 do main 2
and its mut ants R139A and R 139K based o n radioactiv ity in 16 % Tricine-SDS-
PAGE analysis. Per re plicate, inten sities were nor malised to the r S1 D2 WT
band intensity. A two-sided t-test wa s performed at p signif. < 0.05 indicating
signi fican tly decre ased RNAylati on of R139 mut ants of rS1 domai n 2 (p-value=
0.0003 (W T vs. R139A) and 0.0074 (W T vs. R139K)). n=3 of biolo gically
independent replicates. c, RNAylation of E. coli cell l ysate by ModB usi ng
3′-Cy5-labell ed NAD-R NA (schematic ally shown in upper p anel). A time cours e
of E. coli cell lys ate RNAylation by Mo dB in the presen ce of either a
5′-monophosphor ylated RNA 10 mer (P-10mer-Cy5, middle pan el) or 5′-NAD-
capped R NA 10mer (NAD -10mer-Cy5, lower panel), eac h with a 3′-fluorescen t
(Cy5) label. rS1 R NAylated with a n NAD-10mer-Cy5 is app lied as a referenc e
(ref). The ti me course of lys ate RNAylation wa s analysed by 12 % S DS-PAGE,
protein vis ualised by Cooma ssie staining a nd RNAylation reco rded via
fluore scence (Cy5). n=3 of biolo gically indep endent replica tes. NAD
concentration in the lysates exceeds the utilised NAD-10mer concentration
by 48-fold. NA D concentrat ion in the lysate s of 22.5 µ M (n=1 biologically
indepen dent replicate s, n=3 technical re plicates) was de termined usin g the
NAD/NA DH-Glo ass ay (Promega).The sch ematic protein an d tube in c were
created us ing BioRender (https://biorender.com).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 9 | Se e next page for capti on.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 9 | Char acteris ation of the spe cifici ty of ModB- mediated
RNAylation in E. coli lysates. a, RNAylation of E . coli cell lysat e in
the presen ce of ModB WT or i nactive ModB R 73A, G73A or the abs ence of ModB
using 3′-Cy5-la belled NAD -10mer or P-10mer. Time point 0 show s lysate before
addition of M odB, 60min shows R NAylation aft er 60min incubat ion with ModB .
Reactio ns were analysed by 1 2 % SDS-PAGE, protein v isualised by Co omassie
staining a nd RNAylation rec orded via f luorescence (Cy5 ). n=2 of biologically
independent replicates. b, Samples from lysate R NAylation with Cy 5-labelled
5′-NAD- or 5′-P-10mer (as presen ted in Extend ed Data Fig.8c) before ad dition of
ModB (0min) and af ter 60min incubat ion in the presen ce of ModB (60min)
were analyse d by 10 % SDS-PAGE and RNAylatio n monitored by f luorescenc e
(Cy5, here shown in re d). Subsequen tly, Western blottin g was performe d and
ADP-ribo sylation was de tected usin g pan-ADPr bindin g reagent (MAB E1016,
shown in grays cale). n=2 of biologic ally independ ent replicate s. Different
band patte rns were obser ved for ModB-me diated RNAylati on and ADP-
ribosylation in E. coli lysat es indicatin g a distinct t arget specif icity of Mod B for
RNAylation . NAD concen tration in the lys ates exceeds ut ilised NAD -10mer
concentr ation by 48-fold. NA D concentra tion in the lysat es of 22. 5 µM (n=1
biologic ally independ ent replicate s, n=3 technical re plicates) was det ermined
using the NA D/NADH- Glo assay (Prom ega). c, Lysate RNAylation by Mo dB in the
presenc e of various molar exce sses of NAD over N AD-10mer-Cy5 rang ing from
48-fold (native lys ate) to 1000-fold via ad ditional spike-in N AD (n=2). Cy5
represen ts RNAylation . TCE stain indicat es protein load, w hich is enabled by
binding of tr ichloroethan ol in the gel to try ptophan residu es in proteins whi ch
enhance s their fluor escence und er UV light and th ereby enables th eir
detection53. 700-fold mo lar excess of NAD re duces RNAylat ion to 67 % (n=2
biologic ally independ ent replicate s) compared to “native ” lysate. Total Cy5
signals for e ach lane were quanti fied to dete rmine and comp are RNAylation
levels. d, Lysate RNAylation and AD P-ribosylati on in the presenc e of various
ModB conc entrations (8 50, 85 and 8.5 nM) m onitored by f luorescenc e (Cy5 for
RNAylation) and We stern blot (pan-A DPr for ADP-ribo sylation). TCE stain
indicate s protein load. In aver age, RNAylation i s reduced to 8.6 % an d ADP-
ribosyla tion is reduce d to 6.9 % in lysates wit h ModB concen trations that
approximate t he cellular condit ions (85 nM). Total Cy5 or pan-AD Pr signals
(excluding ModB A DP-ribosylati on signal) for each l ane were quantif ied to
determine and compare RNAylation or ADP-ribosylation levels, respectively.
n=2 biologic ally independ ent replicate s.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extende d Data Fig. 10 | Se e next page for captio n.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extende d Data Fig. 10 | Sco pe of ModB RN Aylation target s in E. coli.
a, RNAylation o f E. coli riboso mes by ModB. R NAylated protein is s hifted upon
incubatio n with NAD- 40mer compa red to NAD-10mer w hich itself inc reases
protein weig ht by approx. 3kDa. Rel ative enrichme nt of RNAylated t arget
protein was a ssessed by su bjecting R NAylated protein ba nds and respec tive
control ban ds generated in th e absence of R NA to in-gel diges t and LC-MS/MS
analysis (n=2). b, Plot of the en richment of fr actional spe ctral count s for 50S
ribosom al protein L2 (rL2) ba sed on in-gel-dige st and LC-MS/MS a nalysis
presente d in a. Enrichme nt is calculate d for RNAylation wit h NAD-10mer
(A/C) or NAD -40mer (B/ D), relative to the respec tive, non-RN Aylated control
bands bas ed on spectr al counts from S caffold (n=2). c, Analysis of th e invitro
RNAylation o f rL2 by ModB in the pre sence of NA D-8mer. RNAylated rL 2
proteins have re duced elect rophoretic mob ility during SD S-PAGE. Protein was
visualis ed by fluores cent protein st ain (Flamingo) and prot ein ladder visua lised
by Coomass ie staining. S ignals were quant ified usin g ImageLab indi cating that
about 80 % of rL 2 is RNAylated by M odB invitro (n=3). Band pattern s indicated
that rL2 ca n be RNAylated on ce or even twice invitro. d–f, Tandem MS-ba sed
identif ication of R NAylated rL2 pep tide. d, MS/M S fragment io n spectrum
(spectru m ID: 8679) of RNAylated rL2 p eptide WRGV RPTVR c arrying A DP-
ribose pl us cytidine -monophosphat e and a 3′-phosphate grou p. The spect rum
shows marker ion s of adenine (A’) and cy tosine (C’) as well as A MP and CMP.
The precur sor ion ([M+xH]x+) is de tected uns hifted, shif ted by the mass of A DP-
ribose (*) and by A DP-ribose w ith adenine los s (**). Also, precurs or ions show a
specif ic loss of 42.02 1798 Da, which can b e explained by a loss o f CH2N2 at the
modifi ed arginine. e, I sotopic peak p attern of the pr ecursor ion show n in d, as
detect ed in the corres ponding MS prec ursor ion scan . f, Schematic se quence
and RNA add uct represe ntation of the R NAylated pepti de shown in d and e
including an notations of f ragment ion s. The fragm entation pro ducts obs erved
in the MS/M S spectru m, shown in d, of the ADP-ribose+CMP+3′-phosphate
adduct are in dicated in the s tructure by ligh t blue (mass loss) and dark b lue
(mass adduc ts) lines. g, Sele cted RNAylate d residues of rL 2 identifi ed by LC-
MS/MS . The cataly tically impo rtant H22 9 is 11.1Å apar t from R22 1. rL2
struct ure derived from a 1 .98Å cryo-EM struc ture (7K00)66.The schem atic
protein in c was c reated using Bi oRender (https://biorender.com).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extended Data Table 1 | ADP-ribosylation of endogenously His-tagged rS1 during T4 phage infection
MaxQuant intensities are presented for T4 phage-infected and uninfected samples in biologically independent triplicates (n=3). R139/R142 located in rS1 domain 2 and R485/R487 in rS1 domain
6 appear as ADP-ribosylation sites on rS1 invivo in all three replicates. The ratio comparing intensity of ADP-ribosylated and unmodiied species of the same peptide is computed for each
sample and peptide.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Extended Data Table 2 | ADP-ribosylation of rS1-WT, -R139K and -R139A during T4 phage infection
MaxQuant intensities are presented for T4 phage-infected and uninfected samples in biologically independent triplicates (n=3) only for the respective peptide of the R139 mutation site which is
expected for the respective rS1 version. ADP-ribosylation of the peptide in rS1 is observed invivo in all three replicates. However, ADP-ribosylation at position 139 is abolished by R139A or R139K
mutations (mutation indicated in red). The intensity of ADP-ribosylated peptide relative to the intensity of the corresponding unmodiied peptide species is at least 3-fold reduced upon R139
mutation. One may speculate that R142 is nevertheless ADP-ribosylated in the mutated rS1 proteins but overall ADP-ribosylation yield at the peptide may be reduced as the potentially predomi-
nant ADP-ribosylation site (R139) is not available for modiication.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Article
Extended Data Table 3 | Comparison of ModB protein intensities in lysate assay and invivo via proteomics
Normalised Log2 intensities for selected E. coli proteins and ModB found in proteomic analysis of E. coli cell lysates for invitro RNAylation (n=1) and in a previously published data set of the E.
coli and T4 phage proteome 5min post-infection38 (n=3). Intensity of ModB is divided by the intensity for various E. coli proteins. At 8.5 nM ModB, the ratios approximate conditions found invivo.
Raw data is presented in Supplementary Table5.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Content courtesy of Springer Nature, terms of use apply. Rights reserved

Content courtesy of Springer Nature, terms of use apply. Rights reserved

Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-scale
personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these purposes, Springer
Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue, royalties,
rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any other, institutional
repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or content on
this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke this
licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied with
respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed from
third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com