Structure; 2014 May Vol 22: 719-30

5.

Maria Romano1,*, Robert van de Weerd2,*, Femke C.C. Brouwer2, Giovanni N. Roviello1, Ruben Lacroix2

, Marion Sparrius2, Gunny van den Brink-van Stempvoort2, Janneke

J. Maaskant2, Astrid M. van der Sar2, Ben J. Appelmelk2,† , Jeroen J. Geurtsen2,3, Rita Berisio1

1 Institute of Biostructures and Bioimaging, National Research Council, Naples, Italy. 2Department of Medical Microbiology and Infection Control, VU University Medical Center, Amsterdam, The Netherlands. 3 Bacterial Vaccine Development, Crucell, Leiden, The Netherlands.

*These authors contributed equally to this work

Running title: Structure and function of mycobacterial RNase AS

Structure and function of RNase AS, a polyadenylate-specific

exoribonuclease affecting mycobacterial virulence in vivo

133132

55

ABSTRACTThe cell-envelope of Mycobacterium tuberculosis plays a key role in bacterial virulence and antibiotic resistance. Little is known about the molecular mechanisms of regulation of cell-envelope formation. Here, we elucidate functional and structural properties of RNase AS, which modulates M. tuberculosis cell-envelope properties and strongly impacts bacterial virulence in vivo. The structure of RNase AS reveals resemblance to RNase T from Escherichia coli, an RNase of the DEDD family involved in RNA maturation. We show that RNase AS acts as a 3’-5’-exoribonuclease that specifically hydrolyzes adenylate-containing RNA sequences. Also, crystal structures of complexes with AMP and UMP reveal the structural basis for the observed enzyme specificity. Notably, RNase AS shows a novel mechanism of substrate recruitment, based on the recognition of the hydrogen bond donor NH2 group of adenine. Our work opens a field for the design of novel drugs able to reduce bacterial virulence in vivo.

INTRODUCTION Tuberculosis (TB) is an infectious and often lethal disease whose threat encountered a resurgence, due to spreading of drug-resistant strains (Hett and Rubin, 2008). The causal agent of TB is Mycobacterium tuberculosis which exists in its human host in a variety of physiological states, including a latent infection affecting one third of the world’s population (Chao and Rubin, 2010; Squeglia et al., 2011). The chronic character of the disease, coupled to the insensitivity of mycobacteria to many antibiotics, makes TB a colossal public health problem. For this reason, the development of new and effective therapies targeting essential biosynthetic pathways of mycobacteria is an urgent need. A compelling strategy that offers potential for improving TB therapy is the use of multidrug cocktails acting together using different mechanisms of action to quickly kill mycobacteria. In this context, targeting the mycobacterial cell envelope biosynthesis is a cornerstone of TB therapy, as the complex cell envelope of M. tuberculosis is a key component for mycobacterial survival in the host and virulence of envelope mutants is severely attenuated in vivo (Brennan and Crick, 2007). The envelope consists of three parts: the cytoplasmic membrane, a peptidoglycan-arabinogalactan complex covalently linked to mycolic acids that form a second membrane (the mycomembrane), and finally an overlaying capsule (Brennan and Nikaido, 1995; Hoffmann et al., 2008). The outermost layer of the cell envelope, the capsule, is a network of polysaccharides, i.e., α-glucan, arabinomannan, and a minority of glycolipids and proteins (Lemassu and Daffé, 1994; Ortalo-Magne et al., 1995; Sani et al., 2010). Several of these components are associated with, or mediate binding to, host cells and modulate immune responses (Berisio, 2012; Daffé and Etienne, 1999; Esposito et al., 2008; Gagliardi et al., 2007; Stokes et al., 2004). Therefore, understanding the molecular mechanisms of the cell envelope formation and its role in host-pathogen interaction is vital for future anti-tuberculosis drug development.

In the framework of an on-going project on structural and functional properties of mycobacterial cell wall components (Correale et al., 2013; Squeglia et al., 2013), we here show that deletion of a single gene in the genome of Mycobacterium marinum (MMAR_3223), a widely used model organism for M. tuberculosis (Bouley et al., 2001; Broussard and Ennis, 2007), affects mycobacterial cell surface properties in vitro and virulence in vivo in the zebra-fish model. MMAR_3223 has been shown to be up regulated inside granulomas during long-term frog infection by M. marinum (Chan et al., 2002), whereas its ortholog in M. tuberculosis, rv2179c, is essential for mycobacterial growth (Sassetti et al., 2003) and up-regulated in the hyper virulent Beijing strains of M. tuberculosis during macrophage infections (Alonso et al., 2011). However, the exact role played by this gene in mycobacterial pathogenicity is hitherto unknown.

In this work, we use a plethora of structural biology, biochemical and microbiological techniques to characterize the protein encoded by rv2179c. By combining x-ray crystallography and HPLC analyses, we prove that this protein is a ribonuclease with strong specificity for poly-adenylate

135134

55

sequences; therefore we here denote it as A-specific RNase, RNase AS. Consistently, the crystal structure shows that RNase AS shares a similar fold with RNase T from Escherichia coli, despite the low amino acid sequence identity between the two proteins. Moreover, the structures of RNase AS complexes with AMP and UMP provide key information on the protein:ligand interactions important for catalytic function. Importantly, these data deliver a clear structural basis to explain the enzyme specificity, as they identify the critical molecular requirement of its substrates. This work highlights an important role for RNase AS in the modulation of cell envelope composition as well as bacterial virulence in vivo. Based on our structural data, targeted inhibition of RNase AS can be achieved by small molecule inhibitors that selectively bind the enzyme catalytic cleft.

RESULTSRNase AS modulates mycobacterial cell envelope propertiesWe created transposon (Tn) mutant libraries in M. marinum (a model organism for M. tuberculosis) and M. smegmatis and screened in a double filter assay for mutants with altered production of capsular glucan. Using an anti-glucan monoclonal antibody, we identified several capsular glucan mutants (Table S1). The list shows selected mutants, one of them is strain MMAR_3223::Tn, expressing less capsular α-glucan in double filter assay. As compared to wildtype strain, this mutant grew slower in vitro both on solid medium and in liquid culture (Figure S1). In a similar screen on Tn mutants in M. smegmatis, we isolated one mutant over-secreting capsular glucan where the Tn insertion was located in MSMEG_4245. Interestingly, MSMEG_4245 and MMAR_3223 are orthologous to M. tuberculosis, rv2179c. As anticipated in the Introduction, this gene encodes for a protein we here denoted as RNase AS. This protein is predicted to be essential for M. tuberculosis growth in vitro, and hence cannot be inactivated in that species (Sassetti et al., 2003). Therefore, we focused on MMAR_3223::Tn as a knockout strain, since the gene product of MMAR_3223 (MMAR_3223) shares a high sequence identity, of 79.2 %, with RNase AS.

Knockout of M. marinum MMAR_3223 affects bacterial virulence in vivo To investigate the effect of RNase AS on virulence, we carried out parallel experiments by injecting either M. marinum wild type strain or knockout strain MMAR_3223::Tn into zebra-fish embryos. By monitoring fluorescence at 5 days post infection, we observed that the infection of MMAR_3223::Tn mutant is strongly reduced in comparison to WT (Figure 1A and 1B). Quantification of the infection level by fluorescence measurements shows an infection of 4.7 % in comparison to the WT (Figure 1C). In this particular experiment the bacterial input/embryo was 161 and 169 CFU for WT bacteria and MMAR_3223::Tn strain, respectively; fluorescence data were measured on nine embryos for WT and mutant. To further confirm these data, we quantified by plating in an independent experiment live bacterial counts (CFU) of embryos infected with WT and MMAR_3223::Tn strain (Figure S2). Consistent with the fluorescence data, the MMAR_3223::Tn knockout strain shows strongly decreased CFU counts per fish. In

Figure 1 | Effect of inactivation of MMAR_3223 on virulence of M. marinum in the zebra-fish embryo. MCherry expressing M. marinum (Mma20) WT or MMAR_3223::Tn mutant live bacteria were injected into zebra-fish embryos and the bacterial infection monitored by fluorescence or plating. Inactivation of MMAR_3223 leads to attenuation in virulence. (A) Bright-field (left) and fluorescent (right) pictures of 5 days post-infection (dpi) embryos infected with M. marinum WT or (B) mutant MMAR_3223::Tn bacteria. The bacterial injection inocula were 161 and 169 CFU, respectively. (C) Quantification of infection with M. marinum WT or MMAR_3223::Tn in embryos at 5 dpi. The fluorescence intensities were quantified by proprietary imaging software and SEMs are shown. (D) Infection of M. marinum WT, MMAR_3223::Tn mutant and MMAR_3223::Tn complemented with M. tuberculosis rv2179c, measured by CFU plating of 5 dpi old zebra-fish embryos. Each data point represents the CFU count of one infected embryo. The CFU counts are graphically displayed in a log scale. Means are represented with bars and mean values are shown in numbers. The embryos were inoculated with 63, 47 and 62 CFU for WT, mutant and complemented strain, respectively; ** p< 0.00004, unpaired students t-test. Clearly, inactivation of M. marinum MMAR_3223 leads to a strong attenuation in vivo and this can be complemented with M. tuberculosis rv2179c (the gene product of which is RNase AS). See also Figures S1-S2 and Tables S1-S2.

a final in vivo experiment, we also tested MMAR_3223::Tn complemented with M. tuberculosis rv2179c. Clearly, virulence of the MMAR_3223::Tn mutant strain is strongly attenuated in vivo

A

B

C D

137136

55

and the knockout can be complemented by M. tuberculosis rv2179c (Figure 1D). Altogether, these data strongly suggest that the gene products of M. marinum MMAR_3223 and M. tuberculosis rv2179c (i.e. RNase AS) are indeed functional orthologs.

RNase AS is a dimer in solution and in the crystal stateGiven the importance of RNase AS for bacterial virulence, we expressed and purified RNase AS of M. tuberculosis for biochemical and structural analyses. Structural features of RNase AS in solution were checked using circular dichroism (CD) and light scattering studies. CD spectra showed that RNase AS adopts a well folded conformation, with two clear minima at 210 and 222 nm, indicative of an alpha-beta structure (Figure S3A). Analytical size-exclusion chromatography (SEC), coupled with multi-angle light scattering (MALS) was carried out to investigate the oligomerization state of RNase AS in solution. The on-line measurement of the intensity of the Rayleigh scattering as a function of the angle as well as the differential refractive index of the eluting peak in SEC was used to determine the molecular weight (MW)(Figure S3B). This analysis produced an MW of 38460 ±231 Da, which corresponds to a dimeric organization of the RNase AS protein in solution. RNase AS was crystallized in the P212121 space group with one dimer in the asymmetric unit. The structure was solved by single-wavelength anomalous dispersion (SAD) analysis of Europium-derivatized crystals and refined to a resolution of 2.1 Å (Table 1). As observed in solution, the crystal structure shows that RNase AS is a dimer, with each chain adopting a compact alpha/beta fold (Figure 2 and S3C).

Overall structure of RNase ASOverall, RNase AS exhibits a dimeric hearth-like shape, with the two monomers organized about a two-fold non-crystallographic axis. Each monomer adopts a compact α/β fold in which a central four-stranded β-sheet with topology β5β1β2β3 is surrounded by eight α-helices (Figure 2). Dimer assembly involves the burial of 20% of the total surface area (3080 Å2) and the formation of 13 hydrogen bonds and 11 salt bridges. The interface features complementary protuberances and cavities that come together to allow the formation of a highly stable dimer, with a gain of free energy of solvation DiG=-10.6 kcal/mol. Superposition of the two RNase AS molecules in the asymmetric unit shows a highly conserved structure, with overall backbone root mean square deviation (RMSD) values of 0.4 Å. An analysis of the degree of conservation of surface residues of the protein, carried out with ConSurf (Goldenberg et al., 2009) clearly identifies the catalytic site clefts of RNase AS as two conserved clusters of residues, one in each lobe of the hearth-like structure, formed by Asp6, Glu8, Asp95 and Asp145 (Figure 2). This finding univocally classifies RNase AS among RNases belonging to the DEDD superfamily, a large family of 3′-5′ exonucleases (Hsiao et al., 2011). Among conserved residues in the catalytic site cleft is also His140, which sorts RNase AS in the DEDDh family (Figure 2). Surface conservation analysis also shows a high sequence conservation for interface residues, like Tyr94, Val97, Met106 (Figure 2B). This finding suggests that the dimer organization we observe both in solution and in the crystal state is conserved in homologues sequences.

Table 1. Data collection and refinement statistics.

Values in parentheses are for highest resolution shells: 2.37-2.30 Å, 2.14-2.10 Å, 1.73-1.70 Å, 2.24-2.20 Å for EuCl3 derivative, native, AMP and UMP complexes, respectively.

A. Data collection

EuCl3 derivative Apo AMP complex UMP complex

Space group P212121 P212121 P212121 P212121

a, b, c (Å) 42.4, 76.7, 104.4 42.4, 77.1, 104.6 42.4, 77.0, 104.6 42.5, 76.8, 105.2

Resolution (Å) 50.0-2.3 50.0-2.1 50.0-1.7 50.0-2.2

Average redundancy 12.0 (11.2) 6.6 (6.6) 4.9 (2.8) 6.8 (6.9)

Completeness (%) 100.0 (100.0) 99.9 (99.9) 99.9 (99.9) 99.9 (100.0)

Rmerge (%) 7.2 (45.0) 4.6 (46.0) 4.5 (45.1) 5.8 (46.0)

Average I/σ(I) 25.1 (6.3) 31.8 (5.1) 25.8 (2.3) 26.9 (5.3)

B. Refinement Apo AMP complex UMP complex

Resolution range (Å) 15.0-2.1 15.0-1.7 15.0-2.2

Rwork, Rfree (%) 18.6, 23.2 18.5, 23.1 19.9, 24.0

No. atoms (protein, water)

r.m.s. deviations

2660, 360 2660, 372 2660, 360

Bond lengths (Å) 0.02 0.02 0.02

Bond angles (°) 1.9 1.9 1.9

Table 1 | Data collection and refinement statistics. Values in parentheses are for highest resolution shells: 2.37-2.30 Å, 2.14-2.10 Å, 1.73-1.70 Å, 2.24-2.20 Å for EuCl3 derivative, native, AMP and UMP complexes, respectively.

RNase AS specifically hydrolyses poly (A) in vitroEnzymes of the DEDD superfamily have essential functions in degrading DNA and/or RNA in prokaryotes and eukaryotes (Hsiao et al., 2011; Zuo et al., 2007). We checked for the ability of RNase AS to hydrolyze nucleic acids using HPLC chromatography coupled with mass spectrometry. HPLC analysis after incubation with poly (A) evidenced the accumulation of a degradation product at 4 min. This product, identified both by its retention time in HPLC and by LC-ESI-MS, is highly distinctive of AMP (m/z= 348.01) (Figure 3A). Interestingly, ESI-MS results showed the sole accumulation of AMP and not of poly-nucleotides. This indicates that RNase AS is also active on short poly-nucleotides to fully degrade them to AMP.

139138

55

Figure 2 | RNase AS is a dimer in the crystal state. (A) Cartoon representation of RNase AS crystal structure. Chains A and B of the dimer are shown in orange and green, respectively. The inset shows an enlargement of one of the two catalytic sites, with catalytic DEDDh residues shown in stick representation. (B) Degree of residue conservation on the protein surface (Consurf analysis) plotted on chain B, whereas chain A is shown as blue ribbon. Residue coloring, reflecting the degree of residue conservation, ranges from magenta (highly conserved) to cyan (variable). Conserved residues in the catalytic site and in the interface region are labeled. See also Figure S3.

A

B

To experimentally prove that RNase AS belongs to the DEDDh superfamily, we mutated the catalytic His140 to alanine (H140A) and checked its activity against poly (A). Results clearly showed that poly (A) remains fully un-degraded also after long incubation times (Figure 3A). Most members of the DEDD exo-nucleases have distinctive substrate preferences for specific structures or sequences of nucleic acids (Hsiao et al., 2012; Hsiao et al., 2011; Zuo et al., 2007). In order to understand the possible specificity of the enzymatic reaction

catalyzed by RNase AS, we carried out parallel digestion experiments using poly (C), poly (G) and poly (U) as possible substrates. Similar to experiments with poly (A), the incubated mixtures were analyzed using HPLC (Figure 3A). However, results were strikingly different from those observed for poly (A). Indeed, the UV chromatograms show that the total amounts of added polynucleotides were still present in the mixtures and no extra peaks corresponding to degradation products appeared after incubation (Figure 3B). These data showed that RNase AS has a strong specificity towards poly-adenylate nucleotides. To corroborate this result, we carried out HPLC and mass spectrometry analyses on the asymmetric oligonucleotide AAAACAAAAA. The choice of the oligonucleotide sequence had a twofold purpose: (i) to confirm that oligonucleotide degradation stops at C and (ii) to experimentally determine direction of degradation. The oligonucleotide was incubated with RNase AS and products were analyzed using HPLC and mass spectrometry. HPLC profiles show the formation of a peak at 4 min corresponding to AMP and a further peak at 11 min (Figure 3C). Mass spectrometry analysis allowed us to establish that this peak corresponds to a degradation product containing four A and one C nucleotide (Figure 3C). This finding evidences that RNase AS is a 3’-5’-exoribonuclease.

RNase AS strongly discriminates between RNA and DNATo investigate the specificity of RNase AS towards either RNA or DNA sequences, we carried out digestion experiments with two oligonucleotides: the dodeca-ribonucleotide A12 and the dodeca-deoxyribonucleotide dA12. HPLC analyses were carried out at increasing incubation times of protein-oligonucleotide mixtures. Consistent with studies conducted with poly (A) (Figure 3A), A12 is fully degraded to AMP (Figure 4A). Degradation is complete after 3 hours incubation, when the peak corresponding to non-degraded A12 disappears completely (Figure 4A) and the area of the peak corresponding to the degradation product reaches a stable value. Parallel HPLC analyses carried out on the incubated mixtures with dA12 show that a weak peak, corresponding to dAMP, appears after 3 hours incubation (Figure 4B). On the other hand, monitoring the reaction in time shows that the peak of non-degraded dA12 does not significantly change even after 72 hours (Figure 4B). These data show that RNase AS is barely capable of degrading dA12 and that kinetics is extremely slow.

Structural basis of RNase AS specificity: crystal structures of RNase AS complexesIn order to elucidate structural determinants responsible for the observed strong specificity of RNase AS for sequences of adenylate residues, we determined the crystal structures of the enzyme in complex with AMP and UMP. The analysis of the x-ray structure of the AMP complex shows clear electron density for two AMP molecules bound to the B chain of the enzyme (Figure 5A). On the other hand, only one AMP molecule binds the A chain, due to a crystallographic artifact (See Figure S4 and its legend for details). Analysis of binding to chain B shows that the two AMP moieties mimic the post-cleavage state, after hydrolysis of the phosphodiester bond between the two nucleotides. Therefore, we denoted the two molecules as AMP(-1) and AMP(+1) (Figure 5B). The two nucleotides establish interactions

141140

55

Figure 3 | RNase AS selectively hydrolyses poly(A). (A) HPLC profiles (280 nm) of poly(A) after 48h incubation with RNase AS (red), with the RNase AS mutant H140A (black) and in protein buffer (grey). Electrospray ionization mass (inset) clearly identifies AMP (m/z 348.01) as a reaction product. (B) HPLC profiles (280 nm) of poly(U), poly(G) and poly(C) after 48 hours incubation with RNase AS (upper curve, red) and protein buffer (bottom curve, black). An arbitrary shift was applied for clarity. (C) HPLC profiles (280 nm) of after 48h incubation with RNase AS with the polynucleotide AAAACAAAA (red) and in protein buffer (black). Degradation products are labelled. The reported value of m/z (1559.49, negative ion) is in agreement with the expected value for the product AAAAC ([C49H61N23O29P4-H]-, m/z 1559.05).

with both chains of the enzyme (orange and green in Figure 5), a feature which likely makes the enzyme active only as a dimer. Both are embedded in narrow clefts by establishing hydrogen bonding and hydrophobic interactions (Figure 5). Notably, the ribose moiety of AMP(-1) forms hydrogen bonds with the enzyme both through its 2’ and 3’ OH. In particular, the 2’OH binds the backbone carbonyl carbon of Phe9, which also locks the position of this sugar through Van der Waals interactions (Figure 5C). The 3’OH of AMP(-1) is hydrogen bonded to the backbone nitrogen of Phe9 and to the side chain Oɛ1 of Glu8. Also, the adenine moiety of AMP(-1) is stacked against the side chain of Trp46 (Figure 5B). These interactions make the 3’OH atom of AMP(-1) completely solvent inaccessible, a feature which confirms that, like other ribonucleases of the DEDD family, RNase AS is 3’-5’ exonuclease (Zuo and Deutscher, 2002a, b, c). Likely, RNase AS tightly locks the nucleotide at the 3’ end of the polynucleotide, thus bringing the first phosphodiester bond to be cleaved in the vicinity of the catalytic residues (Figure 5C). As observed for other members of the DEDD family (Hsiao et al., 2011; Zuo et al., 2005), two divalent (Mg2+) ions are present in the catalytic site and stabilize nucleotide binding (Figure 5C). The first Mg2+ ion (MGA) coordinated by the side chain of the catalytic Asp6, by the 2’ and 3’OH atoms of the ribose moiety of AMP(+1) and by the phosphate moiety of AMP(-1). Further coordination sites are occupied by water molecules. The second Mg2+ ion (MGB) is coordinated by the side chains of catalytic Asp6, Glu8 and Asp145, and by the phosphate moiety of AMP(-1). Notably, the full involvement of 2’OH atoms of both nucleotides in hydrogen bonding interactions, either with the protein structure or with MGA, provides a structural explanation for the strong preference of RNase AS for RNA rather than DNA (Figures 4 and 5). An interesting feature of AMP(+1) is that its binding site is composed of both chains of the enzyme, as its adenine moiety is fully buried by establishing interactions with the side chains of Met106 and Arg115 of the adjacent RNase AS chain (chain A, orange in Figure 5C). In particular, the NH2 group of the adenine base of AMP at the (+1) site forms a hydrogen bond with the backbone carbonyl oxygen of Met106 (Figure 5C). Since a carbonyl oxygen can act only as a hydrogen bond acceptor, this interaction poses serious restrictions on the nucleotide which can bind to the (+1) site.

A single hydrogen bond is responsible for A specificityTo corroborate our hypothesis that the adenine moiety on the nucleotide (+1) is the determinant for specificity of recognition by the enzyme and that its NH2 group is a strict requirement as a hydrogen bond donor, we performed HPLC studies using poly-inosine, a polymer of an intermediate of AMP biosynthesis, which holds all characteristics of AMP but carries a nucleobase having in its predominant keto form an oxygen in place of the adenine NH2 group (Figure 6). As predicted, the results clearly show that RNase AS is unable to degrade poly-inosine. Indeed, UV chromatograms of incubated mixtures showed no peaks corresponding to degradation products (Figure 6).To investigate the binding mode of pyrimidine nucleotides, we also determined the crystal structure of RNase AS in complex with UMP. Electron density maps clearly showed the presence of UMP at the site (-1) of both chains A and B, whereas no

A

B

C

143142

55

Figure 4 | RNase AS strongly discriminates between RNA and DNA. (A) HPLC profile (280 nm) of A12 after 3 hours incubation with RNase AS (upper curve, red) and with the protein buffer (bottom curve, black). (B) HPLC profiles after incubation of dA12 with RNase AS, in the time range 3-72 hours. In both panels, UV profiles after incubations were arbitrarily upshifted, for clarity.

electron density was observed for the site (+1) of chain B, likely due to low binding affinity of UMP for this site (Figure 7A). The observed conformation of UMP(-1) is superimposable with that of AMP(-1), with the pyrimidine base forming a stacking interaction with the side chain of Trp46 (Figure 7B). However, consistent with the observed binding mode of AMP, no electron density is observed for UMP in the (+1) site, this suggesting that interactions observed for AMP with Met106 and Arg115 are unfavorable for pyrimidine binding. This result explains the inability of RNase AS to degrade poly(U) and poly(C) (Figure 2B) and suggests that tight binding to both (-1) and (+1) sites is necessary to hold oligo-nucleotides in the right conformation for catalysis.

RNase AS structurally resembles RNase TA search for similar folds in structural data bases revealed a strong structural relationship between RNase AS and subunit RNase T from E. coli (PDB code 3V9X, Z= 23.0 and r.m.s.d.=4.1 Å upon overall backbone superposition), despite there being low sequence identity between the two, i.e. 13% (Figure 8). Compared to RNase T, RNase AS has several deletions, including the N- and C-terminal ends, which are 12-14 residues shorter in RNase AS (Figure S5). Overall, RNase AS displays a more compact structure than RNase T (Figure 8), with extra inter-subunit interactions formed in RNase AS by the loops L1, between β-strands β1 and β2 and by

A

B

A B

C

Figure 5 | Binding mode of AMP nucleotides. (A) Omit (Fo-Fc) electron density map, contoured at 2.0 σ, of AMP(-1) and AMP(+1) nucleotides. (B) Stick representation of AMP(-1) and AMP(+1) in the catalytic site cleft of RNase AS. Cartoon and surface representations of A and B chains of RNase AS are shown in orange and green, respectively. (C) Hydrogen bonding interactions and coordination with Mg2+ sites (MGA and MGB, in yellow) of AMP(-1) and AMP(+1) with the enzyme. See also Figure S4.

145144

55

Figure 6 | RNase AS does not degrade polyinosine. HPLC profiles (280 nm) of poly-inosine after 48h incubation with RNase AS (red) and in protein buffer (blank, black). An arbitrary shift was applied for clarity. The inset shows the chemical structure of inosine mononucleotide.

the α-helix H6 (Figure 8A). An interesting structural difference involves the binding site of AMP(+1). In RNase AS, Met106 is preceded by a proline residue (Pro105). Consequently, a long α-helix embedding residues 121-136 of RNase T is broken in two helices in RNase AS (H5 and H6), to contribute to form the cleft hosting AMP(+1) adenine base (Figures 8A and S6). Both RNase AS and RNase T are highly negatively charged proteins, with theoretical pI values of 5.3 and 5.2, respectively. However, the two proteins display uneven electrostatic potential surfaces (Figures 8B-C). Indeed, RNase AS does not exhibit a characteristic positively charged patch (NSB patch) of RNase T (Figure 8C), which has been shown to be involved in nucleotide binding and is associated with the preference of RNase T to bind long nucleotides (Zuo et al., 2007). Instead, a positively charged patch is located just below the catalytic cleft and is formed by Arg131 and Arg135 (Figure 8B). These differences likely account for the different preferences of the two enzymes for oligonucleotide of different lengths. Consistently, LC-ESI-MS show only the accumulation of AMP upon degradation of poly (A) and not of larger nucleotides. This suggests that, different from RNase T (Zuo et al., 2007), RNase AS is also active on short nucleotides.

DISCUSSIONMycobacteria rely on their complex cell envelope for shape and structure. Also, the low permeability of the cell envelope appears to be essential for mycobacterial survival in

Figure 7 | Binding mode of UMP. (A) Omit (Fo-Fc) electron density map, contoured at 2.0 σ, of UMP nucleotide at the (-1) site of RNase AS. (B) Superposition of complexes of RNase AS with AMP (navy) and UMP (cyan). Whereas UMP does not bind the (+1) site, it shares the same conformation and hydrogen bonding interactions with RNase AS in the (-1) site. Cartoons are shown in light green (UMP complex) and forest green (AMP complex). Binding of either UMP or AMP at the (-1) site induces slightly different conformations of the side chains of Trp46, His140 and Glu8, which are shown in stick representation.

A

B

the host. Therefore, the coordination of enzymes affecting cell envelope composition is a critical process for bacterial survival. By screening genes affecting cell envelope of M. marinum, we identified a protein with a close homologue in M. tuberculosis, here denoted as RNase AS. In this paper, we show that RNase AS is a 3’-5’ exoribonuclease with a strong specificity for poly-adenylate sequences. By combining x-ray crystallography with HPLC chromatography, light scattering, mass spectrometry and microbiology studies, we characterized the structure and the functional properties in vitro and in vivo of RNase AS and highlighted an important role of this enzyme in bacterial virulence.

147146

55

A

B C

Figure 8 | Structural comparisons between RNase AS and RNase T. (A) Superposition of RNase AS structure (orange) to that of RNase T (magenta). The inset shows a detail of structural differences of the region involving the AMP(+1) binding pocket. (B,C) Electrostatic potential surfaces of RNase AS and RNase T, respectively.

The crystal structures of RNase AS, determined in its free and complexed forms, highlighted a structural resemblance with RNase T from E. coli (Zuo et al., 2007), that could not be predicted based on sequence alignments. Structural features of the enzyme resemble those of an independently determined structure reported in the Protein Data Bank while our functional characterization was ongoing (PDB code 4HEC)(Abendroth et al., 2014). Strong conservation of a characteristic patch of residues on the structure surface (the DEDD patch) shows that RNase AS, like RNase T, belongs to the superfamily of DEDD exonucleases (Figure 2). Consistently, phylogenetic analysis shows that RNase AS clusters with the RNA-editing DEDDh exonucleases such as REX and the Poly (A) nuclease PAN (Abe et al., 2010; van Hoof et al., 2000; Yoda et al., 2013). Since most members of the DEDD exonuclease family have distinctive substrate preferences for specific structures or sequences of nucleic acids, we analyzed the in vitro activity of RNase AS in degrading polynucleotides (Hsiao et al., 2011). Results evidenced a strong specificity of the enzyme for RNA sequences, rather than DNA sequences.

Also, RNase AS readily degrades poly-adenylate but no other homo-polynucleotides (Figure 3). The crystal structure of RNase AS in complex with AMP provides a straight forward mechanistic explanation for the observed specificity. Namely, two AMP molecules bind to the catalytic site cleft of the enzyme in positions corresponding to that upstream (+1) and downstream (-1) the cleavage site. Among interactions stabilizing the AMP(+1) conformation, a single hydrogen bond between the NH2 group of the adenine moiety and the backbone carbonyl oxygen of Met106 forces restrictions on the nature of this base. Indeed, even guanine would not fulfill the requirement of having a hydrogen bond donor at the position corresponding to adenine NH2 group. Consistently, we show that the enzyme is unable to degrade poly-inosine, a polynucleotide in which the adenine NH2 group is replaced by a carbonyl oxygen, unable to act as a donor in hydrogen bonding interactions (Figure 6). This simple recruitment mechanism of RNase AS differs from that proposed for RNase T (Hsiao et al., 2011). Indeed, the strong preference for poly-adenylate is also held by other RNases of the DEDD family, but the structural basis for this specificity has only been recently explained for RNase T (Hsiao et al., 2011). By preparing complexes of RNase T with non-preferred single strand DNA ending with cytosine (C), it was shown that the binding of the 3’ C induces a conformational change disrupting the active site structure and directing the catalytic His181 away from the catalytic center (Hsiao et al., 2011). In RNase AS, the α-helix H6, embedding Met106, represents one of the several structural differences with RNase T. Indeed, Met106 is a one-residue insertion preceded by a proline residue (Pro105) in RNase AS (Figure S5). Consequently, a long α-helix embedding residues 121-136 of RNase T is disrupted in RNase AS, to contribute to form the cleft hosting AMP(+1) adenine base (Figures 8A and S6). This structural difference renders the mechanism of nucleotide selection by RNase AS much simpler. Consistent with this recruitment mechanism, the crystal structure of RNase AS in complex with UMP shows the presence of this nucleotide only at the (-1) site and not at the (+1) site, which dictates specificity (Figure 7), thus explaining the inactivity of the enzyme towards poly(U) and poly(C).

The 3′-ends of both prokaryotic, including mycobacterial, and eukaryotic mRNA are polyadenylated, although features of prokaryotic and eukaryotic polyadenylation are distinct. For example, addition of poly(A) tails increases the stability of transcripts in eukaryotes, whereas it promotes instability in prokaryotes (Adilakshmi et al., 2000; Mohanty et al., 2012). The poly (A) tail is an important structural element affecting translational efficiency of mRNAs. Therefore, deadenylation is the first and rate-limiting step of mRNA turnover and an important control point during gene expression (Mohanty and Kushner, 2013). Like other members of the DEDD family, RNase AS is likely involved in mRNA turnover through processing of 3′-ends of mRNA. However, the transcription of which genes are regulated by RNase AS is hitherto unknown.

We investigated the effect of RNase AS on virulence, by injecting the knockout strain MMAR_3223::Tn (being MMAR_3223 highly homolog to RNase AS) of M. marinum into

149148

55

zebra-fish embryos. Notably, we observed that knockout leads to a dramatic depression of bacterial virulence in vivo and that the knockout can be complemented by the gene encoding for RNase AS (Figure 1). What is the molecular cause of the decreased virulence is still under investigation, but it is known that cell envelope components may play a crucial role in virulence. For example, capsular α-glucan interacts with phagocytic complement receptor 3 (CR3), thus mediating binding of M. tuberculosis to host immune cells (Cywes et al., 1997; Ehlers and Daffe, 1998). It also blocks dendritic cell (DC) functions (Gagliardi et al., 2007) and interacts with the C-type lectin immune receptor DC-SIGN (Geurtsen et al., 2009). Consistently, a M. tuberculosis mutant with decreased capsular α-glucan levels displayed reduced virulence in vivo in mice (Sambou et al., 2008). The RNase AS-mediated deadenylation process is likely involved in determining mRNA stability of multiple transcripts and is therefore capable to influence additional mycobacterial virulence determinants. Consistently, the gene region of RNase AS is clustered with genes that affect the mycobacterial cell surface. The gene rv2181 (orthologous of MMAR_3225), located next to the operon rv2179c-2180c, codes for an α-1-2 mannosyltransferase from the glycosyltransferase family C (GT-C) (Kaur et al., 2006; Mishra et al., 2011) with a functional dual role in mannose capping of lipoarabinomannan (LAM) and mannan-core branching of LAM and lipomannan (LM) (Kaur et al., 2008). Its inactivation leads to decreased virulence of M. marinum in the zebrafish (Stoop et al., 2013). However, we observed no differences in LAM or LM biogenesis (data not shown), suggesting that LAM and LM are not likely responsible for the reduced virulence observed in M. marinum RNase AS mutant.

Future efforts should be directed toward identifying the genes responsible for the dramatic impact on bacterial virulence, which are deadenylated by RNase AS (Figure 1) and whose deadenylation is likely critical to their stability. A full comprehension of the molecular details of RNA maturation processes will be essential to devising novel strategies for targeted modulation of RNA processing for therapeutic purposes. In this context, the availability of the crystal structure of RNase AS, whose inactivation has strong functional implications, aids in rational drug design and provides a strong practical advantage in high-throughput docking and lead optimization studies.

EXPERIMENTAL PROCEDURESCloning, expression and purification of RNase ASThe gene coding for RNase AS from M. tuberculosis (rv2179c) was cloned into a pETM11 vector to express a His6-tag RNase AS fusion protein. For protein expression, the plasmid was transformed into E. coli BL21 (DE3) pLysS cells. 10mL overnight culture of BL21 (DE3) pLysS E. coli/pETM11/His-tag-RNase AS was inoculated in 1l of Luria-Bertani (LB) medium, supplemented with 50μg/ml kanamycin and 33 μg/ml chloramphenicol and subsequently

grown at 37°C. When the cultures reached an OD600 ~ 0.65, the expression of the recombinant protein was induced by adding 0.4 mM IPTG at 20°C and incubated for 16 h.

Cell pellets were re-suspended in lysis buffer (10mM Imidazole, 20mM HEPES pH 7.5, 300mM NaCl, 5% glycerol, 10mM MgCl2, 0,01% CHAPS, 1mM DTT, pH 7.5), including a protease-inhibitor cocktail (Roche Diagnostic) and DNase 20mg/mL. Subsequently, suspensions were incubated on ice for 30 min and additionally, the cells were sonicated and centrifuged at 30,000 x g for 30 min. The protein was purified from a clarified lysate by affinity chromatography, using a 5 ml Ni-NTA (Ni2+-nitrilotriacetate) column (GE Healthcare) with an “on-column cleavage” procedure. In detail, the supernatants were loaded onto an affinity column and subsequently washed with buffer A (30mM Imidazole, 20mM HEPES pH 7.5, 300mM NaCl, 5% glycerol, 10mM MgCl2, 0,01% CHAPS, 1mM DTT, pH 7.5). Next, the TEV protease (TEVprot-RNase AS ratio 1:30) was added to the column and incubated for 16-20 hours at 20°C. Then, the target protein (His-tag cleaved RNase AS) was eluted with the same buffer A, leaving His containing elements and the protease on the column. Subsequently, RNase AS was purified by dialysis overnight at 20°C against buffer (20mM HEPES pH 7.5, 300mM NaCl, 5% glycerol, 5mM MgCl2 , 0,0025% CHAPS, 1mM DTT, pH 7.5) and was analyzed by SDS-PAGE for protein homogeneity. The expression vector encoding the H140A mutation was generated by site-directed mutagenesis of wild-type plasmid pETM11-RNaseAS using the Stratagene QuikChange kit and the mutagenic primers: fw 5’-CGGCCACGCGACGTCGCCGACGCGCTGGTCGATG-3’ and rv 5’ CATCGACCAGCGCGTCGGCGACGTCGCGTGGCCG-3’. DNA sequencing confirmed the mutation into the recombinant plasmid sequence. Expression and purification of the mutant were performed in the same conditions as for the un-mutated enzyme.

Crystallization, data collection and processingCrystallization trials were performed at 293 K using the hanging-drop vapour-diffusion method. Preliminary crystallization conditions were set up using a robot station for high throughput crystallization screening (Hamilton STARlet NanoJet 8+1) and commercially available sparse-matrix kits (Crystal Screen kits I and II, Hampton Research, Index). Optimisation of the crystallization conditions was performed manually by tuning protein and precipitant concentrations. Best crystals were grown using a protein concentration of 14 mg/mL in 0.1M HEPES pH 7.5, 10% w/v polyethylene glycol 8,000, 8% v/v ethylene glycol. For structure solution, Europium chloride derivative crystals were prepared by soaking a native crystal in a solution containing 9 mM EuCl3, 0.1M HEPES pH 7.5, 10% w/v polyethylene glycol 8,000, 8% v/v ethylene glycol for 2 hours. A Single-wavelength Anomalous Diffraction experiment (SAD) was recorded in-house at 100K using a Rigaku Micromax 007 HF generator producing Cu Kα radiation and equipped with a Saturn944 CCD detector. The data sets were scaled and merged using HKL2000 program package (Otwinowski and Minor, 1997) (Table 1). Crystals of complexes of RNase AS with AMP and UMP were prepared by soaking crystals of native protein

151150

55

in a solution containing AMP or UMP at 20 mM. Diffraction data were collected using the same protocol used for the native crystal.

Structure determination and refinementPhasing was achieved using in-house SAD data, using a previously adopted protocol (Ruggiero et al., 2011; Squeglia et al., 2014). Using these data, both SHELXD (Sheldrick, 2008) and SOLVE (Terwilliger, 2004) identified five Europium ions. Phases, improved by phase extension and density modification by RESOLVE (Terwilliger, 2004) and wARP (Langer et al., 2008), allowed us to trace nearly the entire molecule structure. Crystallographic refinement was carried out against 95% of the measured data using the REFMAC program implemented in the CCP4 program suite (Potterton et al., 2003). The remaining 5% of the observed data, which was randomly selected, was used in Rfree calculations to monitor the progress of refinement. Structures were validated using the program PROCHECK (Cheng et al., 2008). Structures of complexes with AMP and UMP were determined using Fourier difference techniques and refined using REFMAC (Table 1). Structures of free RNase AS and of its complexes with AMP and UMP were deposited in the PDB with codes 4OKJ, 4OKE and 4OKK, respectively.

CD spectroscopy To analyze the conformational state of RNase AS, far-UV CD spectra were registered at 20°C. All CD spectra were recorded with a Jasco J-810 spectropolarimeter equipped with a Peltier temperature control system (Model PTC-423-S). Molar ellipticity per mean residue, [q] in deg cm2•dmol-1, was calculated from the equation: [q] = [q]obs•mrw•(10•l•C)-1, where [q]obs is the ellipticity measured in degrees, mrw is the mean residue molecular mass (116.0 Da), C is the protein concentration in g•L-1 and l is the optical path length of the cell in cm. Far-UV measurements (190-260 nm) were carried out at 20 °C using a 0.1 cm optical path length cell and a protein concentration of 0.2 mg•mL-1.

Mass spectrometry and HPLC analysesRNase AS 2 mM was incubated for 48 hours at 4°C with 200 mM poly(A) (concentration expressed in A nucleotides) in protein buffer (50 mM Tris HCl, 200 mM NaCl, 5% glycerol, 1 mM DTT, 0.01% CHAPS, 12.5 mM MgCl2, pH 7.8). As control experiments, the protein alone and polynucleotides (poly(A), poly(U), poly(G), poly(C)) 200 mM were stored at 4°C in the protein buffer for 48 hours. After incubations, all samples were analyzed by LC–MS, performed on an MSQ mass spectrometer (Thermoelectron, Milan, Italy) equipped with an ESI source operating at 3 kV needle voltage and 320 °C, and with a complete Surveyor HPLC system, comprising an MS pump, an autosampler, and a PDA detector, by using a Phenomenex Jupiter C18 300 Å column (5 mm, 4.6 x 250 mm) column. Gradient elution was performed (monitoring at 280 nm) by building up a linear gradient from 2 to 50 % of buffer B (0.05 % TFA in acetonitrile) in buffer A (0.05 % TFA in water) over 15 min with a flow rate of 0.8 ml/min. Free mononucleotides were used as reference for mass spectrometry.

Degradation experiments on oligonucleotides AAAACAAAAA, A12 and dA12 were carried out in the same experimental conditions used for polynucleotides. The ESI mass spectrum for the AAAACAAAAA degradation product was obtained using a LCQ DECA ion trap instrument (Thermoelectron, Milan, Italy), operating in negative ion mode. The entrance capillary temperature was 280 °C, the source voltage was 4.5 kV and the capillary voltage was -15V. The sample was dissolved in 20 mM ammonium acetate (pH=7.4) to obtain a 31 mM concentration of oligonucleotide.

Multiple Angle Light Scattering experimentsPurified RNase AS was analyzed by size-exclusion chromatography (SEC) coupled to a DAWN MALS instrument (Wyatt Technology) and an OptilabTM rEX (Wyatt Technology). 600 μg of sample was loaded a S200 10/30 column, equilibrated in 50 mM Tris HCl, 200 mM NaCl, 5% glycerol, 1 mM DTT, 0.01% CHAPS, 12.5 mM MgCl2, pH 7.8. A constant flow rate of 0.5 ml/min was applied. The on-line measurement of the intensity of the Rayleigh scattering as a function of the angle as well as the differential refractive index of the eluting peak in SEC was used to determine the weight average molar mass (Mw) of eluted protein, using the Astra 5.3.4.14 software (Wyatt Technologies) software.

Sequence conservation studiesSequence conservation studies were carried out using the software ConSurf (Goldenberg et al., 2009). The homologue search algorithm CSI-BLAST was used to retrieve sequences from the UNIREF-90 sequence database using an E-value cutoff of 0.001 and minimal sequence identity of 35% (150 sequences). Sequences were aligned using MAFFT-L-INS-I alignment method.

Bacterial strains and growth conditionsM. marinum (strain Mma20, (van der Sar et al., 2004)) was grown at 30°C on Middlebrook 7H10 agar plates (Difco) or in liquid medium (7H9 Middlebrook) medium supplemented with 10% albumin-dextrose-catalase (ADC; BD Bioscience) and 0.05% Tween-80. M. smegmatis mc2155 was likewise grown but at 37 °C. For bacterial strains grown under selective pressure, hygromycin or kanamycin were added at 50 µg/ml. The E. coli strains were grown at 37°C in LB broth or on LB agar (Difco). When required, hygromycin was supplemented at 100µg/ml.

Transposon mutagenesis and capsular α-glucan mutant screenTransposon mutagenesis was performed according to Abdallah et al. (Abdallah et al., 2006). The obtained libraries of M. marinum and M. smegmatis were screened in a double filter assay for expression of capsular α-glucan. Concisely, the mixture of bacteria and transposon carrying-mycobacteriophages was incubated and plated on nitrocellulose filters (Millipore HATF08250) on top of 7H10 plates supplemented with kanamycin and were grown till visually discernable

153152

55

colonies appeared. Subsequently, the filters were removed and placed on top of new filters covering fresh 7H10 plates and incubated at 30oC or 37 °C. During this period soluble, secreted cell surface components including capsular glucan but not intact bacteria diffuse from the top to the bottom filters. Next, the bottom filter was incubated with monoclonal antibody (Mab) IV58B6 against glycogen (Baba, 1993); this Mab cross-reacts with mycobacterial capsular glucan (Sani et al., 2010). After DAB/CNPO staining, filters were checked for colonies that showed more or, in contrast, less capsular α-glucan as compared to WT strain. The transposon insertion sites of the mutant strains were identified by ligation mediated (LM) PCR as earlier described (Abdallah et al., 2006); for primer and linker sequences used see Table S2.

Construction of a genetically complemented strain in M. marinumMutant M. marinum MMAR_3223::Tn was complemented with its M. tuberculosis rv2179c ortholog. In short, primers were designed (see Table S2) and rv2179c was amplified from M. tuberculosis genomic DNA by PCR. The PCR-products were purified and digested with BamHI and NheI. pSMT3 E. coli-Mycobacterium shuttle vector was purified from E. coli strain Dha and digested with BamHI and NheI. The digested plasmid and the insert were ligated by T4 ligase and then transformed into E. coli XL-10 by heat-shock; in this plasmid, rv2179c is located downstream of a lac promotor. The transformants were selected on LB plates supplemented with hygromycin and the pSMT3 complementation plasmid was confirmed by sequencing. This plasmid was electroporated into M.marinum MMAR_3223::Tn. The complemented strain (coded M. marinum MMAR_3223::Tn+rv2179c) was confirmed by PCR (Table S2).

Zebra-fish embryo infection and quantification of bacterial infection levelM. marinum (Mma20) wild-type and M. marinum MMAR_3223::Tn expressing the fluorescent marker mCherry were grown to the logarithmic phase and microinjected into Danio rerio zebra-fish embryos and, subsequently, the infection was analyzed according to Stoop and colleagues (Stoop et al., 2011). Briefly, zebra-fish embryos were inoculated in the caudal vein and the bacterial injection load was determined by plating on 7H10. At 5 days after the bacterial injection (dpi), the infection was monitored by fluorescence microscopy (Leica MZ16FA) and subsequent bright-field and fluorescence pictures were created by a specialized camera (Leica, DFC420C). The quantification of the bacterial infection level in these embryos was monitored in two different ways. In real-time, it was analyzed with fluorescent imaging software and secondly, zebra-fish embryos were disrupted and plated for the determination of colony forming units (CFUs) as published earlier (Stoop et al., 2011). The zebra-fish experiments were performed according the guidelines of the Dutch animal welfare laws and were done at two independent occasions.

ACKNOWLEDGMENTSThe authors thank the Mizutani Foundation for Glycoscience and the COST Action BM1003 (COST-Grants-BM1003-00772) for financial support to R. B. and the Netherlands Organization

of Scientific Research, NWO, for financial support to R.v.d.Weerd (NWO Grant 819.02.004). The authors also thank Christina Vandenbroucke-Grauls and Wilbert Bitter for helpful discussions. We thank Roy Ummels, Marnix Lameijer, Susanne Raadsen, Louis Ates and Eveline Weerdenburg for providing technical assistance.

155154

55

REFERENCES1. Abdallah, A.M., Verboom, T., Hannes, F., Safi, M., Strong, M., Eisenberg, D., Musters, R.J.,

Vandenbroucke-Grauls, C.M., Appelmelk, B.J., Luirink, J., et al. (2006). A specific secretion system mediates PPE41 transport in pathogenic mycobacteria. Molecular Microbiology 62, 667-679.

2. Abe, M., Suzuki, H., Nishitsuji, H., Shida, H., and Takaku, H. (2010). Interaction of human T-cell lymphotropic virus type I Rex protein with Dicer suppresses RNAi silencing. FEBS Lett 584, 4313-4318.

3. Abendroth, J., Ollodart, A., Andrews, E.S., Myler, P.J., Staker, B.L., Edwards, T.E., Arcus, V.L., and Grundner, C. (2014). Mycobacterium tuberculosis Rv2179c Protein Establishes a New Exoribonuclease Family with Broad Phylogenetic Distribution. The Journal of Biological Chemistry 289, 2139-2147.

4. Adilakshmi, T., Ayling, P.D., and Ratledge, C. (2000). Polyadenylylation in mycobacteria: evidence for oligo(dT)-primed cDNA synthesis. Microbiology 146, 633-638.

5. Alonso, H., Aguilo, J.I., Samper, S., Caminero, J.A., Campos-Herrero, M.I., Gicquel, B., Brosch, R., Martin, C., and Otal, I. (2011). Deciphering the role of IS6110 in a highly transmissible Mycobacterium tuberculosis Beijing strain, GC1237. Tuberculosis 91, 117-126.

6. Baba, O. (1993). Production of monoclonal antibody that recognizes glycogen and its application for immunohistochemistry. Kokubyo Gakkai Zasshi 60, 264-287.

7. Berisio, R. (2012). Editorial: learning from pathogens: exploiting host-pathogen liaison for therapeutic development. Current Protein Peptide Science 13, 697-698.

8. Bouley, D.M., Ghori, N., Mercer, K.L., Falkow, S., and Ramakrishnan, L. (2001). Dynamic nature of host-pathogen interactions in Mycobacterium marinum granulomas. Infection and immunity 69, 7820-7831.

9. Brennan, P.J., and Crick, D.C. (2007). The cell-wall core of Mycobacterium tuberculosis in the context of drug discovery. Current Topics Medicinal Chemistry 7, 475-488.

10. Brennan, P.J., and Nikaido, H. (1995). The envelope of mycobacteria. Annual Review of Biochemistry 64, 29-63.11. Broussard, G.W., and Ennis, D.G. (2007). Mycobacterium marinum produces long-term chronic

infections in medaka: a new animal model for studying human tuberculosis. Comparative Biochemistry Physiology C Toxicology Pharmacology 145, 45-54.

12. Chan, K., Knaak, T., Satkamp, L., Humbert, O., Falkow, S., and Ramakrishnan, L. (2002). Complex pattern of Mycobacterium marinum gene expression during long-term granulomatous infection. Proceedings of the National Academy of Sciences of the United States of America 99, 3920-3925.

13. Chao, M.C., and Rubin, E.J. (2010). Letting sleeping dos lie: does dormancy play a role in tuberculosis? Annual Review of Microbiology 64, 293-311.

14. Cheng, T.J., Sung, M.T., Liao, H.Y., Chang, Y.F., Chen, C.W., Huang, C.Y., Chou, L.Y., Wu, Y.D., Chen, Y.H., Cheng, Y.S., et al. (2008). Domain requirement of moenomycin binding to bifunctional transglycosylases and development of high-throughput discovery of antibiotics. Proceedings of the National Academy of Sciences of the United States of America 105, 431-436.

15. Correale, S., Ruggiero, A., Capparelli, R., Pedone, E., and Berisio, R. (2013). Structures of free and inhibited forms of the L,D-transpeptidase LdtMt1 from Mycobacterium tuberculosis. Acta Crystallographica Section D, Biological Crystallography 69, 1697-1706.

16. Cywes, C., Hoppe, H.C., Daffé, M., and Ehlers, M.R. (1997). Nonopsonic binding of Mycobacterium tuberculosis to complement receptor type 3 is mediated by capsular polysaccharides and is strain dependent. Infection and Immunity 65, 4258-4266.

17. Daffé, M., and Etienne, G. (1999). The capsule of Mycobacterium tuberculosis and its implications for pathogenicity. Tuberculosis and Lung Disease 79, 153-169.

18. Ehlers, M.R., and Daffé, M. (1998). Interactions between Mycobacterium tuberculosis and host cells: are mycobacterial sugars the key? Trends in Microbiology 6, 328-335.

19. Esposito, C., Pethoukov, M.V., Svergun, D.I., Ruggiero, A., Pedone, C., Pedone, E., and Berisio, R. (2008). Evidence for an elongated dimeric structure of heparin-binding hemagglutinin from Mycobacterium tuberculosis. Journal of Bacteriology 190, 4749-4753.

20. Gagliardi, M.C., Lemassu, A., Teloni, R., Mariotti, S., Sargentini, V., Pardini, M., Daffé, M., and Nisini, R. (2007). Cell wall-associated alpha-glucan is instrumental for Mycobacterium tuberculosis to block CD1 molecule expression and disable the function of dendritic cell derived from infected monocyte. Cellular Microbiology 9, 2081-2092.

21. Geurtsen, J., Chedammi, S., Mesters, J., Cot, M., Driessen, N.N., Sambou, T., Kakutani, R., Ummels, R., Maaskant, J., Takata, H., et al. (2009). Identification of mycobacterial alpha-glucan as a novel ligand for DC-SIGN: involvement of mycobacterial capsular polysaccharides in host immune modulation. Journal of Immunology 183, 5221-5231.

22. Goldenberg, O., Erez, E., Nimrod, G., and Ben-Tal, N. (2009). The ConSurf-DB: pre-calculated evolutionary conservation profiles of protein structures. Nucleic Acids Research 37, D323-327.

23. Hett, E.C., and Rubin, E.J. (2008). Bacterial growth and cell division: a mycobacterial perspective. Microbiology and Molecular Biology Reviews: MMBR 72, 126-156, table of contents.

24. Hoffmann, C., Leis, A., Niederweis, M., Plitzko, J.M., and Engelhardt, H. (2008). Disclosure of the mycobacterial outer membrane: cryo-electron tomography and vitreous sections reveal the lipid bilayer structure. Proceedings of the National Academy of Sciences of the United States of America 105, 3963-3967.

25. Hsiao, Y.Y., Duh, Y., Chen, Y.P., Wang, Y.T., and Yuan, H.S. (2012). How an exonuclease decides where to stop in trimming of nucleic acids: crystal structures of RNase T-product complexes. Nucleic Acids Research 40, 8144-8154.

26. Hsiao, Y.Y., Yang, C.C., Lin, C.L., Lin, J.L., Duh, Y., and Yuan, H.S. (2011). Structural basis for RNA trimming by RNase T in stable RNA 3’-end maturation. Nature Chemical Biology 7, 236-243.

27. Kaur, D., Berg, S., Dinadayala, P., Gicquel, B., Chatterjee, D., McNeil, M.R., Vissa, V.D., Crick, D.C., Jackson, M., and Brennan, P.J. (2006). Biosynthesis of mycobacterial lipoarabinomannan: role of a branching mannosyltransferase. Proceedings of the National Academy of Sciences of the United States of America 103, 13664-13669.

28. Kaur, D., Obregon-Henao, A., Pham, H., Chatterjee, D., Brennan, P.J., and Jackson, M. (2008). Lipoarabinomannan of Mycobacterium: mannose capping by a multifunctional terminal mannosyltransferase. Proceedings of the National Academy of Sciences of the United States of America 105, 17973-17977.

29. Langer, G., Cohen, S.X., Lamzin, V.S., and Perrakis, A. (2008). Automated macromolecular model building for X-ray crystallography using ARP/wARP version 7. Nat Protoc 3, 1171-1179.

30. Lemassu, A., and Daffé, M. (1994). Structural features of the exocellular polysaccharides of Mycobacterium tuberculosis. Biochemical Journal 297, 351-357.

31. Mishra, A.K., Krumbach, K., Rittmann, D., Appelmelk, B., Pathak, V., Pathak, A.K., Nigou, J., Geurtsen, J., Eggeling, L., and Besra, G.S. (2011). Lipoarabinomannan biosynthesis in Corynebacterineae: the interplay of two alpha(1-->2)-mannopyranosyltransferases MptC and MptD in mannan branching. Molecular Microbiology 80, 1241-1259.

32. Mohanty, B.K., and Kushner, S.R. (2013). Deregulation of poly(A) polymerase I in Escherichia coli inhibits protein synthesis and leads to cell death. Nucleic Acids Research 41, 1757-1766.

33. Mohanty, B.K., Maples, V.F., and Kushner, S.R. (2012). Polyadenylation helps regulate functional tRNA levels in Escherichia coli. Nucleic Acids Research 40, 4589-4603.

34. Ortalo-Magne, A., Dupont, M.A., Lemassu, A., Andersen, A.B., Gounon, P., and Daffé, M. (1995). Molecular composition of the outermost capsular material of the tubercle bacillus. Microbiology 141, 1609-1620.

35. Otwinowski, Z., and Minor, W. (1997). Methods in Enzymology 276, 307–326.

157156

55

36. Potterton, E., Briggs, P., Turkenburg, M., and Dodson, E. (2003). A graphical user interface to the CCP4 program suite. Acta Crystallographica, Biological Crystallography 59, 1131-1137.

37. Ruggiero, A., Squeglia, F., Marasco, D., Marchetti, R., Molinaro, A., and Berisio, R. (2011). X-ray structural studies of the entire extracellular region of the serine/threonine kinase PrkC from Staphylococcus aureus. Biochemical Journal 435, 33-41.

38. Sambou, T., Dinadayala, P., Stadthagen, G., Barilone, N., Bordat, Y., Constant, P., Levillain, F., Neyrolles, O., Gicquel, B., Lemassu, A., et al. (2008). Capsular glucan and intracellular glycogen of Mycobacterium tuberculosis: biosynthesis and impact on the persistence in mice. Molecular Microbiology 70, 762-774.

39. Sani, M., Houben, E.N., Geurtsen, J., Pierson, J., de Punder, K., van Zon, M., Wever, B., Piersma, S.R., Jimenez, C.R., Daffé, M., et al. (2010). Direct visualization by cryo-EM of the mycobacterial capsular layer: a labile structure containing ESX-1-secreted proteins. PLOS Pathogens 6, e1000794.

40. Sassetti, C.M., Boyd, D.H., and Rubin, E.J. (2003). Genes required for mycobacterial growth defined by high density mutagenesis. Molecular Microbiology 48, 77-84.

41. Sheldrick, G.M. (2008). A short history of SHELX. Acta Crystallographic A 64, 112-122.42. Squeglia, F., Bachert, B., De Simone, A., Lukomski, S., and Berisio, R. (2014). The crystal structure

of the Streptococcal collagen-like protein 2 globular domain from invasive M3-type group A Streptococcus shows significant similarity to immunomodulatory HIV protein gp41. The Journal of Biological Chemistry 289, 5122-5133.

43. Squeglia, F., Marchetti, R., Ruggiero, A., Lanzetta, R., Marasco, D., Dworkin, J., Petoukhov, M., Molinaro, A., Berisio, R., and Silipo, A. (2011). Chemical basis of peptidoglycan discrimination by PrkC, a key kinase involved in bacterial resuscitation from dormancy. Journal of the American Chemical Society 133, 20676-20679.

44. Squeglia, F., Romano, M., Ruggiero, A., Vitagliano, L., De Simone, A., and Berisio, R. (2013). Carbohydrate recognition by RpfB from Mycobacterium tuberculosis unveiled by crystallographic and molecular dynamics analyses. Biophysical Journal 104, 2530-2539.

45. Stokes, R.W., Norris-Jones, R., Brooks, D.E., Beveridge, T.J., Doxsee, D., and Thorson, L.M. (2004). The glycan-rich outer layer of the cell wall of Mycobacterium tuberculosis acts as an antiphagocytic capsule limiting the association of the bacterium with macrophages. Infection and Immunity 72, 5676-5686.

46. Stoop, E.J., Mishra, A.K., Driessen, N.N., van Stempvoort, G., Bouchier, P., Verboom, T., van Leeuwen, L.M., Sparrius, M., Raadsen, S.A., van Zon, M., et al. (2013). Mannan core branching of lipo(arabino)mannan is required for mycobacterial virulence in the context of innate immunity. Cellular Microbiology 15, 2093-2108.

47. Stoop, E.J., Schipper, T., Huber, S.K., Nezhinsky, A.E., Verbeek, F.J., Gurcha, S.S., Besra, G.S., Vandenbroucke-Grauls, C.M., Bitter, W., and van der Sar, A.M. (2011). Zebrafish embryo screen for mycobacterial genes involved in the initiation of granuloma formation reveals a newly identified ESX-1 component. Disease Models and Mechanisms 4, 526-536.

48. Terwilliger, T. (2004). SOLVE and RESOLVE: automated structure solution, density modification and model building. Journal of Synchrotron Radiation 11, 49-52.

49. van der Sar, A.M., Abdallah, A.M., Sparrius, M., Reinders, E., Vandenbroucke-Grauls, C.M., and Bitter, W. (2004). Mycobacterium marinum strains can be divided into two distinct types based on genetic diversity and virulence. Infection and Immunity 72, 6306-6312.

50. van Hoof, A., Lennertz, P., and Parker, R. (2000). Three conserved members of the RNase D family have unique and overlapping functions in the processing of 5S, 5.8S, U4, U5, RNase MRP and RNase P RNAs in yeast. EMBO Journal 19, 1357-1365.

51. Yoda, M., Cifuentes, D., Izumi, N., Sakaguchi, Y., Suzuki, T., Giraldez, A.J., and Tomari, Y. (2013). Poly(A)-specific ribonuclease mediates 3’-end trimming of Argonaute2-cleaved precursor microRNAs. Cell Reports 5, 715-726.

52. Zuo, Y., and Deutscher, M.P. (2002a). Mechanism of action of RNase T. I. Identification of residues required for catalysis, substrate binding, and dimerization. The Journal of Biological Chemistry 277, 50155-50159.

53. Zuo, Y., and Deutscher, M.P. (2002b). Mechanism of action of RNase T. II. A structural and functional model of the enzyme. The Journal of Biological Chemistry 277, 50160-50164.

54. Zuo, Y., and Deutscher, M.P. (2002c). The physiological role of RNase T can be explained by its unusual substrate specificity. The Journal of Biological Chemistry 277, 29654-29661.

55. Zuo, Y., Wang, Y., and Malhotra, A. (2005). Crystal structure of Escherichia coli RNase D, an exoribonuclease involved in structured RNA processing. Structure 13, 973-984.

56. Zuo, Y., Zheng, H., Wang, Y., Chruszcz, M., Cymborowski, M., Skarina, T., Savchenko, A., Malhotra, A., and Minor, W. (2007). Crystal structure of RNase T, an exoribonuclease involved in tRNA maturation and end turnover. Structure 15, 417-428.

159158

55

SUPPLEMENTARY DATA

0 50 100 150 2000

1

2

3

4

5

WT Mma20Mma20 MMAR_3223::Tn

time

OD

600

Figure S. Effect of inactivation of MMAR_3223 on growth of M. marinum.

Figure S1 | Effect of inactivation of MMAR_3223 on growth of M. marinum. M. marinum strains Mma20 wild type (WT, u) and Mma20 MMAR_3223::Tn transposon mutant (∆) were grown in liquid 7H9 culture medium and growth was monitored by measurement of optical density over time (OD600; time indicated in hours). Inactivation of MMAR_3223, an ortholog of M. tuberculosis rv2179c, decreases bacterial cell growth. The growth curve are representive for two independent experiments. Related to Figure 1.

WT (Mma2

0)

MMAR_322

3::Tn

100

1000

10000

100000

CFU

per

fish

7786

19000

MMAR_333

4::Tn

** 580

Figure S2. Infection of M. marinum WT, MMAR_3223::Tn mutant and control mutant MMAR_3334::Tn, measured by CFU plating of 5 dpi old zebra-fish embryos.

Figure S2 | Infection of M. marinum WT, MMAR_3223::Tn mutant and control mutant MMAR_3334::Tn, measured by CFU plating of 5 dpi old zebra-fish embryos. Each data point represents the CFU count of one infected embryo. The CFU counts are graphically displayed in a log scale. Means are represented with bars and mean values are shown in numbers. The embryos were inoculated with 80, 102 and 126 CFU for WT, mutant and control, respectively; ** p< 0.012, unpaired students t-test. Clearly, inactivation of M. marinum MMAR_3223 leads to a strong attenuation in vivo. Related to Figure 1.

figure S3

Figure S3 | Fold and oligomerization state of RNase AS. (A) CD spectrum of RNase AS (0.2 mg/mL) measured in 10 mM sodium phosphate. (B) Molecular masses as a function of elution time derived by analytical SEC-MALS experiments. Related to Figure 2.

A B

161160

55

Figure S4

Figure S4 | Binding modes of AMP molecules in the two chains of RNase AS dimer. (A) Overall view of AMP binding to chain A (blue, only binding at the (-1) site) and chain B (green, binding at both (-1) and (+1) sites). (B) Superposition of the two chains show different conformations of residues lining the binding pocket (+1). In chain A, the (+1) site is not accessible to AMP binding due to a crystallographic artifact. Indeed, residues of the (+1) site of chain A (e.g. Arg121 and Arg135) are involved in crystal contacts with symmetry-related residues. Related to Figure 5.

A

B

Figure S5 Secondary structure of RNase AS, based on its crystal structure (A).

Figure S5 | Secondary structure of RNase AS, based on its crystal structure (A). Sequence alignment (catalytic residues drawn in red) (B) and structural comparison (C) between RNase AS (orange) and RNase T (purple). Compared to RNase AS, RNase T presents extensions at its N- and C-termini (magenta in panel B) and three main insertions in secondary structure elements (red, green and blue in panels B and C). Insertion of Met106 in RNase AS is shown in gray. Related to Figure 8.

A

B

C

163162

55

Figure S6 Superposition of the B chain of RNase AS in complex with AMP (green) and the B chain of RNase T in complex with TAAA (blue) (pdb code 3V9X).

Figure S6 | Superposition of the B chain of RNase AS in complex with AMP (green) and the B chain of RNase T in complex with TAAA (blue) (pdb code 3V9X). The arrows describe the displacement of A chain of RNase AS (orange) compared to the A chain of RNase T (magenta) upon superposition of B chains. The inset shows an enlargement of binding modes of AMP to RNase AS (green) and TAAA to RNase T (blue). Related to Figure 8.

1

2

3

No Gene Insertion position Primer H37Rv homologue Putative function

Mutant No.S1

2000-B

MMAR_3334

2000-6

MMAR_0992

pSalg

pSalg

MMAR_3223 pMyco

Rv2241 pyruvate dehydrogenasesubunit E1 (aceE)

Rv0653c transcriptional regulatory protein

Rv2179c hypothetical protein

Table S1| Transposon screen mutants secreting less capsular alpha-glucan.

Table S1 | M. marinum transposon mutants selected in a library screen for decreased expression of secreted capsular alpha-glucan in a double filter assay. In one particular M. marinum mutant expressing less glucan, a transposon was inserted in MMAR_3223. The transposon insertion site was identified with the primers pMyco or pSalg and subsequently sequencing. The gene MMAR_3223 is orthologous to M. tuberculosis rv2179c. Related to Figure 1.

1

2

3

4

CAGCACGCGTCGGTACTTCTACGACACC

NameNo Primer description Primer sequence

MTUB_Rv2179c_fw_c

MTUB_Rv2179c_NHis_rev_c

rv2179c complementation

complementation: forward

complementation: reverse CAGCACGCGTTCGACGGTCCAGTTCATC

pSalg

pMyco

GCTTATTCCTCAAGGCACGA

CCGGGGACTTATCAGCCAAC

Tn identification

transposon insert

transposon insert

Table S2| List of primers used in this study

Table S2 | Primers used in this study. Related to Figure 1.