The protein regulator ArgR and the sRNA derived from the 3’-UTR region of its gene, ArgX, both regulate the arginine deiminase pathway in Lactococcus lactis

Small regulatory RNAs (sRNAs) and their enormous potential and versatility have provided us with an astounding insight in the complexity of bacterial transcriptomes. sRNAs have been shown to be involved in a variety of cellular processes that range from stress to general metabolism. Here we report that the gene encoding the transcriptional regulator ArgR is immediately followed by the gene of the small regulatory RNA ArgX. The latter is transcribed from its own promoter. The production of ArgX is induced by increasing arginine concentrations and repressed by CcpA. Previously, ArgR was shown to act as a transcriptional repressor of the catabolic arginine deiminase pathway (arc operon) by binding in the promoter region of arcA. Here we demonstrate that ArgX downregulates arc mRNA levels. Furthermore, ArgX putatively blocks the translation of one of the genes in the operon, arcC1, a process that would redirect an intermediate in arginine degradation, carbamoyl phosphate, towards pyrimidine synthesis. Our findings exemplify, for the first time, the combinatorial power of a transcription factor and a small regulatory RNA derived from the 3’-UTR region. The regulators ArgR and ArgX share a common target, but act on transcription and on RNA level, respectively.


Introduction
Post-transcriptional regulation by regulatory RNAs has added huge complexity to gene regulatory networks, which contained until recently mainly information of protein regulators such as transcription factors (TFs). Regulatory RNAs influence gene expression by base pairing to target mRNAs and thereby affect the stability of the transcript or its translation [1]. This process can result in gene activation or -repression [2][3][4]. Some regulatory RNAs are transcribed from the DNA strand opposite to the coding strand of the gene they regulate and, thus, (partially) overlap with the gene transcript. These so-called antisense RNAs (asRNAs) have the potential to base-pair with the mRNA by means of a perfect match [5,6]. In contrast, small regulatory RNAs (sRNAs) from intergenic regions usually form a duplex with, often, different a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 reversed in the presence of arginine, which acts as a co-repressor and binds to AhrC. ArgR in a complex with arginine-bound AhrC shifts its preference to ARG boxes. In this model, ArgR acts as a DNA binding protein, while AhrC senses and binds arginine [20]. The argR gene is located upstream of the arc operon, with only the argS gene intervening.
Here we show by transcriptome and proteome studies that ArgX affects arc/Arc expression, and verified these results using an arc-sfgfp fusion in various genetic backgrounds. Furthermore, we examined the promoter of ArgX and show that it behaves strikingly similar to P arcA ; it responds to arginine and is controlled by the transcription factors CcpA and ArgR.

Materials and methods
Bacterial growth, plate reader assays and microscopy Table 1 presents an overview of strains used in this study. L. lactis was routinely grown as standing cultures at 30˚C in CDMPC [21] or M17 broth (Difco, Becton Dickinson, Le Pont de Cm r , high copy number cloning vector [23] pIL253 Em r , medium copy number cloning vector [26] pCS1966 Em r , oroP, integration vector [27] pSEUDO Em r , vector for integration in the pseudo_10 locus [28] pSEUDO-GFP Em r , vector for integration of gfp fusion constructs in pseudo_10 [28] pVE6007 Cm r , plasmid with thermo-sensitive replication [29] pJP005 Cm r , pNZ8048, with recT under control of P nisA [30] pSVDM5003

General DNA techniques and L. lactis strain construction
Plasmid DNA and PCR fragments were purified with the NucleoSpin Plasmid kit and NucleoSpin Gel and PCR Clean-up kit (Machery-Nagel GmbH, Düren, Germany). The enzymes that were used were produced by Fermentas/Thermo Scientific, Vilnius, Lithuania, unless stated otherwise.
The strain with a mutation in the -10 box of the promoter of ArgX was made by applying the recombineering technique [30,31]. To enable recombineering, recT was expressed from pSVDM5003. This plasmid was constructed by amplification of the recT gene from pJP005 [30], introducing a BamHI and an XhoI site. The BamHI and XhoI digested recT gene was then inserted by restricting pVE6007 [29] with the same enzymes, followed by ligation by T4 DNA ligase. For recombineering transformation, 100 μg of phosphorothioate-modified singlestranded DNA oligonucleotide (Biolegio, Nijmegen, The Netherlands) was used, which was introduced via electroporation using a Bio-Rad Gene Pulser (Bio-Rad Laboratories, Richmond, CA) at 2.5 kV, 25 μF and 200 Ohm. Cells with the anticipated mutation were cured from the pVE6007_recT (pSVDM5003) plasmid by growing them overnight in non-selective medium (GM17) at 37˚C.
L. lactisΔArgX was made by double crossover recombination (DCO). To facilitate DCO, the flanking regions of ArgX were amplified; the restriction sites XbaI and PstI were introduced for the upstream region, and PstI and XhoI site for the downstream region. These fragments were then cloned in plasmid pCS1966 [27] by XbaI and XhoI restriction, followed by ligation and used to transform E. coli. The plasmids were isolated from E. coli and integrated in the L. lactis chromosome using erythromycin selection for integration and 5-fluoroorotate counter selection for the excision step, respectively.
The L. lactis ΔArgX complementation strain was constructed by amplifying a 388-bp fragment containing ArgX and its promoter and inserting it in pSEUDO [28] using the restriction enzymes BamHI and EcoRI. This plasmid was constructed in E. coli, isolated and integrated into L. lactis cells using a double cross-over method [28].
The remaining L. lactis mutants were made using restriction-and-ligation-independent cloning [32], by separately amplifying the vector backbone and the required insert(s) with the polymerase pfuX7 [33], which were subsequently treated with the USER enzyme mix (New England Biolabs, Ipswich, MA). Ligation mixtures were used to transform L. lactis by electroporation. A transcriptional fusion was made between P ArgX and the gene for superfolder GFP (sfgfp) [34] and integrated in the transcriptionally silent pseudo_10 locus [28] in order to examine the activity of P ArgX . Overexpression of ArgX, argRX and argRX with a mutated start codon (pSVDM5005, pSVDM5006 and pSVDM5007) was done by introducing these genes under the nisin inducible promoter P nisA on the high copy plasmid pNZ8048 [23]. To test the effect of the overproduction of these three variations of ArgX on arc-sfgfp expression, the region of P arcA until the start codon of arcC1 was fused to the sfgfp gene and inserted in pIL253 [26], resulting in pSVDM5004. Table 1 and S1 Table provide an overview of strains and oligonucleotides used in this study, respectively.
First dimension electrophoresis (iso-electric focusing) and second dimension sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) were performed as described before [35]. Differentially expressed protein spots were identified and statistically analyzed using the Delta2D Image analysis program (Decodon GmbH, Greifswald, Germany), excised, de-stained with 50 mM ammonium bicarbonate in 50% acetonitrile/water and dried prior to overnight digestion with 10 μl of 10 ng/μl trypsin. The obtained peptide mixture was purified with ZipTips (Merck Millipore, Darmstadt, Germany) and spotted on a MALDI plate. MAL-DI-TOF analysis was performed on a Voyager DE Pro (AB Science, Paris, France) and protein identification was done using the Mascot database search on www.matrixscience.com.

RNA isolation and quality control
RNA was isolated as described before [14]. In short, frozen cell pellets were re-suspended in 400 μl TE-buffer (10 mM Tris, 1 mM EDTA; pH 8.0), and added with 50 μl of 10% SDS, 500 μl phenol/chloroform and 0.5 g glass beads (75-150 μm, Fischer Scientific). The cells were disrupted in a Biospec Mini-BeadBeater using 2 cycles of 45 sec with a 1-min interval on ice. Nucleic acids were recovered by chloroform extraction and treated with DNase I supplemented with RiboLock RNase inhibitor (Fermentas/Thermo Scientific)) for 30 min at 37˚C. RNA was retrieved using standard phenol/chloroform extraction and sodium acetate/ethanol precipitation. RNA pellets were dissolved in elution buffer from the High Pure RNA Isolation Kit (Roche Diagnostics, Almere, the Netherlands) and subsequently stored at -80˚C. RNA concentration was measured using a Nanodrop ND-1000 (Thermo Fischer Scientific). RNA quality was assessed by checking the integrity of the 16S/23S rRNA and the presence of any DNA contamination on a 1% agarose/1% bleach gel [36].

Northern hybridization
Separation of total RNA (10 μg) was performed on 8% or 12% polyacrylamide gels in TAE buffer (40 mM Tris, 20 mM acetic acid, and 1 mM EDTA). As a denaturing agent, 1% bleach was used to replace 7 M urea. The separated RNAs were then transferred to a positively charged Zeta-Probe nylon membrane (Bio-Rad), using semi-dry electroblotting (Bio-Rad). RNAs were covalently cross-linked to the nylon membranes at 1200 mJ in a UVC-508 Ultraviolet Crosslinker (Ultra-Lum Inc., Carson, CA). ssDNA oligonucleotides (See S1 Table) were labeled with 32 P-γATP using Polynucleotide kinase (Fermentas/Thermo Scientific), according to instructions of the manufacturer. The membranes were incubated overnight for hybridization at 42˚C in PerfectHyb Plus Hybridization buffer (Sigma-Aldrich Chemie Gmbh, Munich, Germany) with 9 μl 0.16 pmol/μl of the labeled probe. Membranes were washed twice in 2x saline sodium citrate (SSC) buffer with 0.1% SDS, after which they were exposed to a Phosphor Screen overnight. A Cyclone Plus Phosphor Imager and OptiQuant software (PerkinElmer, Groningen, NL) was used for imaging.

RNA deep sequencing and data analysis
RNA samples were sequenced at the Primbio Research Institute (Exton, PA), who performed Ribo-Zero rRNA removal and library preparation using the AmpliSeq kit (ThermoFischer Scientific). The cDNA libraries were sequenced on an Ion Proton sequencer (ThermoFischer Scientific). Raw sequence reads were analyzed for quality, trimmed with a PHRED score >28 and aligned to the genome L. lactis NZ9000 using Bowtie 2 [37]. RKPM values were used as an input for the T-REx analysis pipeline for statistical analysis to determine differentially expressed genes [38]. For the T-REx analysis, a text file describing the factors, contrasts and classes specifying genes from the arc (red) and arg (blue) operons were written. These text files, together with the RKPM values are available in S2 Table. The RNA-seq data have been uploaded under GEO accession number GSE104515.

ArgX biogenesis and homology in other L. lactis species
Previously, we have identified the putative small regulatory RNA LLMGnc_172 by differential RNA sequencing [14]. This sRNA of~66-nt, which we rename here as ArgX, is expressed from the same strand as the immediate upstream gene, argR, and overlaps with the argR 3'-UTR, thus sharing a common terminator sequence (Fig 1A and 1B). Northern analysis using a probe for ArgX shows that an additional larger band exists, which likely represents the argR transcript including ArgX. To provide evidence that both ArgX and argR overlap, we designed a probe that would anneal to the coding region of argR. Since we observed identical sizes on both blots, we conclude that ArgX indeed overlaps with argR ( Fig 1C).
ArgX was identified in a 5'-enriched fraction of total RNA of L. lactis that resulted in primary transcript reads due to a treatment with Terminator 5´-Phosphate-Dependent Exonuclease (TEX). A nearly perfect -10 RNA polymerase recognition sequence (TATACT) was present upstream of the ArgX transcription start site [14], indicating that ArgX is transcribed from its own promoter. We mutated the putative -10 promoter sequence to investigate the possibility that ArgX might also be derived from a processing event, as has been described for CpxQ [11]. Northern analysis revealed that ArgX is not produced in the strain with the -10 box mutation and that, thus, ArgX is derived only by activity from its own promoter, not by processing from the larger argX transcript (Fig 1D). In an L. lactis ΔArgX complementation mutant, carrying ArgX including a 246-bp region upstream region, a band of the size of ArgX reappeared in the Northern blot ( Fig 1D). We also observe an increase of ArgX expression under high arginine conditions, a result that is further examined below.
A GLASSgo search [39] revealed that ArgX is highly conserved in other strains of L. lactis, especially when comparing the argR coding region immediately upstream of the ArgX gene ( Fig 1E). Notably, L. lactis subsp. lactis strains differ from L. lactis subsp. cremoris strains by an A to G mutation in the stem of the terminator (Fig 1E), although this change does not alter its structure or stability, as predicted by Mfold [40]. We could not identify homologs in other, more distant bacteria, using BLAST search and the GLASSgo output. The promoter of ArgX is highly regulated and responds to arginine Various sRNAs are controlled by transcriptional regulators, as well as the other way around [41]. For instance, the sRNA RyhB, which is involved in iron homeostasis in E. coli, is repressed by the ferric uptake regulator Fur [42], of which the translation is negatively influenced by RyhB [43]. Other sRNAs are induced by a sigma factor, such as MicA and RybB, which are regulated by RpoE involved in extracytoplasmic stress [44][45][46]. In our previous work we predicted the presence of a Catabolite Repressive Element (CRE-box) in the ArgX promoter, slightly upstream of the -10 sequence [14], suggesting that the carbon catabolite protein, CcpA, controls ArgX expression. CcpA is a pleiotropic transcription factor that acts as a repressor when glucose is available as a carbon source [19,47]. The chromosomal P ArgX -sfgfp fusion was used to examine the activity of P ArgX . A heterogeneous gene expression pattern was apparent in cells growing in the presence of moderate concentrations (2 mM) of arginine. This heterogeneity in P ArgX activity was not observed at high levels (25 mM) of arginine (Fig 2A). This behavior relays to the macroscopic level as very bright patches of GFP-producing cells in otherwise non-fluorescent colonies (Fig 2B). The strain containing P ArgX -sfgfp was grown in CDMPC with different concentrations of arginine and with glucose as a carbon source. These studies revealed that P ArgX is induced by arginine, in a concentration dependent response ( Fig  2C). This effect was only seen in stationary phase, which suggests that CcpA might have a repressive effect during the exponential growth phase.
Subsequently, we performed transcriptome analyses on L. lactis mutants in which the gene of one of three transcriptions factors, CcpA, CodY or ArgR was deleted from the chromosome. In the latter mutant, ArgX was still intact [25]. The deletion of ccpA (+ 3.9-fold) or codY (-3.2-fold) had a significant effect on ArgX expression, whereas ArgR deletion did not (see S3  Table). An arginine-dependent response of P ArgX does suggest involvement of ArgR in regulating ArgX. Using the ARC box sequences (half ARG box sites) of P arcA [20] we identified a consensus sequence, WGHATADW, that was used to scan ArgX promoter region. This consensus sequence largely overlaps with the -10 sequence of P ArgX . We therefore integrated P ArgX -sfgfp in the pseudo_10 locus in the ccpA, argR or codY deletion backgrounds and measured GFP activity in the various strains in the stationary phase in the presence of high (25 mM) or low (1 mM) concentrations of arginine. All mutations have an influence on the expression of P ArgX under both conditions, albeit that removal of argR or codY has a minimal effect while deletion of ccpA has by far the highest impact (Fig 2D). It furthermore appears that the effect of CodY as measured with the transcriptional fusion is not in line with the decreased expression of ArgX in a ΔcodY mutant that we observed in the transcriptome data. We note here that the difference in media, rich GM17 media for the transcriptome study and CDMPC for the plate reader assay, could be at the basis of the observed difference. Also, it is possible that this discrepancy could be due to potential indirect effects caused by CodY that influence the promoter activity of ArgX.

Expression of arc/Arc is elevated after deletion of the ArgX gene
To examine the potential targets of ArgX, an ArgX mutant strain was constructed that lacks the -10 promoter sequence and the first 32 nucleotides of ArgX (L. lactisΔArgX). The mutation was made such that it did not touch the terminator structure in order not to affect transcription termination of the argR transcript. The transcriptome and proteome of L. lactisΔArgX were compared to those of the wildtype strain L. lactis NZ9000 using RNA-Seq and 2D gel electrophoresis, respectively. For the analysis, the strains were grown in rich GM17 medium. It is important to mention that no effects were seen at the level of argR mRNA or on the ArgR protein level. The arcABD1C1C2 (6.5 ±0.7), argGH (9.3 ±2.1), gltS/argE (8.5 ±0.6), argFBDJC (22.7 ±12.5) and gltQP (8.2 ±2.0) gene clusters were upregulated in the exponential growth phase in the mutant strain relative to the wildtype (see S3 Table). The transcripts that encode genes for arginine biosynthesis are possibly induced, since the level of arginine has most probably decreased due to the high expression of arc in L. lactisΔArgX. One of the most affected genes, with an almost 1500-fold upregulation, was llmg_1128. This hypothetical gene has been previously correlated with strain robustness [48], and is possibly expressed by the stress caused by disrupted arginine regulation. Overall, the changes are strikingly similar to those observed in the L. lactis ccpA and argR mutants. In the stationary phase, pleiotropic effects were observed in the strain lacking ArgX, including elevated arc expression (3.8 ±0.9) (Fig 3A).
We also studied the proteome to uncover possible differences as a consequence of the absence of ArgX in the exponential growth phase. Eleven protein spots had changed significantly; these were analyzed by MALDI-TOF and identified by using the Mascot server. For three spots, no protein could be identified. Four spots were identified as representing ArcA and another four as ArcB, with an average increase of 4.4 (± 1.5) and 4.9 (± 1.5) fold, respectively, in the ArgX deletion mutant relative to the wildtype strain (Fig 3B). The increase in the amount of ArcA/B proteins reflects the changes of their transcripts but other significant protein changes were not detected.

ArgX affects arc expression directly and decreases growth rate when arginine is absent
ArgX is predicted by TargetRNA 2 [49] to interact with the RBS of arcC1, which is the fourth gene encoded on the arc transcript, downstream of arcA, arcB and arcD1 (Fig 4A). To study the direct effect of overexpression of ArgX RNA on the expression of arc, a plasmid was constructed containing a DNA fragment with the arc promoter region and arc genes until the start codon of arcC1 fused to the sfgfp gene (pIL253::P arcA -arcABD1-RBS arcC1 -sfgfp). We decided to construct this fusion on a plasmid because it would otherwise destroy the native arc operon. In addition, plasmids were made that contained the ArgX gene or the argR gene including the ArgX sequence, the latter with or without a disrupted start codon AAG (argR(Δstart)X). These genes were each placed under control of the nisin-inducible promoter P nisA . As a control, the empty expression vector pNZ8048 was used [23] (Fig 4B). The results (Fig 4C) show that induction with nisin of ArgX expression leads to a decrease of arc expression. Lower GFP fluorescence was measured under uninduced and induced conditions in strains SVDM2014 and SVDM2015 containing the mutated and the intact argR gene, probably because ArgX sRNA is produced from its own promoter in these constructs. Since the media contained 25 mM of The sRNA ArgX regulated the arc operon in Lactococcus lactis arginine, ArgX expression is high. Lowest fluorescence was observed when expression of argR was induced. Overexpression of argR(Δstart)X did not result in a drop of the fluorescent signal in comparison with the un-induced sample.
The expression behavior of arc measured by the readout of the GFP in the same plasmid used above, did not show any significant and reproducible differences in the deletion mutant compared to the wildtype strain.
To obtain insight in the biological relevance of ArgX, we grew L. lactis and mutants of ArgX under different arginine regimes. In the absence of arginine, the ArgX deletion strain grew significantly faster during the exponential phase and reached a higher final OD. By contrast, the same strain grew slightly slower at high arginine concentrations of 10 or 50 mM, while ultimately reaching comparable end ODs (Fig 4D).

Biogenesis and processing of ArgX
Most trans-encoded regulatory sRNA genes are located in intergenic regions in the chromosome. Recently, a few examples of regulatory RNAs were reported that overlap with the 3'-UTRs of coding transcripts [13]. We have previously identified ten 3'-UTR-overlapping sRNAs in L. lactis by dRNA-seq. Since the RNA used in this study was enriched for primary transcripts, sRNAs derived from processed 3'-UTRs could not be detected and, if they exist in this organism, remain to be identified [14]. Here we characterized LLMGnc_172 (ArgX), an sRNA that overlaps with the 3'-UTR of the argR transcript. ArgX and argR use the same terminator sequence. ArgX is expressed from its own promoter as a 66-nt transcript and is not formed via processing of the longer argR transcript, as we did not observe any transcripts by Northern analysis after disrupting the -10 sequence in the ArgX promoter. As we observe two distinct ArgX bands on a (12%) polyacrylamide gel it could be that processing takes place at the 5'-end of the ArgX transcript, cleaving off one or only a few nucleotides. Such a processed form of ArgX would carry a monophosphate (5'-P) at its 5'-end while the primary product carries a tri-phosphate (5'-PPP) group. This difference could be biologically relevant as it has been shown in E. coli that RNase E degrades target mRNAs in response to the 5'-P of the involved sRNA. Moreover, in the absence of its target mRNA, an sRNA that carries a 5'-P is more prone to degradation by RNase E [50]. It is possible that 5'-processing of ArgX would change its molecular function. We did not observe a fixed ratio between the two forms of ArgX by Northern analysis, albeit that the upper band is higher under high arginine availability. Possibly, the ArgX processing itself is regulated and depending on the requirement by the cells for e.g., arginine or one of its derivatives. Adding to the complexity of the system, both arc and ArgX are heterogeneously expressed, at least under "normal" arginine concentrations (1-2mM). This could mean that some cells in a population produce more ammonia by catabolizing the carbamoylphosphate, e.g. to neutralize the self-produced acid in the environment, while others may invest in the production of proteins and nucleotides using arginine as the precursor. Whether or not such a strategy would provide a benefit for the whole population is interesting to further investigate.

Regulation of arginine metabolism and the effect of the sRNA ArgX
Arginine metabolism in L. lactis is regulated by the carbon catabolite repressor CcpA [19], the transcriptional repressor CodY [17] and by the arginine repressor ArgR and its protein partner, AhrC [20]. We show here that ArgX expression is mostly affected by CcpA. The ArgX promoter responds to arginine availability in a strikingly similar fashion as the promoter of the arginine catabolic arc operon. This is remarkable as our results show that arc is also regulated by ArgX. We hypothesize that the aligned expression could function in arginine homeostasis. An in silico prediction using TargetRNA2 hinted at a possible base pairing between ArgX and the RBS of arcC1, the fourth gene in the arc operon. Deletion of ArgX resulted in an increase in the number of arc transcripts as well as a rise in the proteins ArcA and ArcB. Overexpression of ArgX, the argR transcript with an intact gene or one in which the argR start codon (argR(Δstart)X) had been mutated all led to a lower expression of arc, indicating that ArgX regulates arc directly. The possibility that the parental argR mRNA is also able to regulate arc by a base pairing reaction, was ruled out since we did not observe any decrease in arc expression upon overexpression of argR(Δstart)X in comparison with the un-induced mutant. Overexpression of argR(Δstart)X leads to a stronger repression of arc than when ArgX was overexpressed. This can be explained by the fact that the ArgX promoter is still present in this construct, probably resulting in a very high amount of ArgX transcripts also because of the high level of arginine (25 mM), a condition that was shown to boost ArgX expression. Nevertheless, the repressing effect on arc transcription by the protein regulator ArgR appears to be stronger than that of the sRNA ArgX at least under the conditions tested here.

Arginine, a versatile molecule in carbon and nitrogen metabolism
Arginine has previously been shown to play a role in acid stress tolerance in L. lactis [51]: The ammonium produced as one of the end products of arginine catabolism can be used to counteract acidification. Ammonium can also be produced for example through conversion of glutamine to glutamate. CcpA represses transcription of the arc operon in the presence of abundant glucose during the exponential growth phase [52], while pH neutralization might already be helpful at this stage of growth. Whenever the glucose level drops, and arginine is present, ArgR repression is relieved and arc is expressed [20]. The ArgX and arc promoters both become activated by increasing amounts of arginine in the stationary phase. The control of arc by CcpA suggests that arginine is mainly used as a carbon and/or energy source, at least during the stationary phase. The arginine deiminase pathway imports and catabolizes arginine. However, levels of arginine and its metabolites could rise to undesirable heights, especially if sufficient arginine is available for protein production. The role of ArgX could be to stop the uptake and catabolism of arginine by inducing the degradation of arc transcripts. In a less critical situation or perhaps under conditions with high arginine in combination with low glucose, ArgX could redirect the carbamoyl phosphate towards pyrimidine metabolism by blocking translation of arcC1, the carbamate kinase that converts carbamoylphosphate into ammonia, ATP and CO 2 . It has been shown that disruption of arginine regulation in argR and ahrC mutants of L. lactis increases the activity and gene expression of the de novo pyrimidine enzymes PyrE and PyrF [18]. ArgX could function to control the way arginine is utilized by the cell: as a precursor for pyrimidine synthesis or for ATP and ammonia production.

sRNAs that regulate arginine metabolism in other organisms
Homologs of ArgX were not detected in genomes of other bacterial species. As this search was performed at the level of nucleotide sequence identity we cannot rule out the possibility that ArgX genes are present in other, more distant species. An sRNA was found to be involved in arginine metabolism in Bacillus subtilis. In this soil bacterium, the sRNA SR1 blocks the translation of the transcriptional activator ahrC, resulting in a decrease of the arginine catabolite gene clusters rocABC and rocDEF [53,54]. Repression by SR1 on ahrC has similar consequences on arginine catabolism in B. subtilis as does ArgX in L. lactis, although it takes place at an earlier stage of growth and without the possibility to redirect arginine side-products. Interestingly, SR1, like ArgX, is also repressed by CcpA. Whether the L. lactis genome harbors an SR1-homolog, whether the rocR transcript from B. subtilis contains an ArgX-homolog or whether ArgX and SR1 are functional homologs are questions still to be answered. In E. coli, an antisense RNA against the argR transcript was identified using dRNA-seq [55]. This transcript, as-argR was specifically expressed at an OD 600 of 0.4, while it was not detected in cells of a culture at an OD 600 of 2.0. Although no functional analyses were performed, it is expected that as-argR targets and helps degrading the argR mRNA.

Conclusions
The 3'-UTR region of the L. lactis argR gene, coding the arginine repressor ArgR, harbors a non-coding small RNA regulator gene called ArgX. This sRNA downregulates mRNA levels of arc, the operon that specifies the arginine deiminase pathway. ArgX adds another layer to the complex regulation of arginine metabolism. In Fig 5, we present an overview of the current model of arginine metabolism regulation in L. lactis. Besides regulation of arc by various protein transcription factors [17][18][19], ArgX allows for post-transcriptional regulation by acting on arc transcript stability and/or translation of arcC1. By blocking arcC1 translation, carbamoyl phosphate can be directed towards pyrimidine metabolism. We propose that ArgX ensures arginine homeostasis by interfering with the breakdown of intracellular arginine into ammonia, CO 2 and ATP. The expression of ArgX (and arc) depends on relief of carbon catabolite repression by the preferred carbon source via CcpA, and on arginine availability to release repression by ArgR/AhrC. Under low arginine conditions, expression of arc is low such that the amino acid can be used for protein synthesis. When arginine is abundant, arc is highly expressed. However complete breakdown of arginine might not always be preferred. Re-direction of carbamoyl phosphate towards pyrimidine metabolism allows saving energy. While the conversion of carbamoyl phosphate by ArcC yields 1 molecule of ATP, the production of carbamoyl phosphate for the synthesis of pyrimidines from glutamine consumes 2 ATP molecules. Also, if all arginine would be fully converted, protein synthesis might fall short. We hypothesize that ArgX, of which expression is based on carbon source and arginine availability, can steer arginine towards ATP and ammonia, to pyrimidine metabolism or to protein synthesis.
Altogether, ArgX is a fascinating example of how an RNA from a 3'-UTR region can function as an RNA regulator molecule that is regulated by and intertwined with the function of the gene product of the parental mRNA, in this case argR.
Supporting information S1