An in vivo brain-bacteria interface: the developing brain as a key regulator of innate immunity

The number of XL2- or CC3-positive cells was quantified and normalized to the tail area in pixels2 using ImageJ software. Leukocytes and apoptotic cells were counted along the 2/3 posterior of the tail (Fig. 4a). To quantify number of GFP-positive cells (myeloid cells) and density of peripheral neural network in xlurp::GFP embryos, tails were subdivided in two independent regions: center and periphery (Fig. 4k, l). The somatic muscle or chevron-shaped myotomes composed ‘center’ region; ‘peripheral’ region was referred to the fin, both dorsal and ventral to the myotomes. Peripheral neural network was evaluated on images of stage 46–48 xlurp::GFP embryos that were immunoreacted against Tub, using gray-level measures (OD) as previously described.36,84–88 Two OD-mean values per animal were calculated after multiple measurements taken along the anteroposterior axis: one for center or somite region and one for peripheral fin. Analysis was done on raw black-and-white 8-bit images. Each measurement consisted of the mean value of the pixels of a fixed-size window. The size of the window remained constant across subjects. OD values ranging from 0 (black, no expression) to 255 (white, maximal expression). All individuals among whom comparisons are being made were produced in the same batch, treated identically for processing and imaging conditions were not changed. In addition, any potential aberrations originating from the optical system were corrected by background image subtraction.

Following infection and surgeries at st. 25 (brain removal, spinal cut, or tail amputation), embryos were grown for 3 h at 21 °C, then their anterior parts, including brain (to ensure that differences in RNA-Seq signature among samples were not coming from the brain tissue itself), were amputated and flash-frozen in groups of 20. RNA-Seq analyses were performed by The MIT BioMicro Center (Boston, MA). Samples were quality controlled on a Fragment Analyzer (Advanced Analytical) to determine DV200 scores. The mRNA from 1 μg of total RNA was isolated using Illumina human/mouse/rat RiboZero Gold and prepared into Illumina libraries using the Kapa RNA HyperPrep kit and 10 cycles of amplification. Final libraries were pooled, and quality controlled using qPCR (Roche LC480II) as well as the Fragment Analyzer and sequenced on HiSeq2000.

Alignments were conducted by Genotypic Technology (India). Eight samples were aligned to the reference transcriptome Xenopus_laevis_v2, downloaded from NCBI [https://www.ncbi.nlm.nih.gov/genome/?term=xenopus+laevis]. There were two biological replicates per treatment (brain intact, not infected or Ctrl NI; brain intact infected or Ctrl UTI; brainless, not infected or BR– NI; brainless, infected or BR– UTI). The raw data from Illumina were checked for quality using FastQC1 and pre-processed, which included removing the adapter sequences and removing the low-quality bases (<q30). Pre-processing of data was done with Cutadapt2. Mapping was performed with HISAT2 to align the high-quality data to the reference genome with the default parameters. Reads are classified into aligned reads (which align to the reference genome) and unaligned reads. HISAT2 is a fast and sensitive alignment program for mapping next-generation sequencing reads (both DNA and RNA) to a single reference genome). Based on an extension of BWT for graphs,89 a graph FM index (GFM) was implemented. HISAT2 uses a large set of small GFM indexes that collectively cover the whole genome. These small indexes (called local indexes), combined with several alignment strategies, enable rapid and accurate alignment of sequencing reads. This new indexing scheme is called a Hierarchical Graph FM index (HGFM).

Cufflinks was used to calculate transcript abundance. It results in normalized read count in the form of FPKM values. FPKM is a unit of measuring gene/transcript expression. Four comparisons were conducted (1) Ctrl NI vs. Ctrl UTI, (2) BR– NI vs. BR– UTI, (3) Ctrl NI vs. BR– NI, and (4) Ctrl UTI vs. BR– UTI. Cufflinks includes a script called “Cuffmerge” that can be used to merge several Cufflinks assemblies together. The main purpose of this script is to develop an assembly GTF file suitable for use with Cuffdiff. Merged GTF files produced by Cuffmerge were used as input in Cuffdiff. Cuffdiff was used to calculate the differentially expressed transcripts and generated p-values, FDR corrected p-values (q value) and the log2fold change values. RNA-Seq unprocessed and processed data have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) and is accessible through GEO Series accession number GSE119729. Supplementary Data 1 contains all processed transcript data, including fold change, p-value, and q-value.

Sub-network enrichment analysis (SNEA) was used to identify networks enriched in each of the four comparisons. Analyses were conducted in Pathway Studio 10.0 (Elsevier) using the ResNet 11.0. The number of gene symbols that mapped to the mammalian homologs in the program ranged from ~12,300 to 12,500 for each of the four datasets using the official gene Name + Alias. Duplicates were addressed using the default “best p-value if present, or maximum fold change” in Pathway Studio. Sub-networks related to cell process were queried and there were 1000 permutations of the data using Kolmogorov–Smirnov algorithm to generate the distributions. SNEA identifies gene networks related to cellular processes or diseases that change with a treatment or disease; these networks are pre-defined molecular networks (expression patterns, binding, or involvement in common pathways) based upon the literature that are focused around gene hubs. All transcripts identified by RNA-seq were used in the analysis and acted as the background list for enrichment. See Supplementary Methods 1 for Reads and quality of sequencing.

Liquid chromatography–mass spectrometry (LC–MS/MS): UTI-infected Ctrl and BR– embryos were harvested at early st. 35 (~20 h post-surgery, 44 h post infection) immediately before the peak of infection-induced death in BR– animals and the significant differences both in survival rate and apoptosis are reached respect to Ctrl embryos (experimentally defined in advance; see Fig. 2a, b). Three biological replicates per experimental group were analyzed for DA level quantification. Each replicate consisted of 30 embryos. Their anterior parts, including brain, were amputated to ensure that differences in DA levels among samples were not coming from the brain tissue itself, and then flash-frozen in groups of 30 embryos. LC–MS/MS was performed by The Small Molecule Mass Spectrometry in the Faculty of Arts and Sciences at Harvard University (Cambridge, MA). LC–MS grade dopamine hydrochloride (DA) and Dopamine-1,1,2,2-d4 hydrochloride (d4) were purchased from Millipore Sigma (cat. D-081, 73483). The working standard (DA, 0.61 ng/mL) and working internal standard solutions (DA-d4, 50 ng/mL) were diluted in LC-MS grade water. Sample preparation and protein precipitation were performed by adding 50 μL of 100 pg/μL of DA-d4 and 200 μL of cold methanol. The LC–MS/MS analysis was done on an Agilent 6460 Triple-quad mass spectrometer (Agilent Technologies) coupled to an Agilent 1290 uHPLC. The chromatographic separation was performed using an Agilent Eclipse XDB (2.1 mm internal diameter × 50 mm length × 2.7 μm particle size) C18 column. A constant flow rate of 0.400 L/min was used, along with a 10 μL injection volume. Mobile phase A was 0.1% formic acid in water v/v and B was 0.1% formic acid in acetonitrile v/v. Initial conditions were 98% A, 2% B.

Compounds were eluted by increasing mobile phase composition to 98% B over 5 min, and the columns re-equilibrated to starting conditions for 3 min prior to the next injection. Targeted analysis using multiple reaction monitoring (MRM) monitoring two transitions for DA and DA-d4 using the following transitions: DA: 154→137 and 154→91, and DA-d4: m/z 158→141 and 158→95. The optimized collision energy was 9 and 25 V, respectively, for the first and second transition. A fragmentor value of 90 V was employed. The standard curve was measured between 0.1 pg/μL and 50 pg in water. The quantitative analysis was performed in MassHunter Quantitative Analysis software (Agilent Technologies, Santa Clara, USA). Data were collected from compounds with a S/N ratio of >10. A linear curve fit with a 1/x weighting was used. See Supplementary Methods 2 for LC–MS/MS conditions and parameters.

Drug Exposure. Uninfected and UTI-infected Ctrl and BR− Xenopus embryos were exposed to specific pharmacological agents, targeting the type-1 or type-2 family of dopamine (DA) receptors (D1R, D2R), from st. 25 (immediately after brain removal) to st. 48. The drugs were refreshed every day. We used specific agonists and antagonists of each dopamine receptor: 10 μM SKF-38393, D1R agonist (SKF, Tocris 0922); 10 μM SCH-23390, D1R antagonist (SCH, Tocris 0925); 10 μM Quinpirole, D2R agonist (Quin, Tocris 1061); 1 μM L-741,626, D2R antagonist (L741, Tocris 1003). All drug treatments were performed using embryos from mixed batches of fertilizations and using three biological replicates per drug. Each replicate consisted of 30 embryos. Stock solutions of SKF, SCH and Quin were created by dissolving the compound in Millipore water to a final drug concentration of 100 mM for both SCH and Quin, and to 25 mM for SKF. Stock solution of L741 was created by dissolving the compound in DMSO to a final drug concentration of 100 mM. All aliquots were stored at −20 °C. Further dilution of all compounds was made in 20 ml of normal frog media (0.1X MMR). Control experiments were performed using embryos in 0.1X MMR (no drug), both for Ctrl and BR− groups. Drug concentrations were determined through toxicity screens and were applied at levels that did not result in lethality or observable developmental defects.

All statistical analysis was performed using GraphPad Prism (GraphPad Software, Inc., CA, USA). Animal numbers were chosen based on previous studies as well as sample size estimates given a type I error rate of 5% and power of 0.8. Each dish of tadpoles was considered a replicate. After applying Bartlett’s test (for equal variances), data from various replicates and multiple independent groups were analyzed by one-way or two-way ANOVA test (if equal variances) or Kruskal–Wallis test (if significantly different variances). When P < 0.05, post hoc multiple comparisons were performed by Bonferroni or Dunn test, respectively. The significance level (α) was set to 0.05 in all cases. P value on graphs is after post hoc Bonferroni or Dunn analysis. The statistical values are reported as mean ± SD. Number of replicates (r), number of animals per replicate (n), number of total animals (N = r × n), and specific analysis used for each experiment are stated in the “Results” section and/or figure legends.

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

We thank Erin Switzer and Rakela Colon for Xenopus husbandry and general laboratory assistance; Matthew M. Mulvey (University of Utah) for providing the E. coli strains; Enrique Amaya (University of Manchester) for providing the spiba and mmp7 probes; Joan Lemire for assistance with molecular biology; Vikas Trivedi and Todd Chappell for assistance with E. coli preparation; Kelly McLaughlin and Joshua Finkelstein for their helpful comments on the paper; Kelly Chatman and Sunia Truger for performing the LC–MS/MS experiments. XL-2 antibody was a generous gift from Makoto Asashima’s laboratory at Tokyo University. The methods were performed in “accordance” with relevant guidelines and regulations. We gratefully acknowledge support of the Templeton World Charity Foundation Independent Research Fellowship to C.H.-R. (TWCF0241) and the Allen Discovery Center program through The Paul G. Allen Frontiers Group (12171), as well as DARPA (W911NF-16-C-0050), and the NIH (AR055993, AR061988).

C.H.-R. and J.-F.P. performed experiments. C.J.M. analyzed RNA-seq data and prepared several figures and tables. M.L., C.H.-R. and J.-F.P. designed the experiments and interpreted data. S.K.J., C.H., and A.F. assisted with the immunofluorescence experiments and cell counting. V.K. and A.D. performed bacterial injection and analysis. C.H.-R., C.-J.M., J.-F.P. and M.L. wrote the paper together.

The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information Files or from the corresponding author upon reasonable request. RNA-seq data have been deposited into the NCBI Gene Expression Omnibus (GEO) database (Accession number: GSE119729).

Supplementary information is available for this paper at 10.1038/s41536-020-0087-2.

The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information Files or from the corresponding author upon reasonable request. RNA-seq data have been deposited into the NCBI Gene Expression Omnibus (GEO) database (Accession number: GSE119729).