Identification of regulatory genes through global gene expression analysis of a Helicobacter pylori co-culture system

Helicobacter pylori is a gram-negative bacterium that establishes life-long infections by inducing immunoregulatory responses. We have developed a novel ex vivo H. pylori co-culture system to identify new regulatory genes based on expression kinetics overlapping with that of genes with known regulatory functions. Using this novel experimental platform, in combination with global transcriptomic analysis, we have identified five lead candidates, validated them using mouse models of H. pylori infection and in vitro co-cultures under pro-inflammatory conditions. Plexin domain containing 2 (Plxdc2) was selected as the top lead immunoregulatory target. Gene silencing and ligand-induced activation studies confirmed its predicted regulatory function. Our integrated bioinformatics analyses and experimental validation platform has enabled the discovery of new immunoregulatory genes. This pipeline can be used for the identification of genes with therapeutic applications for treating infectious, inflammatory, and autoimmune diseases.


142
The WT group showed a minimal increase at all timepoints relative to time 0 ( Figure  1F). In contrast,

143
WT macrophages displayed a dramatic increase of the anti-inflammatory cytokine IL-10 at 60 min 144 post-H. pylori co-culture, that was significantly diminished in LysCre+ BMDM ( Figure  1E). Therefore,

145
H. pylori promotes the activation of cytokine-driven regulatory mechanisms in WT macrophages, that 146 modulate the immune response and generate a regulatory microenvironment that facilitates bacterial 147 proliferation reaching the highest peak at 2 hours post gentamycin treatment.

149
To elucidate the underlying regulatory molecular mechanisms activated upon H. pylori co-culture in 150 WT macrophages, we performed a global transcriptomic analysis on a gentamycin protection assay 7 time course (0, 60, 120, 240, 360 and 720 min). RNAseq analysis demonstrated important differences 152 in the gene expression profile within both genotype ( Figure  1C) and treatment ( Figure  1D). Almost 153 50% of genes exhibited a significant differential expression based on the treatment, with a substantial 154 upregulation after H. pylori challenge. Thus, H. pylori strongly influences the macrophage transcription 155 profile, resulting in drastic modifications in macrophage function that favor the generation of a 156 regulatory phenotype. Therefore, we sought to utilize the new co-culture system to explore novel 157 regulatory pathways activated upon H. pylori infection to discover new host regulatory genes that 158 modulate the immune response, with the potential to become molecular targets for the development 159 of therapeutics for infectious, inflammatory and autoimmune diseases.

161
Validation of experimental co-culture system with identification of differential expression 162 patterns in characterized antimicrobial genes.

163
Expression of regulatory and pro-inflammatory genes is tightly regulated and coordinated over time.      Initially, we performed an analysis based on the fold change of gene expression between genotypes 209 for each gene of both NLR and PPAR pathways across the entire time course presented in the form 210 of heat maps (Figure 3A, D) where blue represents genes downregulated in WT compared to 211 LysCre+, while red represents upregulation of gene expression in WT related to LysCre+. We 212 anticipated the presence of inverted patterns of regulatory and pro-inflammatory genes between both 213 genotypes, where regulatory genes would be increased in WT compared to PPARg-deficient, and pro-214 inflammatory genes overexpressed in PPARg-deficient group. Indeed, the bioinformatics analysis 215 revealed specific expression patterns in both signaling pathways that were clustered in groups. The

216
NLR pathway includes two well-defined clusters based on the distinct genotype gene expression 217 ( Figure  3A). The orange box, at the top, contains genes upregulated in LysCre+ macrophages, and 218 the green box, at the bottom, contains a second class of genes with greater expression in the WT 219 group. Interestingly, the LysCre+ upregulated genes have a delayed expression pattern, while, WT 220 upregulated genes presented an earlier peak. The two NLR clusters are represented in the PPAR 221 pathway, also depicted in orange and green boxes ( Figure  3D). The analysis of genes associated 222 with PPAR revealed an additional third cluster, highlighted in purple, including a group of genes 223 characterized by a dysregulated pattern. Particularly, those genes exhibited oscillating expression 224 kinetics in each genotype among the entire time course. A plausible explanation is the existence of a 225 strong PPARg interaction with these genes. Therefore, the absence of the transcription factor in 226 LysCre+ macrophages could alter the expression of the genes, due to direct activation, inhibition or 227 even due to the upregulation of compensatory mechanisms, that result into a fluctuating expression 228 pattern.

230
Gene selection was narrowed to the green box cluster (i.e. genes upregulated in WT), to include 231 genes with a positive response to H. pylori in WT macrophages that resembled the peak of bacterial 232 loads reported in this genotype ( Figure  1A). The choice included 7 NLR ( Figure  3B) and 10 PPAR 233 ( Figure 3E) pathway genes highlighted in red, which were defined as seed genes. Based on the 10 expression patterns of the seed genes, we built an initial dataset that comprised both these original 235 genes and a group described as linked genes obtained from the global transcriptome dataset. Linked

249
Based on the pattern analysis, five genes from groups 1 and 2 of the final dataset were identified as 250 potential new regulatory leads for further validation. Plexin domain containing 2 (Plxdc2, Figure  4A),

251
V-set and immunoglobulin domain containing 8 (Vsig8, Figure 4C), Ankyrin repeat domain 29 252 (Ankrd29, Figure 4D) and C1q and tumor necrosis factor related protein 1 (C1qtnf1, Figure 4E) share 253 an early expression peak in WT macrophages, abrogated in LysCre+, that coincides with the bacterial 254 burden spike in the gentamycin protection assay. The kinetics of Protein phosphatase 1 regulatory 255 subunit 3E (Ppp1r3e, Figure  4B) is slightly different since it was still upregulated in the last timepoint.

267
To perform a validation of the five selected genes, we initially measured their expression, by qRT-

294
To further characterize the potential regulatory functions of the selected genes, we sought to assess 295 their behavior under inflammatory conditions in a controlled environment in vitro.

299
In contrast, the IL-10-induced peak reported in WT macrophages is abrogated by the lack of 300 PPARg ( Figure 6H)  PPARg-deficient macrophages. However, no differences were reported for Ankrd29 ( Figure  6D). As

504
Samples were adjusted to 7.5x10 5 cells/mL with cold cRPMI_M supplemented with 25 ng/mL of 505 recombinant mouse colony-stimulating factor (m-csf, Peprotech, Rocky Hill, NJ) and cultured at 37ºC, 506 5% CO 2 and 95% humidity to allow their differentiation. At day 3 fresh m-csf-supplemented media 507 20 was added. On day 6, plates were washed to remove non-adherent cells and BMDM were harvested.

508
Cells were re-suspended in cRPMI_M and seeded in triplicate in 12-well plates (5x10 5 cells per well).

580
Hierarchical based clustering was employed to obtain differentially expressed patterns within the 581 combined initial NLR and PPAR genes and the linked dataset generated. The hclust method with 582 Ward's minimum variance method and Manhattan distance metric in R were used to cluster the data.