Antigenic variation by switching inter-chromosomal interactions with an RNA splicing locus in trypanosomes

Highly selective gene expression is a key requirement for antigenic variation in several pathogens, allowing evasion of host immune responses and maintenance of persistent infections. African trypanosomes, parasites that cause lethal diseases in humans and livestock, employ an antigenic variation mechanism that involves monogenic antigen expression from a pool of >2500 antigen coding genes. In other eukaryotes, the expression of individual genes can be enhanced by mechanisms involving the juxtaposition of otherwise distal chromosomal loci in the three-dimensional nuclear space. However, trypanosomes lack classical enhancer sequences or regulated transcription initiation and the monogenic expression mechanism has remained enigmatic. Here, we show that the single expressed antigen coding gene displays a specific inter-chromosomal interaction with a major mRNA splicing locus. Chromosome conformation capture (Hi-C), revealed a dynamic reconfiguration of this inter-chromosomal interaction upon activation of another antigen. Super-resolution microscopy showed the interaction to be heritable and splicing dependent. We find that the two genomic loci are connected by the antigen exclusion complex, whereby VEX1 associated with the splicing locus and VEX2 with the antigen coding locus. Following VEX2 depletion, loss of monogenic antigen expression was accompanied by increased interactions between previously silent antigen genes and the splicing locus. Our results reveal a novel mechanism to ensure monogenic expression, requiring the spatial integration of antigen transcription and mRNA splicing in a dedicated compartment. These findings suggest a new means of post-transcriptional gene regulation.

specific genome organization is required for monogenic VSG expression. The single active 84 VSG gene is transcribed in an extranucleolar Pol I compartment known as the expression site 85 body 11 . In those very rare cases (<10 -8 ) where two VSG genes are simultaneously active, 86 both co-localize at the expression site body 12,13 . In addition, the transcribed chromosome core 87 regions and the sub-telomeric regions coding for the large reservoir of silent VSG genes, 88 appear to fold into structurally distinct compartments 5 , similar to active A and silent B 89 compartments described in mammalian cells 14 . While the nature of the expression site body 90 has remained enigmatic, a protein complex specifically associated with the active VSG gene 91 was identified recently. VSG-exclusion 1 (VEX1) emerged from a genetic screen for allelic 92 exclusion regulators 15 while VEX2 was affinity-purified in association with VEX1 16 . The 93 bipartite VEX protein complex maintains mutually exclusive VSG expression 16 but it remains 94 unclear how these proteins exert their function. In this study we aimed to identify the 95 mechanism that connects RNA maturation, genome architecture and the VEX complex to 96 ensure monogenic antigen expression. Given the well-characterized role of promoter-enhancer interactions in the selective 98 regulation of genes, we set out to identify specific DNA-DNA interactions with a regulatory role 99 in monogenic VSG expression. To this end we used a T. brucei culture homogenously 100 expressing a single VSG gene for chromosome conformation capture (Hi-C) analysis. In 101 addition, we employed the mHi-C analysis pipeline, which allowed us to retain many multi-102 mapping reads and greatly increased the read coverage across repetitive regions of the 103 genome 17 . 104 In order to visualize specific interaction patterns of loci of interest (viewpoints) in the 105 Hi-C dataset, we applied a virtual 4C analysis pipeline to extract genome-wide interaction 106 profiles for chosen viewpoints. To identify VSG gene specific interaction patterns, we chose 107 the active and several inactive VSG genes located in expression sites as viewpoints and 108 plotted the extracted virtual 4C interaction data onto the genome. As expected, we observed 109 a distance-dependent decay of intra-chromosomal interactions between each viewpoint and 110 its upstream and downstream genomic region ( Fig. 1a and Extended Data Fig. 1a). 111 Strikingly, we found the active VSG-2 gene located on chr. 6 in expression site 1 to 112 very frequently interact with a single, distinct locus on chr. 9 ( Fig. 1a-b). Levels of interaction 113 frequency were higher than intra-chromosomal interactions of VSG-2 with its genomic location 114 on chr. 6, pointing to a strong and stable inter-chromosomal interaction. The locus on chr. 9 115 interacting with the active VSG gene is the SL-RNA array, a genomic locus essential for RNA 116 maturation. This locus contains a cluster of ~150-200 tandemly repeated genes encoding the 117 spliced leader RNA (SL-RNA). SL-RNA is an RNA Pol II-transcribed ncRNA that is trans- 118 spliced to the 5´-end of all trypanosome mRNAs, conferring the 5´-cap structure required for 119 RNA maturation, export and translation 8 . Conversely, VSG genes residing in inactive 120 expression sites interacted less frequently or at background levels with the SL-RNA locus 121 ( Fig. 1a-b and Extended Data Fig. 1a). In agreement with these observations, when we 122 chose the SL-RNA locus as viewpoint, we found it to interact more frequently with the active 123 6 VSG expression site than with any inactive VSG expression site (Extended Data Fig. 1b). 124 Thus, the Hi-C analysis revealed a strong and selective interaction between the Pol I- 125 transcribed active VSG gene and the Pol II-transcribed SL-RNA locus located on a different 126 chromosome. 127 To visualize the spatial proximity between the active VSG gene and the SL-RNA locus 128 at the level of individual cells and with an independent assay, we performed super-resolution the core of chr. 9 (Fig. 1c). By scoring nuclei for overlapping, adjacent or separate VSG and 138 SL-RNA transcription compartments, we found that one of the SL-RNA transcription 139 compartments was adjacent to the VSG transcription compartment in the majority of cells (Fig.   140 1c). However, during DNA replication in S phase, the VSG and SL-RNA transcription 141 compartments were detected in separate locations in >50% of nuclei ( Fig. 1c and Extended 142 Data Fig. 1c). Therefore, throughout this study, immunofluorescence assays (IFAs) were 143 subsequently performed in G1 cells, unless indicated otherwise. Taken together, IFAs 144 supported the findings made by Hi-C, suggesting that the Pol I VSG transcription compartment 145 interacts with one of the Pol II SL-RNA transcription compartments. Further, they suggest that 146 the interaction between both compartments is resolved during S phase and successfully re-  representative of two independent biological replicates (≥100 G1 or S phase nuclei); error bars, SD. Detailed n and 165 p values are provided in Data S1 sheet 3. DNA was counter-stained with DAPI; the images correspond to maximal 166 3D projections of stacks of 0.1 μm slices; scale bars 2 μm. N, nucleus; K, kinetoplast (mitochondrial genome).

168
To determine whether the interaction with the SL-RNA transcription compartment is 169 specific for the active VSG gene, and therefore changes following a VSG switching event, we 170 performed Hi-C experiments using an isogenic T. brucei cell line expressing a different VSG 171 isoform, VSG-13 (Fig. 2a) 19 . VSG-13 resides within expression site 17, which is located on to specifically select for parasites expressing VSG-13 through drug selection (Fig. 2a). The 175 exclusive activity of expression site 1 or 17 was verified by RNA-seq (Fig. 2a). 176 Hi-C analysis revealed that VSG-2 -SL-RNA interactions dropped 20-fold to average 177 inter-chromosomal interaction levels in parasites expressing VSG-13, while interactions 178 between the newly activated VSG-13 and the SL-RNA locus increased 36-fold (Fig. 2b, left   179 and middle panel). We found that upon activation of each expression site, the bin harboring  To further explore the relationship between SL-RNA interaction frequency and gene 189 expression, we performed Hi-C analyses using insect stage parasites that do not express any 190 VSGs, but instead express a different group of surface antigens called procyclin genes.

191
Confirming the importance of the SL-RNA interaction, the GPEET and EP1 procyclin genes

208
Previously, we had shown that the bipartite VEX-complex is associated with the 209 actively transcribed VSG gene and maintains monogenic VSG expression but that VEX1 and 210 VEX2 only partially overlap each other 16 . Given a similar juxtaposition of the VSG transcription 211 and the SL-RNA transcription compartments, we sought to investigate the relationship 212 between the VEX complex and these transcription compartments in more detail. Using 213 optimized immunofluorescence staining protocols and super-resolution microscopy, we were 214 able to detect two VEX1 foci in the majority of G1 cells (55 +/-4 %). These VEX1 signals 215 specifically co-localized with the SL-RNA transcription compartments (Fig. 3a). In contrast, 216 the majority of G1 cells (97 +/-1 %) only had one VEX2 focus, which specifically co-localized 217 with the VSG transcription compartment (Fig. 3b). As expected, one VEX1 focus was adjacent  Although VSG and SL-RNA transcription compartments separate during S phase ( Fig.   228 1c), VEX1 does not separate from the SL-RNA transcription compartment and VEX2 does not 229 separate from the VSG transcription compartment ( Fig. 3a and b). Also, consistent with the of at least two independent biological replicates and detailed n and p values are provided in Data S1 sheet 3.

258
Given the close spatial proximity between the site of VSG transcription and the site of 259 SL-RNA transcription, we next questioned whether the splicing process itself impacts the 260 connection between these compartments. We found that inhibition of trans-splicing with 261 sinefungin 20 disrupted both VEX1 (Fig. 3d) and VEX2 (Fig. 3e) localization within 30 min, 262 while the tSNAP transcription factor was not affected under the same conditions (Fig. 3f). 263 Notably, inhibition of splicing by sinefungin also disrupted the connection between the VSG 264 and SL-RNA transcription compartments, revealed by separation of the Pol I and tSNAP 265 signals (Fig. 3g); neither the VEX, nor the tSNAP protein levels were affected by sinefungin 266 treatment (Extended Data Fig.3f). Thus, VEX protein localization and the juxtaposition of the 267 VSG and the SL-RNA transcription compartments are dependent on mRNA splicing activity.

268
Next, we aimed to investigate the mechanism by which the VEX complex ensures 269 monogenic VSG expression. Previously, we found that VEX2 depletion leads to a strong     To explore the role of VEX2 in controlling interactions between antigencoding genes 313 and SL-RNA loci, we performed Hi-C analyses in VEX2-depleted cells. After 24 hours of VEX2 314 depletion, all previously silent expression sites displayed increased interaction frequencies 315 with the SL-RNA locus. (Fig. 4c). The interaction between VSG expression site 3 and the SL-

321
Besides the VSG genes located in previously silent expression sites, expression site 322 associated genes were also strongly upregulated following VEX2 knockdown 16 . In line with 323 this finding, we observed the largest increase in SL-RNA interactions for the regions upstream 324 of the VSG gene in each de-repressed expression site, where expression site associated 325 genes are located (Fig. 4d, Extended Data Fig. 5b). Thus, VEX2 restricts interactions 326 between silent VSG expression sites, the expression site associated genes in particular, and 327 the SL-RNA locus.

328
As a third group of RNA Pol l transcribed genes, insect stage specific procyclin genes 329 are upregulated upon VEX2 depletion 16 . Correlating with these data, following VEX2-330 knockdown, we found GPEET and EP1 procyclin genes to exhibit strongly increased 331 interaction frequencies with the SL-RNA array and also with VSG expression sites (Fig. 4e,   332 Extended Data Fig. 5c). Thus, our data suggest that VEX2 may have a dual function: inter-chromosomal interaction, bringing together two nuclear compartments to ensure efficient 344 VSG mRNA processing at only one expression site (Fig. 4f). In the VSG transcription           Fig. 3c) were generated using the R library circlize 37 and bedgraph 530 files for log2 fold change (Fig. 3c, Extended Data Fig. 3c and Extended Data Fig. 6b) were 531 generated using deeptools2 38 . Bedgraphs were generated with 1kb bins and the option 532 smoothLength 5000. Spliced leader RNA sequences were annotated using the sequences: between the ESB and tSNAP compartment (Fig. 4c), a control measurement (Extended Data 602 Fig. 4d) was performed to make sure that the increase in the distance between the two protein 603 condensates following VEX2 or VEX1 / VEX2 RNAi was not a mere consequence of a 604 decrease in the ESB focus diameter. Moreover, the ESB / tSNAP localisation analyses 605 following VEX RNAi or sinefungin treatment were restricted to G1 cells to exclude any cell 606 cycle bias, as these protein condensates can separate during S phase (Fig. 1c).