Mycobacterium tuberculosis cording in the cytosol of live lymphatic endothelial cells

The ability of Mycobacterium tuberculosis to form serpentine cords is intrinsically related to its virulence, but specifically how M. tuberculosis cording contributes to pathogenesis remains obscure. We show that several M. tuberculosis clinical isolates form intracellular cords in primary human lymphatic endothelial cells (hLEC) in vitro and also in the lymph nodes of patients with tuberculosis. We identified via RNA-seq a transcriptional programme in hLEC that activates cellular pro-survival and cytosolic surveillance of intracellular pathogens pathways. Consistent with this, cytosolic access of hLEC is required for intracellular M. tuberculosis cording; and cord formation is dependent on the M. tuberculosis ESX-1 type VII secretion system and the mycobacterial lipid PDIM. Finally, we show that M. tuberculosis cording is a novel size-dependent mechanism used by the pathogen to evade xenophagy in the cytosol of endothelial cells. These results provide a mechanism that explains the long-standing association between M. tuberculosis cording and virulence.


Introduction 34
Mycobacterium tuberculosis is one of the most successful bacterial pathogens of humankind 35 and still constitutes a global health challenge (WHO, 2017). A striking phenotype of M. 36 tuberculosis growing in nutrient broth is the ability of this pathogen to form serpentine 37 cords, a morphological observation originally described by Robert Koch (Koch, 1882). This 38 cording phenotype is intimately associated with virulence and immune evasion (Glickman et 39 al., 2000). The first morphological descriptions of M. tuberculosis growth in liquid and solid 40 media described a distinct ability of tubercle bacilli to form large and elongated structures 41 by Middlebrook, Dubos and Pierce in the mid-1940s (Middlebrook et al., 1947. Cording is a 42 complex phenotype involving many mycobacterial factors including lipids such as the "cord-43 factor" glycolipid trehalose dimycolate (TDM) (Hunter et al., 2006a;Hunter et al., 2006b;44 Indrigo et al., 2002) and a series of chemical modifications such as cyclopropanation of 45 mycolic acids in the cell wall (Glickman et al., 2000). 46 47 Similar cording has been reported in other pathogenic mycobacteria, primarily in liquid 48 media or extracellularly in various cell and organism models of infection. In zebrafish, M. 49 abscessus released from apoptotic macrophages grows extracellularly, forming cords 50 (Bernut et al., 2014). It is postulated that apoptosis of infected macrophages is a key event 51 in the release of extracellular bacteria and subsequent initiation of cord formation. There 52 are, however, a few reports showing that cording can also occur intracellularly. In 1928, 53 Maximow and co-workers first reported intracellular cording in tissue culture (Maximow, 54 1928). In 1957, Shepherd studied this phenomenon in HeLa cells and found that only fully 55 virulent M. tuberculosis strains formed cords. Moreover, Ferrer and co-workers (Ferrer et 56 al., 2009) showed that an attenuated mutant of M. tuberculosis formed cords in fibroblasts. 57 58 Overall, extracellular cording has been shown in mycobacteria to be anti-phagocytic and to 59 be a trigger of extracellular trap formation in macrophages (Bernut et al., 2014;Kalsum et 60 al., 2017;Wong and Jacobs, 2013). Although proposed as a virulence mechanism, this does 61 not explain why an intracellular pathogen such as M. tuberculosis would prefer to replicate 62 in cords in the relatively nutrient poor extracellular space to avoid phagocytosis. 63 Here we discovered that M. tuberculosis forms large intracellular cords consisting of up to 86 thousands of individual bacteria arranged end-to-end in hLEC in vitro and in biopsies of 87 tuberculosis patients. Intracellular cording is common to all tested clinical isolates and 88 'virulent' lab strains of wild-type M. tuberculosis that had not lost the ability to produce 89 phthiocerol dimycocerosates (PDIMs) during laboratory sub-culturing. We identified a 90 transcriptional signature from the host consistent with M. tuberculosis membrane damage 91 and escape from the phagosome into the cytosol and used correlative light electron 92 microscopy (CLEM) to determine that intracellular cords are formed of chains of individual 93 M. tuberculosis which are only present in the host cell cytosol. M. tuberculosis mutants 94 lacking ESX-1 or PDIMs that cannot access the cytosol are incapable of cording unless co-95 We next sought to understand M. tuberculosis factors that contributed to the intracellular 160 cording phenotype in hLEC. We have previously shown that the ESX-1 secretion system, 161 encoded in the RD1 genomic region, and the cell wall lipid phthiocerol dimycocerosate 162 (PDIMs) are required for intracellular replication of M. tuberculosis in hLEC (Lerner et al., 163 2016;Lerner et al., 2018). Infection with the M. tuberculosis ΔRD1 mutant that lacks the 164 ESX-1 secretion system was not able to form cords but instead exhibited smaller clumps of 165 bacteria sometimes with a mesh-like appearance (Fig. 3a). The phenotype of the M. 166 tuberculosis mutant lacking PDIM also presented a clumpy mesh-like phenotype with an 167 increased number of individual bacteria that were not organised in cords (Fig. 3a). M. 168 tuberculosis mutants lacking either the ESX-1 secretion system or the virulence-related lipid 169 PDIM (Astarie-Dequeker et al., 2009) failed to cord intracellularly (Fig. 3b). The lack of 170 cording observed with the RD1 mutant was not due to the reduced bacterial burden, since 171 increasing the multiplicity of infection did not increase cord formation although significant 172 bacterial growth was observed (Fig. 3c, d, e, f). Moreover, we found that the up-regulation 173 of some genes in hLEC after infection (Fig. 2b) such as interferon-beta (IFN-) or interleukin-174 6 (IL-6) was RD1 and PDIM dependent (Supplementary Fig. 2). For other genes, ESX-1 and 175 PDIM seem to play a suppressive role, suggesting that other Mtb factors are involved in the 176 activation of immune pathways. Altogether, in hLEC, the ability of M. tuberculosis to form 177 intracellular cords requires both the ESX-1 system and the lipid PDIMs. 178

M. tuberculosis intracellular cords are localised in the cytosol 180
Given that at least two critical M. tuberculosis virulence-associated factors that contribute 181 to cytosolic localisation were required for intracellular cording and the significant 182 upregulation of cytosolic pathogen surveillance during cording, we next sought to define the 183 subcellular compartment within which M. tuberculosis cords were localised in hLECs. By 184 using a correlative imaging approach (correlative light and electron microscopy, CLEM), we 185 determined that M. tuberculosis intracellular cords were localised in the cytosol of hLEC in 186 long structures that (in this example) looped around the host cell nucleus (Fig. 4a). In 187 contrast, small groups of M. tuberculosis containing relatively low numbers of individual 188 bacteria were localised in a membrane-bound compartment (Fig. 4b) as reported before 189 (Lerner et al., 2016). The cords are usually formed of a bundle of several parallel chains of 190 individual bacteria. Interestingly, the volume of 25 individual bacteria from a cord compared 192 to 25 from a membrane bound compartment (non-cord, displayed as coloured 193 reconstructions) as measured by three-dimensional serial block face (3D SBF) CLEM was 194 significantly lower (Fig. 4d) (Fig. 5a). Strikingly, if hLEC are co-infected with 209 RFP-M. tuberculosis WT and with E2-Crimson-M. tuberculosis ΔRD1 or GFP-M. tuberculosis 210 ΔPDIM, the M. tuberculosis mutants lacking either ESX1 or PDIM were now able to clearly 211 form intracellular cords (Fig. 5a). Consistent with these observations, the feret diameter of 212 E2-Crimson-M. tuberculosis ΔRD1 or GFP-M. tuberculosis ΔPDIM in co-infected cells was 213 similar to RFP-M. tuberculosis WT whereas GFP-M. tuberculosis ΔRD1 or ΔPDIM alone had 214 low feret diameter measurements (Fig. 5b). Importantly, in co-infected cells, both the M. 215 tuberculosis ΔRD1 or ΔPDIM were able to replicate more efficiently (Fig. 5c). By CLEM, we 216 confirmed that the RFP-M. tuberculosis WT was localised in the cytosol and defined at the 217 ultrastructural level that the cords formed by GFP-M. tuberculosis ΔRD1 in co-infected cells 218 were now localised in the cytosol (Fig. 5d, e) Notably, when we co-labelled ubiquitin and p62 in cord-containing cells, we found that both 234 markers selectively associated with only small bacterial groups and not M. tuberculosis 235 cords (Fig. 6a). Strikingly, large M. tuberculosis cords (as defined by having a feret diameter 236 of greater than 10 µm), were devoid of the selective autophagy markers ubiquitin, p62, 237 Galectin-8, NDP52 and LC3B as well as the late endosomal/lysosomal markers LAMP-2 and 238 cathepsin D (Fig. 6b). In contrast, we found that, whereas none of the markers analysed 239 localised to the cords, some of the markers localised to a single or small group/clump of 240 intracellular M. tuberculosis with a lower feret diameter (Fig. 6b). These data indicated that, 241 although large and long M. tuberculosis cords were present in the cytosol, these were not 242 recognised by xenophagy. Consistent with the cords being negative for the autophagy-243 related host-cell markers tested, live cell imaging in hLECs expressing RFP-p62 revealed that 244 the intracellular cords form from bacteria which have either completely evaded p62-positive 245 compartments as a readout of autophagic targeting (Fig. 6c, Movie S1) or which have 246 initially been growth-restricted in a p62-positive state ( Fig. 6d) but subsequently became 247 p62-negative, where this process can also cycle several times (Fig. 6e, Movie S2). Crucially, 248 the M. tuberculosis cords only ever form once the bacteria lost p62 (Fig. 6e, Movie S3) 249 suggesting that cording is a consequence of avoiding an autophagic state or that cord 250 formation blocks autophagic targeting, potentially by being too large to encapsulate and 251 recapture from the cytosol. 252

253
Since the identification of M. tuberculosis as the etiologic agent of human TB, the 255 phenomenon of cording has attracted significant interest because of its association with 256 virulence and infection in vivo. Whereas there are many studies that implicate cording as a 257 mechanism to subvert phagocytosis, there is little evidence that in the infected host, M. 258 tuberculosis can freely replicate and cord in the extracellular milieu to avoid phagocytosis. 259 We show here that M. tuberculosis intracellular cords are a size-dependent mechanism of 260 evasion of endothelial host cell intracellular innate immune defences such as xenophagy. in the cytosol it is unlikely that ubiquitination will play a major role in xenophagic targeting. 309 We reason that if the bacteria themselves are being recognised, then why is only a 310 subpopulation targeted to autophagy? What is different about them? We hypothesise that 311 it is the ESX-1 mediated damaged membranes surrounding bacteria that are recognised, and 312 if M. tuberculosis is in close proximity to this it will be 'captured' with it. This process may be 313 cyclical, with M. tuberculosis then damaging the autophagic compartment to escape again. 314 However, if M. tuberculosis can get away from the damaged membranes after cytosolic 315 translocation, it may be able to evade autophagic capture. This is likely to occur for the 316 majority of the M. tuberculosis, hence why only a relatively small population are targeted to 317 autophagy. It is unlikely that dead bacteria or those that do not damage the phagosomal 318 membrane will be targeted to autophagy because it is ESX-1 and PDIM dependent; these 319 populations are thus likely to mature into phagolysosomes. Although the cording 320 phenotype seems to be unique for pathogenic mycobacteria, it remains to be determined if 321 other cytosolic pathogens also evades autophagy in a size-dependent manner as shown 322 here. with 10% (v/v) heat inactivated foetal calf serum (FCS) were seeded onto 10 mm diameter 341 #1.5 glass coverslips. 342

Mycobacterium tuberculosis strains 344
This study used the following EGFP tagged strains as described previously (Astarie-Dequeker Aldrich, #A9434) and then permeabilised with 0.01% saponin (Sigma-Aldrich, #84510) 1% 381 the primary antibody (diluted in PBS with 0.01% saponin, 1% BSA) was added onto the 383 coverslips for one to two hours at room temperature (detailed in Table 1). Following this, 384 three PBS washes preceded addition of the secondary antibody (diluted in the same way as 385 the primary antibody) for one hour at room temperature. The coverslips were again washed 386 three times in PBS, before an optional staining step for F-actin using a 1:250 dilution of 387

Electron microscopy (EM) of single-infected cells 442
Electron microscopy was performed exactly as previously described (Lerner et al., 2016). Selected bacteria were segmented manually from slices of SBF SEM datasets and 3D 455 reconstructions were made using the 3dmod program of IMOD (Kremer et al., 1996). Each 456 dataset was first de-noised with a 0.5 pixel Gaussian blur filter applied in Fiji (ImageJ; 457 National Institutes of Health). 2 datasets from each of 2 independent samples were then 458 segmented for each of the cord and membrane-bound bacteria conditions. The dataset xy 459 pixels were 9.9 nm and 8.7 nm for cord bacteria, and 5.4 nm and 6.3 nm for membrane for approach imaging even though the cells were not processed for this method), then the 486 cut face was aligned to a diamond knife in a UC7 ultramicrotome (Leica Microsystems) and 487 70-80 nm sections from the field of interest were collected. The sections were stained with 488 lead citrate and imaged in a TEM (Tecnai G2 Spirit BioTwin; Thermo Fisher Scientific) using a 489 charge-coupled device camera (Orius; Gatan Inc.). For CLEM overlay, TEM images were 490 assigned to confocal slices manually in z. The confocal slice was then processed and aligned 491 with TurboReg in Fiji as above. 492

Histology, immunohistochemistry and analysis 494
The study was performed using excised cervical lymph node tissue stored within the 495

Department of Anatomical Pathology at Groote Schuur Hospital (Cape Town, South Africa). 496
All of these biopsies were taken for clinical indications. Residual paraffin-embedded blocks 497 of these specimens were stored for further processing. This study complied with the 498

Declaration of Helsinki (2008), and ethics approval was obtained from the University of Cape 499
Town Human Research Ethics Committee (REC187/2013). Informed consent was waived, as 500 this was a retrospective study of formalin-fixed paraffin-embedded tissue samples collected 501 during the course of routine clinical practice. Patient identifiers were unavailable to 502

investigators. 503
Formalin-fixed paraffin-embedded tissue sections from patients diagnosed as tuberculosis 504 culture positive and/or acid-fast bacilli positive (AFB+) were selected for the study and 505 processed as described before (Lerner et al., 2016). Briefly, tissue sections were 506 deparaffinized in xylene (2 x 10 min, 100%, 95% and 80% ethanol (2 min each). Tissue 507 sections were then placed into an antigen retrieval buffer (Access super antigen solution, 508 Menarini diagnostics, UK) in a decloaking chamber (Biocare Medical, CA, USA); incubated at 509 110 degrees for 10 min and allowed to cool for 60 min. Sections were permeabilized in PBS-

RNA-Seq data analysis 528
The RNA-Seq data in this paper have been deposited in Gene Expression Omnibus repository 529 with accession number GSE110564. The quality of the Illumina-produced fastq files was 530 assessed using FastQC (v0.11.5) and adapter trimmed using Trimmomatic (v0.36). The 531 resulting reads were then aligned to the human genome (Ensembl GRCh38 release 88 build) 532 using STAR aligner (v2.5.2a). Gene counting was done using RSEM (v1.2.29) and expected 533 read counts were normalized using DESeq2 (v1.18.1), which also determined the log2 fold 534 change and statistical significance between the infected and uninfected samples. Canonical 535 pathway and functional process analyses were performed using IPA Ingenuity (QIAGEN) and 536

Real-time polymerase chain reaction (RT qPCR) 539
Isolated RNA was processed with QuantiTect™ Reverse Transcription Kit (Qiagen).                       WT and 6,470,9,472,11,759 for ∆RD1 at MOI:10,20 and 40,701 respectively. ImageJ was used to first select only the GFP channel corresponding to the bacteria. The GFP 768 was subjected to a pixel threshold, dilated and eroded (add and then remove 1 pixel to the 769 outline, to link up any incorrectly thresholded pixels) and then outlined to form 'particles' 770 distance between the two furthest apart pixels in the particle. 772 Human lymphatic endothelial cells (hLEC) expressing p62-RFP (red) infected with M. LGALS3   *** ***