Single-cell analysis of intestinal immune cells during helminth infection

Single cell isolation from helminth infected intestines has been notoriously difficult, due to the strong anti-parasite type 2 immune responses that drive mucus production, tissue remodeling and immune cell infiltration. Through the systematic optimization of a standard intestinal digestion protocol, we were able to isolate millions of immune cells from heavily infected tissues. Using this protocol, we validated many hallmarks of anti-parasite immunity and analyzed immune cells from the lamina propria and granulomas during helminth development, as well as acute and chronic worm infection.

mouse strains or the assessment of secondary locations like the draining lymph nodes, blood or spleen 21,25,26 , which might only partially reflect local immunity. As helminth infections are strongly linked to chronic impairments that affect nutrition availability 27,28 ; memory, cognition and physical development [29][30][31] ; changes in the microbiota 32,33 and modulation of local and systemic immunity 26,34 , an optimized digestion protocol is needed to further investigate the infected intestinal tissue.
Difficulties with intestinal digests during helminth infection have been associated with a strong anti-parasite type 2 immune response that drives mucus production 35,36 , alters the epithelium 37,38 , induces immune cell infiltration 39 and causes tissue remodelling 40,41 (Fig. 1c). In order to investigate a model of both acute and chronic helminth infection, we infected C57BL/6 mice with Heligmosomoides polygyrus, a naturally occurring rodent parasite with an exclusive intestinal life cycle 42,43 . Infective L3 larvae penetrate the intestinal tissue of the duodenum within 24 hours of ingestion, undergo larval development in the muscularis externa and return to the lumen within 10 days post infection, where the adult worms mate and develop a chronic infection in C57BL/6 mice 43,44 . The peak of acute immunity is usually studied around day 14 post infection and we focused on this time point and the heavily infected duodenum 45,46 , to optimize our digestion protocol.
In order to develop a digestion protocol for heavily infected intestines we followed a systematic approach and optimized each step of the standard intestinal digestion protocol (Fig. 1d). First, we modified the EDTA wash steps to remove the increased amount of mucus but did not observe an improvement in cell yield ( Fig. 1e and Supplementary Fig. 1; digestion protocols #1, 2, 6). This was followed by testing a variety of collagenases that have been reported for intestinal digests, as we hypothesized that the intestinal remodeling that occurred during helminth infection could negatively impact the digestion procedure. Indeed, we found that Collagenase A from Clostridium histolyticum ( Fig. 1e and Supplementary Fig. 1; digestion protocols #7, 8), but not Collagenase VIII, Collagenase D, Dispase or Liberase TM ( Fig. 1e and Supplementary Fig. 1; digestion protocols #3-5), showed an increase in cell yield when used in conjunction with the standard digestion protocol. To further optimize the protocol, we increased and modified the wash steps and observed a further increase in cell yield ( Fig. 1e and Supplementary Fig. 2; digestion protocols #9-12). Importantly, strong vortexing after each wash step significantly improved the outcome of digestion ( Fig. 1e and Supplementary Fig. 2; digestion protocol #13), suggesting that the epithelium is harder to remove in helminth infected tissues, likely due to its strong activation and remodelling 37,38 . Indeed, observations from Stat6ko mice confirmed that the physiological changes that impair the intestinal digest using the standard protocol, were all linked to type 2 immune responses, as intestines from infected Stat6ko mice could readily be digested ( Supplementary Fig. 3). Several intestinal cell isolation protocols utilize a final gradient centrifugation step to further isolate immune cells 17,20 . However, in our hands this resulted in a dramatic drop in cell yield and was therefore omitted ( Fig. 1e and Supplementary Fig. 2; digestion protocol #14). Our optimized lamina propria cell isolation protocol for H. polygyrus infected intestines thus included 3-4 10-minute 2mM EDTA wash steps (each followed by vigorous vortexing) and a 30-minute digest with 1mg/ml Collagenase A, 20% FCS and 0.05mg/ml DNase (see Supplementary Protocol 1 for step-by-step instructions).
When we compared intestinal digests from naïve intestines using the standard or optimized cell isolation protocol, we observed highly comparable outcomes ( Fig. 1e; digestion protocols #1 and 13). Both digestion protocols resulted in a cell yield of 3-6 million live cells per naïve duodenum with 70-80% viability and 20-30% frequency of CD45+ cells. To assess the effectiveness of our digestion protocol during the different stages of H. polygyrus infection, we harvested the duodenum from naïve C57BL/6 mice and at day 7, day 14 and day 28 post infection, which represented time points of larval development in the muscularis externa, acute as well as chronic adult worm infection, respectively. We observed that all time points could be successfully digested using our optimized digestion protocol and that duodenal digests from 14-and 28-days post infection yielded 3-6 million live cells per sample (Fig. 1g). We furthermore observed a consistent doubling of the cell count to 8-11 million live cells per duodenum at day 7 post infection and observed a similar trend when we quantified CD45+ cells in cryosections from these time points (Fig. 1g, h).
In order to understand these differences and validate that our protocol was suitable for subsequent single cell analysis and immunophenotyping, we characterized the isolated cells further using a 23-color spectral flow cytometry panel that incorporated many of the hallmark surface and intracellular markers for type 2 immune responses that have been associated with helminth infections 25,43 (see Supplementary Table 1 for details regarding markers, fluorophores, clones and staining concentrations used). As collagenase digests can negatively affect surface epitope integrity, we tested our optimized digestion protocol on splenocytes and compared digested to non-digested cells. While we observed a reduction in the MFIs of a few markers (namely Ly6G, MHCII, CD45 and CD127), all positive stained cell populations could be clearly identified ( Supplementary Fig. 4). Isolated intestinal lamina propria cells also proved a challenge for intracellular staining, as different commercial intracellular staining kits significantly affected the cellular, but not debris, scatter profiles and varied in the resolution of intracellularly antibody staining ( Supplementary Fig. 5). In our hands, the eBioscience FoxP3/Transcription Factor Staining Buffer Set yielded the best results and was used henceforth.
We used a combination of high-dimensional analysis tools and manual gating strategies to analyze the isolated lamina propria cells from the three main stages of H. polygyrus infection ( Fig. 2a and Supplementary Fig. 6 and 7). In line with previous findings 39 , we observed a strong infiltration of immune cells such as neutrophils and monocytes at day 7 post infection, which we verified in cryosections and were primarily localized around the developing larvae explaining the increase in total cell number (Fig. 2a-c and Supplementary Fig. 8). At later time points this inflammatory response receded, which is likely linked to the worms exiting the intestinal tissue and inhabiting the lumen. Peak expression of RELMa in resident macrophages, which is a hallmark for their alternative activation and wound repair responses 47,48 , was observed at day 14 and was again localized within the granulomas (Fig.  2b,c and Supplementary Fig. 8). While type 2 innate lymphoid cells did not increase in frequency over time, ki67 expression increased, suggesting their proliferation and activation (Fig. 2b), as previously described 36,49 . GATA3+ Th2 cells, important drivers of type 2 immunity 43,50 , could be observed at day 7 post infection, increased in frequency over time and showed high ki67 expression (Fig. 2b). Interestingly, ki67 expression strongly decreased at day 28 post infection for all cell types analyzed ( Fig. 2b and Supplementary Fig. 9), which . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted August 24, 2019. ; https://doi.org/10.1101/746479 doi: bioRxiv preprint could be linked to the strong immunomodulatory properties reported during chronic worm infection 25,51 (Fig. 2e).
Importantly, we were able to characterize intestinal innate and adaptive immune responses for all stages of H. polygyrus infection using our optimized lamina propria cell isolation protocol, which has previously not been possible. We were able to analyze immune cell infiltrations in the villi as well as the granulomas and could observe strong inflammatory responses during H. polygyrus development in the muscularis externa, followed by strong acute Th2 responses and a dampening of the immune response later on. Many of these changes were specific to the infected tissue and were not observed to the same extent in the draining lymph nodes (Supplementary Fig. 10 and 11), highlighting the potential of our protocol for future studies that utilize current single-cell analysis tools. We furthermore anticipate that this method will be applicable to study the immune responses to other intestinal helminths or type 2 immune conditions, such as Nippostrongylus brasiliensis infection or food allergy models, and after determining the correct collagenase compositions, could be also applied to human samples.   Mice. C57BL/6 (C57BL/6JOIaHsd) and C57BL/6.Stat6KO mice were bred at the Malaghan Institute of Medical Research, Wellington, New Zealand. Mice were housed under specific pathogen free conditions and age-matched female adult animals were used in each experiment.

Heligmosomoides polygyrus infection.
Mice were infected with 200 L3 larvae by oral gavage at 6-8 weeks of age and intestines and draining lymph nodes were harvested at the indicated time points.
Conventional and spectral flow cytometry. For conventional flow cytometry cells were resuspended in 0.5ml of 20µg/ml DNase containing FACS buffer, stained with DAPI to identify dead cells, filtered and analyzed using a BD LSRFortessa SORP ™ flow cytometer. For spectral flow cytometry, intestinal and lymph node samples were washed in 200μL FACS buffer and incubated with Zombie NIR Fixable Viability dye (Biolegend) for 15 minutes at room temperature. After washing, cells were incubated with Fc block (clone 2.4G2, affinity purified from hybridoma culture supernatant) for 10 minutes followed by the incubation of surface antibodies (see Supplementary Table 1) for 25 minutes at 4°C in the presence of 20μg/ml DNase and Brilliant Buffer Plus (BD Biosciences). Cells were fixed and permeabilization with the FoxP3/Transcription Factor Staining Buffer Set (eBioscience) according to manufacturer's instructions and incubated with intracellular antibodies (see Supplementary Table 1) for 45 minutes at 4°C. Cells were then resuspended in FACS buffer, filtered, and analyzed on a 3laser Aurora spectral flow cytometer (Cytek Biosciences).
Data analysis. FCS files were manually analyzed using FlowJo (v10.6, Tree Star) or evaluated with high-dimensional data analysis tools using Cytobank (v7.2, Cytobank Inc.). After compensation correction in FlowJo, single, live, CD45+ events were imported into Cytobank and transformed to arcsinh scales. FlowSOM analysis was performed on 1,200,000 concatenated lamina propria and 1,000,000 concatenated lymph node cells, with an equal distribution of samples. Different cluster analyses were performed and 121 clusters were identified as the most representative for both data sets.
Statistical analysis. Experimental group sizes ranging from 3 to 5 animals were chosen to ensure that a two-fold difference between means could be detected with a power of at least 80%. Prism 6 Software (GraphPad) was used to calculate the s.e.m. and statistical differences between groups and samples were calculated using Kruskal-Wallis tests followed by the appropriate multiple comparisons, with P≤0.05 being considered as significant.
Source Data. Lamina propria and lymph node data sets used for manual and FlowSOM analysis can be downloaded from flowrepository (http://flowrepository.org/id/FR-FCM-Z28B). All other datasets are available upon request.   not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted August 24, 2019. ; https://doi.org/10.1101/746479 doi: bioRxiv preprint