Distinct mechanisms of dysfunctional antigen‐presenting DCs and monocytes by single‐cell sequencing in multiple myeloma

Abstract Multiple myeloma (MM) is an incurable plasma cell malignancy with the hallmark of immunodeficiency, including dysfunction of T cells, NK cells, and APCs. Dysfunctional APCs have been reported to play a key role in promoting MM progression. However, the molecular mechanisms remain elusive. Here, single‐cell transcriptome analysis of dendritic cells (DC) and monocytes from 10 MM patients and three healthy volunteers was performed. Both DCs and monocytes were divided into five distinct clusters, respectively. Among them, monocyte‐derived DCs (mono‐DC) were shown to develop from intermediate monocytes (IM) via trajectory analysis. Functional analysis showed that, compared with healthy controls, conventional DC2 (cDC2), mono‐DC, and IM of MM patients exhibited impaired antigen processing and presentation capacity. Moreover, reduced regulon activity of interferon regulatory factor 1 (IRF1) was found in cDC2, mono‐DC and IM of MM patients according to single‐cell regulatory network inference and clustering (SCENIC) analysis, while the downstream mechanisms were distinct. Specifically in MM patients, cathepsin S (CTSS) was markedly downregulated in cDC2, major histocompatibility complex (MHC) class II transactivator (CIITA) was significantly decreased in IM, in addition both CTSS and CIITA were downregulated in mono‐DC based on differentially expressed genes analysis. In vitro study validated that knockdown of Irf1 downregulated Ctss and Ciita respectively in mouse DC cell line DC2.4 and mouse monocyte/macrophage cell line RAW264.7, which ultimately inhibited proliferation of CD4+ T cells after being cocultured with DC2.4 or RAW264.7 cells. This current study unveils the distinct mechanisms of cDC2, IM, and mono‐DC function impairment in MM, offering new insight into the pathogenesis of immunodeficiency.


| INTRODUC TI ON
MM is the second most common blood cancer and accounts for 1% of neoplastic malignancies, 1 characterized by the abnormal expansion and accumulation of malignant plasma cells in BM.
Immunodeficiencies are the common hallmarks of MM, which include (1) dysfunctional state of T cells and natural killer cells, 2 (2) impaired capacity of APCs, 3 (3) disturbed production of tolerogenic cytokines 4 and (4) increased infiltration of myeloid-derived suppressor cells in the BM niche. 5 Due to the capacity to induce T-cell immune responses, the functional status of APCs plays a pivotal role in anti-myeloma immune surveillance.
APCs including DCs and monocytes bridge innate and adaptive immunity. They capture, ingest, and present antigens to T cells via MHC molecules, then initiate allogeneic T-cell responses. 6 Several researchers have investigated the number, 7,8 phenotypic profile, 8,9 and functional status [10][11][12][13] of DCs and monocytes during the development of MM. A previous study has indicated an ~50% decline in myeloid DCs (mDC) in peripheral blood (PB) of MM patients independent of disease stage. 7 In BM, mDCs and plasmacytoid DCs (pDC) accumulate with an immunosuppressive phenotypic during MM development. 8 Moreover, immunological inhibitory cytokines in the TME impair DC activation and maturation. 10,11 For example, TGFβ and interleukin (IL)-10 produced by myeloma cells are responsible for defective upregulation of CD80/CD86 during DC activation. 11 IL-6 promotes the differentiation of CD34 + monocytic cells but inhibits the proliferation of CD34 + DC progenitors, which results in adequate phagocytosis but inadequate antigen presentation. 10 In addition, CD14 + monocytes with lower MHC II expression acquire protumor potential and suppress T-cell activation in the MM condition. 12 These studies all illustrate that DCs and monocytes in the MM condition exhibit dysfunction status.
However, the molecular mechanisms underpinning the dysfunction of APCs in MM remain elusive. In the current study, we found distinct patterns of DCs and monocytes with impaired antigen presentation capacity in MM through single-cell sequencing, which might provide novel insights into understanding the pathogenesis and progression of MM.  Table S1. PBMCs were obtained from fresh BM and PB samples via Ficoll-Paque density gradient centrifugation as soon as possible after the specimens were obtained. mRNA was prepared following the Chromium instrument and the Single Cell 3′ reagent kit (V3) was used. To assess quality control and concentration, we used an Agilent Bioanalyzer 2100 system and the resulting scRNA-seq libraries were sequenced on a NovoSeq 6000 instrument.

| Cell clustering
After quality control analysis, the unique molecular identifier (UMI) count matrix was log normalized. FastMNN was used to integrate data and Seurat (3.2.3) was applied to label each cluster of DCs and monocytes based on the classical marker genes. In each case, clustering was done following principal components analysis.

| Gene set variation analysis (GSVA)
We applied GSVA, a GSEA-based method to quantify the activity of various pathways over different populations in an unsupervised way. 15 In this study, the target pathways include 186 KEGG gene sets and the GSVA R package was implemented.

| Enrichment analysis and estimation of antigen presentation score
DEGs of cell groups were recognized using the "FindAllMarkers" function in Seurat and ranked according to the log 2 FC. The "cluster-Profiler" function (v3.18.1) was used for KEGG analysis. The antigen presentation scores of DCs and monocytes were determined by the "AddModuleScore" function in the Seurat package.

| SCENIC (single-cell regulatory network inference and clustering) analysis
The R SCENIC package (version 1.1.2-2) and hg19-500bp-upstream-10species databases for RcisTarget, GRNboost, and AUCell were used to perform transcription factor activity analysis. The input matrix was the normalized expression matrix from Seurat.

K E Y W O R D S
dendritic cells, immunodeficiency, impaired antigen presentation, monocytes, multiple myeloma

| Pseudotime trajectory analysis
We applied Monocle 2 (v2.18.0) for trajectory analysis to identify the differentiation relationships between mono-DCs and different clusters of monocytes. The "plot_genes_in_pseudotime" function was applied to discover transitional changes in gene expression levels among different clusters.

| Cell-cell interaction analysis
Cellular interactions between different cell types were performed based on ligand-receptor co-expression using the CellPhoneDB (v2.0) tool with normalized count data for scRNA-seq. The default database and parameters were used.

| RNA isolation and qPCR
Total cellular RNA was extracted using an RNeasy Kit (Qiagen) according to the manufacturer's instructions and then reverse transcribed into complementary DNA (cDNA) using the HiScript III All-in-one RT SuperMix (Vazyme). Each cDNA sample was quantified in triplicate using the ChamQ SYBR Color qPCR Master Mix (Vazyme) and subjected to quantitative real-time PCR. The sequences of the PCR primers are listed in Table S2.

| Antigen-presenting analysis
The antigen-presenting function of DC2.4 and RAW264.7 cells was assessed by allogeneic T-cell activation responding to a specific antigen.

| Statistical analysis
Data analyses were computed in R software (Version 4.0.2) and GraphPad Prism (Version 9), a p-value < 0.05 suggested statistically a significant difference. Significance was calculated using the indicated statistical tests.

| Single-cell sequencing profiling of dendritic cells and monocytes
We applied 10× genomics scRNA-seq to study DCs and monocytes from BM and PB samples of 10 MM patients and three healthy controls. After quality control and filtration, 1429 DCs and 42,464 monocytes were obtained. DCs and monocytes were both divided into five clusters according to their gene expression profiles ( Figure 1A,B). Among DCs, AXL + SIGLEC6 + DC (ASDC) was identified according to the previous research, 16 while pDC was characterized by the high expression of IL3RA and CLEC4C. In addition, mDC was previously described as ITGAX + DC, in which cDC2, mono-DC and conventional DC1 (cDC1) expressed CD1C, CD14, 17 and THBD respectively ( Figure 1C and Figure S1). Monocytes were also classified into five clusters, including NCMs, IM, and three clusters of CMs according to the expression of CD14 and CD16. Three clusters of CM were featured by CD74/HLA-DRA (CM1), S100A12/S100A8 (CM2), and IFI44L/IFI6 (CM3), respectively ( Figure 1D and Figure S1). Next, we examined the distribution of each subset in BM and PB samples, and found that the frequency of cDC2 was significantly decreased in BM (p = 0.046) and PB (p = 0.005) of MM patients. VCD treatment partially restored the reduction of cDC2 proportion, while mono-DCs were enriched after treatment ( Figure 1E and Figure S1).
However, there was no significant difference in the percentage of monocyte subsets in either BM or PB ( Figure 1F and Figure S1).

| Cell-state transition of mono-DCs and monocyte subsets
In an inflammatory state, monocytes may develop into mono-DC. [18][19][20] To identify which subset of monocytes that mono-DC derived from, we conducted the pseudotime analysis and observed the gradual transition of CM toward IM and then mono-DC or NCM ( Figure 2A and Figure S2A). Based on the DEGs along the developmental trajectory, we found that expression of CD1C and MHC II molecules (such as HLA-DQA1) increased in mono-DC, while high expression of CD16 and low-expression of CD14 in NCM were observed ( Figure 2B,C). However, compared with healthy controls, pseudotime trajectory analysis of MM patients showed nondifference in developmental branches ( Figure S2B). Thus, these data suggested that mono-DC may derive from IM.

| Dendritic cells and monocytes exhibit impaired antigen presentation capacity in MM
We performed GSVA 15 utilizing KEGG gene sets to estimate pathway activities over different clusters among DCs and monocytes. The results confirmed that cDC1 and cDC2 were the major APCs within DCs ( Figure 3A) and IM showed a higher level of antigen presentation pathway activity compared with the other monocyte subsets, suggesting a higher potential in antigen presentation capacity ( Figure 3B).
Next, compared with healthy controls, the score of antigen processing and presentation pathway according to the KEGG gene set was all decreased in total DCs (p < 0.001, Figure S3A) and total monocytes (p < 0.001, Figure S3B) in MM patients, which indicated an impaired function of DCs and monocytes. To clarify the phenotype characteristics of the dysfunctional DCs and monocytes, we conducted KEGG enrichment analysis. Results showed that the antigen processing and presentation pathway was significantly downregulated in cDC2 ( Figure 3C), mono-DC ( Figure 3D), and IM ( Figure 3E) of MM patients, while no significant difference in the pathway was observed between MM patients and healthy controls in cDC1 ( Figure S3C). In conclusion, cDC2, IM, and mono-DC showed impaired antigen processing and presentation capacity in MM patients.

| Downregulation of IRF1 in cDC2, IM, and mono-DC with MM
After observing the impaired antigen presentation capacity of cDC2, IM, and mono-DC in MM patients, we next explored the mechanisms. By mapping malignancy-specific regulon networks by SCENIC, we identified several key transcriptional factors (TF) with reduced regulon activities in cDC2 of MM ( Figure 4A), such as BACH1 which was associated with the generation of APCs. 21 Among them, only the expression of IRF1 was downregulated (p = 0.035) in MM samples ( Figure 4D and Figure S4A). Similarly, in IM and mono-DC, we found that the regulon activity ( Figure 4B,C) and gene expression ( Figure 4D and Figure S4A) of IRF1 were both reduced in MM patients. Moreover, the expression of IRF1 was increased in IM after treatment (p < 0.001), while there were no significant changes in cDC2 or mono-DC ( Figure S4B). We further inspected the expression of IRF1 in cDC2, IM, and mono-DC from MM patients with different treatment responses after two cycles of the VCD regimen ( Figure S4C). Notably, the expression levels of

| The potential role of IRF1 in regulating antigen processing and presentation
Given that IRF1 was correlated with antigen presentation, we further

| Perturbation of cDC2-CD4 + T-cell and IM-CD4 + T-cell interactions in MM
On the condition that APCs-CD4 + T-cell costimulatory interaction was impaired in MM patients, we supposed that an inhibitory interaction between APCs and CD4 + T cells might occur in TME.
Therefore, we performed cell-cell communication analysis using CellPhoneDB 27 and found prominent interactions between T cells and DCs and monocytes ( Figure S6A). To infer the communications between different subsets of CD4 + T cells and APCs, we first clustered CD4 + T cells into seven subsets ( Figure S6B). The results showed several upregulated inhibitory ligand-receptor pairs in MM patients such as LGALS9_HAVCR2, PDCD1_FAM3C, and CCL4L2_ VSIR between cDC2 and CD4 + T cell communications ( Figure 6A).
Analysis of the interacting molecules between IM-CD4 + T cells showed LGALS9_HAVCR2, PDCD1_FAM3C, and PVR_TIGIT pairs were enhanced in MM patients ( Figure 6B). However, the interaction of mono-DC and CD4 + T cells between MM patients and healthy control showed no significant difference ( Figure S6C).
Collectively, these results indicated that cDC2 and IM were associated with the exhausted phenotype of CD4 + T cells in TME via inhibitory ligand-receptor pairs. Accumulating evidence indicates that monocytes differentiate into mono-DC under inflammatory conditions. 18-20 Enrichment of mono-DC was found in tuberculosis, 20 atopic dermatitis, and psoriasis patients. 18 Researchers also reported the engagement of mono-DC in various tumors, including lung cancer 28 and ovarian cancer. 19 In vitro experiments have demonstrated that circulating CD14 + monocytes are programmed to spontaneously differentiate into CD16 + monocytes. 29 Collectively, these observations support the notion that monocytes can develop into mono-DC, while rare study concentrated on tracking the fate of mono-DC. In our  42 Although earlier studies had confirmed that IRF1 regulates the endogenous expression of cathepsin S(CTSS), 22,23 there is no proof that the IRF1-mediated CTSS pathway regulates antigen presentation. Nevertheless, DC maturation and antigen processing and presentation are ascribed to the regulation of CTSS in earlier studies. 43 Our data showed that downregulated IRF1 resulted in the decreased expression of CTSS in cDC2 and eventually suppressed the activation of CD4 + T cells.

ACK N OWLED G M ENTS
We are greatly indebted to all subjects who participated in this study. We thank Professor Lei Liu, Dr. Gang Liu, and Dr. Yunhe Liu (Fudan University, Shanghai, China) for post-processing and quality control of raw reads in scRNA-seq data. We also thank the Core

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon request. The raw scRNA-seq data are de-