The non-canonical Wnt pathway leads to aged dendritic cell differentiation

Hematopoietic systems undergo a series of changes as they age, including (1) increased hematopoietic stem cell (HSC) pool size, especially for the long-term HSCs (LT-HSCs); (2) skewing of hematopoiesis toward the myeloid lineage and an increase in the myeloid-dominant HSCs in the HSC pool; and (3) decreased repopulation potential per HSC. In line with these changes, myeloid cells (mainly granulocytes) increase, whereas lymphoid cells (B cells) decrease in aged mice.^1,2 Studies of the heart, muscle, skin and blood have demonstrated that changes in signaling pathways (i.e. the Wnt, Notch, TGF-β, NFκB and mTOR pathways) may underlay (stem) cell aging, and excitingly, the modulation of these pathways may restore tissues to a more youthful state.^3,4 Among these pathways, small RhoGTPase cell division control 42 (CDC42) has been identified as a key regulator of aging in HSCs.² This molecule is downstream of the non-canonical Wnt pathway. CDC42 activity increases in the bone marrow and other tissues with age, and this increase has been causally linked to HSC polarity, differentiation, engraftment, and aging.² The elevated activity of CDC42 in aged HSCs seems to be a direct consequence of increased stem cell-intrinsic expression of WNT5A and, thereby, a shift from canonical to non-canonical Wnt signaling. The mechanism by which WNT5A expression in aged HSCs is induced remains largely unknown, but possibly involves epigenetic modifications.⁵

A previous study by Xiao et al. reported that the differentiations of both plasmacytoid dendritic cells (pDCs) and CD8α⁺ conventional dendritic cells (cDCs) are deficient in aged mice. Moreover, these authors found that WNT5A inhibited DC differentiation in vivo and in vitro. Pharmacological inhibition of the non-canonical Wnt pathway partially rejuvenated aged DC differentiation.⁶ Because higher expression of WNT5A was also found in aged HSC and WNT5A pre-treated lymphocyte-primed multipotent precursor (LMPP) cells displayed a significant decrease in DC differentiation similar to that of aged mice, the WNT5A-CDC42 pathway therefore represents an important pathway that affects DC differentiation.

In the classic model for DC development, both cDCs and pDCs can be generated from the fms-like tyrosine kinase 3 (Flt3) expressing early myeloid or lymphoid progenitors. HSCs differentiate into LMPPs, and LMPPs give rise to common myeloid precursors (CMPs) and common lymphoid precursors (CLPs). CMPs then produce common dendritic precursors (CDPs), which are the direct precursors of both cDCs and pDCs. Moreover, some pDCs can originated from CLPs.⁷ Therefore, the inhibition of CDP differentiation upon WNT5A treatment could explain the decreased differentiation of cDCs and pDCs. At the CMP stage, the biased differentiation into GMPs and MEPs may partially explain the decreased CDP number, although this issue is far from clarified. As research progresses, the classic model has been challenged in recent years. Onai et al. noted that the distinct precursor can be produced within one division from the upstream LMPP population, so there could be a direct route to the production of an M-CSFR⁻ DC precursor that bypasses the conventional CDP (defined as Lin⁻Flt3⁺IL-7R⁻CD11c⁻M-CSFR⁺).⁸ This derivation from LMPPs is of particular interest because in a study using ‘barcoding’ of LMPP cells, a remarkable extent of pre-commitment to DC production was found at this early stage of development. Naik et al. demonstrated that LMPPs are highly heterogeneous in terms of the cell types that they produce, which include lymphoid-, myeloid- and DC-biased producers.⁹ In vivo analysis of the output of sibling cells derived from single LMPPs has demonstrated that these cells often share a similar fate, which suggests that the fate of these progenitors is imprinted. Furthermore, because this imprinting is also observed for DC-biased LMPPs, DCs may be considered a distinct pre-committed lineage based on their separate ancestries. These studies suggest a ‘graded commitment’ model of hematopoiesis in which heritable and diverse lineage imprinting occurs earlier than previously thought. Recently, Grajkowska et al. demonstrated that the long form of TCF4, which is a key transcription factor specific for pDC differentiation, increased gradually in the different hematopoietic precursors in a positive feedback manner,¹⁰ which further supports the new model. In a paper from Xiao et al., it was revealed that genes that are important for DC differentiation, including Flt3 and Ikaros, can also be inhibited by WNT5A in the LMPP and HSC stages. The decreased expansion of Flt3L-treated LMPPs when co-cultured with WNT5A further indicated that the Flt3 pathway might be affected by WNT5A.⁶ According to the new model of DC differentiation, the commitment of DC precursors probably occur in the LMPP stage. Increasing evidence suggests that LMPPs are a highly heterogeneous group of cells, and single-cell RNA sequencing analysis might help to identify the molecular features of DC-biased precursors in the LMPP stage. Further study may help to clarify the detailed mechanisms by which WNT5A inhibits DC differentiation in aged mice.

In addition to DC differentiation, the non-canonical Wnt signal pathway can also affect DC function. Valencia et al. reported that WNT5A skewed monocyte derived DC differentiation to a non-conventional phenotype with tolerogenic features.¹¹ Hack et al. also reported that WNT5A inhibited CpG oligodeoxynucleotide-triggered activation of human pDCs.¹² Although Xiao et al. reported that WNT5A negatively regulated pDC activation, the inhibition of the WNT5A-CDC42 pathway could not reverse the pDC activation, which suggests a different mechanism for the regulation of pDC function. Furthermore, the effect of the non-canonical Wnt pathway on the activation of other DC subsets, such as the cDC subsets, could be another important aspect that requires further investigation for clarification of the function of this pathway in DC activation.

Conflict of interest

The authors declare no conflict of interest.