Phloem loading via the abaxial bundle sheath cells in maize leaves

Leaves are asymmetric, with differential functionalization of abaxial and adaxial tissues. The bundle sheath (BS) surrounding the vasculature of the C3 crop barley is dorsoventrally differentiated into three domains: adaxial structural, lateral S-type, and abaxial L-type. S-type cells seem to transfer assimilates towards the phloem. Here we used single-cell RNA sequencing to investigate BS differentiation in C4 maize. Abaxial BS (abBS) cells of rank-2 intermediate veins specifically expressed three SWEET sucrose uniporters (SWEET13a, b, and c) and UmamiT amino acid efflux transporters. SWEET13a, b, c were also identified in the phloem parenchyma (PP). Thus maize acquired a unique mechanism for phloem loading in which abBS cells provide the main pathway for apoplasmic sucrose transfer towards the phloem. This pathway predominates in veins responsible for phloem loading (rank-2 intermediate), while rank-1 intermediate and major veins export sucrose from the phloem parenchyma (PP) adjacent to the sieve element companion cell (SE/CC) complex, as in Arabidopsis. We surmise that abBS identity is subject to dorsoventral patterning and has components of PP identity. These observations provide first insights into the unique transport-specific properties of abBS cells and support for a modification to the canonical phloem loading pathway of maize, which may be generalizable to other C4 monocots.


INTRODUCTION
sheath identity (Fig. 1c). The distribution of marker genes was consistent with the roles of 100 mesophyll and BS cells in C4 photosynthesis ( Supplementary Fig. 2, Supplementary Table 2). The 101 ratio of mesophyll to BSC was ~75:1, indicating a low efficiency of BSC retrieval. To our surprise, 102 no vascular cells were recovered. The BS cluster was further divided into two subclusters, the 103 "upper" and "lower" subclusters, which later were assigned as abaxial ( ab BS) and adaxial ( ad BS) 104 bundle sheath cells (Fig. 1d, Fig. 2) (see below).

123
It may be interesting to explore further whether the subclustering of mesophyll cells represents 124 developmental trajectories or physiological differences. We did not identify an obvious pattern that 125 could be attributed to, for example, dorsoventral patterning due to developmental gradients or due 126 to changes in light properties as it passes through the leaf. Similar observations regarding the 127 presence of multiple mesophyll clusters were made for scRNA-seq analyses of Arabidopsis leaves, 128 however, it was not possible to assign palisade and spongy parenchyma to any of the mesophyll

184
Other BSC-specific genes such as RBCS1 showed equal transcript distribution across all BS cells, 185 excluding the possibility of an artefact, e.g., a gradient of cells differing in UMI counts. This 186 included UmamiT20a, which was BS-enriched but equally expressed across both BS subclusters.

187
The lack of specificity of many genes for subsets of BSC is consistent with published data from in

191
To test the hypothesis that the two BS subclusters may represent spatially discrete BSC populations, 192 in situ hybridization was used to localize the mRNAs of the SWEET13a, b, and c genes ( Fig. 2b-    in parallel using the same method from leaves at the same stages of development. SUT1 RNA was 222 typically found in one or two cells in the phloem, which most likely represent companion cells,

223
where it is responsible for phloem loading. In rank-1 and major veins, SUT1 mRNA was also 224 detected in the medial vascular parenchyma, where it likely contributes to sucrose retrieval (Heyser 225 et al., 1977). In our experiments, SUT1 transcripts were not detected in bundle sheath cells, 226 consistent with SUT1 expression below the detection limit in of our BSC single cell dataset ( Fig.   227 2g-h; Supplementary Fig. 5).

228
Abaxial BS transcripts are co-regulated during the sink-source transition

229
In Arabidopsis, many PP-specific genes were found to also be co-regulated (Kim et al., 2020). We 230 therefore tested whether several of the transporter genes identified in ab BS might also be co-

263
In rank-2 intermediate veins of maize, it appears that the ab BS cells may have recruited sucrose-264 transporting SWEETs to export sucrose toward the abaxially localized phloem (Fig. 6). This 265 presents a unique pathway, in which BSCs likely export photosynthetically derived sucrose to the 266 apoplasm of the phloem on the abaxial side of the leaf. Rank-2 veins are thought to be an emergent 267 phenomenon of C4 grasses (Sedelnikova et al., 2018). Rank-2 veins increase the ratio of BS to MS 268 cells, the vein density, and the capacity for nutrient transport. They appear to be the main path for 269 sucrose phloem loading. It is thus conceivable that the unique phloem loading pathway coevolved 270 with the evolution of the rank-2 intermediate veins.

271
Given the findings of Williams et al (Williams et al., 1989), which indicate that barley uses adaxial 272 and medial BSCs for phloem loading, our results suggest that the two species use distinct sets of 273 BSCs for transferring sucrose from the BS to the phloem. This may be generalizable to other C3 274 and C4 species, and it will be interesting to explore whether SWEETs are also present in medial

330
The raw data that support the findings of this study are available from the corresponding author 331 upon reasonable request. All sequencing data will be deposited in the Gene Expression Omnibus 332 GEO (www.ncbi.nlm.nih.gov/geo/) and the accession number will be updated here.

434
Tissue was sampled from the distal portion of leaf 2 (from 1 cm to 7 cm, as measured from the tip)

482
To aim for partitioning of 7,000 cells, with the expectation that 3,500 cells would be sequenced, 6 483 µL of the protoplast suspension with an estimated 1,200 cells/µL was applied to the 10x Genomics         Genes of interest differentially expressed between clusters ab BS and ad BS. 9 out of 39 genes specific to the ab BS cluster are specific to transmembrane transport (bold). Criteria for inclusion were average log fold change > 0.5 for all cells in subcluster and FDR-adjusted p-value < .01. The ab BSC specificity was validated for three genes, SWEET13a, b and c. Whether genes with lower FDR-adjusted p-values also show high specificity will require experimental validation.            Inset panel shows detail of sucrose movement from bundle sheath cells into apoplasm via SWEET13 transporters, or to phloem parenchyma (teal) via plasmodesmata, where it is then effluxed to the apoplasmic space by SWEETs. Sucrose in the apoplasm is taken up by SUT1 into the companion cells-sieve element (orange, pink) complex for long distance transport.

Supplementary text A subset of mesophyll cells appears to be specialized in iron metabolism
Subcluster MS5 (visible in the lower left corner of Figure 1b) shared mesophyll identity but appeared be specialized in metal accumulation and transport, as indicated by high and specific expression of four nicotianamine synthase genes, NAS1, 2, 9, and 10, which are involved in iron chelation; one iron phytosiderophore transporter. YS1 (yellow stripe 1); and various additional genes involved in metal transport and metabolism (Supplementary Table 1). M5 cells could either represent a cell type with a specific localization in the leaf or correspond to cells that contain different levels of iron.