The three main types of phloem loading currently recognized are (i) active apoplasmic, (ii) active symplasmic, and (iii) passive symplasmic ( Figure 1). The mechanism of active apoplasmic loading ( Figure 1) is well understood with proton-coupled sucrose transporters (SUTs), also referred to as SUCs, concentrating sucrose into SECCCs ^{8– 10} released from surrounding cells by sucrose uniporters, termed sugars will eventually be exported transporters (SWEETs) ^{11, 12} . In contrast, the function of the two symplasmic loading types has been established only recently.

A remaining major question for active symplasmic loading is how PD enable steric filtering to allow sucrose movement into CCs while preventing the synthesized sugar oligomers from leaking back to the mesophyll cells ( Figure 1). No method exists to resolve PD canal sub-structure to the degree required to answer this question, and current insights rest solely on mathematical modelling. For instance, Liesche and Schulz ¹³ showed that a PD structure that is highly restrictive to solute movement, with cytosolic channels of 0.65 nm in diameter compared with diameters of 2.5 nm in PD between leaf mesophyll cells ¹⁴ , could enable steric sugar filtering while accommodating observed phloem loading rates. However, this extreme PD anatomy might not be necessary as differences in intracellular sugar concentrations along the loading pathway from mesophyll to phloem could be sufficient to drive a combination of diffusion and mass flow of molecules (convective flow) into SECCCs. As a consequence, diffusion of sugar oligomers out of the SECCC, against the direction of the convective flow, would be limited ^{15, 16} .

New evidence regarding passive symplasmic loading ( Figure 1) was provided by the finding that sucrose export from poplar leaves was not compromised by blocking apoplasmic loading of sucrose ¹⁷ . This was achieved through expressing a cell wall invertase to cleave sucrose into hexoses, thereby making the sugar unavailable for uptake by SUTs. Furthermore, microfluidic experiments showed that passive phloem loading is feasible with convective flow through PD to the SECCCs ¹⁸ . However, it is unclear whether this concept is universally applicable across all species exhibiting features of passive symplasmic loading. For example, in several designated passive loaders, higher sugar concentrations were detected in phloem compared with mesophyll cells ^{19, 20} . In addition to methodological controversies surrounding measurements of intracellular sugar concentrations, a switch between loading types or even their concomitant operation could explain the conflicting conclusions ²¹ .

Coexistence of phloem loading types has become apparent for many plant species ²² . For instance, two anatomical types of CCs have been detected in minor veins in 35% of 320 Asteridae species exhibiting PD characteristics consistent with active symplasmic or apoplasmic loading ²³ as well as other species of various growth forms ^{24, 25} . Moreover, switching from active symplasmic to apoplasmic phloem loading, putatively by modifying PD conductance in the same CC, has been described for virus-infected melon plants ²⁶ . A switch between loading types in the same CC is thought to be impossible without adaptation of the relevant PD ^{21, 27} . Consequently, elucidating this potential mechanism of switching between loading types is a major task for future research on phloem loading.

In addition to switching between loading types, concomitant operation of passive and active loading by the same CC has been proposed ²⁰ and shown to be theoretically possible though attenuated by a low efficiency ²¹ . Passive loading could function as a fallback mechanism as demonstrated by continued growth, at a reduced rate, of plants in which active loading is blocked through introduction of mutations that render either the relevant SUT or enzymes involved in sucrose oligomerization non-functional ^{22, 28, 29} .

No satisfactory answer has been provided to account for why there are different phloem loading types. Slewinski et al. ²² theorized that some plants load certain molecules symplasmically in addition to those transported across the PM of SECCCs, but no candidates have been identified so far. Analysis of growth form concluded that herbaceous plants generally are not passive symplasmic loaders ³⁰ . Presumably, this is because active loading allows the plant to keep carbon content low in leaves, which provides a substantial, positive effect on growth rate ³¹ . However, this does not mean that the slower-growing trees are all passive loaders. Indeed, half of the families dominated by woody plant species exhibit an active loading mechanism ^{21, 30} . Evolutionary and distribution analyses similarly do not provide a clear explanation for why certain families use a particular phloem loading type ³⁰ . A closer examination of a plant’s ecology might reveal ecophysiological adaptations that different loading types offer. For example, a link between loading type and a plant’s adaptive environmental response has been demonstrated in a comparative study of minor vein structure developed under different temperature and light conditions ³² .

While sugar export rates are a function of sugar availability and sink demand ^{33, 34} as well as phloem loading capacity ^{35, 36} , the underlying molecular mechanisms are poorly understood. SUT activities are expected to be principal regulators of sugar export in apoplasmic loaders ⁶ as demonstrated by SUT overexpression ³⁷ . In this context, transcriptional SUT regulation by phytohormones and corresponding changes in sugar export rates were observed in potato ³⁸ . Several studies suggest that post-translational regulation of SUTs contributes to determining export rates (for example, Sakr et al. ³⁹ ). How post-translational regulation of SUTs occurs in vivo is unclear but could involve differential intracellular localization ⁴⁰ , dimerization ⁴¹ , and protein-protein interactions ⁴² . However, other factors should be considered. For instance, simultaneous upregulated expression of SUTs and SWEETs, in response to water deficit, increased phloem loading ⁴³ . Furthermore, constitutive overexpression of a proton-pumping pyrophosphatase in Arabidopsis thaliana leaf CCs increased phloem loading possibly by enhancing the proton motive force driving sucrose flux through SECCC localized SUTs ^{44, 45} . Regulation of sucrose exchange between cytosol and vacuole of cells in the pre-phloem pathway influences phloem loading rates even in apoplasmic loaders ^{46, 47} . So far not demonstrated is the short-term regulation of phloem loading through modification of PD permeability, which would be especially relevant in symplasmic loaders ²¹ . After elucidation of the molecular mechanisms regulating phloem loading, the important question will be how these are linked to sink demand ⁴⁸ .

Osmotically driven pressure flow has been widely accepted as the mechanism of phloem transport in herbaceous plants ^{49, 50} . However, in regard to trees, where distances between source and sink can extend up to 100 m, there are doubts about whether a hydrostatic pressure potential sufficient to drive flow could be generated ^{51, 52} . A variety of approaches have been employed to answer this question. Simple theoretical models of Münch-type pressure flow agree with measurements of stem diameter variations ⁵³ or SE anatomy ^{54, 55} . Importantly, the main prediction of pressure flow, that SE conductivity scales with transport distance ⁵⁶ , could be confirmed. The scaling of SE conductivity with tree height was shown within a single tree ( Figure 2A; ^{57, 58} ), within a species ⁵⁹ , and across species ( Figure 2B; ^{55, 60} ), confirming that resistance decreases to accommodate mass flow in larger trees. Furthermore, it was recently shown in mature, field-grown Scots pine trees that there is an osmotic pressure gradient along the phloem pathway from leaves to the stem base ( Figure 2C; ⁶¹ ). The osmotic pressure gradient, supported by gravity, was calculated to be large enough to overcome the xylem water pressure potential and establish a phloem turgor pressure gradient that drives mass flow according to the Münch mechanism at all times across the diel cycle ⁶¹ . Taken together, these results confirm that Münch-type pressure flow works even in the tallest angiosperm and gymnosperm trees, although transport speed might be up to 10 times lower than in herbaceous plants ⁶² . However, this is in agreement with equally lower rates of photosynthesis and growth ⁶³ .

Despite major progress in determining how phloem anatomy facilitates osmotically driven pressure flow, not all questions regarding its functional anatomy have been resolved. For example, sieve plates forming the axial connection between adjoining SEs have been assumed to evolve toward a simple form associated with lower hydraulic resistance—a paradigm overturned by an anatomical analysis of 447 species ⁶⁰ . Rather, the distinct organizational patterns of sieve plates might relate to general vascular anatomy (for example, SE length or xylem cell anatomy) instead of differences in conductivity ⁶⁰ . Another aspect that is relevant for phloem transport is the symplasmic coupling along the transport path. Even in apoplasmically loading herbaceous plants, unloading along the stem can switch between apoplasmic and symplasmic unloading pathways, depending upon the source/sink ratio, with high ratios favouring symplasmic unloading ⁶⁴ . In many tree species, especially conifers, the presence of PD connecting the SECCC with surrounding cells ⁶⁵ indicates symplasmic coupling, presumably to efficiently supply the cambium with the large amounts of sugars needed for wood formation. The question is in how far symplasmic coupling along the path compromises whole-plant phloem transport. A recent theoretical analysis showed that unloading along the transport path considerably influences flow if it is not balanced by reloading ⁶⁶ . The most likely explanation is that unloading is symplasmic only in case of high sugar concentrations in the ground tissue, which then could also have a function in homeostatically maintaining SE turgor ⁶⁷ .

Another central issue is the relationship between fibrous SE protein (p-protein) agglomerations, considered to occlude sieve pores to prevent sap loss in the event of damage ⁶⁸ . Careful preparation of material for electron microscope examination reveals unobstructed SE lumens bounded by a parietal cytoplasm. However, Froelich et al. ⁶⁹ showed that in A. thaliana phloem, the p-protein, AtSEOR1, forms a meshwork at the margins and clots in the lumen of intact SEs and their presence does not impede longitudinal flow. Together with discrepancies between theoretical flow speeds calculated according to SE conductivity and observed flow speeds ⁷⁰ , these observations have led to the conclusion that phloem transport in herbaceous plants and grasses is not limited by SE conductivity but only by sink strength ⁷¹ .

Knowledge of phloem sap composition is relevant to reach a quantitative understanding of resource allocation and inter-organ signalling. Sugar concentrations are typically around 20%, but the sap contains other molecules (for example, amino nitrogen compounds, proteins, and RNAs) and mineral ions ⁷² . Determining the precise abundance of the different sap components remains one of the biggest challenges in phloem research as all methods available so far are prone to artifacts ^{73– 75} . Recently, laser-capture microdissection was used to uncover the link between seasonal differences in phloem metabolites and phloem formation in Norway spruce. However, since sampling could not be restricted to SEs, results were not representative of phloem sap ⁷⁶ . In the future, insight on relative sap composition might be gained from using rapid freezing and nanoscale secondary ion mass spectrometry (NanoSIMS), which enables semi-quantitative imaging of ions as well as isotope-labeled nitrogen and carbon compounds within SEs ⁷⁷ . Gaining a complete picture of the spatiotemporal dynamics of phloem sap composition will be instrumental in refining multi-compartmental metabolic models describing source-to-sink metabolite flows (for example, Zakhartsev et al. ⁷⁸ ).

Controversy persists surrounding the function of phloem-mobile proteins and RNA ⁷⁹ . Recently, a meta-analysis comparing proteomes of xylem, phloem, and leaf apoplasmic saps highlighted phloem sap as being the most enriched in signalling proteins of which 13 were conserved across experiments and that included the ubiquitous phloem-mobile protein FLOWERING LOCUS T ⁸⁰ . Grafting studies suggest phloem mobility of a multitude of additional proteins ⁸¹ . However, without further validation of their signalling function, many of these proteins could have entered the translocation stream “by accident” rather than by a control mechanism located at PD interconnecting CCs with SEs ⁸² . Of the proteins identified in phloem sap by proteomics, those associated with redox regulation and plant defense consistently have been found across species ⁸⁰ , suggesting that many phloem proteins function in the response to biotic and abiotic stress. Prominent examples of phloem-mobile RNAs include the transcription factors BEL5, which influences potato tuber induction ⁸³ , and the Cucurbit NACP regulating apical meristem development ⁸⁴ . Recently 2,006 A. thaliana phloem-mobile mRNAs were identified ⁸⁵ with a conserved tRNA-derived sequence conferring mobility ⁸⁶ . However, since most of the mRNAs are expressed in mesophyll cells, it is unclear how they could enter the phloem stream in the apoplasmic loader A. thaliana. In addition to those many phloem-mobile proteins and mRNAs, the role of microRNAs ⁸⁷ and lipids ⁸⁸ as systemic signals needs to be further explored.

Possible mechanisms contributing to efflux across any cell PM, including SECCC PMs, are passive diffusion, transporter-mediated facilitated diffusion, and energy-coupled movement. Passive diffusive fluxes of molecules and ions are the product of their membrane permeability coefficients and trans-membrane gradients in chemical potential for non-electrolytes or electrochemical potential for electrolytes ⁹⁰ . To illustrate the magnitude that these passive fluxes could reach for apoplasmic SECCC unloading of sucrose along the entire phloem pathway from source to sink, a plausible maximum of 1 M for the SECCC trans-PM sucrose concentration difference and a membrane permeability coefficient of 10 ⁻¹⁰ m s ^{−1 91} predict a sucrose efflux of 1 × 10 ⁻⁷ mol m ⁻² s ⁻¹. This matches the maximal membrane flux for loading sucrose into SECCCs (2.3 × 10 ⁻⁷ mol m ⁻² s ⁻¹; Giaquinta ⁹² ). Thus, it is imperative to experimentally determine rates of diffusive leak from SECCCs and include this component in theoretical models of phloem transport.

For facilitated membrane efflux of sugars by uniporters, SWEETs ^{11, 12} have been detected in transport phloem parenchyma cells ⁹³ , but whether these function in efflux to, or retrieval from, the phloem apoplasm remains to be determined. However, sucrose/proton symporters occur in SECCCs of root and stem transport phloem (for example, 94, 95). Depolarized membrane potentials (−55 mV) of root SECCCs predict that sucrose/proton symport could reverse to an efflux mode driven by a trans-membrane sucrose concentration difference of 85.5 mM (for more information, see Carpaneto et al. ⁹⁴ ). In contrast, symporter reversal in stem phloem is highly unlikely as their SE/CC membrane potential of −110 mV ⁹⁶ would require a sucrose concentration difference of 7,200 mM, a concentration difference that far exceeds physiological limits.

During sugar accumulation in fleshy fruits, phloem unloading follows an apoplasmic route ^{64, 97– 100} . Whether the sucrose leak from fruit SECCCs is augmented by reversal of sucrose/proton symporters ^{101, 102} cannot be evaluated, as their membrane potentials are unknown. However, pharmacological studies suggest that sucrose efflux from vascular bundles of apple fruit and grape berries is mediated by energy-coupled carriers ^{97, 103} . Resolving the identity (or identities) of the putative effluxer (or effluxers) is central to acquiring a full understanding of phloem unloading in fleshy fruits. What is clearer is that monosaccharide/proton symporters retrieve inverted disaccharides from the fruit apoplasm into storage parenchyma cells ^{104– 107} and, together with cell wall invertases, function to co-regulate phloem unloading ^{106, 108, 109} .

In all cases, unloading of osmotic solutes must be accompanied by a proportionate loss of phloem water to maintain water potential equilibrium. The exit of water likely occurs through aquaporins localized in PMs of the unloading SEs ( Figure 3A and, for example, 110, 111).

In recent years, it is becoming increasingly clear that a major component of symplasmic unloading from SECCCs is contributed by bulk flow. For instance, experimental manipulations of hydrostatic pressure gradients in root tips and stems cause changes in phloem unloading rates consistent with bulk flow (reviewed by Patrick ¹¹² ). In developing wheat seeds, hydrostatic pressure differences of up to 1.0 MPa between SEs and vascular parenchyma cells account for observed PD volume flow rates ¹¹³ . In contrast, modelling flows through large-diameter funnel-shaped PD, interconnecting SEs, and adjacent pericycle cells in A. thaliana root tips predicted that hydrostatic pressure differences of only 0.05 to 0.2 MPa were required ¹¹⁴ . Whether these model-based hydrostatic pressure differences can be reconciled with measured hydrostatic or osmotic pressures of 1.3 and 0.6 MPa, respectively, between SEs and surrounding cells in barley and maize root tips ^{115, 116} awaits determination.

For the apoplasmic step in the mandatory phloem-unloading pathway of developing seeds ( Figure 3B), sucrose release from maternal seed coats occurs by a combination of facilitated diffusion and sucrose/proton antiport while retrieval by embryo/endosperm is mediated by sucrose symporters ¹¹⁷ or SWEETs ¹¹⁸ or both. Cloned sucrose facilitators (SUFs) from grain legume seed coats exhibited transport properties consistent with those found for facilitated diffusion of sucrose in native membranes of seed coats and were expressed in cells considered responsible for sucrose efflux ^{119, 120} . In developing A. thaliana seeds, SWEETs are positioned to provide a cascade of sucrose transport from the outer (SWEET15) and inner (SWEET12) integuments and uptake into the filial tissues (SWEET11 and SWEET15 ¹²¹ ) while SWEET11 facilitates release from maternal tissues of developing rice seeds ¹²² . Consistent with relative contributions of facilitated diffusion to sucrose effluxed from grain legume seed coats ¹¹⁷ , seed weights, but not number, were depressed by 50% in the A. thaliana triple-mutant sweet11, 12 and 15, and by 65% in the rice sweet11 mutant ^{121, 122} . The challenge now is to identify the gene or (genes) encoding the membrane protein (or proteins) responsible for energy-coupled sucrose/proton antiport from maternal seed tissues ¹²³ that may also operate in fleshy fruit (see previous section). Membrane transport of solutes is matched by water movement through aquaporins located in PMs of maternal and filial tissues of developing seeds ( Figure 3C and, for example, 124, 125).